SUITABILITY OF THE H 2 S METHOD FOR TESTING UNTREATED AND CHLORINATED WATER SUPPLIES

Wat.Sci.Tech.44(6): 9-6 SUITABILITY OF THE H S METHOD FOR TESTING UNTREATED AND CHLORINATED WATER SUPPLIES J NAIR, R GIBBS, K MATHEW, AND G E HO Institute for Environmental Sciences, Murdoch University Murdoch, Western Australia SUMMARY Rain water, bore water and catchment water are used for domestic water supply purposes with or without treatment in remote areas around the world. These places seldom have any facilities for routine testing of their drinking water. A simple on-site testing method is highly required in such areas. The H S method has been tested for treated drinking water (Grant and Ziel, 996) and was found to have a good correlation with the standard methods. The present study was aimed at assessing the suitability of the H S method for testing different sources of drinking water. Since these types of water may contain H S producing bacteria not of faecal origin the occurrence of false results in this method cannot be overruled. Therefore it was worthwhile to study whether the positive results are true positive results and what percentage of false positive and false negative results could be expected while using this test for routine analysis of water samples. Results were compared with the results using standard procedures for testing total coliforms, Escherichia coli and Salmonella spp. The present experiment analysed rainwater samples, 7 bore water samples, 4 catchment water samples and 74 remote Aboriginal community water samples. Rain water, bore water and catchment water samples gave true results of 78.5%, 8.3% and 8.5% respectively while the treated and untreated community samples gave true results of 93.7 and 84.6% respectively. It was concluded that in the developing countries where the acceptable level of total coliform is <MPN, the H S method would be a good test to identify microbial contamination. In other regions, the H S method could be used as a screening test for drinking water supplies. INTRODUCTION In remote areas around the world people use treated and untreated water for domestic purposes. Although there are many chances that the water in these systems will be polluted, the water is seldom tested for microbial quality. There is a need for a simple and inexpensive on-site methods for routine testing of water in these situations. The H S method developed by Manja et al. (98) if found reliable, would be a suitable method for routine testing of water quality. Untreated water may contain bacteria from both faecal and non faecal sources such as soil, plant and animal matter. Since the water sample might contain H S producing bacteria other than of faecal origin the occurrence of false results in the H S method cannot be overruled. Therefore it is worthwhile to study whether the positive results shown by the H S method are true positive results and what percentage of false positive and

-- false negative results could be expected while using this test for routine analysis of water samples from different sources. Rainwater is a source of drinking water in remote areas of Australia as well as in other parts of the world. Rainwater is usually considered as a pure source of drinking water. But there are many chances of contamination, which might affect the physical, chemical and microbial quality of the water stored in tanks. Since rainwater varies in quality in different tanks, rainwater samples forms a good range of samples to be analysed for testing the efficiency of the method. In rural and remote areas and in many parts of the developing countries, bore wells are the only source of drinking water. Even in metropolitan areas of the developed countries this is a source of water for washing. Therefore testing the bore water that is used for household purposes whether it is for drinking or washing is highly essential especially in areas using septic tanks. Water from catchment dams is supplied to metropolitan areas after treatment. The water is frequently tested before and after treatment to ensure that it is safe to be distributed. These samples form a good source of untreated water samples to study the efficiency of the method. Similarly most of the Aboriginal communities of Western Australia are situated in isolated remote locations, which makes it expensive and difficult to conduct routine testing. Although some kind of water treatment is carried out, proper maintenance of the treatment system is often difficult in these