USVI Public Records

Document text

•\.' Advanced Concepts in j Environmental Microbiology . k Manual to Laboratory Training Seminar #2 Presented by: Robert H. Ruskin,. Jr. •'!..'.' ' and the staff '*..*" of the Water Resources Research Center Caribbean Research Institute University of the Virgin Islands Agreement No. 14-08-0001-G 1455 I-T^ie research on which this report is based was financed in - pa'rt by the United States Department of the interior, Geological Survey, through the Virgin Islands Water Resources Research Center. Contents of this publication do not necessarily reflect the views and policies of the U.S. •" Department of the Interior, nor does mention of trade names .or commercial products constitute their endorsement by the ^"United States Government. X MARCH 1975 The Necessity Bacterial Potable Waters Edwin £ Geldreich, Harry D. Nash, Donald J. Reasoner, andRaymondH. Taylor An article based on a paper presented at the annual meeting of the Amer. Soc. forMicrobiology, May 12-17, 1974, Chicago. 111., andcontributed to the JOURNAL on Dec. 13. 1974, by Edwin e'. Geldreich (Active Member, AWWA). consult, bact; Harry D. Nash, res. bact.; Donald J. Reasoner (Active Member, AWWA), res. bact; and Raymond H. Taylor, res. bact (Active Member! AWWA); all of the Wtr.Sply.Res.Lab., NERC, ORD, USE?A, Cin cinnati, Ohio. Part of the Safe Drinking Water Act stipulates that communities provide for emergency supplies of water that are safely free of bacteria. This article, in addition to documenting findings about bacterial counts for bot tled water, provides data and recommendations con cerning emergency water supplies. The growing public concern over water pollution, the fact that numerous municipal water supplies use polluted streams for source water, opposition to the addition of fluoride to some municipal water supplies, and the taste and odor problems in some community water supplies have stimulated growth of the bottled-water industry in the US. In 1972 there werean estimated700 water-bottling plants in the US, which accounted for at least SI07 million in total sales. In southern California, where the sales of bottled water are the largest in the nation, more than 100 mii galof bottled waterare sold yearly to an estimated 930 000 customers.1 E. E. GELDREiCH ET AL "11 7 * Some of the large bottled-water companies now reach large regional markets throughout the US. With the increased geographical distribution and growth in bottled-water sales,an increasing proportionof the population is exposed to a potable- water source of uncertain bacteriological quality. The need for systematic bottled-water-quality surveillance is a growing concern. Some states require that bottled-water- quality standards comply with the USPHS;3 however, most states either have.no regulations, or the existing regulations are ineffective or hot enforced. The state regulations range from programs with a poorly defined sampling frequency to others that specify a minimum of five samples per month if the population served is fewerthan 6000andrequire an increasein sampling frequency proportional to the increase in the popula tion served.3 The number of samples required is based on the sampling frequency curve from the USPHS Drinking Water Standards,as shown in Fig. 1.Unfortunately, few states specify a definite sampling frequency for bacteriological examination or repeat testing when the initialsample results areunsatisfac tory. The assumption that bottled water is of unquestionably good qualitysimply because the source.water is a spring or artesian well is untenable. The labeling of some bottled waters may even imply that the bottled waters are derived from pristine sources that do not require treatment and consequently are superior to municipal water supplies that utilize polluted raw water sources. The implication is that bottled-water source waters are initially of better quality and that the finished prod ucts will retain this high quality during shelf life before purchase or when used in water cooling dispensers. The. purpose of this study, conducted over a 1 1/2-year period, was four fold: 1.Investigate the bacteriological qualityof a variety ofbrands of bottled water purchased from retail outlets 2. Investigate the variability in bacteriological quality of freshly bottled water 3. Characterize changes in the bacterial density of bottled water during storage 4. Examine the bacteriological quality of stored emergency water (civil defense). Materials and Methods S.mpU collection.Half- or full-gallon (1.9-3.8-1) containers of available brands of bottled water were purchased from local retail outlets for bacteriological examination. After initial ex amination, some bottled water samples were examined repeat edly over a period of 30 days or longer to evaluate changes in bacterial density during storage. In addition, samples of freshly bottled water, obtained directly from the bottler, were examined for bacteriological quality. A portion of these samples was collected in conjunction with a nationwide bottled-water study conducted by the Water Supply Div. USEPA,4 in which 25 water bottlers were surveyed. Five water bottlers in the Cincinnati, Ohio, area were included in that survey. Source-water samples collected in sterile, 1-1 polypropylene bottles also were obtained from the water bottlers. All bottled water and source-water samples were transported (unrefriger- ated) to the laboratory within 6 hr for analysis. Some samples of freshly bottled water were held in the laboratory at 23 ± 2C for repeat sampling to determine how the bacterial populations changed during storage; After the initial examination, repeat examinations were conducted over a period of 63 days. To obtain samples of civil defense emergency drinking 118 MANAGEMENT Mmnnuffl NwnMf ol Sampkt par Month Fig. 1. Recommended MinimumMonthlySamples per Population Served by Water Supply TABLE 1 Effect ofIncubation Time on Bottled Water Plate Counts at 35C Count No. oT JUtio No. of. Ratio No. of - Ratio SKM Analyses 24:72 Hr Analyses 41:72 Hr Analyses 120:72 Hr 30-100 21 0.25 37 0.53 17 1J2 101-500 14 0.25 28 0.6S 14 IJ3 501-1000 10 0.22 13 0.61 3 1.09 1001-JO 000 33 0J0 44 0.60 12 US 10 000-50 000 21 0.23 37 0M 7 1.11 50000.100000 11 0J1 17 0.81 • 6 0.99 >100000 12 0.06 19 0.62 8 1.03 Tout 129 ISO 51 Avenge ratio 0J3 0.67 r.i5 TABLE 2 Ranges ofStandardBacterialPlate CountsinFreshly Bottled Water Samples* Count No. of Samples Cumulative SPOmt Samples • ptrcait percent <l 23 17.8 17.8 1-10 56 43.4 61.2 10-100 28 21.7 819 100-500 8 6J 89.1 500-1000 5 3.9 93.0 1000-10000 7 5.4 • 98.4 <10000 2 1.6 100.0 'All samples were plated within 48 hr o( bottling. JOURNAL AWWA g Initial Count Q Maxium Count-Room-TtmiMr*tur« Storag* g MuNnumCount-flafr^araMStortgt Fig. 2. The Storage s BL Samol* NuniMr Effect of Refrigerated and Room-Temperature on the Standard Plate Count of Bottled Water Fig. 3. Free-Chlorine Residual From Chlorine Tablets and Liquid Bleach in Civil Defense Shelter Water-Storage Drums/:: Liquid bleach:-—; tablet:-~ MARCH 1975 water,the drum lid wasrempyed and the top ofthe inner poly ethylene storage bag was swabbed with ethanol; a 2.5-cmcut was madein the polyethylene bagwithastenie kniife blade; ap proximately 400 ml of water waspipetted (using a sterile 100- mil volumetric pipet) irqmjthe storage drum into asterile 500- ml flask; the opening in :thei polyethylene liner,was'resealed with tape; and the drum lidwasreplaced. Sample analyses were initiated within 1 hr after collection. -..-.. Bacteriological taata. 1. Standardplate count(SPC). The SPCof bottled (freshly bottled and retail-purchased)-and-stdred emergency drinking-water samples was determined using the pou r plate procedure described in Standard Methods} Sample aiiquots of 1.0, 0.1, and 0.01 ml were planted, five replicates for each dilution,in disposable 15- x 100-mmplastic petri dishes. A 15-ml volume of sterile plate-count agar (used for all SPC determinations), tempered, to 44-46Q was poured . into each plate and thoroughly mixed with the sample aliquot The agar was allowed to solidify and theplates inverted and in cubated at 35 ±0J>C.The plates were examined and colonies" counted after 24, 48,17, and 120 hr incubation. Results were expressed as SPC/ml. .2. Total and fecal coliform. Total- and fecal-coliform-count determinations were conducted according to Standard Methods membrane filtration procedures.5 Separate 250-rhl volumes of samples were examined for fecal and totaicoliform determina tions for bottled water. For each fecal- and total-coliform deter- mination of source-water samples, a 100-ml volume was ex amined. Results were calculated and expressed as total col-- iforms/100 ml and fecal coliforms/100 ml. 3. Pseudomonas aeruginosa counts. The membrane-filter method of Levin and CabehV utilizing M>PA agar, was used to ' examine bottled,source,andstored emergency-drinking-water samples for the presence of Pseudomonas aeruginosa. Sample volumes of 250 ml of bottled water and 100 ml of source water and emergency drinking water were examined for Pseudomonas aeruginosa. Results Incubation tlma. At the start of this study,-considerable difficulty was encountered in determining the SPC beca *. < the slow development of the bacteria" on plate-count agai ifte; an incubation periodofonly 24 hr at.35C Consequently, the in cubation period at 35C was extended to as long as 120 hr (five, days).Colony counts were made after incubation periods of 24, 48,72, and 120 hr to determine the optimum incubation period needed to obtain valid counts...._.. The data summarized from SPC analyses of 180 bottled- water samples are shown in Table 1.'Data were derived from initial examinations of bottled waters and subsequent repeat examinations after various storage intervals at.23 ± 2C. F bacterialcount used was an averageof five replicate plates,*. .u only plates that had counts of 30-300 colonies after a 72-hr in cubation at 35C were included. The data in Table 1show that the bacterial counts.after a24- and48-hrincubation averaged only 22and67 percentV.respec- tively, of the 72-hr count. Increasing bacterial densities generally had little effect on these ratios.The notable exception was the 24:72-hr count ratio for densities^lOO 000 bacteria/ml. This indicates that when the bottled-water bacteriar de'nsify'is very large, the .24-hrcount at 35Capproximated fewer than ,10 per cent of the 72-hr count. The incubation period for.the bacterial plate count was ex tended to 120 hr (five days) to determine if a longerincubation period would result in further increases in the plate counts of E. E. GELDREICH ETAL 119 1ABLfci Standard Bacterial Plate Counts in Bottled Water Obtained From Retail Outlets* Count No. of Samples Cumulative SPOml Samples pvetnt porcoi/ 0 11 10.9 105 1-10 20 19.S 30.7 10-100 19 IE. J 49.5 100-SOO 9 8.9 St.* 300-1000 6 5.9 64J 1000-10000 16 13.9 noi > 10 000 20 19.8 100.0 •Samples purchased after undetermined periods is stock bottled-water samples. Data from 51 samples (Table I) indicate that increasing the incubation period from 72 to 120 hr pro ducedonly a 15 per cent increase in count The sample-to-sam plevariation in the percentage increase in bacterial count was found to be greater between the 72- and 120-hr incubation than between the 48- and 72-hr incubation time. Because of the relatively small increase in bacterial count, extending the incubation period from 72 hr to 120 hr hadlittle advantage for most samples. Froshiy bottlad watar. A total of 129 freshly bottled water sam ples was collected directly from 25 different bottlers. Of them, only 14(10 percent) had an initial SPCgreater than 500bacte ria/ml (Table 2). Total coliforms were detected in six samples; only two of them exceeded trie USPHS Drinking Water Stan dards cbliform limit (for an individual sample) of 4 col iforms/100 mil. Also, one of the two coliform-positive samples contained fecal coliform bacteria, and the other coliform-posi tive sample contained 42 Pseudomonas aeruginosa/100 ml. Bottlad-watar aampiat of unknown aga. The bacterial densities of bottled water purchased from retail outlets were highly variable.The data in Table 3 show that 42 of 101 samples (41.6 per cent) contained more than 500 bacteria/ml, and about 11 per cent contained fewer than 1 bacterium/ml. Three samples had total coliform counts that exceeded the USPHS Drinking Water Standards. Fecal coliform bacteria were not detected, but one sample contained Pseudomonas aeruginosa at a concentra tion of 13/100 ml. Table 4 shows the variability in bacterial density and the fre-. quency of coliform-positive samples among the 19 brands of bottled water analyzed during this study. Examination of the density ranges of bacteria reveals the variability of the SPC from brand to brand as well as the variability within the same brand where multiple samples were analyzed. Comparison af the data in Tables 2-* shows that the bac teriological quality of bottled water examined within 48 hr of bottling was not representative of the. quality of bottled water obtained from the retail shelf after an undertermined period in stock. Forty-four bottled-water samples purchased from retail stores were stored at room temperature for 30 days and sam pled weekly. The maximum SPC/ml recorded for stored bot tled-watersamples are grouped by range in Table 5-The SPCin 33 of the 44 stored samples (75 per cent) increased to more than 500 bacteria/ml during the storage period; this indicates that sufficient nutrients were present in the bottled waters to support significant increases in bacterial densities. In addition, changes in bacterial populations of six freshly bottled water samples stored at 23 ± 2C were observed for a period of 63 days. These samples were collected and examined in conjunction with a nationwide pilot survey of water bottlers and bottled water.4 The data in Table 6 show the bacterial den- 120 MANAGEMENT TABLE 4 Comparison ofStandardPlate-Count Variability AmongBrands. ofBottledWater CoofonB Sample No. of Count Ranget rosidv* Brand Type* Samples SPOM Simplest A fresh 91 00-23000 3/91 .retail 16 00-28000 0/16 B retail 6 O0-260 000 0/4 C fresh 1 <10 0/1 retail 11 1200-160000 0/11 D retail 5 31-650 0/3 E fresh I <10 0/1 retail 2 100000-390000 2/2 F retail 2 O0-12000 0/7 G fresh 2 oo 0/2 retail 1 <10 0/1 H retail 9 00-390 1/9 I retail 2 oo on I retail 2 O0-2000 1/2 K retail 2 O0-12 1/2 L retail 4 13-4300 0/4 M retail 1 oo on n . retail 1 <10, 0/1 0 retail 1 1X106 1/1 p . retail 1 1600 . 0/1 Q. retail 1 <10 0/1 ;.' retail 1 oo 0/1 retail . 1 12 0/1 'Fresh-samples direct from bonier, examined within 24 hr; retail - samples of unknown ate purchased from retail outlets. tSundard plate count (SPC) rante values represent average count of bacteria/ml. calculated from five replicate plates incubated for 72 hr at 33C using plate count agar. . tOne or more cotifonns/lOO mL {Imported bottled water, carbonated. TABLE 5 Maximum Standard Plate Counts Attained in Retail Purchased Bottled Waters DuringRoom Temperature Storagefor 30 Days Maximum Count•*- No.of Samples Cumulative SKM Samples perctnt ptrctnt <1 2 4.5 4.5 1-10 . 4 9.1 13.6 10-100 4 9.1 22.7 100-300 1 13 25.0 300-1000 1 2J 27J 1000-10000 13 29.5 56.8 > 10000 13 43.2 100.0 TABLE 6 Variation ofSPC inFreshBottled WaterSamplesFrom TheCincinnati. Ohio. Area During Storage at 23C ± 2C* Storage Period. Sample Number days 1 2 3 -4 5 • 6 0 4f 2- 3 1 1 3 3 20000 350 11000 5 3 4 5 52000 250 21000 31 39 4 7 43000 1300 1)000 36 25 69 10 76000 670 70000 130 7 2700 14 47 000 710 18000 1100 39 29000 21 40000 17D0 13000 2200 2200 . 100000 28 35 000 2100 2200 1200 1300 53 000 . 35 33 000 1400 1400 . 690 880 56000 42. 44000 1900 1400 650 1100. 53000 49 48000 49O0 . 660 700 1100 46000 56 46000 1000 490 310 460 50000 63 . 61000 1200 490 240 360 44 000 'Storage study period for all samples began within 24 hr of bottling. tAverage counts/ml, calculated from five replicate plates incubated for 72 hr at 35C with the use of plate count agar. JOURNAL AWWA TABLE 7 Standard Plate Counts in StoredCivil. DefenseEmergency Water Avg. Count 72-Hr Incubationt Five-day Incubation! Count Ratio STOW No. Samples ptrctnt No. Samples ptrctnt 72:120 Hr 0-10 4 13J 2 9J 1.00 10-100 .6 20.0 2 9J 0.78 100-500 1) 36.7 3 14.3 0,50 500-1000 2 6.7 6 216 0.38 1000-10000 I 3J 2 9J 0.10 > 10000 6 20.0. 6 216 0.44 'Each vahao used was aa average of dye replicate plates, incubated for either 72 hr or 120 hr (flvo days) at 35C tlncrudes all samples for which 72-hr SPC was determine! ttacludol only samples for which both the 72-hr and five-day SPC were determined. TABLE 8 Oianges in Bacterial Quality Following Initial Disinfection of Civil Defence Emergency Stored Water Occurrence—per cent Count . SKJM Months •Years 1 3 4 6 8 12 3 ' 6 9 <> 45.1 39J 29.2 29.6 36.4 29.4 71.8 73.6 10.0 1-50 17.2 22.8 9.7 16.2 117 19.1 4.9 none 10.0 51-100 11 U 13 •11 3J 19 1.0 5.3 15.0 101-500 .9.4 10.1 13.9 17 14J 13.2 5.8 none 30.0 501-1000 11 6J 4.2 17 3.2 4.4 1.0 none 5.0 1001-5 000 6J 8.9 13.9 203 117. 17.7 • 17 5J none 5001-10000 1.7 6.3 SJ 6.8 3.2 • U 4.9 none 10.0 10001-50000 9.4 1J 11.1 6.8 7.9 7.4 1.0 105 15.0 >50 000 4.7 3.8 1.4 6.8 6.4 !• 4.4 none SJ 5.0 Total samples 64 79 72 74 63 68 103 19 20 TABLE 9 Emergency Disinfection Treatment forStored Supplies of Drinking Water Approximate of Available Treatment Commercial Disinfectant Disinfectant Volume Product ptrctnt Quantity* Treated—xo/ Source Availability High-test 70 1 tablet 80 janitorial, hotel, restaurant. hypochloridc dairy, and swimming. tablets pool supplies Iodine or chlorine —. 2 tablets 1/4 drug store and sporting tablets • goods store iodophoret - kO-1.6 1 lbs 4-45 drug store, dairy and chemical supplies. Liquid chlorine 5J Ijbs 20 grocery store laundry bleach Tincture of iodine 2 1 tbs 8 drug store Chlorine based. 1.0 1 tbs • 4 drug store astisepuc powder t Ten-mg/l dose. 30-miri contact time tbodine, Blair Co: Wescodyne. West Chemical tZonite. Norcliss Ca MARCH 1975 sity changes thatoccurred in each of the sixbottledwater sam ples starting within 24 hrofbottling. Each standard plate count . in the table represents the average of.five replicate plates incu bated for 72 hr at 35C These results indicate that bacterial densities uvbottled water increase appreciably during storage and raise questions con cerning (1) possible interference with coliform detection, and (2) the possibility ofincreased risk of human exposure to bac teria known to be secondary pathogens. Storage effects. To determine if the increase in bacterial den sity in bottled water could beminimized during storage, seven pairs of bottled-water samples were purchased. One bottle from each pair was stored at refrigerator temperature (4 ± 2C) and the other stored at room temperature (23 ± 2C) for the dura tion of the storage experiment Samples from each bottle were examined weekly for eight weeks to observe changes in bacte rial density. Results from six of the paired samples showedthatthe max imumbacterial density attained in thestored refrigerated sam ple was significantly lower than that attained in the stored unrefrigerated sample (Fig. 2).In theremaining paired sample, maximum counts of less than20bacteria per ml were observed in both the room temperature and refrigerated samples. The peak bacterial density in the refrigerated samples occurred after an average storage of 26 days, as opposed, to only eleven days storage for the room-temperature samples. Emargancy watar. Watersamples werecollected from 17.5-gal (66.2) drums of civil defense emergency water that had been stored for ten years.The ranges and the distribution of SPCs for 30 emergency water samples collected during, the period Jul. 1972 to Apr. 1973 are listedin Table 7.SPCs afterincubation at 35C for 72 hrand 120 hr(five days) and theaverage 72:120-hr count ratios for the platecount rangesare alsoshown. The data show that 21 of 30 samples (70 percent) had a SPC of fewer than 500 bacteria/ml after 72-hr incubation at 35C.Of the sam ples for which SPCs were determined after both 72-hrand five- dayincubation,however,only 7 of 21 (33.3 percent) contained fewer than 500 bacteria/ml.These data indicate that the five- day-incubation SPC at 35C provides a better estimate of the bacteriai densityjof stored emergency water than does the 72- hr incubation. Total coliform, fecal coliform, and Pseudomonas aeruginosa were not detected in any of the 30 emergency water samples analyzed. . Discussion Bottiad watar. fre»h and itorad. Concern over the bacteriologi cal quality of bottled water dates back at least to 1916.7 How ever, there is little documentation of the variability, of bottled- water quality. These results indicate (Table 1) that bottled water samples should be incubated at 35C for 72 hr to obtain reliable SPC data. Many of the bacteria.detected in bottled water grow slowly when inoculatedinto the nutrient-rich medium used for the SPCanalysis.Therefore, SPCanalyses ofbottled watersam ples using 35C for 24 hror20C for a48*hr incubation generally will not provide an accurate estimate of the bacterial densities in the samples. The sameconclusion applies totheexamination of stored, emergency water samples (Table 7J,_except thatin-. . cubation for 120 hr (five days) at 35C is needed to obtain relia ble counts. Therefore, minimum incubation periods of 72 hr for SPC analyses of bottled-water samples and.120 hr for stored emergency water samples at 35C are recommended. The plat ing medium should be plate-count agar ortryptone glucose ex tract agar.5 E. E. GELDREICH ET AL 121 Although a specific statement may not be present on the brand label ofabottled-water product, thelabel mayimply that thebottled-water source isof better.quality than the municipal supply source and that the finished product will retain the ini tial high quality throughout itsshelf-life. The results obtained do not support this implication: the bacteriological quality of the freshly bottled product varied greatly from brand tobrand and from sample to sample within the same brand. Although the data are not presented here, source-water samples from brands A, C, E, and G exhibited wide variability in bacterial content The method of treatment before bottling appeared to be more important than source-water quality in determining the bacterial quality of the finished product The type of treatment used for the production of bottled water was variable. The data collected during the bottled-water survey4 show that the use of a single means of treatment was infrequent—usually a series of treatment steps was used. The bottled-water treatment process ranged from no treatment to a combination of treatment steps that might include softening or ion exchange, filtration, and disinfection. Ozonation orultraviolet irradiation was most frequently used as the final disinfection step before bottling. The bacterial quality ofbottled waters disinfected byeither ofthese methods was highly variable among bottlers and for a given bottler at different times. Of the various bottled waters examined in the Cincinnati area, one brand was pasteurized and its usually good quality was maintained during storage. In general, bottled waters treated by. an ion-exchange column (or bed) or by filtration alone were poorest in quality because ofthe bacterial growth on the ion ex changer or filter medium asaresult ofinfrequent orinadequate maintenance of the equipment Wallis and associates8 have shown that charcoal filters used to remove objectionable tastes and odors from drinking water can support large bacterial populations. The charcoal filters concentrateboth bacteriaand organic nutrients and then pro vide a place for the bacteria to multiply. Bacterial concentra- • tions as high as 7.0 x 10* cells/100 ml were detected in the effluent from a charcoal filter only six days after installation. Beds or columns of ion-exchange resins or activated carbon may support similar growths ofbacteria unless properly main tained and serviced. Good quality bottled water was generally produced byacombination of filtration or distillation and ozon ation of ultraviolet irradiation. The use of protective sealed caps on bottles and covered storage during transit to the consumer have reduced the risk of bottled-water contamination. Unprotected storage in business establishments and homes, poor handling techniques, and wa ter dispensers ofquestionable cleanliness may, however, result: in rapid deterioration of bottled-water quality. Few coliform bacteria were recovered from bottled water during this study. During astudy conducted in 1936,9 coliform bacteria weredetected in 32 percent of 212bottled-water cool ing dispensers, and 8.5 per cent of thesamples contained ap proximately 1000 coliforms/100 ml. A more recent study re ported coliform bacteria in only 6.4 per cent of 1040 bottled- water cooling dispensers, and counts ranged from 2.2 to 240 coliforms/100 ml.10 Apparently, better cap protection, im proved dispenser designs, nonreturnable plastic containers, and improved handling and sanitation of water-cooler dispen sers have reduced the coliform occurrence. In somecases, however, new nonsterileplastic bottles can be a source of contamination. The bottles are shipped by the manufacturer in unsealed cardboard boxes containing six un- 122 MANAGEMENT Capped bottles; the caps, packaged in plastic bags, are shipped separately in bulk quantities. Eight unused bottles were examined by rinsing each bottle with a volume of sterile buffered water followed by (1) plating 1-ml aliquots in plate-count agar andincubating at 35C for 72 hr or (2) filtration of 50-ml aliquots through sterile 0.45-/tm membrane filters that were then placed on a modified plate- count medium and incubated at 35C for 72 hr. These new plastic bottles contained a low density of bacteria (counts ranged from <1 to 5 bacteria/ml) that contributed to the con tamination of the finished product and resulted in decreased product quality. Persistence of initial bacterial contamination, whether col iform or nonconform, will be determined by the availability of bacterial nutrients, water temperature, pH, and antagonism or competition of otherbacteria presentin the water.Inonestudy of bottled spring water, coliform bacteria in the contaminated bottled-water samples persisted for periods of fewer than two weeks.10 A die-awayinterval of nine to fourteen days was noted for fecal coliforms in spring water from another geographical area,11 but the total coliform population persisted for fourto six weeks, undergoinga declining cyclic survival pattern.Numbers of parallel studies in different types of water have shown that fecal coliform and Salmonellasurvival rates are essentially iden tical;12 therefore the occurrence of fecal coliforms in spring water could indicate the possible occurrence of pathogens. In addition to indicator bacteria, a large, general, bacterial population may occur inbottled water; thispopulation depends on the kinds ofbacteria present and the availability of nutrients in the water. Pigment-producingbacteria are frequently found inbottled,distilled, andstoredcivil-defense water.15-16 When bottled-water samples examined during this study contained pigmented bacteria, the proportion of the standard plate count represented by the pigmented forms ranged from only a few per cent to nearly 100 per cent. In some bottled- water samples, the bacterial population included yellow, orange, or pink pigmented forms. Yellow forms were most often encountered, but in two instances a pink pigmented bac terium accounted for nearly 100 per cent of the bacterial popu lation. The pigmented forms were predominantly gram-neg ative rods that varied widely in size. Bacteriological results (Table 7) obtained during this study from stored emergency water-supply samples were in good agreement with results obtained for a previous study. In this previous study the quality of emergency water sup plies located in twenty cities was examined over a twelver month period to determine the effects of long-term storage.17 Periodic coliform and SPC analyses were performed on samples from twenty 17.5-gal (68-1) water-storage drums during the first year of storage. The water, which was obtained from the. local community water supply, wasstored in a polyethylene bag surrounded by an outer polyethylene liner. At the time of fill ing, 7 ml (1 teaspoon) of 5.25 per cent chlorine bleach was added to each drum of water to protect against accidental con tamination during filling and handling operations. The long-term protectionafforded by chlorinedisinfection is, however,limited by the chlorine demand of (1) organic and in organic substances in the water, (2) the polyethylene bag, and (3) dust contamination accidentally introduced during long- . term storage. Studies indicate that effective free chlorine re- siduals.resulting from liquid laundry bleach or chlorinetablets placed into civil defense water-storage drums17 probably per sists forseveral weeks (Fig. 3). These dataalsoindicate that the liquid chlorine solution was instantly soluble and gave aslightly JOURNAL AWWA higherconcentration of available chlorinethan didthe chlorine tablets. ' In samples from a total of 87 storage drums in the twenty cities,unsatisfactory coliformresultswere recorded for only six samples.One supply had coliform densities of 15/100 ml in the first month, 38/100 ml in the fourth and 9/100 ml in the twelfth month. Coliforms were not detected in the initial sample or in the three-, five-, or eight-month samples, which suggests that contamination of that particular water drum may have recurred at times during the twelve-month storage period. The generalbacterialquality of stored emergency watersup plies was variable and related directly to the quality of the source waters used. Source-water quality was examined in 62 instances during initial filling of civil defense water drums in 1962. The community water supply in only one city had an ex cessive bacterial count, which ranged from 2 100 to 10000/mL Thirty source waters (48.4 per cent) contained 1 organism/ml, and 24 samples (38.7 per cent) had bacterial counts that ranged from 1 to 16 organisms/ml. A significant decline in the overall quality of water in these stored supplies .occurred during the first year of storage (Table 8). Apparently, the bacterial flora initially present adjusted to this particular water environment and, with changes in dominance of microbial species, reached a stabilized density. Stored-water samples were obtained after five and six years' storage from two different metropolitan areas not included in the earlier investigation. The bacteriological data obtained from these samples generally agreed with the data from the previous study. The results indicate that long-term storage produced a substantial increase in the number of stored water supplies that had less than 1 bacterium/ml. However, occasional samples from five- and six-year-old stored-water supplies had apprecia ble bacterial densities. Examination of the bacterial populations that persisted in stored water after approximately five years revealed that 22.8 per cent of 167 samples contained species of Flavobacterium and 16.2' per cent contained species of Pseudomonas. Flavobac- teria were frequently dominant and counts ranged from 10 to 26000/ml. . Recommendations Bottlad watar. The results from the bottled-water survey and the storage studies support two recommendations for the bot tled-water industry and retailers. First, refrigerated storage will minimize bacterial multiplication in bottled water from the time of bottling to sale. The consumer should also be advised to keep bottled water refrigerated. Second, each container should be marked with the bottling date of lot number or both to assist the retailer and consumer in determining freshness. The lot number is desirable as an aid to the recall of a given lot of bot tled