Hostname: page-component-54dcc4c588-2ptsp Total loading time: 0 Render date: 2025-09-12T16:07:06.095Z Has data issue: false hasContentIssue false
  • English
  • Français

Cannabis use and cognition in older adults: Preliminary performance-based neuropsychological test results and directions for future research

Published online by Cambridge University Press:  01 September 2025

Kyler Mulhauser Open the ORCID record for Kyler Mulhauser [Opens in a new window] ,
Daniel Sullivan ,
Jessica L. Bair ,
Anthony N. Correro II ,
Subhamoy Pal ,
Jonathan Reader ,
Benjamin M. Hampstead  and
Bruno Giordani
Kyler Mulhauser*
Affiliation:
Department of Psychiatry, University of Michigan, Ann Arbor, MI, USA Michigan Alzheimer’s Disease Center, University of Michigan, Ann Arbor, MI, USA
Daniel Sullivan
Affiliation:
Department of Psychiatry, University of Michigan, Ann Arbor, MI, USA
Jessica L. Bair
Affiliation:
Mental Health Service, Iowa City VA Healthcare System, Iowa City, Iowa, USA
Anthony N. Correro II
Affiliation:
Department of Neurology, Division of Neuropsychology, Medical College of Wisconsin, Milwaukee, WI, USA
Subhamoy Pal
Affiliation:
Michigan Alzheimer’s Disease Center, University of Michigan, Ann Arbor, MI, USA Department of Neurology, University of Michigan, Ann Arbor, MI, USA
Jonathan Reader
Affiliation:
Michigan Alzheimer’s Disease Center, University of Michigan, Ann Arbor, MI, USA Department of Neurology, University of Michigan, Ann Arbor, MI, USA
Benjamin M. Hampstead
Affiliation:
Department of Psychiatry, University of Michigan, Ann Arbor, MI, USA Michigan Alzheimer’s Disease Center, University of Michigan, Ann Arbor, MI, USA Department of Neurology, University of Michigan, Ann Arbor, MI, USA Department of Veterans Affairs, Veterans Affairs Ann Arbor Healthcare System, Ann Arbor, MI, USA
Bruno Giordani
Affiliation:
Department of Psychiatry, University of Michigan, Ann Arbor, MI, USA Michigan Alzheimer’s Disease Center, University of Michigan, Ann Arbor, MI, USA Department of Neurology, University of Michigan, Ann Arbor, MI, USA
*
Corresponding author: Kyler Mulhauser; Email: mulhause@med.umich.edu
Rights & Permissions [Opens in a new window]

Abstract

Objective:

We evaluated performance-based differences in neuropsychological functioning in older adults (age 65+) across the dementia continuum (cognitively intact, mild cognitive impairment, and dementia) according to recent cannabis use (past six months).

Method:

A sample of 540 older adults from a well-characterized observational cohort was included for analysis. Participants completed a standardized questionnaire assessing cannabis use in the six months prior to the study visit and completed a comprehensive neuropsychological assessment. We used traditional cross-sectional analyses (multivariate, univariate) alongside causal inference techniques (propensity score matching [PSM]) to evaluate group differences according to recent cannabis use status. We also examined whether cannabis-related problem severity, a risk factor for cannabis use disorder (CUD), was associated with cognitive outcomes among those reporting recent cannabis use.

Results:

Approximately 11% of participants reported using cannabis in the prior six months, with the median user consuming cannabis two to four times per month. Participants with recent cannabis use performed similarly across all five domains of neuropsychological functioning compared to those with no cannabis use. Among older adults reporting recent cannabis use, those with elevated risk for CUD demonstrated lower memory performance.

Conclusions:

These preliminary results are broadly consistent with other findings indicating that low-frequency cannabis use among older adults, including those along the dementia continuum, is generally well tolerated from a cognitive perspective. However, among older adults who used cannabis, elevated symptoms of CUD may negatively impact memory performance. Future research should explore how variations in cannabis use patterns, individual characteristics, and clinical phenotypes influence cognitive outcomes.

Keywords

Older adultscognitioncannabisneuropsychological assessmentcannabis use frequencyaging and memory

Information

Type
Research Article
Information
Journal of the International Neuropsychological Society , First View , pp. 1 - 8
Creative Commons
Creative Common License - CCCreative Common License - BY
This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution and reproduction, provided the original article is properly cited.
Copyright
© The Author(s), 2025. Published by Cambridge University Press on behalf of International Neuropsychological Society

Statement of research significance

Research Question(s) or Topic(s):

This study investigated whether older adults who use cannabis perform differently on neuropsychological tests compared to those who do not. Main Findings: Older adults who reported using cannabis in the past six months showed no differences in cognitive test performance compared to non-users across five cognitive domains. Among older adults reporting recent cannabis use, greater severity of cannabis-related problems was associated with lower memory performance. Study Contributions: This study expands the research base on cannabis use and cognition in older adults. Specifically, this study used a comprehensive battery of neuropsychological tests to evaluate for cognitive differences in older adults according to recent cannabis use status and risk factors associated with cannabis use disorder. Consistent with other research, these preliminary findings suggest that low-frequency cannabis use is generally well tolerated, although hazardous/problematic cannabis use behaviors were associated with lower memory performance.

Introduction

Older adults (age 65 and older) are using cannabis at unprecedented rates, representing the fastest growing age segment among both new cannabis users and daily cannabis consumers (Han et al. Reference Han, Sherman, Mauro, Martins, Rotenberg and Palamar2017; Han & Palamar, Reference Han and Palamar2018, Reference Han and Palamar2020; Khoury et al. Reference Khoury, Maliha and Ibrahim2022; Statistics Canada, 2019; Statistics Canada & Rotermann, Reference Rotermann2020). Approximately 2 – 7% of older adults in North America report using cannabis in the past month, with 2019 estimates revealing that 27% of older adults using cannabis were first-time users (Jeffers et al. Reference Jeffers, Glantz, Byers and Keyhani2021; Maxwell et al. Reference Maxwell, Jesdale and Lapane2021; Statistics Canada, 2019). Older adults who consume cannabis now use at rates comparable to that of younger-aged peers, with approximately 60% reporting frequent or daily use (Jeffers et al. Reference Jeffers, Glantz, Byers and Keyhani2021). Among older adults who use cannabis, 75% do so for medical purposes – primarily symptom management for pain, sleep, and mood – although the reported therapeutic benefits and harms of cannabis use are mixed (Briscoe & Casarett, Reference Briscoe and Casarett2018; Minerbi et al. Reference Minerbi, Häuser and Fitzcharles2019; Tumati et al. Reference Tumati, Lanctôt, Wang, Li, Davis and Herrmann2022; Wolfe et al. Reference Wolfe, Corace, Butler, Rice, Skidmore, Patel, Thayaparan, Michaud, Hamel, Smith, Garber, Porath, Conn, Willows, Abramovici, Thavorn, Kanji and Hutton2023).

