Risk factors for gastric aspirate culture contamination in children evaluated for tuberculosis in Botswana

• Published in: The international journal of tuberculosis and lung disease 2018-09, Vol.22 (9), p.1044-1050
• Main Authors: Ho-Foster, A., Tenforde, M. W., Arscott-Mills, T., Maramba, M., Sedigeng, P., Mbeha, B., Banda, F., Steenhoff, A. P.
• Format: Article
• Language: English
• Subjects: Bacteriological Techniques - methods; Botswana - epidemiology; Child; Child, Preschool; Children; Collection; Contamination; Cross-Sectional Studies; Culture; Diagnostics; Female; Gastric Juice - microbiology; Humans; Infant; Laboratories; Liquid culture; Logistic Models; Male; Median (statistics); Multivariate Analysis; Mycobacterium tuberculosis - isolation & purification; Pediatric; Regression analysis; Regression models; Resource-Limited Settings; Risk analysis; Risk Factors; Sputum; Sputum - microbiology; Statistical analysis; Statistical methods; Sub-Saharan Africa; Transport; Tuberculosis; Tuberculosis - diagnosis; Tuberculosis - epidemiology
• ISSN: 1027-3719; 1815-7920
• DOI: 10.5588/ijtld.18.0036

SETTING: Gastric aspirate (GA) sample culture is commonly performed in children evaluated for tuberculosis (TB) who cannot expectorate sputum. Contamination limits culture yield and negatively impacts care.OBJECTIVE: To evaluate the proportion of and factors associated with GA contamination at a central TB reference laboratory in Botswana.DESIGN: This was a 5-year cross-sectional study of untreated children aged 12 years evaluated for TB with the first GA sample registered at the National Tuberculosis Reference Laboratory. We performed descriptive statistics to assess the risk of contamination with patient age, sex, transport time and distance, culture medium, and facility type. We generated multivariable logistic regression models using generalized estimating equation extension.RESULTS: We analyzed 3642 samples. The median age was 2 years (interquartile range [IQR] 1-4), median transport time was 4 days (IQR 2-7), and 64.1% of samples were from clinics or health posts. TB culture positivity was 1.6% (60/3642), and contamination was observed in 35.6% (1298/3642). Hospital collection was associated with lower contamination risk (adjusted OR [aOR] 0.53, 95%CI 0.40-0.69) and Mycobacteria Growth Indicator Tube vs. Löwenstein-Jensen medium with higher risk (aOR 1.88, 95%CI 1.51-2.34).CONCLUSION: In routine care settings, high sample contamination and low TB culture yield were observed. This raises questions about the collection technique and storage in lower-level facilities and affirms higher risk with a liquid culture medium.