remote areas. This is often the case of remote communities throughout the world. Therefore analysing the efficiency of the H S method in testing the community water supply, which is affected by human activities, should be conducted. Furthermore the H S method is primarily intended for such remote villages where routine maintenance of the water supply system is lacking. The present study was therefore aimed at assessing the suitability of the H S method for testing rainwater, bore water, catchment waters and community water supply in comparison with the standard methods currently used for testing total coliforms, Escherichia coli and Salmonella spp. MATERIALS AND METHODS.. Sample collection Rainwater and bore water samples (5ml) were collected from household tanks in and around Perth, Western Australia. The samples were analysed within 4 hours of collection and were refrigerated if stored for more than 4 hours. The catchment water samples were obtained from the PathCentre (State Health Laboratories), which received samples from the Water Corporation of Western Australia for routine testing for the presence of total coliforms and E.coli. An additional ml of the sample was collected for this study every time a sample was collected for the Pathcentre. The community water samples were also collected by the Water Corporation of Western Australia. They were brought to the Pathcentre in cold boxes within 4 hours of collection. The present experiment analysed rainwater samples, 5 bore water samples, 4 catchment water samples and 75 Aboriginal community water samples... Testing procedure The rainwater and bore water samples were tested for total coliforms and E coli by the membrane filtration method using mfc agar, Salmonella concentrations were determined by the method described in the Anon

-3- (98). Each sample was simultaneously tested using the H S method. Whenever there was an extra ml sample, the sample was also tested using the medium M (see below). The catchment and the community water samples were tested for total coliforms and E.coli by the Pathcentre as per the Australian Standard (995). The tests for Salmonella sp and the H S method were conducted at Murdoch University..3. The H S method The H S media (Mand M) were prepared as described by Pillai et al. (999). Rainwater samples were analysed using both media. For testing the catchment and community water samples only medium M was used because of the scarcity of the water samples. M was selected for the experiments because in previous experiments it was found to be more effective (Pillai et al., 999). The H S bottles were incubated at 37 C and examined after 4 and hours for positive results. The bottles that did not turn positive after hours were considered as negative. The data was analysed to find the correlation of the H S method with the presence of total coliforms, E.coli or Salmonella spp. Some identification of the bacteria that caused positive results was carried out. True results were noted when both the H S method and the standard method for total coliform bacteria gave the same result. False positive results were when the H S method gave a positive result when there was no coliform bacteria. False negative was when the H S method failed to give a positive result in the presence of coliform bacteria. The results were analysed following a method described by Mack and Hewison (988). Sensitivity is the ability of a test to determine a true positive result while specificity is the ability of the test to determine a true negative result. The Predictive Value (PPV) is the ability of a positive test to predict the presence of coliforms, whereas the Predictive Value (NPV) is the ability of a negative test to predict the absence of coliforms. They are calculated as follows H S method Laboratory Result (Standard Methods) a b c d a+c b+d Sensitivity = a/(a+c) X Specificity = d/(b+d) X PPV = a/(a+b) X NPV = d/(c+d) X According to the WHO (993), all water intended for drinking must not contain E.coli or thermotolerant bacteria in a ml of sample, and all treated water must not contain any total coliform bacteria in a ml sample. However in a great majority of rural water supplies in developing countries, this quality could not be achieved. In most developing countries such as Thailand (Sivaborvorn, 988), India (Manja et al. 98) and Malaysia (Anon, 985) a total coliform count up to MPN/ ml is acceptable. In the present experiment the correlation of the H S method with the presence of coliform bacteria was examined at a total coliform count of per ml as well as to a limit of per ml. 3. RESULTS AND DISCUSSION The H S medium (M) was used to test rainwater samples out of which 56 were tested with medium M also (Table ). High contamination with total coliforms (>CFU/ml) was noticed in 35 samples. In