water from retail shelves if the bacteriological quality is found unacceptable. These measures which are designed to dis courage retail overstocking and shorten storage time would minimize increases in bacterial populations in bottled-water products. Bottled drinking water should be analyzed for bacteriological quality at the same frequency per month as is required by the USPHS Drinking Water Standards (Fig. 1) and include a repeat sampling program and a follow-up.sanitary survey when the., results are unsatisfactory.2 At the time of bottling, the water should contain less than one coliform per 100 ml and have an SPC of fewer than 500 bacteria/ml. Samples taken from supermarket, drugstore, or restaurant supplies should have fewer than 1 cpliform/100 ml MARCH 1975 and an SPC of fewer than 1000 organisms/1 mL Establishment of an SPC limit is suggested for control of bacterial quality deterioration during storage. The storage limit of 1 000 bacte ria/ml for bottled water is higher than the 500-bacteria/ml Hmit suggested for treated distribution water1'because bottled waters may normally be stored for alonger periodthan the resi dence time for community finished waters in distribution lines. In addition to this, bottled water usually does not contain residual chlorine to control bacteria and is not subject to other problems associated with distribution lines such as sediment buildup, line breaks, back sip nonage due to low pressure, or Other related problems. Concern over high bacterial counts in bottled-water relates to (1) the possible loss ofcoliform test sensitivity in waters with excessively high bacterial populations and (2) the increased risk of human exposure to organisms that are considered sec ondary pathogenic invaders.15The bottled-water standards adopted by the USFDA, which became effective on May 22, 1974,** do not include an SPC limit for bacteria, although the coliform limit is similar to that of the present USPHS Drinking Water Standards. This is unfortunate because in many cases bottled water is recommended for use in preparing baby for mulas, prescriptions, and fruit juices, coffee, tea, and other beverages. Thus, some infants and many older people who use bottled water are in a position of increased health risk because of possible exposure to bacteria that may be secondary patho gens. Both young and elderly individuals are relatively more susceptible to secondary pathogens than the rest ofthe popula tion. Storad amarganey watar. The requirements for stored emer gency water supplies set aside for use at time of natural dis asters or during war-time devastation are unusual. These stored supplies may be used for drinking water or for medical treat ment of injured personnel. Water used to treat the injured must be of good overall bacteriological quality, and quality sur veillance requires an examination of the water for more than fecal contamination. Radiation exposure may reduce the pa tient's white-cell count, impair body resistance to bacterial in vasion, and thus enable apparently harmless bacteria to become serious pathogenic invaders.20"2* Ensuring the continual availability of a high-quality emer gency water supply over an extended storage period,is difficult Initially a high quality source water must be used and care must be exercised to prevent contamination when filling arid sealing storage containers. Nutrient concentrations in stored water supplies, although low, are sufficient to permit slow growth of some bacteria ini tially present in the source water or introduced by contamina tion during storage. For comparison, the nutrients available in distilled water may be present in trace amounts, yet examina tion of these waters shows that sizable bacterial populations can develop.13_15-19 Because the organisms in stored water sup plies have adjusted to slow growth rates, the need to incubate pour plates for 120 hr (five days) to obtain a reliable bacterial count is re-emphasized. The data available on stored, emergency drinking-water sup plies indicate that water-quality deterioration occurred in a sig nificant number of the filled water drums. Contamination dur ing storage may occur, but it is probable that most bacterial oc currences relate directly to the initial source-water quality. In either case, disinfection is recommended before emergency use, particularly if the water must be used for first-aid treat ment, for drinking, and for bathing purposes by disaster casu alties. In emergencies, disinfection is best accomplished by E. E. GELDREICH ETAL 123 boiling the water for at least 1 min or by using commonly available disinfectants24such ashousehold laundry bleach, swimming-pool disinfectants, tincture of iodine or various trade-name iodine orchlorine tablets/The data inTable 9indi cate the quantity ofsome oftheproducts required torelease 10 mg ofCl2 Or Ij per litre of water. Recommended contact time should be 30min to ensure maximum killrates for various or- : ganisms that might be present Finally, emergency water stockpiles mustreceive periodic in spections for evidence ofwater leakage and evaporation. Occa sional inspection of some supplies revealed empty or partially filled water drums. One recommendation proposes acontinual monitoring program for these supplies, including semiannual bacteriological examination17 ofsamples taken at random from each supply: two, three, or four drums per shelter would beex amined when the stockpile contaius 1-100, 100-1 000, or 1000-5 000 drums, respectively. Acritical needexists toreduce thebacterial population tothe lowest level possible in all emergency supplies. During an an nual or semiannual inspection a systematic effort should be made toreestablish a free chlorine residual, preferably 10 mg/1, in each water drum. Also water drums depleted because of water loss or fouled in dirty or flooded storage areas should be routinely refilled. Such inspections would involve carefully opening and reclosing the plastic water bags to avoid accidental contamination of the water. Summary Although coliforms were infrequently detected in bottled water, the general bacterial population often exceeded 1000 or ganisms/ml.The initial quality ofany bottled water is related, in part, to the quality ofsource water, but the type and effective ness ofdisinfection are much more important. Contamination during bottling or storage may result from poor plant sanitation, reuse ofunclean glass bottles, or improper maintenance ofdis infection or other treatment equipment In spite of the possibilities for contamination, low bacterial densities found in many freshly bottled water samples indicate that good quality bottled water containing fewer than 500 bac teria/ml can be produced. There will generally be some deterioration in bacterial qualityduring storage before sale, but by use of good plant sanitation and refrigerated storage, the bacterial densities in bottled water can be held to fewer than 1000/ml. Thus, the authors recommend aworking limit of500 bacteria/ml for freshly bottled water and apermissible increase during storage to amaximum of 1000 bacteria/ml. In general, the pasteurized bottled water tested was of ex cellent quality and did not deteriorate during storage. Based on data from apilot survey of 25 bottlers,6 ultraviolet irradiation or ozonation orboth, however, most frequently was used todisin fect source water, and the bacterial quality of bottled water treated by either method was highly variable. Bottled waters prepared using ion-exchange columns (or beds) were poorest in quality because of bacterial growth in. the ion-exchange media and infrequent or inadequate maintenance ofthe equip ment This problem iseasily remedied butoften overlooked or ignored until aregulatory agency, such as astate health depart* ment, enforces proper replacement or maintenance ofdeioniz- ing media and equipment Thelack ofsystematic surveillance ofbottled waters isamat ter ofgrowing concern. These sources ofdrinking water should be analyzed at the same monthly sampling frequency as that re quired bythe USPHS Drinking Water Standards for community water supplies, including repeat sampling and afollow-up sani- 124 MANAGEMENT . tarysurvey when results are unsatisfactory. The bacteriological quality ofemergency watersupplies ma* deteriorate during storage and may present a health hazard if such supplies are ever needed. Because the quality of stored water may be difficult to protect and maintain over indefinite storage periods, all emergency stored water supplies should be given supplemental disinfection during annual inspections and before use in an emergency. Referenced I. Hetherinoton, B. O. Bottled Water in California. Jour. Envir. • Health. 29:256 (Nov.-Dec. 1966). I USPHS Drinking Water Standards, Pubn. No. 956. Deptof HEW, Govt Pmtg. Ofce. Washington, D.C (Rev. 1962). 3. Bumstead, J. C; Syrotynski, S.: & McKeefe, F, Owners and Operators of Bottled and Bulk Water Facilities. Public Wtr. Sply. Guide, New YorkState Deptof Health (Jan. 1971). 4. A Pilot Survey of Water Bottlers and Bottled Water. Bottled Wtr. Rprt in Bottled Water Study. Wtr. Sply. Div„ USEPA (Sep. 1972). 5. Standard Methods for the Examination of Water and Wastewater. APHA, AWWA, WPCF, Washington, D.C (13th ed„ 1971). 6. Levin, M. A. & Cabelu, V. J. Membrane Filter Technique for Enumeration of Pseudomonas aeruginosa. AppL MknbhL 24:6:861 (Dec. 1972). • . . 7.0BST, M. M. Bacteria in Commerical Bottled Waters. Jour. Bact. 1:103 (Jan. 1916). . 8. Wallis. C; Stagg, C H.; &Melnkk, J. L,The Hazards of Incor porating Charcoal Filters into Domestic Water Systems. Wtr. Rex. 8:111 (Feb. 1974). '••,,,,„•.-. 9. ZoBell, CE.&. Ford, G. C A Sanitary Survey of 212 Bottled Drinking Water Dispensers. California State DeptPublic Health Weekly Bull.. 15:141 (Oct 3, 1936). 10. Console, O. M. A Study of Sanitary Conditions in Bottled Water Cooling Dispensers. Jour. Envir. Health. 26:326 (Mar.-Apr. 1963). II. Geldreich, E E &Clarke N.A. Bacterial Pollution Indicators in the Intestinal Tract of Freshwater Fish. AppL MiaobtoL, 14:429 (May 1966). ^ II Geldreich, E E et al. The Bacteriological Aspects ofStormwater Pollution. Jour. WPCF. 40:1861 (Nov. 1968). 13. Leifson, E The Bacterial Flora of Distilled and Stored Water. Part 1: General Observations, Techniques and Ecology. Inu, Bufl. Bac terial Nomenclature andTaxonomy, 12:133 (1962). 14. Leifson, E The Bacterial Flora of Distilled and Stored Water. Part 2: Caulobacter vibrioldes (Henrici and Johnson 1935) in Distilled Water. Intl.'Bull. Bacterial Nomenclature and Taxonomy, 12:155 (1962) 15. Leifson, E TheBacterial Flora of Distilled and Stored Water. Part 3: New Species of the Genera Corynebacterium, Flavobocterium, Spirillum, and Pseudomonas. Intl. Bull. Bacterial Nomenclature and Taxonomy, 12:161(1962). . f • 16. Geldreich, E E &Clark, H, F. Distilled Water Suitability for Microbiological Applications. Jour. Milk and Food TechnbL, 28:351 (Nov. 1965). ^' " J ' 17. Taylor, F. B. Shelter I. USPHS-Office of Civil Defense Study of Storage ofDrinking Water inFallout Shelters. Dept ofHEW, Govt Pmtg. Ofce., 1-220. Washington, D. C (Nov. 1963). 18. Geldreich, E E et al. The Necessity of Controlling Bactenal Populations in Potable Waters: Community Water Supply. Jour. AWWA. 64:596 (Sep. 1972). 19. Bottled Water Quality Standards: Addition of Fluoride, and Current Good Manufacturing Practice Regulations. Dept of HEW, Fed. Rgtr. Pt III, 38:226:32558 (Nov. 26,1973). 20. SMrm, W. W. Survival after Radiation Exposure-Influence of a Disturbed Environment Nucleonics. 10:80 (Nov. 1952). 21. Gonshery, L.; Marston. R.Q.; &Smith, W. W. Naturally Occur ring Infections inUntreated and Streptomycin-Treated X-Irradi- ated Mice. Amer. Jour. PhysioL 172:359 (Feb. 1953). .21 Wensinck, F.; Van Bekkum, D. W.; &Renaud, H; ThePrevenuon of Pseudomonas aeruginosa Infections in Irradiated Mice andRats. Roan. Res..7:491 (Nov. 1957). . 23 KovaL, B. Physicians (and Engineers) Save Radiation Vtcums: Chem. Engrg.. 64:245 (Apr. 21 1968). 24 SafeDrinking Water in Emergencies. Deptof HEW, USPHS Pubn. No. 387, Govt. Pmtg. Ofce. Washington, D. C (Rev. 1964). JOURNAL AWWA AWWA Organisms in Water Committee Because ofthe importance ofthe microbiological qualityofwater, theAWWA Organisms in Water Committee elected to review the various options and approaches that had been proposed in connection with revised drinking water regulations. This report addresses these proposals, as well as current practice, and summarizes the committee's concerns and recommendations. As required by the Safe Drinking Water Act, the US Environmental Pro tectionAgency(USEPA)is involved in a comprehensivereassessment of federal drinking water regulations.1 A signifi cant part of this reassessment deals with microbial degradation of source water quality, waterborne disease out breaks (Giardia, Yersinia. Campylobacter, Legionella, and viral agents), adequacy of the coliform indicator, the impact of alternative treatment processes, mea surement of microbial quality, monitor ing of public water supplies, and the approach to publicnotification. Manyof theproposed optionsandnewapproaches relating to these issues were addressed in the NATO/CCMS Drinking Water Pilot Project,?the USEPA Microbiology Workshop,3 and the AWWA Research Foundation-USEPA Workshops on the Revised.Primary Drinking Water Regu lations.4 Because of the importance of these issues, the AWWA Organisms in Water Committee elected to review these proposals and existing practices and to present a synthesis of concerns and recommendations. Treating and protecting watersupplies Increasing urban populations often place a burdenon water suppliers, both in terms of the quantity demanded and in terms of contamination potential through the use of less desirable raw waters derived from polluted rivers, lakes, and groundwater. Wastewater treatment plants, urban and rural runoff, feedlots, and a host of other activities often discharge to a watercourse that may be the source for a public water supply.Eventhe mostpristine watershed is potentially contaminated with or ganisms pathogenic to humans. Surface waters. Surface water is at risk from microbiological contamination. Giardia lambda, for example, has been recovered from isolated lakes and streams where there is little human activity. This pathogenic protozoan is less susceptible to conventional disin- MAY1987 feetion practices than are coliforms and other bacteria. Further, particulate matter can physically protect even sus ceptible pathogenicbacteria and viruses from the disinfecting agent. The com bined effect of these two factors has resulted in outbreaks of waterborne disease involving surface supplies, even when disinfection produces water meet ing the coliform standard. The AWWA committee agreed that a dual barrier is desirable to provide a consistently high quality water supply. Minimal treatment of surface waters from restricted watersheds should con sist of coagulation and rapid sand fil tration to entrap cysts and turbidity, followed by disinfection to inactivate any remaining cysts, coliforms, patho genicbacteria,and viruses.Slowsand or diatomaceous earth filtration, in con junction with disinfection, is a viable treatment alternative, which would in addition eliminate the need for chemical pretreatment. Exceptions to these treat ment requirements would be based on a past record ofnogiardiasiscausedbythe public watersupply,periodic monitoring of treated water for Giardia, a watershed management program to minimize the potential for contamination by enteric- viruses and Giardia cysts in the source water, demonstration that the water characteristics will not interfere with effective disinfection, and alternatives to filtration that will meet the require ment of a thousandfold reduction in microbial contamination. Wastes must be treated properly to minimize pathogen releases. Steady progress is being made to bring on-line more waste treatment facilities nation wide. Most important, treated waste waters should be disinfected, unless it can be demonstrated on a case-by-case basis that there will be no adverse environmental effects. Variance from this requirement should depend on considerations of time and distance, water temperature, the microbiological nature of the discharge (e.g., pathogenic bacteria,, viruses, or parasitic proto zoans), and the efficiency of the available water treatment barriers. Several factors justify the disinfection of wastewater effluents. First is the philosophy of maximizing public health protection through a multiple barrier concept. Disinfection of treated waste water provides the initial barrier to the transmission of waterborne disease. Removal of this barrier simply transfers an additional burden to the potable water purveyor. The second concern is that wastewater treatment is often not precise and reliable, with some municipal plants being unable to meet permit limitations, thus making disinfection less effective. Efficient disinfection of the treatment planteffluentensures controlled minimal protection for. downstream water users and reduces microbiological contamina tion during periods when treatment processes are not operating properly. Where feasible, chlorinated sewage ef fluent should beretained onthe premises until the chlorine residual has dissipated or, upon release of the effluent, is not detectable in the discharge for more than 330 ft (100 m) downstream. Tri- halomethanes formed during chlorina- tion of wastewater may be significantly dissipated in the receiving stream pro vided there is adequate dilution or volatilization. When there is concern over the effect ofchlorine-based disinfectants on aquatic life, alternative processes or disinfec tants, e.g., dechlorination, ozone, or ultraviolet light, can protect both the native aquatic biota and the downstream water supplies. The AWWA committee believes that the overriding concern is public health protection and that the water purveyor should not be forced to demonstrate a public, health threat to make upstream disinfection of waste water mandatory. Surface waters receiving discharges from municipal wastewater treatment plants and industrial and agricultural activities require more complete treat ment, including coagulation, floccula- tion, sedimentation, filtration, and disinfection to minimize organics and fluctuating densities of microbial con-, taminants. Such serial treatment pro vides the multiple barriers needed to COMMITTEE REPORT 81 ensure the microbiological quality ofthe water. In the case of source waters that need special treatment to reduce sus pected carcinogens, the useofaeration, ozonation,activated carbon adsorption, oranother effectiveprocessbeforedisin fection may be desirable. However, microbial activity doesoccur ingranular activated carbon (GAC) processes, pos sibly resulting in regrowth related to biodegradable organics, especially in warm-waterperiods. Inaddition,carbon fines may transport bacteria in GAC product water during stabilization of a . new bed of virgin GAC or replacement with reactivated GAC. Neither problem isbeyondcontrolifreasonabletreatment precautions are taken. Groundwaters. Groundwaters derived from deep aquifers aregenerally ofgood bacteriological qualitybecause percola tion ofwater through soil results in the removal ofmuch microbialpollution. As Given the widespread potential for groundwater contamination, it is pru dent to assume that, unless otherwise demonstrated,groundwatersuppliesalso contain microbiological contaminants . and need to be treated. Specifically, all treatment plants forpublicgroundwater supplies should incorporate disinfection or at least have disinfection capability An allowable variance to this require ment would be based on increased bacteriological monitoring, norecord of anoutbreakwith the current operational configuration, andannual sanitary sur veys todemonstrate continued integrity of the system. .. Given the widespread potential for groundwater contamination, it is pru dent to assume that, unless otherwise demonstrated,groundwater suppliesalso contain microbiological contaminants and need to be treated. Specifically, all treatmentplantsforpublicgroundwater supplies should incorporate disinfection or at least have disinfection capability. Variance to this requirement would be basedon increased bacteriologicalmoni toring, norecord ofanoutbreak withthe current operational configuration, and annual sanitary surveys to demonstrate continued integrity of the system. Finished waters. For either surface water.or groundwater, the ideal disin fectant should becapableofmaintaining a residual throughout the distribution system to control microbiological re- growth and protect against other con tamination events such as cross-con nection orlowlinepressure. Disinfection has been one of the most effective public health practices of modern times; this fact must not be lost >n the current controversy surrounding chlorination by-products and carcinogenesis. Whatever raw water quality problems arecorrected byadequatetreatment, all is for naughtunlessproper distribution system water protection practices are 82 RESEARCH AND TECHNOLOGY followed. Maintaining a disinfectant residual in the distribution system is critical but it is just one step in proper systemmanagement. Toensuredelivery ofa high quality water toeachconsumer,. managersofpublicwater supply systems must be vigilant for any contamination in the distribution network and for evidence of water.quality degradation. This is made difficult by the complexity of a distribution system, which is a network of mains, fire hydrants, valves, auxiliary pumping or chlorination Sub stations, storage reservoirs, standpipes, and service lines. Properly constructed storage reser voirs are needed to ensure that inad vertent, or even intentional, contam ination does not occur. The well-known problems associated with animals and children around finished water reser voirs have been increased by acts of vandalism. Today,morethan ever before, all finished water storage reservoirs should be covered, with all access ports locked and all vents screened. In addition, there should be an ade quate program for flushing the distri bution system, reservoir cleaning, and disinfecting all new and repaired water mains, and a viable cross-connection controlprogram. This preventive main tenance program is critical to the con tinuedprotection ofwaterduringdistri bution to the consumer. Because cross- connections may bewidespread, it is not only desirable to identify them at the water treatment plant (short-circuiting of process waters) and throughout the distribution system but alsotoeliminate them through repairs and backflow preventer devices. High risk locations, e.g., hospitals, mortuaries, car washes, and wastewater treatment plants, should be priorities. It would, of course, be prudent to select a nonpolluted raw water source. Such sources are rare, if they exist at all, but the best available source should be used. A safe water supply, therefore, requires proper treatment, effective operation, andsafepassage through the distribution system to the consumer's faucet. Types of pathogens Specialattention is beingfocused on viralagents, Giardia lamblia, Legionella, andopportunistic pathogens, whichare amongthe newlyidentified waterborne pathogens. Consideration hasbeen given to including these organisms in the Primary Drinking Water Regulations, but no recommendation to do so is now being made except for recommendations pertaining to additional treatment pro cesses to control pathogen occurrences. Viruses.Viral agents may beimportant because of the growing reuse of water, particularly when reuse is associated with decreased wastewater disinfection. Numerous publicsurface water supplies are neither filtered nor disinfected. Some water utilities (small groundwater sup plies in Illinois, for example) are even. reducing chlorination of potable water. Still other communities are shifting to combined chlorine disinfection toachieve trihalomethane reduction. Changed dis infection practices and other treatment modifications should be evaluated as microbial barriers wherever the risk of human viral contamination in source waters is real. It would also be desirable for the state authority to conduct pro spective epidemiological surveysof the incidence of viral disease in such com munities to identify special risks. Addi tional research is needed on virus meth odology. The usefulness of a surrogate such as the coliphage test is unproven at this time. Giardia. Detection of Giardia in source water, treatment processes,and finished drinking water is labor intensive, is variable in recovery efficiency, and requires a special professional skillthat maynotyetbeavailable tomanyutilities. No appropriate surrogate indicator has been found. Because of this problem of analytical methodology, it is more ap propriate to use the bestavailable treat ment technology tocontrolinfection and to make periodic prospective epidemic- . logical surveys.Although properly oper ated treatment schemes, includingchem ical additional, filtration, and chlorina tion with a free residual, are capable of significant Giardia reductions, water suppliersarealsoadvised touseturbidity monitoring for filtration performance evaluation. If continuous turbidity moni toring is used,the turbidityvalueat 4-h intervals (or a shorter regular time interval) is critical fordetermining com pliance with performance criteria. For systems usingslowsandordiatomaceous earth filtration•, the sampling frequency for turbidity could be limited to. one sample per day. It must beremembered, however, that breakthrough in filter beds is greater in smallplants and that Giardiaoutbreaks have occurred in small systems using improper filtration. Fur thermore, monitoring of filtered water for Giardia cysts (where washwater is recovered) could lead to a decision not to recover backwash water in order to prevent cyst contamination of the pro cessed water. Legionella and otheropportunistic patho gens. Anxiety about Legionella may represent only the tip of the iceberg as far as unknownpathogens areconcerned. Classic water treatment practice has focused on reducing enteric pathogens. The coliform group has been used as an indicator of either source water accept ability or disinfectionefficiency, because if coliforms are present, enteric patho gens may also bepresent and may have survived treatment. Such is not the case JOURNAL AWWA with all pathogens. Legionella and Le- gionella-Wke organisms are pathogens that are transmitted through aerosol- ization and. inhalation rather than by ingestion. Other pathogens with poten tial for transmission through.inhalation of drinking water aerosols are Myco bacterium, Pseudomonas, and Klebsiella. Microbial allergens (a cause of allergic- type reactions from industrial air filters used in humidification.devices) may also represent a problem. The committee feels that if the water treatment plant operator has responsibility for minimiz ing microbial dissemination via the aerosol route, then acceptable indicator systems and required levelsoftreatment to be applied at sites of high risk need to bedefined. Because few data are available to define the infectious dose of such respiratory pathogens, a rational water quality criterion needs to be developed. When they come into body contact, opportunistic pathogens also are a con cern, especially for hospital patients. Waterborneorganisms that have caused secondary infections in hospitals include Pseudomonas putida, P. multophila, P. aeruginosa, Acinetobacter cakoaceticiis, Alcaligenesfaecalis, and Flavobacierium species. The problem maybecomplicated further if opportunistic pathogens be come antibiotic resistant through ac quisition of plasmids from resistant bacteria, thereby making patient treat ment more difficult. Drinking water treatment is intended to produce water that is free from pathogenic microorganisms but not sterile. Bacterial regrowth in the distri bution system may not beofsignificance for public health but may be of concern .in hospitals or kidney dialysis clinics. Thus, it may be necessary for hospitals to consider installation of booster chlo- rinators.to restore lost disinfectant residuals for improved reduction of all heterotrophic organisms, including op portunistic bacteria, and to maintain an in-house flushing program to prevent colonization. Particular attention should begiven to routine monitoring for disin fectant residuals in the building plumb- . ing network and to periodic flushing of all reservoirs and hot water tanks to remove sediment accumulations that protect bacterial survivors. Hospital personnel need to establish effective monitoring programs for water supplies and to plan corrective action. Although water quality deterioration beyond the service lines is not a utility responsibility, better communication is desirable be tween water purveyors and special user groups, e.gj, hospitals and food process ing and pharmaceutical plants. The AWWA committee suggested that pe riodic meetings address the utility's wa ter quality objectives, review facility plumbing flushing and corrosion control programs, and identify specialized treat- MAY 1987 ment approaches to be implemented by user groups to minimize amplification of organisms(Legionella, opportunistic bac teria, and fungi) associated with water inhalation and body contact infections. Conventional criteria of treatment effectiveness Total coliforms. The total coliform group of bacteria remains the best avail able indicator of treatment effectiveness. Although there is substantial evidence that coliforms are not an adequate indicator of Giardia occurrences and there are documented instances in which disease outbreaks occurred in the ab sence of any indication of a problem by coliform results, it is likely that similar situations would occur regardless of the indicator used. No indicator is absolute in its ability to predict the presence or absence of pathogens or to predict the occurrence of a waterborne outbreak. The most effective preventative measure would be to expand treatment barriers. Testing larger sample volumes of plant effluents and reemphasizing use of the sanitary engineering survey'would be valuable adjuncts. Heterotrophic bacteria (standard plate count). The heterotrophic plate count (HPG) is a good operational tool for measuring microbial breakthrough, eval uating process modifications, and de tecting loss of water main integrity. This general measure of bacterial water quality can also be an important early sign of excessive microbial growth on distribution system pipe walls and in sediments, and it can indicate the pres ence of opportunistic pathogens. It is recommended that baseline data be gatheredon the density of heterotrophic bacteria in the distribution system and that the information be utilized to measure treatment train effectiveness and bacterial quality changes. The com mittee recommends that a minimum of two samples or 10percent ofthe coliform samples, whichever is greater, be exam ined quarterly, including one sample from the plant effluent and one from storage facilities in the distribution system. The density of heterotrophic bacteria in the plant effluent can easily be maintained at very low levels (<10 organisms/mL) with adequate, disinfec tion. When normal background hetero trophic plate counts in the distribution system become greater than 103 organ- isms/mL (a hundredfold increase) and this is confirmed by a second sample, action should be initiated immediately to resolve the microbial regrowth prob lem. One committee member disagreed with the recommendation to establish a limit for the heterotrophic bacterial population, claiming that colonization exists in the distribution system at all times and cannot be prevented. Turbidity. Particles in water protect bacteria and promote theirgrowth in the