Use of simulated sputum specimens to estimate the specificity of laboratory-diagnosed tuberculosis

• Published in: The international journal of tuberculosis and lung disease Vol. 14; no. 8; pp. 1016 - 1023
• Main Authors: DEMERS, A.-M, BOULLE, A, WARREN, R, VERVER, S, VAN HELDEN, P, BEHR, M. A, COETZEE, D
• Format: Journal Article
• Language: English
• Published: Paris, France IUATLD 01.08.2010; International Union against Tuberculosis and Lung Disease
• Subjects: Adolescent; Bacterial diseases; Biological and medical sciences; Computer Simulation - utilization; Cross-Contamination; Diagnosis, Differential; False Positive Reactions; Human bacterial diseases; Humans; Incidence; Infectious diseases; Medical sciences; Mycobacterium tuberculosis - isolation & purification; Pneumology; Quality Assurance; Quality Assurance, Health Care; Respiratory system : syndromes and miscellaneous diseases; Sensitivity and Specificity; Simulated Sputum; South Africa - epidemiology; Sputum - cytology; Sputum - microbiology; Tuberculosis; Tuberculosis - diagnosis; Tuberculosis - epidemiology; Tuberculosis - microbiology; Tuberculosis and atypical mycobacterial infections; South Africa; Respiratory disease; simulated sputum; Mycobacterial infection; Laboratory; Contamination; Infection; Quality assurance; Specificity; Tuberculosis; Sputum; cross-contamination; Bacteriosis; Diagnosis; Pneumology
• ISSN: 1027-3719; 1815-7920

SETTING: Cross-contamination is not uncommon in mycobacteriology laboratories of high-income countries, as documented by bacterial genotyping. The extent of this problem in low-income countries is largely unknown, where this method is impractical.OBJECTIVE: To estimate the rate of cross-contamination in a high-volume tuberculosis (TB) laboratory in South Africa.DESIGN: Simulated sputum specimens labelled with false names were sent from a TB clinic, interspersed with patient samples, and processed for culture and microscopy. Results were interpreted in the context of the observed proportion of samples with positive microscopy and culture results.RESULTS: With microscopy, 6/190 (3.2%) simulated specimens were positive (estimated specificity = 96.8%). Considering the 881 positive microscopy results in 6093 clinical samples, we extrapolate that 19.3% (95%CI 7.0-42.8) of positive smears were false-positives. On culture, 2/190 (1.1%) of the simulated specimens were positive for Mycobacterium tuberculosis (estimated specificity = 98.9%). Considering the 1862 positive cultures from 6093 clinical samples, we estimate that 2.4% (95%CI 0.3-8.8) of positive cultures were false-positives.CONCLUSION: Simulated specimens offer a simple means of estimating the proportion of false-positive results, providing information on all sources of potential error from the clinic, through the laboratory and to reporting of results.