-4- addition to the coliforms, Enterobacter cloacae, E.amnigenous, Proteus, Citrobacter diversus, C. freundii, Serratia sp., Erwinia nigrifens, Xantho maltophilia, Chromobacterium violaceum were isolated from some of the positive bottles. Salmonella arizona was present in two samples. Four samples that were false negative contained E.coli. Proteus was a common bacteria that was isolated from the H S positive samples even in the absence of coliforms. The percentage of true results and false results observed by the H S method are presented in Tables & 3. When comparing the two media, M gave more true results when the bottles were incubated for hours whereas the percentage of true results was higher after 4 hours of incubation with M (Table ). False negative results were greater with M whereas false positive results were greater with medium M. Both media gave a similar percentage of true results after 4 hours of incubation of rainwater samples (Table ). The percentage of false positive results was higher after hours of incubation whereas false negative results were more after 4 hours of incubation with both media. False negative results were obtained in 7 samples, out of which 5 samples contained low level of coliform bacteria (<5 CFU/mL), and no E.coli were found in those samples. Both M and M gave a higher percentage of true results after 4 hours of incubation when a total coliform count of <CFU/mL was considered (Table ). In bore water samples false positive results were obtained in 4 samples after hours of incubation. True results were higher with M whereas the percentage of false positive and false negative results was higher with M. The true results with M was 73.3% after 4 hours of incubation at <CFU/mL whereas the true results increased to 86.6% at total coliform count of <CFU/mL (Table ). At both levels of coliform bacteria, the medium M gave a higher percentage of true results. There was no false negative results at the level of < coliform /ml. The water supplies at the Aboriginal communities were both treated (49 samples) and untreated (6 samples). Out of the 75 samples tested, 7 samples gave true results in comparison to the presence of total coliforms at <CFU/mL whereas 68 samples gave true results at <CFU/mL (Table). The treated samples gave 93.7% true results at both the coliform levels whereas the percentage was lower for untreated samples ie, 84.6 and 9.3% respectively at the coliform level of < and <CFU/mL. False positive results were obtained with the treated samples whereas the percentage of false negative results was nil at a coliform count of <CFU/mL (Table & 3). Salmonella sp. was observed in positive samples. Table. Percentage of true and false results for the H S test in comparison to the standard method for testing coliforms at a total coliform count of CFU/mL Samples True results(%) False positive(%) False negative(%) Rain water 4 hours hours 4 hours hours 4 hours hours M ( samples) 78.5 75. 5.7 7.3 5.7 7.4 M (56 samples) 78.5 8.3 3.5.7 7.8 8.9 Bore water-m (5 73.3 6. 6.6 6.6. 3.3 samples) M (6 samples) 7.4 57. 4. 8.5 4. 4. Catchment water-m (4 78. 8.9 9.7.. 4.8 samples) Community water Treated (49 samples) Untreated (6 samples) 93.8 84.6 4.. 5.4

-5- Of the 4 catchment water samples tested, samples were free from coliforms and Salmonella sp. Four samples showed false positive results and five samples showed false negative results at <CFU/mL of total coliforms (Table). The true results were higher (78.%) at a coliform level of <CFU than at the level of < CFU (58.5%). The false positive results were found to be very high (36.5%) at total coliform count of <CFU/mL and false negative results were high at <CFU/mL of total coliforms (Tables & ). Salmonella sp. was observed in sample, which was also positive by the H S method Table. Percentage of true and false results in comparison to the standard methods for testing coliforms at a total coliform count of CFU/mL Samples True results(%) False positive(%) False negative(%) Rain water 4 hours hours 4 hours hours 4 hours hours M 73.5 64.4 9. 3.4 7.4 4. M 75. 6.5 7.8 3. 7. 5.3 Bore water- M 86.6 6. 