distribution system. For this reason, turbidity in finished water, should be minimized. Well operated treatment plants consistently achieve a finished water turbidity <1 ritu. Where Giardia cysts or asbestos fibers may be present in the raw water, a 0.2-ritu standard may be necessary (dependingon the treatment applied) to ensure minimal risk. Good quality groundwater supplies can be expected to have a turbidity of <0.5 ntu when they enter the distribution net work. Exceptions for cases of turbidity >1 ntu may be made as long as the turbidity does not interfere with coliform detection or contribute to disinfectant demand in the finished water. Turbidity monitoring in the distribution system, although not required by regulation, is a good quality control practice; the goal should be 2 ntu. Values >5 ntu would signal the need to flush the distribution system and to search for areas of pipe corrosion that must be brought under control. Disinfectant residuals. Measuring resid ual disinfectant is undoubtedly the most important test used in water supply monitoring. Emphasis should be on observing and interpreting sudden changes in the disinfectant residual of finished water and water in the distri bution system. Disinfectant measure ments should be used as a routine diagnostic tool. The practice of adding ammonia to convert free chlorine to combined chlorine to minimize trihalo- methane formation will increase C • T disinfectant values. Several very large water systems, however, have used chloramines for years with no evidence of quality deterioration or public health problems. Recent case histories docu mentingcoliform colonies indistribution networks have renewed interest among water suppliers in maintaining disin fectant residuals to the ends of the system, including dead-end sections, in the attempt to reduce microbial growth in such places. Measuring microbial quality The bateriological quality of water was initially measured by the multiple tube or most probable number (MPN) method. Regulations defined an unsatis factory sample as one producing gas positive results in more than three tubes of a five-tube test. A water supply was out of compliance when >10 percent of the tubes in all MPN tests performed on distribution system samples over a 30- day period were positive. With the introduction of the membrane filter (MF) procedure, quantitative density limits, were recognized: An unsatisfactory sample was defined as one containing more than 4 coliforms/100 mL, and a water supply was out of compliance COMMITTEE REPORT 83 when the monthly average of all MF tests exceeded 1 coIiform/100 mL. Since the mid-1960s, many laboratories have replaced the MPN procedure with the less labor-intensive MF technique for total coliforms. With more than 50 percentofall laboratories now using the MFprocedure,there is greater emphasis on the quantitative aspects of coliform occurrences. To equalize the sample volumes for testing by either the MF or MPNprocedure,the sixteenth edition of Standard Methods has introduced a 10- tube multiple tube test and includes a . table so that expanded MPN values can be calculated. This expanded multiple tube test extends quantitative results from a maximum value of 16/100 mL to 23/100 mL and improves the 95 percent confidence range. The committee sug gests using the expanded multiple tube test as aquantitative measurement that is morecomparable to the MF technique inprecisionand samplevolume(100mL) requirement. Observation of coliform bacteria has received recent attention, particularly in regard to resolving problems of infre quent and minimal sampling of small water supplies(serving 25-3300 people). Regulating the percentage occurrences for any.density of coliforms found by either testing procedureover 12months, rather than limiting coliform density to a 30-day period, needs to. be fully eval uated in different geographical areas of the United States. Frequency of occur rence is measured by a presence-absence test, the result of which is only qualita tive. A new presence^absence test me dium may improve recovery of stressed coliform bacteria and could be designed for use in a delayed incubation procedure after samples have been transmitted via mail to a central laboratory. Statistical interpretation of the results would be based on a 12-month moving average; the watersupply would be in compliance if <5 percent of all presence-absence : tests were positive. The committee believes that before the frequency of occurrence concept is adopted, more water supply data from different geo graphical areas and different water treatment processes should be analyzed to determine thecompliance equivalency offrequency ofoccurrence with present regulation requirements. This could be done by simply recalculating MPN or MF "presence" results for any coliform occurrences and "absences" from all negative tests. Monitoring publicwater supplies Monitoring the quality of public water supplies is essential. Water plant oper ators and microbiologists.agree that there continues to be a need for operational monitoring methods that will rapidly characterize product water; Unfortu nately, rapid methods are unavailable, 84 RESEARCH AND TECHNOLOGY or they fail to defect coliforms at the concentration of 1 organism/100 mL. Finished waters. Although large water systems monitor the plant effluent for coiiforms, turbidity, and chlorine resid ual several times a day, smallersystems may make such analysesonce every 24 h or only several times a week. The re- q jired frequency of coliform analysis for y/ater plant effluents (population served, . J;0'000 or more people) should be at least me sample per day, collected at different times within the operational period. A more desirable approach might be com posite sampling of the plant effluent over a 24-h period, with analysis of 0.25-1.25-gal (1-5-L) portions by the MF procedure. The purpose of this would be the identification of low density coliform breakthroughs. The frequency of moni toring plant effluents from small water systems would be determined by the state authority and would be intensified if the system had a water quality problem. The turbidity of the finished water produced from surface water sources should be measured continuously or once per operating shift. The turbidity value reported for maximum contaminant levels would be the average value per day. For small systems serving <1000 persons, a grab sample,' preferably taken toward the end of the filter run or once per shift, would be adequate. Aside from the mandatory requirement to measure finished water turbidity, it is desirable to measure turbidity at all stages of the treatment train to.determine process effectiveness. Distributionwater. Water quality moni toring of the distribution system should continue to be based on population served, but consideration should also.be given to the length of the distribution system and to any other mitigating circumstances that might dictate addi tional sampling. Stringent quality con trol and sample collection guidelines need to be developed and imposed. For small watersystems, minimum sampling should be increased to five samples per month, with state discretion permitted to reduce the number on the basis of knowledge and experience. The current requirement of only one sample per month (or only one sample per quarter) does not provide adequate information if there is a water quality problem. The highest priority, however, should be placed on providing good water quality, not just on increased monitoring. Samples must be taken at representa tive sites in the system^ A portion of these sites should be. fixed locations that are selected with reference to pressure zones, potential sources of contamina tion; high risk areas (hospitals, clinics), and the past history of coliform occur rences. Other sampling locations should be varied so that all parts of the distri bution system are sampled. Routine monitoring should include, measuring coliforms, turbidity, and, with leaser frequency, heterotrophic bacteria. Turbidity samplingof the distribution system should be done at 10 percent of the sites used, in the coliform monitoring program, the turbidity value being in dexed to corrosion, chlorine demand, bacteriological regrowth, and the need to flush the distribution lines. With respect to monitoring the heterotrophic bacterial population, a data base must be estab lished for all seasons and for all sites where samples are collected for coliform analyses; then testing can be reduced to 10 percent of the sampling frequency recommended for total coliform deter minations. During periods of observed microbial regrowth, sampling for coli form and heterotrophic bacteria should be intensified until appropriate measures are found to.be effective in eliminating the colonization. Public notification The National Primary DrinkingWater Regulations require public notification, through announcements included with water bills or through other appropriate means, whenever a violation occurs. This requirement is applied whether the violation is minor or major, isolated or persistent. Maximum contaminant level (MCL) violations in^addition require publication of the notice in newspapers and provision of the notice to the elec tronic media. \ Public notification relates to the bac teriological MCL in two ways: (1) public notice is required of systems that fail to monitor as prescribed, and (2) public notice is required of systems whose samples fail to meet the bacteriological MCL. The net effect of this requirement from the consumer's point of view is that all violations aregiven equal weight, when, in fact, only MCL violations imply any potential health risk. Thus, the current requirement appears to be coun terproductive. Whatever benefits are gained by public awareness of a water system's lack of samplingare more than offset by either public, perception that such failure constitutes a serious health hazard or, worse, public.belief that notices by water purveyors are routine and should not be considered significant. In either case, such notice does not achieve the principal desired effect, namely, community awareness of a potential health hazard. Public notice should be used with discretion to ensure that legitimate threats to.public health receive the required visibility and attention. There are two levels of concern and each requires a different level of action: (1) utility accountability to the public (Continued on page 88) JOURNALAWWA (Continued from page84) served, and (2)the prompt notification of a potential health problem. At the first level, any^adverse trend inwater quality should beTeviewed by utility, state,and federal water supply representatives and state and local health officials. A joint action plan would then be formulated that should result in the intensification of data gathering (including monitoring for any fecal coliform occurrences), a sanitary (engineering) survey of the sys tem, and daily review of the results with appropriate corrective measures. The second level of action would be immediate public notification if an outbreak occurs, if fecal contamination is detected in the water, or if the utility fails to follow corrective strategy defined by joint agreement of designated authorities. . In short, the AWWA committee con cluded that public notices should be applied only when some risk to com munity health is definable. They should not be used as a regulatory punishment for minor or nonhealthrelated viola tions. The key question that needs to be addressed to protect public health through safe water supplies is not whether the regulations have been vio lated, but rather the cause and signif icance of these violations in terms of public health. References 1. US Environmental Protection Agency. National Interim Primary Drinking Wa ter Regulations. EPA-570/9-76/003. Office of Water Supply. Washington. D.C. (1976). 2. Committee on the Challenges of Modern Society. Drinking Water Pilot Study Summary. NATO/CCMS Drinking Wa ter Pilot Project Series CCMS 130. EPA 570/932007. Office of Drinking Water. USEPA. Washington. D.C. (1983). 3. Assessment of Microbiology-and Tur bidityStandards forDrinking Water(P.S. Berger and Y.Argaman, editors). Proc. of a Workshop. December 2-4. 1981. EPA- 570/9-83001. Office of Drinking Water. USEPA.Washington. D.C (1983). 4. AWWA Research Foundation-USEPA Workshopson Revising the Regulations. Philadelphia. Pa.. Sept. 21-23. 19S3; St. Louis. Mo.. Oct. 4-6. 19S3; Reno. New. Oct. 31-Nov. 4.19S3: Winter Park. Fla.. Nov. 27-30.19S3. About the authors: This report was prepared by Ihe A \V\VA Organisms in Water Commillcc. Commillcc, members were E.E. Geldreich (chairman). A.E. Grccnbcrg, .C.N.Haas, J.C. Hoff R.J. Karim.J. Marlin, E. Means. R.H. Moscr, P. Rcgunathan, K. Reich, and H. Viclorccn. Monitoring for Indicator Bacteria in Small Water Systems* Edwin E* Geldreich, Eugene W. Rice and Eleanor J. Read** The problem of monitoring for water quality in small or rural water supplies is a great concern because these are the public water supplies that experience most of the waterborne disease outbreaks. The reasons for the unsatisfactory state of public health may be found either in the use of poor quality raw source waters that are not ade quately treated or in the ineffectual operation of existing plant processes. Major waterborne agents contaminating water supplies include pathogenic bacteria, virus, protozoa and blood flukes,.any of which cause a variety of gastroentestinal illnesses. These organisms origi nate in the feces discharged from infected humans, animal pets, farm animals and wild life and are transported by sewage and stormwater to the receiving waters that may be used as a raw source water in water supply. Engineered process barriers are designed to intervene in the passage of waterborne risks through public water supply. How effective these barriers are in prevention of pathogen passage in drinking water is the key purpose of the monitoring program. While monitoring for pathogens in.water is desirable, it is not practical for many reasons. First, the number of waterborne pathogens identified in water is estimated to be several hundred or more when consideration is given to the numerous species of Salmonella and types of viral agents. This list of pathogens has expanded in recent years with the discovery of new bacterial and viral agents that include Yersinia, Campylobacter, rotavirus, reovirus, parvovirus and the proto- zoansj Giardia and Cryptosporidium. Other pathogens will be discovered in the future as hew breakthroughs in methodology unravel the mysteries of unidentified agents associated with waterborne outbreaks; In the United States alone (Table 1), unidentified etiologic agents accounted for-46.5% of all outbreaks during 1961-1983 and caused 86,740 individ ual illness cases (Craun, 1985; Li.ppy and Waltrip, 1984). The problem of pathogen monitoring is further complicated by the. lack of a single test that will detect all bacterial pathogens, viral *Presented at the International Conference :on Resource Mobilization for Drinking Water Supply.and Sanitation in Developing Nations, San Juan, Puerto Rico, May 26-29, 1987. **EPA Senior Microbiologist; Research Microbiologist; Water Engineering Research Laboratory and Senior Statistician, Computer Sciences Corp oration, Cincinnati, Ohio, respectively. Geldreich agents or pathogenic protozoans and the knowledge that a negative pathogen test is inconclusive, using current state-of-the-art tech niques. Monitoring sewage for pathogens currently prevalent in the community is not completely adequate because there will be no input from pathogen shedders in the infected farm animal and wildlife popula tion. . While some methods are available for specific pathogens (Salmo- nella, Yersinia, Campylobacter and enterovirus) these tests are best done in the specilaized laboratory, not by the average technician using the limited resources available in the small laboratory. Cost becomes^ the other limitation on direct monitoring for pathogens. Estimates for bacterial pathogen examination in water samples can range from $20 to $50 per sample while a virus or Giardia examination can cost $100 or more. Obviously, these costs can not be reconciled to small water system operating budgets. Therefore, the only viable alternative is to select a surrogate organism or bacterial indicator of fecal contami nation that can be detected in a simple, inexpensive laboratory test. Table 1. Waterborne Outbreaks in the United States During 1961-1983* Etiologic Agents Outbreaks Cases Deaths Bacterial Shigella Salmonella Campylobacter Toxigenic E. coli Vibrio . Yersinia Viral Hepatitis A Norwalk Rotavirus Protozoan Giardia 84 22,897 0- Entamoeba 3 39 2 Chemical Inorganic (metals, nitrate) 29 891 0 Organic (pesticides, herbicides) 21 2,725 7 Unidentified agents 266 86,740 0 52 7,462 6 37 19,286 3 5 4,773 0 5 1,188 4 1 17 0 1 16 0 51 1,626 1 16 3,973 0 1 . 1,761 0 TOTAL 572 153,394 23 *Data adapted from Craun (1985); Lippy and Waltrip (.1984). Geldreich . The perfect surrogate would be one that completely mimics the occurrence and survival pattern of all fecal pathogens thatmig.ht be waterborne and.traceable through inadequate treatment barriers. Unfor tunately,, such an organism or indicator system does not exist. What is available are several indicator systems (total coliform, fecal coliform and fecal, streptococcus) that are found in all warm-blooded animal feces with densities ranging from 106 .to 108 organisms per gram (Kowal, 1982; Geldreich, 1978). Historical studies of these candidates has placed general emphasis on the wide spectrum of intestinal organisms • that comprise the total coliform population. Their detection in water supply would suggest fecal contamination has occurred and.this contami nation very likely includes some intestinal pathogen.risk to water con sumers. Furthermore, adequate treatment will, remove ail. coliforms detectable in 100 mL, including those that are of fecal origin as well as the more ubiquitous environmental strains. While this assumption has generally been accurate, the concept is occasionally flawed by virus or protozoan occurrences not associated with detection of coli forms in 100 mL samples. However, an overwhelming data base from systems world-wide has supported the general assumption that coliform absence in public water supply indicates there is reasonable assurance . that the water is safe and carries minimal public health hazard from pathogens (World Health Organization 1984). Basic Principals for Coliform Detection Total coliform detection in water supply is not complicated nor . does it require a large amount of specialized equipment or a profes sional microbiologist to perform the test. There are two approaches: . measurement of coliform density in a 50 or 100 mL test portion or the simple determination of the presence or absence of coliforms in 100 mL sample volume (Figure 1). For quantitative measurements, many labora tories use a membrane filter (MF) technique which involves the culti vation of differentiated coliform colonies on the. filter surface using a lactose type medium and incubation at 3.5°C for 24 hours before colony counts are established (American Public Health Association, 1985). The multiple tube fermentation test (FT) has been in use by some labora tories for many years as the alternative method for detecting coliforms (American Public Health Association, 1985). This procedure is based on detecting gas produced by coliforms growing in a medium containing lactose. Five tubes of the sterile medium are inoculated with 10 mL sample portions and then the set of tubes is incubated for 24-48 hrs at 35°C to permit any coliforms.present to ferment the lactose with re lease of gas into the medium and entrappment in a fermentation vial. If growth and gas are produced, those individual positive cultures are verified by growth transfers into a second, more restrictive lactose medium for further evidence; of gas production. This confirmation pro vides assurance the sample contained coliform bacteria, not some other bacteria causing a false positive reaction. The density of coliform bacteria per 100 mL is then estimated from the number of tubes positive by using a most probable number statistical, table. The FT procedure requires at least 48 to 96 hours to. determine a test result and.is more labor intensive than the MF procedure. Geldreich A more simplified test (Figure 1) and one that can be a very attractive operational test for small water.plant operators is the presence-absence (P~A) test (American Public Health Association, 1985). While this procedure appears to be somewhat similar to a multiple tube fermentation test, the involvement in laboratory work is much less complicated. Basically, the procedure consists of inoculating 100 mL of sample into a bottle containing the appropriate concentration of a lactose type medium and a fermentation tube for gas entrappment. The bottle, with 100 raL sample, is incubated for 24 to 48 hours at 35°C and " inspected for growth and gas production. If gas is noted in the fermen tation tube or a color change (acid reaction) is observed, a small inoculumn of the culture is transferred to.a tube of brilliant green lactose broth for verification that gas production again occurs and was related to coliform occurrence. Results of this test are completed within 96 hrs and reported as coliform present or absent. The only equipment needed is a 35°C incubator, sterile sample containers, bot tles of P-A medium and a supply of sterile brilliant green lactose broth in culture tubes containing fermentation vials for gas entrap pment. Media could be prepared and sterilized in a small autoclave at the water plant or obtained prepared for use from a central laboratory. If the P-A culture bottles are carefully premarked at 150 mL capacity (100. mL for sample plus 50 mL for medium with adjustment made for fermentation tube displacement in liquid) the samples could be added directly into the P-A bottle at the site of sample collection. With this accomplished, sterile sample containers would not be a necessary item in the test. Any chlorine residual in the water would be immedi ately, neutralized by the medium constituents. P-A Concept vs P-A Test It is important to separate two different aspects associated with presence-absence information. The P-A concept is concerned with the frequency of coliform occurrence in a water supply .over a specified time span. Such data can be obtained from conventional bacteriological tests using either the membrane filter or multiple tube procedure, simply by translating any coliform count or positive tube results into a coliform occurrence. This concept places equal emphasis on all posi tive samples, regardless of density, with a limit defined by a specific percentage of positive coliform occurrences permitted!. For example, the presence-absence record of compliance could be based on degree of treatment: ;5% positive occurrences permitted if the system uses com-. plete treatment, 3% if treatment consists only of filtration and disin fection of surface waters and 1% for water supplies using disinfection as the only treatment control barriers. For the public health author ity, information on how often there are coliform occurrences, over the long term, is an important indication of treatment effectiveness and operator skill in providing a continuous supply of safe.drinking water. From the operators viewpoint, it is desirable to have the capabil ity for a simple bacteriological test which provides frequent checks on water quality produced. Before acceptance of the P-A test, two ques tions need to be addressed: (1), what are the technical considerations in performing the test and (2), how do the results of the P-A test compare with .'the more exacting laboratory procedures using either the Geldreich membrane filter or multiple tube test? In a review of technical con siderations, the P-A test ranks high for ease of examination. The sample can be added directly to the culture bottle at the field site, thus dispensing with the.sterile sample bottle and dechlorinating agent. Preparation of medium in sterile culture bottles could be dele gated to a cooperating public health laboratory and sufficient supplies stored in a dark, cool storage area for no more than four weeks. As another option, at least one commercial venture is exploring the poten tial for manufacturing prepared medium in a disposable culture bottle < so that no medium preparation need be done in the water plant. The answers to the secon/t question can be found In a statisitcal review of several evaluation'studies now completed (Jacobs, et al., 1986; Pipes, et al., 1986; Caldwell and Seidler, 1987). How Valid is the P-A Test Operational tests used by the small water plant, operator should not only be inexpensive and easy to perform but also acknowledged to produce data that is equivalent or better in precision to that of the standard laboratory procedures. It therefore becomes important to understand the validity of P-A test results obtained from a variety of water supplies in different geographical areas. This consideration is critical to both the operator proposing to use the test and the public health authority who needs to evaluate the water supply quality. Test sensitivity to coliform detection and parallel examinations by the P-A test and either the MF or FT procedure (or both) were done in three : widely divergent geographical areas. A data base of 1,483 samples in Vermont (Jacobs, et al., 1986), 1,560 samples in Oregon (Caldwell and Seidler, 1987). and 2,601 samples in eastern Pennsylvania (Pipes, et al:, 1986) obtained from small water systems, were analyzed statistic ally to determine if significant differences could be established between any of the three testing procedures;. Comparative data, in the Pennsylvania study did not include parallel examination by the FT test. The McNeraars statistical test (Fleiss, 1981) was selected as the most appropriate way .to compare the different coliform detection methods, since each water sample was examined in parallel, by more than one method. As noted in Figures 2 and.3, the P-A test significantly outper formed, both. the. multiple tube and membrane filter tests for coliform detection in water samples from supplies in Vermont and Oregon. Details of the statistical analyses are given in Table 2. Further analysis of these data also revealed that the multiple tube fermentation test detected more samples containing coliforms than the conventional meat- brane filter procedure using M-Endo medium. The statistical analysis of the eastern Pennsylvania data (Figure 4.) indicated that there was no significant difference in coliform detection by the P-A test and the MF method. The Chi-square value was only 1.27, signifying that compar ative results, between the P-A and MF test results were essentially, equivalent. Why there was no clear cut superiority to the P-A test as the best method for coliform detection in all instances may be related to .the state of.vigor for coliforms in the water supply. For the eastern Pennsylvania study, many of the positive results were obtained from samples of a small water system that did not chlorinate or other- Geldreich wise treat the groundwater supply. Regardless of this fact, these field studies on test performance do demonstrate, that the P-A method equals or exceeds the results obtained by the recognized conventional procedures. Therefore, there should be no reason not to use the P-A test as either an operational tool in the small water plant or as part of the official monitoring data. Table 2. McNemar1s Test Results Method McNemar1s Comparison Study X2 p-value P-A vs FT Vermont 5.3 0.02 Oregon 12.7 <0.01 P-A vs MF Vermont 52.0 <0.01 Oregon 39.9 <0.01 Pennsylvania 1.3 0.26 FT vs MF Vermont 31.4 <0.01 Oregon 8.3 <0.01 Operator Tests for Water Quality Lacking information on plant effluent quality seriously restricts the rural water plant operator's ability to promptly respond to unsat isfactory water quality conditions through treatment adjustments. Per haps this position leads to blind faith that water treatment can run automatically and produce a satisfactory water supply, of uniform qual ity. This may be true for protected groundwater supplies but in many small water systems, surface and ground raw water quality does fluctu ate because of poorly designed wells that do not protect water quality, from contamination by improperly operated wastewater treatment systems, stormwater runoff and animal activity in the area adjacent to water supply intakes. * Monitoring finished water quality for several key characterisitcs is. not beyond the realm of possibility in small water systems. There are available several basic techniques that water plant operators can learn to use with minimal training that will give them a measure of water quality being released as public water supply. Much critical information of immediate value can be obtained by frequent measurements for free chlorine residual, turbidity and coliform bacteria in the plant effluent. These three measurements will, provide information respectively on the continued maintenance of a treatment control process, Interference and protective shielding of microorganiras and verification that the treatment process is effectively removing coli form bacteria. Chlorine and turbidity measurements can be made at frequent intervals in the day and provide immediate information, using basic test kits that are readily available and easy to use. While Geldreich coliform testing requires a minimum of 24 hours of processing time it is an important record of treatment effectiveness for controlling the microbial quality of water produced and should be done at least once, per week. Bacteriological testing of public water supplies in.rural or re- . mote areas is not beyond the reach of operator capability. Both the MF and FT procedures have been packaged into commercially available field kits that may be used. For sheer simplicity, however, the < presence-absence (P-A) coliform test may be the method of choice. Adapting the P-A test into a simplified operational test would be ideal for on-site monitoring although it does not quantify the extent of contamination events that might suggest treatment breakdown, loss of distribution integrity or biofilm development. The procedure tests a single 100 mL sample, not 50 mL which must be divided into five replicate portions of 10 mL required in the multiple tube test. Mater ials required for the P-A test (media and associated glassware items) are more readily available and cost less in many third world countries than membrane filters and associated filtration equipment. While, the P-A concept was originally developed with a modified lactose broth (P-A broth), there is no reason why the test could not be applied to other coliform broth formulations (lactose broth, lauryl tryptose broth or MacConkey broth)to which bromcresol purple is added to indicate acid production* To further simplify the P-A test application to small water systems use, it should be permissable to utilize positive test information obtained from culture bottles after 24 to 48 hours incuba tion at 35°C without any further positive, result confirmation. While these positive results might occasionally contain a false positive reaction (gas produced by a non-coliform, organism) the error would be on the conservative side for safety considerations. These Options would provide the opportunity to use the P-A test in remote locations in the world, where regional suppliers Of laboratory materials carry limited choices of bacteriological media. The only critical consider ation would be in the preparation of triple strength medium to.be dispensed to the culture bottles in 50 mL volume. Medium dilution by the addition of 100 mL sample would create normal strength medium; P-A Test For Remote Monitoring Regional or national public health authorities need to monitor the wate*r quality of small systems in all areas of the District or Nation. This essential program should define a minimal number of water samples to be collected (on a monthly basis) from all public water supplies, . regardless of their remote locations and submitted to the central lab oratory for examination. Unfortunately, some samples may be in transit for several days before they reach the laboratory. During this time, . the microbial flora in the water changes, often leading to adverse, uncharacteristic test results that suggest the water sample meets national drinking water standards when indeed it does not (McDaniels . and Bordner, 1983). Since preservation of bacteriological samples is difficult to achieve without loss of low density coliform occurrences through nutrient depletion, extended contact time with chlorine, toxicity of Geldreich heavy metal impurities or microbial flora antagonism, the only other approach is to permit controlled growth during transits The P-A test could be a solution to this problem. The sample is added directly to the medium, then the culture bottles transported (maximum of 5 days) back to a central laboratory. The procedure is uncomplicated and requires very minimal effort by the water plant operator to perform (Pipes, et al., 1986). During transit, ambient air temperatures will influence the magnitude of sample culture growth, being slow during winter temperatures below 10°C and accelerated at summer or tropical temperatures that may reach normal (35 C) incuba tion temperature. Unon. arrival in the laboratory, observation, for gas production is made t.