Evaluation of chromogenic technologies for use ... - Wiley Online Library

Journal of Applied Microbiology ISSN 1364-5072

ORIGINAL ARTICLE

Evaluation of chromogenic technologies for use in Australian potable water G. Hallas*, S. Giglio*, V. Capurso, P.T. Monis and W.L. Grooby Australian Water Quality Centre, SA Water Corporation, Salisbury, SA, Australia

Keywords chloramine, chromogenic, Colilert-18, water. Correspondence Steven Giglio, Australian Water Quality Centre, SA Water Corporation, Private Mail Bag 3, Salisbury, SA 5108, Australia. E-mail: [email protected]

*These authors contributed equally to this study. 2007 ⁄ 0972: received 20 June 2007, revised 10 March 2008 and accepted 11 March 2008 doi:10.1111/j.1365-2672.2008.03842.x

Abstract Aims: To compare the use of MI agar, Membrane Lactose Glucoronide Agar (MLGA), CM1046 agar and Colilert-18 (Defined Substrate Technology, IDEXX Laboratories Pty. Ltd., Sydney) on Australian potable water. Methods and Results: Both potable (n = 369) and nonpotable waters (n = 35) were analysed by membrane filtration using chromogenic agars as well as Colilert-18 over a period of 12 months. Recoveries of stressed organisms on these chromogenic media were also investigated. Agar-based chromogenic technologies compared favourably to Colilert-18 for chlorinated waters, but there are possible limitations when using these agars for chloraminated waters. Additionally, the breakthrough of problematic organisms, especially oxidase positive organisms, may lead to misrepresentation or over-estimation of E. coli and total coliforms, particularly on MLGA and CM1046. The recovery of stressed organisms was favoured in the Colilert-18 system when compared to chromogenic agars. Conclusions: MI agar performed better than the other chromogenic agars with respect to recovery and colour identification and discrimination of organisms, and compared favourably with Colilert-18. The use of chromogenic agars in chloraminated waters should be done cautiously. Significance and Impact of the Study: This study provides comparison data for laboratories looking to adopt chromogenic technologies, and is especially important for Australian laboratories wanting to uptake the use of MI agar (as used in USEPA method 1604) for routine use and for gaining accreditation. Additionally, to the best of our knowledge, this is the first reported evaluation of these agars in chloraminated waters and is especially timely as the use of this disinfection agent is increasing.

Introduction The use of Escherichia coli (E. coli) as a primary indicator for faecal contamination and Total Coliforms (TC) as a secondary indicator of water quality and system performance continues to underpin monitoring programs worldwide to minimise public health risks (Smith et al. 1973; Evanson and Ambrose 2006). In Australia, this is embodied in the Australian Drinking Water Guidelines (ADWG) (http://www.nhmrc.gov.au/publications/synopses/ eh19syn.htm) that is prescribed under various regulatory acts in different states and territories. Water quality 1138

incidents, or detection of E. coli or TC specifically, require swift notification and remedial action, and rapid technologies for the detection of these indicator organisms are therefore of great interest to water utilities, laboratories and public health authorities. Recently in the United States and United Kingdom, and more recently in Australia, chromogenic substrate technologies have been accepted as standard methods (Brenner et al. 1996; Oshiro 2002; Anon., 2005). The United States Environmental Protection Authority (USEPA) method uses membrane filtration (MF) and MI agar (BD DifcoTM, Becton Dickinson (BD) Diagnostics,

ª 2008 The Authors Journal compilation ª 2008 The Society for Applied Microbiology, Journal of Applied Microbiology 105 (2008) 1138–1149

G. Hallas et al.

USA) and the Drinking Water Inspectorate in the United Kingdom has approved the use of membrane lactose glucuronide agar (MLGA) (Oxoid, UK) (Sartory and Howard 1992). These technologies use synthetic substrates for specific enzymatic reactions (such as b-glucosidase or b-glucuronidase activity) and rely primarily upon cleavage of the conjugated chromogen or fluorogen for discrimination, rather than phenotypic traits such as fermentation or the production of gas for organism confirmation. MI requires no confirmation, whereas MLGA requires any suspect coliforms or E. coli to be confirmed by traditional methods. Uptake of these methods by water utilities is traditionally slow, mainly due to the lack of specific data related to evaluation in respective countries. In Australia, data are only available for Colilert-18 (Defined Substrate Technology, IDEXX Laboratories Pty. Ltd., Sydney) (Adcock and Saint 1997), limiting the choices available for alternative technologies to be implemented for routine use. Further comparative studies may assist Australian water authorities in the expanded routine use of these alternative methods, which have already been rigorously tested by the USEPA and UK DWI (Drinking Water Inspectorate) (Sartory and Howard 1992; Oshiro 2002). Our laboratory has recently compared four defined chromogenic technologies (Colilert-18, MI agar, MLGA and CM1046 (Oxoid, UK)) against a variety of waters from chlorinated and chloraminated distribution systems. With the advent of automated colony counters and equipment to process water by membrane filtration, our main focus was to intensively examine these technologies for drinking water (and nonpotable water to a limited extent), as an alternative to Colilert-18. These chromogenic technologies were evaluated over 12 months to include any seasonal variation in bacterial communities according to criteria defined in ISO ⁄ TR 13843 (ISO ⁄ TR, 2000), AS ⁄ NZS 4659:1999 parts 2 and 3 (Anon. 1999a,b). These latter documents are used for guidance of microbiological method validation in Australia. Additionally, statistics as recommended by ISO 17994:2004 ‘Water Quality – Criteria for establishing equivalence between microbiological methods’ (ISO, 2004) were employed to determine the equivalency between chromogenic technologies, with a ‘D’ value set at 10% for potable water as stipulated in section 4. Materials and methods Potable and nonpotable water sampling A total of 404 randomly selected routine samples were collected from various sites in South Australia over a 12 month period. Potable samples included chlorinated

