Flexible piecewise linear model for investigating dose-response relationship in meta-analysis: Methodology, examples,...

Area of Science:

• Epidemiology
• Biostatistics
• Public Health

Background:

• Dose-response meta-analysis (DRMA) is crucial for understanding exposure-outcome relationships.
• Existing DRMA methods are limited for discrete exposures, particularly with non-linear trends.

Purpose of the Study:

• To introduce and validate a piecewise linear DRMA model for discrete exposures.
• To address limitations of linear models in capturing complex dose-response patterns.

Main Methods:

• Developed and illustrated one-stage and two-stage piecewise linear DRMA models.
• Employed slope equality tests to identify "piecewise effects" versus linear trends.
• Applied the model to sleep and parity data, comparing with linear and restricted cubic spline models.

Main Results:

• Both one-stage and two-stage piecewise linear DRMA models demonstrated good fit and similar results.
• Significant "piecewise effects" were detected, indicating non-linear relationships.
• The piecewise linear model showed improved fitting and reliable results compared to the simple linear model.

Conclusions:

• Piecewise linear DRMA is a valid and practical method for discrete exposures, especially when linear models are inadequate.
• It offers a superior alternative to linear models and a comparable alternative to non-linear models in DRMA.