Re: Trends in cause specific mortality across occupations in Japanese men of working age during period of economic stagnation, 1980-2005: retrospective cohort study
Koga speculated that changes in smoking prevalence could be responsible for the reversal of trends in occupation-specific cancer mortality rates. We assessed trends in smoking prevalence by occupation using the Comprehensive Survey of Living Conditions on Health and Welfare, a nationally representative repeated cross-sectional survey in Japan. The age-adjusted smoking prevalence amongst managers and professionals has steadily decreased in parallel with other occupations between 1986 and 2007. Were there a cohort effect of higher smoking prevalence among managers that corresponded with the onset of the period of economic stagnation, the Japanese population would experience an increased mortality due to excess smoking which was not the case.
Koga is correct that risk reversal for cardiovascular mortality takes place rather quickly after smoking cessation, but this does not necessarily indicate that the rate of changes in smoking prevalence will match that in mortality. We agree with the author that disentangling the impact on health outcomes of socio-economic determinants and proximal risk factors such as smoking is a very important future research agenda to develop strategies to reduce the health impact of economic crises.
Ito and Oshima addressed very important issues in analyzing health outcomes by socioeconomic status. First, we assessed numerator-denominator bias in our study1, showing that the results remained the same under even the most extreme assumptions about the direction and magnitude of the numerator-denominator bias. Second, as opposed to the claim by the authors, linking death certificates and census data will not necessarily solve the problem of the numerator-denominator bias, partly because occupation is based on self-reported responses, and individual respondents give different occupations at different recording times.
Thus, we believe that sensitivity analysis would be necessary even if census and registration data were linked at the individual level. Nevertheless, we fully agree with Ito and Oshima that micro data that are individually linked with the death certificate registration database are an invaluable source of policy-relevant research. Although such data are not yet available in Japan, the Japanese Diet is discussing the introduction of a bill on the national identification number system, which is a big step toward facilitating the efficient use of available data sources2. We strongly believe that only with public understanding of the relevance of and support for such a system will linked micro data become available to promote evidence-based health policies.
1. Wada K, Kondo N, Gilmour S, Ichida Y, Fujino Y, Satoh T, et al. Trends in cause specific mortality across occupations in Japanese men of working age during period of economic stagnation, 1980-2005: retrospective cohort study. BMJ 2012;344:e1191.
2. Anonymous. Single tax number eyed for firms. Japan times, 10 Jan 2011.
Competing interests: No competing interests