Letters in Applied Microbiology ISSN 0266-8254

ORIGINAL ARTICLE

Performance validation of chromogenic coliform agar for

the enumeration of Escherichia coli and coliform bacteria
B. Lange1, M. Strathmann2 and R. Oßmer3
1 Department of Water Quality, IWW Water Centre, Muelheim an der Ruhr, Germany
2 Department of Applied Microbiology, IWW Water Centre, Muelheim an der Ruhr, Germany
3 BioMonitoring R&D, Merck KGaA, Merck Millipore, Darmstadt, Germany

Significance and Impact of the Study: The international standard for the detection and enumeration of
E. coli and coliform bacteria by membrane filtration (ISO 9308-1) is currently under revision and will be
published in 2014. In the new standard, lactose–triphenyl tetrazolium chloride (TTC) agar will be
replaced by a CCA. A performance validation of this revised method according to ENV ISO 13843 is pre-
sented in this study to determine fundamental data on its applicability and to provide reference data
for secondary validation by users of this method.

Keywords Abstract
chromogenic coliform agar, coliform bacteria,
drinking water, Escherichia coli, performance, The performance of chromogenic coliform agar (CCA) for the enumeration of
validation. Escherichia coli and coliform bacteria was validated according to ENV ISO
13843 using pure cultures and naturally contaminated water samples. The
Correspondence results indicate that for the detection of E. coli and coliform bacteria,
Bernd Lange, IWW Water Centre, Moritzst-
respectively, the method is sensitive (94 and 91%), specific (97 and 94%),
rasse 26, 45476 Muelheim an der Ruhr, Ger-
many. E-mail: [email protected]
selective (selectivity 0
78 and 0
32) and efficient (96 and 92%). Relative
recovery of E. coli and coliform bacteria on CCA in comparison with tryptone
2013/1025: received 24 May 2013, revised soy agar (TSA) was good (104 and 94% in mean, >80 and >70% in all cases),
1 August 2013 and accepted 12 August and repeatability and reproducibility were sufficient. The linear working range
2013 was defined for 10–100 total target colonies per 47-mm membrane filter. A
high precision of the method was confirmed by low overdispersion in
doi:10.1111/lam.12147
comparison with Poisson distribution. The robustness of the method with
respect to the variable incubation time of 21  3 h was found to be low,
because an incidental increase in presumptive colonies especially between 18
and 21 h was observed. In conclusion, the CCA method was proved as a
reliable method for the quantification of E. coli and coliform bacteria.

presence of coliform bacteria, although not a proof of

Introduction
faecal contamination, may indicate problems in the treat-
The presence and extent of faecal contamination is an ment process or during drinking water distribution.
important issue for the assessment of the quality of a To enable a reliable detection and quantification of
water sample and of the risk to human health due to E. coli and coliform bacteria, an appropriate analytical
infection. It is widely accepted in practice and in literature method is necessary. Different methods were reviewed in
to use Escherichia coli and coliform bacteria as indicators the study by Rompr
e et al. (2002).
for a faecal pollution (Byamukama et al. 2000; Geissler et ISO/DIS 9308-1 (2012) specifies a method for the
al. 2000). Other coliform bacteria than E. coli can also be enumeration of E. coli and coliform bacteria based on
found in the environment, for example soil or surface membrane filtration and subsequent culture on a chromo-
water (Geissler et al. 2000). This makes interpretation genic agar medium. It is especially suitable for water sam-
of results for coliform bacteria more difficult, but the ples with low bacterial numbers (e.g. drinking water,

Letters in Applied Microbiology 57, 547--553 © 2013 The Society for Applied Microbiology 547
Performance of chromogenic coliform agar B. Lange et al.

