Diversity of urine bacteriology practice in Ontario: An external quality assessment

Members of the Microbiology Committee: Robert Lannigan MD FRCPC, Jane Gun-Munro MLT FIMLS FCSLT, Donald E Low MD FRCPC, Josephine Palmer MLT ART, Donna Thompson MSc MLT Correspondence and reprints: Dr Harold Richardson, Director, Laboratory Proficiency Testing Program, Suite 501, 250 Bloor Street East, Toronto, Ontario M4W 1E6. Telephone 416-323-9540 ext 242, fax 416-323-9324, e-mail richardson@lptp.on.ca Received for publication December 14, 1995. Accepted March 8, 1996 H RICHARDSON, CA FLEMING, AMR MACKENZIE AND THE MICROBIOLOGY COMMITTEE. Diversity of urine bacteriology practice in Ontario: An external quality assessment. Can J Infect Dis 1996;7(4):243-246.

C ulture is usually performed on voided urine, therefore, quantification and specific identification of the organisms in the urine are necessary to discriminate between contaminants and true pathogens (1).A count of 10 5 colony forming units (CFU)/mL is generally accepted as the criterion for the diagnosis of cystitis even though this may not be valid (2).In consequence, quantitative urine culture is practised widely, yet it is expensive, time-consuming and labourintensive.These challenges were compounded by the consensus decision arising from the Metric Commission Canada to adopt the Système International d'Unités (SI) in the health sector and to begin to report in these units effective 1981.The SI convention requires a urine bacterial colony count to be expressed as number per litre rather than number per millilitre.The number cannot exceed three digits, and exponents of the power of 10 are to be expressed in simple multiples of three; thus, a count of ³10 5 CFU/mL is expressed as ³100´10 6 CFU/L.A decade and a half later this requirement is not widely known.This external quality assessment of urine bacteriology practice in Ontario was carried out to assess the ability of licensed laboratories to quantify uropathogens accurately and to issue appropriate quantitative reports in compliance with SI requirements.

MATERIALS AND METHODS
Ontario laboratories licensed to perform urine bacteriology received a simulated urine sample as one of four separate challenges in a proficiency testing survey.The survey start date was May 12, 1995 and results were required to be returned to the Laboratory Proficiency Testing Program (LPTP) within 17 days.The sample was Mueller-Hinton broth containing a strain of Escherichia coli at a target concentration of 100´10 6 CFU/L stabilized by the addition of boric acid (290 mmol/L) and sodium formate (132 mmol/L) (3).The sample was prepared and distributed by the Department of Microbiology, University Hospital, London, Ontario under contract to LPTP as the supplier of microbiology testing materials.The distribution occurred during May to avoid wide temperature fluctuation.A courier service provided next-day delivery after dispatch from the supplier with the samples held at ambient temperature during transit.Stability of the target count was established through repeated testing at intervals by the seven designated LPTP reference laboratories before the participant survey.
Participants were instructed to assume that the sample was a freshly collected 'clean-catch' midstream urine specimen for culture and susceptibility testing.It was to be processed by the routine procedures used in each laboratory.The date and time of receipt of the sample, the date tested, the bacterial count, the associated units and the method used to determine the count were required to be recorded on the regular fixed format LPTP Analysis Worksheet.Additionally, participants were required to submit to LPTP a copy of their findings using the format they would normally use in reporting to a user-physician.
Consensus reference values were derived from the results of the reference laboratories managed by four designated teaching hospitals, one community hospital, one public health laboratory and the supplier.Data capture, analysis and review of results followed previously described LPTP procedures (4).Reference results were sent to all participating laboratories within two days of the closing date of the survey.A Survey Report, Committee Comments and the cumulated results along with recommendations for 'best practice' were distributed within six weeks of the survey.

RESULTS
Data for analysis were available from 182 laboratories.The following aspects of the survey were evaluated separately: the appropriateness of the methods used for quantification and the accuracy of the counts reported on the Analysis Worksheet; and the conformity of the final reports with SI convention.