The acute cognitive effects of cannabis use are well established, with impairments predominantly in the areas of attention, executive functioning, learning, and memory (Crean et al. Reference Crean, Crane and Mason2011; Volkow et al. Reference Volkow, Swanson, Evins, DeLisi, Meier, Gonzalez, Bloomfield, Curran and Baler2016; Zhornitsky et al. Reference Zhornitsky, Pelletier, Assaf, Giroux, Li and Potvin2021). Residual effects of heavy cannabis use on cognition typically resolve within three weeks with sustained abstinence (Crean et al. Reference Crean, Crane and Mason2011; Meier et al. Reference Meier, Caspi, Danese, Fisher, Houts, Arseneault and Moffitt2018; Ross et al. Reference Ross, Ellingson, Rhee, Hewitt, Corley, Lessem and Friedman2020). However, some evidence supports persisting negative impacts on cognition for long-term, heavy users who initiate use during the neurodevelopmental period (age 25 and younger) and those with significant symptoms of cannabis use disorder (CUD) (Crean et al. Reference Crean, Crane and Mason2011; Kroon et al. Reference Kroon, Kuhns, Hoch and Cousijn2020; Lovell et al. Reference Lovell, Akhurst, Padgett, Garry and Matthews2020). Notably, the vast majority of cannabis research focuses on adolescent and young adults, leaving a significant gap in understanding its cognitive and neurobiological effects in older adults.

Emerging evidence suggests that the effects of cannabis on the aging brain may differ meaningfully from those observed in younger populations. Reviews focused on the cognitive effects of cannabis use in older adults, including those with cognitive decline and dementia, have concluded that low-dose, short-term medical cannabis use is generally well tolerated in older adults and does not confer significant risk for adverse cognitive outcomes (Scott et al. Reference Scott, Brennan and Benitez2019; Vacaflor et al. Reference Vacaflor, Beauchet, Jarvis, Schiavetto and Rej2020). Reviews incorporating both human and animal studies have identified potential neuroprotective properties of cannabis on cognition. For example, Pocuca et al. (Reference Pocuca, Walter, Minassian, Young, Geyer and Perry2021) noted methodological limitations impacting the largely null effects of cannabis use on older adult cognition in human studies, while observing that better controlled animal models indicate that very low doses of THC improved cognition in very old rodents while slightly higher chronic doses of THC improved cognition in moderately aged rodents. Similarly, Weinstein and Sznitman (Reference Weinstein and Sznitman2020) also note important brain-age differences and emphasized the potential for cannabis to confer neuroprotective effects in older adults specifically. Relevant to older adult cognition across the dementia spectrum, some cannabinoids, including THC and CBD, have been associated with reduced neuroinflammation, enhanced cholinergic transmission, and reduced beta-amyloid aggregation, with exploratory application in human participants with Alzheimer’s disease (Abate et al. Reference Abate, Uberti and Tambaro2021; Aso & Ferrer, Reference Aso and Ferrer2014; Bahji et al. Reference Bahji, Meyyappan and Hawken2020; Weinstein & Sznitman, Reference Weinstein and Sznitman2020). These findings suggest differential effects of cannabis on cognition and brain health that vary by dose and age. However, rigorous controlled studies in older human populations remain limited.

Among older adults who reported any use of cannabis in the past year, subjective memory concerns (SMC) were endorsed at higher rates relative to non-users, with recent estimates ranging from 13.0% to 17.5%, compared to a significantly lower rate of 6.4% among non-users (Mulhauser et al. Reference Mulhauser, Hampstead, Coughlin and Ilgen2023; Vacaflor et al. Reference Vacaflor, Beauchet, Jarvis, Schiavetto and Rej2020). Among a representative sample of US adults aged 50 years and older, 8.2% reported using cannabis in the past year, and use was associated with a two-fold increase in reporting SMC; however, this relationship was attenuated after accounting for demographic, health, and psychiatric factors, and the relationship was nonsignificant when evaluating past-month use of cannabis (Mulhauser et al. Reference Mulhauser, Hampstead, Coughlin and Ilgen2023). Even so, the literature on older adults who use cannabis remains comparatively small and is limited by several factors, including a lack of representation specific to older adults, limited information on cannabis use patterns among older adults, and minimal or no performance-based measures of neuropsychological functioning (Mulhauser et al. Reference Mulhauser, Hampstead, Coughlin and Ilgen2023; Scott et al. Reference Scott, Brennan and Benitez2019; Volkow et al. Reference Volkow, Swanson, Evins, DeLisi, Meier, Gonzalez, Bloomfield, Curran and Baler2016; Weinstein & Sznitman, Reference Weinstein and Sznitman2020).

We sought to address these gaps in the literature by evaluating cognitive differences in older adults with and without recent cannabis use based on comprehensive neuropsychological test performance. Based on the literature indicating that low-dose cannabis use in older adults is generally well tolerated from a cognitive perspective, we hypothesized that older adults with recent cannabis use would demonstrate comparable neuropsychological test performance to non-users after accounting for relevant covariates. Furthermore, we explored how responses on the Cannabis Use Disorders Identification Test–Revised (CUDIT-R), a frequently used measure of cannabis-related problem severity and indicator of CUD, may be associated with cognitive outcomes.