disinfected pool water or finished water from drinking for membrane filtration and cultivation on CCA. Typical
water treatment plants). Due to the low selectivity of the results are shown in Fig. 1. In the case of E. coli, all tested
differential agar medium, background growth can interfere strains revealed dark-blue- to violet-coloured colonies,
with the reliable enumeration of coliform bacteria and and in the case of coliform bacteria, pink- to red-col-
E. coli, for example in surface waters or shallow well oured colonies. Nontarget bacteria produced beige- to
waters. This method is not suitable for these types of yellow-coloured colonies or even no growth on the CCA
water. medium. In conclusion, using the selected chromogenic
The current third version of ISO 9308-1 cancels and enzyme substrates contained in the CCA medium, a clear
replaces the second edition (ISO 9308-1 2000), which has identification and differentiation of the two groups of
been technically revised. This revision was necessary, target organisms and nontarget flora were possible.
because the previous version was based on the use of lac-
tose–TTC agar and showed too low selectivity and prob- Sensitivity, specificity and selectivity studies
lems with high accompanying flora (Pitk€anen et al. 2007). The fundamental parameters of the method were deter-
In that method, the detection of coliform bacteria relies mined as described in ENV ISO 13843 section 9.2. Drink-
on lactose fermentation and negative oxidase reaction and ing water spiked with naturally contaminated ambient
of E. coli on the detection of tryptophanase by indole river water samples from 10 different locations, each hav-
reaction. In addition, on ISO level, it was decided already ing different levels of contamination, was used for these
in 2006 to define coliform bacteria on their b-D-galactosi- studies, covering concentrations within the whole working
dase activity (EC 3.2.1.23) and no longer on lactose range of the method. Fifteen different independent sam-
fermentation. Further, E. coli should be identified by its ples were processed according to ISO/DIS 9308-1 (2012),
additional b-D-glucuronidase activity (EC 3.2.1.31). and in total, 220 colonies were randomly selected, includ-
The membrane filtration method provided in ISO/DIS ing typical E. coli and coliform bacteria and atypical colo-
9308-1 (2012) meets these requirements. This method is nies. All colonies were tested regarding their oxidase
based on the following principles: (i) the use of a suitable activity, Gram staining, cell morphology, b-D-galactosi-
growth medium containing peptones, pyruvate, sorbitol dase and b-D-glucuronidase activity. Using these results, it
and phosphate buffer to support rapid colony growth, was possible to group the colonies into four groups: (i)
even for the sublethally injured coliforms, (ii) inhibition true positives, (ii) false negatives, (iii) false positives and
of accompanying Gram-positive bacteria flora as well as (iv) true negatives. The results obtained for readings of
some Gram-negative bacteria flora by the use of Tergitol® coliform bacteria (incl. E. coli) and E. coli are compiled
7, which has no negative effect on the growth of the in Tables 1 and 2, respectively, and were used to calculate
coliform bacteria, (iii) the use of Salmon-GAL (6-chloro- sensitivity, specificity, false-positive rate, false-negative
3-indoxyl-ß-D-galactopyranoside) and isopropyl-b-D-thiog- rate, efficacy and selectivity (Table 3). The raw data of all
alactopyranoside (IPTG) substrate, which can be cleaved tested colonies are compiled in the supporting informa-
by b-D-galactosidase, resulting in a pink to red coloration tion Table S1.
of the coliform colonies and (iv) the substrate X-glucuro- Six and four colonies from the 117 and 65 typical colo-
nide (5-bromo-4-chloro-3-indoxyl-ß-D-glucuronide) for nies for coliform bacteria and E. coli, respectively, were
the detection of b-D-glucuronidase. Escherichia coli cleaves identified as noncoliform species. Eleven and four colo-
both Salmon-GAL and X-glucuronide; thus, positive colo- nies from the 103 and 155 atypical colonies were actually
nies will show a dark blue to violet colour. The perfor- found to be coliform bacteria and E. coli, respectively.
mance of the ISO/DIS 9308-1 (2012) method was In general, performance results were better for the
validated in this study according to the requirements of detection of E. coli compared with total coliform bacteria.
ENV ISO 13843 using selected pure cultures and water The outcomes of this validation indicate that the ISO/DIS
samples contaminated with ambient river water. 9308-1 (2012) method is sensitive (94 and 91%) and spe-
cific (97 and 94%) for the detection of E. coli and coli-
form bacteria, respectively. The method is also selective,
Results and discussion
having a value of 0
32 and 0
78 for coliform bacteria
and E. coli, respectively, which is in both cases consider-
Target organism identification
ably better than the guidance value of 1 suggested by
Pure culture challenge ENV ISO 13843 for colony count methods (data not
Typical characteristic reactions of target and nontarget shown). It has to be noted that selectivity for E. coli
bacteria were confirmed using bacterial suspensions of seems to be worse in comparison with coliform bacteria.
pure cultures of selected reference strains of E. coli, coli- This is due to the fact that the CCA medium is not
form bacteria and noncoliform Gram-negative bacteria intended to be selective for E. coli alone, but for total

548 Letters in Applied Microbiology 57, 547--553 © 2013 The Society for Applied Microbiology
B. Lange et al. Performance of chromogenic coliform agar

