Requiring continuous enrollment during follow-up? A call for robust research methods.

Area of Science:

• Health outcomes research
• Biostatistics
• Epidemiology

Background:

• Managed care strategies require evidence on treatment outcomes and costs.
• Secondary data sources like claims and EHRs are commonly used.
• Excluding participants lost to follow-up via continuous enrollment is standard practice.

Purpose of the Study:

• To illustrate immortal time bias and other time-related biases.
• To explain how continuous enrollment requirements contribute to bias.
• To provide guidance on evaluating and mitigating attrition bias.

Main Methods:

• Examination of reasons for participant attrition (death, insurance changes).
• Analysis of bias mechanisms introduced by attrition.
• Delineation of considerations for evaluating attrition impact.

Main Results:

• Continuous enrollment criteria can introduce immortal time bias.
• Attrition due to death, insurance termination, or switching can bias study estimates.
• Standard exclusion criteria may compromise study validity.

Conclusions:

• Researchers must address time-related biases stemming from attrition.
• Recommendations include reporting attrition tables and reevaluating study designs.
• Strategies for avoiding bias and redesigning vulnerable studies are discussed.