Cause-deleted proportional mortality analysis and the healthy worker effect

Area of Science:

• Occupational Health
• Epidemiology
• Biostatistics

Background:

• Proportional mortality indices are frequently used in occupational health research when complete denominator data are unavailable.
• The validity of these indices, such as the proportional mortality ratio (PMR) and mortality odds ratio (MOR), as surrogates for the standardized mortality ratio (SMR) is often questioned.

Purpose of the Study:

• To evaluate the relationship between proportional mortality indices and the standardized mortality ratio (SMR).
• To determine the conditions under which the proportional mortality ratio (PMR) and mortality odds ratio (MOR) are valid estimators of the standardized mortality ratio (SMR).
• To assess the utility of these indices when certain causes of death are excluded from the analysis, particularly to mitigate bias from the 'healthy worker effect'.

Main Methods:

• The study mathematically examines the conditions required for the proportional mortality ratio (PMR) and mortality odds ratio (MOR) to accurately estimate the standardized mortality ratio (SMR).
• Analysis includes scenarios where specific causes of death are deleted from the dataset to adjust for potential confounding factors like the 'healthy worker effect'.

Main Results:

• Cause-deleted proportional mortality ratios (PMRs) and mortality odds ratios (MORs) can provide unbiased estimates of their corresponding standardized mortality ratios (SMRs).
• Excluding certain causes of death (e.g., heart disease) can correct for the inflationary effect of the 'healthy worker effect' on PMRs for other causes (e.g., cancer).

Conclusions:

• Cause-deleted proportional mortality indices (PMRs and MORs) are valuable tools for estimating standardized mortality ratios (SMRs) in occupational settings.
• These adjusted indices are particularly effective for evaluating specific occupational hazards and relatively infrequent causes of mortality.