~; determine what processing will be required. If turbidity and gas production is evident, the culture is confirmed in brilliant green lactose broth for evidence of coliform occurrence. If no gas production is noted, the culture bottle is incubated for 24 to 48 hrs at 35°C; then confirmed, if positive. Those culture bottles with no acid production or no visible turbidity but. gas in the inner tube indicate air was shaken into tube during shipment, so these must be inverted to release the entrapped air and incubated 24 to 48 hrs as for the other negative cultures received for processing. The procedure, is not labor intensive and would provide information.on any coliform occurrences that wera related to the original water sample. Summary Development of a practical microbiological monitoring program for small water systems is urgently needed at the local level for prompt detection of contamination followed by appropriate remedial actions. Three operational measurements are within the capability of the water plant operator: chlorine residual, turbidity and a basic test for/ coliform occurences using a presence or absence (P-A) concept. Evalua tion of the P-A test in three different geographical areas demonstrated the procedure to outperform both the membrane filter and fermentation ^ tube (MPN) tests for coliform detection. The use of aP-A test approach is recommended for small water system personnel because of its sim plicity and adaptability to minimal resources available to the public system sector in rural and remote regions world-wide... Further adapta tion of the test can be made to initiate field inoculation of the sample to be sent to a central laboratory for final processing and data gathering on water quality nationwide. References American Public Health Association (1985). Standard Methods for the Examination of Water and. Wastewater. 16th ed., 1268 pp., Washington, D;C. Caldwell B. A. and Seidler, R. J. (.1987) Comparison of Bacteriologi cal Assays and Sampling Regimes, for. Increased Coliform Detection in Small Public Water Supplies. Appl. Environ. Microbiol. (In Press;. Craun G. (1985) An Overview of Statistics on Acute and Chronic Water Contamination Problems, p 5-15. In Fourth Domestic Water Quality,, q ' Geldreich Symposium: Point-of-Use Treatment and Its Implications. Water Quality Assoc., Lisle, 111. Fleiss, J. L. (1981) Statistical Methods for Rates and Proportions. 2nd ed. John Wiley and Sons, New York. Geldreich, E. E. (19 78). Bacterial Populations and Indicator Concepts in Feces, Sewage, .Scortnwa'ter'and Solid Wastes. In Indicators of Viruses in Water and Food, G* Berg ed;* 422 pp., Ann Arbor Science Publishers Inc., Ann Arbor, MI. Jacobs, N. J., Zeiglei-i W. L. , Reed, F. C. Stukel, T. A, and Rice, E. W; (1986) Corapari5o:i of Membrane Filter, Multiple-Fermentation-Tube, and Presence-Absence Techniques for Detecting Total Coliforms in Small Community Water Systems. Appl. Environ. Microbiol., 51:1007-1012. Kowal, N. E. (1982) Health Effects of Land Treatment: Microbiological. U.S. Environmental Protection Agency, EPA-600/1-82-007, Health Effects Research Laboratory, Cincinnati, Ohio. Lippy, E. and Walt rip, S. (1984) Waterborne Disease. Outbreaks, 1946 - 1980: A Thirty-Five-Year Perspective. Jour. Araer. Water Works Assoc, 76:60-67. McDaniels, A. E. and Bordner, R. H. (1983) Effects of Holding Time and Temperature on Coliform. Numbers in Drinking Water. Jour. Amer. Water Works Assoc, 75:458-463. Pipes, W. 0., Minnigh, H. A., Moyer, B. and Trog, M. A. (1986) Compari son of Clark's Presence-Absence Test and the Membrane Filter Method for Coliform Detection in Potable Water Samples. Appl. Environ. Microbiol. 52:439-443. Pipes, W. 0., Minnigh, H. A. and Troy, M. (In press). Field Incubation Of Clark's P-A Test for Coliform Detection. Proc. Water Quality Technology Confr., Portland, OR. World Health Organization (1984) Guidelines for Drinking Water Quality, Vols. 1-3. Geneva, Switzerland. Geldreich i-1 o H o Membrane Filtration Test h- ! CO i m i M - Endo Agar h- ; 35 C 24 HRS < ! 1 Z i O ' Coliform Colony H '• Count < ! oc : UJ i OL • o : Calculate Conform Denslty/100 ml FIGURE 1. PATHWAYS FOR TOTAL COLIFORM DETECTION IN WATER SUPPLY so. P-A VERwCHT 1ITIVE TEST p.A FT METHOO ORECON STATE P-A ONLY CS3 rt ONLT ricur.E 2. p-a vs n ueihoo comparisons SO S 20 E 10 S n ur VERMONT 01EGON POSITIVE TEST CS3FT ONLY FT UT P.» Mf- P-A Uf P-A HF -ETMCO VERMONT OftEGON PENNSYLVANIA STAlE positive test can* P-A ONLY cz=3 wf ONLY FIGURE i. P-A VS WF UETHCO COUFAPISOnS ufTMCO STATE 1 Wf ONLY FIGURE * FI VS MF UfTKOO CGWPAXlSCWS 11 Geldreich Applied and Environmental 'Microbiology. May. 1986, p. 1007-1012 0099-2240/86/051007-O6S02.00/0 Copyright © 1986, American Society-for Microbiology Vol. 51, No. 5 Comparison of Membrane Filter, Multiple-Fermentation-Tube, and Presence-Absence Techniques for Detecting Total Colifprms in Small Community Water Systems NICHOLAS J JACOBS.1-*-WANDA L. ZE1GLER,1 FRANK C. REED.2 THERESE A. STUKEL,3 and EUGENE W. RICE"* Department ofMicrobiology, Dartmouth Medical School, Hanover, New Hampshire 037561; Connecticut River Watershed Council, Easthampton. Massachusetts 010272; Department ofCommunity and Family Medicine, Dartmouth Medical School, Hanover, New Hampshire 037563; and Drinking Water.Research Division, Water Engineering Research Laboratory, U.S. Environmental Protection Agency, Cincinnati, Ohio 45268* Received 2 August 1985/Accepted 25 February 1986 Methods for detecting total coliform bacteria in drinking water were compared using 1,483 different drinking water samples from 15 small community water systems in Vermont and New Hampshire. The methods included the membrane filter (MF) technique, a 10-tube fermentation tube (FT) technique, and the presence-absence (P-A) test. Each technique was evaluated using a 100-ml drinking water sample. Ofthe 1,483 samples tested, 336 (23%) contained coliforms as indicated by either one, two, orall three techniques. The FT detected &2%> the P-A detected 88%, and the MF detected 64% of these positives. All techniques simultaneously detected 55% ofthe positives. Evaluation ofthe confirmation efficiency ofthe P-A technique showed 94% ofthepresumptive positives confirming ascoliforms. Thirteen different species ofcoliforms were identified from the 37 testsin which the P-Awas positive but the MF and FT were negative. The P-A test was simple to inoculate and interpret and was considerably more sensitive than the MF and slightly more sensitive than the FT in detecting coliformsin this type of drinking water supply. Currently, the membrane filter (MF) technique and the multiple-fermentation-tube (FT) technique, as described in Standard Methodsfor theExamination of Water and.Waste- water (1), are the only procedures approved for monitoring drinking water systems for total coliforms under the Safe Drinking Water Act (14)^ Both techniques have been evalu ated and compared in severalstudies(4, 13, 19, 20, 22. 24). The conclusions from these.studies have suggested modifi cations of the media and procedures of both methods for moreaccurate results (11,.13,17,18, 22).Other investigators have proposed a presence-absence (P-A) technique as an alternative (7, 8, 23, 25). The P-A technique, a basic simplification of the FT procedure, was developed by J. A. Clark as a qualitative means of monitoring drinking water systems. Although it wastested inparallelwiththe MFon different drinkingwater systems and found to be as sensitive (7), it has not been compared with both the FT and.MF together. In this study, the sensitivity of each of the methods (MF, FT, and P-A) was compared using drinking water samples from a variety of small community water systems. The methods used were the conventional MF technique, a 10- tube FT technique, and the most recently proposed P-A technique (1).Smallwater systems located in rural areas arc quite variableand are often less well protected than munic ipal supplies..In addition, appropriate laboratory facilities for monitoring are not readily available. The resiilts pre^ sented in this study should assist in developing appropriate methods for improving this type of water supply. MATERIALS AND METHODS Samples. Over a 1-year period, a total of 15 small community water systems in Vermont and New Hampshire * Corresponding author. weresampled.These water systems (Table1)each serve less than 1,000 persons and; have more than 10 service connections. Sources of water for these systems include shallow wells, deep wells, and springs. Most systems did not chlorinate or filter their water. Samples were collected from one location ineach system on a weekly basis. Once a month, five sampleswere collectedfrom five sites in the distribution network of each system, including the weekly site. Samples were collected according to the guidelines in Handbook,for Evaluating Water Bacteriological Laborato ries (16). Sterile 500-ml plastic bottles (polymethylpentene) containing sodium thiosulfate (16) were used as sample containers. Samples were kept in an ice chest, transported to the laboratory within 2 to 3 h, and.analyzed within 5 to 6 h after collection. Microbiological procedures. Each sample was analyzed by each of the techniques, using 100-mlwater portions for each test. The MF procedure Was performed by methods detailed in references 1 and 5. HA membrane filters (Millipore Corp., Bedford, Mass.) were placed on sterile pads (Millipore Corp.) saturated with M-endobroth (Millipore Corp.; Difco Laboratories, Detroit, Mich.) and incubated for 24 h at 35°C. A minimum of five of. the typical green-metallic sheen . colonies from each positive sample were transferred to lauryl tryptose broth (LTB; Difco) and brilliant green bile lactose broth (BGLB; Difco) for.verification as coliforms (5). Production of gas in LTB and BGLB withiri 48 h was considered a positive test (5). At least one colony per positive sample wasexamined for lactosefermentation after growthon Levineeosirt methyleneblue (EMB)agar (Difco). The FT technique involved 10 tubes, each containing 10 ml of double-strength LTB and a fermentation tube. The addition ofa 10-ml sample of water to each tube aliowed a total of 100 ml of water to be examined. For each positive sample, all presumptive positive tubes up to a maximum of 1007 1008 JACOBS ET AL. Appl. Environ. Microbiol. TABLE .1. Description of the 15.drinking water systems System no. Year established . No. of service connections Apprbx. no. of Water source" Chlorination* Filtration0 Oil 1928 11 021 1930 70 031 1960 20 041 1946 238 051 1965 .38 061 1960 :i / 071 1965 ia 081 1900 2S0 091 1949 3:j0. 1011 1977 40 1021 1968 . 30 1031 1952 13 1041 1935 70 1051 1970 98 1061 1948 135 30 1 240 1 55 1/5 910 1 90 2/5 70 • 3 55 1/5 650 1 930 1/5 75 r 65 • 1/5 30 1/5 325 •? 350 • •? 420 1/5 "Water source: 1, deep well; 2, shallow well;3. springs; 4. wells to reservoirs; 5. wells to coveredreservoirs. * Chlorination: -..none; +. chlorination with hypochlorite. ' Filtration: -, none; -t-, yes. five were confirmed for gas production in BGLB. As indi cated in Results, these water, samples demonstrated a high confirmation efficiency, and therefore the confirmation of five tubes was sufficient to judge a.given Sample as contam inated with confirmed coliforms. At least one BGLB tube was transferred to Levine EMB agar. The number of coliforms per 100 ml was estimated from a 10-tube most- probable-number (MPN) table (1). The, procedure was the same as previously described (5), with the omission of the final step of the completed test, in which typical colonies on EMB agar are transferred to LTB. To assess the validity of this modified procedure, a resampling scheme was con ducted inwhich 49positive samples (14% ofthe totalnumber of positive samples of this study) were carried through to the final step of the completed test (see results below). The P-A technique used was a recent modification of the original test proposed by Clark (7). The test consisted of a single culture bottle (250-mJ milk dilution bottle), containing a 50-ml portion oftripie-strertgth medium plusa fermentation tube (12 by 75 mm). The formulation of the medium at single strength was: 13.0 g of lactose broth (Difco), 17.5 g of LTB, and 0.0085 g of bromocresol purple (Difco) (dissolved in 10 ml of 0.1 N NaOH before addition to broth) in 1 liter of distilled water. The prepared bottle was autoclaved for 12 min at 121°C and was stored until used. Before samples were added, the bottle was inverted to empty the medium out of the fermentation tube. After the bottle had been inoculated with a 100-ml sample portion, it was inverted to fill the fermentation tube. The bottle was incubated at 35 ± 0.5°C and inspected after 24 and 48 h for production of acid or acid plus gas. Bottles showing any degree of color change from purple to yellow or brownish yellow, were subcultured for confirmation. Inoculum from a presumptive positive test was transferred to BGLB for confirmation and then to Levine EMB agar for detection of lactose-fermenting colonies. In cases in which the P-A test was positive and the MPN (FT) and MF methods were negative, organisms were iso lated in pure culture from Levine EMB agar and identified using API 20E identification test strips'(Analytab Products, Inc., Plaihview, N.Y.). The species names given-are those obtained from the API identification scheme. A quality assurance program was performed as outlined in reference 5. Statistical methods. McNemar's test (15) was used to compare the overall proportion of positive samples detected by different methods. This test does not take into account the disagreement on individual samples. To study patterns of agreement between methods, kappa measures of inter-rater agreement (2.15) were used. Kappa is an index taking values between -1 and 1, indicating the relative agreement between two techniques beyond chance agreement, k = 1 indicates maximum possible agreement; values greater than 0.75 rep resent excellent agreement, and values below 014"represent poor agreement beyond chance, k = 0 indicates chance agreement only. A further analysis was performed to study where disagree ments of the P-A and the FT with MF occurred. This analysis was based on the obviously unproven assumption that MF gives the correct conclusions. The MF technique was chosen as the standard for comparison because of its wide usage among water laboratories. Conditional kappa measures of agreement (2) with FT were calculated given MF counts of 0, 1 to 4 (inclusive), and >5 per 100 ml. By collapsing the 1 to 4 (inclusive) and >5 categories into a ^1 category, an additional measure of agreement was computed conditional on a positive MF count. Due. to the dichotomous nature of the. P-A test, conditional kappa measures could only be computed given zero and positive MF counts. RESULTS Comparison of the MF, FT, and PrA methods. Of the 1,483 samples analyzed in this study, 336 (23%) confirmed positive samples were detected by either one, two, or all three of the techniques. The total coliform counts of these samples ranged from 1 to over 300 organisms per 100 ml. A compar ison of the three methods in detecting the presence of coliforms is shown in Fig. 1. The FT technique detected 275 positive samples (19% of the total samples, 82% of all positives), whereas the MF technique detected 216 positive samples (15% of the total samples, 64% of all positives) and the P-A technique detected 296 positive samples (20% of the total samples, 88% of all positives). All three techniques simultaneously detected coliforms in 185 samples. There was a statistically significant difference, when analyzed by McNemar's test (15), in the detection rate of positive sam ples between each pair of methods. The MF detected fewer Vol. 51, 1986 COLIFORM DETECTION IN SMALL COMMUNITY WATER SYSTEMS 1009 ..FIG. 1. Comparison of three methods for detection of *ater samples positive for coliforms. The number'of positive samples Sled by each technique is given in the text, ^uve^mpks are defined as at least 1coliform per 100 m! ot the MPN (FT) and M^ tests, ora positive reaction inthe P-A'.est. positive samplesthan either the MPN (FT) (p <0.001) or the P-A (p < 0.001); the P-A detected slightly more positives lh?0la^a^pa°Son ofsamples with both low and high coliform densities, the coliform counts for the FT and the MF test were divided into three categories: negative, ==5, and 1 to 4 (inclusive) per 100 ml. The '^ter represents an arbitrary figure for identifying systems, with |ow cohform den ties. Figure 2indicates the cross-classified coliform counts for each pair oftechniques. These data were analyzed statistically (see below), but ™\c^™%*^°™i >? P-A test was always positive when the MF count was so PA PA MPN VALUES MPN VALUES . 0 1-4 >5 1208 MF COUNTS 0 1-4 >5 1167 20 0- | 100 114 sil 134- 82 0 1-4 >5 1267 MF COUNTS 0 1-4 >5 1184 24 0 | 53 42 2 j 30 68 80 J 1267 134 82 1187 296 1483 1208 97 178 1483 TABLE ^ Kappa measures of agreement and their standard errors for -pairwise agreement between the three techniques Comparison Observed k (SE) PA vs MPN (FT) (Ovs-ir... ••• .0-83 0-02 PA vs MF (0 vs ^1) : 074 0 02 MF vs MPN (FT) (0 vs >U X» J02 MF vs MPN (FT) (overall)0 ••• 0.59(0.02) FIG. 2.' Cross-classified coliform counts for each pair of micro biological techniques. •MOvs 21)indicates negative versus positive niwi;><5 "Classification of samples into negative. 1 to 4 (mclusrve). and a5 categories. . • (Fie ?B). Furthermore, there were 30 samples where the MPN (FT) value was s5, but the MF was negative (Fig. 2C). The statistical analysis for the data in Fig. 2is presented as kappa values (see Methods) in Table 2. The agreement between P-A and FTfor a simple negative-positive classifi cation (0 versus al) was excellent. Agreement between MF and the other two techniques for the same binary classifica tion was not as strong but still good. Moderate agreement only was found between FT and MF for overall classification of samples into negative, low-coliform (1 to 4), and high- coliform (s5) categories. In Fie ^C the counts below thediagonal are much higher than thoS; above the diagonal: of the 177 total disagree- ments 151 (85%) occurred below the diagonal. This indi cates that when .there was disagreement on a"^particular sample, coliform counts for FT were usually higher than, those for MF. The same pattern of disagreement occurred between the P-A and MF (Fig. 2B): there were 120 disagree ments 100 (83%) of which showed MF negative and P-A positive. It should be noted- however, that.agreement.be- tween FT and P-A was very strong.(Fig. 2A and Table 2). There were fewer (75) overall disagreements; 48 of these (64%) showed P-A positive and FT negative. The lower sensitivity of the MF test is also apparent from the data in Fig. 1, which compares all three techniques as either positive or negative. There were 63 samples (19% of the positives) for which both the FT and P-A tests together were positive while the MF test was negative. In addition, for 120 samples (36% ofthe positives) the MF was negative and either the MPN (FT) or the P-A gavc: P05^^"11^ Furthermore, there were only 13 samples for which the MF was positive alone, 7 samples for which the MF and FT alone were positive, and 11 samples where the MF and P-A tests alone were positive. . • In summary, both the MPN (FT) and the. P-A methods appear to be more sensitive than the MF test. That the FT and P-A strongly agree gives some evidence that the Mr- could be less sensitive. In addition, the.P-A appears to be a more sensitive test than the FT. Aformal analysis ofthe disagreements between the MF and the P-A and FT techniques is shown in Table 3, using conditional kappa values (see Materials and Methods). When MF > 5, agreement with the FT Avas excellent, i.e., MPN value ofs5. When MF > 1,agreement was very good, i e P-A positive; MPN value of>1. However, when MF - 0, agreement (i.e., P-A negative; MPN = 0) was only moderate. In summary, when the MF technique indicated a positive sample, the.other two techniques showed strong agreement: whea the MF technique indicated a negative sample, agreement was not as strong. Furthermore^the higher the coliform count for the sample as determined by MF. the stronger the agreement with the FT technique. Very poor aereement was found between Fr and MF when MF density was between 1 and 4 (inclusive) (see Fig. 2C and HtlU IACOHS II ai I AMI I * riiiiilitiKii.il k.i|»|».i ni.MM'.u-. ••! .-y.u•<. nirii! Willi Ml .i> • tin- sl.nnl.ii«l' H|.M-nnl • Nl I. Ml ..Mllll ... MI'N ill I :ik.« «•«•"«••'" . I • if-n.-miii 0 I • I l) ».s <0 0»l l> .'(» (0 I)"- (I S<. lit »>U 0 •).' ((> ill) (11.1 ill I).') • N .V il SS lil IM> N A S,v M.il.ll.lls.lil.l M.Ul.xK I"" . \i'l.«:»..ii.'.' •' « ••,l.l1!-...ii k.i|.j..i iih-.imii. ihI l.'i llu- iim- ol Ml .is iI**" sLnnl.iu! I>" >•""•i'-•: ••"' "• N \. Not ipplu .tl'l.' lahlc \). although 110 of IU iS2- ,_)»»*. liusc samples tested positive In 1-1. <>S of llu- HI lM'• >samples had ^ colifonus In llu- II lest Again. H .was seen that the IT consistently read nunc cohloinis for'a sample ilian llu- Ml ( mtiiriiialinii tHitioncY and importance of" gas production. A useful screening test should exhibit a high (alio-of pic sumplive positives toconfirmed postiivcs. We (. aktilatcd the percentage of presumptive positive tests thai wni' suhse qucntlv confirmed (coiilit malum cl!k icncv) foi each lech nique\Table I). Phc continuation dtini'iKV ol llu- Ml technique wnv'li'l, with <>>•!.ol 702 suspected colonies picked from llu- membrane tillers conliiiucd as cohfoiins. llu; confirmation\lVuioiuv of the .l-T lechm'quc was «> \'- . wilh 1.04* ol"l.l2> presumptive positive lubes continued as • coliforuis. llu- clViciencv «>t Ihc PA lest.compared favorably. Of ihe U(> presumptive positive PA tests that showed acid (either .. strong orweak) ami gas. •M'-.'- were conlirmcd.ascolifornis Twocategories" ol positive PA tests were noted Ihe vasi •majority showed both a strong acid tvellow color! and gas reaction. Of 277 of these tests, only S weie not confirmed, giving al>7*", eonlinualion ellicieiicv loi this category. Ilow ever, there were '.'»')'borderline tests whieh hail relatively slight color chanties (a brownish yellow) with a small amount of gas (4 lob-mm gas bubble in fermentation tube). Ofthese W* cases. 27 were continued, giving.a <»l>'j confirmation elliciencv. In contrast, there were M tesis tn which only acid reactions ami no lias pnuluclion was ohscrvcil. None ol lliese was conlirmcil. iiuhcalinii the critical importance ol examininti the PA test lot tias pH>thiction. A resamplinti schemewasconducted to assess the validity IAIM T -4. fontiniialion oHieieiK-K-s *»l'the I'-A. MI'N ', I Ml- techniques L|')..aiul I'-A MI'N it'n Ml' IV<v-|l|«lli>ll t'l U->l K-M|ll Slr»>iig acid, ti'as Slight acid.'' gas Sit one or sliiihl acid. •g;is'-" Stroiti: or slight acid..no gas lubes with gas Metallic-sheened colonics 702 .lll'KvUil N... ceil •. i on i> inntit liiimsl tn mill m.iltiin ._. , — , ..- m 2'<K> «)".' V) 27 (»'» }\u'- :•>(.' >M M 0 0 1.125 1.0-H (i54 l>\ Slight ;wkl w;is iiulKMlCil In .i hi-o\Miisli->elK»\v sol»r. Siinii'l'lhc iwo lines alnno . . Al'l'l'. liNVIRON. MltHOHIOI.. ol oui moililicaiion of the IT. technique. As indicated in Maleuals and Melluuls, we omitted the final s'lep of ihe. completed lest, in which typical isolated colonies on KMB at-ai aie liansleiied l«( I 111: l-ach ol" the •l-S waler systems was usampled an equal number of limes. A iolal of 49 positive samples vveie obtained for Which at least one IIIhe gained thuMiiih to Ilie entire completed step (5). In this usamplmn scheme, we subcnlluieil all IUJLFJ lubes positive at the coiiliuuetl slaj-'e. These samples showed a Iolal iil 147 lubes dial earned through to Ihe lirsl step of the completed U-si. wheie typical ctilonies wereohscived on llMHajiar. All of these typical colonies weie transferred lo I.TH, Ihe final step ol the completed lest, and produced na> within 4X h. 'Ihese usiills uulicate that the obseivalion of typical co|o- nus on l-MH a>',ai was a.line rellection of ihe presence of cohlorms Ilie'relore, for the water samples we investigated, oui modilicalioii of Ihe completed step ol" Ihe IT procedure was valid. Organisms isolati-d IV llu- PA Usl. Organisms were isolated ami identified I'romTlie <7 PA tests thai were positive when Ihe Mb and l"l tests were negative. The API system for identification, and.nomenclature was used. Table, s lists Ihe organisms along with the number of limes they, were isolated-. ( Hinh,u trr liriunlii was the most frequently isolated organism, followed by If.nlrnthtu Irr ny.nitimrnnix, and Srmilin I'lvimitliii*!. Most («I"J) of these M presump tive positive P A tests were noticed oil day 2of incubation, rather than at 21 h. The color of these positives ranged from brown to vellow.-and gas was detected in each case. Ivach ol the pure isolates-which"'was designated as a species ol the genus Sririiliii was rciuoculalcd intolactose broth loconlivm its ability to produce gas from lactose. All Ihese organisms produced gas upon rclesiing The identities of the .V. phniuthifti species were confirmed by.Ihe Analylab Aerobe I.aboralory (Aualyrab Products. Plainview, N.Y.). Comparison oftin- IO-tiil>c VY wilh ilu- 5-luhf Kl'tests, the I'T technique tested in this study used 10 10-inl lubes, as opposed tt> the l-T technique which utilizes only S 10-ml lubes (I). As a means of comparing the lO-luhe wilh Ihe s tube l-T. we separately tabulated Ihe number of water samples which would have tested positive ifonly5tubeshad been analyzed instead of 10. This comparison was done by separating Ihe 10-lube method into two sets of 5 tubes, consisting of the even- and odd-numbered tubes. Of the 275 water samples positive by IT. both setsof livetubes showed 'I'AIM.I-i s ()ig;iiiisins isolated Irom positive I'-A tests when the Ml aii«J MI'N <I-T) lesls were negative. No. ol nines ls,,l-,k" ' isoliilcd. Iiiltinhdi In itwItWH ntii.\ (> I., ,),„i, tic .......:, • "* ' •/•.. tirmi'riU'M • • (ijiiihiutrrliriniilii... * list licrii liin <oli 1 klihsiiiln niiruniiniiiif . • i h . OZlHIUIC ......... : : ' K. I'Wlniii. :•-. ••• ^ Sitrutiii plyinnthh-ii.. ' '•••.- '» .V. '1'iuilit olti • • •. • • ' ,S. nihiiliiti' • • .S. inliriji-Ki ..'.. •• -. 1. Iln/niii alvii '• • • t •• li'ditilkil accoiilinu lo ihc prolilcnumbers dclcrmined by API. Vol. 51, 1986 COLIFORM DETECTION IN SMALL COMMUNITY WATER SYSTEMS 1011 TABLE 6. Positive samples detected by MF, FT, and P-A ' techniques in each of the 15 water systems System No. ofpositive samples by. MF. '. FT P-A on 18 •' 33 • 38 021 17 ... 21 • 22 031 55 57 62 041 0 1 0 051 1 . 2 •• 2 061 82 100 96 . 071 6 i5 16. 081 6 2 ' 5 091 9 .: 12 12 1011 2 •i 8 1021 15 24 • 25 1031 4 • 2 3 1041 0 ' • 4 • ' 4 1051 0 0 0 1061 1 .1 3 at least one positive tube in 207(76%) of the samples. On the other hand, there were 68 (24%) samples for which only one set of five tubes showed one or more positive tubes. Thus, assuming an equal distribution of col.iform in the tubes, on average 88%of the positivesfrom the 10-tube methodwould have been detected using only 5 tubes (88% = 76% + 1/2 24%). Comparison of the three methods in individual water sys tems. Approximately 100samples were collected from each system.The results for each individual system are compared in Table 6. Systems 061 and 031 were the most frequently contaminated, and the MF technique compared favorably with the P-A technique in detecting positives in these two systems (Table 6). In contrast, the P-A technique detected many more positives than the MF technique in some of the less frequently contaminated systems. For instance,, system 011 waspositive 38timesby the P-A biitonly18times by the MF procedure (Table 6). DISCUSSION The results indicate that, under the circumstances of our tests, the P-A is much more sensitive than the MF and slightly more sensitive than the FT. There are several factors that may account for this observation. The MF technique gave negative readings in many sam ples for which the MPN (FT) and P-A methods were positive. Previous studies have also.reported the lower sensitivity of the MF technique in recovering conforms. Reasons suggested for this failure have included the survival. of coliforms on a membrane filter surface compared to survival when in broth (23), the failure to revive injured coliforms or weakened cells (19, 24), orthe possibility that the M-endo broth used in the MF test is a selective medium which may be inhibitory to' stressed coliforms (3, 4, 24). Inoculation of a sample into an enriched broth-based me dium with a prolonged incubation period may enhance the recovery of indicator organisms (3, 24). These organisms may have been injured coliforms that needed more time to grow. The 10-tube FT technique appears as an optional proce dure in the 16th (1985) edition of Standard Methods for the Examination of Water and Wastewater. (1). This 10-tube technique provides more precise MPN values than the 5-tube FT. With the particular set:of drinking water samples that we studied, 12% of the positive samples on average would have been read as coliform negative if.the 5-tube FT had been utilized instead of the 10-tube method. Since this 10-tube FT procedure uses a 100-ml sample of. water, it is comparable to the MF and P-A methods, which also use 100-ml samples. The FT was considerably more sensitive in terms of recovery than the MF procedure. It was surprising that the FT was slightly less sensitive than the P-A test, since they are both broth-based methods relying on gas production and are both incubated for up to 48 h. A possible explanation is the slight difference in medium composition between the P-A and FT .tests. There is a.slightly higher lactose concentration in the. P-A (0.75%) than in the FT (0.5%). It has previously been found that isolates anaerogenic in the FT medium (LTB) were aerogenic in a broth containing a slightly higher lactose concentration (12). Members of the genera Citrobacfer, Enterobacter, and Kleb siella were most frequently isolated (12). These organisms were also isolated in the present study from the positive P-A tests when the MF and FT were negative. Other possible explanations for the failure of the FT technique to detect coliforms are given in other studies (6, 11, 13, 1.7, 21). : The P-A test showed a high confirmation efficiency, which is an important characteristic for a screening technique. This high confirmation efficiency was found only in tests that showed both acid and gas production. P-A tests that showed acid but no gas were not confirmed and probably contained lactose-fermenting species other than coliforms (8-7IO). Another important characteristic for a useful screening technique is the ease of detection of a positive test. With the P-A test, we. found that a large percentage of all positive samples (91%) showed a distinctive yellow color and gas. Clear negatives (77% of all samples) were readily noted by no change in the purple color and no gas. However, there were a small percentage (3% of all samples), which showed only a borderline color change (yellow-brown or brown), but did show some gas production. These were usually true positives, with a 69% confirmation efficiency. However, when no gas was observed, these borderline color changes did not confirm. These findings have important practical implications for the reading of borderline P-A tests in the field. Borderline color changes which' showed a small amount of gas represented approximately 10% of our posi tive P-A tests. Only 69% of these borderlines were confirmed as coliforms. Therefore, it is particularly important to carry all P-A tests with borderline color changes through to the confirmation step before making any decision about the presence of coliforms. ACKNOWLEDGMENTS We thank Elizabeth Fuller, Raymond Neff, and Ray Gleason for their aid in sampling and statistical analyses. C. T.. Gray and E. Robert Greenberg are also acknowledged for their collaborative efforts. This project was supported by ILS. Environmental Protection Agency .through a cooperative agreement (CR-810805) wilh the Connecticut River Watershed Council. LITERATURE CITED 1. American Public Health Association. 