Evaluation of chromogenic technologies for Australian water

(n = 183) and chloraminated (n = 186) supplies, and nonpotable samples consisted of surface water (n = 10), waste water (n = 10), groundwater bores (n = 6) and tanks (n = 9). Potable supplies that did not have either E. coli or coliform results were re-spiked using environmental wild-type strains to give counts in the range of 1–80 CFU 100 ml)1, as suggested in section 5.2 of ISO 17994:2004. The nonpotable supplies were primarily included for analysis of chromogenic colour reactions for comparison with manufacturer’s specifications. All samples were analysed by all chromogenic technologies as described below. Samples were collected in sterile polypropylene bottles dosed with sodium thiosulphate (100 mg l)1) according to AS ⁄ NZS 5667.1:1998 and AS ⁄ NZ 2031:2001 (Anon. 1998, 2001), transported on ice and analysed within 6 h of collection. Preparation of water samples for the determination of the limit of detection for chromogenic technologies American Type Culture Collection (ATCC) or environmental wild-type isolates were obtained for the following: Enterobacter aerogenes (ATCC 13048), E. coli (ATCC 11775), Klebsiella ozaenae, Citrobacter braakii, Pantoea sp., Serratia marcescens and three additional strains of E. coli. Cultures were grown overnight at 35C in brain heart infusion broth (Oxoid, UK) and enumerated by heterotrophic colony counts using R2A agar and incubated at 35C for 48 h. Overnight cultures were consistently found to have levels of 109–1010 CFU ml)1. Each of the broths was serially diluted in physiological saline and then added to filter sterilised potable water to obtain a final spiked concentration of approximately 1–20 CFU 100 ml)1, 20–80 CFU 100 ml)1 and 80–500 CFU 100 ml)1. Samples (n = 63) were analysed in triplicate by all chromogenic technologies. Determination of the recovery of stressed organisms in chlorinated and chloraminated water supplies South Australian potable waters are supplied as filtered chlorinated, unfiltered chlorinated, filtered chloraminated, and unfiltered chloraminated water. For the investigation of the chromogenic technologies to isolate stressed organisms from these different matrices the protocol of Adcock and Saint (Adcock and Saint 1997) was followed to recover 20–80 colonies per 100 ml on the media, the only exception being that 90 s of contact time was used for chloraminated water instead of 60 s. Determination of chlorine residual levels pre and post addition of sodium thiosulphate were performed by titrating N,N-diethylp-phenylene (DPD) with ferrous ammonium sulphate (FAS) at pH 6Æ2–6Æ5. Organisms used in these investigations were both ATCC and wild-type E. coli and


1139


Enterobacter cloacae (as above). All analyses (n = 96) were performed in triplicate. Colilert-18 Samples and dilutions were added to Colilert-18 sterile vessels, and Colilert-18 powders were aseptically added and mixed by shaking until completely dissolved. The vessel was aseptically emptied into a Quanti-Tray and heat-sealed according to manufacturer’s instructions, then incubated at 35C for 18 h. The Quanti-Tray was inspected for yellow wells indicating the presence of coliforms, and under long wave UV (366 nm) light for fluorescent wells indicative of E. coli. Estimated counts were determined using most probable number tables supplied by the manufacturer. Chromogenic agars All chromogenic agars were prepared in accordance with the respective manufacturer’s instructions in 90 mm petri dishes. Samples were filtered through 0Æ45 lm sterile filters (Millipore, USA) according to Standard Methods for the Examination of Water and Wastewater method 9222B (Eaton et al. 2005). All agar plates were incubated for 24 h at 37C according to manufacturers’ instructions, before being examined under natural and long-wave UV (366 nm) light. All colonies on MI agar (BD DifcoTM, Becton Dickinson (BD) Diagnostics, USA) were counted as per manufacturer’s instructions and the USEPA Method 1604 guidelines for interpretation (Oshiro 2002). In brief, blue ⁄ green fluorescent colonies were designated E. coli, and white fluorescent colonies were designated coliforms unless otherwise indicated. For MLGA, colonies were examined in accordance with manufacturer’s instructions and the United Kingdom Drinking Water Inspectorate guidelines (Anon. 2002). All yellow colonies (however faint and irrespective of size) were recorded as TC and green colonies were recorded as E. coli. For CM1046 – chromogenic E. coli ⁄ coliform selective agar (Oxoid, UK), typical pink colonies were designated as coliforms and blue ⁄ purple colonies as E. coli. For all agars both typical and atypical colonies were investigated. Typical and atypical colonies were subcultured to Cysteine Lactose Electrolyte Deficient (CLED, Oxoid, UK) agar with andrade indicator and incubated at 35C for 24 h. Pure isolates were then identified using API20E (bioMerieux, France), unless otherwise indicated. Chromogenic colony colour investigations Individual chromogenic technologies and their colour discrimination specifications (according to manufacturers’ 1140

