Setting: A tuberculosis programme run by a non-governmental organisation in eight hill and mountain districts of eastern Nepal.
Objective: To assess the impact of contact screening on case-finding.
Design: A retrospective cohort study of contacts of smear-positive, smear-negative and extra-pulmonary tuberculosis patients diagnosed and registered during 1996-1998 ('index cases'). Contacts, defined as household members identified by index cases, were screened by sputum examination; two positive smears were taken to indicate smear-positive pulmonary disease.
Results: Approximately 50% (668) of registered cases identified contacts; 75% (2298) of the contacts identified provided one or more sputum specimens. An overall smear-positive case yield of 0.61% (14) was obtained from contacts tested, all except one of which were contacts of smear-positive index cases. For smear-positive index cases with a smear grading of > or = 2+, the yield was 7.2 times greater (P = 0.04) than for those with a grading of 1+.
Conclusion: In this setting, sputum examination of household contacts of smear-negative and extrapulmonary tuberculosis cases is not justified. Further assessment is needed to evaluate the utility of testing contacts of smear-positive cases without symptom screening, and whether cost effectiveness can be improved by restricting testing to contacts of cases with high bacterial (> or = 2+) loads.