3.3 4. M 66.6 33.3 33.3 66.6 Catchment water- 58.5 53.6 36.5 43.9 4.9.4 M Community water Treated Untreated 93.7 9.3 Sensitivity, Specificity, PPV and NPV values are shown in Table 3. The Sensitivity and NPV were % with bore water using M at both the coliform levels whereas with M % sensitivity was obtained only at < CFU/mL. The case was similar with treated and untreated community samples were only M was used to test the samples. The sensitivity and NPV had higher values when incubated for hours with all samples,whereas the specificity and PPV were better when incubated fr 4 hours. Table 3. Sensitivity, Specificity, PPV and NPV values of the H S method compared to the total coliform counts 6. 7. Sample Media Hrs < CFU Sensitivity Specificity PPV NPV < < < < < < CFU CFU CFU CFU CFU CFU < CFU Rain water M 4 67.7 83.8 76.4 8.3 9. 76. 74.3 53.8 9.3 8. 56.5 43.7 69.7 79. 87.8 87.5 M 4 Bore water M 4 M 4 Catchment M 4 water Communit M y Treated Untreated 7. 86. 4. 6. 83.3 93.3 66.6 63.6 76.9 9.8 87.5 93.7 87.7 6.5 75. 33.3 9. 6. 58.3 54.5 95.5 7.9 53.6 33.3 5. 85.7 57. 4. 8. 89.4 93.7 86.5 7.7 66.6 4. 66.6 4.8 86. 84.8 5. 56.6 47. 6. 5. 5. 6.6. 45.4 77.7 33.3 7.4 76.5 75. 75. 63.6 75. 97.9 78.9 86.7 89.8 83.3 87.5

-6- It appeared that L-cystine accelerated H S production and therefore more positive results were observed at 4 hours for M whereas with M the H S production was slightly slower and therefore took hours. It was observed (Pillai et al., 999) that M was better able to detect faecal coliforms to lower concentrations and at a wider range of incubation temperatures. For this reason more false negative results were observed with M. Since the method detected H S producing bacteria, false positive results observed could be due to the presence of other H S producing bacteria such as Enterobacter sp., Proteus and Citrobacter, which were isolated from some of the positive bottles of rain water samples. Although all samples were tested for Salmonella sp. only contained S. arizona. The groups of bacteria that belong to Enterobacter and Serratia that have been isolated from some of the samples are common in soil and water and sometimes occur in the intestinal tract. The Proteus sp. which was common in many of the H S positive samples were probably soil inhabitants but were found in particular abundance in decomposing animal material. Some phytopathogens such as Erwinia sp and Chromobacterium were also isolated from the positive bottles. The source of these bacteria could be the leaves that were carried into the tanks. Many phytopathogens as well as the other bacteria identified from the positive samples are of human health concern. Therefore false positive results as compared to the coliform test would be due to contamination of the rainwater tanks by soil, plant matter or dead animals such as the frogs or reptiles which are washed down to the tank from the roof. Fujioka et al. (995), reported that C. perfringens and RNA phage were absent in all the rain water samples tested, which indicated that the source of the contamination was not human faeces or sewage, and therefore rainwater is not likely to contain human faecal pathogens. The quality of water in rainwater tanks changes according to the season. During the dry season, because of the lower water levels and the higher water temperatures, bacteria grow faster than in the rainy season where in most cases the tanks overflow. Therefore there is a need for frequently testing water in the tanks. When comparing rainwater results from the two media at a coliform level of < and < CFU/mL, M showed a higher percentage of true results after hours of incubation. But the presence of higher percentage of false negative result with M is a concern, as it shows the inability of the medium to detect some of the positive samples. With bore water M was definitely better in relation to the percentages of true, false positive and false negative results. When comparing the incubation period, the true results were almost similar after 4 and hours of incubation except in bore water, using M. The relatively higher percentage of true results after 4 hours could be due to the low number of samples tested (7). The higher percentage of false negative results noted after 4 hours suggest that it will be worth examining the test again after hours. Although there is an increase in false positive results after hours of incubation, it could be due to the presence of bacteria from other sources such as soil, leaves or decaying animal matter that are not advisable to be present in a safe drinking