Abstract
Objective:
Although inattention symptoms have been previously linked to cognitive performance in younger samples, few studies have examined links between ADHD symptoms and cognitive performance for middle aged and older adults.
Methods:
In this study, we drew from a nationally representative sample from the Health and Retirement Study (HRS) of ~1,400 middle to older adults (Mage = 66.9, SD = 8.4; 41.4% male; 60.7% White) who completed a set of cognitive measures and an ADHD symptomatology questionnaire in the 2016 Wave of the HRS. A multigroup path model was run by examining the association between self-reported ADHD symptom subscale scores for inattention and hyperactivity/impulsivity as well as self-reported depressive symptoms and cognitive outcomes across three groups: middle age, young-old, and middle-old.
Results:
Inattention symptoms were significantly associated with Serial 7s and Immediate Recall, however the constrained model was the best fitting model, suggesting no differences in the associations between self-reported inattention symptoms and cognitive outcomes by age.
Conclusion:
These results are consistent with previous work on the links between ADHD symptoms and cognitive performance in younger populations and add to the literature on ADHD in later life. This may have implications for clinicians and practitioners as well as future research on older adults with ADHD.
Introduction
ADHD
ADHD is a neurodevelopmental disorder that is characterized by executive dysfunction, where individuals have age-inappropriate and interfering difficulty with avoiding distractors, planning and accomplishing tasks/goals, with such difficulty starting in early childhood (i.e., prior to age 12 years) and being present across multiple settings (American Psychiatric Association [APA], 2022). Although ADHD was initially thought to only affect children and adolescents, it is now recognized as being present throughout adulthood, including older adulthood (Barkley et al., 2002; Callahan & Plamondon, 2019; Fischer & Nilsen, 2024). Worldwide, it is estimated to impact 2.6% to 6.8% of adults (Song et al., 2021). To meet criteria for an ADHD diagnosis, children/adolescents must display at least six symptoms out of the nine total inattention symptoms and/or nine total hyperactive/impulsive symptoms; whereas adults have a lower symptom threshold (i.e., five out of nine possible symptoms of inattention and/or hyperactivity/impulsivity) as they can experience ADHD related impairments with fewer symptoms (APA, 2022). Additionally, ADHD symptoms must not be better explained by any other psychological or health disorders (e.g., depression and traumatic brain injury). ADHD impacts a variety of life outcomes including educational attainment and workplace performance; health issues including injuries, functional limitations, and comorbidities; and quality of life (Brod et al., 2012; de Zeeuw et al., 2017; Fredriksen et al., 2014; Jangmo et al., 2021; Landes & London, 2021).
Given the far-reaching effects of ADHD on daily life, ADHD diagnoses are essential for improving outcomes throughout the lifespan (Faraone et al., 2021). For example, Pawaskar et al. (2020) showed that middle-aged individuals diagnosed with ADHD report better functioning and productivity at work with a higher health-related quality of life and self-esteem than their undiagnosed peers with elevated ADHD symptoms. Consistent with this, a qualitative study with 21 adult participants who were interviewed regarding their experiences before and after their diagnosis concluded that there were major positive consequences of diagnosis compared to being undiagnosed, especially with regards to the treatment of their symptoms (Halleröd et al., 2015).
Adult ADHD
The presentation of ADHD symptoms changes in adulthood, such that inattention symptoms tend to persist while hyperactivity and impulsivity symptoms typically diminish (Callahan & Plamondon, 2019; Cheung et al., 2015; Semeijn et al., 2016). Additionally, ADHD symptoms are generally more heterogeneous for adults, and may be typified by feelings of restlessness or issues such as choosing busy lifestyles (hyperactivity), making decisions without thinking through the options and consequences (impulsivity), or having trouble concentrating/focusing (inattention; Adler, 2004). This may in turn lead to unfavorable outcomes such as low occupational performances that lead to lower incomes, higher unemployment rates and financial issues, and increased risks for comorbid mood disorders and suicide (Leffa et al., 2022).
Research on adult ADHD has been increasing since the turn of the century but has not quite yet focused on ADHD in older adulthood (i.e., in individuals ages 65+ years), which has an estimated prevalence of about 3% (Kooij et al., 2016; Michielsen et al., 2012). Although several validated adult ADHD screening tools have been developed, there is no validated scale specifically for older adults, which may be problematic due to the differential presentation of symptoms in later life that may be linked to age-related cognitive decline characterized by Mild Cognitive Impairment (MCI) or dementia instead of ADHD (Kooij et al., 2016; Sibley, 2021). Indeed, there are multiple factors to consider for ADHD in later life: a study conducted in the Netherlands reported age effects such that young-old adults (aged 60–70 years) had higher ADHD symptoms than their older counterparts (aged 71–94 years; Michielsen et al., 2012). This suggests a potential link of ADHD to earlier mortality, which is corroborated by literature (Barkley, 2020; Dalsgaard et al., 2015; Leffa et al., 2022; O’Nions et al., 2025). Additionally, using data from the same study of older adults, Michielsen et al. (2013) reported positive associations between ADHD diagnosis and symptoms and clinical levels of depression and anxiety, such that higher ADHD symptoms meant higher depressive and anxiety symptoms on average. This finding underscores the importance of considering differential diagnosis (i.e., alternate explanations for ADHD symptoms) when assessing ADHD in older adults, and accounting for comorbid internalizing symptoms when examining the links between ADHD and cognitive functioning.
