Response to ‘Cancer incidence rates among South Asians in four geographic regions: India, Singapore, UK and US’

I strongly disagree with the statement ‘The low rates in India compared with US whites and SA in UK and US may be due partially to under diagnosis but may also be due to lifestyle and environmental factors’.¹ According to reasons elucidated below, the low rates in India are due to gross under-representation of the vast majority of the population, underreporting of cases and many missed cases due to survival bias.

Almost all of the population-based registries from India contain data from mostly cancers reported from conurbations of the respective city. The reports produced by these registries cover a population of 48.5 million which amounts to <5% of the total population of the country.^2,3 On the other hand in Singapore and UK, the registries provide national population-wide figures, and in the United States the registries comprise a greater number of cities than in India. Hence the comparison between Indian registries with the other registries is misleading.

India currently does not have successful population-wide mass screening programs for cancer detection even in the cities mentioned in the study. Hence the new cases of cancer detected by registries under-represent the total number of cases, and may over-represent the less severe cases or cases from upper socioeconomic strata who are able to afford health care.

In India, it can be assumed that severe cases either die at home before detection (more so in rural areas) or die at hospital before diagnosis. The cases that are represented in the registries are only those who survived long enough to get detected and hence there is a potential for severe survival bias in the Indian registry data.

Furthermore, the lifestyle and environmental factors in urban areas are probably much worse or certainly no better than the other countries in the study. For example, the air pollution rates and smoking rates in Bangalore, Mumbai and Delhi are among the worst in the world. Other lifestyle habits in these cities are also unlikely to be much different from those in cities in other countries. Hence the suggestion that environmental factors play a role in the low incidence of cancer rates in India lacks merit and supporting data.

The authors’ report that the age-adjusted incidence rate of cervical cancer was 65.5 in Ambillikai, a rural area. This is probably due to high rates of illiteracy and ignorance among rural population regarding risk factors for cervical carcinoma and lack of simple screening tests in most rural areas. In India, 72% of the population lives in rural areas and the rates of many cancers, including carcinoma cervix, for this vast population remains either absent or limited. The representativeness of data collected by a few urban cancer registries for the entire country is a question that can only be answered if future research focuses on these rural areas.

The authors correctly argue that since it is an ecological study, the interpretations cannot be causal. But, the inferences are not valid at the level of groups either since the registries differ from each country in terms of geographical area covered, lack of homogeneity within the group and the process of reporting to the registry. For example, the UK registry contains data for all South Asians, the Singapore registry clumps Indians, Pakistanis and Sri Lankans together and the US registry contains only Indians and Pakistanis. I presume that when comparisons are made with Indians in India and these heterogeneous groups, the results cannot be internally valid for only Indians or even for South Asians across any groups. In summary, the authors do not appear to have sufficient information to tackle the question they posed.