Abstract
Posttraumatic stress disorder (PTSD) is associated with both cognitive deficits and an increased risk of dementia. Few studies, however, have examined the association between PTSD and cognitive decline in the context of parameters important to brain aging, including health conditions and genetics (e.g., APOE Ɛ4 status). National Alzheimer's Coordinating Center data were used to investigate the associations between lifetime PTSD status and working memory, immediate and delayed episodic memory, and executive functions over 7 years in 11,961 older adults with (n = 179) and without PTSD. Inverse probability weighting was used to mitigate confounding variables. Linear mixed-effects models were fit to weighted data. Sex, race, and APOE Ɛ4 status were examined as moderators. Lifetime PTSD was associated with an additional 0.031 standard deviations of decline in working memory annually, B = −0.031, 95% CI [−0.055, −0.007]. There was no significant PTSD x Time interaction for other cognitive domains. Sex moderated the associations between PTSD and working memory, B = 0.067, SE = 0.03, and delayed recall, B = 0.063, SE = 0.03, such that, among individuals with PTSD, men demonstrated faster decline than women. APOE Ɛ4 moderated the associations between PTSD and delayed recall, B = −0.106, SE = 0.03, and executive functions, B = 0.061, SE = 0.02; among individuals with PTSD, APOE Ɛ4 carriers showed faster and slower decline, respectively, than noncarriers. PTSD in older adults is associated with accelerated decline in working memory. Men and/or APOE Ɛ4 carriers may be important targets for early cognitive decline prevention.
| Original language | English |
|---|---|
| Pages (from-to) | 646-658 |
| Number of pages | 13 |
| Journal | Journal of Traumatic Stress |
| Volume | 38 |
| Issue number | 4 |
| DOIs | |
| State | Published - Aug 2025 |
Bibliographical note
Publisher Copyright:© 2025 International Society for Traumatic Stress Studies.
Funding
This work was funded by the National Institute on Aging (NIA) at the National Institutes of Health (NIH; Mentored Research Scientist Development Award K01AG070279, P30 AG072946. Deidentified data were extracted from the June NACC UDS 2023 data freeze (Beekly et al., 2007). The NACC UDS sample comprises individuals who volunteer for cognitive aging studies at National Institute on Aging–funded Alzheimer's disease research centers (ADRCs) across the United States and undergo approximately annual assessment of mental status, cognition, and health status during a series of clinical interviews. We excluded participants who did not reach 55 years of age during the study or who had a baseline diagnosis of dementia or a history of cerebrovascular accident, Down syndrome, Parkinson's disease, Huntington's disease, schizophrenia, or neurodevelopmental disorders. We additionally excluded participants who did not have an available APOE genotype (n = 5,161). Remaining participants included 179 (1.5%) with lifetime PTSD (baseline age range: 48–87 years) and 11,782 (98.5%) without PTSD (baseline age range: 48–102 years). All procedures for the current study were approved by the University of Kentucky Institutional Review Board (IRB). Participants provided written informed consent to share their UDS data under local IRB oversight at each ADRC. Deidentified data were extracted from the June NACC UDS 2023 data freeze (Beekly et al., 2007). The NACC UDS sample comprises individuals who volunteer for cognitive aging studies at National Institute on Aging–funded Alzheimer's disease research centers (ADRCs) across the United States and undergo approximately annual assessment of mental status, cognition, and health status during a series of clinical interviews. We excluded participants who did not reach 55 years of age during the study or who had a baseline diagnosis of dementia or a history of cerebrovascular accident, Down syndrome, Parkinson's disease, Huntington's disease, schizophrenia, or neurodevelopmental disorders. We additionally excluded participants who did not have an available APOE genotype (n = 5,161). Remaining participants included 179 (1.5%) with lifetime PTSD (baseline age range: 48–87 years) and 11,782 (98.5%) without PTSD (baseline age range: 48–102 years). All procedures for the current study were approved by