1985. Standard methods for the examination of water and wastewater,. 16th ed. American .Public Health Association, Inc.. Washington, D.C. ' 2. Bishop, Y. M., \V.. E.. Fienberg, and P. W. Holland. 1975. Discrete multivariate analysis: theory and practice. MIT Press, Cambridge, Mass. 3. Bissonette, G. K., j. J. Jezeski, G. A. McFeters, and D. G. Stuart. 1975. Influence of environmental stress on enumeration 1012 JACOBS ET AL. of indicator bacteria from natural waters. Appl. Microbiol. 29:186-194. 4. Bissonette, G. K., J. J. Jezeski, G. A. McFeters, and D. G. Stuart. 1977. Evaluation of recovery methods to detect coliforms in water. Appl. Microbiol. 33:590-595. 5. Bordncr, R., and J. Winter (ed.). 1978.Microbiological methods for monitoringthe environment—water and wastes. U.S. Envi ronmental Protection Agency, Cincinnati. 6. Chambers, C. W. 1950. Relationship of coliform oactcria to gas production, in media containing lactose. Public Health Rep. 65:619-627. 7. Clark, J. A. 1968. A presence-absence (P-A) rest providing sensitive and inexpensive detection of coliforms feca!coliforms and fecal streptococci in municipal drinking water supplies. Can. J. Microbiol. 14:13-18. 8. Clark, J. A. 1969.The detection of various bacteria indicative of water pollution by a presence-absence (P-A)procedure. Can. J. Microbiol. 15:771-780. 9. Clark, J. A. 1980. The influence of increasing numbers of non-indicator organisms upon the detection of indicator organ ismsby the membrane filterand presence-absence tesls. Can. J. Microbiol. 26:827-832. 10. Clark, J. A., and L. T. Vlassoff. 1973. Relationships among pollution indicatorbacteria isolated from raw water and distri bution systems by the presence-absence (P-A) test. Health Lab. Sci. 10:163-172. 11. Evans, T. M., M. W. LeChevallier, C. E. Waarvick, and R. J. Seidler. 1981. Coliform species recovered from untreated sur face water and drinking water by the membrane filter, standard, and modified most-probable-numbcr techniques. Appl. Environ. Microbiol. 41:657-663. 12. Evans, T. M., R. J. Seidler^ and M. W. LcChcvallier. 1981 Impact of verification media and resuscitation on accuracy of the membrane filter total coliform enumeration technique, Appl. Environ. Microbiol. 41:1144-1151. 13. Evans, T. M., C. E. Waarvick, R. J. Seidler, and M. W. LeChevallier. 1981. Failure of the most-probable-number tech nique to detect coliforms in drinking water and raw water Appl. Environ. Microbiol. supplies. Appl. Environ. Microbiol. 41':130-138. 14. Federal Register. 1975. National interim primary water regula tions. Fed. Reg. 40:59566-59588. 15. Fleiss, J. L. 1981. Statistical methods for rates and proportions, 2nd ed. John Wiley and Sons, New York. 16. Geldreich, E. E. 1975. Handbook for evaluating water bacte riological laboratories, 2nd ed. U.S. Environmental Protection Agency, Cincinnati. 17. Geldreich, E. E., H. D. Nash, D. J. Reasoner, and R. H. Taylor. 1972. The necessity of controlling bacterial populations in potable waters: community water supply. J. Am. Water Works Assoc. 64:596-602. 18. LeChevallier, M. W., S. C. Cameron, and G. A. McFeters.1983. New medium for improved recovery of coliform bacteria from drinking water. Appl. Environ. Microbiol. 45:484-492. 19. McFeters, G. A., S. C. Cameron, and M. W. LeChevallier. 1982. Influence of diluents, media, and membrane filters on the detection of injured waterborne coliform bacteria. Appl. Environ. Microbiol. 43:97-103. 20. McFeters, G. A., and D. G. Stuart. 1972. Survival of coliform bacteria in natural waters: field and laboratory studies with membrane filter chambers. Appl. Microbiol. 24:805-811. 21. Meadows, P. S.\ J. G. Anderson, K. Patel, and B. W. Mullins. 1980. Variability in gas production by Escherichia coli in enrichment media and its relationship to pH. Appl. Environ. Microbiol. 40:309-312. 22. Morgan, G. B., P. Gubbins, and V. Morgan. 1965. A critical appraisal of the membrane filter technic. Health Lab. Sci. 2:227-237. 23. Pipes, W. O., and R. D. Christian. 1984. Estimating mean coliform densities of water distribution systems. J. Am. Water Works Assoc. 76:60-64. 24. Snipe, E. L., and G. M. Cameron. 1954. A comparison of the membrane filter with the most probable number method for coliform determinations from several waters. Appl. Microbiol. 2:85-88. 25. Weiss, J. E., and C. A. Hunter. 1939. Simplified bacteriological examination of water. J. Am. Water Works Assoc. 31:707-713. American Journal or Epidemiology Copyright c 1982 by The Johns Hopkins University School of Hygiene and Public Health All rights reserved Vol 115. No. 4 Printed m I'S.A. SWIMMING-ASSOCIATED GASTROENTERITIS AND WATER QUALITY' V. J. CABELLI.* A. P. DUFOUR.1 L. J. McCABE.3 and M. A. LEVIN4 Cabelli, V. J. (Oept. of Microbiology, U. of Rhode Island, Kingston, Rl 02881), A. P. Dufour, L. J. McCabe and M. A. Levin. Swimming-associated gastroen teritis and water quality. Am J Epidemiol 1982;115:606-16. Adirect, linear relator)fth,JQ hoti^ftn cujjrnminrj.»c«nfif*tt>ri qagtrpjntARtinal ..ness and the quality of the bathing water was obtained from a multi-year. muHipfe-lftcation prospective epidemiologic-microbiologic research program rnndnrted in New York Citv. 1973-1975. Lake Pontchartrain. Louisiana. 1977- 1978, and Boston. Massachusetts, 1978. Several microbial indicators were used in attempting to define the quality of the water; and, of those examined, enterococci showed the best correlation to total and "highly credible" gastro intestinal symptoms, 'the frequency of gastrointestinal symptoms also had a high degree of association with distance from known sources of municipal wastewater. A striking feature of the relationship was the very low entero- coccus and Escherichia coii densities in the water (10/100 ml) associated with appreciable attack rates (about 10/1000 persons) for "highly credible"' gastrointestinal symptoms. Moreover, the ratio of the swimmer to nonswimmer symptom rates indicated that swimming in even marginally polluted marine bathing water is a significant route of transmission for the observed gastro enteritis. gastroenteritis; swimming; water microbiology In ah earlier report (1»., the authors presented evidence from a prospective epidemiologic-microbi'-'ogic study that there are measurable lealth effects as sociated with swimming in sewage: polluted waters. In some cases, these ef fects were observed even in waters that were in compliance with existing recrea tional water quality guidelines and stan- Received for publication October 16. 1980. and :r. final form October 7. 19H1 ' From th«j US Environmental Protection Agency. Health-Effects Research laboratory'. Marine. Field Station. West Kingston. Rl '•'Present address. Department of Microbiology University of Rhode Island. Kingston. Rl 02881 'Reprint requests to Dr. Cabelli.) 3 Present address. Health Effects Research Labo ratory. US Environmental Protection Agency. Cin cinnati. OH. * Present address. Office of Research and Devel opment. US Environmental Protection Agency. Washington. DC The authors thank the many individuals who con tributed to the conduct of the individual studies. dards (2). The swimming-associated ill ness observed was an acute, relatively benign gastroenteritis which had a short incubation period and duration. The ac companying symptoms, as pointed out in another report <3), suggested that tne etiologic agent might be the human rotavirusps or Norwalk-Hke viruses. The water-related nature of one of these agents, the Xorwalk-like virus, recently has been confirmed in a shellfish-asso ciated outhrpak nf pr?mrroPnrPriHs jn Australia of some 2000 cases (4•• and in several outbreaks associated with drink ing water (5?. The objective of the overall research program was to determine if there are illnesses associated, with swimming in sewage-polluted water and, if so, whether their rates can be quantitatively related to some measure of the quality of the ba thing water^ This question has been the^ subject of controversy since the 1950s 606 SWIMMING-ASSOCIATED GASTROENTERmS 607 when Stevenson (6) arid Moore (7) ob tained seemingly contradictory results. Therefore, studies similar to those re ported for New York City beaches (8, 9) were conducted at two other locations in the United States: Lake Pontchartrain, Louisiana and Boston, Massachusetts. The results obtained at these two sites were essentially the same as those found in the New York City study. This report describes the quantitative relationship of the swimming-associated gastroenteritis to the mean enterococcus density of the water as obtained from all the epidemiologic-microbiologic studies conducted in the United States. Materials and methods Study sites. Studies were conducted at three general locations: New York City (beaches on Coney Island and the Rocka- ways) in 1973-1975; Lake Pontchartrain. Louisiana (Levee Beach and Fontain- ble^u Beach) in 1977 and 1978; and Bos ton. Massachusetts (Revere Beach and Nahant Beach) in 1978. The beaches were chosen because they were near large met ropolitan areas and, hence, used by large numbers of individuals who swam on weekends but not during midweek days. This was an essential requirement of the experimental design for reasons to be given. The sources of pollution reaching the beaches in the New- York. City study were reasonably well defined as those emerg ing from the mouth of the Hudson River They were less defined in the Boston study, and least defined in the Lake Pontchartrain study. Moreover, in order to determine which, if any, symptoms or groups of symptons were both swimming- associated and pollution-related (a major objective of the first two years of the New York City study), a relatively un polluted beach at the fcockaways was' paired with a barely acceptaDle'byafrr on L'oney island. Tne latter was adjacent to a beach area classified by local au: thorities as unsafe for swimming. The paired beaches were also chosen so that they would have demographically similar populations. The results obtained, with reference to this objective of the New York City study have been published (1). Study de$\gp.. A prospective cohort de sign was used in all the studies. The es sential features of the design, which have been described previously (1, 9. 10), are as follows: 1. Discrete trials wprp pnr|ijrffr| opjv on Saturdays and Sundays. Potential par- ticipants were recruited at the beach, preif- erabiy as family groups. Trials were lim ited to weekend days to mayim^ tfift ™7* of the beachgoinfl population especially the portion that comes to the beach only nnwfiffl^pndfi. Rv excluding from the study those individuals who swam in the five midweek days before and after the weekend in question or at other locations on either weekend day, exposure to ba thing water was limited to that at the specific beach during a single day. or two days at the most. This decreased the con founding effect of beach-to-beach and day-to-day variability in pollution levels on the illness-pollution relationships sought. In addition, it allowed the anal yses of the data by trials <study days) or by groups of trials when the pollution levels as indicated by the mean indicator densities in the water were similar. 2. Demographic information was ob tained at the initial beach interview and during the' subsequent telephone follow- up survey. The information included age. sexT ethnicity and socioeconomic status, as determined from a persons-to-rooms ratio, 3. Information on bathing activity was obtained at the initial beach interview. Swimming was stringently defined as complete exposure of the head to the Wjitex This characteristic was deter- mined oy direct inquiry and by observa tion, i.e., whether or not the hair of the subject was wet. Individuals who did not SWIMMING-ASSOCIATEDGASTROENTERITIS 609 tinal symptoms and gastroenteritis will be used synonoraously. Water quality monitoring. \YaXer.sam- pies were collected periodically on trial ~days during the time of maximum swim ming activity at the beaches,. This was generally between the hours of 11 a.m. and 5 p.mUsually, three to four samples were collected at two or three sites from each beach at chest depth approximately four inches below the surface ol the water. Upon collection, the samples were, iced and delivered to the laboratory, where they were assayed within eight hours of collection. Potential water quality indi- cators that were examined are shown in table 2. Total coliform and fecal coliform densities «"»^ obtained using the most probable number or membrane filter Lynn's, asdescribed (ID. The densitiesof total coliforms and the component genera of that group (Escherichia. Klebsiella, ntrnhacter-Enterobacter) were also measured using the membrane filter pro- rpdure.for coliforms imC) of Dufour and Cabelli (12). Aftpr 1974. Escherichia coli densities wore determined by the mem brane filter method for thermotolerant. E. coli (mTECi (13), Ehterococcj (14), 5). Aeromonas Pwudomonas aeruginosa h\d~rophila (16). Clostridium perfringens__ •1T iand Vibrio parahemolyticus <18) were assayed-using membrane filter methods. .Vmry«'g Th^Plafiorpfnp of swim". nung-associated 'swimmer minus non- cwimmen gastrointestinal symptom rates to the mean indicator densities in tjje water wasexamined by regression analy- sis. Because the participants were re cruited at the beach on weekends and individuals who uere swimming in the midweeks before and after the one in cuestion were eliminated from the study. the symptom rates for a given weekend dav (trial) and the associated mean indi catordensity could have been analyzed as a point on the regression line. In fact, this was not possible with most.of the trials because the number of nonswimming par- Table 2 Potential water qualityindicators usedat the New York City, LakePontchartrain, Louisiana and Boston, Massachusetts beaches, 1973-1975 Indicators New York City Lake Pontchartrain Boston Enterococci Escherichia coli Klebsiella sp. Enterobacter sp. ' Citrobacter sp. Total conforms Clostridium perfrmgens Pseudomonas aeruginosa Fecal coliforms Aeromonas hydrophila Vibrio parahemolyticus • * shows that measurements were made for this indi cator at the specified location. ticipants wastoo small.Thisproblem was circumvented by grouping the trials. Single-day trials were arrayed according to increasing indicator densities. Groups were selected by utilizing "natural breaks" in the array; in this way. those trial days with similar indicator densities formed a group of data from which a geo metric mean density and the associated rates for gastrointestinal andhighly cred ible gastrointestinal symptoms could be calculated. This arbitrary grouping of trials was done for each of the indicator organisms. The attack rajies for gastrointestinal and highly credible gastrointestinal, symptoms were regressed against the mean indicator density. The log-linear regressh : equation _ . Y = a ••*- o l°g,ft 1) was used in which X was the mean indi cator density and Y the gastrointestinal symptom rate. Results Studies were conducted over several years at three locations, in the United States. The locations; beaches, study- years, follow-up percentages and number of usable responses are shown in table 3. 610 CABELtJ. DUFOUR. McCABE AND LEVIN Table 3 Location of beaches and the number of usable responses by beach and study year, 1973-1978 Location Beaches % Follow-up during study year respom 1 No. of usable sea during study year 1 2 3 2 3 New York City* Coney Island Rockaway8 82.3 78.3 86.6 82.9 78.3 641 681 3146 6491 4923 Lake Pontchartrain, LA* Levee Fontainbleau 77.2 77.9 _§ 3432 2768 551 Boston. MA+ ' Revere Nahant 81.2 81.2 1824 2229 Coney Island. 1973-1975; Rockaways. 1973-1974 • Levee. 1977-197S; Fontainbleau. 1978. t Revere, 1978: Nahant. 1978. § Included with Levee Beach. The degree of association of the mean indicator densities to swimming-asso ciated gastrointestinal symptoms in the three years of the New York City study was used to reduce the number of in dicators examined in subsequent studies. Thecorrelationcoefficientsfor thf> varinqs indicators obtained from these regression^ analyses are shown in table 4. It can be seen that enterococci was the best indi cator of those examined. Equally impor tant, fecal coliform densities, the basis for most federal and state guidelines and standards (2),correlated very poorly with swimming-associated gastrointestinal symptoms. The rates for total and highly credible gastrointestinal symptoms among swim mers and nonswimmers and the r.esiduais •swimmer minus nonswimmer rates) for the grouped trials are given in table 5. Also included are the corresponding means and ranges of the enterococcus den sities and the number of trials (days) in each group. Similar data for E. coli are given in table 6 by way of contrast. In a number of instances, the swimmer and nonswimmer rates were significantly dif ferent from each other. This was more frequent for residual rates associated with high enterococcus densities and with total gastrointestinal symptoms. The regression lines obtained from the data given in tables 5 and 6 are shown in figure 1 along with their correlation co efficients (r). In addition to having higher r values, the enterococcus regression lines, differ from those for E. coli in two other ways. The E. coli lines have shallower slopes and intercept the X axis at much lower densities. However, in the regres sion lines for both indicators, rather low densities are associated with appreciable attack rates. Attack rates for highly cred ible gastrointestinal symptoms of about _. Table 4 Correlation coefficients for totalgastrointestinal <GI- symptoms and the "highly credible" gastrointestinal •HCGI• />).-•*;.•••; against the mear. indicator densities for'siudieo jr .\W York Cih beech?*. 1973-1975 Indicator. Enterococci Escherichia coli Klebsiella Enterobacter-Citrobacter Total coliforms Clostridium perfringens* Pseudomonasaeruginosa Fecal coliforms Aeromonas hydrophila Vibrio parahemolyticus* * No data for 1973. Correlation coefficients <- HCGI GI No. of points 0.96 0.81 9 0.58 0.51 9 0.61 0.47 11 .0 64 0.54 13 0.65 0.46 11 0.01 -0.36 8 0.59 035 11 051 0.36 12 0.60 0.27 11 0.42 0.05 7 Su ary*"* z::^;::r;::;;;::,,::::;—'r -^™?*;*-*•—*- «•»—- -««-*•* ... _. ' '"'""""" "i ^'""ics on suimmiiiL'ussociatpd tfrtM/minta^;nr,i m„ «ti ,,.-« . • " " ' " '""••»»"«/"<»« 'on- relationships among swimmers, nonswim •imi.-n-.i i,,„ls ,njitudu>xjm swuiminga.ssuciated gastrointestinal illness. 1973 1978 Studv Meacli Kntriiii-iirni* (Irii.siU [110 ml M.-:iri Kjin>»c N«*w York City R<irk«wnvM ("miry Island l«»V;i! !i| J. :t« V I) I < . I 2 (i 2 7 Id .1(1 •r>0 I HI) l.itkc IVnt- I'tiurtrttin. I. A I*'VIS" 1«*'%(•«• PoiituinMciiu liUVvl- Buxton, MA Itl'Vrrp Nahnnt Revere I «♦•;■! IHVh IOVH IM '•». :".r i •ni. mi ii i I li: i i l:io :ill .1 3 C7 i; i.i ii :im /ii* MS ."II Mil .1.1 :io:i 12 Trails «ilay«) clustered 8 K ' 3 n 2 14 10 4 8 4 8 ft 4 ;i 2 1 No. «if swimmers 481 17.1 i:<9i or. i 02 ft h:ii 2232 .1806 57(1 87« 7211 89ft 12:10 24* 81)1 697 11.HI 222 */' ' 0.0ft; *> -. 0.01 • Study populiiliun lo» miiiuII I., ^,,..1,., iril,K |,v •uiiil/ir 1i11ln.it.,r densities. N... of : non- . Kwirrtmera c • Swimmers 107 107 711 111(10 410 410 93ft 678 101 ifti 4ft« 164 4)5 .103 •122 529 1009 376 81 72 27 .18 42 43 63 59 60 86 108 108 75 81 112 83 71 108 Symplon) rntc-H m caoea'1000 Total GI Non- swimmers 46 24 23 :h 17 2.1 55 37 31 61 50 54 34 63 60 66 67 74 Highly credible Gi Residuals Swimmers n' swimmers 35 48* 4 4 2.V 20 8 22* 20 3fi* 68** 54*» 41" 18 62" 17 4 34* 30.4 46 4 7 6 Ml ft Mid 18 I 18.8 14.8 34. ft 32 0 31 9 3ft 8 366 44 3 42 4 23.0 33 0 41.0 162 180 4.2 fl.9 2.4 19.3 7.4 111 8.8 8.6 14.5 23.1 1ft ft 11.0 28.0 13.0 Residuals 16 2 284 3 4 3.6 13.6 18.1* 06 7.4 346* 20.9* 23.1* 27.2" 22.1 • 21.2 28.9* 12.0 6.0 28.0* o> Taiii> 1? Summary ,if the mean /v. n-h tf,-n it\ t;tisin>uit- until <(!!) symptom nit,- relationships among swimmers, nnnswimmerit and residuals olitmnt'ti /mm rlit>ti '<•./ trials m studies on swimming-associated gastrointestinal illness, 197.1 197H Study Bern Ii Win M.MH Itiinuif clustered BW,mmern swimmers Swimmera Non- R,,BiduH|B Swimmers Non" Reaiduale o > New York City Itockiiwiivs 1;•V *> JIM .1 :i| H 484 197 81 48 36 30.4 16.2 16.2 Co r r* Coney Island Lake Pont- 1974 • ~. • "n'I"'.*" Z'T '*.* ' '. * — :;•:—".. ._«.r^^rr.:^rr.:'.v:..-t:.. ;•".:;:. 7~*r_. : ..... . »...«—»....-,.-. -., v 100 ml Kanxe Trials uiuys) clustered No »f swimmers No of non- swimmers Swimmers Symptom Total GI Non- Residu swimmers rates in.caBus/100 MiB» hIr Swimmers nv 0 ili-n-m M.m.i ly credible Gl Non" Residuals trimmers 21 M .1 .11 H 484 197 81 48 35 30.4 16.2 16.2 17 1 r.o yon 8 474 167 72 24 48* 46.4 18.0 284 2 '.'. 1 :» 6 2614 1641 25 34 •9 8.0 3.7 43 13 1 VI l«- 4 1304 104ft 38 29 9 14 1 5.7 8.4* :lu V :\u. .if. 2 600 42ft 6ft 33 32* 23 3 2.4 209 li. -. 22 Mil 10 194ft moo ftft ftl 4 13 4 178 -44 J 42 ii;". n;<i H • . 77ft 104 76 41 35 24 ft 10.3 14.2 :'.1H 2iim mm; M 1049 330 ftft 24 31* 21.0 3 0 18.0- Ml in n.'.'i 4 937 271 68 65 13 24 ft • 7.4 17.1 I—• to clinrlrain, LA Uvee 1077 4 1 33 ftl 2 372 222 132 46 87** 32 3 9.0 23.3 CO JJI 3 910 306 120 6ft - 55** ' 52 7 22 8 29.9* •4M7 • II;. rir.li 4 ft74 307 85 46 40* 32 8 130 19.8 :in!M in:li) 42 71' ft 419 204 88 83 5 31.0 4.9 26.1 KiintaiiiMcmi 1078 '1 M 1 2.1 8 248 303 81 63 18 44 3 23.1 212 1,4'VOf 32 i. 1/ H7 ft 1123 382 78 44 34* 38 3 20.9 17.4 0.1 V r>3 177 . 4 918 3S5 103 36 67** 392 8.5 30.7 Boston, MA I97H r. ft 4 7 2 541 874 72 63 9 39.0 29.0 10.0 / II ft 0 2 477 410 86 68 18 23 0 100 130 1 .' • i:i 22 •j ft80 22ft 70 67 3 27 0 27 0 0.0 211 :> 2H .11 2 442 •19ft 93 71 22 32.0 140 16.0 */i • 0.0ft; *> • 0.01 ' Sludy |Ki|iiilntiiin liio small M rluslrr lii.il. In mihiIhi iihIh ilnr densities. SWIMMING-ASSOCIATED GASTROENTERITIS 613 STUDY ce r- <n ir> .<• o z o rr or UJ o a. U. o uj o H- o < £E V. to O UJ r- 2 o < r- <L O *? <J </) en > to < _J o < 7 <L 2 r- i UJ 5 r- Z (9 O © o HCGI A A A A NEW NEW NEW LAKE LAKE YORK CITY YORK CITY YORK CITY PONTCHARTRAIN PONTCHARTRAIN YEAR 1973 1974 1975 1977 1978 1978 MEAN ENTEROCOCCUS DENSITY PER 100 ml 0' I02 MEAN. DENSITY r=25 £ co1' PER 100ml Figvri 1. Repression of swimming-associated 'swimmer minus nonswimmer) rates for gastrointestinal Gl)svmptomsonthe mean enterococcus and E. coh densitiesir the water.Coordinatepoints are fromtables 5 and 6. Correlation coefficients in are as given. HCGI. high; credible gastrointestinal. 10 1000 (1 per cent) are associated with enterococcus densities ofabout 10/100 ml. The enterococcus regression lines, their formulae, the r and p values and 95 per cent confidence limits for trip linps are <hown in figure 2. These relationships "predict the illness rates from the mean rnterococcus densities.. The relative importance ol" swimming in sewage-polluted water as a route of transmission for enteric illness was de^ termined by examining the ratio of swimmer to nonswimmer gastroenteritis rates against the mean enterococcus den- ^irji. It was assumed that all the cases ac quired by all the routes other than swim ming in sewage-pollu.ted waters were in cluded in the nonswimmer rates. The regression lines obtained for the trials clustered by indicator densities are shown, in figure 3. It can be seen that the rates for both total and highly credible gastro intestinal symptoms were equal at a mean enterococcus density of about 1100 ml. At a level of 10 100 ml. the rates for total and highly credible gastrointestinal symp toms were 1.5 times for swimmers and rwict- ihose for nonswimmers. respec tively. The higher ratios for highly credi ble than for total gastrointestinal symp toms are of interest because of their im plications concerning the reliability of the respondents" information to -the illness queries. Discussion The results clearly show that the risk of gastroenteritis associated with swimming in marine waters impacted with munici pal wastewaters is related to the quality of the water as indexed by the mean en terococcus density in the water. More- 614 CABELLI. DUFOUR. McCABE AND LEVIN W70 Z o v> rr £60 *- O % ^ o|40 o ? 30 O </) V) en _, «g2o O H i 2 .0 5 o REGRESSION LINE (Y ON X) 95% C L AROUND LINE TOTAL Gl SYMPTOMS HIGHLY CREDIBLE Gi SYMPTOMS » *th Mil —I—I I I Mill i i i li.nl 10' I02 I03 MEAN ENTEROCOCCUS DENSITY / 100 ml Ficl-re 2. Regression of swimming-associated gastrointestinal (GI» symptom rates on the mean en terococcus densities in the water. Data are fromall US studies!The 95 percent confidence limits iCL« forthe' lines areas shown The slopes,intercepts, r andp values for Gl symptoms are. respectively. 24.2. -5.1. 0.82 and '<6.i»l. For highly credible GI symptoms, they are. respectively. 12.2. 0.2. 0.75 and - 0.00L. over, the risk is detectable at extremely low levels of pollution. According to the criteria suggested by Hill (19), there is a strong suggestion of causality. _£irjst. the association is a good one; in some trials, the swimming-associated gas troenteritis rate was three to four times greater than the nonswimming rate. second, t) r-re was a consistency in the association in that it was observed ait multiple locations over multiple years. Thjrrd. the association between enteric disease and fecal contamination is a rea sonable one bv its very, nature. Fou^tn, the association is a coherent one since there is a precedent for such a relation- ship hv other waterborne routes of trans mission i.e.. in shellfish (20) and potable water 121). It was also understandable that, of the indicators examined, enterococcus den sities in the water correlate^ best with the rates for the swimming-associated gastroenteritis. The two salient indicator characteristics required for the specific association obtained are a consistt nt fecal source and "good" survival during sewage treatment and transport in the aquatic environment. Of the indicators examinpri enterococci and E. coli hpst satisfy thP first requirement (22. 23); and, of the two. enterococci have the best survival charac teristics (24>. although their densities in raw or treated sewage are 1-2 orders of magnitude less than those off, coli i2oT These two < ffererices are consistent with those observed in the slopes andX axis in tercepts of the regression lines for the two indicators. That is, the slopes of the re gression lines should become shallower and the lines should, cross the X axis at lower indicator densities as the survival SWIMMING-ASSOCIATED GASTROENTERITIS 615 CO £• n • > CO UJ rr A 6.5 A O TOTAL Gl . A HIGHLY CREDIBLE Gl 5 £r- V) CO z o z (XL UJ 2 CO UJ r- Z o QC r- CO < P rr o u. 3 - I <* --0^g£ i i i linn in mm i i Mini o° io1 i.62 io3 MEAN ENTEROCOCCUS DENSITY/100ml Figure 3. The relative risk of swimming in sewage-polluted waters as shown by the regression of the ratio of swimmer to nonswimmer (background' rates for gastrointestinal <GI> symptoms on the mean en terococcus density in the water. Data are from table 5 characteristics of the indicator become poorer, relative, to those of the etiologic a'gentts-. ThprP arP two implications frorn thp finding ofrather high gastroenteritis.fates 1 per ppnrrajwwjgtpH with the ingestion or one to five enterococci (the accidental ingestion of 10- 15 ml of water <261 whose enterococcus density was about 10 100 ml'. The- first is that even enterococci may not survive as well.as the etiologic Tgent for ihe gastroenteritis. The second ..- ;hai tht- agent must be present m the nathing waters and, hence, municipal wastewaters in very large numoers. pe Highly, infectious, survive very well in ihe marine environment or. most prob ably, a combination ofall Ijhrpp, The analysis of the ratios of the swim mer to nonswimmer gastroenteritis rates would suggest that, for individuals of "swimming age," swimming in sewage- polluted waters is not an insignificant, route of transmission • r the disease. Moreover, the risk of gastroenteritis is present even at relatively low pollution levels, as seen from the indicator den sities. The higher ratio, of the swimmer to nonswimmer rates observed with highly credible as opposed to total gas trointestinal symptoms suggests that. nausea, stomachache and even diarrhea are disproportionately reported bv non- swimmers,. This, in turn, suggests that the swimming-associated rates for total gastrointestinal symptoms are under estimated Finally, the finriimr of swimming- related rat*-s of gastroenteritis "associated with very low indicator densities, i.e.. the inge.stion of one to \")ve enterococci. has some interesting implications with regard to the existence of sporadic cases of this illness by the other potential water- associated routes of transmission, e.g., shellfish, drinking water and even aero sols generated, from municipal sewage and its receiving waters. These possibil ities should be pursued by prospective epidemiologic investigations. 616 CABELLI. DUFOUR. McCABE AND LEVIN References 1. Cabeili VJ. Dufour AP, Levin MA. et al. Re lationship of microbial indicators to health ef fect? at marine bathing beaches. Am J Public Health 1979:69 690-6 2. US Environmental Protection Agency. Quality criteria for water. Washington. DC: USEPA 1976 42-50. 3 Cabelli VJ. Evaluation of recreational water quality, in: James S. Evison L. eds. Biological indicators of water quality. London Wilev. 1979 14-1 to 14-22. 4 Murphy AM. Grohmann GS. Christopher PJ. et al. An Australia-wide outbreak of gastroen teritis from ovsters caused bv Norwalk virus . Med J Aust 1979:2:329-33, 5. Halfy CE. Gunn RA. Hughes JM, et al. Out breaks of waterborne disease in the United Stares. 1978. J Infect Dis 1980:141:794-7. 6. Stevenson AH. Studies of bathing water quaiv and health. J Am Public Health Assoc 1953.43:529-36. 7. Moore B. Sewage contamination of coastal ba thing waters in England and Wales: a bac teriological and epidemiological studv. J Hvg 1959:57:435-72. 