ADHD and Cognitive Performance
ADHD is linked to impairments in cognition that manifest as measurable differences in cognitive functioning between individuals with and without ADHD, typically across the lifespan (Pazvantoğlu et al., 2012). Basic capacities such as processing speed as well as higher order functions appear to be impaired (Butzbach et al., 2019; Mohamed et al., 2021). In their systematic review, Onandia-Hinchado et al. (2021) reported that in addition to deficits in attention and sub-domains of EF such as working memory and inhibitory control, individuals with ADHD may also show deficits in verbal memory, language, arithmetic, social cognition, and processing speed when compared to their non-ADHD counterparts. In comparisons of performance on working memory tests like the Differential Abilities Scale II and the NIH List Sorting task (Kofler et al., 2024), children who don’t have ADHD are often reported to have outperformed those who did have ADHD (Jusko et al., 2021; Reddy et al., 2008; Weisen, 2008); however, these tests were designed for younger populations and have not yet been used as extensively in adult and older adult samples.
For middle and older adults, Serial 7s is often used as measure of arithmetic (and working memory components of arithmetic; Bristow et al., 2016; Saito et al., 2020) as well as processing speed (Williams et al., 1996) both of which may be impacted in ADHD. Similarly, Number Series is linked to quantitative reasoning (Schrank & Wendling, 2018) and fluid intelligence (Fisher et al., 2013; Langa et al., 2020). A recent factor analysis of cognitive measures in the Health and Retirement Study, including the Number Series, suggested that it was potentially linked to set shifting as well (Jones et al., 2024). However, the few studies that have assessed Number Series performance in children and young adults with ADHD have not found significant differences in performance on Number Series between individuals with and without ADHD (Hart, 2020; Spenceley et al., 2022).
Finally, there is at least one recent study that has identified links between ADHD and verbal memory. Mendonca et al. (2021) reported verbal memory deficits, particularly in processes involving acquisition, delayed recall, and recognition for individuals with ADHD when compared to control participants without ADHD. They suggested that this may be due to inefficient memory strategies that may be associated with weak encoding of stimuli.
ADHD and Cognitive Performance in Older Adulthood
Although it is now established that adults with ADHD may have deficits in cognitive domains of attention, inhibition, memory, and executive functioning, two studies reported that the link between ADHD and performance on processing speed, memory, and executive functioning tasks were actually explained by depressive symptoms. Specifically, Semeijn et al. (2015) examined the relationship between depressive symptoms, ADHD diagnosis, and cognitive performance using data from the Longitudinal Aging Study Amsterdam of 231 older adults between the ages of 60 and 94 years. They found that attention and working memory performance, as measured by the Mini Mental State Examination, Raven’s Colored Progressive Matrices, and Stroop Color-Word, was not significantly associated with ADHD diagnosis or severity when controlling for depressive symptoms. This finding was further supported by another population study in Australia, which conducted similar analyses but stratified their sample into two age bands: middle aged (n = 2,182, age range = 48–52 years) and older adults (n = 1,973, age range = 68–72 years; Das et al., 2014). Das et al. (2014) aimed to not only validate the Adult ADHD Self-Report Scale (ASRS; Kessler et al., 2005) in older adults but also observe the associations between inattention and hyperactivity/impulsivity and cognitive performance by age group. Notably, the cognitive tasks included Immediate and Delayed Recall, among others, which our study also included. Das et al. (2014) found that the effects of ADHD symptoms on cognitive performance were weaker in older adults compared to middle-aged participants. Most importantly, their findings consistently showed that the relationship between ADHD symptoms and cognitive performance was indirect, mediated by the strong association between depressive symptoms and cognition, or that depression is a confounding variable in the link between ADHD and cognitive performance.
Although Das et al. (2014) divided their sample into middle-aged and older adults, given the changes in cognitive functioning in later life, further stratification within older adulthood may be pertinent to better observe the links between cognitive performance and ADHD symptoms. Furthermore, the extant work on ADHD and cognitive functioning in later life has been conducted on samples located outside the United States (U.S.; Das et al., 2014; Michielsen et al., 2013; Semeijn et al., 2015). And so, it is imperative to explore how these associations exist in a large national sample of U.S. older adults in the Health and Retirement study, as well as whether they are impacted by co-occurring disorders such as depression. The authors systematically evaluated published studies using this publicly available dataset (through September 25, 2025) and warrant that this study presents unique analyses that provide a distinct contribution to the literature.
Research Questions and Hypotheses
In that vein, this study builds on the limited existing research by examining a broader age range of older adults, and by examining the association between ADHD symptoms and cognitive functioning in middle-aged and older adults in the U.S. while accounting for depressive symptoms as well as participants’ race and ethnicity.
