REGISTRATION OF CANCER MORTALITY DATA IN A DEVELOPING-AREA - CHENNAI (MADRAS, INDIA) EXPERIENCE

Objectives: This study was carried out to evolve a method to improve t he registration of cancer mortality data in Chennai (Madras, India). M ethods: Data on cancer deaths have been collected from the Vital Stati stics Department (VSD) by a population-based cancer registry (PBCR) in Chennai only since 1982. The low mortality-to-incidence ratio during 1982-84 suggested under-registration of mortality data. Since 1985, th e PBCR has taken special effort to ascertain the vital status of cance r cases by sending reply-paid postcards and/or making house visits. Th e data on all deaths occurring in Chennai, irrespective of stated caus e of death in the death certificate, have been collected from the VSD since 1992. Results: Deaths that occurred in Chennai and obtained by s ending reply-paid postcards and/or making house visits were registered in VSD as non-cancer causes of death; hence, these data were not coll ected from VSD. The sensitivity and positive predictive values of deat h certificates on cancer diagnosis based on 1992 and 1993 mortality da ta were 57 percent and 99.5 percent, respectively. Conclusion: Since t he accuracy of death certificate information on cancer diagnosis is re latively low in a developing country such as in India, collecting data on all deaths will improve the mortality data registration in PBCRs.