There's Corn in My Wheat: Data Sources for Evaluating the HIV Continuum of Care and the Issue of Silos

Wednesday, June 25, 2014: 10:30 AM
109, Nashville Convention Center
Anne Rhodes , Virginia Department of Health, Richmond, VA
Kate Gilmore , Virginia Department of Health, Richmond, VA
Lauren Edwards Yerkes , Virginia Department of Health, Richmond, VA
Jennifer Kienzle , Virginia Department of Health, Richmond, VA
Jennifer Loney , Virginia Department of Health, Richmond, VA
Bryan Collins , Virginia Department of Health, Richmond, VA, Afghanistan
Neehar Gollapudi , Virginia Department of Health, Richmond, VA
Raid Mohaidat , Virginia Department of Health, Richmond, VA

BACKGROUND:   The National HIV/AIDS Strategy requires tracking HIV positive persons through the continuum of care, including diagnosis, linkage to medical care, retention in care, access to medications, and viral suppression. Determining the HIV care status of a person requires utilizing data systems that have historically been designed for separate, often disparate purposes.  The Virginia Department of Health received funding to improve outcomes across the continuum, but needed to determine how to assess a person’s current status on the continuum. 

METHODS:   Staff worked to integrate data from HIV surveillance, care, prevention, testing and other areas, at the person-level. Virginia is a medium morbidity HIV state, with over 24,000 persons living with HIV as of 12/31/2012.  Initial assessment of the care continuum numbers were done utilizing only HIV surveillance data and then re-assessed after the addition of data from the Ryan White program.  Issues were explored about how to match persons across datasets, along with how to resolve conflicting information regarding key demographic variables.  A Care Markers database was established, which utilizes exact and fuzzy matching routines and establishes a hierarchy of variable selection. 

RESULTS:   Initial HIV care continuum numbers for 2011 found a retention in care rate of approximately 40% and a viral suppression rate of 38%.  After adding in exact matches from the Ryan White care data, these numbers increased to 51% and 44%, respectively.  Over 10% of cases, however, were determined to be fuzzy matches, where one of the key variables including name, date of birth, gender, race and/or other key variables did not match.  Data managers from each database, including additional ones to be matched, met and designed a process for reviewing fuzzy and unmatched persons, which is currently being implemented.   

CONCLUSIONS:   The development of the Care Markers database established processes that are applicable in many areas of applied public health where data to evaluate a problem often reside in multiple systems.  The process also brought together data managers and analysts from different areas who now work together more easily, including designing a lab management system for HIV and sharing of code and resources across programs. Data sources were identified through a series of cross-unit meetings that allowed staff across program areas to work together.  The combination of data sources led to a more complete investigation of the continuum of care issue and more accurate data to characterize those living with HIV.