Aims: The prediction of individuals' use of medical services and associated costs is crucial for medical systems. We modeled a risk assessment equation that included patient sociodemographic characteristics and health risk behaviors (obesity, smoking, and alcohol abuse) to strengthen the power of self-reported health status to predict healthcare resource use. We also sought to uncover gender-specific differences in the predictive value of the models. Methods: Before their first primary care visit, 509 new patients were interviewed. Data collected included sociodemographics, self-reported health status Medical Outcomes Study Short-Form (MOS SF-36), body mass index (BMI), and screening for alcoholism and smoking. Subsequent use of healthcare services for 1 year was determined by reviewing medical and billing records. Results: Generalized linear models and two-part regressions were estimated relating the five types of charges (plus total charges) to self-reported physical health status, controlling for gender, age, education, income, obesity, smoking, alcohol abuse, and mental health status. Lower physical health status was associated with higher charges for primary care (p=0.0022), specialty care (p=0.0141), diagnostic services (p<0.0001), hospitalizations (p=0.0069), and total charges (p<0.0001). For female patients, the regression equation predicted 14% of the variation in total medical charges compared with 28% for males. Female patients had higher charges for primary care (p=0.0019), diagnostic services (p=0.0005), and total charges (p=0.0180). Conclusions: Health status and patient gender were significant predictors of healthcare use and charges. The R2 of total charges was two times higher for men vs. women. This research has policy implications for healthcare organizations in predicting the usage patterns.
ASJC Scopus subject areas