A model approach to calculate cancer prevalence from 5 years survival data for selected cancer sites in India--part II.

Abstract

Prevalence is a statistic of primary interest in public health. In the absence of good follow-up facilities, it is often difficult to assess the complete prevalence of cancer for a given registry area. An attempt is made to arrive at the complete prevalence including limited duration prevalence with respect of selected sites of cancer for India by fitting appropriate models to 1, 3 and 5 year cancer survival data available for selected registries of India. Cancer survival data, available for the registries of Bhopal, Chennai, Karunagappally, and Mumbai was pooled to generate survival for the selected cancer sites. With the available data on survival for 1, 3 and 5 years, a model was fitted and the survival curve was extended beyond 5 years (up to 30 years) for each of the selected sites. This helped in generation of survival proportions by single year and thereby survival of cancer cases. With the help of estimated survived cases available year wise and the incidence, the prevalence figures were arrived for selected cancer sites and for selected periods. In our previous paper, we have dealt with the cancer sites of breast, cervix, ovary, lung, stomach and mouth (Takiar and Jayant, 2013). The prevalence to incidence ratio (PI ratio) was calculated for 30 years duration for all the selected cancer sites using the model approach showing that from the knowledge of incidence and P/I ratio, the prevalence can be calculated. The validity of the approach was shown in our previous paper (Takiar and Jayant, 2013). The P/I ratios for the cancer sites of lip, tongue, oral cavity, hypopharynx, oesophagus, larynx, nhl, colon, prostate, lymphoid leukemia, myeloid leukemia were observed to be 10.26, 4.15, 5.89, 2.81, 1.87, 5.43, 5.48, 5.24, 4.61, 3.42 and 2.65, respectively. Cancer prevalence can be readily estimated with use of survival and incidence data.