Method

Participants and procedure

We extracted data from 588 participants spanning the dementia continuum from the University of Michigan Memory and Aging Project (UM-MAP) databank. UM-MAP is the primary longitudinal observational cohort of the Michigan Alzheimer’s Disease Center (MADC), which contributes common data elements to the National Alzheimer’s Coordinating Center (NACC) Uniform Data Set (UDS) (NACC, 2024). The first study visit in which participants reported their cannabis use history were included. UM-MAP participants were recruited from communities in southeastern Michigan, the neuropsychology and neurology clinics at the University of Michigan, the Healthier Black Elders Center through the Institute of Gerontology at Wayne State University, and the University of Michigan Health Research website. Inclusion criteria for participation in UM-MAP were aged 55 years or older, a research diagnosis consistent with NACC-defined criteria (see below), and accompaniment by a knowledgeable informant (e.g., spouse, partner, or child).

Prior to starting the evaluation, participants or their legally authorized representative provided informed consent. All human data included in this manuscript were obtained with approval from the University of Michigan Institutional Review Board (IRB). The study adhered to the ethical principles outlined in the Declaration of Helsinki. Study visits occur annually and consist of a standardized clinical evaluation comprised of demographic data, history and physical examination (including neurological examination), dementia staging, behavioral and functional assessments (including self- and informant-report questionnaires), and neuropsychological testing. Only data from the first time participants completed the cannabis questionnaire were included for the current analyses.

Measures

UDS neuropsychological battery

The rationale, development, validation, harmonization, administration and scoring procedures, and cognitive status staging of the UDS neuropsychological battery are described in detail elsewhere (Dodge et al. Reference Dodge, Goldstein, Wakim, Gefen, Teylan, Chan, Kukull, Barnes, Giordani, Hughes, Kramer, Loewenstein, Marson, Mungas, Mattek, Sachs, Salmon, Willis‐Parker and Welsh‐Bohmer2020; Kiselica et al. Reference Kiselica, Webber and Benge2020; Staffaroni et al. Reference Staffaroni, Asken, Casaletto, Fonseca, You, Rosen, Boxer, Elahi, Kornak, Mungas and Kramer2021). The UDS neuropsychological battery is derived from neuropsychological instruments and paradigms commonly used in clinical practice. The current version (UDSv3) consists of the following tests: Montreal Cognitive Assessment (MoCA), Craft Story 21 Recall, Benson Complex Figure, Number Span Test, Category Fluency, Phonemic Fluency, Trail Making Test A & B, and Multilingual Naming Test (MINT).

Prior factor analysis of the UDSv3 neuropsychological battery (Kiselica et al. Reference Kiselica, Webber and Benge2020) identified a higher-order factor comprised of a supraordinate general cognitive factor and five lower-order factors: speed/executive (Trail Making Test A & B, Phonemic Fluency for both “F” and “L”), visual (Benson Complex Figure Copy and Recall trials), attention (Number Span Test forward and backward), language (Category Fluency for both animals and vegetables, MINT), and memory (Craft Story 21 Recall Immediate and Delayed). For the current study, raw scores for these measures were standardized into z-scores and then averaged to create composite z-scores for the five lower-order neuropsychological domains identified by Kiselica et al. (Reference Kiselica, Webber and Benge2020). As is common in clinical practice, the raw scores (seconds) for Trail Making Test parts A and B were truncated at 150 and 300 s, respectively. These two z-scores were also reversed to ensure that higher z-scores reflected better performance across all tests and composites.

Questionnaires

Cannabis use in the prior six months

Participants were asked, “Have you used any cannabis over the past six months?” Responses included “Yes” and “No.”

Cannabis Use Disorder Identification Test–Revised (CUDIT-R)

Participants who responded “Yes” to using cannabis in the past six months were then administered the CUDIT-R (Adamson et al. Reference Adamson, Kay-Lambkin, Baker, Lewin, Thornton, Kelly and Sellman2010), an 8-item self-report inventory that assesses behaviors and symptoms concerning for CUD over the past six months. Each item is scored on a 5-point scale ranging from 0 to 4, with higher scores reflecting greater frequency or severity of cannabis-related problems. Total scores range from 0 to 32; scores ranging from 8 to 11 indicate hazardous cannabis use, scores ≥ 12 indicate possible cannabis use disorder.

Geriatric Depression Scale–Short Form (GDS-SF)

The GDS-SF is a 15-item self-report inventory of depression symptoms commonly observed in older adults. Scores on the GDS-SF range from 0 to 15. Scores ≥ 5 indicate elevated levels of depression, while scores ≥ 9 warrant clinical attention and safety evaluation by a licensed mental health provider.

Covariates

Demographic information (i.e., age, sex, race, years of education) and clinical characteristics (i.e., depression, general cognitive functioning) were obtained from questionnaires and clinical interviews in accordance with NACC administrative procedures (see UDS references above). Alcohol and tobacco use were considered as potential covariates but ultimately excluded due to temporal and measurement inconsistencies. Per NACC protocol, these variables were recorded only at the initial study visit and carried forward across subsequent visits, which resulted in misalignment with the cannabis and cognitive data used in this study. Differences in measurement timeframes and formats (e.g., lifetime tobacco use, 3-month alcohol use, and 6-month cannabis use) further limited their appropriateness for inclusion in adjusted models.

Data analytic plan

We screened the initial dataset to exclude participants who did not meet inclusion criteria. Participants with missing data on cognitive tests were excluded, as this precluded calculation of a domain composite score. There was no significant difference in the frequency distribution of cannabis use (users vs. non-users) by cognitive test data missingness (missing data vs. complete data), χ2(1, N = 588) = 0.29, p = .59.

We used multivariate, univariate, and propensity score matching (PSM) analyses to evaluate the associations between cannabis use and cognition in older adults. First, we used multivariate analysis of covariance (MANCOVA) to determine whether differences existed in cognitive performance by cannabis use (users vs. non-users). Age (years), sex (male vs. female), race (White, Black, Asian, Other), education (years), depression (GDS-SF total score), and global cognitive functioning (MoCA total score) were entered as covariates. The MoCA was included to account for diagnostic heterogeneity across participants, as our sample included individuals with normal cognition, MCI, and dementia. Although cannabis groups did not differ significantly in the distribution of diagnostic classifications, there was considerable variability in global cognitive functioning. Including the MoCA helped reduce confounding due to overall cognitive status while avoiding criterion contamination that would arise from using diagnosis – partially defined by the cognitive outcomes themselves – as a covariate. Assumptions of multivariate normality and homogeneity of variance–covariance matrices were tested using Mardia’s test of kurtosis (α = 0.05) and Box’s M (α = 0.05), respectively. Multivariate outliers were identified using Mahalanobis distance (α = 0.05). A significant omnibus MANCOVA was followed by five multivariable linear regressions for each cognitive domain. All significance tests were two-tailed with α = 0.05.