G. Hallas et al.

instructions) were assessed by selecting representative colonies isolated from all water matrix types and comparing biochemical identification (API 20E) with specified colour reactions. This study concentrated on organism counts that are typically less than 80 CFU 100 ml)1 in selected water types, and as such did not have a large number of other bacteria to raise issues of reliable colour discrimination from overcrowding, or the out-competing of recalcitrant bacteria. Issues with problematic organisms and the effects on colony colours and overcrowding of bacteria have been discussed in previous studies (Sartory and Howard 1992; Brenner et al. 1993, 1996; Buckalew et al. 2006). Statistical analysis The recovery data in CFU per 100 ml)1 for all media were tested for normality by using the Shapiro-Wilk test. Data sets that were not normally distributed were analysed using nonparametric Kruskal-Wallis analysis of variance (KW-anova) with the Dunn’s post hoc test for significance, and Spearman-Rank correlation coefficient (rs). Normally distributed data were analysed using anova with Tukey’s post hoc analysis and Pearson’s correlation coefficient (r). Agreements of methods were analysed using Passing and Bablok analysis with the CUSUM (cumulative summation) test for linearity. A CUSUM value of P > 0Æ1 indicated agreement between methods. All statistical analyses were performed using Analyse-it for Microsoft Excel (http://www.analyse-it.com) and GraphPad Prism Version 4.03 for Windows (GraphPad Software, San Diego, CA, USA). Additionally, the relative mean difference test outlined in ISO 17994:2004 was applied to the data set. Results Evaluation of chromogenic technologies on potable and non potable water systems – field trials A combined total of 369 split potable water samples were collected and analysed for TC and E. coli. The chlorinated supply (n = 183) and chloraminated supply (n = 186) required a large percentage of negative results to be re-spiked in accordance with criteria defined in ISO 17994:2004 section 5.2. The resultant data set, including the re-spiked samples (n = 295), were analysed using the mean relative difference to compare the relative performance of the trial method being chromogenic agars and the reference method Colilert-18. MI, MLGA and CM1046 agars and Colilert-18 were ‘not different’ (–10% £ xL £ 0 and xH > 0) for E. coli and coliform counts with the lower and higher confidence intervals of


G. Hallas et al.


Table 1 Mean relative difference analysis (Trial Method – Reference Method) of paired sample results (CFU 100 ml)1) from the Trial Methods (Chromogenic agars) and the Reference Method (Colilert-18) for drinking water samples analysed according to ISO17994:2004 (ISO 2004) MI (n = 295)

Mean relative difference SD U (expanded uncertainty) XL XH Outcome

MLGA (n = 295)

CM1046 (n = 295)

E. coli

Coliforms

E. coli

Coliforms

E. coli

Coliforms

)4Æ71 29Æ03 3Æ38 )8Æ09 20Æ94 Not different






the expanded uncertainty around the mean (xL and xH) for E. coli and coliforms summarised in Table 1. Further statistical analysis of the data in CFU per 100 ml)1 did not yield any significant differences (anova P > 0Æ05) between any of the chromogenic technologies when comparing specific drinking water (chlorinated and chloraminated) samples (Table 2). Pearson coefficient indicated very good correlation (r = 0Æ87–0Æ99) between the counts and all the methods tested. Passing–Bablok analysis demonstrated close agreement between the methods (chlorinated potable supply regression lines: MI, CM1046 and MLGA for E. coli and TC; y = x and y = 0Æ7391x, y = 1Æ25x and y = 1Æ3478x; y = 2Æ5x and y = x. chloraminated potable supply regression lines: MI, CM1046 and MLGA for E. coli and TC; y = x and y = x, y = x and y = 1Æ1111x; y = x and y = 1Æ3333x) and the CUSUM test for linearity showed that the observed slopes were not significantly different from the perfect-fit lines for all methods tested (P > 0Æ1). Analysis of nonpotable water matrices (surface water, wastewater, groundwater bores and tanks, n = 35, Table 3) consisted of a data set that was too small to apply statistics according to criteria defined in ISO 17994:2004. These water types were primarily included in the study for reliable colour discrimination data, however a preliminary analysis using traditional statistics failed to yield any significant differences (KW-anova, P > 0Æ05) between or within any of the matrix types. Spearman rank coefficients indicated the lowest correlation results occurred between MLGA vs CM1046 and CM1046 vs

Colilert-18 (rs = 0Æ67; rs = 0Æ66) in the detection of TC for municipal long term storage supply tanks. These results were in contrast to the correlation between MI and MLGA agars when compared to Colilert-18 for the same matrices (rs = 0Æ87; rs = 0Æ95). Passing and Bablok analysis of the data indicated poor to moderate agreement between all the methods, and is likely due to the small data set investigated. Further work would be required to reliably compare CM1046 on nonpotable matrices in Australian waters, and previous studies have covered Colilert-18, MI agar, and MLGA agar on other waters (Sartory and Howard 1992; Brenner et al. 1993, 1996). Evaluation of the limit of detection of chromogenic technologies Spiked organism data were evaluated using statistical analysis recommended by ISO ⁄ TR 13843 and considered outside the scope of ISO 17994:2004. Detection levels of E. coli and TC in potable water supplies are often less than 20 CFU 100 ml)1 and therefore this low level recovery of these spiked organisms in the chromogenic technologies was evaluated in detail. The CFU per 100 ml recovery counts were tabulated for the low detection limit (1–20 CFU 100 ml)1) (Table 4). The resultant spike levels were less than 20 CFU 100 ml)1 for each of the organisms used and any count of 0 CFU 100 ml)1 was removed from the data set. No significant differences (anova, P > 0Æ05) in the limit of detections were evident between all of the technologies. MLGA agar showed the