Methods
Sample
The sample was drawn from the Health and Retirement Study (HRS), a publicly available dataset collected by the University of Michigan. It is a nationally representative longitudinal survey that has been collecting data since 1992 from approximately 20,000 older adults. The survey uses randomized stratification to capture a wide range of measures, ranging from cognition and lifestyle activities to biomarkers and finances (Fisher & Ryan, 2018). The data used for this analysis was drawn from the 2016 Wave which also included an ADHD self-reported symptoms questionnaire in the Experimental Module (sent to a random 10% of the Core HRS sample) as well as the cognitive and demographic measures collected in the Core HRS survey.
To refine the sample, an inclusion criterion was applied where selected participants also had completed the experimental module and answered the questions related to the self-reported ADHD symptoms, did not report a diagnosis of dementia or Alzheimer’s, and were the primary respondent for the cognitive assessments. Additionally, only individuals who were between the age of 55 and 85 in the 2016 wave were included. Age bands were created with 55 to 64 as middle-aged adults, 65 to 74 as young-old, and 75 to 84 as middle-old adults. Participants older than 85 are considered to be in the old-old age band; however, considering that this sample was already a subsample from the larger HRS sample (containing participants for the Experimental Module) as well as the additional inclusion criteria, very few participants were present in the old-old age band (n = 98). Hence, participants older than 85 were excluded from analysis in the current study. Similarly, participants younger than 55 were excluded since cognitive data is collected only from participants older than 50 and so, a sufficient sample size did not exist for the 45 to 54 age band of middle-aged adults. The final sample size consisted of 1,326 participants (Mage = 66.9, SD = 8.4; 41.4% male; 60.7% White) who had competed at least a high school level of education (Myears = 12.9, SD = 2.9). A more detailed breakdown of the final sample’s demographics is presented in Table 1.
Sample Descriptives by Age Band.
Note. CESD = Center for Epidemiology Studies Scale for Depression; ASRS = Adult ADHD Self Report Scale.
p < .1. *p < .05. **p < .01. ***p < .001.
Cognitive Outcome Measures
For cognitive measures, the cognitive tests administered to most of the sample were selected as outcomes, these include Number Series and Serial 7s, which utilize working memory and auditory attention, and similar to Das et al.’s (2014) study, the Immediate and Delayed Recall tasks were utilized to measure memory performance quality.
Number Series
Number Series involves having participants fill in missing numbers among a series of numbers by determining the relationship between them (Schrank et al., 2014). As such, Number Series is considered a measure of quantitative reasoning, a component of fluid intelligence (Gf; Schrank & Wendling, 2018) that may have some overlap with set shifting (Jones et al., 2024). The HRS Number Series is a six-block test that adapts according to the participants’ ability. The score is a standardized logit scale score that takes into account the probability of answering the item correctly, and as such, ranges from 390 to 580, with higher scores indicating better performance (Fisher et al., 2013). The Number Series score has been used as a continuous measure in the analyses and was rescaled (divided by 100) to aid parameter interpretation and model convergence by ensuring the scale was in proportion with the other variables in the model.
Serial 7s
Serial 7s involves participants subtracting the number 7 backwards starting from 100, for a total of five trials. The final score is the number of correct trials of subtracting 7 backwards and ranges from 0 to 5. Participants have to hold the answer of each trial in their head (without any prompting or help from the interviewer) while subtracting 7 from it; as such, this test is considered to be closely related to working memory as well as mental processing (Karzmark, 2000; Manning, 1982). The Serial 7s score has been used as a continuous outcome in the analyses.
Immediate and Delayed Recall
Immediate Recall involves listening to a list of 10 random words with no overlapping content or meaning and recalling as many of them as possible in no particular order. As such, it is considered to be a test of immediate episodic recall. Delayed Recall involves asking participants to recall the list of 10 words read to them around 5 min earlier in any order (for the Immediate Recall test). Delayed Recall is considered to be a measure of episodic memory (Ofstedal et al., 2005). The final score for both Immediate Recall and Delayed Recall is a sum of the number of words participants recall correctly, which ranges from 0 to 10. Scores on Immediate and Delayed Recall were used as continuous measures in the analyses.
Primary Predictor
ADHD Symptoms
The ASRS is an 18-item self-report questionnaire that identifies the frequency of DSM symptoms in adults. The scale is often divided into two subscales, each based on nine questions specifically related to inattention and hyperactivity/impulsivity symptoms. Participants were asked to respond with a yes (coded as 1) or no (coded as 0) if they had experienced any of the symptoms; the responses were then summed up to create the final score separately for each subscale. The ASRS total score provides a dimensional measure of ADHD symptoms, with higher scores indicating a greater number of ADHD symptoms (Kessler et al., 2005). A threshold of five symptoms or more is considered to be clinically significant for each subscale (13.2% of sample for inattention, 15.1% of sample for hyperactivity/impulsivity). However, it is important to note that symptom scores on the ASRS that are equal to or above 5 may not necessarily reflect a diagnosis of ADHD, for which a clinical assessment is required. Chamberlain et al. (2021) reported that the ASRS may over-identify ADHD by capturing elevated inattention and hyperactivity/impulsivity symptoms due to other conditions (e.g., depression), which may not be distinguishable through a self-reported questionnaire. Reliability of the ASRS in the current study was acceptable for inattention (α = .84) and questionable for hyperactivity/impulsivity (α = .69); it was acceptable overall across the 18-items (α = .84).