5 Anderson AC. Ktsanes VK. Diem JE, et al. Bacterial indicators and health effects among swimmers. Abstracts of the Annual Meeting of the American Soctetv for Microbiology. 1979 222 9. Cabelli VJ: Health t-;Tects criUr: a for marine recreational waters. Cincinnati OH: HERL. United States Environmental Protection Agency. 1980 EPA-600'l-80-031>. 10. Cabelli VJ. Levin MA. Dufour AP. et al. The development of criteria for recreation waters. :In: (Jameson H.. ed. International symposium.on distharee of sewage from sea outfalls. London. Perjamon. 1975:63-73 i Air.v::.an Public Hca:f.h Association. >iar.a.srd •»»e:r.' d? f.r ih*« examination of'w-ater and' rtjf'cM'd'.r:. 14th ed Washington.-DC Arr.'.-n- • cur. P:h!ic Heahh Association. 1976. 1- D>>:- \:r AP. Cabc-lli. VJ.' Membrane filter pn,it~ •r.;.rf •-r rnuir.vrn'.ini.Mi: component c.-nera ! 'f- i"t/«r-rf! gr-i.p u. -«.-;i-.v .iter Appl N!;ct' r•••; ! - !)•.::..," AP. Strickiar.c ER. Cabelli V.J M,-. hrar.c Hirer r.n-thi.d for enumerating £?• fhcr.hia -.••h. Appl Environ Microbiol 19M: 41.T152-8. 14. Levin MA. Fischer JR. Cabelli VJ! Membrane filter technique for enumeration of enterococci in marine waters. Appl Microbiol 1975:30:66-71. 15 Levin MA. Cabelli VJ. Membrane filter tech nique for enumeration of Pseudomonas aeru ginosa. Appl Microbiol 1972:24:864-70. . 16. Rippey, SR, Cabelli VJ. Membrane filter proce dure for enumeration ofAeromonas hydrophila. '.. Appl Environ Microbiol 1979:38:108-13. 17. Bisson JW. Cabelli VJ. Membrane filter enu meration method for Clostridium perfnngens. Appl Environ Microbiol 1979:37r55-66 IS Watkins WD. Thomas CD. Cabelli VJ. Mem brane filter procedure for.enumeration of Vibrio pnrnhaemoiyticus. Appl Environ Microbiol ;97r-:32:679-S4.. 19. Hill AB The environment and disease, associa tion or causation? Proc R Soc Med 1965:58:295- 300. 20 Verber JL. Shellfish borne disease outbreaks Internal report. Davisville. Rl- Northeast Tech nical Service Unit. USFDA. 1972. 21. McCabe LJ. Epidemiological consideration in the application of indicator bacteria in Nonh America. In: Hoadley AW, Dutka BJ. eds. Bacterial indicators.health.hazards associated with water. Philadelphia: American Society for Testing &. Materials. 1976:15-22. 22. Dufour AP. E,cok: the fecal coliform. In: Hoad- ic> AW. Dutka BJ. eds. Bac.prial indicators, health hazards associated with water. Philadel phia: American Society for Testing it Materials. 1976:4*.5S. 23 Cabelli VJ. Indicators of recreational water quality. In: Hoadley AW.. Dutka BJ. eds. Bacte rial indicators,health hazards associated- with water. Philadelphia- American Society for Test-, ir.g & Materials. 1976:222-3* _'4. H;::v s NB Fraeai.i R. Effect . f ^awater con- i. t! i.l :r.d:..atr,: bactcr:^. J •'1 Fid \V*~ A9 <J7- i04. ii V.J. Enu rococci and other •>rs in nr'jrncipai «ewa«;i' P.illut C.:,':/..-,•..i Fed lli*^ ;n -•vi.. :t .- \{.\ Bavt-rv.ii uoiii ;\ •ih)i"-«::v»'.< i'<>r i'r.^ ' >]•:: . River A iju:'l'e lor T.< '.'valuation of-sani tary londition ..-.♦' waters used lor citable Mipphes and recreational uses. Cincinnati. Ohio, uhio River Valle*. Sanitation v. ummissiun. jy51. i'iv.1 jtior. un sur xVrtV-r P.i|! .it <Y.r.: Ui.-scu-r J- . Cab.. '...< r'lhi.'ii iniiici -T.ueM* •! Water ArruniMtranftKviiinv.'July \97tt. p W-71 -i'tqtyrichl C I97.V Ani<-n««n S.« »-l\ ft* Mw-r>J>i<4i^v Vd. an. No. i Membrane Filter Technique for Enumeration of Enterococci in Marine Waters M. A. LEVIN.' J. R. FISCHER, and V. 4. CABELL! Notional Marine Water Quality Laboratory, West Kingston, Rhode Island (12892 Received for publication 12 February 1975 A memhrane filter procedure i* described for Ihe enumeration of enteron>cci in marine waters. The procedure utilizes a highly selective and somewhat differen tia) primary isolation medium followed by an in situ substrate test for identifying colonies of those organisms capable of hydrolyzing esculin. The procedure (mE) was evaluated with known streptococci strains and field samples with repard to its accuracy, sensitivity, selectivity,-specificity, precision, and comparability to existing methods. EssentialK quantitative recovery was obtained with seawater- stressed cells of Strcptoa>ciu< faecali*- and S.faecium Neither .S bo: is, S. equmus. S. milis, nor 5. salnarms grew on the medium The selectivit\ of the medium was such that a N'.OOO-fold reduction in backcround organisms was obtained relative to a medium which contained no inhibitors and wa« incubated at 3n C. About 90f~ of those typical colonies designated a> enterococci confirmed as such and about 12^ of the colonies not so designated were, in fact, identiiied as Enterococci. Plate to plate variability across samples approximated that ex pected by chance alone. Verified recoveries of enterococci from natural samples by the mE procedure, on the average, exceeded those by the KF method by one order of magnitude. Shortly after J: P. Laws and F. W. Andrewes first reported streptococci from the gastrointesti nal tract. Houston noted that these organisms t ppeared to be characteristic of sewage and ani- tial fecal wastes. He suggested that they were ndicative of dangerous pollution because they are readily demonstrable in recently polluted waters and seemingly absent from waters above suspicion of contamination (18). Since that time, a volume of data has been accumulated categorically demonstrating that, fecal strepto cocci are associated with the fecal wastes of man and lower animals and that they can be isolated from polluted water.containing such wastes (14. 20). Nevertheless, this group of organism? has hot been generally accepted as an indicator of fecal contamination for at least two reasons First, coliforms and coliform biotypes have been a more attractive means of identifying fecal contamination because early workers found them easier to quantily and they are present in larger numbers in feces, sewage, and polluted waters. Second, there has been a good deal of confusion concerning the identity of the fecal streptococci, particularly as it relates to their ecological distribution. This is reflected in the fact that those streptococci which cnn he found in the feces of man and other warm-blooded animals have been referred to variously as enterococci,. fecal streptococci and. more re cently, group D streptococci. The composition of these three group- is shown in Fig. 1. It can be seen that two organisms not associated with humans. Streptococcus bocis and S. equinus. and two organisms associated with humans, but not exclusively with fecal wastes, S. mitis and S. salivarius, are included, in one or more of the groups. Ideally, the indicator should be limited lo the fewest number of species or biotypes which are most closelv or exclusively associated wit h the fecal wastes of man. i.e.. S. faecalis and S. faecium. Facile methods, to do this have not been available Hence, broader groups of strep tococci have been used at times, because thes* were t he biotypes recovered by the procedures available. The taxonomy and distribution of this group of organisms has been reviewed by Hartmanet al. (8), Kjellander (11). and revised recently in Ber/iey's Manual for Determinative Bacteriol ogy, 8th ed. (21. Enrly attempts to quantify fecal streptococci relied on enrichment tube procedures associ ated with the use of the most probable number method. In 1940 Mallmann (15) suggested the use of azide lactose broth. This was later $6 Vol. 30. 1976 FECAL STREPTOCOCC E r N S. foecolis T • E S. foecium n o1- C ! group O streptococci o' er C S- bovis S: rntfis £. sdlivortus G R 0 U P D S T R E P T 0 c 0 c c F«s. 1. Distribution of streptococci species omonf rtrre groups. This study was completed prior to the release of the 8th edition of Bergey's Manual for Determinative Bacteriology (2), which hot classified S. rymogenet as S. foecalis. subspecies 2ymogenes «nrf S. durans as equivalent to S. foecium. modified by Mallmann and Seligmann (16) osing Roth's basic azide dextrose medium, followed by a confirmation in ethyl violet azide fcroth as described by Litsky et al. (13). This procedure has been evaluated in marine waters 13). Hajna and Perry (7) developed the SF medium; and Winter and Sandholzer (23) de scribed presumptive and confirmatory media, winch use sodium azide and/or high concentra tions of NaCI as inhibitors. Tbe quantification of enterococci by a mem brane filter procedure was first reported by Slanetz and Bartley in 1957 (21): and. in 1961. Kenner et al. (10) described the KF method for the enumeration of fec81 streptococci. In 1966 hrnberget al. (9) reported a plating procedure IPSE) for the quantification of group D strepto cocci. This medium offered increased recoveries and easier differentiation based on the use of tfce esculin reaction. The recoveries by these and other procedures have been compared by a somber of workers (19. 22). Impetus for the present investigation devel- MEMBRANE FILTRATION TECHNIQUE €7 oped from a comparison of the KF methods and PSE medium (used in conjunction with mem brane filtration) for the isolation of fecal strep tococci from marine waters at beaches in the vicinity of New York City. In this preliminary investigation, it was observed that the con firmed recoveries of fecal streptococci by the modified PSE procedure exceeded those by the KF method by about one order of magnitude. However, the modified PSE procedure as used was deficient in that overgrowth by background microorganisms was a serious problem. The present report describes the evaluation of a procedure for the enumeration of enterococci designed to obviate the problems noted above. In addition, an attempt was made to make the method more specific, i.e.. to eliminate the recovery of the viridans group (S. mttis. S. soharius) and the two organisms exclusively associated with animal feces IS. equinus and S. bovis). Tbe method (mE) was.eyaluated against the following criteria: (i) accuracy, recovery of at least 75^- of the viable S. foecalis and S. foecium cells following a stress imposed by exposure to sea water for 48 h at 4 C; (ii) selectivity, the reduction of background orga nisms in naturally polluted waters by at least three orders of magnitude (1.000-fold). (Hi) specificity, colonies designated as positive should verify as such at least 75**. of the time, and no more than lO^i of those designated as negative should verify as enterococci: <iv) preci sion, with, field samples, the D* (4) value, distribution approximates that estimated by chance; and (v) comparability, the accuracy and sensitivity of the method be equal to or greater than existing membrane filter methods. MATERIALS AND METHODS Cultures »nd field samples. The recovery and accuracy studies were performed with cultures of S. faecalis, S. foecium, S. bovis-, S. equinus, S. mitis, and S. salivorius provided by R. FacklBm (Center for Disease Control. Atlanta. Ga.). and with a strain of 5. foecoltt isolated from New York Habor Suspensions of these organisms were prepared from brain-heart infusion broth (BH1, Difco) cultures incubated at 37 C for 20 h. After incubation, the organisms were washed three times in sterile phosphate-buffered saline <NaH,PO.. OiSBg: N«,HPO,.0:25g; NaCI. 0.95 g: distilled water. 100 ml). Aliquots of the.resultant suspensions were delivered into flasks of sea water passed through a 0.2-nm membrane filter. These were held at 4 C and sampled .periodically by the mE method and a control procedure (spread plates on BHI agar) to determine.the number of recoverable organisms. Field samples were collected from marine and 68 LEVIN. FISCHER. AND CABELLI ettuarine waters in sierile container* held at 4 to 6 C and assayed within 12 h of collection. Membrane filtration. Appropriate volume* of the test suspensions or water samples used in evaluating the experimental medium were passed through mem brane filters (47-mm diameter. 0.45-nm pore size). When the portion of the water sample to be filtered was lev than 20 ml. it was brought to at least that volume wilh phosphate-buffered saline to wash reside ual organisms onto the membrane. The glass filter holders were sterilized for 2 min in an ultraviolet sterilizing apparatus (Millipore Filter Corp.). The membrane fillers were obtained presterilized from the manufacturer Recovery media: mE. The formulae and methods of preparation of mE medium and the esculin-iron agar substrate used in the in situ esculirr test are given in Table 1. After a basa! medium containing peptone, sodium chloride.'yeast extract, esrulm. and ferric ammonium citrate was shown to quantitatively recover the organisms, a search (or appropriate inhibi tors was initialed. A number of candidates were screened.for this purpose initially by using a modifica tion of the gradient plate method of Szybalski as described by Levin and Cabelli (12). Nalidixic acid and sodium azide are used to inhibit gram-negative organisms and artidione to inhibit fungi. Triphenvl tetrarolium chloride in the concentration used colors tbe colonies, differentiates enterococci from other streptococci based upon its reduction, and has.a slight inhibitory effect on some background microorga nisms Esculin is included to induce the enzyme catalyzing iis hydrolysis. The mE plates are incu bated for. 4? h at 41 C: the elevated incubation temperature also inhibits some of the indigenous microbial f1"ra In situ esculin substrate. The hydrolysis of esculin is used in the characterization of enterococci. Ini tially, the esculin indicator system was included in Tabu: 1. Preparation of mSD medium and ElA substrate. mSI) E1A Ingredients g1nei lnrrfdients* g/lim Agar 15.0 Agar 15 Peptone. 10.0 Esculin 10 NaCi 15.0 Ferric citrate 0-5 Esculin 1.0 Distilled water 1.000 ml Yeast extract 30.0 Aclidione 0.050 Sodium azide 0.150 Distilled water 1.000 ml •Autoclave at 121 C for 15 min. After autoclaving. add nalidixic acid. 0.24 g. and triphenvl tetrazolium chloride: 0.1.5 g. Adjust pH to 7.1 * 0,1 and pour in 3.5-ml amounts to50-mm membrane filler plates. 'Adjust pH to 7.1 d 0.1 before autoclaving at 121 C for 15 min. Pour in 3.5-ml amounts to 50-mm mem brane filter plates. Am. MtcaoBOL the primary medium. This resulted in dark nt colonies approximately 2 mm in diameter with black halos in the medium resulting from the reaction e? cumarin with the ferric chloride. When more thanJP colonies were present, however, the zones coalesced making it impossible to determine which of thr colonies was positive. This problem was overcomeU the use of an in situ test in which, after incubation. tbe membrane is transferred to an csculin-iron sis? ptete (Table 1). After 20 to 30 min at 4) C. sms! black spots appear under the positive colonies, pet milting enumeration of at least 80 enterococcus eeu> nies per platewithout problems of coalescence. Control media. KF (Difco) and PSE (Pfiwrl media were prepared and used following instruction from the manufacturers and Standard Methods fr the Examination of Water and Wastewater (1). A» noted earlier, the PSE medium »«« used in a mer* brane filter procedure. • Verification of colonies. Verification of colonie?* enterococci was accomplished b> using thr bik- esculin medium of Schwan in combination with* modification of the procedure of Facklam and Moah (5). i.e.. (i) growth al 45 C in BHI broth:, (ii)« negative catalase test: fiitl' esculin hydrolysis: |rr» growth on 40** bile-blood agar.• (v) an acid reactions . litmus milk: and (vi) a positive Gram stain. RESULTS The accuracy of the mE method was deter mined by comparing the recoveries obtained by this procedure to those, observed when eight species of streptococci were spread plated n. BHI agar. The suspensions, whose initialed! densities varied between 10* to 10* per ml. wt prepared in filtered estuarine water and held at. 5C for periods up to 9 days. The averagr relative recovery of S. faecali* and S. foecium over the 9-day period was 102*5-; S. boots, S. equinus, S. mitis, and S. salivarius recoveris were 0.0001*5 or less (Table 2). The selectivity of the mE method was suck that a 10.000-fold reduction in background organisms was obtained relative to that ob served when the inhibitors were omitted frea mE medium and tbe plates were incubated* 35 C- This 99.99<Tr reduction was obtained witk samples whose initial background densiuo were 10* to 10* cells per 100 mi The specificity of the mE procedure «» examined by determining (i) the percentage«f typical colonies which did not verify as mem bers of the enterococci group (false positives! and (ii) the percentage of other colonies, thaar which did not possess the typical colonial char acteristics which, in fact, were enterococci (fake negatives). The 2.231 colonies examined wen isolates from polluted marine and estuariar water samples collected at six locations aloof the east coat of the United States. Most of tk Vol 30. 1975 Tabu 2. Therecovery of stressed* fecal sirrptorocci on mSD medium Organism 1 Rtco\ery' ofm exposure <d»ys> 0 1. .1 6 9 S.forcalis 107 ND' 94 123 87 S foecium 102 ND 126 73 109 S. foecium' 114 ND' 108 46 100 S.foecolis 117 ND 118 ND 113 • subsp lymogenes S boiis 0.001 <0.0001 ND ND ND S. equinus < 0.0001- ND ND ND ND S. solivariu* < 0.000) ND ND ND ND 5. mitis <0.0U0! ND ND ND ND •Suspensions heid in seawater at 5 C for number of ibys as indicaied 'Relative to recoverv on BHI spread plates 'Obtained in 1973 from R. Fackiarn as S. durans. •No data isolates came from samples collected at beaches is the vicinity of New York City. In general, all tbe colonies on a given plate were examined. Ninetypercent of the typical colonies and 11 7^ <af the other colonies were enterococci. (Table 3). Neither.S. 6oi't$, S. equinus, S. mitts, nor S. saliorius were encountered. The precision of the mE method was deter mined from D' values for assay variability calculated from the following equation as given k* Eisenhart and Wilson (4): D' - A'IA'1: - CXilVIXi. where ZXi is the summation of tbe plate counts A,. X, X„, and A' (the somber of replicate plates per sample) was 5. The D* values calculated from the examination :• «f 15 polluted, marine, and estuarine water cunples are displayed in Fig 2 along with the expected D* control limits for P = 0.005. 0.025 and 0.5. In the event of excessive variability aziong the five replicate determinations rotates), hence. poor reliability of a sinple termination, the observed D: values should brre exceeded the control limits more Ire qortith than expected by chance alone By dbaace alone, one D7 value in 40 would have k»w expected to exceed the P'« 0.005 limit Bwrever. it can be seen that none of the values esreed the limits, and that they are evenly Attributed around the P = 0.5 limit (Fig. 2). Sixteen Water samples collected from a vari- «* of sources were assayed in parallel by the KF. PSE. and mE methods. Typical colonies as described for the various procedures were veri fied as stated in Materials and Methods. Thus, tie data presented are derived from verified Rcoveries. As can be seen from Table 4. neither Mfi-<,\f •,\ ) I';:-: 69 the rSK nnr (hr Kf recc\erics approached those oblained uilh the rnF. method, although those by the PSE and KF methods did approxi mate each other. The average number of colo nies on the filters in these 13 trials was 70. ranging from 9 to 110. In a second, more extensive set of trials, fecal streptococcus densi ties obtained by the KF method were compared Tabije 3. Specificity of the mSD procedure for the enumeration of group D streptococci No of colonies a [>d ** verification Genera! sample Typical colonie*. Other colonies- location No. "• Faise No 1r' Pais* examined rx*:!i\f examined inegstue | Cone\ Island 1.22-' 11 107 j 1? NY. RusPark 577 10 104 11 NY 1 Miami. Flo 16 U' 16 0« Boston Harb.. 59 3.0 ND' ND Mass. Connecticut 67 4.0 20 0* Rhode Island 30 0* 10 0- Overall 1.974 10.0 257 11.7 •No false negative or positive colonies *N(>-data 0* VALUES FOR «SD AGAR 10 P«.0O5 P-.025 a .\ p.oj t < LO > ftjii i i i i i i i i i i i i i i 5 10 15 SAMPLE NUMBER Fir. 2. Precision of the msD procedure for en terococci os extended from dispersion of Dl i-olues. Symbols (91 D* value calculated from /ire replicate plates at each point: •--•: control limits when proba bility is as stated. 70 LEVIN. FISCHKK.' AND CA'RKI.M Tablc 4. Companion of verified recoveries of group D streptococci by the mSD. PSE ond KF mrthints • n mSD Location Trial recovery recoveries* by plate) PSE KF Providence River 1 45 1 i 82 62 23 70 74 3 90 68 58 4 55 78 94 5 70 49 73 Goneylsland 1 110 59 61 2 82 60 77 . 3 21 45 • 67 Boston Harbor 1 89 8* 75 2 97 59 70 3 73 47 73 Miami 1 9 ND' 66 2 107 ND' 33 Stoningion 1 . 9 39 j 75 . Harbor ! Ne* London 1 61 j 25 38 Harbor i Mvstic Harbor 1 85 j 6* 9 *•»' 70 60 63 •Values obtained from the average of triplicate plate counts. •No data'because the PSE plates were overgrown with background organisms directly to those bv t he mE procedure (Table 5) In only three instances did the KF recoveries exceed the mE recoveries: on the average, the mE recoveries were about 10 times greater than those by the KF procedure. The samples re ported in Table 4 were obtained during the winter months and those reported in Table 5 were obtained in July and August. DISCUSSION The mE procedure has satisfied most of the predetermined criteria for a primarv. selective differential method for the enumeration of en terococci in marine waters. The onl\ exception was the rate of false-negative colonies Essentially quantitative recovery of 5. foecalis and S foecium was obtained with unstressed and stressed cells. Although this approach is basically artificial in that the work was done with pure cultures rather than natural samples, it is not logistically feasible to perform such experiments with natural samples 5. bovis and S. equinus were not recovered in significant numbers of mE-medium, nor were they isolated from natural snmples". Therefore, it may be assumed that these two species, whose origin is animal feces, are not included in Am.. Mkxosml densities of enterococci as obtained by the mE procedure. Relative lo other methods, this in creases the specificity of (he mE procedure for enterococci of human origin. However. S. faecalts and S. faecium do occur in animal fecn (2). In addition, biotypes of 5. foecalis hav* been reported (4. 15) as being associated with vegetation unpolluted with human fecal wastes and with insects. The former are differentiated by their ability to hydrolyze starch, a character istic which could be performedas a subsequent in situ test. The selectivity criterion for the evaluation of mE methods assumes that the marine recres tiona) waters to be examined for enterococri generally would contain less than 10' back ground organisms per ml (those bacteria which grow in 48 h on the mE medium when the inhibitors are omitted and the plates are incu bated at 35.C). The required 1.000-fold (99.9*!) reduction in the levelof these organisms, to be achieved by the combination of inhibitors and Table 5 Comparison of verified recoveries offeed streptococci and other organisms by the mSD andKf methods Recovery per 100ml by Trial rroup and method Location . Feral Other Mreptocorn ' tolonie> mSD KF mSD KF Conev Island. 1 38 <1 54 13 NY 2 38 ND- 35 TN» 3 270 6 190 <1 ' 4 660 ND TN TN 5 670 ND 240 TN 6 310 20 430 <10 7 . 70 4 123 <1 8 7 1 4 <1 9 70 4 70 <l 10 1.510 6 310 <1 Rockawav*'. 1 17i» 12 SMO <1 N.\ 2 230. 17 1.090 <1 3 44i'i . 8 TN <1 4 51 5 450 <r 5 25 61 • 56 <10 6 49 62 24 <10 7 31 2 <1 <1 . 8 16 8 127 <I 9 32 5 100 <1 10 47 2 53 <l 11 36 8 2 <r 12 12 57 89 <i "N«» data, confluent growth of background on* ni*ms •Too numerous to count Vol. 30. 1975 the elevated, incubation temperature, would result in 50 colonies per filter when a 50-ml water sample is assayed. The diflerential char acteristics of the medium then should permit the detection of a single group enterococcus colony, on such a filter. The actual reduction OD-W'*) exceeded this requirement, thereby permitting the detection of group D streptococci in the presence of 600.000 background orga nisms deposited on the filter. Although the rate of false-negative colonies (11.7^) slightly exceeded the specificity crite rion (10%). the false-positive rate (10'V.I was markedly less than the specified limit.'(255). Therefore, it may be assumed that verification of a number of typical and' other colonies is necessary only when an operator is being trained. In the absence of verification, estimate? of enterococci densities should be designated as presumptive. After verification, the estimates would be considered as confirmed. The results demonstrate (fig. 2» the precision (reproducibility) of the mE procedure The plate to plate variability over the samples examined was that expected by chance alone; that is, the D1estimates distributed equally on both sides of the P - 0.5 limit of 3.2 and none exceeded the P = 0.025 limit. The recovery efficiency of the mE method - comparable to. or better than, that by the K method for the examination of enterococci in estuarine waters: in addition; it provides higher confirmed recoveries. The difference between recovery ratios (rnE/KF) observed at Coney Island in the summmer months (1.5: Table 4) and winter months (54: Table 5) may be a function of a seasonal variation (water tempera ture, rainfali). As noted previously, the method measures 8 more specific portion of the fecal streptococcus population and one that appears to be a close association with the fecal wastes of humans. LITERATURE CITED 1 America*; Publir Health Association 1P"1 Standard methods for the examination of water and wastewater. 13th ed p 719. American Publir Health Association. New York 2 Buchanan. R. E. and N E Gihhon*. 19*4 Ber*e>'». manual of determinate haneriolofv 8th ed . p 490-509 Williams and Wilkins Co 3. Buck. J. D. 1972. Selective detection of enterococci in marine water* Am J. Puhlit Health 62:419--121 A. Eisenhart. C. and l». W. Wilvm, ISM.t StaliMiral MEMBRANE RLTRATION TECHNIQUE 71 method* and mntnd in bacteriulnsv. Bacterinl. Rev. 7:57-137. 5. Facklam. R. R.. and M. D. Moody. 1970. Presumptive identification of jrroop D streptococci: the bile-esculin test. Appl. Microbiol. 20:245-2V> 6. Geldreich, E.. and B. A. Kenner. 1969. Conceptsof fecal streptococci in stream pollution J. Water Pollut. Con trol Fed. Res 4USuppl.):R336-362 7. Hajna. A. A., and C. A. Hero. 1943. Comparative study of presumptive and confirmatory media for bacteria of the cotifarm rroup and for fecal streptococci. Am. J. Public Health i3(5):.SAO-5.S6. fi. Han roan. P. A.. G W. Reinbold. and D. S. Saraswat. 1966 Indicator organisms—a rrview. 1. Taxonomy of the fecal streptococci. Int. J. Svst. Bacteriol. 16:197-221. 9. Isenberc. H. D.. D. Goldberg, and J. Sampson 1970 Laboratory studies with a selective enterococcus me dium Appl. Microbiol 20:433-4.16 10 Kenner. R A . H F. Clark, and V W Kabler. 1960 Feca! streptococci 1 Cultivation and enumeration of strepto cocci in surface waters Appl Microbiol. 9:1 5-2<» 11. Kiellander. J. 1960. Enter<* streptococci as indicators of fecal contamination o( water. Aria Pathol. Microbiol Scnnd 48iSuDpl.i:13£ 1124 12 Levin. M. A . and VJ, Cabelli 1972 Membrane filter technique for enumeration of Pteudomonat aeruginosa. Appl Microbiol. 24:864-870. 13. Utsky. W.. W. L. Mallmann. and C W. Fifteld 1953. A . new medium for the detection of enterococci in water. • Am J Public Health 43:873-879 14. Litsky. W.. W. L. Mallmann. and C W. Fifrtld 1955. Comparison of the most probable numbers of Esche richia coli and enterococci in river waters Am J. Public Health 45:1049-1053. 15 Mallmann. \V. L. 1940 A new yardstick for measuring sew ace pollution. Sewaee Work* J. 12:875-P7s. 16 Mallmann. W. L. and E. B Sehgmann. Jr 1950. A comparative study of media for the detection of strep tococci in water and sewage. Am, J. Public Health 40:2*6-289 17. Papavassilous J. 1962. Species differentiation ofgroupD streptococci. Appl.Microbiol. 10:65-69. 18. Prescott S. C. A, C. Window, and M. H. Macrady. 1946. Water bacteriology, p: 207-212. John Wiley and Sons. New York 19. Sabbaj. J.. V. L. Sutter, and S M. Finegold 1971. Companion of selective media for isolation of presump tiye group D streptococci from human feces Appl. . Microbiol 22:100? 105 i 20. Schwan. A 19.S4. The use of a bileaesculin medium and of Mailed s technique of Lanrefield groupinc in the identification of enterococci (Group U streptococci>. J Clin Pathol 7:16*^-16*^ 21 Slaneu. L W.. and C. H Bart lev 1957. Numbers of enterococci in water, sewage, and feces determined by the membrane filter technique with an improved me-, dium. J. Bacteriol 74:59.1-595. 22 Swiuer. R E. and J B Evans 1974 Evaluation of selective media for enumeration of rroup D strepto cocci in bovine feces Appl Microbiol. 28:1088-1067. 23. Winter. C. E . and L A. Sandholier 1946. Recommended procedure for detertinc the presence.of enterococci. TL2. U.S Dept. of Interior. Fish, and Wildlife Service. Washington. D. C. November ABSTRACTS OF THE ANNUAL MEETING-1960 20S Q 46 Microbial Degradation of High Molecular Weight Polycaprolactones. C.V. Benedict* and j.A. Cameron, University of Connecticut, Storrs, CT Two polyca'prolactones, PCL-700 (HW 40,000) and PCl- 300 (HW 10,000) were the sole carbon source in individual, enrichment cultures of lake and river sediments. Degrada tion has been quantitatively monitored using hsoh pressure liquid chromatography. Both mixed enrichment systems and Individual isolates obtained from them have b«r assayed, for degradattve capacity under a variety Of cipditions. In the absence of alternate carbon sources, die higher nolecularweight polymer appears more readil* degraded; 100*. degradation has beer, seen in mixed cultfte systems after!one months' incubation. A yeast, idenff'fied as Cryptscoccus laurentii. has been shown to ey.ensive.iy degraje both polymers. Preliminary evidence indicates that me degree of crystallinity is an impytant char- acterBtic affecting polymer'degradation. Jrwith current polymeti preparation techniques, PCI-300 i# more crystal line thV PCt-700. Most single isolates Mo not have the degradawve capacity of the mixed culturdr. Of nineteen ergarismV eighteen are coryneforms andjfne is a yeast, TrichospcVm pullulans. Among these, a#->inoie corynefonn Mas able \"degrade PCL-300 to a greater extent than the vised cultne from which it was obtain*:). QW The Effect of Nutrient Codbttions upo* Biode- gradation of a Synthetic Jo'yme'". o.A. Marchui' and J.\ Cameron, University o"f Connecticut, .Stor", CT . The effect 0% nutrient conditions upon the ability of ricroorganismslto decide synthetic polymers has beep studied. Twelve ojcterial andAeast isolates from- enrichment culture! were teste/ for polycaprolactone degradation by zona clearing•#• an. agar'containing polymer. Yeast isolates demqfstratec orowtr> ans degra dation of polycaproVctone ijf a m»nina^ meoiun. whereas th? bacteria did nc&\ Bacty.sl degradation xyi/'o"'be.- promoted by the addition ofr^sno, di and tri-carboxylic acids and aHno ae'dite tip•medium. Moo ar.c*.d'- •saccharides did not plom:t/ bacterial dtcraoat'or o* the poly7-?"" and inhibwee/east rieqracation, ho»*ver trtriouic'. could be ovVy--e by a low concentration of succinate in the'meditA/ Inhititior. o* degradation also occurrec upon the Opinion of caproate te tne tedium, suggesting a feJjJiack inhibition mechanist. Q 68 Scanrinf. Electrof-Hiftbicoric Vis.uel:;atton • o* EioJec'-adaticr of V.