This paper reports on a study conducted in the year 1975 to estimate yield of tuberculosis cases from multiple sputum specimens, and work out correction factors to be applied to estimates based on small number of specimens. Eight sputum specimens were collected within a fortnight from each person with an abnormal chest X-ray during an epidemiological survey in 77 villages in a district of south India. Each specimen was examined by Ziehl-Neelsen technique of microscopy and culture. In all, 3,199 persons were referred for sputum examination and results of all the eight specimens were available for 1,652. Of the latter, 64 were culture positive.

The first specimen detected 58% of the culture positives and the additional positives by later specimens generally decreased. The contribution from the first specimen was 71% for cultures showing good growth and 19% for cultures with scanty growth. Similarly for positives on both culture and microscopy, first specimen detected 87% whereas the corresponding proportion was 32% for those positive only on culture. The type of specimen (viz., spot or overnight) and age or sex of the case did not influence the yield from multiple examinations. The precision of an estimate of prevalence will depend on the number of specimens on which it is based and the coverage obtained in the collection and examination of specimens. Correction factors to be applied to such estimates based on one or two specimens, for various levels of coverage have been presented. For example, an estimate of prevalence based on one sputum specimen with 90% coverage will have to be nearly doubled to get a more precise estimate. Using these correction factors, revised estimates of prevalence have been presented for a number of prevalence surveys conducted in India. It has been estimated that the total number of infectious cases in India at present may be at least 3 million, as against 2 million according to earlier estimates.