SUMMARY
Guidelines available for the diagnosis of pulmonary tuberculosis (TB) in children vary widely. In an area of high prevalence, pulmonary TB was often suspected but difficult to confirm. In a prospective study, diagnostic methods were recorded in 144 consecutive children admitted and diagnosed as having pulmonary TB. Six of the 144 children had bacteriologically confirmed tuberculosis. Seventy-five children had probable tuberculosis, the diagnosis being based on the tuberculin response in 19, a history of family contact in 34 and a diagnostic chest radiograph in 22. The remaining 63 children had suspected tuberculosis: 23 of these had an inconclusive chest X-ray. The suspected group were significantly younger than the probable group (mean ages 2·8 and 4·4 years respectively), experienced a longer delay between admission and the start of anti-tuberculous chemotherapy and suffered a significantly higher mortality (30% and 8% respectively). A diagnostic process is proposed which takes account of the high proportion of non-responders to tuberculin in a young, malnourished population. A trial of anti-tuberculous therapy is accepted as a valid diagnostic manoeuvre in suspected cases who are malnourished or have recently had measles.