AK Chakraborty: Indian J TB 1993, 40, 215-25.

In a chronic disease like tuberculosis, the exact levels of prevalence or incidence of infection and disease are of lesser importance than its time trend. Surveys should be conducted repeatedly if possible, in order to study the latter. Longitudinal surveys, conducted by National Tuberculosis Institute (NTI) & New Delhi TB Centre, could provide information only on the incidence and prevalence of the disease & infection and not on the time trend due to inadequate sample size of the population selected for the surveys. To measure an annual decline of 1% after 12 years, NTI should have taken a population of 4,45,000 for Tumkur survey instead of 35,000 actually taken. An attempt to measure the trend with the help of epidemetric model also suffers from the inherent infirmity of the small population size. It gave little statistical support to the coefficient of variations of the observed rates, thus imparting little discriminatory power to the observed rates. The error of taking inadequate sample size of the population for these surveys, could be attributed to: (1) The statistical concept of epidemiological assessment through repeated measurement of TB problem had not yet concretised in the minds of the Epidemiologists and Programme Planners. (2) A very high rate of decline was expected after the implementation of the District TB Programme (DTP). (3) The purpose of longitudinal surveys was to get information only on the incidence of infection & disease and not to measure the change. (4) It was not envisaged in 1962 when DTP was being formulated, that there would be no change situation in the prevalence rate of tuberculosis after implementation of DTP from that found in National Sample Survey carried out during 1955-58. The hypothesis underlying static situation was formulated by the Indian epidemiologists later taking their clue from Grigg's momentous work.

Mean time it was established that the Annual Risk of Infection (ARI) holds the key for evaluating the epidemiological trend in a community. From the available data from Longitudinal Survey of NTI it has been found that almost identical rates of ARI were calculated as incidence rates of infection actually observed during the initial surveys. Over a period of 23 years, there has been an annual decline in the risk of infection for the area at the rate of 3.2%. Estimation of incidence of smear positive cases on the basis of the ARI could be made (1% ARI being equivalent of 50 cases per 100,000 population). The findings commensurate with observations made 23 years later, wherein incidence of cases was observed 23/100,000 population and ARI of 0.6% (a parametric relationship seen). The programme operation of average 33% efficiency for nearly three decades would give an annual declining trend of the following extent: 1.4% in case rate, 2.0% in smear positive case rate and 3.2% in ARI. Alternatively the above trend could also represent the natural dynamics.