Initiator Types and the Causal Question of the Prevalent New-User Design: A Simulation Study

Area of Science:

• Pharmacoepidemiology
• Health Services Research
• Biostatistics

Background:

• New-user designs are preferred for nonexperimental drug comparative safety and effectiveness studies, minimizing confounding by indication and healthy-adherer bias.
• The prevalent new-user design expands sample size by including patients switching from or previously using the comparator treatment.
• This design may reduce external validity, especially for newly approved treatments compared to standard ones.

Purpose of the Study:

• To describe three initiator types: new users, direct switchers, and delayed switchers.
• To articulate causal questions addressed by prevalent and new-user designs.
• To compare the causal inference capabilities of both designs.

Main Methods:

• Description of initiator types: new users, direct switchers, and delayed switchers.
• Articulation of causal questions for prevalent and new-user designs.
• Simulation study to assess bias from conditioning on time since comparator initiation.

Main Results:

• The prevalent new-user design can answer distinct causal questions compared to the new-user design.
• Conditioning solely on time since comparator initiation, without full treatment history, can lead to biased treatment effect estimates.
• Proper implementation of the prevalent new-user design yields valid causal estimates.

Conclusions:

• The prevalent new-user design, when correctly applied, offers a valuable approach for pharmacoepidemiologic research.
• It enables the estimation of novel causal effects distinct from traditional new-user designs.
• Careful consideration of patient initiation types and conditioning strategies is crucial for accurate inference.