4. Comparative Quantification of Mortality and Burden of Disease Attributable to Selected Risk Factors

Sources: Table 1 in Ezzati and others 2002, and individual risk factor chapters in Ezzati and others 2004 for data sources.

Note: BMI = body mass index, COPD = chronic obstructive pulmonary disease, IHD = ischemic heart disease, KJ = kilo joules, SD = standard deviation. New disease outcomes are used here when more recent epidemiological analyses enabled improvements over those of the CRA project (for example, multiple cancer sites assessed separately, versus grouped together, for smoking). Several risk factors (lead exposure, global climate change, and selected occupational risks) were included in the CRA project but not in the current analysis because of conceptual and empirical difficulties in converting estimates of exposure and/or hazards from the GBD analysis subregions to the World Bank regions.

a. Outcomes in italic are those that are likely to be causal but not quantified due to lack of sufficient evidence on the magnitude of hazardous effect.

b. The resulting hemoglobin levels vary across regions and age-sex groups because the other risks for anemia (for example, malaria) vary.

c. In estimating years of life lost because of disability (YLD) due to smoking in the CRA project (Ezzati and Lopez 2004), population attributable fractions (PAFs) for disease incidence were assumed to be the same as PAFs for mortality for cancers and COPD, and one-half of PAFs for mortality for all other diseases. In the current analysis, PAFs for disease incidence were assumed to be the same as those for mortality for cancers, COPD, and cardiovascular diseases—for which smoking is expected to increase mortality through increasing incidence—and zero for all other diseases. The estimated total disease burden from smoking is robust to assumptions about the effects of smoking on incidence of diseases other than major chronic diseases (cancers, COPD, and cardiovascular diseases).

d. The theoretical-minimum-risk exposure level for alcohol is zero, the global theoretical minimum. Specific population subgroups or diseases may have a non-zero theoretical-minimum-risk exposure (see figure 4.2) (Rehm and others 2004).

e. In the CRA project, PAFs for indoor smoke from household use of solid fuels were applied to COPD and lung cancer mortality and disease burden after subtracting the smoking-attributable burden (Smith, Mehta, and Maeusezahl-Feuz 2004). This overlooks multicausality of COPD and lung cancer, which is also illustrated empirically in the analysis of the hazards of smoking in China (Liu and others 1998). Therefore, we applied PAFs to total mortality from COPD and lung cancer.