Correction factors for self-selection when evaluating screening programmes.

Area of Science:

• Epidemiology
• Biostatistics
• Public Health

Background:

• Screening programmes are susceptible to participant self-selection bias, commonly known as the healthy screenee bias.
• Existing evaluation methods like Intention-to-screen and per-protocol are also prone to self-selection bias.
• The post hoc approach, comparing participants versus non-participants after screening implementation, is particularly vulnerable.

Purpose of the Study:

• To provide an overview of approaches for quantifying and correcting self-selection bias in screening programmes.
• To demonstrate the interconvertibility of different bias correction methods.
• To derive correction factors applicable to various screening scenarios.

Main Methods:

• Consideration of four distinct methods for quantifying and correcting self-selection bias.
• Mathematical simplification revealing the identity of these correction methods.
• Derivation of correction factors for broader application, contingent on specific assumptions.

Main Results:

• The German Neuroblastoma Screening Study exemplifies the application, showing no significant reduction in mortality or stage 4 incidence due to screening.
• The most substantial bias, favoring screening, was identified when comparing participants against non-participants.
• Identified that different bias correction methods are mathematically equivalent.

Conclusions:

• Correcting for self-selection bias is essential, particularly for post hoc evaluation approaches.
• The post hoc method often lacks complete data, necessitating external data or additional assumptions for accurate bias estimation.
• Accurate evaluation of screening programmes requires addressing inherent self-selection biases.