Quantitative urine results:
The results obtained by the reference laboratories are shown in Table 1.Counts were determined over a seven-day period to evaluate the stability of the material over time.The reference laboratories reported a variety of methods, primarily the semiquantitative method using a calibrated 0.001 mL loop.The mean count for all reference laboratories on repeat testing varied from 95 to 139´10 6   Richardson et al mant colonies (CFU)/L a été soumis à des laboratoires de bactériologie ontariens autorisés.Les résultats sur l'identification et le dosage de l'organisme pathogène ont été demandées à l'intérieur d'un temps précis.Les valeurs de référence ont été déterminées par une entente consensuelle sur les résultats des sept laboratoires désignés.PARTICIPANTS : Le test a été offert à 182 laboratoires ontariens autorisés à effectuer des analyses bactériologiques sur l'urine.On n'a établi aucune stratification selon le type ou la complexité des installations.PRINCIPAUX RÉSULTATS : Les échantillons ont été traités selon la marche à suivre habituelle, la date et l'heure de réception de l'échantillon, la date de l'exécution du test, la numération bactérienne, les unités des dosages qui y sont associées, la méthode utilisée étaient les données demandées.On a également demandé que les résultats soient imprimés sur la formule de rapport habituellement utilisée par le laboratoire à l'intention des médecins utilisateurs.RÉSULTATS : L'organisme a été isolé et correctement identifié par 179 laboratoires.Cinquante-huit pour cent seulement des laboratoires ont signalé une numération de 100x10 6 CFU/L ou plus, 42% signalant une numération située entre 10 et 100x10 6 CFU/L.La majorité ont utilisé la méthode en boucle à 0,001 mL standard.Quatre-vingt-sept participants seulement ont déclaré utiliser la notation correcte en unités SI, bien que 65 autres ont présenté les résultats en CFU/L.CONCLUSIONS : La variété des formules de rapport utilisées semble poser des problèmes.Le traitement et la consignation des résultats devraient être standardisés.Les laboratoires devraient également fournir une note explicative ou une interprétation lorsque la nomenclature ou le format de rapport est modifié.CFU/L.The mean count in individual reference laboratories varied from a low of 94´10 6 to a high of 195´10 6 CFU/L.The true value as determined by the supplier, using a calibrated pipette and pour plate procedure, was 125´10 6 CFU/L.The intrareference laboratory coefficient of variation for repeat testing varied from 4.4 to 31.1.The inter-reference laboratories coefficient of variation ranged from 8.1 to a high of 41.5.The target count did not decay over the seven days of repeat testing by the reference laboratories.
Colony counts reported by participants on the fixed format LPTP Worksheet are summarized in Table 2. Since the Analysis Worksheet allowed for reporting counts as less than, equal to or greater than, it is not possible to provide a frequency distribution of the counts.To enable easier comparison, all counts have been converted from the units reported by the participants into CFU/L.Eight laboratories did not provide a count and 21 participants reported counts using incorrect units that could not be converted into CFU/L.These results have not been included in the analysis.Sixty-four of 152 evaluable counts were less than 100´10 6 CFU/L and, of these, 58 were below the lowest mean value reported by any of the reference laboratories.Only 88 participants elected to report a count of 100´10 6 CFU/L or more.
The quantitative and semiquantitative methods used by the participating laboratories were variable.One hundred and twenty-eight participants used a 0.001 mL loop method, with 94 of these opting for a calibrated nondisposable loop.There was no difference in the distribution of counts between users of disposable and nondisposable loops.Twenty-seven used the optional 0.01 mL loop and 25 some form of a dipslide.One participant used a 0.025 mL loop and one did not specify the method.

Review of laboratory report forms:
The reports submitted by laboratories, using the regular form and format they would use if reporting to a physician-user, were reviewed and data abstracted by the LPTP microbiology technologist.A summary of the various reporting formats and frequency of use by laboratories is presented in Table 3.Only 87 of 182 participants reported the count using the correct notation of SI units, although a further 65 did report as CFU/L.Of these 65 laboratories, 47 had the count correct but used the wrong exponent to express the result.Three laboratories continued to report as CFU/mL.Six reported a count but did not indicate the associated unit volume.Sixteen of the 18 that did not provide a count did indicate a heavy or significant growth.Two did not indicate a count on the report form and two did not submit the requested document.