Table 2 Comparison of E. coli and TC counts in chlorinated and chloraminated drinking water samples of MI, MLGA and CM1046 chromogenic agars against Colilert-18 MI Drinking water type

E. coli

MLGA TC

Significance level (correlation coefficient) of E. coli and TC Chlorinated >0Æ05 (0Æ99) >0Æ05 (0Æ92) Chloraminated >0Æ05 (0Æ99) >0Æ05 (0Æ99)

CM1046

E. coli

TC

E. coli

TC

>0Æ05 (0Æ99) >0Æ05 (0Æ99)

>0Æ05 (0Æ93) >0Æ05 (0Æ99)

>0Æ05 (0Æ99) >0Æ05 (0Æ99)

>0Æ05 (0Æ87) >0Æ05 (0Æ96)


1141


G. Hallas et al.

Table 3 Recovery counts (CFU 100 ml)1) for nonpotable water matrices Water matrix

Method

n

Organism

Mean

SD

SE

Surface water

MI Agar

10

CM1046

10

MLGA

10

Colilert-18

10

MI Agar

10

CM1046

10

MLGA

10

Colilert-18

10

E. coli Coliform E. coli Coliform E. coli Coliform E. coli Coliform E. coli Coliform E. coli Coliform E. coli Coliform E. coli Coliform E. coli Coliform E. coli Coliform E. coli Coliform E. coli Coliform E. coli Coliform E. coli Coliform E. coli Coliform E. coli Coliform

98Æ00 4632Æ00 81Æ00 8350Æ00 66Æ17 10 550Æ00 90Æ67 5132Æ00 4Æ30 321Æ70 4Æ70 602Æ50 4Æ20 1212Æ00 5Æ40 52Æ40 16Æ66 26Æ83 13Æ33 24Æ16 1Æ67 21Æ66 4Æ83 24Æ83 0Æ00 13Æ33 0Æ00 6Æ56 0Æ00 7Æ67 0Æ00 8Æ89

153Æ20 9513Æ00 143Æ50 18 470Æ00 111Æ60 22 790Æ00 170Æ00 11 700Æ00 9Æ09 754Æ60 9Æ93 1723Æ00 9Æ30 2962Æ00 11Æ53 85Æ08 40Æ82 39Æ33 32Æ65 38Æ63 4Æ08 38Æ90 11Æ83 40Æ45 – 13Æ75 – 6Æ98 – 7Æ91 – 8Æ88

62Æ54 3883Æ00 58Æ57 7541Æ00 45Æ57 9306Æ00 69Æ42 4776Æ00 2Æ88 238Æ64 3Æ14 545Æ00 2Æ94 936Æ90 3Æ65 26Æ90 16Æ66 16Æ06 13Æ33 15Æ77 1Æ67 15Æ88 4Æ83 16Æ51 – 4Æ59 – 2Æ33 – 2Æ64 – 2Æ96

Waste water

Groundwater bores

Tanks

MI Agar

6

CM1046

6

MLGA

6

Colilert-18

6

MI Agar

9

CM1046

9

MLGA

9

Colilert-18

9

highest correlation coefficients for spiked E. coli and TC (r = 0Æ97 and 0Æ97 respectively) followed by MI (r = 0Æ87 and 0Æ99) and CM1046 agars (r = 0Æ6 and 0Æ99) when comparing individual technologies to Colilert-18. Passing and Bablok analysis showed good agreement (Fig. 1) between all of the technologies at the low end detection limit, with CUSUM values >0Æ1 for all. Further count ranges were investigated at 20–80 CFU 100 ml)1, which according to AS ⁄ NZS 4276.1 and USEPA Method 1604 represents the optimum count range for membrane filtration techniques (Table 4). The high end range 80– 500 CFU 100 ml)1 required serial ten-fold dilutions onto the chromogenic agars, however Colilert-18 ⁄ 2000 trays do not require dilutions and have an estimated count range of 0–2400 CFU 100 ml)1. Colilert-18 recovered significantly more TC and E. coli (anova P < 0Æ05) at the high end of the limits of detection compared to the MI, MLGA and CM1046 agars. No significant differences (anova P > 0Æ05) could be detected in spiked potable supply samples at the low and mid range detection limits 1142

95% CI of mean )62Æ76 )5351Æ00 )69Æ55 )11 040Æ00 )50Æ98 )13 370Æ00 )87Æ78 )7144Æ00 )2Æ21 )218Æ10 )2Æ41 )630Æ40 )2Æ45 )906Æ80 )2Æ86 )8Æ47 )26Æ17 )14Æ44 )20Æ94 )16Æ38 )2Æ62 )19Æ16 )7Æ59 )17Æ62 – 2Æ76 – 1Æ19 – 1Æ59 – 2Æ06

258Æ80 14 610Æ00 231Æ50 27 740Æ00 183Æ30 34 470Æ00 269Æ10 17 410Æ00 to 10Æ805 to 861Æ5 to 11Æ806 to 1835 to 10Æ84 to 3332 to 13Æ65 to 113Æ2 to 59Æ51 to 68Æ11 to 47Æ60 to 64Æ71 to 5Æ951 to 62Æ49 to 17Æ258 to 67Æ29 – to 23Æ90 – to 11Æ92 – to 13Æ74 – to 15Æ71

during the limit of detection trials. MI agar showed the highest correlation (r = 0Æ91–0Æ99) followed by MLGA and CM1046 (r = 0Æ86–0Æ99 and r = 0Æ78–0Æ99 respectively) when comparing individual technologies to Colilert-18 (unpublished data). Recovery of stressed micro-organism in chlorinated samples Stressed organism data were evaluated using traditional statistical analysis recommended by ISO ⁄ TR 13843 and considered outside the scope of ISO 17994:2004. In filtered chlorinated water, TC counts were overall slightly decreased for the chromogenic agar technologies when compared to Colilert-18. These differences for TC and E. coli were not statistically significant (anova, P > 0Æ05) and there was good correlation (r = 0Æ90–0Æ96) between all of the technologies for the filtered chlorinated supply. Passing and Bablok analysis showed good agreement (spiked stressed chlorinated regression lines: MI, CM1046


