Research: Funded Research

Enhancing Retrospective Life History Data in the Health and Retirement Study

Theory suggests that autobiographical memory (AM) involves two interconnected memory systems: episodic memories of events from specific remote times in an individual?s life (e.g., first day at school); and semantic memories of the important facts and themes that define an individual?s life history. In the absence of cognitive decline, older adults generally exhibit a shift from recalling the details of specific episodes to reporting more generalized semantic memories. There is considerable debate and inconsistent findings in the literature about the specific patterns of episodic and semantic AM disruption associated with progression to Alzheimer?s and related dementias. Whereas individuals with diagnosed CIND generally have preserved semantic AMs, findings regarding AD patients are mixed. In part, these mixed findings may be associated with study sampling differences, but they have also been associated with the specific features of the semi-structured interview measures of AM used. Using data from the Health and Retirement Study (HRS), we will examine subgroup differences based on cognitive status (normal vs CIND vs suspected dementia) in patterns of episodic and semantic AMs collected in a structured retrospective life history questionnaire (LHMS) developed in our R01 project. The LHMS embeds elements of existing, frequently used, semi-structured interview measures of AM (Kopelman et al, 2009; Levine et al., 2002) into a structured life grid and checkbox response format. The life-grid method and checkbox formats provide more support for retrieval from long-term semantic memory than semi-structured interview methods. Participants will be assigned to cognitive status groups (normal, CIND, and suspected dementia) based on validated cut-points for performance on standard cognitive tests included in HRS core waves (Langa et al., 2017). We will test hypotheses about cognitive status and age-related subgroup differences in reporting of AM semantic content, episodic details, and narrative themes collected in the spring 2017 LHMS sample (N = 3844). We expect older age groups with normal cognitive status and CIND to have more missing entries for episodic than semantic AMs. In addition, we expect older age groups with normal cognitive status and CIND will show a temporal gradient favoring early childhood AMs. Unless they have completed the LHMS with a proxy, participants with suspected AD/ADRD will likely have the most missing and inconsistent responses on the LHMS compared to other subgroups regardless of their age, the life course epoch of the question, and the AM type (episodic, semantic). We also expect a magnification of cognitive status differences in life grid compared to the checkbox questions. We will extend our analyses to examine if the HRS 2016 HCAP classification of cognitive status reveals different subgroup patterns of AM when data for these additional cognitive status measures are available. To our knowledge, there few population-based studies that examine cognitive status and AM. We anticipate that dissemination of our study findings will foster future innovative studies of cognitive decline and AM that leverage the extensive data available in HRS.


Health and Human Services, Department of-National Institutes of Health

Funding Period:

09/14/2018 to 12/31/2019