Estimation of odds ratio from group testing data with misclassified exposure

Area of Science:

• Biostatistics
• Epidemiology
• Public Health

Background:

• Estimating odds ratios for low prevalence diseases is challenging due to data limitations.
• Observational studies often suffer from misclassified exposure status and imperfect diagnostic tests.
• Group testing offers a potential cost-effective alternative to individual testing for disease surveillance.

Purpose of the Study:

• To develop and evaluate a group testing methodology for estimating odds ratios in low prevalence settings.
• To account for both misclassified exposure status and imperfect diagnostic test characteristics.
• To compare the performance of group testing against individual testing.

Main Methods:

• A statistical model was developed to incorporate imperfect sensitivity and specificity of diagnostic tests and exposure misclassification.
• Internal validation data, obtained via simple random sampling, was used for model identifiability.
• Pseudo-maximum likelihood estimation was employed for parameter estimation.

Main Results:

• The proposed group testing methodology demonstrated reliable estimation of odds ratios under various parametric configurations.
• Performance comparisons indicated the efficiency of group testing relative to individual testing.
• The methodology was illustrated using COVID-19 prevalence data.

Conclusions:

• The developed group testing approach provides a robust method for odds ratio estimation in low prevalence diseases, effectively handling data imperfections.
• This methodology offers a valuable tool for epidemiological studies, particularly when resources are limited or disease prevalence is low.
• The findings have implications for disease surveillance and risk factor analysis in public health.