Background: Tuberculosis (TB) is highly endemic in Rio de Janeiro State prisons. In addition to TB screening at entry and passive case detection, active case identification may be warranted.
Objectives: To develop and evaluate performances of scores aimed at identifying "tuberculosis suspects" in order to target TB screening among inmates.
Methods: Systematic chest X-ray screening was carried out in two prisons (n=1910). TB was diagnosed among individuals with X-ray abnormalities by sputum microscopic examination and culture or, if bacteriological results were negative, by response to TB treatment. Using this strategy as a reference, the clinical score proposed in WHO guidelines "TB Control in Prisons" was evaluated. Using the same variables in a logistic regression comparing TB and non-TB cases, another score was developed and evaluated. Finally, a 'new score', based on socio-demographic and clinical variables was developed and evaluated.
Results: When applied to our study population (prevalence of active TB: 4.6%), these scores missed many TB cases (sensitivities: 56%, 72%, 74%, respectively). Among the "TB suspects", the probability of finding TB cases was low (positive predictive value: 10%). The scores had high negative predictive values (>97%); specificities (75%, 60%, 67%) were low. Performances were similarly poor for smear-negative and smear-positive cases.
Conclusion: The scores investigated performed poorly and would be unhelpful to target TB screening. Therefore, systematic X-ray screening may be considered, at least during the initial stages of the reinforced TB programme, in order to reduce the impressive burden of TB.