-nthetiC Po-lyr*«-s by FungiJ k'.J. Coof anoi^. Jarrett, Um ve^sUy o* Connecticiit.fctorrs, CI The polyefter.s polycaprolact^-t 700 (40,000 daltons) and polycaproBctone 300 (10,000 dlHons} were used as polymer model/to demonstrate the modegradation process by fungi. Saeral fungal species, Isolated fror. poly- caprolactone^OO enrichment cultureaof sewage or pond • water showedjfgrowth and signs of degwdatibn on agar plates coat* with a polycaorolactoneViln1. Specimens fpr microsefcy were prepared by ccatifk basal rinimal salts - noble agar plates cr acid washadglass slides •with a polwe'- film, inoculating with afungal isolate, tnt incubating in a moist chamber for alno'nth. After incubationffsamptes were removed from thaaaar and slides, air-driedJsputtered with gold and viewe*with a scanning rflectron microscope (SEH). TheVolecules of a polymer cajst film organise into trie typic» supra- aolecular structures called spherulites deVmstratinq crystalline (ordered) and non-crystal line (\nordered) regions. Biodegradation was seen on the PCL700 and PC1300. It appears that deterioration of the amorphous areas of the spherulites preceded the biodegradation of tbe crystalline areas. Q M A 24-hour Membrane.Filter Procedure for enumera ting Enterococci. A.P. DUFOUR. Marine Field Station. HERL-Ctn., West Kingston. Rl 02892. A membrane filter procedure <»£.) for .enumerating enterococci was modified and simplified by eliminating . .the Jm situ esculin test and decreasing-the Incubation period from 48 hours to 24 hours. Indoxyl-B-D-glucoside. 0 chronogenic cellobiose analogue, was substituted for esculin In the primary medium. The mE procedure and the modified method were com pared for the recovery of enterococci from marine, fresh' and estuarine water sables. Slightly increased recov eries were observed with the modified procedure. Colonies isolated by both methods were submitted to biochemical testing to confirm their identity. Ninety-four percent of the isolates examined from each nethod were, confirmed as enterococci. The false negative rate was less than 41 with the modified procedure. The modified mE procedure is a sped fie,facile method that can be utilized for enumerating enterococci In marine, estuarine or freshwater environments. Q 70- The Enumeration and Tderr if icatie-r cocci frorr. Landfill Leachates. J•; a. DOyr.'.LV,< P. v. SCAFP1S?. and D. BRIBER. Univ. of Clncin^rl and VSEFA, Cincinnati, Ohic. lncreaslnc use of landfills to deposit ses9.se sludge, jir.:c:pal and hospital solid vast'es, and othe# containing p»£orer.lc microbes enhances the posslbll iry#hat these pacmMicns will leave trie landfill m the ]ej#hate and con tainir\j * ground and surface water supplies^ Fecal atrep- tococc^vere found to survive longer In l*dfill«- than most r.rjV-negative rod«, including ccl!f«r--.«, asd their presence^" leachate limited tt>e feral y.icif. cf the wasre. cacerial fV the landfill, Identification ot streptococcal species furVft indicated the veretablf < animal or hu-.3T; bricir. rfthe^Jtachate. Several qualitative strerto^ec- cal procedurtsV er* cor.ppred with ty Standard Methods <297r.' Most PrcVMe Nut-her (*!":>> jft:. Uarhates were". obtained'from la9^ra:erv and laitf scale field lysitneters, and a commercial JVifill. As eoT-l violet a?. Id* broth fro* the VPN test v\r.-'.i -IvayaTsrecific when used vim leachate. it. was rep^^ed b\ sjVeptoccccal identifying .media as bile, esculin T&idt, »:tii!crixoctus agars. leachate. ail plate count^ •higher counts than the Stai m»ter* s^-owed e hieher rerfeVs:f rf Croup D'streptococci. while t»>t co^anfrcial lar»y. Il^krr.riired a creater variety cf strtptocov ci. inz\ui)Jz C.roUft.r. ar.i vlridans- Q. 71 Men" vc-..' QUV- Cclorad.-. St a if Ur] Fort Collins. streptococcal and Tsb streptococcal count \ia gave two-fold to ten-fold' -d Method? MPS. Older Jvsi- 7cchr.:s::c fc% t!:e Isolation of ;i.*t<:r 1. \± F-ARUrv, T J. r-:L:iss. a::: s.. m/V-p'p.jsov. r\ersi:i and Center for tmeease Control, A nrr;brary fil ter procedure was developcdVrr the isolatior of ffn'.nn<a cnic.'i?zei.iJ*ca froia aquat] environments/^ Primary differentiation was based^r the fermentatio^of sorbitol, the absence of lysine deocr- bo.tyla«c aafi arginjne dihvdfolasc activities, and tT product ioi/of urease. Sodiuir. desoxycholatc wa.< incoi poratcd y an inhibitor.of background organisms. The presumpt/. r identification of V. cr.tcrecrC/f<evt was acconplA.'.cd in $'.' hours, at 2$°Z, and the rate of con- firinat»>r. of identity of typical pesitive colonies was 90*.. /he mean recover) rate of 1^ strains of the organ ism f/or phosphate buffer suspensions vas 91\. The tech- niqu/usvd to isolate 33 cultures of V. ertfctrcet<^tca froa" ib cf 2~ river water samples and from prechlorinated sewage effluent. Two isolates were identified as sero type 0:4 (or 0:4.32), 2 were 0:17. and the fifth was 0:40. as of 1 Jan 1981 Media Preparation: mE Medium-Modified Bacto-Agar 15g Bacto-Peptone lOg Yeast Extract 30g NaCl 15g Sodium Azide G.15g Actidione(cycloheximide) 0.05g Deionized Water 1000.0ml Autoclave for ISmin, cool to 50 C. Aseptically add the following: .1. Nalidixic Acid 0.24g, weigh into a sterile test tube, add 3ml sterile water. Add 0.2ml ION NaOH, vortex to dissolve. Add to cooled medium and rinse tube with medium. 2. Triphenyi Tetrazolium Chloride 0.02g, weigh into a sterile test tube. Add to cooled medium and rinse tube with medium. 3. Indoxyl-B-D-Glucoside 0.75g, weigh into a sterile test tube, add 5ml of 95% EtOH. Dissolve with mixing and add 5ml sterile water, vortex. Add to medium and rinse tube with the medium. Swirl to mix or use stir plate, aseptically adjust pH to 7.1+0.1, dispense 4.5mls into 50x9mm plates using a sterile Cornwall, syringe. Store inverted in a foil covered basket at 4 C, will keep about 1 month. Incubation- Place inverted plates in stacks of 2 in. 41 C incubator for 48hrs. Counting- count all typical colonies: pink center, gray fringe, . blue halo or those that are dark with blue halos. No minimum size. References: 1. Levin, M.A., et al, Applied Microbiology vol 30:1 pg. 66- 71, 1975. 2. Dufour, A.P., Abstracts Annual Meeting American Society for Microbiology, 1980, abstract number Q69. TEST METHOD ENTEROCOCCI IN WATER BY THE MEMBRANE FILTER PROCEDURE METHOD 1106.1 1985 1. Citation 2. Scope and Application 2.1 This method describes a membrane filter (MF) procedure for the detection and enumeration of the enterococci bacteria in water. The enterococci are commonly found in the feces of humans and other warm-blooded animals. Although some strains are ubiquitous and not related to fecal pollution, the presence of enterococci in water is an indication of fecal pollution and the possible presence of enteric pathogens. 2.2 Vhe enterococci test measures the bacteriological quality of recre ational waters. Epidemiological studies have led to the development of criteria which can be used to promulgate recreational water C+anriarfic b2ctt/* r"% ^^° »e*»kl iehoH ••als+innrKin t\n+Mekf%r\ Sa»1 *h effects and water quality. The signficance of finding enterococci in recreational water samples is the direct relationship between the density of enterococci in the water and swimming-associated gastro enteritis studies of marine and fresh water bathing beaches (1). 2.3 The test for enterococci can be applied to potable, fresh, estuarine, marine, and shellfish growing waters. 2.4 Since a wide range of sample volumes or dilutions can be analyzed by the MF technique, a wide range of enterococci levels in water can be detected and enumerated. 3. Summary 3.1 The Vf method provides a direct count of bacteria in water based on the development of colonies on the surface of the membrane filter (2), A water sample is filtered through the membrane which retains the bacteria. Following filtration, the membrane containing the bacter ial cells is placed on a selective medium, mE agar, and incubated for 48 h at 41°C. After Incubation, the filter 1s transferred to EIA agar and held at 41°C for 20 min. Pink to red enterococci colonies will develop a black or reddish-brown precipitate on the underside of the filter. These colonies are counted with a fluorescent lamp and a magnifying lens. 4. Definition 4.1 In this method, enterococci are those bacteria which produce pink to red colonies with black or reddish-brown precipitate after incubation . 14 on n£ agar and subsequent transfer to EIA medium. Enterococci include Streptococcus faecal is, Streptococcus faecium, Streptococcus avium, and tneir variants. 5. Interferences 5.1 Water samples containing colloidal or suspended particulate materials can clog the membrane filter.and prevent filtration, or cause spread ing of bacterial colonies which could interfere with Identification of target colonies. 6. Safety Precautions 6.1 The analyst/technician must know and observe the normal safety procedures required in a microbiology laboratory while preparing, using and disposing of cultures, reagents, and materials, and while operating sterilization equipment. 6.2 Mouth-pipetting is prohibited. 7. Apparatus and Equipment 7.1 Glass lens with magnification of 2-5X or stereoscopic microscope. • £ Ldmp, W1 V" O WW I , nil I WW I IUVI COV.CII l» UJUC. 7.3 Hand tally or electronic counting device. 7.4 Pipet container, stainless steel, aluminum or borosilicate glass, for glass pipets. 7.5 Pipets, sterile, T.D. bacteriological or Mohr, glass or plastic, of appropriate volume. 7.6 Graduated cylinders, 100-1000 mL, covered with aluminum foil or kraft paper and sterile. 7.7 Membrane filtration units (filter base and funnel), glass, plastic or stainless steel, wrapped with aluminum foil or kraft paper and. sterile. 7.8 Ultraviolet unit for sterilizing the filter funnel between filtrations (optional). 7.9 Line vacuum, electric vacuum pump, or aspirator for use as a vacuum source. In an emergency or 1n the field, a hand pump or a syringe equipped with a check valve to prevent the return flow of air, can be used. 7.10 Flask, filter, vacuum, usually 1 L, with appropriate tubing. A filter manifold to hold a number of filter bases 1s optional. 15 7.11 Flask for safety trap placed between the filter flask and the vacuum source. 7.12 Forceps, straight or curved, with smooth tips to handle filters Without damage. 7il3 Ethanol, methanol or Isopropanol 1n a small, wide-mouth container, for flame-sterilizing forceps. 7.14 Burner, Bunsen or Fisher type, or electric incinerator unit for sterilizing loops and needles. 7.15 Thermometer, checked against a National Bureau of Standards (NBS) certified thermometer, or one traceable to an NBS thermometer. 7.16 Petri dishes, sterile, plastic, 50 x 12 mm, with tight-fitting lids. 7.17 Bottles, milk dilution, borosilicate glass, screw-cap with neoprene liners, marked at 99 ni for 1-100 dilutions. Dilution bottles marked at 90 nt or tubes marked at 9 mL may be used for 1-10 dilutions. 7.18 Flasks, borosilicate glass, screw-cap, 250-2000 mL volume. 7.19 Membrane filters, sterile, white, yriu marked, 47 mm diameter, with 0.45 + 0.02 pm pore size. 7.20 Inoculation loops, at least 3-mm diameter, and needles, nichrome or platinum wire, 26 B A S gauge, in suitable holders. 7.21 Incubator maintained at 41 +0.5°C. 7.22 Waterbath maintained at 44-46°C for tempering agar. 7.23 Test tubes, 150 x 20 mm, borosilicate glass or plastic. 7.24 Caps, aluminum or autoclavable plastic, for 20 mm diameter test tubes. 7.25 Test tubes, screw-cap, borosilicate glass, 125 x 16 mm or other appropriate size. 8. Reagents and Materials 8.1 Purity of Reagents: Reagent grade chemicals shall be used in all tests. Unless otherwise indicated, reagents shall conform to the specifications of the Committee on Analytical Reagents of the American Chemical Society (3). The agar used 1n preparation of culture media must be of microbiological grade. 8.2 Whenever possible, use commercial culture media as a means of quality control. 16 8.3 Purity of Water: Reagent water conforming to Specification D1193, Type II, Annual Book of ASTM Standards (4). . 8.4 Buffered Dilution Water 8.4.1 Composition: Sodium Dihydrogen Phosphate 0.58 g Sodium Monohydrogen Phosphate 2.50 g Sodium Chloride 8.50 g 8.4.2 Preparation: Dissolve the ingredients in T L of reagent water in a flask and: dispense 1n appropriate amounts for dilutions in screw-cap bottles or culture tubes and/or into containers for use as rinse water. Autoclave after preparation.at 121°C (15 lb pressure) for 15 min. The final pH should be 7.4 + 0.2. 8.5 m£ Agar (Difco 0333-15-1) 8.5.1 Composition of Basal Medium: Peptone 10.0 g Sodium Chloride 15,0 g Yeast txtract 30.0 y Esculin 1.0 g Actidtone 0.05 g Sodium Azide 0.15 g Agar 15.0 g 8.5.2 Preparation of basal^medium: Add 71.2 g of dehydrated mE basal medium to l^t of reagent grade water in a flask and heat to boiling until ingredients dissolve. Autoclave at 121°C and 15 lb pressure for 15 min and cool in a 44-46°C water bath. / ,/Zlr • 8.5.3 Reagents added after sterilization:v Mix 0.25 g nalidixic acid 1n 5 mL reagent grade water, add O..2'mt.of lO^N^aOH. to dissolve, and add to the basal medium. Add oTlb% triphenyl tetrazoliurn chloride separately to the basal medium and mix. 8.5.4 Preparation of mE Agar: Pour the mE agar into 50 mm petri dishes to a 4-5 mm depth (approximateYy 4-6 mL), and allow to solidify. The final pH of medium should be 7.1 + 0.2. Store in a refrigerator. 17 8.6 EIA Substrate Agar (Difco 0488-15-4) 8.6.1 Composition: Esculin 1.0 g />5> Ferric Citrate 0.5 g_' #i<- Preparation: Add 16.5 g of dehydrated EIA medium to 1 L of reagent grade water 1n a flask and heat to boiling until Ingredients are dissolved. Autoclave the EIA medium at 121°C and 15 lb pressure for 15 min and cool In a 44-46PC water bath. After cooling, pour the medium into 50-mm petri dishes to a depth of 4-5 mm (approximately 4-6 mL) and allow to solidify. The final pH should be 7.1 +0.2 before autoclaving. Store in a refrigerator. 8.7 Brain Heart Infusion (BHI) (DifcO 0037-02, BBL 11058) 8.7.1 Composition: Calf Brain Infusion 200.0 g Beef Heart Infusion 250.0 g Peptone 10.0 g Sodium Chloride K n n Disodium Phosphate 2.5 g Dextrose 2.0 g 8.7.2 Preparation: Dissolve 37 g of dehydrated brain heart infusion in 1 L of reagent grade water. Dispense 1n 8-10 mL volumes in screw-cap tubes and autoclave at 121°C (15 lb pressure) for 15 min. If the medium is not used the same day as prepared and sterilized, heat in boiling water bath for several min to remove absorbed oxygen, and cool quickly without agitation, just prior to inoculation. The final pH should be 7.4 +0.2, 8.8 Brain Heart Infusion (BHI) Broth with 6.5% NaCl 8.8.1 Composition: Brain heart infusion broth with 6.5% NaCl is the same as BHI broth in 8.7 with additional NaCl* 8.8.2 Preparation: Add 60.0 g NaCl per liter of medium. Since most commercially available dehydrated media contain sodium chloride, this amount 1s taken Into consideration in determining the final NaCl percentage above. 8.9 Brain Heart Infusion Agar (Difco 0418-02, BBL 11064). 8.9.1 Composition: Brain heart infusion agar contains the same components as BHI (see 8.7) with the addition of 15.0 g of agar per L of BHI broth. 18 8.9.2 Preparation: Heat to boiling until Ingredients are dissolved. Dispense 10-12 mL of medium in screw-cap test tubes and steri lize for 15 min at 1216C (15 lb pressure)'. Slant after sterilization. The final pH should be 7.4 + 0.2. .8.10 Bile Esculin Agar (BEA) (Difco 0879) 8.10.1 Composition: Bacto Beef Extract 3.0 g Bacto Peptone 5.0 g Bacto Oxgall 40.0 g Bacto Esculin 1.0 g Ferric Citrate 0.5 g Bacto Agar 15.0 g 8.10.2 Preparation: Add 64.5 g of dehydrated BEA to 1 L reagent water and heat to boiling to dissolve completely. Dispense in 8-10 mL volumes in tubes, for slants or into flasks for subsequent plating. Autoclave at 121°C at 15 lb pressure for 15 min. Overheating may cause darkening of the medium. Cool to 44-46°C and dispense Into sterile petri dishes. The final pH should be 6.6 + 0.2. Store in a refrigerator. 9. Sample Collection, Preservation and Holding times 9.1 Sampling procedures are described in detail In the USEPA micro biology methods manual, Section II, A (5). Adherence to sample preservation procedures and holding time limits Is critical to the production of valid data. Samples should not be analyzed if these conditions are not met. 9.1.1 Storage Temperature and Handling Conditions Ice or refrigerate bacteriological samples at a temperature of 1-4°C during transit to the laboratory. Use Insulated con tainers to assure proper maintenance of storage temperature. Take care that sample bottles are not totally immersed in water during transit or storage. 9.1.2 Holding Time Limitations Examine samples as soon as possible after collection. Donot hold, samples longer than 6 h between collection and initiation of analyses. 10. Calibration and Standardization 10.1 Check temperatures 1n Incubators daily to insure operation within stated limits. 19 10.2 Check thermometers at least annually against an NBS certified thermometer or one traceable to NBS. Check mercury columns for breaks. 11, Quality Control 11.1 See recommendations on quality control for microbiological analyses in the USEPA microbiology methods manual, Part IV, C (5). 12. Procedure 12.1 Prepare the mE agar as directed 1n 8.5. 12.2 Mark the petri dishes and report forms with sample Identification and sample volumes. 12.3 Place a sterile membrane filter on the filter base, grid-side up and attach the funnel to the base; the membrane filter is now held between the funnel and the base. 12.4 Shake the sample bottle vigorously about 25 times to distribute the bacteria uniformly, and measure the desired volume of sample or dilution into the funnel. 12.5 For ambient surface waters and wastewaters, select sample volumes based on previous knowledge of pollution level, to produce 20-60 enterococci colonies on membranes. Sample volumes, of 1-100 mL are normally tested at half log intervals, for example 100, 30, 10, 3mL, etc. 12.6 Smaller sample size or sample dilution can be used to minimize the interference of turbidity or high bacterial densities. Multiple volumes of the same sample or dilution of sample may be filtered and the results combined. 12.7 Filter the sample and rinse the sides of the funnel at least twice with 20-30 mL of sterile buffered rinse water. Turn off the vacuum and remove the funnel from the filter base. 12.8 Use sterile forceps to aseptically remove the membrane filter from the filter base and roll it onto the ME agar to avoid the formation of bubbles between the membrane and the agar surface. Reseat the membrane if bubbles occur. Close the dish, invert, and incubate at 41 + 0.5°C for 48 h. 12.9 After incubation, transfer the membranes, to EIA agar plates which have been:at room temperature for 20-30 min, and incubate at 41°C for 20 min. 12.10 After incubation, count and record colonies on those membrane filters containing, if practical, 20-60 pink-to-red colonies with 20 black or reddish-brown precipitate on the underside of the membrane. Use magnification for counting and a. small fluorescent lamp to give maximum visibility of colonies. 13. Calculation of Results Use the following general rules to calculate the enterococci count per 100 mL of sample: 13.1 Select and count membranes with Ideally 20-60 pink to red colonies with black or reddish-brown precipitate on the underside. Calculate the final value using the formula: Enterococci/100 mL «. No. of enterococci colonies xioq Volume ot sample filtered (mL) 13.2 See the USEPA microbiology manual, Part II, Section C, 3.5, for general counting rules.5 14. Reporting Results 14.1 Report the results as enterococci per 100 mL of sample. 15. Verification Procedure 15.1 Pink to red colonies with black or reddish-brown precipitate after incubation on EIA agar can be verified as enterococci. Verification of colonies may be required in evidence gathering, and is also recommended as a QC procedure upon initial use of the test and with changes in sample sites, lots of commercial media, or major ingredients in media compounded in the laboratory. The verification procedure follows: 15.2 Using a sterile inoculating needle, transfer cells from the centers of at least 10 wel1-isolated typical colonies Into a brain heart infusion broth (BHI) tube and onto a BHI slant. Incubate broth tubes for 24 h and slants for 48 h at 35 + 0.5°C. 15.3 After 24 h Incubation, transfer a loopful of material from each BHI. broth tube to: a. B1le Esculin Agar (BEA) and Incubate at 35 + 0.5°C for 48 h. b* BHI Broth and incubate at 45 + 0.5°C for 48 h. c. BHI Broth with 6.5% NaCl and incubate at 35 + 0.5<>C for 48 h. 15.4 Observe for growth. 21 15.5 After 48 h incubation, apply a gram stain to growth from each BHI agar slant. 15.6 Gram positive cocci which grow in BEA, BHI Broth at 45°C, and BHI Broth +6.5% NaCl, and hydrolyze esculin, are verified as enterococci. 16. Precision and Bias 16.1 Performance Characteristics 16.1.1 Precision - The degree of agreement of repeated measurements of the same parameter expressed quantitatively as the standard deviation or as the 95% confidence limits of the mean computed from the results of a series of controlled determinations. Precision of the mE method was established by Levin et al.' (2) who indicated that the method did not exceed the expected limits for counts having the Poisson distribution. 16.1.2 Bias - The persistent positive or negative deviation of the results from the assumed or accepted true value. The bias of the enterococci MF method with m£ Agar has been reported to be +2% of the true value (2). 16.1.3 Specificity - The ability of a method to select and/or distin guish the target bacteria from other bacteria in the same water sample. The specificity characteristic of a method is usually reported as the percent of false positive and false negative results. The specificity for this medium as reported for various environmental water samples was 10% false positive and 11.7% false negative (2). 16.2 Collaborative Study Data 16.2.1 A collaborative study was conducted among eleven volunteer laboratories, each with two analysts who independently tested local fresh and marine recreational waters and sewage treatment plant effluent samples, in duplicate. The data were reported to the Environmental Monitoring and Support Laboratory - Cincinnati, U.S. Environmental Protection Agency, for statistical analyses. 16.2.2 The results of the study are shown In Figure 1 where S0 equals the pooled standard deviation among replicate counts from a single analyst for three groupings (counts less than 30, counts from 30 to 50, and counts greater than 50) and Sb equals the pooled standard deviation between means of duplicates from analysts in the same laboratory for the same groupings. The precision estimates from this study did not differ with the water types tested. 22 16.2.3 By linear regression, the precision of the method can be generalized as: S0 « 0.103 count/100 mL +2.42 (dilution factor) and Sg * 0.152 count/100 mL + 5.16 (dilution factor) where dilution factor =volurne 0f original sample filtered 16.2.4 Because of the instability of microbial populations In water samples, each laboratory analyzed its own sample series and no full measure of recovery or bias was possible. However, all laboratories analyzed a single surrogate sample prepared from a freeze-dried culture of Streptococcus faecalis. The mean count (x) and the standard deviation of the counts (Sj) (including the variability among laboratories for this standardized enterococci sample) were 32.5 colonies/ membrane and 9.42 colonies/membrane, respectively. 23 25 20 15 10 'I iii!!i Jiiu ii.! iii ii; >i ' • \ • IO; ft* . j!1 tail I : : I . i | . Ijniiii': 111. i; S!'M:i. IW-i i : . : : ..Till::-: iliu • i > ; 111! i: ,: j 11; . I • • ' ! ! i ;• !••!• 1: 11 S-;b: Standard:devi •1077-rj—• $i-ifi&tAHdaird devj ' .-j jIja|ra;lyst$;Jn 1• J ; ; ' : j '.-.'. i i >• •«! I! I i te counts! from ^s 1 • *.;•-.• it'ipn between the.neans of captivate the 'same lab. »! i : I ! ; Averag i-.ili.« • iii; 1i ; j ; J infllje; i4na if ,...,1.1 mill MT11 1« 1 i 11 ...III! |!|i| j ! ' i 1I • rj 'Ml ! 1! ! I !| ! MM'1 !! |!m r••; ' 1 .• • 1 ' ' i • • 1 1 ; i!HTi[ 11 • • , 1: Mu ll i I ! 1 I • I •MMiil! 1rr 1pi iji ! !il • •• 1 : ; : ; ! 1 ! is : : ! i • ' ' ; : 1 ' ! I J I : : : I i i I 1 ii::}:: t 1! |.! ; : . 1.1 1 1,1 • , . . i'Njii! 1—r • :! ill ' ! ' • • 1 • I i! !!!! ji II L !! 1 10 20 30 Colony (ount per ~"~~40 "•"'•" Plate 50~""""'"" bO 70 80 90 H -Li. 1 I .U- 100 FIGURE 1. Precision Estimates for Enterococci In'Water by the Membrane Filter/mE Procedure References 1 Cabelli, V. J., A. P. Dufour, M. A. Levin, L. J. McCabe, and P. W. Haberman, "Relationship of Microbial Indicators to Health Effects at Marine Bathing Beaches/ Amer. Jour. Public Health, 69, 690-696, 1979. 2 Levin, M. A., J. R. Fischer and V. J. Cabelli, "Membrane Filter " Technique for Enumeration of Enterococci 1n Marine Waters," Appl. Microbiol. 30. 66-71, 1975. 3 "Reagent Chemicals," American Chemical Society Specifications 6th Edition, Am. Chem. Soc., Washington, D.C., 1981. For suggestions of the testing of reagents not listed by the American Chemical Society, see Reagent Chemicals and Standards, Joseph Rosin, D. Van Nostrand, Co.."Inc., Princeton, H.J., 156/, and the United States Pharmacopeia, Nineteenth Edition, United States PharmacOpeial Convention, inc., Rockville, MO, 1974. 4 Annual Book of ASTM Standards, Vol. 11.01, Water, American Society for / Testing and Materials, Philadelphia, PA., 1985. 5. Bordner, R., J. A. Winter and P. V. Scarpino (eds.), "Microbiological Methods for Monitoring the Environment, Water and Wastes, EPA-$00/8-7fl-bl7, O.S. Environmental protection Agency, Office of Research and Development, Environmental Monitoring and Support Laboratory - Cincinnati, Cincinnati, Ohio, 1978. 25 Am u-'n an'p.Hnvikonmi m m Mit kokioi of.v. Apr. 1986. p.. 832-K40 (kiw-::4osmw(ik.>:-()9So:.(ki(» •• Copynjih: « I'M*. AnVcriiiiii Si>ciol\ for Micr(ibioU»j:> Vol. 51. No. 4 Comparative Study of Selective Media for Enumeration of Pseudomonas aeruginosa from Water by Membrane Filtration ANTONIO di VICENTE." JUAN J. BORREGO. FRANCISCO ARRABAL. and PEDRO ROMERO Dcporiiimvnto dv Microhiologia. Facultad'dc dentins, Universidad dc Malaga, Malaga, Spain Received 25 March 1985/Accepted 10 December 1985 In the present study, mPA-Dand mPA-Eagar, modifications of mPA-C agar that reducebackground fecal streptococci that interfere with the differentiation and enumeration of the Pseudomonas aeruginosa colonies grown in other mPA media, are proposed for use in analyzing natural water samples. In addition, the efficiencies of.several culture media for the recovery of/*, aeruginosa in water after membrane filtration and multiple-tube techniquesarecompared.The degree of selectivity, precision, efficiency, and sensitivity achieved withthe proposed mediaexceeded that achieved by current methods. Furthermore, they yielded equal rates of accuracy and specificity. Incubation at 36°C resulted in an improved recovery of stressed P. aeruginosa. In conclusion, wepropose the useof mPA-D andmPA-E agar, bothincubated at 36CC for24to 48h, foranalyzing river water and seawater, respectively. Pseudomonas aeruginosa is considered to be a ubiquitous and easily detectable microorganism in waters and soils, although it is not autochthonous to these environments (16). Among its most important habitats are human and animal fecal wastes, which are the main source of pollution in natural surface waters (3. 10. 16). The importance of the study of P. aeruginosa in natural surface waters used for swimming is based on its potentiality in originating different kinds of infections, especially otitis medium for the isolation and enumeration of P. aeruginosa from natural waters that was later modified by Dutka and Kwan (11) and by Brodsky and Ciebin (4). Even so. these media do not provide suitable results,.because of the growth of background fermentative microorganisms (A. de Vicente, J. J. Borrego; and P. Romero, Abstr. 9th Congr. Nac. Microbiol, 1983,436, p. 929-930). To eliminate thisproblem. mPA-D agar, in which sugars present in mPA-C agar(4) are suppressed, and mPA-E agar, with xylose as the unique • TABLE 1. Average percent recovery of P. aeruginosa strains from different stressed suspensions" 9c Recovery of strain: Incubation <°C'h> Avg ATCC10145 ATCC14216 J7-5 T26 N61 ^recovery mPA • 41.5/48 5.14 22:41 18.41 5.50 4.90 11.07 mPA-B 41.572'' 96.42 84.44 89.55 87.58 90.10 89.62 mPA-C 41.5/24 94.45 74.55 82.06 85.13 98.49 86.94 mPA-D 36''24 87.55 87.10 74.34 80.80 67.47 79.45 36'48 94.98 . 88.51 75.01 . 82.08 . 72.26 82.57 41.5/24- 85.42 80.92 76.08 81.22 96.86 84.10 - 41.5/48 89.87 • 83.20 77.05 81.38. 99.35 86.17 mPA-E 36'24 87.52 83.77 87.28 91.60 72.34 . . 84.75 . 36/48 90:61 85,21 88.26 93.09 76.64 £6.97 41.5/24 92.54 76.44 80.21 90.68 92.53 87.08 • 41.5/48 94.32 83.33 82.90 95.27 93.47 89.45 Calculated with equation 1.Ten samples, of each strain wereusedto calculate the average percent recovery. Similar values were obtained at 96 h. (5, 20. 25). Ii has also been considered to be a water quality indicator microorganism (3. 13, 21). Several different methods, including the multiple-tube technique (1. 9) and membrane filtration (4. 11, 23. 24). have been developed for the enumeration i-f P. aeruginosa in water (10, 17). However, most of them show some disad vantage, and none of them are completely accepted (10). Generally, the counting techniques based oh most probable number (MPN) are considered to be less precise and effec tive than those based on membrane filtration (6. 14). In 1972. Levin and Cabelli (23) developed the mPA * Corresponding author. sugar, are proposed for analyzing river water and seawater samples^ respectively. MATERIALS AND METHODS Field samples. The number and kinds of water samples provided were as follows: 10 samples of polluted river water, 6 samples of seawater, and 9 samples of sewage water, all collected during an 8-week period. The samples were refrigerated at 4°C and processed within 6 h ofcollection. All samples were collected from superficial layers (up to 30 cm below the water surface). Microorganisms. The following five strains of P. aerugi nosa were used for the evaluation of the accuracy of the 832 Vol. 51, 1986 SELECTIVE MEDIA FOR ENUMERATION OF P, AERUGINOSA 833 TABLE 2 Effect ofexposure todifferent stressed suspensions ohthe recovery of P. aeruginosa strains" %Recovery of strains from stressed suspensions of: Avg % recover Medium Incubation('C/h) Control* Seawater stressed for (h): 0 6 24 Freshwater stressed for (hi: 0 6. 24 Distilled water stressed for (h): 0 6 24 mPA mPA-B mPA-C mPA-D mPA-E 41.5/48 41.5/72* 41.5/24 36724 36/48 41.5/24 41.5/48 36724 36/48 41.5/24 41.5/48 0.00 86.85 74.10 84.86 94.69 82.87 85.89 87.53 90.14 93.15 96.25 34.29 92.46 103.80 78.38 80.25 S7.73 90..39 87.<M 90.