(a) (b)

(c) (d)

(e) (f)
Figure 1 Typical appearance of colonies of
target and nontarget organisms grown on
membrane filters on chromogenic coliform
agar (CCA); (a) Escherichia coli WDCM 00012
(target organism), (b) Citrobacter freundii
DSM 30039 (target organism for coliform
bacteria), (c) Enterobacter aerogenes WDCM
00175 (target organism for coliform bacteria);
(d) Aeromonas hydrophila WDCM 00063
(nontarget organism), (e) Staphylococcus
aureus WDCM 00032 (nontarget organism);
(f) Naturally contaminated water sample
obtained from the River Ruhr.

Table 1 Experimental data for the determination of coliform bacteria Table 2 Experimental data for the determination of E. coli obtained
obtained for randomly selected typical and atypical colonies (n = 220) for randomly selected typical and atypical colonies (n = 220)

Presumptive colonies Presumptive colonies

Typical Atypical Total Typical Atypical Total

Confirmed colonies Confirmed colonies

Positive 111 11 122 Positive 61 4 65
Negative 6 92 98 Negative 4 151 155
Total 117 103 220 Total 65 155 220

Letters in Applied Microbiology 57, 547--553 © 2013 The Society for Applied Microbiology 549
Performance of chromogenic coliform agar B. Lange et al.

Table 3 Performance characteristics for the use of chromogenic coli- of counting repeatability and reproducibility using the
form agar for the enumeration of Escherichia coli and coliform bacte- relative standard deviations (RSD) of repeated counts and
ria according to ISO/DIS 9308-1 (2012)
recommends that when using mixed populations, RSD
Coliform should ideally be <0
1 (i.e. not more than 10% deviation).
Parameter E. coli bacteria The calculated values of 0
035 and 0
046 for the individ-
ual counting repeatability for the determination of E. coli
Identification*
Sensitivity 93
8% 91
0%
and coliform bacteria, respectively, are considerably lower
Specificity 97
4% 93
9% than the recommended value of 0
1 given in ENV ISO
False-positive rate 6
2% 5
1% 13843 Annex B, indicating that reliable counting of both
False-negative rate 2
6% 10
7% target organisms can be achieved with the ISO/DIS 9308-
Efficiency 96
4% 92
3% 1 (2012) CCA method (Table 3). The calculated values of
Selectivity 0
78 0
32 0
127 and 0
114 for the intralaboratory counting repro-
Counting uncertainty (RSD)
ducibility for the determination of E. coli and coliform
Repeatability 0
046 0
035
Reproducibility 0
127 0
114
bacteria, respectively, are only slightly higher than the
Recovery >80% >70% guided value, thus, confirming this result. Raw data for
Range of quantitative determination 10–100 10–100 the determination of counting repeatability and reproduc-
(colonies per 47-mm membrane filter) ibility are summarized in the supporting information
Robustness of incubation time Incidental increase in Tables S3 and S4, respectively.
presumptive colonies
especially between
18 and 21 h Working range and linearity
RSD, relative standard deviations. Linearity is defined as the ability of an analytical proce-
*Number of investigated colonies: n = 220. dure to obtain test results within a given range, which
are directly proportional to the concentration of analyte
in the sample. The number of bacteria in a sample
coliform bacteria including E. coli. With a calculated effi- where linearity is not given any more can be regarded
ciency value of 96 and 92% for the detection of E. coli as the upper boundary of the working range. The work-
and coliform bacteria, respectively, the data suggest that ing range and thus the linear range covered by the
the method is highly efficient. False-negative results as method were determined by checking the proportionality
well as false-positive results were considerably low. of results using serial dilutions of samples of drinking
water spiked with naturally contaminated ambient river
water. One representative result is shown in Fig. 2, indi-
Relative recovery
cating that colonies of E. coli and coliform bacteria
The recovery of target organisms grown on CCA was could be counted in a reliable manner in a range of up
determined relatively in comparison with the same sam- to 100–300 colonies per membrane filter in this case. In
ples cultivated on nonselective TSA medium. Two E. coli total, seven samples of which were four water samples
strains and four coliform bacteria that are not E. coli were contaminated with ambient river water, two pure cul-
tested, each in at least ten replicate sample filtrations. tures of coliform bacteria (Klebsiella pneumoniae round
Mean relative recovery rates on CCA of 104 and 94% for robin test NLGA Aurich and Citrobacter freundii DSM
E. coli and coliform bacteria (incl. E. coli) were observed, 30039) and one pure culture of E. coli WDCM 00012
indicating a high recovery of target organisms in general. were tested for linearity in the same way. Quantitative
In all cases, relative recovery was >70% for coliform interpretation of the results was conducted by calculating
bacteria and >80% for E. coli (Table 3). Raw data for the log-likelihood index (G²) and comparison of the
determination of the relative recovery rates are summa- results to guide values, obtained by referring to tables of
rized in the supporting information Table S2. chi-square with n 1 degrees of freedom, as described
in ENV ISO 13843 example B.6. Values exceeding the
tabulated value indicate departure from proportionality
Counting uncertainty
at the chosen probability level. For counting of E. coli
Repeatability is the agreement in counts obtained by colonies, linearity in the range of 10–100 colonies per
repeated counting performed by the same analyst, and filter was confirmed to be >95%, whereas the counting
reproducibility is the agreement between subsequent of coliform bacteria colonies showed lower linearity of
counts obtained by two or more different analysts. ENV >50% in this range (data not shown). In conclusion,
ISO 13843 Annex B provides guidance on the assessment 10–100 total target colonies per 47-mm membrane filter

