Session: Abstract of Distinction: Lessons Learned from the Last Decade of Data Harmonization and Integrated Data Analysis in HIV and Addiction (Society for Prevention Research 26th Annual Meeting)

INTRODUCTION. New requirements for open access to research data from Federal agencies in the last several years has created a rising tide of change with respect to research data sharing. This is important for the future of scientific investigation because many areas of research have been limited by small sample size and lack of sample heterogeneity. The increasing public availability of research data can allow for new scientific discoveries when can be data combined and as comparability of measures (across time, studies, populations, and data collection sites) increases. As evidence of this changing resource base, the National Addiction and HIV Data Archive Program (NAHDAP) at the University of Michigan has been involved in building an infrastructure to support data harmonization and its archived data resources have led to a number of new analyses. As such, improving data integration and availability has enabled research and evaluation in the US across a range of hard to study prevention research problems, including substance abuse. The goal of this session is to present promising approaches that have emerged in the field of substance abuse that have allow for available data, from NAHDAP and elsewhere, to be combined to address new research questions.

One approach has been large scale efforts to architect methodological common ground across studies so that key data can be harmonized. Paper One describes lessons learned from harmonizing data from the 21 studies that are part of National Institute on Drug Abuse-funded Seek, Test, Treat and Retain cohort consortium studying substance use and focused on HIV treatment cascade outcomes. 11 common domains across the studies included: demographic characteristics, criminal justice involvement, HIV risk behaviors, HIV and/or hepatitis C infections, laboratory measures of CD4 cell count and HIV viral load, mental health, socioeconomic status, healthcare access, and substance use. STTR data were harmonized to allow for complex analyses of outcomes over diverse populations.

The field has also seen the development and implementation of statistical methods to facilitate integrated data analysis even when measures are across studies (or time points) are not comparable. It is possible to create a commensurate latent variable across multiple different measurement scales as long as there are items in common using moderated non-linear factor analysis (MNLFA). Paper Two uses simulated data to represent multiple samples arising from independent substance use studies which also differ from one another in measurement and in terms of relevant demographic characteristics. The simulated data are pooled, analyzed using MNLFA, and accuracy of the scoring is evaluated. Paper Three uses MNLFA with integrated data from publicly available national data to study nicotine dependence in Hispanic teens.

CONCLUSIONS. Together, these three papers suggest a forward path for secondary researchers; where publicly available data can be combined to address past data limitations leading to greater understanding of substance abuse within and across sociodemographic groups.