water. False negative results observed are a concern for using this method for testing rainwater. This showed that the method did not correlate to the presence of coliforms. The majority of these samples (5/9) contained low numbers of faecal coliform bacteria. But the presence of E.coli in 4 samples that were false negative is a controversy. In the community water samples out of the two false negatives, one contained E.coli and the other sample had 5 total coliforms in the ml sample. Ziel et al. (995) reported that the H S method correlated 87.7 % with the presence absence test (P/A) for coliform. According to Martins and Pellizari (99) percentage agreement varied between 66.7-9% with raw waters and a 9-94 % with drinking water samples. More positive results were obtained with the H S test than the other tests, could be the false positives as obtained in the present work. Grant and Ziel (996) observed that

-7- out of the 4 well water samples tested there was false positive and false negative result whereas 7.% false negative results. Rijal and Fujioka (995) by testing 5 different rainwater tanks for five weeks observed that the concentration of total coliforms correlated well with the H S producing organisms. But since the same five cisterns were tested for five weeks the bacterial population could almost be same and the number of false positive and false negative results were not identified separately. Wallis (99) reported that on testing rainwater tanks in Thailand % false positive and 4 % false negative results were obtained. Hazbun and Parker (983) recommended that increasing the incubation temperature to 37 C for 4 hours would reduce the false negative results, as was done in the present work. It was observed that the percentage of true results was higher in treated and untreated community water samples than the natural waters. The only sample with false negative result with treated water contained one coliform bacteria. This shows that treatment reduced the false results and therefore correlated better with the standard methods. The percentage of true results was higher with bore water and catchment water than with the rainwater samples. According to Mack and Hewison (988), for a test to be useful, the sensitivity and specificity should be 8% or better. To accurately screen the water samples the PPV and NPV should be %. When testing the drinking water samples, they received a sensitivity and specificity of 6.5 and 6.9% respectively with the H S method. In the present study better percentages were obtained with all the water samples. Since the sensitivity is the ability of the test to determine the true positive and the NPV is the ability of the negative test to predict the absence of coliforms, the test can be considered to be good in these cases. As Mack and Hewison (988) suggested, the lower values obtained with specificity and PPV could be due to the false results caused by the presence of environmental bacteria that release hydrogen sulphide as a metabolic product. It was observed that the percentage of true results is higher with the community water than the rainwater and the borewater samples. Although rainwater is a clean source of drinking water, since thetank is a confined environment the bacteria multiply rapidly and a complex bacterial population persists. This could be the reason for the poorest correlation obtained for rainwater than the deep and shallow wells and the pond water by Mack and Hewison (988). In the community water samples the two false positive results obtained with the treated water samples had no faecal coliforms. This shows that the treatment was effective to destroy the coliform bacteria but not the hydrogen sulphide producers. The positive results obtained with treated water may indicate some kind of contamination if not of faecal origin. 5. CONCLUSION The medium M gave better correlation with the coliform bacteria when testing borewater and the correlation was almost similar for rainwater samples. The results described here indicate that the H S method (M) could be used as screening method for treated drinking water, where a 93.7% correlation was obtained. The false positive and false negative results were 4.% and.% respectively at <CFU whereas the false negative results were nil at <CFU/mL of coliform bacteria. This indicates that in the developing countries where the acceptable level of total coliform is <MPN, the H S method would be a good test to identify microbial contamination. In other regions, the H S method could be used as a screening test that could be carried out by the house owners to assess the quality of drinking water in the rainwater tanks or in remote communities where no other facilities are available. 6. ACKNOWLEDGEMENTS