the University of Kentucky Institutional Review Board (IRB). Participants provided written informed consent to share their UDS data under local IRB oversight at each ADRC. Psychiatric disorders, including alcohol and substance use, neurodevelopmental, anxiety, depressive, and schizophrenia and other psychotic disorders, as well as PTSD, were based on a combination of existing diagnoses (self-reported, informant-reported, and medical record review) and unstructured clinical interviews. Clinical interviewers consulted the Diagnostic and Statistical Manual of Mental Disorders (5th ed.; DSM-5; American Psychiatric Association [APA], 2013) as needed. A participant was classified as having lifetime PTSD if, based on the clinical interview, the clinician determined they had a current or past diagnosis of PTSD. Demographic characteristics were recorded at the first UDS assessment. Medical history and diagnoses were based on the clinical interviewer's judgment after administering the medical history interview. Syndromic cognitive diagnoses (i.e., normal cognition, MCI, impaired cognition but not MCI, or dementia) were determined by a single physician or a consensus panel; these diagnoses were determined based on a combination of cognitive testing, neurological examination, and clinical interview by a neurologist; informant report functional assessment; and medical history (Morris et al., 2006). APOE genotype data were downloaded from the NACC database. Neuropsychological testing at ADRCs was performed by certified psychometricians overseen by neuropsychologists. Cognitive test scores were first converted to z scores based on the mean performance of participants determined to have “normal” cognition, per syndromic cognitive diagnoses, for their age at their first clinical interview and were then incorporated into domain scores. To measure working memory, the present study used “digit span forward,” which Mirsky and Duncan's (2001) nosology of disorders of attention conceptualizes as measuring the “encode” element of attention and designates as “working memory.” Working memory and immediate and delayed episodic memory scores were collapsed across tests to accommodate changes in the test battery with the adoption of UDS-3 (Monsell et al., 2016). Therefore, working memory was measured using two variations of digit span (Snow et al., 1989; Wechsler, 1987; Weintraub et al., 2018). Episodic memory was measured using two tests of immediate and delayed story unit recall: the Wechsler Memory Scale–Revised Logical Memory I and II total recall scores (Wechsler, 1987) and Craft Story 21 immediate and delayed recall total recall scores (Craft et al., 1996). Executive functioning measures were unchanged across UDS versions and calculated using animal and vegetable naming for semantic fluency (Morris et al., 1989) and the trail making test (TMT; Reitan & Wolfsan, 1985) using the following formula: Executivefunctioning=animals,totalcorrect+vegetables,totalcorrect+timetocompletion,TMT,PartB−PartA.$$\begin{eqnarray*}{\mathrm{Executive functioning}} &=& \left[{\mathrm{animals}},{\mathrm{ total correct}}\right. \\ && \left. +\; {\mathrm{vegetables}},{\mathrm{ total correct}}\right. \\ && \left. +\; \left({\mathrm{time to completion}},{\mathrm{TMT}},\right.\right. \\ && \left.\left. {\mathrm{ Part B}}-{\mathrm{Part A}} \right) \right].\end{eqnarray*}$$ The Logical Memory I and II and digit span subtests have demonstrated adequate reliability and validity (Butters et al., 1988; Wechsler, 1987), the TMT B-A has been found to be a valid indicator of pure executive control (Sanchez-Cubillo et al., 2009), and semantic fluency tests incorporate executive function components (Aita et al., 2019). Test–retest reliability averages.80 for TMT Parts A and B (Wagner et al., 2011), and.70 for fluency (Harrison et al., 2000). The newly adapted NACC measures also demonstrated construct validity, with new tests shown to have good-to-very-good correlations with prior tests (Monsell et al., 2016). This retrospective cohort study analyzed data including up to the first seven consecutive study visits for each participant during 2005–2023. Data were truncated to seven visits to maintain sample size throughout follow-up and account for the lower average number of visits in the PTSD group. Missing covariate