550 Letters in Applied Microbiology 57, 547--553 © 2013 The Society for Applied Microbiology
B. Lange et al. Performance of chromogenic coliform agar

400

350
y = 2·6222x
R ² = 0·9856
300

250

Colony counts
y = 2·2715x
R ² = 0·9991
200

150

Figure 2 Linear working range of the 100

method using drinking water spiked with a y = 0·7572x
50 R ² = 0·9965
naturally contaminated surface water sample
obtained from the River Ruhr; mean values of
three independent counts per sample are 0
shown: (♦) Escherichia coli, (■) coliform
0 16 32 48 64 80 96 112 128
△
bacteria, ( ) total colony counts. Relative sample volume

can be defined as a suitable working range for the CCA Detailed data on these calculations can be found in the
method according to ISO/DIS 9308-1 (2012). Because supporting information Table S6. In summary, for the
E. coli and other coliform bacteria are determined on determination of E. coli and coliform bacteria, overdisper-
the same filter, both will contribute to this number of sion was found to be 9
3 and 3
1% in excess of the
colonies. In dependence on the ratio of E. coli to other Poisson distribution, respectively, in mean of the ten
coliform bacteria, the countable number of E. coli will investigated samples. Using the regression approach to
vary with respect to the other coliform colonies. take into account the whole set of investigated samples,
an overdispersion of 16
6 and 8
4% was observed for the
determination of E. coli and coliform bacteria, respec-
Precision
tively. These values indicate only minor deviation of
In general, precision is defined as the closeness of agree- uncertainty of results obtained by CCA method from the
ment between independent test results obtained under theoretically expected one due to Poisson distribution,
stipulated conditions. For microbial assays, a basic random thus proving a good precision of the method.
variation is unavoidable due to the random distribution
of bacteria in the tested sample as described by the Pois-
Robustness of incubation time
son distribution. Technical imperfections (e.g. variation
in media and sample processing) and other causes (e.g. Robustness means tolerance towards slight changes in
variations in laboratory staff) are responsible for addi- procedure or towards unavoidable variations in condi-
tional variation. Parallel determinations can vary even tions of the laboratory environment. The aspect of
more than is explained by the Poisson distribution. This robustness that is of major relevance for the CCA method
situation is called overdispersion. Parallel determinations is the variance of incubation period of 21  3 h, as
involving the whole analytical procedure cannot be described in ISO/DIS 9308-1 (2012). To investigate this
expected to follow the Poisson distribution. The quantifi- influence, six samples of drinking water spiked with dif-
cation of the overdispersion can be used as a means to ferent naturally contaminated surface water samples were
describe the precision of an analytical procedure. In this processed in tenfold replicates and were each counted
study, the overdispersion at detector level was determined after 18, 21 and 24 h (raw data and significance values
from 10 drinking water samples spiked with naturally are presented in the supporting information Table S5).
contaminated river water samples, each processed tenfold In the case of determination of E. coli, an increase in
in parallel, and each counted by at least two different presumptive colonies of 11
0% (6
6%) was observed
technicians. Quantitative interpretation of the overdisper- between 18 and 21 h, whereas the increase between 21
sion was performed by (i) calculation of the Poisson index and 24 h was comparably low [1
4% (3
1%)]. For
of dispersion (Χ²n 1), (ii) calculation of the overdispersion quantification of coliform bacteria (incl. E. coli), an
coefficient (u) by Anscombe’s method I (Anscombe 1950) increase of 28
6% (10
9%) and 18
4% (8
3%) was
and (iii) using the regression approach described in ENV determined between 18-h and 21-h incubation time, and
ISO 13843 section 6.2.3. 21 h and 24 h incubation time, respectively.

