Search tips
Search criteria 


Logo of jclinpathJournal of Clinical PathologyVisit this articleSubmit a manuscriptReceive email alertsContact usBMJ
J Clin Pathol. 2007 November; 60(11): 1290–1293.
Published online 2007 March 2. doi:  10.1136/jcp.2007.046854
PMCID: PMC2095479

Problems of basing patient recruitment for primary care studies on routine laboratory submissions



Use of routine urine submission rates for estimation of patient enrolment in primary care studies of acute urinary symptoms may overestimate patient recruitment rate.


To compare the rates of submission of urines and significant bacteriuria from patients presenting with acute urinary symptoms in study general practices to routine microbiology laboratory urines.


Routine laboratory urine submissions were determined by counting all mid‐stream urine specimens submitted to the laboratory from 12 large general practitioner (GP) practices served by Gloucester and Southmead microbiology departments over two years (2000–02). Urine specimens were requested from all patients with acute urinary symptoms referred at research nurse practice visits over the same time period. The annual study urine submission was calculated using the ratio of the number of nurse practice visits to the annual number of possible consulting sessions. Significant bacteriuria was defined as a urine growing a single organism reported as >105 colony forming units/ml. Rates per 1000 patients were calculated using practice population data.


The urine submission rate from study patients with acute urinary symptoms was one‐third the routine urine submission rate from the same practices. The significant bacteriuria rate attained from the study was less than half the routine significant bacteriuria rate.


Two‐thirds of routine urine samples submitted by GPs are probably not for the investigation of acute urinary symptoms. Basing consultation sample size power calculations for primary care studies or sentinel practice‐based surveillance in urinary tract infection on routine laboratory submissions is unreliable and will lead to significant overestimation of recruitment rate.

The use of microbiology laboratories by general practitioners (GPs) in England for the investigation of urinary symptoms and renal disease varies greatly.1 Patients with more severe symptoms, recurrence and the elderly are more likely to have specimens submitted.2 For the past five years the South West Antimicrobial Resistance Surveillance Group has been investigating sampling strategies in primary care. Patient enrolment from sentinel practices was slower in this study and others3,4,5 than expected from the sample size calculation based on routine laboratory data.

Antimicrobial resistance is a major public health concern, with a relentless rise in the hospital and community settings.6,7,8,9,10 Antimicrobial resistance surveillance systems are designed to inform disease management and guide empirical use of antimicrobials in patients presenting with suspected infections. Selection bias in antibiotic resistance surveillance causes distortion of the resistance rates detected due to the manner in which the patients (and thence isolates) are selected.11,12,13 In the primary care setting the distortion of resistance rates due to patient selection bias may be greater than in the hospital setting, as patients are often given empirical treatment and microbiology specimens are not routinely advised in uncomplicated infections.14

We wished to test the hypothesis that urine samples submitted to the routine diagnostic laboratory are not representative of urine samples that would be obtained if cultures had been provided from all patients presenting with acute urinary symptoms in general practice, and therefore, basing consultation sample size power calculations for primary care studies or surveillance on laboratory submissions is flawed.



This study was part of an ongoing investigation of antibiotic resistance surveillance in primary care and was approved by the Southmead and Gloucester Local Research Ethics Committees (062/99 and 99/51G respectively). Patients gave informed consent to be enrolled into the study.

Selection of the study practices

The study was undertaken in the general practices served by Gloucester and Southmead microbiology departments. We randomly selected practices with a patient population of over 7000. Practices were invited by telephone and then letter to participate. Twelve of the 18 general practices in Avon and Gloucestershire approached agreed to participate in the South West primary care antibiotic resistance surveillance study.3 Lack of time or space for the research nurse was the reason given for non‐participation by practices.

We compared rates of urine specimen submissions and significant bacteriuria, from patients with acute dysuria syndrome presenting to the research nurse in the selected general practices in the South West surveillance study,3 to submission of all mid‐stream urine (MSU) specimens to the two laboratories from the same practices over the same time period (26 months, 1 April 2000–31 July 2002).

Routine laboratory submission rate of urine specimens from general practices

This was determined by counting all urine specimens submitted to the two laboratories over the study period from patients [gt-or-equal, slanted]17 years of age, registered with the same 12 practices. Clinical details were not used to exclude any urine specimens. The submission rate (number of urines submitted per practice/1000 population served/year) was calculated using 2001 population data for each practice obtained from the primary care trusts. Urines were referred to the local microbiology laboratory (Southmead or Gloucester) on the routine transport system which visits practices each weekday. Specimens were cultured every day in the laboratory.

