Setting: Cape Town, South Africa.
Objective: To develop a standardized, reliable measure of household tuberculosis (TB) exposure that considers child-specific risk factors.
Design: We assessed TB exposure in 536 children. Children were considered Mycobacterium tuberculosis infected if two of three tests of infection were positive. Principal component analysis identified a discrete set of components that collectively described exposure and contributed to a composite contact score. Logistic regression assessed the odds of having M. tuberculosis infection given increasing contact score while controlling for age and past TB treatment.
Results: Four components described 68% of data variance: 1) maternal TB and sleep proximity, 2) index case infectivity, 3) duration of exposure, and 4) exposure to multiple index cases. Components were derived from 10 binary questions that contributed to a contact score (range 1-10, median 5, 25th-75th interquartile range [IQR] 4-7). Among children aged 3 months to 6 years with household exposure, the odds of being M. tuberculosis-infected increased by 74% (OR 1.74, 95%CI 1.42-2.12) with each 1-point increase in the contact score.
Conclusions: Well-quantified TB exposure is a good surrogate measure of M. tuberculosis infection in child household contacts in a high-burden setting, and could guide targeted preventive treatment in children at highest risk of M. tuberculosis infection.