Drinking Water Test Methods in Crisis-Afflicted Areas: Comparison of Methods Under Field Conditions

Abstract

To simplify the testing of drinking water in crisis-afflicted areas (as in Kosovo in 2007), rapid test methods were compared with the standard test. For Escherichia coli and coliform pathogens, rapid tests were made available: Colilert^®-18, P/A test with 4-methylumbelliferyl-β-d-glucoronid, and m-Endo Broth. Biochemical differentiation was carried out by Enterotube™ II. Enterococci were determined following the standard ISO test and by means of Enterolert™. Four hundred ninety-nine water samples were tested for E. coli and coliforms using four methods. Following the standard method, 20.8% (n=104) of the samples contained E. coli, whereas the rapid tests detected between 19.6% (m-Endo Broth, 92.0% concordance) and 20.0% (concordance: 93.6% Colilert-18 and 94.8% P/A-test) positive samples. Regarding coliforms, the percentage of concordant results ranged from 98.4% (P/A-test) to 99.0% (Colilert-18). Colilert-18 and m-Endo Broth detected even more positive samples than the standard method did. Enterococci were detected in 93 of 573 samples by the standard method, but in 92 samples by Enterolert (concordance: 99.5%). Considering the high-quality equipment and time requirements of the standard method, the use of rapid tests in crisis-afflicted areas is sufficiently reliable.

Introduction

I n crisis-afflicted and disaster areas, infrastructure as well as equipment and consumables are usually limited compared with common laboratory facilities in Germany. Thus, it is important to be able to analyze samples rapidly but nonetheless reliably under these restricted laboratory conditions. Today, rapid tests for qualitative and/or quantitative detection of indicator bacteria are available. Not only are these tests rapid, but most of the consumables are dry and less voluminous and can be stored at room temperature. Thus, storage and handling are better suited to field laboratory conditions, where logistics and cold storage cannot be guaranteed at any time.

The methods of testing drinking water microbiologically are based on the detection of indicator bacteria in Europe and the United States. In the United States, the Total Coliform Rule (TCR) issued by the U.S. Environmental Protection Agency (EPA) in 1989 is applied. The TCR establishes maximum contaminant levels, and routine tests are required for presence or absence of total coliforms. If a repeat sample is positive for total coliforms, this sample has to be analyzed for fecal coliforms or Escherichia coli. The nondetection of those indicators is used to prove the absence of pathogens in the water to a certain degree. In Germany, the Regulation for the Water Quality for Human Use (2001) applies to drinking water, which can influence human health. This regulation was legislated based on § 37 No. 3 and § 38 No. 1 of the Infection Protection Act (2000) and on § 9 No. 1 of the Foodstuffs and Community Act (2006) and transforms the Council Directive 98/83/EC into German law (1998).

In addition to coliform bacteria like in the United States, in Germany and Europe also E. coli and enterococci are regarded as indicator bacteria in routine samples. A group of microorganisms that are taxonomically different is called coliform bacteria, because a huge amount of them can be found in feces of warm-blooded animals and humans. Further, those bacteria also occur in the environment, but naturally not in water. Coliforms serve as indicators for recent or continued contamination and thus as fecal and environmental indicators.

E. coli has the highest specificity, sometimes representing the fecal origin of all indicator bacteria (Cabelli, 1982; Feng and Hartman, 1982; Edberg et al., 1988; World Health Organization, 2004). High numbers of E. coli are found in fresh feces of endothermic animals and humans, but do not occur in water naturally. Because of their low survival rate in water, they are pure fecal indicators for recent as well as for continued contamination.

Enterococci (Enterococcus faecalis, E. faecium, E. durans, and E. hirae) occur in feces as well as on plants. As they are quite resistant against environmental changes, they can survive in water for longer periods and are regarded as fecal and environmental indicators for long-lasting contamination.

The Regulation for the Water Quality for Human Use (2001) specifies the method, following EN ISO 9308-1 (1997), for investigating E. coli and coliforms. This method describes a standard as well as a rapid test, both of which are based on membrane filtration without enrichment. Enterococci are detected following EN ISO 7899-2 (2000)—membrane filtration method as well. These methods are reliable, but consume time and resource. A large amount of consumables is needed and must be stored cold.

Amongst other rapid methods, E. coli and coliforms can be investigated by Colilert^®-18 (Idexx) (Bernasconi et al., 2006), the Presence/Absence-Test Broth with MUG (Hach), and the PourRite™ m-Endo Broth Test (Hach). Samples are mixed with selective media that select E. coli and coliforms by specific color changes. Positive samples are subcultured and biochemically analyzed by Enterotube™ II (Becton Dickinson). In contrast to the other two tests, the m-Endo Broth Test allows the quantitative determination of bacteria.

