Marginal regression models for clustered count data based on zero-inflated Conway-Maxwell-Poisson distribution with...

Area of Science:

• Public Health
• Biostatistics
• Epidemiology

Background:

• Community water fluoridation is a key public health strategy for preventing dental caries.
• Despite its benefits, public acceptance and understanding of fluoridation remain areas of discussion.
• The Iowa Fluoride Study (IFS) has longitudinally tracked fluoride exposure in children since 1991.

Purpose of the Study:

• Quantify fluoride exposure from various sources in children.
• Correlate long-term fluoride exposure with dental fluorosis and dental caries.
• Develop and validate novel statistical models for analyzing complex health data.

Main Methods:

• Utilized a subset of the Iowa Fluoride Study data.
• Applied a marginal regression model with a zero-inflated Conway-Maxwell-Poisson (ZICMP) distribution.
• Introduced and evaluated two new estimation methods for the ZICMP model through simulations.

Main Results:

• The novel ZICMP modeling approach effectively handles data with excess zeros and overdispersion.
• The methodology provides new insights into the impact of community water fluoridation on dental caries.
• The statistical methods were also successfully applied to a genomic dataset.

Conclusions:

• Advanced statistical modeling can illuminate the complex relationship between fluoride exposure and oral health outcomes.
• The developed ZICMP regression model offers a robust tool for analyzing count data with unique distributional properties.
• This research contributes to a better understanding of community water fluoridation's role in public dental health.