Covariates
Age
As one of the major predictors of cognitive decline (Deary et al., 2002; Reuter-Lorenz & Park, 2014), age is an essential predictor to include in analyses of cognitive functioning in later life. In this study, participants were included from the ages of 55 to 85 and were then stratified into age bands based on commonly determined categories: older middle age (55–64 years), young-old (65–74 years), and middle-old (75–84 years; Hamarat et al., 2002; Koo et al., 2017; Lee et al., 2018). Each age band has a span of 10 years and so to account for the developmental changes that may have occurred within that time frame, participant age (within the range of their corresponding age band) was included as a covariate. Age in the HRS is asked at three timepoints, the interview beginning, middle, and end; in these analyses, age at the end of the interview is used to signify participant age.
Sex Assigned at Birth
Cognitive functioning varies somewhat by sex assigned at birth but especially within the context of cognitive aging (Maitland et al., 2000). McCarrey et al. (2016) report varying trajectories of cognitive decline among individuals assigned as male and female at birth in a clinically normal sample of older adults, particularly in the domains of mental status and visuospatial ability. Within the context of ADHD, sociocultural influences on behavior may influence differences in how ADHD is diagnosed as well as treated for individuals identifying as male or female, with approximately 2:1 diagnoses for male and female individuals (Martin, 2024). Platania et al. (2025) recently reported that there may be differences in ADHD symptoms reported by sex: male participants reported more childhood symptoms whereas their female counterparts had a higher tendency to report inattentive and hyperactive/impulsive symptoms in adulthood. In the HRS, sex assigned at birth is coded as 0 for male and 1 for female.
Education
Education is linked to cognitive performance through multiple pathways; it is often highly correlated with cognitive functioning (Reuter-Lorenz & Park, 2014) and linked to cognitive reserve (Meng & D’Arcy, 2012). It is also considered as a proxy for socioeconomic status (Galobardes et al., 2006), which is another factor that may affect cognitive performance in later life (Krueger et al., 2025). Lower formal education is related to sharper declines in cognition in later life (Alley et al., 2007; Evans et al., 1993; Lövdén et al., 2018) as well as greater incidence of dementia (Caamaño-Isorna et al., 2006). Experience in educational environments may in some ways prepare individuals for cognitive assessments. And so, it is necessary to account for all these factors and include education as a covariate when predicting cognitive outcomes in later life (Ramos-Henderson et al., 2025; Tavares-Júnior et al., 2019). Education is also linked to ADHD such that individuals with ADHD often report lower educational attainment than their peers without ADHD particularly as ADHD symptoms may directly impact schooling experiences (Fredriksen et al., 2014). In the HRS, education is captured by the number of years of education participants report, ranging from 0 to 17, with 17 indicating more than undergraduate education.
Self-Rated Health
Self-rated health predicts not only cognitive performance but also mortality (Idler & Kasl, 1995). As such, it is an important predictor used in older adult populations for functional aging (Arnadottir et al., 2011). Self-rated health is also linked to ADHD such that individuals reporting higher ADHD symptoms also report having poorer self-rated health among other life outcomes (Landes & London, 2021). In the HRS, self-rated health is a five-point scale, where higher scores indicate worse health.
Depressive Symptoms
The Center for Epidemiology Studies Scale for Depression (CESD) was used to determine the extent of depressive symptoms participants reported (Radloff, 1977). In the HRS, the CESD 8-item scale is used, where higher scores indicate more depressive symptoms with a cutoff of 4 being the clinical threshold. Reliability of the CESD in the current study was acceptable (α = .77).
Analytical Strategy
Following descriptive analysis, bivariate Pearson’s correlations were conducted to observe any trends among the variables of interest. A two step process was utilized to run the multigroup path model and assess differences across age bands using lavaan v0.6-19 in R v4.4.3 (Rosseel, 2012). A series of path models were first run to observe the associations between self-reported inattention, hyperactivity/impulsivity, depressive symptoms, and cognitive outcomes independently. All cognitive outcomes were covaried with each other, with the exception of Immediate and Delayed Recall, which had a directional path from Immediate Recall toward Delayed Recall to account for initial encoding of information in verbal recall. Inattention and hyperactivity/impulsivity symptoms were entered simultaneously, followed by depressive symptoms given the robust correlation between inattention, hyperactivity/impulsivity, and depressive symptoms. There is some evidence that inattention may drive hyperactivity/impulsivity in children and adults with ADHD (Michielsen et al., 2013; Sokolova et al., 2016), however this is in contrast to other studies that report that hyperactivity may actually be linked to increased working memory demands in both children as well adults (Hartanto et al., 2016; Hudec et al., 2014; Rapport et al., 2009; Sarver et al., 2015). Initial nested models contained only self-reported ADHD and depressive symptoms (Models A and B respectively); covariates were added in Model C1 and all paths were estimated without any constraints.