«8 93.24 95.96 19.60 103.73 88.41 80.32 81.43 94.35 95.46 89.41 92.08 94.44 94.85 34.33 100.42 99.12 85.68 87.29 87.36 87.96 98.80 99.82 91.32 91.41 7.57 82.34 81.86 66.06 68.19 74.68 75.45 74.05 76.36 74.59 75.91 9.85 85.98 94.86 74.30 78.91 90.99 91.12 71.94 74.06 87.66 88.32 0.00 94.88 73.51 77.32 81.51 59.84 71.07 95.57 96.65 68.34 80.69 2.25 59.43 74.91 90.05 93.71 87.51 88.42 76.80 80.12 88.61 90.57 0.70 93.60 99.80 93.12 94.14 97.75 97.85 98.34 100.14 98.92 100.04 2.09 94.45 79.00 64.43 65.55 77.92 78.07 69.95 72.42 80.52 80.55 11.07 89.62 86.94 79.45 82.57 84.10 86.17 84.75 86.97 87.08 89.45 - Calculated with equation 1. Five samples ofeach strain were used tocalculate theaverage percent recovery. * Unstressed cells in BHI. r Sim^ar values were obtained at % h. different methods: ATCC 10145, ATCC 14216. and J75 from seawater; T26 from river water; and N61 from untreated sewage. Twenty-one Streptococcus strains, including 4 5. foecalis, 7S.faecium, and 10 5. avium strains, isolated from samples ofnatural waters, and S.faecium ATCC 10541 were used for the growth assay. Recovery media. Drake 10 medium (9) and Favero aspar- agine broth (1) were used as recovery media in the MPN technique and incubated for 48 h at 36'C. All the tubes that showederowth with greenish-blue pigment, fluorescence, or both under UV light were subcultivated in acetamide agar and milk agar (1) for confirmation as P. aeruginosa. The following recovery media, with membrane filtration, were used: nalidixic acid-cetrimide agar (24). mPAagar (23), mPA-B agar (11), mPA-C agar (4), and mPA-D and mPA-E agar. mPA-D and mPA-E agar are modifications of the mPA-C agar developed by uswhich suppress all ofthe sugar compounds (mPA-D) or only lactose and sucrose (mPA-E). These two media were prepared by the procedure of Brodsky and Ciebin (4). with suppression antibiotics sulfa-, pyridine and actidione. Both of these media were incubated at 36 and 41.5°C and examined at 24 and 48 h. Methods. The MPN assays were carried out us described in Standard Methods (I). The (titrations were fivefold for each one of the assayed methods with 0.45-u.m membrane filters (H.AWG 047; Millipore Corp.. Bedford. Mass.). Phos phate-buffered saline (23) was used as diluent solution. The. membrane filtration technique was used following the spec ifications described previously (1). Control media and suspensions. P. aeruginosa cultures were prepared by inoculation of. strains into brain heart infusion (BHI) (Difco Laboratories. Detroit. Mich.) and incubated for 18 to 20 h at 36'C. Suspensions and culture dilutions were prepared in phosphate-buffered saline blank tubes. The concentrations of Suspended microorganisms were determined by.the spread plate technique with OA-ml portions of tryptic soy agar (Difco). BHI agar (BHIA; Difco). and King's A agar (22). The recounts were deter mined after 48 h at 36'C with three replicate plates. BHIA. plate count agar (Difco). and mPA-B agar without antibiotics ill) were used to quantify, the total number of organisms in natural water samples by membrane-filtration.- Fermentative colony verification. Ihe fermentative colo nies urowri on mPA. mPA-B.and mPA-C agars were isolated on nutritive agar containing 0.V't glucose. Ihe colonies were ' identified bv the following tests: morphology and Gram stain; motility; oxidation-fermentation test; catalase and cytochrome oxidase tests; xylose, lactose, and sucrose fermentation; growth in BHI atI0°C, at45°C, atpH 9.6. and after 30 min at60°C; growth inmilk-0.1% methylene blue, in broth with 6.5% NaCl, and in0,04% potassium tellurite; and resistance to 0.01% sodiumazide (7, 15). Confirmation was accomplished with the API 20 Strep system (Analytab Prod ucts, Plainview, N.Y.). Streptococcus qualitative growth test. Streptococcus cul tures obtained from BHI containing 0.1% glucose were streaked on the different assayed mPA andm Enterococcus (Difco) media. The growth and fermentation results were recorded at 24, 48. and 72 h. TYPICAL AND NON-TYPICAL COLONIES FROM MEMBRANE FILTER ISOLATION. ON T.S. AGAR I MILK AGAR (1) KING'S A MEDIUM (18) I MORPHOLOGY AND GRAM STAIN (15) MOTILITY (2) GROWTH AT 41.5°C (15) KING'S B MEDIUM (18) O/F TEST (2) GLUCONATE OXIDATION (15) CYTOCROME OXIDASE (2) DENITRIFICATION (2) ARCINIflE DIHYDROLASE (15) GELATIN LIQUEFACTION (15) Presumptive ?. aeA.uoU/10'ia Confirmed 'P. ae/iuoAJiV'tu FIG. I. P aeniuino.ui identification scheme (numbers in paren- ihoses indicate references in Literature Cited). Apim. Environ. Microbiol. S?4 . . l»i VIC'LMl IVI A> VaBI.K 3. Verification as P, acnt^sa ofoMomc^^ different methods Typical colonics .Atypical colonies Medium mPA mPA-B mPA-C mPA-D mPA-E Nalidixic acid-cetrimide • Verified as P. arrupinoui. Incubation rC'hi . No." of colonics .examined 41.5/4K 41.5/72 41.5/24 36'48. 41.5/48 36/48 41..V48 . 36^24 71 377 263 443 101 287 114 283 No. examined . 9J Verified" 19 84.21 162 83.95 12h 90.48 299 95.65 53 98.11 204 98.04 50 90.00 58 87.93 No. examined ^Verified* 52 215 137 144 48 83 64 225 1.92 8.37 8.76 5.55 4.16 7.23 9.37 6.67 RESULTS Accuracy was determined by comparison ot- the P. aeru- vinosa ecoveries obtained with the mPA media with those %SS^le reference media (BHIA try** soy agar, and Kings Aagar) by the following equation. detected in the different selective P. aeruginosa media. These factors are highly variable, depending mainly on the media the incubation conditions, and particularly the nature of the- studied sample. In the two proposed media, these fictors varied between 801 for mPA-E agar (36°C for 48 hi Cultures of the five P. aeruginosa strains, with cell densities nf lOMo 10£ CFU per ml in BHI. were suspended in saw^ter distilled water, and river water: all water samples hid been sterilized by filtration. Appropriate volumesof P. had been sterzeu y^ a tooblam a con. fll'7S"el TJTc Vcells per ml and were stored at •ScTKSriy tiis of strain cultures and suspensions were t !n\ nd 24 hby the spread plate method. The ery of 79.41 (mPADaga-. mf.dia. are given >n Sles iand2. Th" accuracy of ihc'.melhod. a.ways ex- ,Fig. 1), The confirmation>*™»W*^g$^% rinfflon conditions, and 8.3* ^W?«S^ can also be seen that all of the ass*iyed "*•"***e* »£d confirmed typical colonies and that they never exceeded actfMide agar selectivities for«~^^ nosa recovery from natural waters are shown in Table 4_ Such sefectivities were obtained by observing the degree of reduction of background microbial flora in »he assj>cq methods Reduction factors were obtained by calculating the Tate between the average number of colonies per.100 ml detected in control mediatBHIA, plate count agar.andhmPA • base agar) and the total number of colonies per 100 ml x 100'number of samples Hi and -n S00 for mPA-D agar (41.5°C for 24 h) in freshwater simnles The degree of reduction ranged from 1.820 for mPA-D agar (36°C for 48 h) to 14,500for mPA.E agar (41.5 C for 24 h) in seawater samples. In sewage samples, minimal and maximum reduction ^^^VX^^^^'l agar and the values were 1.81 x 101 (36"C for 48 h) and 3.8; x IO7 (41 ^°C for 24 h). respectively. Generally, the average reduction'values of the different water samples, obtained with mPA-D and mPA-E agar, were equal or higher than those with other media. • • . . . . The precision of the different methods was graphically determined from dispersion ofFisher inde*; Z>- values of the assay variability for mPA-B. mPA-C, mPA-D, and rnPA-E agar as calculated from the following equation (12, 23): D- = |An*r-0*iVME*f (:» where X, is the bacterial number obtained from each plate of TABLE 4 Selectivity of methods for enumeration of/>. aeruginosa from natural water samples Medium mPA mPA-B mPA-C mPA-D mPA-E Nalidixic acid cetrimide 41.5/48 41.5/72* 41.5/24 36/24 36/48 41.5/24 41.5/48 36/24 36/48 41.5/24 41.5/48 36/24 Mean background reduction factor inC"r°Tn Freshwater Seawater Sc^ ,C/h> <xl0>) txlO*) <>'<" 30.70 8.36 44.40. 1.85 1.03 23.50 10.60 1.69 0.80 17.10 7.65 0.40 16.90 1.22/ 4.75 ' 2.11 1.82 14.20 5.40 3.13 1.92 14:50 5.81 1.16 27.2<« 5.81 2150 19< 20? 3.41 2.62 U> US i.r 2.3" o.tw ""- Ratio between total colonies recount on control"^^eStSil' BHIA. and mPA-B base agar) and total «^^£**. assayed. Values shown arc the arithmetical mean of five samp. * Similar values wereobtained at 96h. Vol. 51. 1986 SELECTIVE MEDIA FOR ENUMERATION OF P. AERUGINOSA 835 n • 0.005 0.025 0.0 5 o 10 a a o o -G a s .o a a dd Q O o ry-^ _ _T« -_...OQ-. - J 0.5 C) n d a ti a in 3 O o o O a • o 9 o c o O 0.1 • i i i i i •••••• I I I II lil. 10 11 .16 .17 24 Sample no. FIG. 2. Precision of mPA-D agar procedures for P. aeruginosa as estimated from dispersion of D1 values (equation 2). Symbols: •. incubated at 36CCtor 24 h:';. incubated at 36'C for 4S h: B. incubated at 41.5"C.for 24 h; G. incubated at 41.5°C for 48 h; -—. control limits when probability was as stated. Sources o( -.amples: I to 10. freshwater: 11 to 16. seawater: 17 to.24. sewage. Data from'five replicate plates were used to calculate the /): value tor each point. the same portion and sample and (V is the number o( replicate, plates, five for each assayed portion. The D1 values, determined for the different methods with mPA-D and mPA-E agar and 24 analyzed water.samples (river water, numbers I to 10: seawater. numbers 11 to 16: sewage water, numbers 17 to 24) and theoretical D2 values for different probabilities are given in Fig. 2 and 3. D2 values of mPA-B and mPA-C agar are given in Fig. 4. From mPA-D and assayed in parallel by the multjple-tube and membrane filtration procedures with the media described previously. The results of each method, expressed as the relative per cent recovery with respect to the maximum recount for that sample, given as 100% (equation 3) and as the percentage of samples in which P. aeruginosa was detected with respect to the total number of tested samples (equation 4). were com pared as follows (12): IP:aeruginosa concentration for assayed method//0, aeruginosa maxirnum concentration obtained for that sample) number of samples number of samples in which P. aeruginosa was detected x 100 (3) (4) total number of analyzed samples x 100 mPA-E agar results, it may be deduced that there was a uniform distribution and that, from the recount variability results, there 'Mis not a significant effect from plate to plate. On the other hand. mPA-B and mPA-C agar presented an extended dispersion of results because of a more h'cteroge-. neous'recount of the different replicates. •Tuenty-four water samples from different sources were Positive./*, aeruginosa percents recovery jn each method •considering that in every sample P. aeruginosa was de tected by one or more methods) and average P. aeruginosa percent levels in each method compared with the maximum level obtained for the same sample are expressed in Table 5. This comparison was also carried out under different culture conditions!incubation temperature and lime).. ji.V, Di VICENTE EI Al. O 3 n > 10 a i Ari'i. Envi'kon. Microbioi _ 2 rlO.005 0.025 0.05 -0.5 M a E 3 c o O 0.1 M I I ' ' ' •' • ' 1 10 ' i t i t i 11 16 I I I I I I I I 17 24 Sample no. TO.', Precision of r^PA-E •* P—£*,*fi^j^^«g£^»l&Sffi^S&Z ,5£^&"*££££S—:*» '• "~U.0 .6. scawa,,. 17 .o 24.sewae, Da,a fr„m five rcpiica* p.a,es were used to calculate the D: value for each point. Fermentative colonies that interfered with P. aeruginosa recounts appeared after 24 h of incubation on mPA-B and mPA-C agar when different kinds of water were analyzed. When identification tests specified in Bergeys Manual (7) were used, these microorganisms appeared to be of the genus Streptococcus. The percentages ofthe isolated micro organisms were as follows: S. avium. 35.30%: S. facialis, *»3 53%; S. faecium, A\.\19c. Also, fecal streptococci from the same water samples were investigated with m Fnterococcus agar: the percentages of fecal streptococci were as follows: S. avium. 33.33%: S- foecalis. 16.66%: S. faecium. 50%. Todetermine whether these microorganisms were responsible for the interference described above, qual itative growth tests on mPA and mEnterococcus agar were carried out. the results show that there was a delay in and inhibition of streptococcal development on the proposed media which improved P. aeruginosa colony recount (Table 6). DISCUSSION When water samples with high concentrations ofbacteria are studied, detection methods for specific organisms must fulfill two basic conditions: high selectivity and optimal recovery efficiency. Several methods have been proposed to. detect and quantify/*, aeruginosa in water samples, but none of them can be considered optimal. The exclusion of sugar by mPA-D agar delayed the growth ofStreptococcus colonies. Because xylose is metabolized by P. aeruginosa but not by Streptococcus species, we decided tp design mPA-E agar. ... mPA-D and mPA-E agar accuracy were evaluated by- comparison with media used in Other methods (4, 11, 23). Average recoveries with mPA-D agar varied from 79.45 to 86.17% for the different kinds of samplestested, depending on incubation conditions. The range varied from 84.75 to 894S% for mPA-E agar.These percentsare similar to those obtained with mPA-B-(89.62%) and mPA-C (86.949?) agar under the same conditions but are much higher than those obtained with mPA agar (11.07%) which also fluctuated, depending on the suspension tested (Table 2). Analyzing the resultsofthe different P. aeruginosa strains(Table 1).it can be seen that the recovery interval for mPA-Dagar oscillated from 67.47 to 99.35% with the N61 strain. The percents ranged from 72.34% for the N61 strain to95.27% for the T26 strain when mPA-E agar was used, these values are very similar to those obtained with mPA-B and mPA-C agarbut cohsiderablv different from those obtained with mPA agar, ranging from 4.90 to 21.14% for N61.and ATCC 14216, respectively, both under culture conditions proposed by Vol. 51. 1986. SELECTIVE MEDIA FOR ENUMERATION OF P. AERUGINOSA 837 © O D a u O G O D a _&& — a_ Q_l 0.005 0.025 0.05 r lOfcr o D §"~0" u\ a a TJ-" •Q" a 0) VJ3 (0 > ._._ ---^ |o.5 O II a a c o o iii i i i ii i i i i 11 ' i i i i i i i i 10 11 16 17 24 Sample no. . FIG. 4. Precision of mPA-Band mPA-Cagar procedures for P. aeruginosa as estimated by dispersion of D: values (equation 2), Symbols: 0. ir.PA-B agar incubated at 41.5'C tor 48 h; O, mPA-B agar.incubated at 41.5'C for 72 to 96 h: •. mPA-C agar incubated at 41.5'C for 24 h; -13. mPA-C agar incubated at 41.5C for 48 h; "---•. control limits when probability wasas stated. Sources of samples: 1 to 10. freshwater; 11 to Ift. seawater: 11 to 24. sewage. Data from the replicate, plates were used to calculate the D2 value for each point. Levin and Cabelli (23). From these results, it can be seen that these average recoveries exceeded the required levels for the correct accuracy of a medium, except in the case of mPA agar. The accuracy results obtained with mPA-B and mPA-C agar agree with the reports of Brodsky and Ciebin (4). but those obtained with mPA agar are in open opposition to those pointed'out by Levin and Cabelli (23). In the most stressful conditions, the P. aeruginosa recov ery with urtselective media was greater than with selective media (Table 2). However, there are instances in which recovery with the selective media assayed increase's.with time exposed.to.a stressful: environment. These results suggest that injury to the cells occurs during the lag phase and that the percentage of injured cells is lower alter a long exposure time to stressful conditions. The decreased per centage'of injured'cells could be explained by.either the repair or the inaetivation phenomena, subsequent to the cell •injury, as previously reported by Hoadley and Cheng and by .Houdle\ (IS. 19). These results occur more frequently when the incubation temperature is not restrictive, as is 36'C. Spec:licit\ of mPA-D and mPA-E agar was high, because the percentage of colonies confirmed as P.-aeruginosa was above KW > in all culture conditions, and the percentage of false-negative colonics never exceeded !()''•. with over 5' ? of colonies being falsely negative on mPA-D agar and 8% being falsely negative on mPA-Eagar. With the remaining meth ods, the positive verification percentage of typical colonies ranged between 83.95 and 90.48% for mPA-B and mPA-C agar, respectively, and the percentage of atypical colonies identified as P. aeruginosa ranged between 1.92 and 8.76% for mPA and mPA-C agar, respectively (Table 3). Even though these results are slightly lower than those obtained by other investigators (4, 11, 23). they have enough reliabil ity to avoid the routine verification of typical colonies. Furthermore, in disagreement with Dutka and Kwan (11), significative changes in colonies cannot be seen when the incubation temperature is modified. Selectivity criteria of a recount medium state that the background floraof water samples must decrease by at least three orders of magnitude (23). Results of the ratio between the average concentration of microorganisms detected-in control media by membrane filtration and the total concen tration of microorganisms grown on P. aerugitiosa-'ukcuve media are given in Table 4, These results show that there wasa significant decrease of background flora,dependingon the level .of bacteria in the sample and also on the source of :he assayed water. Therefore, recoveries from freshwater and seawater samples Were, similar and lower than those- IUX 01 VJ( l.NTK FT At A pri.. Environ. Microbiol. TABU- y. ( omp:•.n'son of the cffKicncy olf diflcrenl proccdures for recover) of /'. aeruginosa from waters '-• Detection or recoven from the following-samples: ., . Incubation Medium . ff-'h! F-reshw;itci l'n - }(u Seawater <n -- >( Sewage (n r 9l . r,r Overall In - 24) Detection"' Recovery1 Detection" Recovery" Detection" Recovery* Detection" Recovery* mPA 36 4S 10.(M) 5.8S 20.00 25.19 >>.^ 23.98 29.16 21.57 •41.5'4S' 0.00 0.00 0;00 0.00 41.572 0.0(1 0.00 0.00 0.00 . mPA-B 36'48 90.00 46.02 40.00 34.02 100.00 53.39 82.61 47.86 • . 41..V4X 60.00 10.09; 40.00 17.17 55.55 63.72 54.16 31.81 41..V72- 9000 12.34 40.00 26.26 66.66 54.45 70.83 28.84 mPA-C 36/24 100.00 74.50 . 60.00 55.50 87.50 59.02 86.95 . 66.23 36'48 • 100.00 ' 75.46 60.00 65.30 100.00 68.18 . 91,30 71.24 '41.5/24' 50.00 11.66 20.00 4.42 55.55 39.64 45.83 23.72 41.5/48 . 9c.oo 26.68 60.00 6.10 77.77 53.29 79.16 33.13 mPA-D 36/24 iOo.OO 82.16 80.00 50.91 100.00 79.24 95.83 75.58 36/48 100.00 82.71 100.00 46.81 100.00 82.96 100.00 74.99 41.5/24 90.0C 23.90 60.00 28.85 77.77 40.83 79.16 30.92 41.5'48 100.00 47.30 60.00 42.46 100.00 57.10 91.66 50.64 mPA-F 36/24 10-».oo 79.83 100.00 66.36 100.00 81.90 100.00 77.81 36'48' 100.00 77.6] 100.00 96.18 100.00 87.85 100.00 85.21 41.5/24 80.00 51.60 100.00 •13.96 88.88 35,76 87.50 . 36.61 41.5/48 . 100.00 65.72 100.00 35.32 100.00 66.32 100.00 59.61 Nalidixic acid- 36'24 100.00 43.19 60.00 43.46 88.88 57.29 87.50 48.60 '• Cetrimide 36'48 90(10 64.4? 80.00 43.21 87:50 57.31 86.95 57.70 Favero broth 100.00 32.31 20.00 1.73 100.00 38.22 • 83.33 33.44 Drake 10 medium .100.00 36.94 0.00 100.00 16.21 7.9.12 27.12 Percentage of samples in which P. animir.osn wa-- detected (equation 41. Average relative percent recovery withrespect to the ma.ximum recount for each sample, based onlyon samples positive for P. aeruginosa (equation 3). Original!) proposed procedure. from sewage samples. If selectivitves of the different meth ods are compared, similar behavior is observed, because all of the methods presented high selectivity (decrease above 1.000-fold), except in nalidixic acid-cetrimide and mPA-E agar incubated at 36°C for 48 h. which did not achieve such a decrease in river water samples. In all niPA media incu bated at 41.5°C. more than a 1.000-fold reduction of over growth wasobtained, and in sewage samples this reduction was even higher. But when the incubation temperature was lowered to 36SC. selectivities of mPA-D and mPA-E agar were reduced by one order of magnitude in river water and seawater samples, but in sewage samples this decrease was insignificant. To explain this phenomenon, it can be implied that selectivity is affected by the temperature of the water TABLE 6. Qualitative growth of Streptococcus test strains on mPA media at 36 and 41.5CC Time (h* Growth of test strains' Assayed medium S. avium . in = 10) S. faecium • (n * 7) S. ({.stalis U: ~ 4) S. faecium ATCC 10541 . m'PA 24 + . F -*. F + . F -. F mPA-B 24 + . F + . F + . F + . F mPA-C .24 + . F •+.'F + . F • -. F raPA-D 24 - ' - • - - 48 - - 72 .(•*-) • ( + ) . ( + ) • <+ » mPA-E 24 - - - ••' - 48 (*) (+•} (*>• (-"•) 72 . (+) ( + ) ( + ) (+> " Results a: 36 and 41.5"C were.equal. h +*Growth: F. fermentation: -. no growth: (+), weak growth. samples, often over 20°C, arid by a high concentration of thermophilic microorganisms (4. 23). Even though there was not a significant quantitative improvement in selectivity with mPA-D and mPA-E agar, there was a qualitative one, because streptococcal growth that produced fermentative colonies on mPA agar was delayed up to 72 h (Table 6). The Streptococcus species isolated from mPA media were the same species that were detected with m Enterococcus agar and were obtained from the same samples of polluted natural water. The importance of delaying streptococcal growth on.mP'A, mPA-B, and mPA-C agar is that these microorganisms present higher concentrations than does P. aeruginosa in the same natural polluted water samples (8). Also, by allowing incubation at a temperature of 36°C. detection of stressed P. aeruginosa cells, which can be inhibited at 41.5°C. is improved. ^ Experimental D2 values in variability assays (Fig. 2 and 3) are approximately equal to the expected estimates, demon strating the high precision of: mPA-D and mPA:E agar. Comparison of D2 values obtained for mPA-B and mPA-C agar with those obtained for mPA-D and mPA-E agar (Fig. 2 to 4)suggests that mPA-Dand mPA-E agar are more precise, especially when incubation is carried out at 36°C. In this study, a higher variability than that obtained by Brodsky and Ciebin (4) can be seen for mPA-B and mPA-C agar. This may be caused by the variation of typical colony morphology (6). Comparison of mPA-D and mPA-E agar assayed methods with other assayed methods (Table 5) was achieved by means of the efficacy and sensitivity of those methods, designated P. aeruginosa recovery efficiency. The tech niques based on MPN. with Favero asparagine broth and Drake 10 medium, presented a high detection sensitivity, except with seawater samples, but they showed a poor percent recovery, about 30%: Even so, it must be noted that Vol. 51.1986 SELECTIVE MEDIA FOR ENUMERATION OF P. AERUGINOSA 839 interference which other background flora can produce in the test tubes can produce false results, decreasing the sensitivity ofthe method. This could also be caused by the individual or global inhibitory effect ofprocesses ofamensal- ism. competition, or toxic substance inhibition ol the sam ple Another inconvenience of this method is that, because its presumptive test is based on fluorescein production and because for the confirmation test pyocyanine production is required, the existence of apyocyanogcnic strains or the nonproduction offluorescein or both (17) that result in aloss of. sensitivity and security of these methods cannot be forgotten. . «-. ,u Efficiency was much higher for membrane nitration meth ods than for MPN. in agreement with the results reported by other inveslieators (4. 11. 23). mPA agar, used as described by Levin and Cabelli (23). did not detect P. aeruginosa in any ofthe samples. Incubation at 36T for 48 honly achieved 29% sensitivity and 21% recovery. Only four assays reached maximum sensitivity detection (100%): mPA-D agar at 36 C for 48 h, mPA-E agar at 36=C for 24 h mPA-&.agar at 36X for 48 h, and mPA-E agar at 41.5'C for 48 h (1 able 5) The highest percent recovery belonged tomPA-h agar at 36 Ctor 48 h. with an average recovery of 85%: the comparative percent recovery was only above 75% in mPA-D and mPA-E agar incubated at 36°C. so both media in these culture conditions offered higher efficiency in P. aeruginosa enu meration from natural water. mPA and mPA-C agar methods presented a low sensitivity and a slight recovery level compared with the methods proposed in this study. Like wise, the mPA-B agar method showed low sensitivity lor seawater and a degree of recovery ranging from 12 to >4,< depending on the type of sample, it must be noted that nalidixic acid-cetrimide agar has high sensitivity, about 87%. even though it only gives 50% recovery of P,aeruginosa; furthermore, limited definition and concretion ot P. aerugi nosa typical colonies, grown on this medium, must be considered. .,••_•'•• <-.!,„ Behavior of the media varies with the source, of the sample. All methods lose recovery ability or sensitivity or both with seawater samples, except with mPA-E agar incu bated at36SC for 48 h:thus, thismedium isrecommended tor these samples. The method that showed the best results tor freshwater samples was mPA-D agar incubated at 36 C tor 48 h. Both methods can be used with sewage water. We must agree with Levin and Cabelli (23) that, with each medium, sensitivity, of the method is considerably lower when turbid samples with high contents of particulate matter or with sediments are used. This is because particulate matter produces faulty filtration and disguises the normal morphology and colony color of P. aeruginosa in these media. ACKNOWLEDGMENTS This work was supported by Comision .-Wsora do Investigation Cicntificu y lecnica grant number 115881 from Min.Merio de Kducacion y Ciencia delGobierno de Espana. UTKRATfRE CITED \n>erkan Public Health Association. P*Kl. Standard methods tor ihc elimination of water and. wastewater. iMh ed. American Public Health Association. Washington. 1)1 \nwrican Sotietv for \licr..hi.ilt>«y Committee -»n Continuing Education. i9Xl.'identification of glucose n.»n-fermen'.ing grain- 3. 4. negative rods. American Society for Microbiology, Washington. DC. Bonde, G. J. 1977. Bacterial indication of water pollution, p. 273-364 In M. Droop and H. W.Jannasch (ed.). Advances in aquatic microbiology, vol. 1. Academic Press, Inc. (London), Ltd.. London. . . • .- ,. . Brodsky, M. H., and B. W. Ciebin. 1978. Improved medium for recovery and enumeration of Pseudomonas aeruginosa from water using membrane filters. Appl. Environ. Microbiol. 36: 36-4"> Cabelli, V. J., H. Kennedy, and M. A. Levin. 1976. Pseudomo nas aenteinosa-fecal coliform relationships in estuanne and fresh recreational waters. J. Water Pollut. Control Fed. 48: Carson L. A.. N. J. Petersen, M. S. Favero, I. L. DottyD. E. Collins, and M. A. Levin. 1975. Factors influencing detection and enumeration of Pseudomonas aeruginosa by most- probable-number and membrane filtration techniques. Appl. Microbiol. 30:935-942. Deibel, R.»., and H. W.Seeley, Jr. 1974. Family II. Streptococ- cuceae, fam. nov.. p.,490-517. In R. E. Buchanan and N. E. Gibbons (ed.). Bergey's manual of determinative bacteriology, 8th ed. The Williams & Wilkins Co.. Baltimore, de Vicente, A., J. J. Borrego, F. Arrabal.and P. Romero. 1983. Aninitial studyof the inactivation of Pseudomonas. aeruginosa in seawater and its comparison with the usual bacterial indica tors of fecal pollution. Rapp. P.-V. Reun. Comm. Int. Explor. Sci. Mer Mediterr. Monaco 28:49-50. . Drake, C. H. 1966. Evaluation ofculture media for theisolation and enumeration ofPseudomonas aeruginosa. Health Lab. Sci. Dutka B J. 1981. Pseudomonas aeruginosa: a controversial indicator pathogen, p. 119-128. In B. J, Dutka (ed), Membrane nitration: applications, techniques and problems. Marcel Dekker. Inc.. New York. . Dutka, B. J., and K. K. Kwan. 1977. Confirmation of the single-step membrane filtration procedure for estimating Pseu domonas aeruginosa densities in water. Appl, Environ. Micro biol. 33:240-245. .„•-..' A El-Shaarawi, A. H., and W. O. Pipes. 1982. Enumeration and statistical interferences, p. 43-*6. /nW. O. Pipes (ed.). Bacte rial indicators of pollution. CRC Press. Inc.. Boca Raton. Fla. Foster D. H.. N. B. Hanes, and S. M. Lord, Jrf 1971. Acritical examination ofbathing water quality standards. J. Water Pollut. Control Fed. 43:2229-2241. Geldreich, E. E. 1975. Handbook for evaluating water bacteri ological laboratories. 2nd ed. U.S. Environmental Protection Agency. Washington. D.C. Gerhardt P.. R.G. E. Murray. R. N. Costilow, E. W. Nester. VV A Wood, NR.Krieg, and G.B. Phillips (ed.). 1981. Manfial of'methods for general bacteriology. American Society for Microbiology. Washington. D.C. . . . Hoadley A.. VV. 1977. Pseudomonas aeruginosa in surface waters p. "U-57. In V. M. Young (ed.), Pseudomonas aerugi nosa, ecological aspects and patient colonization. Raven Press. New York. . ... Hoadley, A.W. 1977. Potential health ha/ards associated witn Pseudomonas aeruginosa in water, p. 80-114. In A. W. Hoadley and B. J. Dutka (ed.)! Bacterial indicators/health hazards associated wilh water. American Society for Testing and Materials. Philadelphia- . Hoadley, A. W. 1981. Effect ofinjury on the recovery ofbacteria on membrane filters, p.. 413-450. In B. J. Dutka (ed.) Mem brane filtration: applications, techniques and problems. Marcel Dekker. Inc.. New York. Hoadley. A. W.. and C. M. Cheng. 1974. The recovery ot indicator bacteria on selective media. J. Appl. Bacteriol. 37: Itoadlev, A. W.. and D. E. Kni«ht. 1975. Outer (external otitisI ear infections among swimmers and nonswimmers. Ar«.n. Environ. Health 30:445. . m • Kenner. B. A., and HP. Clark. 1^74. Detection and ™»™-™- lion of Snlnionilla and Pu-udomotias aenwnoxa. J. water 8. 9. 10. 11. 12. 13.. 14. 15. 16. 17. IK 19 20 21 8.40 Of VICENTE ET AL/ - S! E^tJ^S^l E: Rane>: 1*4 Two simp*- "• 3k for the demonstration of Pvocyanin>nd fluorescein. ). •• lab Clin Med. 44:301-307. ,''''",• -i Uvih MA^, and V. j. Cabenj.1972. Membrane filter lechnjque to ommeniuon of Psaidonionas. aeruginosa..Apt*- M.crobiol. 24 25 Appl. Environ. Microbi,,, '24'8r>4-r870.' ' Lilly, H. A., and E. J. L. Lowburv. 1972. Cetrimide-nalidu;. acid agar as aselective medium for Pseudomonas aerugihos,, j Med. Microbiol- 5:151-153. . Pipes, VV. O. (ed.). 1982. Bacterial indicators of pollution. ,. 1-19.CRC Press. Inc., Boca'Raibn, Fla. mPA - A 1-Lysine Hydrochloride 0.5 g NaCl 0.5 g Yeast Extract rVoc ~ Sodium Thiosulfatie E io* „• _ 0.125 g Sucrose noc X 0.125 g 0.008 0.08 c 1.5 g Lactose no Phenol Red n'nfi / Ferric Ammonium Citrate . Ir',- _ Agar 1. Suspend in 100 mL deionized water 2. Adjust the pH to 7.2 +0.1 3. Autoclave at 15 lbs for 15 minutes 4. Cool to 50 -60 C. and add the following dry antibiotics 0.0176 g Sulfapyridine 0.00085 g Kanamycin 0.0037 g ."•• Nalidixic Acid 0.0150 g Actidione 5'. Adjust the final pH if necessary to 7.1 + 0.1 and. dispense, into 50 x 9 mm petri dishes mPA -. B 1-Lysine Hydrochloride NaCl Yeast Extract Xylose Sucrose Lactose Phenol Red Ferric Ammonium Citrate Sodium Thiosulfate Mg S04 - 7H20 Agar 1. Suspend in 100 mL deionized water 2. Adjust the pH to 7.1 + 0.1 3. Autoclave at 15 lbs for 15 minutes 4. Co6l to 50 -60 C. and add the following dry antibiotics 0.0176 g Sulfapyridine 0.00085 g Kanamycin 0.0037 g Nalidixic Acid 0.0150 g Actidibne 5. Adjust the final pH if necessary to 7.1 + 0.1 and dispense into 50 x 9 mm petri dishes 0.,5 g 0. 5 g 0. 2 g 0..125 g 0..125 g 0..125 g 0,.008 g 0,.08 g b,.5 g 0,.15 g i,.5 g mPA - C 1-Lysine Hydrochloride NaCl Yeast Extract Xylose Sodium Thiosulfate Sucrose Lactose Phenol Red Mg S04 - 7H20 Ferric Ammonium Citrate Agar 1. Suspend in 100 mL deionized water 2. Adjust the pH to -7.1 + 0.1 3. Autoclave at 15 lbs for 15 minutes 4. Cool to. 50 -60 C. and add the following dry antibiotics 0.00085 g Kanamycin 0.0037 g. Nalidixic Acid 5.7 Adjust the final,-pH if necessary to 7.1 + 0.1 and dispense into 50.x 9 mm petri dishes 0.,5 g 0. 5 g 0.,2 g 0..125 g 0..5 g 0,.125 g 0,.125 g 0,.008 g 0,.15 g 0,.08 g 1 .5 g mPA - D 1-Lysine Hydrochloride 0-jj * NaCl 0*2 g Yeast Extract • * Sodium Th.iosulfate 0 008 q Phenol Red ft-' y Mg S04 - 7H20 "•" l Ferric Ammonium Citrate v* n Agar 1.5 g 1. Suspend in 100 mL deionized water 2. Adjust the pH to 7.1 +0.1 3. Autoclave at 15 lbs for 15 minutes 4. Cool to 50 -60 C. and add the following dry antibiotics 0.00085 g Kanamycin 0.0037 g Nalidixic Acid 0*0150 g Actidione 5. Adjust the final pH if necessary to 7.1 +0.1 and dispense into 5.0 x 9 mm petri dishes mPA - E 1-Lvsine Hydrochloride -- -?"5 g NaCl 0.5 g Yeast Extract • ",*• . Xylose -2-i"-9 Sodium Thiosulfate " «rt2 Phenol Red 0-?98 q Mg S04 - 7H20 °-" 9 Ferric Ammonium Citrate u.uo g Agar 1.5 g 1. Suspend in 100 mL deionized water 2. Adjust the pH to 7.1 + 0.1 3. Autoclave at 15 lbs for 15 minutes 4. Cool to 50 -60 C. and add the following dry antibiotics 0.00085 g Kanamycin 0.0037 g Nalidixic Acid 5. Adjust the final pH if necessary to 7.1 +0.1 and dispense into 50 x 9 mm petri dishes m -CX Bacto Peptone 4.0 g Potassium Sulfate 2.0 g Xylose niaga Mg S04 - 7H20 0-28 g Cetrimide °-°° 9 Bacto Agar 1*36 9 1. Suspend in 100 mL deionized water 2. Adjust the pH to 7.1+0.1 3. Add 1.5 mL of Glycerol 4. Heat to boiling in a water bath . 5. Autoclave at 15 lbs for 15 minutes 6. Cool to 50 - 60 C. and add the folldwing dry antibiotics 0.00085 g Kanamycin 0.0037 g Nalidixic Acid 7. Adjust the final pH if necessary to 7.1 +0.1 and dispense into 50 x 9 mm petri dishes