Letters in Applied Microbiology 57, 547--553 © 2013 The Society for Applied Microbiology 551
Performance of chromogenic coliform agar B. Lange et al.

The data suggest that an incidental increase in pre- samples were pretested by Colilertâ-18 QuantiTray 2000
sumptive colonies especially between 18 and 21 h will (IDEXX, Ludwigsburg, Germany) according to the manu-
occur for both E. coli and other coliform bacteria. For facturer’s instructions, and dilutions were adapted to
E. coli, no significant change in presumptive colonies was obtain required target colony counts.
observed between 21 and 24 h, and thus, the incubation
time is robust in this range. However, for some coliform
Detection of Escherichia coli and coliform bacteria
bacteria that either grow slowly or have reduced b-D-
galactosidase activity, there may be a significant increase The quantification of E. coli and coliform bacteria was per-
in counts within the accepted incubation time. An formed using the ISO/DIS 9308-1 (2012) method. Briefly,
increase in counts over a prescribed incubation period is samples (10–100 ml) were vacuum-filtered through a
a well-recognized phenomenon (e.g. for typical membrane membrane composed of cellulose mixed ester (0
45 lm
filtration methods, such as ISO 7899-2 for enterococci, pore size, EZPak, Merck, cat. no. EZHAWG474), placed on
unpublished data). However, provided that the increase CCA plates (Chromocult® coliform agar, Merck) and incu-
in counts after an extended incubation is not excessive, bated for (21  3) h at (36  2)°C. For some experiments
for practical and operational considerations, counts from (determination of robustness), cultivation time was varied
the CCA method at 21  3 h of incubation can be con- to exactly 18, 21 and 24 h. For some experiments (determi-
sidered acceptable. Nevertheless, the preferred incubation nation of relative recovery), cultivation was performed on
time for water samples should be 21 h. Incubation for nonselective TSA medium.
24 h increases the recovery of the target bacteria especially Counting of b-D-galactosidase-positive colonies (pink
if they are stressed, for example, from disinfected waters. to red) resulted in presumptive coliform bacteria that are
not E. coli. To avoid false-positive results, caused by oxi-
dase-positive bacteria such as Aeromonas spp, the pre-
Materials and methods
sumptive colonies were confirmed by a negative oxidase
reaction. Escherichia coli was quantified by counting ß-
Bacterial strains
galactosidase- and ß-glucuronidase-positive colonies (dark
The following bacterial strains were used in this study: blue to violet). Total coliform bacteria were calculated as
E. coli WDCM 00012, Citrobacter freundii DSM 30039, the sum of oxidase-negative colonies with pink to red col-
Enterobacter aerogenes WDCM 00175, Aeromonas hydro- our and all dark blue to violet colonies. Colonies having
phila WDCM 00063 and Pseudomonas aeruginosa WDCM other colours than typical for E. coli or coliform bacteria
00024 were obtained from the German Collection of were counted as ‘atypical colonies’.
Micro-organisms and Cell Cultures (DSMZ, Braun-
schweig, Germany) and E. coli and Klebsiella pneumoniae
Oxidase test
round robin strains from the Nieders€achsisches Landes-
gesundheitsamt (NLGA, Aurich, Germany). All micro- The oxidase activity of presumptive colonies was tested
organisms were maintained on TSA. using Bactident® Oxidase test (Merck, cat. no.
1.13300.0001) according to the manufacturer’s instructions.
Culture media
Gram staining
Chromogenic coliform agar (CCA) was obtained from
Merck (Chromocult® coliform agar, cat. no. 1.10426.0500, Gram staining was performed using Gram-color gram
Merck Millipore, Germany). TSA was obtained from Oxoid staining kit (Merck, cat. no. 1.11885.0001) according to
(cat. no. CM0131). All culture media were prepared in the manufacturer’s instructions. Samples were investigated
accordance with manufacturer’s instructions. using a Leica Laborlux microscope at 1000-fold magnifi-
cation.
Water samples
Detection of b-D-galactosidase and b-D-glucuronidase
All water samples investigated throughout this study were
activity
prepared using drinking water (tap water, M€ ulheim an
der Ruhr, Germany) spiked with naturally contaminated The principle of the method described in ISO 9308-2
ambient river water samples derived from 10 different (2012) was used to detect the presence or absence o
sampling sites at surface waters located in M€ ulheim an b-D-galactosidase and b-D-glucuronidase activity of pre-
der Ruhr (Germany) and Duisburg (Germany). Concen- sumptive colonies. The proprietary Colilertâ-18 substrate
trations of E. coli and coliform bacteria in river water (IDEXX, cat. no. WP200I) was dissolved in 100 ml sterile