In the second step, the paths for model C1 were fully constrained (Model C2) to observe differences across self-reported symptoms and cognitive outcomes by age bands. Constraints were removed in each successive iteration from inattention (Model D1), hyperactivity/impulsivity (Model D2), both inattention and hyperactivity/impulsivity (Model E), and then depressive symptoms (Model F). Model fit indices were then compared across each model to arrive at the best fitting model. Much of this work was exploratory and we note that the current study and its analyses were not pre-registered.
Results
Sample Descriptive Statistics
Means and standard deviations for each age band as well as the overall sample are displayed in Table 1. Statistical tests were also conducted (one way ANOVAs for education, self-rated health; chi squares for sex, race, ethnicity) to see if there were any differences in demographics across the age bands. There were non-significant differences for sex assigned at birth across the age bands with a χ2(2, 1,329) = 6.00 and p = .050 although adjusted residuals hinted at a greater number of female participants than expected in the young-old age band. Although the male/female ratio of ADHD diagnoses tends to favor males (Martin, 2024; Mowlem et al., 2019), our sample contains a higher rate of female participants. This may be partly due to the greater survival rates for female individuals as compared to male individuals (Feinglass et al., 2007; Murphy et al., 2021). This imbalance in life expectancy may be compounded by experiencing ADHD-related complications in life outcomes including physical health. O’Nions et al. (2025) reported that male participants diagnosed with ADHD in the United Kingdom had a total life expectancy of 73.26 years compared to 80.03 years for non-ADHD matched counterparts. Similarly, there was a 9-year difference for female participants with an ADHD diagnosis, whose total life expectancy of 75.15 years was much lower than the 83.79 years of non-ADHD matched controls. Apart from survival, another reason may be the way HRS samples participants. Recruitment is typically done within the same household as long as one spouse meets the age criteria (51 years and above), which means that spouses within those households are included even if they are less than 50 years old (Heeringa & Connor, 1995). Given female spouses are generally younger than male spouses in the United States (Cherlin, 2010), this may contribute to a greater number of female participants than male participants for this study.
The racial composition across age bands did have significant differences [χ2(4, 1,327) = 70.1, p < .001], with a greater than expected number of participants reporting their race to be Black or Other in the middle age group and White in the middle old group, based on adjusted residuals. A similar pattern was observed for ethnicity, [χ2(2, 1,326) = 11.9, p = .003], with a greater number of participants identifying as Hispanic in the middle age group while more non-Hispanic participants were observed in the middle old age band, based on adjusted residuals. This is likely due to the intentional oversampling of minoritized racial-ethnic populations in the HRS (Ofstedal & Weir, 2011).
Education also differed significantly across age bands [F(2, 1,319) = 4.04, p = .018], with a post-hoc Tukey’s test showing that the young old age band had significantly more years of education as compared to the middle old group (Mean difference = 0.64, p = .013). Self-rated health was also significantly different across the age groups [F(2, 1,326) = 3.24, p = .039] with the middle old age band reporting significantly worser self-rated health than the young-old age band (mean difference = 0.185. p = .041) in a post-hoc Tukey’s test.
Symptoms by Age Band
The mean symptom scores for each age band are also reported in Table 1. In a one-way ANOVA comparing the mean subscale scores of inattention and hyperactivity/impulsivity across the age bands, there was a significant difference in the mean hyperactivity/impulsivity subscale score, F(2, 1,284) = 4.08, p = .017. Probing where the group differences were using a Tukey Test revealed that the middle age and middle-old age bands significantly differed (mean difference = 0.37, p = .015) for the hyperactivity/impulsivity subscale. No significant differences emerged for the mean subscale score of inattention. Depressive symptom scores also varied significantly across age bands when examined using a one-way ANOVA, F(2, 1,326) = 5.20, p = .006. Using the Tukey Test revealed that mean difference of 0.387 (p = .005) between the middle age and young-old age bands.
Descriptive statistics by age band and clinical range are reported in Table 2. The number of participants in the clinical range for inattention and hyperactivity/impulsivity symptoms as well as depressive symptoms also varied by age band, with a higher proportion of middle-old participants (14.1%) falling above the clinical threshold for inattentive symptoms, but the lowest for hyperactivity/impulsivity (8%) and depressive symptoms (11.2%). A chi-square test was run to see whether the number of people above the clinical threshold for self-reported inattention, hyperactivity/impulsivity, and depressive symptoms also differed by age band; such differences were not found to be statistically significant (ps > .107).
Symptom Descriptives by Age Band and Clinical Range.
Note. CESD = Center for Epidemiology Studies Scale for Depression.
Correlation Table
Results for the Pearson’s correlation analysis are depicted in Table 3; among the variables of interest, age was inversely associated with all four cognitive outcomes with small‑to‑moderate negative correlations (r = −.05 to −.29, ps < .01). Higher education was positively correlated with cognitive performance (r = .28 to .46, ps < .01) and negatively with ADHD and CESD symptoms (r = −.19 to −.23, ps < .01). Inattention and hyperactivity/impulsivity symptoms correlated strongly with each other (r = .55, p < .01), and modestly with CESD symptoms (r = .34 to .43, ps < .01). Inattention symptoms were also negatively associated with cognitive measures with (r = −.19 to −.23, ps < .01) while similar associations were observed for hyperactivity/impulsivity symptoms as well (r = −.01 to −.15, ps < .01).