Enterolert™ (Idexx) is a rapid test for enterococci. This qualitative test detects enterococci of the enriched sample by means of the fluorescence technique.

The aim of this study was to compare alternative rapid methods for the testing of drinking water in crisis-afflicted and disaster areas with the standard methods regulated by the German Drinking Water Regulation (2001). One of the top-priority tasks in such regions is to provide both local inhabitants and relief personnel with hygienically safe drinking water. It is particularly important to obtain test results as quickly as possible with an amount of effort that is suitable in an operational environment.

This study was conducted in Kosovo in 2007, when the German Federal Armed Forces were stationed for peacekeeping after the conflicts in 1997 and 1998. As the infrastructure of the country was still bad, this region was regarded as a crisis-afflicted and disaster area. Water samples from different sources in Kosovo were analyzed using the standard as well as rapid methods.

Materials and Methods

Samples

Routine water samples (573) were analyzed during a foreign assignment of the German Federal Armed Forces in Kosovo. The water samples were spiked water samples (n=40 water samples), raw (n=1) or processed (n=352) water from military facilities of KFOR (Kosovo Force), raw (n=78) or processed (n=40) water from Kosovoian civil facilities, or Kosovoian surface water (n=62). About 800 mL water was sampled in eight Thio-bags (Whirl-Pak^®) after cleaning and disinfecting the sampling location and draining the water for 5 min. All samples were transported and stored cold, were split into four aliquots, and investigated by the standard test, the Colilert-18 test, and P/A test with 4-methylumbelliferyl-β-d-glucoronid (MUG), but only 499 samples were also analyzed by m-Endo Broth.

Detection of E. coli and coliform bacteria

E. coli and coliform bacteria were detected by four different methods per sample. The standard method was the DIN EN ISO 9308-1, which was in accordance with the German Drinking Water Regulation (2001). The remaining three methods were rapid techniques, two qualitative and one quantitative approach among them. The rapid methods were chosen, because these tests had been previously used by the laboratory personnel. Every isolated oxidase-negative colony forming unit from each test (i.e., four colonies per sample) was biochemically identified to discriminate between E. coli and other coliforms.

DIN EN ISO 9308-1 standard method (1997)

After filtering a 100 mL sample, the membrane filter was incubated at 36°C±2°C for 21±3 h on Lactose-TTC-Agar (Lactose-TTC-Agar Basis; Merck). Lactose-positive colonies (yellow colonies) were counted and subcultivated on blood agar (oxidase test [Blood agar; Merck]) and Gassner agar (Merck; indol test). After incubation (36°C±2°C for 21±3 h), the oxidase-negative (colorless) colonies of the blood agar were regarded as coliform bacteria. Those colonies that were additionally positive (red) on Gassner agar (incubation at 44°C±0.5°C for 21±3 h) were regarded as E. coli. All oxidase-negative colonies were biochemically analyzed by Enterotube II (Becton Dickinson).

Colilert-18 test (Idexx)

This rapid test is a qualitative approach and requires incubation of 100 mL samples in one capsule of enrichment medium (defined nutrient substrate containing o-nitrophenol-β-galactopyranosid and MUG; 35°C±2°C, 18 h) followed by fluorescence test (365 nm) of yellowy samples (in comparison to control sample). Fluorescence-positive samples were regarded as E. coli. Yellowy samples were regarded as coliforms and biochemically analyzed by Enterotube II (Becton Dickinson) in order to specify additional E. coli more precisely.

Presence/absence test broth with MUG (Hach)

This test is also a qualitative fluorescence test. Samples (100 mL) were incubated in one sterile bottle with Presence-Absence Bouillon (Hach; 35°C±2°C for 24–48 h). The fluorescence test (365 nm) was carried out with yellow or yellowy-brown samples (=coliforms). Positive samples were regarded as E. coli. All coliforms were biochemically identified by Enterotube II (Becton Dickinson).

m-Endo broth test (Hach)

Samples (100 mL) were filtrated (45–47 μm membrane), and the membrane filter was placed on m-Endo Agar (PourRite™ m-Endo Broth Ampules) in a Petri dish. After incubation (35°C±2°C for 24–48 h), lactose-positive colonies were counted (red: coliform bacteria; red including Fuchsine brilliance: E. coli) and biochemically identified by Enterotube II (Becton Dickinson).

