Noise exposure-response relationships established from repeated binary observations: Modeling approaches and...

Area of Science:

• Environmental Health
• Biostatistics
• Epidemiology

Background:

• Noise exposure-response relationships are vital for assessing noise impacts.
• Existing literature often overlooks the statistical methodologies used to derive these relationships.
• Different statistical approaches can lead to varying estimations of noise effects.

Purpose of the Study:

• To provide an overview of two key statistical approaches for establishing noise exposure-response relationships from repeated binary observations.
• To discuss the appropriate applications of subject-specific and population-averaged logistic regression.
• To illustrate the impact of statistical methodology on noise effect estimations using real-world data.

Main Methods:

• Comparison of subject-specific and population-averaged logistic regression models.
• Application of these models to data from three distinct noise effect studies.
• Analysis of the magnitude of differences in results obtained from the two statistical approaches.

Main Results:

• The choice of statistical approach (subject-specific vs. population-averaged logistic regression) can yield significantly different results in noise exposure-response relationships.
• The divergence in results depends on the specific dataset and the probability range of the binary outcome.
• Methodological choices influence the predicted strength of noise effects.

Conclusions:

• The statistical methodology employed is critical for accurately estimating noise exposure-response relationships.
• Selecting the appropriate statistical approach is essential for reliable population risk assessment and individual response estimation.
• Future research must carefully consider and report the statistical methods used to ensure the validity and comparability of noise effect studies.