G. Hallas et al.


Table 4 Recovery counts (1–20, 20–80 and 80–500 CFU 100 ml)1) for low level detection limit ATCC and environmental spiked E. coli and TC organisms Water matrix

Method

n

Organism

Mean

SD

SE

ATCC and environmental spiked organisms (1–20 CFU 100 ml)1)

MI Agar

6 15 6 15 6 15 6 15 6 15 6 15 6 15 6 15 6 15 6 15 6 15 6 15

E. coli Coliform E. coli Coliform E. coli Coliform E. coli Coliform E. coli Coliform E. coli Coliform E. coli Coliform E. coli Coliform E. coli Coliform E. coli Coliform E. coli Coliform E. coli Coliform

2Æ67 1Æ93 1Æ83 1Æ00 2Æ83 2Æ40 3Æ00 2Æ60 49Æ08 30Æ33 53Æ42 30Æ67 52Æ50 31Æ00 60Æ25 47Æ67 287Æ92 184Æ00 317Æ33 179Æ00 283Æ17 197Æ33 405Æ00 307Æ11



CM1046 MLGA Colilert-18


MI Agar CM1046 MLGA Colilert-18


MI Agar CM1046 MLGA Colilert-18

and MLGA for E. coli and TC; y = 1Æ2682x + 0Æ3659 and y = 0Æ962x + 23Æ605; y = 1Æ134x – 0Æ067 and y = 0Æ8571x + 29Æ5; y = 1Æ1351x + 0Æ9324 and y = 0Æ8649x + 27Æ659) with CUSUM values >0Æ1 for all of the technologies against Colilert-18. However in unfiltered chlorinated water, Colilert-18 recovered significantly more TC than MI, MLGA, and CM1046 agars, and differences in correlation coefficients were evident between these sample (anova, P < 0Æ05, r = 0Æ77; r = 0Æ81; r = 0Æ91, Table 5). Passing and Bablok analyses detected a significant bias (CUSUM, P < 0Æ1) for the detection of TC in Colilert18 compared against the MI, CM1046 and MLGA chromogenic agars in unfiltered chlorinated water (Regression lines: y = 14x – 1000, y =)1Æ6889x + 28904 and y = )2Æ9615x + 37519). Recovery of stressed micro-organism in chloraminated samples Statistical analyses of the data from filtered and unfiltered chloraminated water indicated that Colilert-18 performed significantly better than CM1046 and MI chromogenic agars (anova, P < 0Æ05), however MLGA recovery of E. coli organisms were not significantly different (anova, P > 0Æ05). MLGA agar showed the highest correlation coefficients for filtered and unfiltered chloraminat-

95% CI of mean 0Æ83 0Æ42 0Æ03 0Æ16 1Æ03 0Æ16 1Æ67 0Æ15 28Æ23 23Æ67 26Æ78 20Æ52 27Æ24 23Æ73 29Æ44 26Æ46 225Æ28 82Æ99 230Æ05 85Æ38 209Æ55 94Æ43 286Æ83 123Æ80

to 4Æ504 to 3Æ449 to 3Æ641 to 1Æ837 to 4Æ641 to 4Æ643 to 4Æ327 to 5Æ048 to 69Æ942 to 36Æ993 to 80Æ055 to 40Æ812 to 77Æ758 to 38Æ271 to 91Æ058 to 68Æ871 to 350Æ550 to 285Æ015 to 404Æ614 to 272Æ625 to 356Æ781 to 300Æ234 to 523Æ170 to 490Æ427

ed water (r = 0Æ66 and r = 0Æ98) for E. coli (Tables 6 and 7). Passing and Bablok analyses detected a significant bias for recovery towards Colilert-18 when compared to MI, CM1046 and MLGA agars in filtered chloraminated water (E. coli and TC regression lines for MI CM1046 and MLGA; y = 17Æ647x – 5682Æ4 and y = 3Æ4279x ) 19922; y = 13Æ723x – 11579 and y = 3Æ2061x – 19748; y = 7Æ1429x ) 5285Æ7 and y = 2Æ0195x ) 7993Æ7. CUSUM, P < 0Æ1) and also in unfiltered chloraminated samples (E. coli and TC regression lines for MI, CM1046 and MLGA; y = 2Æ9574x ) 9Æ5745 and y = )3Æ785x + 15957; y = 2Æ1067x + 82Æ667 and y = )4Æ8413x + 14614; y = 0Æ6728x + 658Æ01 and y = 7Æ3636x ) 6595Æ5, CUSUM, P < 0Æ1). Further work would be required to assess where the breakpoint relationship for linearity occurs between E. coli recovery counts on MLGA agar compared to Colilert-18 in these water matrices. Identification of colonies on the chromogenic technologies A total of 6833 representative organisms from all technologies were identified using API 20E (Table 8). According to criteria outlined in AS ⁄ NZS 4659:1999 parts 2 & 3, MI and MLGA were comparable to Colilert-18 in sensitivity (100%) and specificity (99%). Oxoid agar CM1046