-8- We acknowledge Mr Kim Patridge, Water Corporation, Western Australia and Mr. Ray Mogyorosy, PathCentre, Western Australia for organising the water samples and the help rendered by them in this study. 7. REFERENCES Anon (98). Reports on Public Health and Medical Subjects No 7, HMSO Books, London Anon, (985). Food Act 983, Kuala Lumpur: Malasian Government, p89. Australian Standard (995). Water Microbiology Method 7: Thermotolerant Coliforms and Escherichia coli Membrane Filtration Method. AS 476.7. Castillo, G., Duarte, R., Ruiz, Z., Marucic, M.T., Honorato,B., Mercado,R., Coloma, V., Lorca,V., Martins,M.T. and Dutka, B.J. (994). Evaluation of disinfected and untreated drinking water supplies in Chile by the H S paper strip test. Water Research. 8:765-77 Desmarchelier, P., Lew, P., Caique, W., Knight, S., Toodayan, W., Isa, A.R. and Barnesn A.(99). An evaluation of the H S water screening test and coliform counts for.water quality assessment in rural Malaysia. Trans. Royal Society of Tropical Medical Hygiene, 86: 4-45. Fujioka, R., Rijal, G. and Ling, B. (995). A solar powered UV system to disinfect cistern water. Proceedings: Rainwater utilization for the world s people, Volume.The International Rainwater Catchment Systems Conference June -5, Beijing, China. Grant, M.A. and Ziel, C.A. (996). Evaluation of a simple screening test for faecal pollution in water. Journal of Water SRT- Aqua.,45(): 3-8. Hazbun, J. A. and Parker, M (983) Simplified test for the detection of faecal pollution in drinking water, Third national rural water supply and sanitation workshop, Solomon Islands, 6-7 June. Kromoredjo,P. and Fujioka, R.S.(99). Evaluating three simple methods to assess the microbial quality of drinking water in Indonesia. Environmental Technology and Water Quality, 6: 59-7. Mack, K. F. and Hewison, K. (988). The PATH/NEVWRP Tests. Thai-Australian Northeast village water resoure project, Report No 47: Evaluation of a hydrogen sulphide screening test. Manja, K.S., Maurya, M.S. and Rao, K.M. (98). A simple field test for the detection of faecal pollution in drinking water. Bulletin of the World Health Organisation. 6: 797-8. Martins, M.T. and Pellizari, V.H. (99). Evaluation of coliphage test and other simple microbiological methods for the examination of drinking water and classification of water sources. In: Dutka B.J. and El-Sharaawi A.H. (eds) The use of simple inexpensive microbial water quality tests. IDRC Report, 99, 79-9. Pillai, J., Mathew, K., Gibbs, R. and Ho, G. E. (999). H S paper strip method-a bacteriological test for faecal coliforms in drinking water at various temperatures, Water Science and Technology, 4(): 85-9 Rijal, G. and Fujioka,R. (995). A homeowners test for bacteria in cistern waters. Proceedings: Rainwater utilization for the world s people, Volume.The International Rainwater Catchment Systems Conference June -5, Beijing, China. Sivaborvorn, K. (988). Development of simple tests for bacteriological water quality of drinking water. (Water Quality Control in South East Asia). Final Technical Report to IDRC. Dept. of Sanitary Engineering, Mahidol University, Thailand Centre, File 3-P-83-37-3 Wallis, I. (99). Simple test to detect contamination of drinking water. Project report Marine and Freshwater Research Center, No /9. Land and Water Resources Research and Development Corporation, Canberra, ACT p. WHO (993). Drinking water quality control in small-community supplies, In: Guidelines for Drinking Water Quality. Volume 3. WHO, Geneva. Ziel, C A. Rijal, G and R. Fujioka (995) Comparison of Manja s Hydrogen sulphide producing bacterial test to coliform test. Proceedings of the 95th General Meeting American Society for Microbiology, Washington DC -5 May, Section 5. Water Quality : Miscellaneous.

-9- Table. True and False results of the H S method as compared to total coliform count with different water samples Samples H S method Lab Method Rain water Bore water M (56 samples) 4 hours M hours M ( samples) 4 hours hours M (6 samples) 4 hours M hours M (5 samples) 4 hours hours Catchment M 4 hours M hours Aboriginal Community Untreated (6) Treated (49) <CFU <CFU 3 3 4 9 6 6 4 8 5 9 3 45 7 3 3 9 5 9 59 56 34 38 9 35 5 44 3 3 3 3 5 5 8 7 4 3 9 4 6 4 7 5 6 5 44 4 5 7 4 6 8 5 8 7 7 3 45