Bivariate Correlation Table.
Note. CESD = Center for Epidemiology Studies Scale for Depression.
p < .1. *p < .05. **p < .01. ***p < .001.
Multigroup Path Model
For the first stage, a series of nested path models were run to assess the associations between the self-reported symptoms and cognitive outcomes across age bands, starting with Model A which had only self-reported inattention and hyperactivity/impulsivity symptoms, and then sequentially adding CESD (Model B), and then the covariates in Model C1 (see Table 4). The final model including self-reported symptoms as well as the covariates and all paths were fully unconstrained (see Table 5). Hu and Bentler’s (1999) suggested fit indices all indicated a good fit for the final full, unconstrained model, with CFI = 1, TLI = 0.992, SRMR = 0.013, and RMSEA = 0.020 with 90% CI [0.00, 0.063]; the χ²(9) was 10.3 (p = .326). Higher levels of inattention symptoms significantly predicted lower performance on Immediate Recall (B = −0.18, p = .001; all coefficients reported are standardized estimates) for the middle age group only, while higher levels of hyperactivity/impulsivity symptoms were significantly associated with lower performance on Delayed Recall (B = -0.11, p = .018). There were no significant associations between self-reported ADHD symptoms and cognitive outcomes for the young-old age band.
Fit Indices for Nested Multigroup Path Models.
Note. IA = inattention symptoms; HI = hyperactivity/impulsivity symptoms; CESD = Center for Epidemiology Studies Scale for Depression.
Final Unconstrained and Constrained Multigroup Path Models.
Note. Coefficients marked with c are constrained to be equal across groups. All models control for age, gender, race, ethnicity, education, and self-rated health. β = standardized path coefficient; SE = standard error; CESD = Center for Epidemiology Studies Scale for Depression.
In the next stage, Model C2 was run with all paths constrained to be equal across the groups. The fit indices suggested the model was a good fit with CFI = 0.998, TLI = 0.996, SRMR = 0.034, and RMSEA = 0.014 with 90% CI [0.00, 0.032]; the χ²(91) was 97.4 (p = .304). Higher levels of inattention symptoms were significantly associated with lower scores on Serial 7s (B = −0.07, p = .039) and Immediate Recall (B = −0.11, p = .002). Since this was a constrained model, these associations were equivalent across the age bands (Table 5).
Successive models released constraints on paths between inattention symptoms and cognitive outcomes (Model D1), on hyperactivity/impulsivity (Model D2) and then both self-reported ADHD symptoms (Model E) and then freeing the paths for CESD symptoms (Model F) and cognitive performance as well. Releasing path constraints within these models (Model D1 to Model F) yielded similar fit indices but slightly larger AIC and BIC values for each model (Table 4). As such, the fully constrained model, Model C2, was retained as the best-fitting and most parsimonious model, suggesting that associations between self-reported inattention symptoms and cognitive outcomes do not differ across age bands (Figure 1).
Fully constrained multigroup path model (C2) with standardized coefficients.
Discussion
Although ADHD may influence cognitive functioning in later life (Callahan et al., 2021; Mendonca et al., 2021), few studies have looked at the association between ADHD symptoms and cognitive outcomes in middle and older adults. In this study, we observed the association between self-reported ADHD symptoms and performance on cognitive tasks in a large, national sample of individuals between the ages of 55 and 84 residing in the US. Participants were divided into three age bands: middle aged for individuals aged 55 to 64 years, young-old for individuals aged 65 to 74 years, and middle-old for individuals aged 75 to 84 years to assess the relationship between symptoms and cognition among different ages.
The first hypothesis was that inattention symptoms would be more closely related to cognitive outcomes than hyperactive/impulsive symptoms. This was supported by the results of the current analyses, where in the final, fully constrained model (Model C2), inattention symptoms were more linked than hyperactivity/impulsivity to the cognitive outcomes, with associations being strongest for the Immediate Recall as well Serial 7s measures (with weaker, non-significant associations for Number Series). This is not entirely unexpected, as links between adolescents and adults diagnosed with ADHD and poorer verbal recall performance have been reported previously as well (Quinlan & Brown, 2003). In an earlier meta-analysis, Skodzik et al. (2013) suggested that verbal memory deficits in adults with ADHD were linked to issues with initial encoding during memory acquisition. Similarly, given that Serial 7s is linked to working memory, it is not surprising that higher inattention symptoms predicted lower performance on the Serial 7s. Interestingly, although hyperactivity/impulsivity symptoms were initially significantly linked to Delayed Recall performance for the middle-old age band in the fully unconstrained model (Model C1), this association was not significant in the final fully constrained model (Model C2).