1143


(a)

Identity line Y=X

6

G. Hallas et al.

(b) 16

Identity line Y=X

14

5 12 Colilert coliform

Colilert E. coli

4 3 2

10 8 6 4

1 2 0 –1 –1

(c)

0

y = 0·6667x + 0·8333 1

3 MI E. coli

–2 –2

5

Identity line Y=X

6

3

(d)

8 MI coliform

13

Identity line Y=X

16

5

14 12 Colilert coliform

4 Colilert E. coli

y = 1·8333x

3 2 1

10 8 6 4 2

0 –1 –1

y = 0·5x + 1·75

1

3 CM1046 E. coli

(e) 6

0 –2 –2

5

Identity line Y=X

(f)

3

16

8 CM1046 coliform

13

Identity line Y=X

14

5

12 Colilert coliform

4 Colilert E. coli

y = 2x

3 2 1

10 8 6 4 2

0 –1 –1

y = 0·8333x + 0·9167

1

3 MLGA E. coli

5

0 –2 –2

y = 1·0545x 3

8 MLGA coliform

13

Figure 1 Passing and Bablok1 method comparison of E. coli and TC recovery counts between MI (a & b), CM1046 (c & d) and MLGA (e & f) agars against Colilert-18 for low level spike detection limit (1–20 CFU 100 ml)1).

1144


G. Hallas et al.


Table 5 Recovery comparison of stressed E. coli and TC in unfiltered chlorinated spiked samples

MLGA E. coli Significance MI MLGA CM1046

Table 6 Recovery comparison of stressed E. coli and TC in filtered chloraminated spiked samples

E. coli

TC

level (correlation coefficient) >0Æ05 (0Æ99) >0Æ05 (0Æ98) – – – –

of E. coli and >0Æ05 (0Æ99) >0Æ05 (0Æ99) –

TC >0Æ05 (0Æ93) >0Æ05 (0Æ97) 0Æ05 (0Æ97) >0Æ05 (0Æ98) 0Æ05 (0Æ97) 0Æ05 (0Æ93) >0Æ05 (0Æ96) – – – –

of E. coli and >0Æ05 (0Æ96) >0Æ05 (0Æ98) –

TC >0Æ05 (0Æ94) 0Æ05 (0Æ78) >0Æ05 (0Æ92) >0Æ05 (0Æ66) >0Æ05 (0Æ79) – 0Æ05 (0Æ78)

MLGA

CM1046

E. coli Significance MI MLGA CM1046

Colilert-18

CM1046

TC

level (correlation coefficient) >0Æ05 (0Æ72) >0Æ05 (0Æ73) – – – –

showed 78% and 88% sensitivity when compared to MI and MLGA agars in drinking water samples (Unpublished data). MI agar and Colilert-18 showed the least variation and only differed for a small percentage of Aeromonas spp. isolated from waste water samples, where for MI agar it should have been a white nonfluorescent colony according to the manufacturer’s instructions, but instead was white fluorescent colony, and the subculture of Colilert-18 wells that were yellow and fluorescent isolated Aeromonas sp., indicating a false-positive result in this system also. These isolates only became problematic for MI agar and Colilert-18 once numbers became extremely high (>10 000 CFU 100 ml)1, unpublished data). There was a 97% agreement between the target organism colour descriptions and the identity of MI agar isolates (n = 1036); 3% of the total isolates, such as A. baumannii and Pseudomonas spp., were problematic but are not considered as coliforms. No misrepresentations of E. coli occurred with MI agar. For MLGA agar 79% of the total number of representative organisms identified matched the target colour descriptions and 21% of organisms were misrepresented. These included TC (21%) such as Klebsiella spp., Citrobacter spp., Enterobacter spp., and miscellaneous organisms including, Pseudomonas spp., Flavobacterium spp., and Aeromonas spp., which are oxidase positive

E. coli

TC

Colilert-18

E. coli

TC

E. coli

TC

of E. coli and >0Æ05 (0Æ78) >0Æ05 (0Æ98) –

TC >0Æ05 (0Æ71) >0Æ05 (0Æ79) 0Æ05 (0Æ4) >0Æ05 (0Æ98) 0Æ05 (0Æ98) 0). The other chromogenic agars had unreliable colour discrimination that would preclude their use in the routine laboratory setting unless subsequent confirmation techniques are employed.


G. Hallas et al.

Examination of non potable waters (n = 35, average recovery counts in the range of 104–106 CFU 100 ml)1) failed to yield any significant differences between the technologies or within any of the matrix types. Some small differences were noted between MI agar and Colilert-18 in tank and waste water in the detection of E. coli and TC; and for MLGA with waste, reservoir and source water in the detection of TC. However none of these biases were considered significant by Passing and Bablok analysis. Although these latter observations may allude to the preferential use of particular technologies for a specific matrix type, dilution errors that invariably occur when analysing water with heavy bacterial loads may have contributed to these results. Also, our data set was small and may not necessarily reflect a detailed investigation of these water types in all Australian waters. Chlorination and chloramination are the most widely used bactericidal agents to disinfect drinking water and protect distribution systems worldwide. Disinfection regimes usually destroy cellular components of the organism and are bactericidal, however the process may instead result in sub-lethal injury, where bacteria retain activity but develop recalcitrance to culture (Watkins et al. 1988; Hoefel et al. 2003, 2005a,b). The sub-lethal injury increases sensitivity of stressed coliforms to selective media and may result in the failure to detect or underestimate their concentration in drinking water (McFeters et al. 1986a,b). This was not apparent in our recovery of stressed organisms when investigating the effects of chlorination, however significant differences were apparent in the stress recovery experiments for chloraminated systems when the agars were compared to Colilert-18 (Tables 6 and 7). This difference between chlorine and chloramine has also been observed by others and is presumably due to the ability of reducing agents to repair oxidative injury caused by oxidation of sulfhydryl groups to disulfides in protein associated chloramine damage (Watters et al. 1989). Colilert-18 contains sulphite compounds, which act as reducing agents that may reverse the damage caused by chloramine action, and it seems that the future development of chromogenic agars may be enhanced if similar reducing agents were included to increase recovery of stressed bacteria. Stressed bacteria become sensitive to the process of MF and are susceptible to drying out on the surface of the membrane, particularly if there is cell wall damage as exerted by common disinfectants, and it seems Colilert18 may aid in the recovery of these organisms (Niemela et al. 2003). Interestingly, the data from our distribution samples did not support these phenomena, probably due to the increased contact time of disinfectants in the distribution system rendering any membrane damage irrevers-