To better balance the groups and control for potential confounding variables, we employed PSM using a 2:1 nearest-neighbor matching strategy (with replacement) between non-users and users of cannabis as a sensitivity analysis (Austin, Reference Austin2010). Specifically, cannabis users and non-users were matched based on the propensity scores (i.e., probability of cannabis use) based age (years), sex (male vs. female), race (White vs. Black), education (years), depression (GDS-SF total score), and global cognitive functioning (MoCA total score). We excluded participants who identified as Asian or Other race due to small cell sizes that precluded matching. After matching, we estimated the Average Treatment Effect on the Treated (ATT) to compare cognitive performance between the matched groups across the same five cognitive domains.

Finally, on an exploratory basis, we examined whether cannabis-related problem severity was associated with cognitive performance among participants who reported recent cannabis use. Participants were categorized using validated CUDIT-R total score cutoffs as described above. We conducted analyses of covariance (ANCOVAs) across the five UDSv3 cognitive domains, retaining the same covariates from prior models. In a secondary exploratory analysis, we assessed bivariate correlations between individual CUDIT-R items and each cognitive domain score to identify potential item-level associations.

Results

Descriptive statistics

Descriptive statistics for the total sample and by cannabis use appear in Table 1. Ages ranged from 55 to 93 years (M = 72.52, SD = 6.45). Biological sex was distributed as 195 (36.11%) male and 345 (63.39%) female. Most participants identified as White (n = 366, 67.78%), with approximately one-third identifying as Black or African American (n = 166, 30.74%). Years of education ranged from 8 to 21 and was truncated at 20 years (M = 16.33, SD = 2.37). Most participants were right-handed (n = 462, 89.71%). Any cannabis use in the prior 6 months was self-reported as either “yes” (n = 60, 11.11%) or “no” (n = 480, 88.93%). MoCA scores ranged from 11 to 30 (M = 25.23, SD = 3.53). Using χ2 tests of independence, cannabis use was not significantly associated with sex, race, handedness, or diagnostic classification (p’s > .05). When using independent-samples t-tests or the Mann-Whitney U test, years of education, number of medical conditions, and MoCA scores were not significantly different between users and non-users of cannabis (p’s > .05). Relative to non-users of cannabis (M = 72.81, SD = 6.59), the cannabis users were younger (M = 70.22, SD = 4.61; p ≤ .001). Cannabis users scored higher (M = 2.07, SD = 2.25) than non-users (M = 1.32, SD = 1.82) on a screening measure of depression (GDS-SF; p = .02).

Table 1. Demographic and clinical characteristics (N = 540)

Note. GDS-SF, Geriatric Depression Scale-Short Form; MCI, Mild Cognitive Impairment. MoCA, Montreal Cognitive Assessment. Age, education, and number of medical conditions are presented as means and standard deviations. Differences in the means or frequencies by cannabis use (users vs. non-users) are reported as Welch’s t-tests and Chi-squared (χ2) tests of independence, respectively.

a The Mann-Whitney U Test was utilized to determine whether differences existed in the number of medical conditions by cannabis use status (n = 516).

Inferential statistics

Group differences by recent use of cannabis

Initial Model

The initial omnibus MANCOVA was statistically significant, Wilks’ Λ = 0.31, F(45, 2356) = 15.50, p ≤ .001. The multivariate R 2 was .69. Box’s M supported homogeneity of variance–covariance matrices (p = .19). However, Mardia’s test of kurtosis (p ≤ .001) for the residuals suggested departure from a multivariate normal distribution. Examination of Mahalanobis distance revealed 18 outliers that exceeded the χ2 critical value (12.83) for df = 5 dependent variables and α = .05. Given the impact of these violations on MANCOVA, these 18 participants (1 cannabis user, 17 non-users) were excluded from this analysis, yielding an analytic sample of 522.

Final Model

Table 2 depicts the MANCOVA results with N = 522. The omnibus MANCOVA was again statistically significant, Wilks’ Λ = 0.32, F(45, 2275) = 14.83, p ≤ .001. The multivariate R 2 was .68. At the multivariate level, age (p ≤ .001; multivariate R 2 = .05), MoCA total score (p ≤ .001; multivariate R 2 = .54), sex (p = .01; multivariate R 2 = .03), and race (p ≤ .001; multivariate R 2 = .12) were significant covariates. Cannabis use was non-significant (p = .30; multivariate R 2 = .01). Mardia’s test for multivariate kurtosis (p = .92) and Box’s M (p = .87) supported multivariate normality and homogeneity of variance–covariance matrices, respectively .

Table 2. Multivariate analysis of covariance (MANCOVA) of five neuropsychological domains by cannabis use and relevant covariates (N = 522)

Note. GDS-SF, Geriatric Depression Scale-Short Form. MoCA, Montreal Cognitive Assessment. Sex was coded as 1 = male, 2 = female. Race was coded as 1 = White, 2 = Black, 5 = Asian, and 50 = Other. The overall multivariate analysis of covariance (MANCOVA) was statistically significant, Wilks’ Λ = 0.32, F(45, 2275) = 14.83, p ≤ .001, Multivariate R 2 = .68. The dependent variables included five neuropsychological domain composite scores: speed/executive functioning, visual, attention, language, and memory.

Given the significant omnibus MANCOVA, five univariate multiple regression models followed (Table 3). Omnibus tests for these five regressions indicated statistically significant models (p ≤ .001) for all five cognitive domains: speed/executive, visual, attention, language, and memory. Regarding covariates, overall cognitive functioning – as measured by the MoCA total score – was a statistically significant positive predictor in all five regression models.

Table 3. Univariate multivariable linear regression of five neuropsychological domains by cannabis use, age, substance use, and depression (N = 522)

Note. Non-use of cannabis was entered as the reference category. The five outcome variables for the univariate models were speed/EF, visual, attention, language, and memory.