Detection of enterococci

The standard method is DIN EN ISO 7899-2 and fulfils the requirements of the German Drinking Water Regulation (2001). The rapid method is a qualitative approach.

DIN EN ISO 7899-2 standard method (2000)

After filtration of 100 mL samples, the membrane filter was incubated on enterococci selective Slanetz and Bartley agar (36°C±2°C for 44±4 h; Merck). All red-, brown-, or pink-colored colonies were counted and regarded as enterococci. To confirm the test, filters were placed on choler-aesculin agar (Oxoid) and incubated for 2 h at 44°C±0.5°C. The yellowy-brown or black-colored colonies were then regarded as intestinal enterococci.

Enterolert test (Idexx)

Samples (100 mL) were incubated with Enterolert at 41°C±0.5°C for 24 h. Positive samples in fluorescence test at 365 nm were regarded as positive for enterococci. To confirm the test, filters were placed on choler-aesculin agar (Oxoid) and incubated for 2 h at 44°C±0.5°C. The yellowy-brown or black-colored colonies were then regarded as intestinal enterococci.

Statistical analyses

Data were stored in MS Excel 2003 (Microsoft). Analyses were carried out by SAS 9.1 TS Level 1M3 (SAS Institute, Inc.). Accordance of results of different methods was tested by McNemar test (Procedure PROC FREQ). The level of significance was 0.05. Further, Kappa coefficient was determined including the 95% confidence limits.

Results

The samples investigated by all tests are displayed in Table 1 (number of samples n=499), the samples analyzed only by the standard test, Colilert-18 and P/A test with MUG (n=573) are shown in Table 2. E. coli was found in 20.84% of 499 water samples, when investigated by the standard procedure (Table 1). The tests Colilert-18, P/A test with MUG and m-Endo Broth detected less positive samples than the standard method (between 19.64 and 20.04% positive samples) (Table 1). Nevertheless, the biochemical analysis showed that 114 or 115 samples, respectively, analyzed by a rapid test contained E. coli, while only 104 samples were positive for E. coli by the standard test. Considering that this method does not produce false positive results, it would appear that some E. coli died during the isolation procedure following the DIN EN ISO 9308-1. Differences between the specific reactions and the biochemical analysis were due to misclassification of E. coli as coliforms by the rapid test. Therefore, no differences occurred between the specific reactions and the biochemical analysis of coliforms.

Table 1.

Detection of Escherichia coli and Coliform Bacteria in Water Samples from Kosovo with Four Different Methods (n =499)

		Escherichia coli		Coliforms
Method		n	%	n	%
German Drinking Water Regulation	Specific reaction	104	20.84	153	30.66
	Biochemical analysis	104	20.84	153	30.66
Colilert^®-18	Specific reaction	100	20.04	155	31.06
	Biochemical analysis	115	23.05	155	31.06
P/A test with MUG	Specific reaction	100	20.04	151	30.26
	Biochemical analysis	114	22.85	151	30.26
m-Endo Broth	Specific reaction	98	19.64	154	30.86
	Biochemical analysis	115	23.05	154	30.86

n, number; %, percentage of positive results; MUG, 4-methylumbelliferyl-β-d-glucoronid.

Table 2.

Comparison of Three Rapid Methods with the Standard Method (German Drinking Water Regulation)

	Method	E. coli (specific reactions)	E. coli (biochemical analysis)	Coliforms
Colilert-18	Concordance	93.59	97.21	98.95
n=573	False negative	1		2
	False positive	15		4
	Kappa	83.91	92.05	97.58
	CI	78.42–89.39	88.22–95.88	95.66–99.51
	p-value	0.8575	0.0005^a	0.4142
P/A test with MUG	Concordance	94.76	97.04	98.43
n=573	False negative	2		6
	False positive	15		3
	Kappa	84.38	91.53	96.35
	CI	78.96–89.80	87.57–95.48	93.98–98.72
	p-value	0.7150	0.0016^a	0.3173
m-Endo Broth	Concordance	91.98	97.39	98.60
n=499	False negative	1		3
	False positive	12		4
	Kappa	76.05	92.40	96.71
	CI	69.00–83.09	88.33–96.47	94.28–99.13
	p-value	0.5271	0.0023^a	0.7055

Concordance: percentage of concordant results; false negative: number of samples negative in rapid test, but positive in standard test; false positive: number of samples positive in rapid test, but negative in standard test; Kappa: Kappa coefficient; CI: 95% confidence interval; p-value: p-value of McNemar's test.

