Assessment of urine analysis for the diagnosis of tuberculosisBMJ 1996; 312 doi: https://doi.org/10.1136/bmj.312.7022.27 (Published 06 January 1996) Cite this as: BMJ 1996;312:27
- Emmanuel Mortier, physiciana,
- Jacques Pouchot, physiciana,
- Laurence Girard, physiciana,
- Yves Boussougant, microbiologistb,
- Philippe Vinceneux, professora
- a Department of Internal Medicine, Hopital Louis Mourier, University of Paris VII, 92700 Colombes, France
- b Department of Microbiology, Hopital Louis Mourier, University of Paris VII, 92700 Colombes, France
- Correspondence to: Dr Mortier.
- Accepted 28 September 1995
The number of reported cases of tuberculosis has been increasing steadily over the past few years. Due to the complex and time consuming procedures needed for the detection and isolation of Mycobacterium tuberculosis, the workload of the microbiology laboratory is substantial. Many recently developed technologies—including blood cultures performed with lysis-centrifugation, genomic detection by polymerase chain reaction, or detection of mycobacterial antigens by immunoassays—will modify the diagnostic approach of tuberculosis.1 As most of these new diagnostic procedures are not yet available routinely, sputum and urine remain the most common clinical specimens submitted to conventional microbiological study. To improve the quality and efficiency of medical care, we tried to assess the usefulness of urine analysis for the diagnosis of tuberculosis.
Patients, methods, and results
In this retrospective study, conducted between July 1983 and July 1993 in a 500 bed hospital, all patients with positive urine cultures for M tuberculosis were identified from microbiology records. During this 10 year period, 7200 midstream morning urine specimens obtained from 2814 patients (mean (SD) age 48.9 (22.9) years; 63% male) were submitted to microbiological examination for a presumptive diagnosis of tuberculosis. All urine specimens were inoculated onto Lowenstein-Jensen medium. Only 65 (0.9%) urine samples obtained from 33 patients (1.2%) yielded positive cultures for M tuberculosis. In this group, acid fast bacilli in direct Ziehl-Nielsen stained smears were detected in eight of the 19 samples processed. Urine analysis performed on non-centrifuged samples of 29 of these 33 patients gave normal results for 21 of them, including three of the seven patients with genitourinary tuberculosis. The urinary sediment of the remaining patients showed pyuria (>/=10000 cells/ml) in all eight and haematuria (>/=5000 cells/ml) in five.
Of the 33 patients with positive urine cultures for M tuberculosis, 22 presented with pulmonary tuberculosis as the main site. Direct smears of sputum were positive for acid fast bacilli in 15 of them (68%). Seven patients had tuberculosis limited to the genitourinary tract as diagnosed by clinical signs and symptoms. No additional patients with genitourinary tuberculosis were identified from pathology records during the study period. The remaining patients had gastrointestinal, pleural-peritoneal, miliary, and meningeal involvement (table). Except for the patients with tuberculosis of the genitourinary tract, for whom urine culture was diagnostic, identification of M tuberculosis in urine did not allow a more prompt diagnosis of tuberculosis already established with other specimens.
Of patients with pulmonary tuberculosis, 5-8% have positive urine cultures for M tuberculosis even though there are no signs, symptoms, or laboratory data that suggest genitourinary tract involvement.2 3 4 No predictive factors of bacilluria in pulmonary tuberculosis have been identified so far. In our study, among the 19 patients with bacilluria and pulmonary tuberculosis in whom urinary sediment analysis was available, only three (16%) had abnormal results on urinalysis, which is consistent with published reports.2 3 4
These data suggest that submission of urine specimen to the microbiology laboratory for identification of M tuberculosis is rarely useful in the diagnostics of tuberculosis. Except in patients with genitourinary tuberculosis, who usually present with clinical signs and symptoms justifying the sampling,5 urine analysis does not provide an earlier diagnosis. Moreover, bacilluria does not indicate the need to modify the recommended therapeutic regimens for the patients with pulmonary tuberculosis, nor does it purport a worse prognosis.
The cost of this diagnostic procedure should also be considered. In our study, the 7200 urinalyses incurred a cost of about 72000 ecu (approximately pounds sterling58000) and were beneficial only in the seven patients with genitourinary tuberculosis.
The authors acknowledge the help of Mrs Cances and Chandemerle and Dr Barge in collecting the data.
Conflict of interest None.