METHODS AND MATERIALS

Media and reagents

UTI medium (CM 949) was from Oxoid, Basingstoke, UK; BBL CHROMagar (254102) was from Becton and Dickinson, Oxford, UK; CPS ID2 (43211) was from bio-Merieux, Basingstoke, UK; Harlequin CLED (HAL 007) was from Lab M, Bury, UK; CLED agar (CM301) was from Oxoid.

Identification study

Isolates and samples

An organism collection of Gram negative urinary isolates saved from our previous study,⁵ and identified using API 20 E and 20NE (bio-Merieux) and BBL Crystal (Becton and Dickinson), was used for the preliminary identification phase. Additional clinical isolates identified by standard laboratory techniques were used as the source of Gram positive organisms and yeasts (table 1).

Media inoculation and incubation

Isolates, which were anonymised by using a numbering system of 1 to 787, were plated on to 5% horse blood agar to check for viability and purity. One colony was inoculated into 5 ml sterile saline to produce a 0.5 Macfarland standard.⁶ Using a 1 μl loop, each of the isolates was cultured on to one quarter of a plate for each of the four chromogenic agars.

All plates were incubated in air at 37°C ± 1°C for 18–24 hours.

Plate reading

Plates of each of the four chromogenic media were read independently and by two fully qualified members of staff. Results were documented according to colour and any discrepancies were checked by the site study coordinator.

Statistical methods

For the identification study, presumptive identification using the manufacturer’s colour criteria was compared with the real organism identity. Media sensitivity and positive predictive values (PPVs) were calculated.

Isolation study

Isolates and samples

In total, 1435 routine urine samples (predominantly in boric acid) received in four laboratories during a two week period in December 2001, from both hospital and general practice, were included in the second phase of our study. Samples were selected on the basis of white blood cell count (WBC) of > 100 × 10⁶/litre, as in our previous study.⁵

Media inoculation and incubation

Using a 1 μl loop, each of 1435 samples containing > 100 × 10⁶/litre WBC were inoculated on to one quarter of a plate for each of the three chromogenic media performing best in the identification study, in addition to the normal CLED subculture. These media were BBL CHROMagar, UTI medium, and CPS ID2.

All plates were incubated in air at 37°C ± 1°C for 18–24 hours.

Plate reading

Each of the chromogenic media and CLED agar plates was read independently and by two members of staff. Data on bacterial growth and significant isolates were recorded at the time of reading and discrepancies verified daily by a separate senior member of staff. Bacterial growth was recorded semiquantitatively on the basis of colony counting—one colony equated to 10⁶ colony forming units (cfu)/litre, increasing to > 10⁸ cfu/litre for more than 100 colonies.

For the purpose of our study, plates were recorded as having NG (no growth), 1, 2, or > 2 colony types according to colonial morphology. The numbers of each colony type were also recorded (0, < 10 cfu; 1, 10–100 cfu; 2, > 100 cfu) to support the evaluation of the contributing organism counts of mixtures.

Statistical methods

For the isolation study, data were collated centrally and entered into an Excel spreadsheet. Differences between laboratories were compared using the χ² test and differences between media using McNemar’s test.

RESULTS

Identification study

Using the colour criteria provided by the manufacturers, table 1 summarises the correlation of the presumptive identification from each of the chromogenic media with the confirmed identification previously assessed using API 20E/NE and BBL Crystal.

Table 2 summarises the sensitivity and PPV of each agar for Escherichia coli, Proteus/Morganella/Providencia spp and Pseudomonas spp.

Because E coli is the most likely organism to be used as a marker for monitoring resistance patterns, the sensitivity and PPV for the identification of E coli were considered most important in evaluating the performance of the agars. Results indicated that all of the chromogenic agars gave > 97% PPV (% of pink isolates being correctly identified as E coli), with Harlequin CLED, CPS ID2, and BBL CHROMagar achieving > 99% PPV. The lower PPV achieved by UTI medium was largely the result of the false presumptive identification of Citrobacter spp (11 of 18 isolates appearing as pink colonies), confirming results achieved from our earlier study.⁵

In terms of sensitivity (how many of the known E coli isolates were pink), UTI medium and BBL CHROMagar achieved > 98% sensitivity, whereas CPS ID2 attained > 95% sensitivity. Harlequin CLED gave the lowest sensitivity (78%), which was the result of E coli isolates producing inconsistent colouration (table 1).

With the exception of Harlequin CLED agar, the sensitivity and PPV for Proteus spp was 100%. By using a supplementary bench oxidase test most Pseudomonas spp isolates could be differentiated. Variations in PPV for Pseudomonas spp largely related to the organisms varying in degrees of pigmentation, so that some isolates appeared colourless, rather than green.

Isolation study

One thousand four hundred and thirty five urine samples were examined from the four laboratories (352, 376, 351, 356, respectively). Statistical analysis indicated some differences in the distribution of significant growths, mixed growths, and no growths across the laboratories. However, the distribution of results by laboratory according to media type were generally similar, and amalgamated results have therefore been used to identify media differences (table 3). Where there are significant differences between laboratories when comparing media these are discussed.

It can be seen that isolation rates (all samples except NG) varied across the four sites (p < 0.01). The comparison of media indicates that overall the highest isolation rate was achieved using BBL CHROMagar (p < 0.01). However, there was laboratory variation, with laboratory 4 giving parity between BBL CHROMagar and CPS ID2, whereas in all other sites the isolation rates were higher on BBL CHROMagar than on CPS ID2. In all cases, isolation rates were higher on BBL CHROMagar than on CLED or UTI medium.

