Data from 62 randomly selected villages in a district of south India, which formed part of a prevalence survey carried out by the National Tuberculosis Institute, Bangalore, during 1960-61, has been made use of. The survey covered 29,813 persons in 5,266 households. There were 70 cases with bacilli demonstrable either in smear or culture and 300 suspect cases. Using the village map (prepared by survey staff), ‘case clusters’ were formed first, with each case household as nucleus and adjacent households within a maximum distance of about 20 meters on either side of the case households. Households closest to the nucleus household on either side have been called as 1st neighbourhood and those coming next in proximity on either side as a 2nd neighbourhood and so on. The case household and its four neighbourhood together was called a cluster. If another case household was found within 4th neighbourhood of the first case the cluster was extended by including the 4th neighbourhood of the new case also. Such clusters were called composite case clusters and clusters with only one case household as simple case clusters. Similarly, suspect case clusters were formed and differentiated as simple suspect clusters or composite suspect clusters. Further, to serve as a control group, non-case clusters were constituted from a systematic sample of 10% households that were not included in case or suspect case clusters.

Out of 60 case clusters formed, only 7 have multiple cases showing that there was no evidence of high concentration of disease in case clusters. While the percentage of child contacts (0-14 years) infected was considerably higher in case clusters (25.8%), there was not much difference between suspect case clusters (14.9%) and non-case clusters (9.8%). Similarly, there was not much difference between simple and composite clusters. Infection among child contacts was higher in case households as compared to their neighbourhoods. To get some idea of the zone of influence of a case or suspect case, prevalence of infection was studied for 10 neighbourhoods, in simple clusters to avoid the influence of multiple cases. It appeared that the zone of influence of a case may extend at least upto the 10th neighbourhood. It was also noted that there was very little difference between zones of influence of suspect cases and non-cases. Case clusters in which the nucleus case had shown activity of lung lesion (evident on X-ray reading) or had cough showed significantly higher infection among child contacts. Clusters around cases positive on both smear and culture did not show higher infection than those around cases positive on culture only. (This may be due to sputum examination of single specimen only).

Out of the total infected persons in the community, only 2% were in case households and 7% in suspect case households, over 90% being in non-case households. The zone of influence of a case extending at least upto the 10th neighbourhood and the overlapping of such zones of influence of cases, present and past, seems to be the most probable explanation for the wide scatter of infection in the community. Prevalence of infection among child contacts was definitely higher in case clusters. But, the significance of this could be understood only from a study of the incidence of disease during subsequent years in different types of clusters. It is significant that only 10% of the total infected persons in the community were found in case clusters. The case yield in general population, cluster contacts, household contacts and symptomatics attending general health institutions have been also compared. The case yield in the last group (10%) is much higher than the case yield from both types of contacts (0.7% and 0.6%) which where only slightly higher than the case yield from the general population (0.4%).