+ p < .10. *p < .05. **p < .01. ***p < .001.

Regarding cannabis use (users vs. non-users), estimated means with standard errors for the five regression models are presented in Table 4. The means are adjusted for covariates (i.e., age, sex, race, years of education, depression, MoCA total score). The differences in z-scores between users and non-users of cannabis ranged from -0.09 to 0.17. None of the tests of differences reached statistical significance (ps > .11). That is, there was no significant difference in cognitive domain performance between users and non-users of cannabis after controlling for relevant covariates.

Table 4. Estimated means and standard errors for five neuropsychological domains by cannabis use in the multivariable linear regression (N = 522)

Note. The estimated means and standard errors for the five neuropsychological domains are represented as z-scores (M = 0, SD = 1). Means are adjusted for covariates in the univariate multiple regression model: age, sex, race, education, depression (Geriatric Depression Scale-Short Form), and global cognitive functioning (Montreal Cognitive Assessment [MoCA] score).

Propensity score matching

PSM was used to estimate mean differences in performance across five cognitive domains between recent cannabis users (n = 59) and non-users (n = 93). PSM controls for confounding variables by matching participants based on their likelihood (propensity) of cannabis use, calculated using covariates including age, sex, race, education, depression score, and global cognitive functioning (MoCA score). Matching was performed using a 2:1 nearest-neighbor algorithm with replacement, meaning that non-users could be matched to more than one cannabis user if they had a similar propensity score. This approach ensures that comparisons are made between individuals with equivalent likelihoods of cannabis use, rather than requiring direct, one-to-one matches between participants. The analyses revealed no significant differences between cannabis users and non-users in any cognitive domain (all p’s > .38; see Table 5).

Table 5. Propensity score matching on five cognitive domains by cannabis use

Note. Differences in the neuropsychological domain performances by cannabis use (users vs. non-users) are estimated using propensity score matching. Cannabis users were matched to non-users in 2:1 nearest-neighbor matching strategy (with replacement) by age, sex, race, education, depression score (Geriatric Depression Scale-Short Form), and global cognitive functioning (Montreal Cognitive Assessment [MoCA] score).

p < .05. Means are presented as z-scores (M = 0, SD = 1).

Cannabis-related problem severity and cognitive performance

To further explore the relationship between cannabis use and cognition, we conducted follow-up analyses using participants’ total scores on the CUDIT-R. Based on established cutoffs, we initially categorized participants into three groups: “within normal limits” (total score of 0–7; n = 45), “hazardous use” (total score of 8–11; n = 11), and “cannabis use disorder” (total score of 12–32; n = 4). Due to small cell sizes, the hazardous use and cannabis use disorder groups were combined, resulting in a dichotomous classification: within normal limits (n = 45, 75%) and hazardous/problematic use (n = 15, 25%). Notably, two participants who endorsed cannabis use in the past six months scored zero on all eight CUDIT-R items, suggesting minimal engagement beyond a single use.

We conducted a series of ANCOVAs to evaluate whether this dichotomous indicator of cannabis use severity was associated with cognitive performance, controlling for age, sex, race, education, GDS total score, and MoCA total score. There were no significant differences between groups for speed/executive, attention, visual, or language domains (all ps > .28). However, a significant small-to-moderate effect was observed for memory performance, F(1, 51) = 6.66, p = .01, R 2 = .12. Estimated marginal means indicated that participants scoring within normal limits on the CUDIT-R performed significantly better on memory tasks (M = 0.53, SE = 0.12) than those with hazardous/problematic cannabis use (M = −0.17, SE = 0.22). See Table 6 for details. This effect remained significant after removing the two participants with CUDIT-R scores of 0, F (1, 49) = 5.77, p = .02, with estimated means of M = 0.50 (SE = 0.12) and M = −0.15 (SE = 0.23), respectively.

Exploratory correlations between CUDIT-R items and cognitive domains

To further explore the relationship between cannabis-related behaviors and cognition, we examined correlations between individual CUDIT-R items and the five cognitive domains. Item 3 (“How often during the past 6 months did you find that you were not able to stop using cannabis once you had started?”) was negatively correlated with performance in four cognitive domains: speed/executive (r = −0.29, p = .024), visual (r = −0.44, p < .001), language (r = −0.28, p = .033), and memory (r = −0.30, p = .022). In addition, memory performance was significantly negatively correlated with Item 1 (frequency of use; r = −0.35, p = .006) and Item 6 (“How often… have you had a problem with your memory or concentration after using cannabis?”; r = −0.43, p < .001) See Table 7 for details.

Table 7. Correlations between CUDIT-R items and cognitive domains (N = 60)

Note. *p < .05. **p < .01.

Discussion

Overall, our results indicate that older adults who reported using cannabis in the prior six months did not differ on neuropsychological testing results when compared to older adults who reported no recent use of cannabis. This pattern held true across both traditional analytical approaches (multivariate and univariate regressions) and propensity score matching, which was employed to balance groups and enhance causal inference. Participants who used cannabis and scored in the hazardous/problematic use range on the CUDIT-R demonstrated lower performance in the memory domain compared to participants who used cannabis but scored within normal limits on the CUDIT-R. Exploratory associations between item level data from the CUIT-R and cognitive outcomes suggest that difficulty stopping cannabis use (CUDIT-R Item 3) may be associated with broader cognitive difficulties across domains (i.e., speed/executive, visual, language, and memory).

Our results are generally consistent with findings from systematic reviews concluding that low frequency cannabis use among older adults is generally well tolerated and is not clearly associated with adverse cognitive outcomes (Scott et al. Reference Scott, Brennan and Benitez2019; Vacaflor et al. Reference Vacaflor, Beauchet, Jarvis, Schiavetto and Rej2020; Weinstein & Sznitman, Reference Weinstein and Sznitman2020). Yet, these findings stand in contrast to increased rates of SMC in older adults who use cannabis (Mulhauser et al. Reference Mulhauser, Hampstead, Coughlin and Ilgen2023; Vacaflor et al. Reference Vacaflor, Beauchet, Jarvis, Schiavetto and Rej2020). While SMC is common in older adults and frequently reported by individuals who perform within normal limits on comprehensive neuropsychological evaluation (Crumley et al. Reference Crumley, Stetler and Horhota2014), SMC has also been shown to be a harbinger of cognitive decline and future diagnosis of MCI and dementia (Lee & Foster, Reference Lee and Foster2023). Therefore, clarifying the relationship between increased SMC but generally similar objective cognition in older adults who use cannabis is worth further investigation. It remains a commonly held perception that cannabis use of any kind is deleterious to cognition (Carliner et al. Reference Carliner, Brown, Sarvet and Hasin2017), and this perception may contribute to increased concern for SMC among older adults who use cannabis. As older adults and their medical providers are increasingly considering cannabis to manage chronic health conditions, additional data on the interactive effects across cannabis use patterns, aging trajectories, co-occurring medical conditions, and cognitive status will help inform treatment decisions and manage potential risks.