Statistically significant, p<0.05.

Regarding coliforms, the rapid tests produced almost the same results as the standard method (Table 1). The analysis showed that Colilert-18 and m-Endo Broth classified more samples positive for coliforms than the DIN EN ISO 9308-1.

Looking for sample-based similarity the statistical test procedure showed significant differences between the standard and each rapid procedure regarding the biochemical analysis (Table 2). No differences could be detected concerning the specific reactions of E. coli and concerning coliforms. Concordance exceeded 90% in every comparison, and the Kappa coefficients were substantially high.

Enterococci were detected in 92 (Enterolert) or 93 (German Drinking Water Regulation) samples, respectively (Table 3). This resulted in a concordance of 99.48% and a Kappa coefficient of 98.07.

Table 3.

Detection and Comparison of Enterococci in Drinking Water Samples from Kosovo with Two Different Test Systems (n=573)

	Enterococci
Method	n	%	Statistical tests
German Drinking Water Regulation	93	16.23	99.48^a
			2^b
			1^c
			98.07^d
Enterolert™	92	16.06	95.88–1.00^e
			0.5637^f

n, number; %, percentage of positive results.

Concordance.

False negative.

False positive.

Kappa.

CI.

p-Value.

Discussion

Detection rates for E. coli and coliform bacteria of three rapid methods were compared with the results of the standard procedure following DIN EN ISO 9308/1 (1997). With the Colilert-18 method, a higher detection frequency was achieved for E. coli and coliforms than with the standard method. Based on the method-specific reactions, the Colilert-18 method yielded false-negative results in the detection of E. coli in 2.79% of all samples, which correspond to the findings of other authors (Fricker et al., 1996; Schets et al., 2002; Chao et al., 2004; Bernasconi et al., 2006). Nevertheless, these false-negative samples in the detection of E. coli were regarded as unsuitable as well, because they were classified as coliform positive. In contrast, the standard method showed 15 false-negative results for E. coli.

By biochemical analysis, more samples could be identified positive for E. coli by using Colilert-18 than by the standard method. Probably, more pathogens died during the standard isolation process than during the alternative procedure. The misclassification of E. coli as coliforms by some Colilert-18 tests can be tolerated, because samples are regarded as unsuitable, if one of these microorganisms is detected. The statistically significant differences between the standard and Colilert-18 are due to the fact that more positive samples were identified by Colilert-18. Similarly, the Colilert-18 detected more coliform-positive samples than the DIN EN ISO 9308/1.

Considering our results and the findings of other authors (Fricker et al., 1996; Eckner, 1998; Schets et al., 2002; Yakub et al., 2002; Chao et al., 2004; Macy et al., 2005; Bernasconi et al., 2006), the use of Colilert-18 detecting coliforms and E. coli is at least equivalent to the standard method of the German Drinking Water Regulation.

The results of P/A test with MUG correspond to those of Colilert-18. The 16 false-negative samples of E. coli were all identified as coliform bacteria and therefore excluded from human consumption. The P/A test with MUG detected more positive samples for E. coli than the standard method did. Therefore, it can be considered that P/A test with MUG detects E. coli and coliforms at least as reliable as the standard method.

Horman and Hanninen (2006) established a sensitivity of 83.6% for m-Endo Broth. In our study, 82.7% of positive samples (standard method) were declared positive by m-Endo Broth. This test had fewer concordant results regarding the specific reactions of E. coli, but not regarding coliforms.

Considering our own results and those of other authors (Sassen, 1999; Horman and Hanninen, 2006), the usage of m-Endo Broth to detect coliforms and E. coli shows as reliable results as the standard method DIN EN ISO 9308/1, but reduced reliability compared with Colilert-18 and P/A test with MUG.

Because of the substantial agreement when detecting enterococci (99.48% concordance), Enterolert can be regarded as an equivalent alternative to the standard method DIN EN ISO 7899/2. Some authors published similar results (Eckner, 1998; Yakub et al., 2002), but Kinzelman et al. (2003) assessed a weaker reliability of Enterolert.

Conclusions

In conclusion, all of the alternative methods tested produced results that were at least equivalent to those obtained by the standard method. Concerning the biochemical analysis for differentiation of E. coli and coliform pathogens (Enterotube II), the method of the Drinking Water Regulation produced less-positive results than the alternative methods.