Rate of acute episodes of suspected urinary tract infection seen in study practices

On the days the research nurse visited, clinicians were asked to refer to her all patients aged [gt-or-equal, slanted]17 years with any signs and symptoms of a urinary infection of less than 10 days duration. Referral slips, which were readily accessible on each clinician's desk, were used to refer patients to the nurse immediately after their consultation with a clinician, during which normal care occurred. The nurse reminded the clinicians about the study at the beginning of each surgery. Patients gave oral consent, were asked to provide an MSU and were asked to complete a short questionnaire. The questionnaire confirmed that their urinary symptoms were indicative of urinary tract infection (UTI), and asked about urine infections in the past 12 months, current and past antibiotic usage and other significant health problems. The research urines were cultured at Southmead microbiology laboratory on the day of collection.

As there was only one research nurse, practices were only visited on a limited number of days (range 26–33 days per practice). Where possible, visits were systematically organised by the statistician so that the nurse visited each practice a similar number of times on each weekday. The number of patient episodes of acute urinary symptoms per year was calculated from the ratio of visits to possible days of practice consulting per year (247).

Significant bacteriuria

Urines were cultured according to national standard operating procedures.15 Significant bacteriuria was defined as a urine growing a single organism reported as >105 colony forming units (CFU)/ml. The number of specimens with significant bacteriuria was obtained from the surveillance study database and the two routine laboratory databases for the 12 practices over the same study period.

Data analysis

Urine specimen submission rates and significant bacteriuria rates per thousand population were determined using the general practice primary care trust 2001 population data. We then calculated the ratio of significant bacteriuria in routine urine specimens to study practice urine specimens.

We used linear correlation to determine if there was any association between the number of patients over 65 years of age in the practice and specimen submission or bacteriuria rate for routine and study specimens.


One study practice in Avon failed to recruit any patients with acute urinary symptoms, and was therefore removed from the analysis. The total patient population was 108 227.

The list size of the 11 practices ranged from 7016 to 13 400. Age/sex make‐up of the 11 practices was similar to the whole population served by the two laboratories and was a mixture of urban and rural practices and range of deprivation (table 11).).

Table thumbnail
Table 1 Comparison of routine mid‐stream urines submitted to the two laboratories, and urines submitted from patients with acute urinary symptoms from the same 11 practices over 26 months

Routine urine submissions

Over the 26 month period, the 11 practices submitted 117 649 urine specimens to the two laboratories, of which 85.5% were MSUs (table 11).). Significant bacteriuria ([gt-or-equal, slanted]105 CFU/ml) was present in 24.2% of MSUs. The rate of routine MSU submissions from the 11 practices to the two laboratories varied fivefold from 34 to 155/1000 patients per year (table 1). The submission rate for the Gloucester practices (mean 52.2, SD 13.1, median 59.0) was lower than that of the Bristol practices (mean 104.4, SD 31.4, median 88.0; p = 0.013). Cases of significant bacteriuria in these routine MSUs varied fivefold from 7 to 36 cases/1000 patients per year (table 11).

Urine specimens from patients with acute urinary symptoms

Over the 26 month study period, 133 patients with acute urinary symptoms were referred to the research nurse. All patients consented to provide a urine sample and each study practice referred between 8 and 19 patients. This equated to an annual submission rate of urine specimens from patients with acute urinary symptoms of between 10 and 38 specimens/1000 patient population (table 11).). Significant bacteriuria was found in 43% (57/133) of the patients with acute urinary symptoms. This equated to a significant bacteriuria rate of between 2.1 and 14.9/1000 patients per year.

Comparison of study practice and routine urine submission and bacteriuria rates

Urine submission rate from the study practices was approximately one‐third of the routine laboratory urine submission rates (table 11).). The significant bacteriuria rate from the study practices was less than half the significant bacteriuria found in routine laboratory urine submissions. It ranged from tenfold lower in practice 6B to equivalent in practice 9G compared to significant bacteriuria rate from the routine laboratory (table 11).

There was no association between the proportion of patients over 65 years in a practice, and practice routine urine submission and bacteriuria rates or study urine submission and bacteriuria rates (r2 values for the proportion of elderly to routine urine submissions 0.072 (p = 0.54); routine urine bacteriuria 0.112 (p = 0.45); study urine submission 0.135 (p = 0.42); study urine bacteriuria 0.229 (p = 0.33)).


Summary of main results

We found that the overall study urine submission rate from patients with acute urinary symptoms presenting to a research nurse in the 11 general practices was one‐third the routine urine submission rate from the same practices. The significant bacteriuria rate attained from the study was less than half the routine urine rate. The difference varied by practice and the study rate in one practice was ten times less than the routine rate. Increased submission was not necessarily related to a larger elderly patient population.