data were minimal overall except for history of TBI and anxiety disorders, which underwent significant changes in the UDS-3 protocol. Missing values for history of anxiety disorder or TBI were plausibly missing at random and were coded as “missing” so that participants missing those data could be retained in the analyses; inverse probability weighting can use partial data when appropriate missing data codes are applied. Inverse probability weighting was used to mitigate confounding by creating a balanced distribution of the included covariates and to limit the number of parameters to be estimated in the main analyses. The propensity of lifetime PTSD was estimated conditional on covariates that significantly differed by lifetime PTSD status: age, sex, race (binarized as White vs. non-White due to sparse cells), ethnicity, history of anxiety or depressive disorders, TBI, diabetes, number of medications, APOE Ɛ4 status, and length of follow-up. Due to their prognostic importance, we also included indicators for heart disease, hypertension, and clinical MCI diagnosis at study baseline. Alcohol and other substance use disorders, documented at baseline as 58 (0.5%) and 13 (0.1%), respectively, were not included in the inverse probability weighting due to low prevalence in the sample. The propensity score model fit in PROC LOGISTIC included linear and quadratic terms for all continuous variables, and the model was not further fitted or reduced; inverse probability weights were based on the probabilities from the propensity score model. Weighted linear mixed-effects models with random intercept terms were fit to the data in PROC MIXED to investigate whether the rate of change in cognitive scores differed by PTSD status. The only independent variables in the mixed models were lifetime PTSD status and time. Time was modeled as study time from the first assessment; age was included in the propensity score model. We used robust standard errors to account for weighting. We examined moderation (interaction) on an additive scale with the goal of assessing whether some subgroups of participants with PTSD may experience more rapid mean cognitive decline. We chose a limited set of variables known to be associated with cognitive performance. Sex, race, and APOE Ɛ4 carrier status were investigated as potential moderators of the association between PTSD and the rate of change in cognitive performance; each potential moderator was assessed individually and removed from the inverse probability weighting for that analysis. Hierarchically well-ordered models with all possible two- and three-way interactions among PTSD status, time, and the moderator were fit to the weighted data. Given that this is a novel analysis of this population and topic, our goal was hypothesis generation as opposed to the generation of definitive results. Therefore, we did not correct for multiple comparisons (Gelman et al., 2012). We performed all statistical tests using SAS (Version 9.4) with a two-sided alpha value set to.05. The authors would like to acknowledge that the National Alzheimer's Coordinating Center (NACC) database is funded by the NIA/NIH (U24 AG072122). NIA‐funded Aging and Disability Resource Centers contribute data to the NACC (P30 AG062429, P30 AG06646, P30 AG062421, P30 AG066509, P30 AG066514, P30 AG066530, P30 AG066507, P30 AG066444, P30 AG066518, P30 AG066512, P30 AG066462, P30 AG072979, P30 AG072972, P30 AG072976, P30 AG072975, P30 AG072978, P30 AG072977, P30 AG066519, P30 AG062677, P30 AG079280, P30 AG062422, P30 AG066511, P30 AG072946, P30 AG062715, P30 AG072973, P30 AG066506, P30 AG066508, P30 AG066515, P30 AG072947, P30 AG072931, P30 AG066546, P30 AG086401, P30 AG086404, P20 AG068082, P30 AG072958, P30 AG072959. This work was funded by the National Institute on Aging (NIA) at the National Institutes of Health (NIH; Mentored Research Scientist Development Award K01AG070279, P30 AG072946.
| Funders | Funder number |
|---|---|
| Ministerul Cercetării, Inovării şi Digitalizării | |
| National Alzheimer's Coordinating Center | |
| TMT Tapping Measuring SARL | |
| NIA | |
| National Institute on Aging | |
| National Institutes of Health (NIH) | 11,782, K01AG070279, U24 AG072122, P30 AG072946 |
ASJC Scopus subject areas
- Clinical Psychology
- Psychiatry and Mental health