DISCUSSION
This was the first time that LPTP participants were presented with the challenge of performing a colony count on a simulated midstream urine specimen, and it was intended to provide an opportunity to evaluate current laboratory practice with regard to quantitative urine testing and reporting.The diversity demonstrated by this external quality assessment indicates that Ontario laboratories show marked variation in reporting colony counts and in complying with the SI unit reporting notation.
The practice of quantitative or semiquantitative bacteriology on midstream urine specimens is well established and clearly described (3,5).It is based on the assumption that, in a correctly taken and expeditiously cultured or appropriately preserved midstream urine specimen, a count of 100´10    More recently the concept of significant bacteriuria has been redefined to take into account the clinical presentation and the epidemiological setting.Optimal colony counts for defining infection differ when patients are subdivided by sex, by presence or absence of symptoms and by method of specimen collection (6).The suspension of organisms used in this survey had a density of just over 100´10 6 CFU/L and was intended to give an unequivocal significant result, even allowing for the revised definitions.The intra-and interreference laboratory coefficients of variation for reported counts confirm the inherent imprecision of the currently recommended methods for quantification of bacteriuria.Only 58% of laboratories reported a count of 100´10 6 CFU/L or more, with 42% reporting a count between 10 and 100´10 6 CFU/L, which many laboratories would consider of 'doubtful significance'.Laboratories obtaining a count less than 100´10 6 CFU/L were strongly advised to review their urine plating procedures for errors known to be associated with low counts.Emphasis was placed on the correct use of a calibrated loop of the recommended size.Most of the participating laboratories used either a nondisposable or a disposable plastic loop for plating the urine.The appropriate loop to use is 0.001 mL.This loop, with an internal diameter of approximately 1.5 mm, is smaller than the loop used for most routine microbiology procedures (an internal diameter of 5 mm or more).A 0.001 mL loop would give a colony count for this specimen of approximately 100.Twenty-seven laboratories reported the use of a 0.01 mL loop.This would have yielded growth beyond the countable range of colonies with this sample (approximately 1000).The use of a 0.01 mL loop is considered inappropriate for routine midstream urine specimens.The recommended method of charging the loop, inoculating the plate and inspection and calibration of the loop must be followed to maintain the integrity of the method.
The variety of reporting formats for quantitative urine results among Ontario laboratories is of concern because of the potential for misinterpretation of these reports by physicians.It is appreciated that laboratories may have come to an agreement with their users and that local practitioners may have become accustomed to particular reporting styles.Nonetheless, physicians may receive urine microbiology reports from more than one laboratory and it is reasonable to expect that all reports will conform to a common format to avoid confusion and the potential for misinterpretation.A few of the reports received were so eccentric in their presentation that it is doubtful whether they would be correctly interpreted by any receiving physician.Interlaboratory standardization of reports is as much a requirement for urine colony counts as it is for other measurements, such as blood glucose.
There is no doubt that the decision, taken in Canada some 15 or so years ago, to adopt the SI notation in the health field continues to cause difficulties for laboratories reporting quantitative urine bacteriological assessment.It is likely that confusion exists among treating physicians since most have learned from the predominantly American medical literature that the significant colony count is greater than 10 5 /mL.The Canadian SI Manual (7) does not make specific reference to quantitative bacteriology and the use of exponents.The general guide applies and reports should be in the format of colony counts expressed as number/L (as opposed to number/mL) with the number not to exceed three digits (ie, 1 to 999) and exponents to be expressed in multiples of three (eg, 10 3 , 10 6 , 10 9 ).A count expressed as 10 7 or 10 8 CFU/L is not a correct use of the SI notation.Many computer systems cannot print superscript index numbers, in which case the format 'n x 10E6 CFU/L' is acceptable with a careful explanation of this terminology.LPTP recommends that laboratories provide an explanatory note or interpretation on all laboratory reports when the nomenclature or format is changed from that previously provided to physician or other health care users.

Urine bacteriology practice in Ontario
6/L or more of a single isolate indicates true infection, a count in the

TABLE 2 Urine colony counts* reported by participants on Laboratory Proficiency Testing Program Worksheets
*Corrected to colony forming units (CFU)/L

TABLE 3 Formats used in reporting to user-physicians
to 100´106/L is commonly associated with infection but requires evaluation on the basis of clinical information, and a count of less than 10´10 6 /L is more likely to indicate urethral or vaginal contamination of the specimen.
CAN J INFECT DIS VOL 7 NO 5 SEPTEMBER/OCTOBER 1996 range of 10