Interpretation of results

The rate of urine specimen submission to the routine diagnostic laboratory is not representative of study urine specimen submission rates from patients with acute urinary symptoms in general practice. This may be due to under‐referral of patients to the research nurse or, alternatively, many of the urine specimens sent routinely to the laboratory are not for the investigation of acute urinary symptoms. Although most urine specimen request forms state “query UTI” in the clinical details, it seems possible that they are submitted to exclude UTI rather than to confirm it. Doctors may send in urine specimens for reasons other than suspected infection (eg, suspected haematuria, investigation of incontinence or other urinary symptoms) or to rule out infection in patients with diagnostic difficulties (eg, non‐specific abdominal pain, or confusion in the elderly). Alternatively, some routine urines may be submitted without a surgery‐based consultation, directly from nursing homes, following a telephone consultation or following a home visit or out‐of‐hours consultation.


The strength of this study is that it was carried out as part of a well‐funded antibiotic surveillance study in a range of urban and rural practices in two geographical areas. Therefore, the results are likely to be generalisable to other UK primary care studies and sentinel general practice surveillance systems. Our routine laboratory submission and bacteriuria rate/1000 was similar to that found in a South West England study.1


We cannot rule out under‐referral of patients to the research nurse, although an audit of medical records of consultations on days on which the research nurse was in attendance in one practice (7G) did not reveal a high level of under‐referral. We consider that any under‐referral to the research nurse would be similar in other general practices involved in studies or sentinel surveillance. If we had more practice visits we may have been able to increase patient recruitment to the study; however, we had expected patient referral to be much greater, and therefore we only had one research nurse visiting the practices. Staffing will be a barrier to recruitment in any study or sentinel surveillance system run by visiting clinicians. As the number of patients recruited in the study practices was so low we were unable to compare antibiotic resistance rates in the two groups. As the specimen pots did not contain any borate, some overgrowth of bacteria may have occurred during transport of the routine urines. However we do not think this would have influenced our results, as there was daily specimen transport to the laboratory during surgery opening hours, and therefore there would have been minimal delays.

Other work in this area

In a spotter practice model, Richards asked clinicians from three general practices to submit MSUs from all patients presenting with symptoms suggestive of UTI.16 This spotter practice system did not involve research nurses and collected over 2000 urine specimens in a year, which is ten times the number in our study. The percentage of patients with significant bacteriuria rate was low (26%), suggesting that many of their patients did not have acute dysuria syndrome.5,14 In contrast, 1991 General Practice Morbidity Statistics found that the consultation rate with a doctor for cystitis was 147/10,000,17 which is lower than our mean urine study submission rate.

A laboratory survey of bacterial susceptibility in urinary tract pathogens showed that the prevalence of trimethoprim resistance in urinary pathogens isolated from primary care specimens doubled when general practices were asked to use urine dipsticks to guide management and to culture urines only in complicated UTI.18 A study investigating clinical outcomes in acute uncomplicated UTI in primary care found that the trimethoprim resistance rate was half that found in routine laboratory data collected over the same time period.5 These studies support the hypothesis that routine laboratory urine samples are not representative of urines from primary care patients with uncomplicated UTI.


Further research is needed to understand the reasons why GPs submit urinary samples to microbiology laboratories. Basing consultation sample size power calculations for primary care based studies in UTI or surveillance on routine laboratory submissions is unreliable and will lead to great overestimation of recruitment rate. It may be more accurate to base recruitment on general practice morbidity statistics,17 or searches based on Reid codes within the local general practices involved.

An alternative approach in antibiotic resistance surveillance is the development of programmes based on the potential pathogens carried within a population, similar to those used successfully for zoonotic organisms in farmed animals and patients in secondary care facilities.19 Most community‐acquired UTIs arise from the perineal flora; antibiotic resistance surveillance of perineal carriage in a representative asymptomatic community population may therefore be more feasible. Bruinsma et al attained a 25% response rate to a postal request for a stool specimen from a randomly selected Dutch population.20 The South West Antibiotic Surveillance Group is now exploring the feasibility and validity of postal surveillance methods, using first pass urine specimens.3

Take‐home messages

  • Urine submission rate from patients with acute urinary symptoms presenting to a research nurse in study general practices was one‐third the routine urine submission rate from the same practices.
  • Urine specimens submitted to the routine diagnostic laboratory are not representative of urine specimens that would be obtained if cultures had been provided from patients presenting with acute urinary symptoms in general practice.
  • Basing consultation sample size power calculations for primary care based studies or surveillance in urinary tract infections on routine laboratory submissions is unreliable and will lead to great overestimation of recruitment rate.
  • To obtain a sufficient sample size, an alternative recruitment approach may be needed in surveillance systems in primary care.


We would like to thank all the general practices that took part in this study; the laboratory staff, and Walter Palmer in microbiology at Gloucestershire Royal Hospital; Rosy Reynolds for statistical support for the paper; and Becky Carron and Emma Henderson for enrolling patients.


