Using HIV Surveillance Data for Public Health Evaluations and Interventions: Common Challenges & Proposed Methodological Solutions

Julia Hood | 2016

Advisor: Stephen E. Hawes

Research Area(s): Clinical Epidemiology, Epidemiologic Methods, Infectious Diseases, Public Health Practice, Social Determinants of Health


In the United States (U.S.), the purpose of HIV surveillance and related data systems is changing. Evolving programmatic priorities and complicated care needs of an aging population has led to greater demand for timely, accurate, and detailed data. My dissertation evaluates three prominent data systems: (1) National HIV Surveillance System (NHSS), (2) Medical Monitoring Project (MMP), and (3) CFAR Network of Integrated Clinical Systems (CNICS). Identification and management of data limitations were goals underlying each aim. My first aim evaluates the burden of diabetes, chronic kidney disease, and hypertension in MMP, a nationally representative sample, and CNICS, a clinical cohort. We encountered and addressed the following challenges: selection bias, missing data, non-standardized case definitions, and dissimilar patient populations. After using a standardized analytic approach, MMP and CNICS yielded similar sub-group specific prevalence estimates. Both data sources suggest considerable disease burden among older adults in HIV care. My second aim used NHSS and US census data to project the demographic composition of the U.S. population of people living with diagnosed HIV (PLWDH) through 2045. The model developed for this aim projects that the US PLWDH population will continue to grow in absolute size and will increasingly be comprised of racial/ethnic minorities; the number of PLWDH 55 years and older is projected to more than double between 2013 and 2045. My final aim used King County HIV surveillance data to explore the origins of NHSS data and how surveillance data is used for HIV control interventions. We discovered that the number of in-migrants with HIV is increasing concurrently with a decrease in the number of new diagnoses; that 12% of cases reported to CDC as newly diagnosed had evidence of a prior HIV diagnosis; and integration of patient care and HIV control activities improved key program metrics. In conclusion, existing data systems to monitor the U.S. PLWDH population have limitations, some of which can be addressed through statistical adjustment and some can only be resolved through adaptation of the data system’s design. As demands on HIV care programs are projected to grow, the programmatic utility of HIV surveillance systems should be enhanced.