Scott et al. (Reference Scott, Brennan and Benitez2019) reviewed 26 studies examining cannabis use and cognitive outcomes in older adults, including participants with normal aging, neurodegenerative diseases, and common medical conditions. Overall, they concluded that “higher doses and heavier use of cannabis are associated with modest negative effects” on cognition, while also noting that “the cognitive deficits associated with heavy, recreational cannabis may not be applicable to medical cannabis users, who may use products with less THC and experience relief from other symptoms, which may contribute to improved cognitive functioning” (Scott et al. Reference Scott, Brennan and Benitez2019, p. 452). Our study found no domain-level group difference in cognition (negative or positive) between older adults with and without recent cannabis use. Potentially consistent with Scott et al. (Reference Scott, Brennan and Benitez2019), our subsample of cannabis users who endorsed elevated risk for CUD demonstrated lower memory performance when compared to cannabis users without increased risk for CUD. The frequency of cannabis use in our sample ranged from less than monthly (41%) to 4 or more times per week (24%) with a median response of 2-4 times per month. Like Scott et al. (Reference Scott, Brennan and Benitez2019), our sample also lacks clear information on relevant use patterns that may inform outcomes. Even so, our study addresses some limitations identified by Scott et al. (Reference Scott, Brennan and Benitez2019) by using comprehensive performance-based neuropsychological testing results in the evaluation of older adults with mixed cognitive status.

In a secondary exploratory analysis, individual CUDIT-R items showed modest associations with cognitive performance, particularly within the memory domain. Notably, self-reported difficulty stopping cannabis use (CUDIT-R Item 3) was associated with lower performance across multiple cognitive domains (i.e., speed/executive, visual, language, memory), suggesting that diminished behavioral regulation around cannabis use may reflect broader cognitive vulnerabilities. In contrast, items reflecting frequency of use and cannabis-related memory complaints were more selectively associated with lower memory performance. These findings are consistent with recent work by Anquillare et al. (Reference Anquillare, Gallegos and Thayer2023), who found that greater cannabis-related problem severity was linked to higher self-reported cognitive failures and reduced cognitive self-efficacy in older adults. Although preliminary, our findings suggest that measurable cognitive differences may be linked to problematic patterns of cannabis use (i.e., impaired control, cannabis-related cognitive concerns), whereas such deficits are not observed in older adults who use cannabis without symptoms of a use disorder.

Several considerations and limitations are important when interpreting our results. First, like many studies of older adult cannabis use and cognition, several salient details about cannabis use patterns in our sample are unknown. For example, this sample does not include data on historical use patterns (e.g., age of onset, history of substance use disorder), motivational factors (e.g., recreational vs. medical), cannabinoid content (e.g., relative THC vs. CBD), or routes of administration (e.g., gummies, joints, tinctures), each of which may inform the associations between cannabis use and cognitive functioning in older adults. Second, our sample of older adults reporting recent cannabis use is small. Even so, this longitudinal cohort continues with recruitment and annual neuropsychological evaluations. As cannabis-related questions are now part of ongoing data collection and rates of cannabis use continue to rise among older adults, future studies will be positioned to examine within-person changes in cannabis use over time and their potential association with cognitive trajectories. Such longitudinal analyses will provide a more nuanced understanding of the dynamic relationship between cannabis use patterns and cognitive aging. Appropriate data on other substance use variables were not available for inclusion as covariates due to significant variability in data collection timing and response formats in the dataset; future studies will benefit from more robust and standardized assessment of behavioral use patterns across substances. Finally, we were not able to evaluate differential outcomes based on cognitive status or neurocognitive diagnosis due to the currently small sample of those reporting use. Further analysis of differences in cognition according to recent cannabis use within each cognitive phenotype may reveal subgroups that experience different risk/benefit profiles. Despite these limitations, this study advances our currently limited understanding of cannabis use and cognition in older adults by addressing current gaps in the literature and identifying areas for future research.

Conclusions

Our study contributes to the literature on cannabis use and cognition in older adults by examining comprehensive neuropsychological outcomes associated with recent (i.e., past six months) cannabis use in a well-characterized sample of older adults spanning the dementia continuum. Across the entire group, cognitive functioning was similar between older adults with and without recent cannabis use across five domains of cognition. Among older adults who reported recent use of cannabis, scoring in the hazardous/problematic use range on the CUDIT-R was associated with lower performance on memory tasks. Future research should incorporate detailed assessment of cannabis use patterns and individual characteristics to better understand the nuances of these relationships and potential differences based on cognitive phenotype.

Table 6. F-ratios from analyses of covariance (ANCOVAs) for cannabis use and cognitive functioning (N = 60)

Note. GDS, Geriatric Depression Scale-Short Form. MoCA, Montreal Cognitive Assessment. Age, sex (male vs. female), race (White, Black, Other), years of education, depression (GDS-SF), and general cognitive functioning (MoCA) were included in all five ANCOVAs as relevant covariates. The values depicted represent the F-ratios in the ANCOVAs.

a Cannabis use disorder was measured as a dichotomous variable (0 = within normal limits, 1 = hazardous/problematic use) using established cutoff values from the Cannabis Use Disorder Identification Test (CUDIT).

p < .05. **p < .01. ***p < .001.

Acknowledgements

None

Funding statement

This work was partially supported by the NIH/NIA funded Michigan Alzheimer’s Disease Center (P30AG053760 and P30AG072931; PI Henry Paulson, MD, PhD).