An analysis of the number of samples containing mixed organisms showed that all three chromogenic media evaluated in this phase detected significantly more mixed growths than did the traditional CLED agar (p < 0.01). This confirms the results of our previous study with UTI medium,⁵ in that CLED agar identifies significantly fewer mixed growths than the chromogenic media. Using the criterion of mixed growth being greater than two organism types (used in our previous study), BBL CHROMagar yielded slightly more mixed cultures than UTI medium and CPS ID2; however, this difference was not significant. By applying the criterion of mixed growth being greater than one organism type, the variation becomes significant, with BBL CHROMagar detecting more mixed growths than UTI medium and CPS ID2 (table 4).

A direct comparison of UTI medium and CPS ID 2 showed that although they each gave the same number of mixed cultures, there was laboratory variation, with two laboratories giving similar numbers and one giving more on UTI medium than on CPS ID2 (p = 0.015) and the other more on CPS ID2 than on UTI medium (p = 0.046).

CHROMagar and UTI medium gave a higher proportion of samples that yielded enterococci than did CPS ID2 (table 5). For all the chromogenic media examined, the identification of enterococci was not impaired in the presence of heavy growths of Gram negative organisms (table 5), but for other organisms (gram positives and yeasts), detection levels decreased as the numbers of Gram negative organisms increased (table 6). This trend was consistent across all laboratories.

DISCUSSION

CLED medium has historically been used in the UK for diagnostic routine urinary bacteriology as a non-selective medium capable of supporting the growth of most urinary pathogens and giving good colonial differentiation without surface spread of Proteus spp. The development of chromogenic agars, combining the basal CLED medium with various chromogenic substrates, has enhanced the presumptive identification of urinary isolates and enabled mixed cultures to be detected more easily. This enhanced identification of different species facilitates the monitoring of bacterial resistance, thus providing a cost effective mechanism of antibiotic surveillance in the routine diagnostic laboratory.

Our previous study had confirmed the superiority of one type of chromogenic agar against CLED,⁵ but the importance of this follow up study was to evaluate alternative media available to determine whether other formulations were superior.

“This enhanced identification of different species facilitates the monitoring of bacterial resistance, thus providing a cost effective mechanism of antibiotic surveillance in the routine diagnostic laboratory”

BBL CHROMagar, with a PPV of > 99% (excluding Gram positive isolates) and a sensitivity of > 98%, was the most accurate of the agars evaluated for the presumptive identification of Gram negative bacilli. This was mainly because of its ability to separate out Citrobacter spp as blue/purple colonies, therefore reducing the false identification of E coli, and the high number of pink isolates that were E coli (only six of 531 E coli isolates were not pink).

UTI medium and CPS ID2 performed similarly, each showing advantages and disadvantages in terms of sensitivity and PPV. Using E coli as an indicator, UTI medium gave a PPV of 97.8% compared with 99.4% for CPS ID2. This was the result of more pink colonies being E coli on CPS ID2 media than on UTI medium. Conversely, the sensitivity of UTI medium for E coli was greater than that of the CPS ID2 medium as a result of more E coli colonies being pink (eight colourless isolates on UTI medium compared with 22 on CPS ID2).

Although Harlequin CLED agar had a high PPV, it was of low sensitivity (only 414 of the 531 E coli isolates were green) and was not considered sufficiently accurate for the presumptive identification of isolates. This medium was therefore not evaluated further.

The interpretation of mixed cultures is controversial in midstream urines. There is no formal definition of a “mixed growth”, with some authors defining significance as a pure growth only, whereas others accept the criterion that mixed growths may be clinically relevant. Often they are regarded as evidence of collection contamination or bacterial overgrowth, and an indication for repeat culture. Graham and Galloway⁷ suggest that the interpretation of mixed growths can be considered as “more of an art than a science”, and that the isolation of a predominant organism within a mixed culture should be considered as clinically relevant. We recommend that antibiotics should be avoided until the significance of a mixed growth has been confirmed through the testing of a repeat sample. With improved methods of detection, there is now a need to re-evaluate the interpretation of mixed growths through a clinical trial of repeat sampling using a range of media in conjunction with clinical symptoms.

Our previous study⁵ confirmed the superiority of chromogenic media over CLED agar in the detection of mixed cultures and these results corroborate earlier studies. This is partially because of the highly visible turquoise colouration of colonies of faecal streptococci. The recovery of enterococci on chromogenic media varies, with CHROMagar and UTI medium performing better than CPS ID2. The detection of “other” organisms (Gram positives and yeasts) diminishes in the presence of increasing numbers of Gram negative organisms, as a result of the white or colourless appearance of the colonies on the chromogenic media. For Gram positive organisms and yeasts, CHROMagar and CPS ID2 performed better than UTI medium.

A cost comparison of the agars suggests that as the use of chromogenic agar in laboratories increases, the purchase cost is decreasing. At present, the cost of using a one quarter BBL CHROMagar plate equates to an additional laboratory cost of approximately 6 pence/sample over the use of CLED agar. This needs to be evaluated against the costs of repeat samples and avoided antimicrobial chemotherapy as a result of improved mixture detection.

“A cost comparison of the agars suggests that as the use of chromogenic agar in laboratories increases, the purchase cost is decreasing”

In conclusion, the comparison of the chromogenic agars indicates that BBL CHROMagar provides enhanced identification of organisms, thus supporting improved accuracy of surveillance monitoring of susceptibility data in support of the national antibiotic strategy.⁸ In addition, it performs better than UTI medium and CPS ID2 in terms of overall isolation rates and the detection of mixtures, and is therefore a suitable medium for routine use in urine analysis.