Targeting TB preventive therapy to those at high risk: derivation and validation of a risk score for predicting tuberculosis disease

Background. Use of tuberculosis preventive therapy (PT) must be urgently scaled up and provided to people at high-risk of developing tuberculosis disease (TB). In high-burden settings, PT is poorly utilised and rarely offered to adult household contacts. Furthermore, the tests for tuberculosis infection used to guide prescription are unreliable for predicting disease, fraught with logistical challenges and unavailable in many settings. In low- and middle-income countries a TB risk score could be used to prioritise PT for contacts most likely to benefit. We therefore aimed to develop a risk score to predict TB disease among adult household contacts of patients with TB in urban Callao, Perú. Methods. We identified index-cases with pulmonary TB (n=715), recruited their household contacts aged ≥15 years (n=2,017) and followed them for TB for a total of 18,988 person-years (PY). Cox proportional-hazards models were fitted to investigate factors associated with TB. 1,009 contacts were selected as a derivation cohort from which a risk score was created. Scores were calculated for each contact and low, intermediate and high-risk groups defined. The score was subsequently validated in the remaining 1,008 contacts. Results. Eight predictors formed the score: age 15-30 or ≥50; history of TB; body-mass-index; prolonged exposure to the index-case; poverty; exposure to indoor air pollution; male index-case and high index-case smear-positivity. In the derivation cohort the incidence rates in the low-, intermediate- and high-risk groups were 331/100,000PY (95%CI:200-550); 942/100,000PY (95%CI: 650-1363) and 2038/100,000PY (95%CI:1501-2768) respectively (p