Very low birth weight (VLBW) and bronchopulmonary dysplasia (BPD) are significant health issues which are also risk factors for oral health disease, including dental caries. In addition, VLBW/BPD is associated both with enamel defects, a biological risk factor for caries, and reduced cognitive development which may hinder the acquisition of healthy oral health behavior. Furthermore, both VLBW and BPD disproportionately affect minorities and those with low SES. The relative contributions or `mediating effects'of the various biologic, cognitive/behavior and psychosocial factors to the incidence and progression of oral disease in VLBW/BPD individuals is poorly understood. We propose to develop a new approach to mediation analysis and to analyze data from a unique cohort study to help illuminate the relationship between VLBW/BPD and dental caries. Our data are from a study which collected dental caries, enamel defects and oral health behavior information on a cohort of 302 adolescents previously assessed on medical, cognitive, and psychosocial outcomes at birth and at eight years of age. We propose a flexible new statistical method which translates and elaborates a proposed causal diagram into a system of easily fitted generalized linear models. This general context will allow us to incorporate specialized models, such as the zero-inflated Poisson, for the DMFT/S and other dental caries outcomes. We use a potential outcomes framework to develop mediation measures which may be expressed as functions of parameters from our general model. An additional aim of our project is to conduct simulation studies to examine the bias and efficiency of our mediation effect estimators, with a focus on robustness to model departures. Our development and study of methods for mediation analysis will facilitate their broad applicability for dental health and other studies. By helping to illuminate causal mechanisms in the relationship between VLBW/BPD and adolescent dental caries, we also hope to contribute to the development of effective new intervention programs for these special needs children.
This project seeks to learn about the reasons (whether they are biologic, oral health behavioral, or both) for the increased risk of dental caries among adolescents with very low birth weight (VLBW) and/or bronchopulmonary dysplasia (BPD). Our specific goals are to develop new statistical methods to handle complex causal models and to use these methods to analyze data from a cohort of adolescents followed from birth. By understanding the mechanisms leading to dental caries in VLBW/BPD children, we can more effectively intervene to improve the oral and general health of these special needs children.
| Albert, Jeffrey M; Wang, Wei; Nelson, Suchitra (2014) Estimating overall exposure effects for zero-inflated regression models with application to dental caries. Stat Methods Med Res 23:257-78 |
| Albert, Jeffrey M (2012) Distribution-free mediation analysis for nonlinear models with confounding. Epidemiology 23:879-88 |
| Wang, Wei; Albert, Jeffrey M (2012) Estimation of mediation effects for zero-inflated regression models. Stat Med 31:3118-32 |
| Albert, Jeffrey M; Nelson, Suchitra (2011) Generalized causal mediation analysis. Biometrics 67:1028-38 |
| Nelson, S; Albert, J M; Lombardi, G et al. (2010) Dental caries and enamel defects in very low birth weight adolescents. Caries Res 44:509-18 |