In the evaluation of water samples based on the results obtained by the alternative methods, no water sample was deemed fit for human consumption, which had been classified as unfit for human consumption by the standard method. The study results indicate that the standard method could have produced false-negative results, which were recognized by the rapid tests. Consumer safety might even be improved by using the alternative methods.

Based on the tests carried out and the results obtained, the use of Colilert-18 and Enterolert—and to a lesser extent m-Endo Broth—for the testing of drinking water in crisis-afflicted and disaster areas is recommended.

Footnotes

Disclosure Statement

No competing financial interests exist.

References

Bernasconi

, Volponi

, Bonadonna

. Comparison of three different media for the detection of E. coli and coliforms in water. Water Sci Technol, 2006; 54:141–145.

BGBl

. [German law to prevention and control of human infectious diseases (infection protection law).] 2000.

BGBl. I. [German Regulation on the quality of water for human use.] 2001.

BGBl. I. [Notification of the amendment of the German food and feed code] 2006.

Cabelli

. Microbial indicator system for assessing water quality. J Microbiol, 1982; 48:613–618.

Chao

, Chao

. Evaluation of Colilert-18 for detection of coliforms and Eschericha coli in subtropical freshwater. Appl Environ Microbiol, 2004; 70:1242–1244.

Council Directive 98/83/EC of 3 November 1998 on the quality of water intended for human sonsumption. Communities

. Official Journal of the Euopean Communities, 1998; 32–54.

DIN EN I.S.O. 7899/2. Water quality—detection and enumeration of intestinal enterococci. Part 2: Membrane filtration method, 2000.

DIN EN I.S.O. 9308/1. Water quality—detection and enumeration of coliform organisms, thermotolerant coliform organisms and presumptive Escherichia coli. Part 1: membrane filtration method, 1997.

10.

Eckner

. Comparison of membrane filtration and multiple-tube fermentation by the colilert and enterolert methods for detection of waterborne coliform bacteria, Escherichia coli, and enterococci used in drinking and bathing water quality monitoring in southern sweden. Appl Environ Microbiol, 1998; 64:3079–3083.

11.

Edberg

, Allen

, Smith

. National field evaluation of a defined substrate method for the simultaneous enumeration of total coliforms and Escherichia coli from drinking water: comparison with the standard multiple tube fermentation method. Appl Environ Microbiol, 1988; 54:1595–1601.

12.

[EPA] United States Environmental Protection Agency. Total Coliform Rule (TCR) 54 FR 27544-27568, 54:124. 1989.

13.

Feng

PCS

, Hartman

. Fluorogenic assays for immediate confirmation of Escherichia coli. Applied and environmental microbiology, 1982; 43:1320–1329.

14.

Fricker

, Illingworth

, Fricker

. Use of 2 formulations of colilert and quantitray for assessment of the bacteriological quality of water. Wat Res, 1996; 31:2495–2499.

15.

Horman

, Hanninen

. Evaluation of the lactose Tergitol-7, m-Endo LES, Colilert 18, Readycult Coliforms 100, Water-Check-100, 3M Petrifilm EC and DryCult Coliform test methods for detection of total coliforms and Escherichia coli in water samples. Water Res, 2006; 40:3249–3256.

16.

Kinzelman

, Ng

, Jackson

, Gradus

, Bagley

. Enterococci as indicators of Lake Michigan recreational water quality: comparison of two methodologies and their impacts on public health regulatory events. Appl Environ Microbiol, 2003; 69:92–96.

17.

Macy

, Dunne

, Angoran-Benie

et al. Comparison of two methods for evaluating the quality of stored drinking water in Abidjan, Cote d'lvoire, and review of other comparisons in the literature. J Water Health, 2005; 3:221–228.

18.

Sassen

. [Microbiological Water Examination: Performance of Alternative for Mobile Use of Suitable Examination-Methods Compared to the Procedures Described in the German Law for Driniking Water “Trinkwasserverordnung.”] Tierärztliche Hochschule. Hannover: Department of Food Quality and Safety, 1999.

19.

Schets

, Nobel

, Strating

, Mooijman

, Engels

, Brouwer

. EU drinking water directive reference methods for enumeration of total coliforms and Escherichia coli compared with alternative methods. Lett Appl Microbiol, 2002; 34:227–231.

20.

World Health Organization. Guidelines for Drinking-Water Quality. Geneva: World Health Organization, 2004.

21.

Yakub

, Castric

, Stadterman-Knauer

et al. Evaluation of Colilert and Enterolert defined substrate methodology for wastewater applications. Water Environ Res, 2002; 74:131–135.