The second hypothesis was that the association between ADHD symptoms and cognitive outcomes would vary by age, with stronger associations for middle-aged adults versus young old or middle old adults. This was not supported by the results of the multigroup path model where the constrained model was the best fitting and most parsimonious model—suggesting that there were no meaningful differences in the associations between inattention symptoms and cognitive performance across age bands. One possibility that may help explain this is that as individuals get older, normative age-related cognitive decline may occur, which may outweigh the effects of ADHD symptoms on verbal learning and memory). Studies have also found previously that Immediate Recall may be more sensitive to age-related declines in memory as well as those that occur in dementia (Degenszajn et al., 2001; Robinson-Whelen & Storandt, 1992). It is important to note that people with a diagnosis of dementia or Alzheimer’s were excluded from this analysis, so findings can only be generalized to individuals without identified cognitive decline. Additionally, there may be a slight survival bias, since individuals with more severe ADHD may have experienced greater impairments in their life course across a broad array of functions and activities and subsequently may have higher likelihood of earlier mortality (O’Nions et al., 2025).
In a departure from previously reported studies on ADHD symptoms and cognitive outcomes, this study found the association between depressive symptoms and cognition to be non-significant in the final model (see Supplementary Tables). Previously, Das et al. (2014) reported that although higher ADHD symptoms were linked to lower cognitive performance in middle aged adults (48–52 years), for older adults (68–74 years) this association seemed to be mediated through the link between depressive symptoms and cognition. Additionally, their multi-group structural equation model indicated limited positive associations between the latent factor for hyperactivity/impulsivity and Delayed Recall for middle-aged participants such that a greater number of hyperactive/impulsive symptoms resulted in poorer performance on Delayed Recall. Meanwhile, Immediate Recall was not significantly associated with either inattention or hyperactivity/impulsivity in both cohorts of adults. Our study found significant negative associations between inattention symptoms and Immediate Recall in the fully constrained model; higher hyperactivity/impulsivity symptoms also significantly predicted lower Delayed Recall performance, but this only occurred for the middle-old group in the unconstrained model and was not significant for the final, fully constrained model.
Similarly, Semejin et al. (2015) corroborated this with their study on older adults where they found that the association between attention or working memory and ADHD symptom severity disappeared when depressive symptoms were accounted for in the model. However, this was not the case with our study: paths between depressive symptoms as measured by CESD were not significantly associated with cognitive performance in the final model and the association between inattention symptoms and cognitive measures remained significant even when CESD was added to the model. In a later study within a middle aged sample from the same Australian population study examined in Das et al. (2014; Das et al., 2015; n = 2,091, age range = 47–54 years), also reported that inattention and hyperactivity symptoms were associated with cognitive outcomes independent of depressive or anxiety symptoms. However, in their study, neither Immediate nor Delayed Recall were reported to be significantly associated with either ADHD symptom subscale. While links between depressive symptomology and cognition were not the focus of our study, future research should explore why there are differences in these associations across these larger population studies.
Clinical Implications
Our study identified close links between ADHD symptoms and immediate verbal recall as well as Serial 7 performance in later life in a national sample of middle-aged and older adults in the US. This has significant clinical implications for not only the diagnosis and treatment of ADHD in adults, but also the interaction between ADHD and comorbid disorders, including anxiety, depression, and perhaps even MCI. Although we found no age-related differences in ADHD symptoms and cognitive performance, late middle age may still be a viable time period for interventions to address ADHD inattention symptoms, which is less frequently discussed relative to earlier periods of life in the current literature on ADHD (see (May et al. [2023] for a review). Given that the closest links identified here are between ADHD and auditory memory measures, it is possible that interventions focusing on memory and learning strategies may be most relevant for this age range. In addition to behavioral and cognitive therapies recommended for adults with ADHD (Nimmo-Smith et al., 2020; Torgersen et al., 2016; Young et al., 2020), memory training methods such as the method of loci may be a helpful strategy to boost memory performance (Gross et al., 2014; Sandberg et al., 2021). Lifestyle and health behavior changes such as increasing physical activity may also serve as a neuroprotective factor as individuals with ADHD symptoms age, given links between higher levels of physical activity and less inattention and better cognitive performance (Tucker et al., 2025; Weuve et al., 2004), and evidence that physical activity may help improve inattention symptoms for children and adolescents (Chang et al., 2010; Gapin et al., 2011; Xie et al., 2021). Additionally, results highlight the importance of promoting evidence-based diagnostic procedures, given that cognitive impairments can better account for decline in older adult samples, and that it may be difficult to distinguish between ADHD and MCI related impairment in middle-old and old-old older adults.
