Zero inflated statistical count models for analysing the costs imposed by GERD and dyspepsia

Area of Science:

• Health Economics
• Biostatistics
• Gastroenterology

Background:

• Gastro-oesophageal reflux disease (GERD) and dyspepsia present significant public health and economic challenges.
• Economic data from these conditions often exhibit count characteristics with zero inflation, necessitating specialized statistical models.

Purpose of the Study:

• To evaluate the performance of various regression models for analyzing healthcare cost data associated with GERD and dyspepsia.
• To identify the most appropriate statistical model for accurately estimating the economic burden of these gastrointestinal disorders.

Main Methods:

• A cross-sectional household survey was conducted in Tehran, Iran, collecting data on gastrointestinal symptoms and associated costs.
• Poisson regression (PR), negative binomial regression (NB), zero-inflated Poisson (ZIP), and zero-inflated negative binomial (ZINB) models were applied.
• Model performance was compared using likelihood ratio tests, Vuong tests, AIC, and BIC.

Main Results:

• The zero-inflated negative binomial (ZINB) regression model demonstrated superior performance in analyzing the cost data compared to PR, NB, and ZIP models.
• Key predictors of healthcare costs included sex, age, smoking status, body mass index (BMI), insurance status, and education level.

Conclusions:

• The ZINB model is the most appropriate for analyzing zero-inflated count data in health economic studies of GERD and dyspepsia.
• Findings highlight the importance of considering unobserved heterogeneity and excess zeros when modeling healthcare expenditures for these conditions.