Continuous outcome logistic regression for analyzing body mass index distributions

Area of Science:

• Epidemiology
• Biostatistics
• Public Health

Background:

• Body mass index (BMI) is crucial for monitoring population weight status and health risks.
• Current logistic regression models often categorize BMI, leading to information loss and limiting study comparisons.
• Ad hoc BMI categorization hinders analysis of continuous BMI distributions and between-study comparability.

Purpose of the Study:

• To propose a continuous outcome logistic regression model for estimating continuous BMI distributions.
• To overcome limitations of ad hoc BMI categorization in regression analyses.
• To facilitate generalized post hoc extraction of parameters and simplify between-study comparisons.

Main Methods:

• Developed and evaluated a continuous outcome logistic regression model for BMI.
• Applied the model to estimate continuous BMI distributions.
• Demonstrated post hoc extraction of parameters like odds ratios for specific categories.

Main Results:

• The continuous logistic regression model effectively estimates continuous BMI distributions.
• This method avoids arbitrary BMI categorization, preserving information.
• Empirical evaluation using Swiss Health Survey data validated the approach.

Conclusions:

• Continuous logistic regression for BMI offers a more robust and flexible alternative to categorized models.
• The proposed method enhances the analysis of population health data and facilitates meta-analyses.
• This approach simplifies comparisons and pooling of data across diverse studies.