Raj Narain, SS Nair, P Chandrasekhar & G Ramanatha Rao: Indian J TB 1965, 13, 5-23.

The incidence of infection with mycobacterium tuberculosis is an index of the risk of infection to which a community is exposed. An accurate estimation of incidence rate is of considerable importance in understanding the epidemiology of tuberculosis in organising control measures. A new method of estimating incidence of infection is discussed. The material from 3 studies of National TB Institute has been utilized. Study I: is a part of a survey of a random sample of 134 villages. No previous tuberculin testing or BCG vaccination had been carried out in the area, but each person was examined for BCG scar in order to exclude persons vaccinated probably from other areas. After a complete census, a Mantoux test with 1 TU of PPD RT 23 with Tween 80 given on two occasions (Round I and II). Those with reaction of 13mm or less at Round I were offered a test with 20 TU with Tween 80 within a week of 1 TU test. The interval between the rounds was about 18 months. From the analysis of the data from the first 50 villages for which complete information for both rounds was available, it was seen that there was a general increase in the size of reactions elicited in the second round. Study 2: tuberculin testing was carried out with 1 TU and 20 TU among selected ‘control’ groups which provided the data regarding the “enhancing of tuberculin allergy” seen in repeat tuberculin tests. Study 3: in the course of the longitudinal “survey reader assessments” were carried out periodically to judge the standards of the tuberculin test readers. Inter & intra-reader comparisons were made. The findings have been used to estimate the magnitude of reader variation. The data was also used to study variations in the technique of testing and reading.

It was estimated that on an average inter & intra-reader variations between the rounds were unlikely to exceed 6mm or more in more than 5% of the observations. The reading errors have an equal chance of being positive or negative except at extreme ends of the distribution where zero readings of Round I can only show an increase, and the very large reactions had a greater chance of showing only a decrease at a subsequent round. The study mainly concerns with the problems of estimating the incidence of tuberculous infection in a community. Calculations based on age-specific prevalence rates or on rates of tuberculin conversion or both subject to gross error, leading to unreliable epidemiological conclusions. For estimating the newly infected, a new approach has been suggested based on the drawing of a curve for the distribution of differences in reaction size from one round of tuberculin testing to another. It is assumed that if new infection causes a distinct rise in the degree of tuberculin sensitivity which is greater than the combined rise due to enhancement and reader variation, the distribution of differences between the rounds should indicate the newly infected. It is shown that the newly infected probably constitute a homogeneous group with an increase in mean reaction size of about 24mm and standard deviation of 4mm. Accordingly, 98% of the newly infected show an increase in reaction size of 16mm or more.