ible. Nonetheless, caution should be exercised if using chromogenic agars for chloraminated systems. Reliable colour discrimination on chromogenic agars is required to avoid lengthy confirmation of isolates and would be a major advantage to the water industry world wide, particularly with recent advances in digital colour analysis colony counters and semi-automation of sample analysis. The results in Table 8 indicate that MI is the most reliable and consistent (>90% sensitivity and specificity) for the discrimination of E. coli and TC in potable water, which lends itself to the use of automated colour colony counters in routine settings. For certain types of water the use of Colilert-18 may also be problematic. False positive results have been reported for Aeromonas sp. (Landre et al. 1998). This study also found the break-through of Plesiomonas sp. and Chryseomonas sp. in the Colilert-18 system (data not shown) and should be used with caution if these organisms are suspected or detected in high numbers. The various effects of each medium can be attributed to the recent inclusion of inhibitory substances in the preparation formulae of each technology. For MI agar the inclusion of 5–10 lg ml)1 cefsulodin has the effect of inhibiting 104–105 CFU 100 ml)1 of oxidase positive organisms (Brenner et al. 1993) and Colilert-18 suppresses the expression of b-galactosidase of some oxidase positive organisms at levels up to 104 CFU 100 ml)1 due to addition of ammonium sulphate, presumably due to their inability to assimilate this compound and induce the galactosidase system (Edberg and Edberg 1988). MLGA and CM1046 do not have such additions and hence the breakthrough of problematic organisms. Chromogenic technologies have greatly simplified both medical and environmental interpretation of microbial flora. While the use of chromogenic agars is expanding in medical laboratories in Australia, there has been a reluctance to accept chromogenic agars by water authorities, primarily due to the lack of appropriate comparative studies under Australian conditions. This study shows that the chromogenic agars investigated are comparable with the current standard method of Colilert-18 when dealing with chlorinated water. The advantages of MI agar over the other agars is primarily due to its ability to inhibit nontarget organisms and the colour specificity observed, without the need to perform confirmatory tests, and as such is an ideal candidate for the examination of indicator organisms in these waters. Acknowledgements The authors would like to thank the Australian Water Quality Centre and South Australian Water Corporation for supporting this work. The technical assistance from


1147


all the laboratory technicians in the microbiology operations laboratory is gratefully acknowledged. Special thanks are extended to Oxoid and Becton Dickinson for their support and for providing chromogenic reagents and other consumables, and to M. Banasiak for critical evaluation of the manuscript. References Adcock, P.W. and Saint, C.P. (1997) Trials of Colilert system. Water, March ⁄ April, 22–25. Anon. (1998) AS ⁄ NZS 5667.1: Water quality - Sampling Guidance on the Design of Sampling Programs, Sampling Techniques and the Preservation and Handling of Samples. North Sydney, NSW, Australia: Standards Australia. Anon. (1999a) AS ⁄ NZS 4659.2: Guide to Determining the Equivalence of Food Microbiology Test Methods, Part 2: Quantitative Tests. North Sydney, NSW, Australia: Standards Australia. Anon. (1999b) AS ⁄ NZS 4659.3: Guide to Determining the Equivalence of Food Microbiology Test Methods, Part 3: Confirmation Tests. North Sydney, NSW, Australia: Standards Australia. Anon. (2001) AS ⁄ NZS 2031: Selection of Containers and Preservation of Water Samples for Microbiological Analysis. North Sydney, NSW, Australia: Standards Australia. Anon. (2002) The Microbiology of Drinking Water - Part 4 Methods for the Isolation and Enumeration of Coliform Bacteria and Escherichia coli (including E. coli O157:H7) in this Series. Nottingham, UK: Methods for the Examination of Waters and Associated Materials. Anon., 2005. AS ⁄ NZS 4276.21: Examination for coliforms and Escherichia coli—Determination of Most Probable Number (MPN) using Enzyme Hydrolysable Substrates. North Sydney, NSW, Australia: Standards Australia. Brenner, K.P., Rankin, C.C., Roybal, Y.R., Stelma, G.N. Jr, Scarpino, P.V. and Dufour, A.P. (1993) New medium for the simultaneous detection of total coliforms and Escherichia coli in water. Appl Environ Microbiol 59, 3534–3544. Brenner, K.P., Rankin, C.C., Sivaganesan, M. and Scarpino, P.V. (1996) Comparison of the recoveries of Escherichia coli and total coliforms from drinking water by the MI agar method and the U.S. Environmental Protection Agency-approved membrane filter method. Appl Environ Microbiol 62, 203–208. Buckalew, D.W., Hartman, L.J., Grimsley, G.A., Martin, A.E. and Register, K.M. (2006) A long-term study comparing membrane filtration with Colilert defined substrates in detecting fecal coliforms and Escherichia coli in natural waters. J Environ Manage 80, 191–197. Chao, K.K., Chao, C.C. and Chao, W.L. (2004) Evaluation of Colilert-18 for detection of coliforms and Eschericha coli in subtropical freshwater. Appl Environ Microbiol 70, 1242–1244.