552 Letters in Applied Microbiology 57, 547--553 © 2013 The Society for Applied Microbiology
B. Lange et al. Performance of chromogenic coliform agar

water and was subsequently divided into portions of 5 ml Geissler, K., Manafi, M., Amoros, I. and Alonso, J.L. (2000)
in test tubes. These test tubes were inoculated with mate- Quantitative determination of total coliforms and
rial obtained from the presumptive colonies, mixed and Escherichia coli in marine waters with chromogenic and
incubated for 20  2 h at 36  1°C. Detection was fluorogenic media. J Appl Microbiol 88, 280–285.
recorded as presence/absence of yellow colour for b-D- ISO 9308-1 (2000) Water Quality - Enumeration of
galactosidase activity and blue fluorescence (excitation Escherichia coli and Coliform Bacteria - Part 1:
at 366 nm) for b-D-glucuronidase activity. Membrane Filtration Method. Geneva: International
Standards Organization.
ISO 9308-2 (2012) Water Quality - Enumeration of Escherichia
Identification of isolates coli and Coliform Bacteria - Part 2: Most Probable
Number Method. Geneva: International Standards
Some presumptive coliform colonies (red) that were
Organization.
tested oxidase positive were identified using the proprie-
ISO/DIS 9308-1 (2012) Water Quality - Enumeration of
tary APIâ 20 E system (Biom
erieux SA, Marcy l’Etoile,
Escherichia coli and Coliform Bacteria - Part 1: Membrane
France) according to the manufacturer’s instructions. Filtration Method for Waters with Low Bacterial
Background Flora. Geneva: International Standards
Data analysis Organization.
Pitk€anen, T., Paakkari, P., Miettinen, I.T., Heinonen-Tanski,
All data analyses were performed using Microsoft Excel H., Paulin, L. and H€anninen, M.-L. (2007) Comparison of
(version 2010). media for enumeration of coliform bacteria and
Escherichia coli in non-disinfected water. J Microbiol
Methods 68, 522–529.
Acknowledgement
Rompr
e, A., Servais, P., Baudart, J., de-Roubin, M.R. and
We thank Fabian Ansorge for technical assistance. Laurent, P. (2002) Detection and enumeration of
coliforms in drinking water: current methods and
emerging approaches. J Microbiol Methods 49, 31–54.
Conflict of Interest
Dr. Oßmer is an employee of the Merck KGaA. Mr. Lange
Supporting Information
and Dr. Strathmann have no conflict of interest.
Additional Supporting Information may be found in the
online version of this article:
References
Table S1 Specificity and selectivity for the determina-
Anscombe, F.J. (1950) Sampling theory of the negative tion of E. coli and coliform bacteria using CCA.
binomial and logarithmic series distributions. Biometrika Table S2 Recovery rates for the determination of E. coli
37, 358–382. and coliform bacteria using CCA.
Byamukama, D., Kansiime, F., Mach, R.L. and Farnleitner, Table S3 Repeatability for the determination of E. coli and
A.H. (2000) Determination of Escherichia coli coliform bacteria using CCA.
contamination with Chromocult Coliform Agar showed a Table S4 Reproducibility for the determination of E. coli
high level of discrimination efficiency for differing fecal and coliform bacteria using CCA.
pollution levels in tropical waters of Kampala, Uganda. Table S5 Robustness of incubation time for the determi-
Appl Environ Microbiol 66, 864–868. nation of E. coli and coliform bacteria using CCA.
ENV ISO 13843 (2001) Water Quality - Guidance on Table S6 Precision testing for the determination of E. coli
Validation of Microbiological Methods. Geneva: and coliform bacteria using CCA.
International Standards Organization.

Letters in Applied Microbiology 57, 547--553 © 2013 The Society for Applied Microbiology 553