In the United States, many aspects of health system performance are the responsibility of local governments, including county governments. In order to be responsive to the health needs of their populations, county-level decision-makers need accurate information about local health trends, health system performance, and whether their local health systems are delivering necessary health interventions to achieve good health outcomes in their counties.
The US County Performance research team compiles available national and local health data from throughout the United States. We develop new methods for small area estimation and apply these methods to track county-level life expectancy, risk factor prevalence, and disease prevalence and control to determine how well counties are addressing the health of their residents.
Key Activities
- Track disparities in the effective delivery of health services
We are creating a cost-effective approach to tracking disease status, intervention delivery, and effective disease control at the county level using existing national and local data sources. This work is part of a National Institute on Aging-funded project to track chronic disease care, vaccinations for children and adults, cancer screening and treatment, maternal care, injury prevention, and response to acute coronary events at the county level in Washington state.
We use validated small area estimation methods with a geospatial component to estimate county-level prevalence, treatment, and control of disease. Where diagnosis bias is an issue, we cross walk between data sources to create improved estimates. For example, to track the fraction of hypertensive individuals taking blood pressure-lowering medication, we construct a predictive model relating clinical status to self-reported covariates in the National Health and Nutrition Examination Survey (NHANES) and apply the model to predict disease status for individuals with self-reported diagnosis and treatment information in the Behavioral Risk Factor Surveillance System. Corrected estimates of disease status and intervention delivery are entered into a geospatial regression model to calculate coverage of blood pressure-lowering medication and control of hypertensive disease.
- Measure health risks: individual total risk score
Preventable risk factors account for over 50% of premature deaths in the US. A health risk score that can predict the individual-level impact of these preventable risk factors on mortality would allow individuals and health professionals to set preventive goals and behaviors at the personal and clinical level. Furthermore, population health risk data are needed to inform health policy and priority setting. To meet these goals, we are developing a prototype risk calculator that computes a summary measure of risk for an individual based on 11 biometric and behavioral risk factors that contribute substantially to avoidable risk of death. This work is part of an ongoing collaboration with the Dartmouth Institute for Health Policy & Clinical Practice.
We calculate the 10-year avoidable risk of mortality for an individual based on disease-specific relative risks from systematic reviews and meta-analyses. We use data from NHANES from 2003 to 2008 to calculate current exposure to risk factors in the US population and compute the fraction of the mortality rate not due to the 11 risk factors. We are finding that elevated avoidable risk of mortality is clustered in a relatively small subset of the population, especially at younger ages. To provide targeted health interventions, local health risk data could accurately stratify individuals according to level of risk.
- Investigate county-level disparities in the US for key causes of death, risk factors, and quality indicators
The US will become the test bed for a combination of analyses that aim collectively to assess subnational variation in health system performance assessment. The research team will generate county-level estimates of mortality and select risk factors, causes of death, and quality indicators. For risk factors, the research team expects to create a continuous time series from 1990 forward. For causes of death, the research team will utilize all data from the ninth and 10th versions of the International Classification of Diseases (from roughly 1980 forward). Key intervention coverage indicators will be used as a measure of quality, using methods developed through the targeted analysis developed as part of the Tracking Disparities in the Effective Delivery of Health Services project.
The ultimate aim is to create a combined metric of health performance relevant to the data at hand, drawing upon methodological development undertaken in the Malaria Control Policy Assessment project and applied to child health interventions in Zambia. The results will also be presented and analyzed as separate indicators, creating combined maps of performance in different counties to more readily understand levels and trends over time. The results will be used to better understand disparities from county to county, to help inform policy debate, and to serve as a centerpiece for future research efforts related to the US.
Impact
The US County Performance research team facilitates data-driven decision-making about health system priorities by providing policymakers with improved information on disease risk, mortality, and health intervention delivery in their communities.