CFU - colony forming unit

GP - general practitioner

MSU - mid‐stream urine

UTI - urinary tract infection


Funding: This study was funded by the NHS South and West Research and Development Directorate.

Competing interests: None. (All authors work in the field of antibiotic resistance and could be considered to have vested interests in investment in this area, whether by governments, charities or industry.)


1. McNulty C A M, Bowen J, Clark G, on behalf of the South West GP Microbiology Laboratory Use Group et al How should general practitioners investigate suspected urinary tract infection? Variations in laboratory‐confirmed bacteriuria in South West England. Commun Dis Public Health 2004. 7220–226.226 [PubMed]
2. McNulty C A M, Freeman E, Nichols T. et al Laboratory diagnosis of urinary symptoms in primary care: a qualitative study. Commun Dis Public Health 2003. 644–50.50 [PubMed]
3. Lovering A M, MacGowan AP and SW England Enhanced Surveillance Group Comparison of enhanced and conventional surveillance for determination of the prevalence of antimicrobial resistance in E coli. 43rd Annual ICAAC, C2‐2098 2003
4. Hay A D, Thomas M, Montgomery A. et al The relationship between primary care antibiotic prescribing and bacterial resistance in adults in the community: a controlled observational study using individual patient data. J Antimicrob Chemother 2005. 56146–153.153 [PubMed]
5. McNulty C A M, Richards J, Livermore D M. et al Clinical relevance of laboratory‐reported antibiotic resistance in acute uncomplicated urinary tract infection in primary care. J Antimicrob Chemother 2006. 581000–1008.1008 [PubMed]
6. Gupta K. Emerging antibiotic resistance in urinary tract pathogens. Infect Dis Clin N Am 2003. 17243–259.259
7. Department of Health Standing Medical Advisory Committee on Antimicrobial Resistance The path of least resistance. Occasional report. London: The Stationery Office, 1998
8. Woodford N, Ward M E, Kaufmann M E. et al Community and hospital spread of Escherichia coli producing CTX‐M extended‐spectrum beta‐lactamases in the UK. J Antimicrob Chemother 2004. 54735–743.743 [PubMed]
9. Donnan P T, Wei L, Steinke D T. et al Presence of bacteriuria caused by trimethoprim resistant bacteria in patients prescribed antibiotics: multilevel model with practice and individual patient data. BMJ 2004. 3281297
10. Magee J T, Pritchard E L, Fitsgerald K A, on behalf of the Welsh Antibiotic Study Group et al Antibiotic prescribing and antibiotic resistance in community practice: retrospective study 1996–8. BMJ 1999. 3191239–1240.1240 [PMC free article] [PubMed]
11. Gaynes R. Surveillance of antibiotic resistance: learning to live with bias. Infect Control Hosp Epidemiol 1995. 16723–726.726
12. Steinke D T, Davey P G. Association between antibiotic resistance and community prescribing: a critical review of bias and confounding in published studies. Clin Infect Dis 2001. 33(Suppl 3)S193–S205.S205 [PubMed]
13. Livermore D M, Threlfall E J, Reacher M H. et al Are routine sensitivity data suitable for the surveillance of resistance? Resistance rates amongst Escherichia coli form blood and CSF from 1991–1997, as assessed by routine and centralized testing. J Antimicrob Chemother 2000. 45205–211.211 [PubMed]
14. Hummers‐Pradier E, Kocken M M. Urinary tract infections in adult general practice patients. Br J Gen Pract 2002. 52752–761.761 [PMC free article] [PubMed]
15. PHLS Standard Operating Procedure: Investigation of urine. Ref No BSOP41, Issue No 1 02.04.1998. Technical Services, PHLS HQ, London. HPA Issue No 5 28.02.05. http://www.hpa‐ (accessed 11 January 2007)
16. Richards J. Monitoring antibiotic resistance in urinary isolates form the community—a spotter practice model. Commun Dis Public Health 2002. 5226–229.229 [PubMed]
17. Office of Population Censuses and Surveys Morbidity statistics from general practice, fourth national study 1991–1992. Series MB5 No 3. London: HMSO, 1995
18. Batchelor B I F, Crook D W M, Jones T. et al Impact of guidelines for the diagnosis of urinary tract infection on trimethoprim susceptibility of Escherichia coli. J Antimicrob Chemother 2002. 49223–224.224 [PubMed]
19. Wray C, Gnanou J C. Antibiotic resistance monitoring in bacteria of animal origin. J Antimicrob Agents 2000. 14291–294.294
20. Bruinsma N, Filius P M G, de Smet P A G M. et al Antibiotic usage and resistance in different regions of the Dutch community. Microb Drug Res 2002. 8209–214.214

Articles from Journal of Clinical Pathology are provided here courtesy of BMJ Publishing Group