PROSPER Award funds database acquisition to address healthy-user effect bias
Rene Soria-Saucedo, M.D., Ph.D., M.P.H., an assistant professor of pharmaceutical outcomes and policy, received a 2018 Excellence Award by the Program for Education and Research, or PROSPER, at the University of Florida College of Pharmacy. This award will allow the acquisition of the Truven Health Risk Assessment, or HRA, database — an integrated data source containing self-reported biometric and health-related behavioral data fully linkable to claims.
The proposal titled “Integrating claims and health risk assessment data to measure the relationship between adherence and utilization outcomes. Quantifying the “healthy user effect (HUE)” bias.” aims to measure more precisely the potential impact, magnitude, direction, and extent of the HUE bias as a confounder. “By linking claims with self-reported data, we will have the largest database of its kind ever assembled to assess this question,” Soria-Saucedo said.
HUE acknowledges the fundamental differences between people who engage in healthier activities —adhere to treatment regimens, exercise regularly, display good nutrition habits, consume lower amounts of alcohol and tobacco products, cope well with stress and avoid reckless behavior — from those who do not practice these activities.
“In observational research, this is a pervasive bias typically very hard to control for,” Soria-Saucedo said. A classic example of healthy user bias in observational studies was the erroneous finding that hormone replacement therapy reduces cardiovascular risk — a conclusion debunked later by evidence from clinical trials.
This work will be innovative in several aspects, Soria-Saucedo said. “By having a large pool of individuals, we should be able to generate very precise parameter estimates. No previous cohort study design has investigated HUE using self-reported information linked to long-term adherence rates extracted from administrative claims. After quantifying the HUE bias in different population subsets and settings, the results can be used to design probabilistic bias analysis or advance the pharmacoepidemiology methods field by offering a new pathway to account for HUE. Finally, the addition of HRA variables should expand the exploration of determinants of poor adherence to psychotropics and allow the development of comprehensive predictive models for non-adherence.”
Faculty members in the department of pharmaceutical outcomes and policy will also take advantage of the availability of this data source.
“One of the primary concerns in observational research is the residual confounding due to unmeasurable information,” said Joshua Brown, Pharm.D., Ph.D., M.S., an assistant professor of pharmaceutical outcomes and policy. “The HRA includes additional health behavior information that can be applied to these analyses and will be novel in the anticoagulation fields. In addition, the HRA includes variables that will expand the opportunities for collaboration and contract research with pharmaceutical companies who are often times interested in patient-reported outcomes.”
“We have secured 12 months of access to this fantastic data source,” Soria-Saucedo said. “We will use this time to assess the quality and robustness of the data, and evaluate whether it adds value to future and ongoing projects.”
Faculty in the UF College of Pharmacy interested in behavioral data should contact Soria-Saucedo to learn more about database access for future projects and collaborations.