1148

G. Hallas et al.

Covert, T.C., Shadix, L.C., Rice, E.W., Haines, J.R. and Freyberg, R.W. (1989) Evaluation of the Autoanalysis Colilert test for detection and enumeration of total coliforms. Appl Environ Microbiol 55, 2443–2447. Eaton, A.D., Clesceri, L.S., Rice, E.W. and Greenberg, A.E. (2005) Standard Methods for the Examination of Water and Wastewater. Washington, DC: American Water Works Association (AWWA). Edberg, S.C. and Edberg, M.M. (1988) A defined substrate technology for the enumeration of microbial indicators of environmental pollution. Yale J Biol Med 61, 389–399. Evanson, M. and Ambrose, R.F. (2006) Sources and growth dynamics of fecal indicator bacteria in a coastal wetland system and potential impacts to adjacent waters. Water Res 40, 475–486. Hoefel, D., Grooby, W.L., Monis, P.T., Andrews, S. and Saint, C.P. (2003) Enumeration of water-borne bacteria using viability assays and flow cytometry: a comparison to culture-based techniques. J Microbiol Methods 55, 585–597. Hoefel, D., Monis, P.T., Grooby, W.L., Andrews, S. and Saint, C.P. (2005a) Culture-independent techniques for rapid detection of bacteria associated with loss of chloramine residual in a drinking water system. Appl Environ Microbiol 71, 6479–6488. Hoefel, D., Monis, P.T., Grooby, W.L., Andrews, S. and Saint, C.P. (2005b) Profiling bacterial survival through a water treatment process and subsequent distribution system. J Appl Microbiol 99, 175–186. ISO (2004) ISO 17994:2004 Water Quality - Criteria for Establishing Equivalence Between Microbiological Methods. Geneva: International Organisation for Standardisation. ISO ⁄ TR (2000) ISO ⁄ TR 13843 Water Quality - Guidance on Validation of Microbiologicla Methods. Geneva, Switzerland: International Organisation for Standardisation. Landre, J.P., Gavriel, A.A. and Lamb, A.J. (1998) False-positive coliform reaction mediated by Aeromonas in the Colilert defined substrate technology system. Lett Appl Microbiol 26, 352–354. Macy, J.T., Dunne, E.F., Angoran-Benie, Y.H., Kamelan-Tano, Y., Kouadio, L., Djai, K.A. and Luby, S.P. (2005) 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 3, 221–228. McFeters, G.A., Kippin, J.S. and LeChevallier, M.W. (1986a) Injured coliforms in drinking water. Appl Environ Microbiol 51, 1–5. McFeters, G.A., LeChevallier, M.W., Singh, A. and Kippin, J.S. (1986b) Health significance and occurrence of injured bacteria in drinking water. Water Sci Technol 18, 227–231. Muirhead, R.W., Littlejohn, R.P. and Bremer, P.J. (2004) Evaluation of the effectiveness of a commercially available defined substrate medium and enumeration system for measuring Escherichia coli numbers in faeces and soil samples. Lett Appl Microbiol 39, 383–387.


G. Hallas et al.

Niemela, S.I., Lee, J.V. and Fricker, C.R. (2003) A comparison of the International Standards Organisation reference method for the detection of coliforms and Escherichia coli in water with a defined substrate procedure. J Appl Microbiol 95, 1285–1292. Oshiro, R.K. (2002) Method 1604: Total Coliforms and Escherichia coli in Water by Membrane Filtration Using a Simultaneous Detection Technique (MI Medium). Washington, DC: USEPA. Sartory, D.P. and Howard, L. (1992) A medium detecting b-glucuronidase for the simultaneous membrane filtration enumeration of Escherichia coli and coliforms from drinking water. Lett App Micr 15, 273–276. Smith, R.J., Twedt, R.M. and Flanigan, L.K. (1973) Relationships of indicator and pathogenic bacteria in stream waters. J Water Pollut Control Fed 45, 1736–1745.


Watkins, W.D., Rippey, S.R., Clavet, C.R., Kelley-Reitz, D.J. and Burkhardt, W., 3rd (1988) Novel compound for identifying Escherichia coli. Appl Environ Microbiol 54, 1874–1875. Watters, S.K., Pyle, B.H., LeChevallier, M.W. and McFeters, G.A. (1989) Enumeration of Enterobacter cloacae after chloramine exposure. Appl Environ Microbiol 55, 3226–3228. Yakub, G.P., Castric, D.A., Stadterman-Knauer, K.L., Tobin, M.J., Blazina, M., Heineman, T.N., Yee, G.Y. and Frazier, L. (2002) Evaluation of Colilert and Enterolert defined substrate methodology for wastewater applications. Water Environ Res 74, 131–135. Yanez, M.A., Valor, C. and Catalan, V. (2006) A simple and cost-effective method for the quantification of total coliforms and Escherichia coli in potable water. J Microbiol Methods 65, 608–611.


1149