Unlike studies on younger samples, studies on cognitive performance of older adults rarely assess current ADHD symptoms or even prior diagnoses. And so, researchers interested in cognitive outcomes in later life should consider adding ADHD symptoms to their measures when testing cognition for older adults. Additionally, further education on ADHD symptoms, comorbid disorders, and their interactive effects on cognitive functioning may help clinicians and physicians when assessing and treating individuals as they age. As Goodman et al. (2024) suggest, diagnosing ADHD in middle age can still assist individuals in accessing much-needed treatments that will improve their quality of life and diminish their ADHD-related impairments, both functionally and cognitively. Given that family history and the genetic nature of ADHD is an important factor, clinicians may consider asking questions about both when assessing older adults and trying to distinguish between ADHD or MCI presentations. Owing to the lifelong nature of ADHD, older adults may have been experiencing cognitive difficulties since their childhood or adolescence; younger family members having an ADHD diagnosis may also be an indicator of a genetic predisposition toward ADHD (Goodman et al., 2016). It is important to note that MCI is typically assessed using neuropsychological performance based tests, whereas diagnosis of ADHD generally utilizes behavioral rating scales and interviews (Dobrosavljevic et al., 2023). It would be optimal to have both approaches be administered for diagnosis of ADHD in older adults; future studies should examine the concordance between these two methods in this population.
Limitations
This study has several limitations, chief of which is that only a random subset of the HRS sample was asked to complete the ADHD self-report measure. Additionally, given that this study was conducted using the Experimental Module of the HRS, there may be self-selection biases. Although self-reported ADHD symptoms have been found to be correlated with reports from spouses or other informants in adults (P. Murphy & Schachar, 2000), this methodology comes with its own set of self-response biases, as well as concerns about performance and symptom validity failures. Fuermaier et al. (2024) reported a failure rate of between 8% and 34% for the Conners’ Adult ADHD Rating Scale in an adult sample. If a similar rate exists for the sample in this study, the results may certainly be biased. This makes it all the more important to note that there are currently no clinical measures (e.g., diagnostic interviews) or assessment of the functional impairment or criterion, age of onset criteria available regarding ADHD in the HRS that can be used as a criteria for diagnosis. This makes it unclear exactly how many participants in the sample may actually meet the criteria for a diagnosis of ADHD. Inclusion of such measures would greatly help future studies on ADHD in later life be more precise in examining how ADHD symptoms may affect older adults. Similarly, since no formal diagnostic measures were included in HRS, we are unable to rule out if other comorbid conditions, such as anxiety or depressive disorders, may explain the elevated ADHD symptoms observed in the current sample.
While the HRS does offer a set of cognitive tasks to most respondents, these tasks are still limited in the domains of cognition that they represent. A broader set of cognitive tasks, the Harmonized Cognitive Assessment Protocol (HCAP) was also conducted in 2016—the same wave from which data in this study was collected—but the overlap between respondents to the Experimental Module containing the ASRS and HCAP resulted in sample sizes that were too low for stratified analyses (n = 277–291). Similarly, the lower end of the age range for this study had to be restricted to late middle age, from 55 to 64 years, due to the low sample size for participants younger than 50 enrolled in HRS. However, examining data from participants below the age of 50 years would have further helped to elucidate the age-related changes in ADHD symptoms along with their link to cognitive functioning (especially middle age onwards) and remains a promising focus for future work examining these links across the life course.
We note that we do not include race or ethnicity variables in the models for our study, following the APA guidelines that suggest avoiding the use of dichotomized race and ethnicity variables to minimize misleading interpretations (Wang & Leath, 2023). However, we do note that the inclusion of these race and ethnicity variables in our models does not change the pattern of results discussed here. Future studies may wish to further examine the influence of a variety of contextual factors when exploring ADHD symptomology in older adults.
Conclusion
This study emphasizes the importance of examining links between ADHD symptoms, particularly inattentive symptoms and cognitive status in later life. Looking forward, larger samples are needed across a broader time frame to better examine the influence of aging on ADHD symptoms. Additionally, mechanistic data that focuses on the neural and physiological differences between ADHD in younger as well as older adult samples may be helpful in addressing salient questions about the nature of ADHD throughout the life course, such as questions regarding differences between childhood onset and late or adult onset of ADHD (Faraone & Biederman, 2016; Pagán et al., 2023; Rivas-Vazquez et al., 2023; Sibley et al., 2020), and whether adult onset ADHD exists or is simply a reflection of late-identified ADHD (Faraone & Biederman, 2016; Mitchell et al., 2021; Pagán et al., 2023; Sibley et al., 2018). Given differences in observed effects between our study and Das et al.’s (2014) study (in an Australian sample), more research needs to be conducted on cultural differences in how ADHD is measured, evaluated, and diagnosed to understand the impact of ADHD in different populations of older adults throughout the world.
Supplemental Material
sj-docx-1-jad-10.1177_10870547251394080 – Supplemental material for Self-Reported ADHD Symptoms and Cognitive Performance in a National Sample of US Older Adults
Supplemental material, sj-docx-1-jad-10.1177_10870547251394080 for Self-Reported ADHD Symptoms and Cognitive Performance in a National Sample of US Older Adults by Marrium Mansoor, Rosanna Breaux, Tae-Ho Lee and Benjamin Katz in Journal of Attention Disorders
Footnotes
Ethical Considerations
Not applicable (the data used in this study is secondary data).
Consent to Participate
Not applicable (the data used in this study is secondary data).
Funding
The authors disclose receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the National Institute of Health [R01AG075000/AG/NIA].
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data Availability Statement
Not applicable (the data used in this study is publicly available on the Health and Retirement Study website).
Supplemental Material
Supplemental material for this article is available online.
Author Biographies
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
