Methodological issues in estimating smoking-attributable mortality in the United States.

Area of Science:

• Epidemiology
• Public Health
• Biostatistics

Background:

• Accurate estimation of smoking-attributable mortality is crucial for public health policy.
• Methodological variations in data sources can influence mortality estimates.
• Confounding factors may affect the precision of smoking-attributable mortality calculations.

Purpose of the Study:

• To compare smoking-attributable mortality estimates derived from two distinct US data sources.
• To assess the impact of residual confounding on disease-specific mortality estimates.
• To evaluate the reliability of different epidemiological methods for quantifying smoking's health burden.

Main Methods:

• Comparison of smoking-attributable mortality estimates from the Cancer Prevention Study II (CPS II) cohort with data from the National Mortality Follow-back Survey (NMFS) and National Health Interview Survey (NHIS).
• Application of a model-based approach to address potential residual confounding in disease-specific mortality estimates.
• Adjustment for confounding factors including education, alcohol intake, hypertension, and diabetes status.

Main Results:

• The CPS II yielded a 19% higher smoking-attributable mortality estimate for four major diseases compared to the NMFS/NHIS.
• Both data sources produced similar smoking-attributable mortality estimates for lung cancer.
• After accounting for age and other confounders, residual confounding had minimal impact on mortality estimates.

Conclusions:

• Methodological differences between large cohort studies and national surveys can lead to variations in smoking-attributable mortality estimates.
• The NMFS/NHIS combined data provide a reliable estimate for smoking-attributable mortality, particularly for specific diseases like lung cancer.
• Age adjustment and control for common confounding factors effectively minimize bias in smoking-attributable mortality estimations.