Competing interests

None

References

Abate, G., Uberti, D., & Tambaro, S. (2021). Potential and limits of cannabinoids in Alzheimer’s disease therapy. Biology, 10(6), 542.10.3390/biology10060542CrossRefGoogle ScholarPubMed
Adamson, S. J., Kay-Lambkin, F. J., Baker, A. L., Lewin, T. J., Thornton, L., Kelly, B. J., & Sellman, J. D. (2010). An improved brief measure of cannabis misuse: The cannabis use disorders identification test-revised (CUDIT-R). Drug and Alcohol Dependence, 110(1), 137–143.10.1016/j.drugalcdep.2010.02.017CrossRefGoogle ScholarPubMed
Anquillare, E., Gallegos, A. C., & Thayer, R. E. (2023). 15 associations among subjective cognitive function and cannabis use in older adults. Journal of the International Neuropsychological Society, 29(s1), 328–329.10.1017/S1355617723004526CrossRefGoogle Scholar
Aso, E., & Ferrer, I. (2014). Cannabinoids for treatment of Alzheimer’s disease: Moving toward the clinic. Frontiers in Pharmacology, 5, 1–11.10.3389/fphar.2014.00037CrossRefGoogle ScholarPubMed
Austin, P. C. (2010). Statistical criteria for selecting the optimal number of untreated subjects matched to each treated subject when using many-to-one matching on the propensity score. American Journal of Epidemiology, 172(9), 1092–1097.10.1093/aje/kwq224CrossRefGoogle ScholarPubMed
Bahji, A., Meyyappan, A. C., & Hawken, E. R. (2020). Cannabinoids for the neuropsychiatric symptoms of dementia: A systematic review and meta-analysis. The Canadian Journal of Psychiatry, 65(6), 365–376.10.1177/0706743719892717CrossRefGoogle ScholarPubMed
Briscoe, J., & Casarett, D. (2018). Medical marijuana use in older adults. Journal of the American Geriatrics Society, 66(5), 859–863.10.1111/jgs.15346CrossRefGoogle ScholarPubMed
Carliner, H., Brown, Q. L., Sarvet, A. L., & Hasin, D. S. (2017). Cannabis use, attitudes, and legal status in the U.S.: A review. Preventive Medicine, 104, 13–23.10.1016/j.ypmed.2017.07.008CrossRefGoogle ScholarPubMed
Crean, R. D., Crane, N. A., & Mason, B. J. (2011). An evidence-based review of acute and long-term effects of cannabis use on executive cognitive functions. Journal of Addiction Medicine, 5(1), 1–8.10.1097/ADM.0b013e31820c23faCrossRefGoogle ScholarPubMed
Crumley, J. J., Stetler, C. A., & Horhota, M. (2014). Examining the relationship between subjective and objective memory performance in older adults: A meta-analysis. Psychology and Aging, 29(2), 250–263.10.1037/a0035908CrossRefGoogle ScholarPubMed
Dodge, H. H., Goldstein, F. C., Wakim, N. I., Gefen, T., Teylan, M., Chan, K. C. G., Kukull, W. A., Barnes, L. L., Giordani, B., Hughes, T. M., Kramer, J. H., Loewenstein, D. A., Marson, D. C., Mungas, D. M., Mattek, N., Sachs, B. C., Salmon, D. P., Willis‐Parker, M., Welsh‐Bohmer, K. A. … National Alzheimer’s Coordinating Center (NACC) (2020). Differentiating among stages of cognitive impairment in aging: Version 3 of the uniform data set (UDS) neuropsychological test battery and MoCA index scores. Alzheimer’s & Dementia: Translational Research & Clinical Interventions, 6(1), e12103.Google ScholarPubMed
Han, B. H., & Palamar, J. J. (2018). Marijuana use by middle-aged and older adults in the United States, 2015–2016. Drug and Alcohol Dependence, 191, 374–381.10.1016/j.drugalcdep.2018.07.006CrossRefGoogle ScholarPubMed
Han, B. H., & Palamar, J. J. (2020). Trends in cannabis use among older adults in the United States, 2015–2018. JAMA Internal Medicine, 180(4), 609.10.1001/jamainternmed.2019.7517CrossRefGoogle Scholar
Han, B. H., Sherman, S., Mauro, P. M., Martins, S. S., Rotenberg, J., & Palamar, J. J. (2017). Demographic trends among older cannabis users in the United States, 2006–13. Addiction, 112(3), 516–525.10.1111/add.13670CrossRefGoogle ScholarPubMed
Jeffers, A. M., Glantz, S., Byers, A., & Keyhani, S. (2021). Sociodemographic characteristics associated with and prevalence and frequency of cannabis use among adults in the US. JAMA Network Open, 4(11), e2136571.10.1001/jamanetworkopen.2021.36571CrossRefGoogle ScholarPubMed
Khoury, R., Maliha, P., & Ibrahim, R. (2022). Cannabis use and misuse in older adults. Clinics in Geriatric Medicine, 38(1), 67–83.10.1016/j.cger.2021.07.003CrossRefGoogle ScholarPubMed
Kiselica, A. M., Webber, T. A., & Benge, J. F. (2020). The uniform Dataset 3.0 neuropsychological battery: Factor structure, invariance testing, and demographically adjusted factor score calculation. Journal of the International Neuropsychological Society, 26(6), 576–586.10.1017/S135561772000003XCrossRefGoogle ScholarPubMed
Kroon, E., Kuhns, L., Hoch, E., & Cousijn, J. (2020). Heavy cannabis use, dependence and the brain: A clinical perspective. Addiction (Abingdon, England), 115(3), 559–572.10.1111/add.14776CrossRefGoogle ScholarPubMed
Lee, C. D., & Foster, E. R. (2023). Subjective memory complaints predict decline in memory, instrumental activities of daily living, and social participation in older adults: A fixed-effects model. The American Journal of Occupational Therapy, 77(4), 7704205100.10.5014/ajot.2023.050151CrossRefGoogle ScholarPubMed
Lovell, M. E., Akhurst, J., Padgett, C., Garry, M. I., & Matthews, A. (2020). Cognitive outcomes associated with long-term, regular, recreational cannabis use in adults: A meta-analysis. Experimental and Clinical Psychopharmacology, 28(4), 471–494.10.1037/pha0000326CrossRefGoogle ScholarPubMed
