Bias from matching on age at death or censor in nested case-control studies

Area of Science:

• Epidemiology
• Occupational Health
• Biostatistics

Background:

• Nested case-control studies commonly use incidence-density sampling based on attained age.
• A new matching criterion, age at death, has been proposed for selecting controls.
• The impact of this additional criterion on exposure effect estimates is not fully understood.

Purpose of the Study:

• To evaluate potential bias introduced by adding age at death as a matching criterion in incidence-density sampling.
• To compare bias between matching on attained age alone versus attained age plus age at death.
• To investigate methods for handling workers with zero exposure due to latency assumptions (lagged-out workers).

Main Methods:

• Simulated occupational cohorts with time-dependent exposures were used.
• Incidence-density sampling was applied with matching on attained age alone or attained age plus age at death.
• Conditional logistic regression estimated exposure effects for unlagged and lagged cumulative exposures.
• Lagged-out workers were either excluded or included with zero exposure.

Main Results:

• Matching on attained age alone generally yielded unbiased effect estimates.
• Adding age at death as a matching criterion resulted in downwardly biased estimates.
• Including lagged-out workers with zero exposure showed similar or less bias compared to excluding them, especially for lagged exposures.

Conclusions:

• Incidence-density sampling with matching on attained age plus age at death introduces bias due to differences in control exposure history.
• Matching on attained age alone, with inclusion of lagged-out workers, provided unbiased estimates in simulations.
• Attained age matching is preferred for accurate exposure assessment in nested case-control studies.