Fred Pampel University of Colorado, Boulder FORECASTING SEX DIFFERENCES IN MORTALITY IN HIGH INCOME NATIONS: THE CONTRIBUTION OF SMOKING PREVALENCE Fred Pampel University of Colorado, Boulder
Goals Use past trends in smoking prevalence of men and women to forecast future trends in smoking mortality sex differences Examine the contribution of smoking mortality of men and women to total mortality sex differences Link differences in cigarette diffusion to national patterns of change in sex differences in mortality
Smoking-Attributed Mortality Accounting for the Harm of Tobacco Trancends Cause of Death Categories Indirect Estimates (Peto et al. 1994) Debates Over Method Excess Lung Cancer Advantages in Forecasting Bias and Relative Rates
Data and Measures Data 21 High Income Nations 1975, 1985, 1990, 1995, 2000 Ages 0-34, 35-69, 70+ Measures Logged Male/Female Mortality Ratios Smoking Prevalence Cigarette Consumption Cigarette Diffusion
Counterbalancing Trends Means 1975 and 2000 (Table 1) Smoking Mortality Declining Female Advantage Other Mortality Rising Female Advantage All Causes Slight Narrowing
National Differences Logged Ratio Regressions (Table 2) Year and Cigarette Diffusion Stage Earlier the Start, Narrower the Gap Trend More Negative at Later Stage Aside on Gender Equality
Logged Mortality Ratios Smoking Mortality (Table 3) Logged Ratio of Smoking Prevalence Explanatory Power Future Trajectory (Figure 1) Non-Smoking Mortality (Figure 2) Linear Year Term Age Differences
Projecting Non-Smoking Ratio Past Trends Ages 35-69 (Figure 3) Linear Increase Alternative Future Trajectories Continued Increase Declining Rate of Increase No Change Decrease
Modeling Logged Total Ratio Determinants (Table 4) Predicted Logged Non-Smoking Ratio Predicted Logged Smoking Ratio Cigarette Consumption Lagged Logged Smoking Ratio x Cigarette Consumption Lagged Predicted Proportional Change
National Differences Country-Specific Change in Logged Total Mortality Ratio (Table 5) Range 35-69: -.432 (Greece) to .193 (New Zealand) Correlation with Cigarette Diffusion r = .590 – Early Stage Narrow Late Stage Widen Simple versus Smoking Forecasts (Figure 4)
Conclusion Past Narrowing: Due to Smoking Smoking-Attributed Mortality Differences: Narrowing Levels Off Total Mortality Differences Increasingly Depend on Non-Smoking Mortality Non-Smoking Mortality: No Evidence of Narrowing Yet National Differences Based on Cigarette Diffusion Crucial for Future Patterns