Maxwell, C. J., Jesdale, B. M., & Lapane, K. L. (2021). Recent trends in cannabis use in older Americans. Annals of Internal Medicine, 174(1), 133–135.10.7326/M20-0863CrossRefGoogle ScholarPubMed
Meier, M. H., Caspi, A., Danese, A., Fisher, H. L., Houts, R., Arseneault, L., & Moffitt, T. E. (2018). Associations between adolescent cannabis use and neuropsychological decline: A longitudinal co-twin control study: Adolescent cannabis use and IQ decline. Addiction, 113(2), 257–265.10.1111/add.13946CrossRefGoogle Scholar
Minerbi, A., Häuser, W., & Fitzcharles, M.-A. (2019). Medical cannabis for older patients. Drugs & Aging, 36(1), 39–51.10.1007/s40266-018-0616-5CrossRefGoogle ScholarPubMed
Mulhauser, K., Hampstead, B. M., Coughlin, L. N., & Ilgen, M. A. (2023). The association between cannabis use and subjective memory complaints in older adults in the United States. Journal of the International Neuropsychological Society, 29(9), 870–877.10.1017/S1355617723000061CrossRefGoogle ScholarPubMed
NACC (2024). About NACC data. National Alzheimer’s Coordinating Center.Google Scholar
Pocuca, N., Walter, T. J., Minassian, A., Young, J. W., Geyer, M. A., & Perry, W. (2021). The effects of cannabis use on cognitive function in healthy aging: A systematic scoping review. Archives of Clinical Neuropsychology: The Official Journal of the National Academy of Neuropsychologists, 36(5), 673–685.10.1093/arclin/acaa105CrossRefGoogle ScholarPubMed
Ross, J. M., Ellingson, J. M., Rhee, S. H., Hewitt, J. K., Corley, R. P., Lessem, J. M., & Friedman, N. P. (2020). Investigating the causal effect of cannabis use on cognitive function with a quasi-experimental co-twin design. Drug and Alcohol Dependence, 206, 107712.10.1016/j.drugalcdep.2019.107712CrossRefGoogle ScholarPubMed
Scott, E. P., Brennan, E., & Benitez, A. (2019). A systematic review of the neurocognitive effects of cannabis use in older adults. Current Addiction Reports, 6(4), 443–455.10.1007/s40429-019-00285-9CrossRefGoogle ScholarPubMed
Staffaroni, A. M., Asken, B. M., Casaletto, K. B., Fonseca, C., You, M., Rosen, H. J., Boxer, A. L., Elahi, F. M., Kornak, J., Mungas, D., & Kramer, J. H. (2021). Development and validation of the uniform data set (v3.0) executive function composite score (UDS3-EF). Alzheimer’s & Dementia, 17(4), 574–583.10.1002/alz.12214CrossRefGoogle ScholarPubMed
Statistics Canada (2019). National cannabis survey, third quarter 11, 1–14.Google Scholar
Statistics Canada, & Rotermann, M. (2020). What has changed since cannabis was legalized? Health Reports, 31(2), 11–20.Google ScholarPubMed
Tumati, S., Lanctôt, K. L., Wang, R., Li, A., Davis, A., & Herrmann, N. (2022). Medical cannabis use among older adults in Canada: Self-reported data on types and amount used, and perceived effects. Drugs & Aging, 39(2), 153–163.10.1007/s40266-021-00913-yCrossRefGoogle ScholarPubMed
Vacaflor, B. E., Beauchet, O., Jarvis, G. E., Schiavetto, A., & Rej, S. (2020). Mental health and cognition in older cannabis users: A review. Canadian Geriatrics Journal, 23(3), 237–244.10.5770/cgj.23.399CrossRefGoogle ScholarPubMed
Volkow, N. D., Swanson, J. M., Evins, A. E., DeLisi, L. E., Meier, M. H., Gonzalez, R., Bloomfield, M. A. P., Curran, H. V., & Baler, R. (2016). Effects of cannabis use on human behavior, including cognition, motivation, and psychosis: A review. JAMA Psychiatry, 73(3), 292.10.1001/jamapsychiatry.2015.3278CrossRefGoogle ScholarPubMed
Weinstein, G., & Sznitman, S. R. (2020). The implications of late-life cannabis use on brain health: A mapping review and implications for future research. Ageing Research Reviews, 59, 1–8.10.1016/j.arr.2020.101041CrossRefGoogle Scholar
Wolfe, D., Corace, K., Butler, C., Rice, D., Skidmore, B., Patel, Y., Thayaparan, P., Michaud, A., Hamel, C., Smith, A., Garber, G., Porath, A., Conn, D., Willows, M., Abramovici, H., Thavorn, K., Kanji, S., & Hutton, B. (2023). Impacts of medical and non-medical cannabis on the health of older adults: Findings from a scoping review of the literature. PLOS ONE, 18(2), e0281826.10.1371/journal.pone.0281826CrossRefGoogle ScholarPubMed
Zhornitsky, S., Pelletier, J., Assaf, R., Giroux, S., Li, C. R., & Potvin, S. (2021). Acute effects of partial CB1 receptor agonists on cognition – a meta-analysis of human studies. Progress in Neuro-Psychopharmacology and Biological Psychiatry, 104, 110063.10.1016/j.pnpbp.2020.110063CrossRefGoogle ScholarPubMed
Figure 0

Table 1. Demographic and clinical characteristics (N = 540)

Figure 1

Table 2. Multivariate analysis of covariance (MANCOVA) of five neuropsychological domains by cannabis use and relevant covariates (N = 522)

Figure 2

Table 3. Univariate multivariable linear regression of five neuropsychological domains by cannabis use, age, substance use, and depression (N = 522)

Figure 3

Table 4. Estimated means and standard errors for five neuropsychological domains by cannabis use in the multivariable linear regression (N = 522)

Figure 4

Table 5. Propensity score matching on five cognitive domains by cannabis use

Figure 5

Table 7. Correlations between CUDIT-R items and cognitive domains (N = 60)

Figure 6

Table 6. F-ratios from analyses of covariance (ANCOVAs) for cannabis use and cognitive functioning (N = 60)