Preface: Expert Advice to Enhance Aging Research and the Health and Retirement Study

Summaries

Response rates and their implications for data quality are the timely topics of the article by Dr. Joseph Sakshaug of the University of Mannheim. Response rates have been declining in the HRS as in most other cohort/panel studies, and increased levels of outreach have not been able to resurrect historically higher response rates. This article addresses challenges associated with sustaining population-based longitudinal datasets, and suggests innovative approaches to improving survey response. The article discusses selection bias in panel studies from the cumulative effects of participant nonresponse across waves, and from difficulties in linking participants to administrative databases. There is a focus on what to do to enhance response and consent rates ahead of data collection, and also on what to do to compensate for non-response after data have been collected. Strategies to increase rates of consent for high-value data, administrative records, and biological specimens are described. Linkage to administrative datasets can help to adjust for non-response bias, but the article notes ethical considerations when using administrative data for non-respondents who have not consented to linkage.

Contextual data to augment the reach of the HRS into exciting new areas of scientific interest is the topic of the article by Dr. Christopher Dick of Demographic Analytics Advisors. Linking contextual data to participants’ locations through techniques such as geocoding can advance research on the impact of the physical and social environment on health and aging, including climate change and sustainability. The article considers nationwide contextual and neighborhood-level datasets that can be readily merged with HRS data. It suggests a major expansion of the existing HRS contextual resource to cover the impacts of both natural and human-induced disasters on participants and the resiliency of their communities. As examples, linkage can be accomplished to measures of a community’s vulnerability to flooding, and a community’s capacity to respond to weather-based disasters, pandemics, or other disasters.

Cognitive data in the HRS intended to support NIA research priorities on dementia have been enhanced via the inclusion of the Harmonized Cognitive Assessment Protocol (HCAP) and current plans include collecting a second wave of HCAP data. HCAP is a sub-study of HRS panel respondents who receive a set of established cognitive and neuropsychological assessments and informant reports to better characterize cognitive function in older adults designed to be comparable with other U.S and international studies. Drs. Jacqueline M. Torres and Maria Glymour (University of California San Francisco) consider how HRS/HCAP can improve survey content to further facilitate the study of epidemiological effects, diagnostics issues, and comparisons across other international studies in the HCAP Network. The author’s recommendations included prioritizing recruitment of racial/ethnic minorities and longitudinal assessments, phasing-in an HCAP baseline assessment for people under age 65 and adding blood and AD biomarkers.

Data on drivers of mortality rates can facilitate research on mechanisms that explain rising rates of working-age mortality in the United States as described in a recent report by the National Academies of Sciences, Engineering, and Medicine (NASEM) funded by NIA and the Robert Wood Johnson Foundation.^[6] Drs. Shannon Monnat (Syracuse University) and Ira Elo (University of Pennsylvania) consider the strengths and limitations of HRS data for supporting such research and discuss recommendations to enhance the HRS to study this high priority research area. Their recommendations include enhancements to the sample and content. Specifically, reducing the age of eligibility for inclusion in the sample to capture more of the “working-age” population, oversampling of rural residents, contextual data on structural determinants of health and new measures that capture drug use, gun ownership, and social media use. They note that the contextual data collection has the benefits of enhancing research opportunities in this area without increasing respondent burden.

Data on the changing nature of work is needed to support studies on recent trends in the US workforce such as contingent work and variation in the nature of work by demographic and socioeconomic groups on life course health, aging and dementia. The COVID-19 pandemic also effected employment differentially across occupations and could have significant impacts on future work and retirement. Dr. Kathleen Mullen (RAND Corporation) considers survey content needed to better enable research on recent trends in the US workforce using the HRS. Her paper provides an overview of the information on working conditions that is currently available in the HRS or can be added by merging information from occupational databases and explores the potential for new content on work arrangements with employers and occupational demands from respondent jobs. Beyond potential connecting to data sources such as O*NET and the Occupational Requirements Survey, she recommends collecting data on work flexibility and work scheduling, the “gig” economy, and technological changes to jobs.