Estimating standardized mortality odds ratios with national mortality followback data

Area of Science:

• Epidemiology
• Biostatistics

Background:

• Proportionate mortality analyses are crucial for cause-specific mortality studies when population denominators are unavailable.
• Existing logistic regression methods for proportionate mortality ratio analysis have limitations, particularly when external standard rates are missing for certain strata.

Purpose of the Study:

• To present an extension of existing proportionate mortality ratio logistic regression methods.
• To describe methods for estimating standardized mortality odds ratios (SMORs) using numerator data from mortality followback studies.
• To address challenges in estimating SMORs when external standard rates are not available for all strata of interest.

Main Methods:

• Utilizes logistic regression on representative mortality followback data, which includes covariates collected from family members of decedents.
• Develops methods to generate fully standardized SMORs with numerator data.
• Explores the generation of SMORs that account for effect modification.
• Proposes using mortality followback data to create substitutes for external standard mortality rate ratios.

Main Results:

• Demonstrates the feasibility of estimating fully standardized SMORs using logistic regression with mortality followback data.
• Shows that SMORs can be generated to allow for effect modification.
• Highlights the flexibility of mortality followback data in overcoming limitations of external standard rates.

Conclusions:

• Mortality followback data, analyzed with logistic regression, provide a robust method for estimating standardized mortality odds ratios (SMORs).
• These extended methods enhance the analysis of cause-specific mortality, especially in situations lacking complete population denominators or external standard rates.
• The approach allows for more nuanced epidemiological investigations, including the assessment of effect modification.