INTRODUCTION

Acetylcholine plays an important role in cognition, including learning and memory (Huang et al., 2022). Cholinergic neurons, which project widely to the cerebral cortex, arise in the basal forebrain (Mesulam, 2004). The hippocampus, a key medial temporal lobe structure implicated in memory (Dickerson & Eichenbaum, 2010), receives inputs from the basal forebrain cholinergic system (Mesulam, 2004). Cholinergic system dysfunction is implicated in aging-related cognitive decline and disorders like Alzheimer's disease (AD) (Bartus et al., 1982; Hampel et al., 2018). Among the evidence for cholinergic system involvement in AD is atrophy of basal forebrain cholinergic neurons in mild cognitive impairment and AD (Grothe et al., 2012; Teipel et al., 2011).

Based on the cholinergic hypothesis of AD, medications used in the symptomatic management of AD include cholinesterase inhibitors (e.g., donepezil, galantamine, and rivastigmine) (Marucci et al., 2021). These medications target the cholinergic system, preventing the enzymatic breakdown of acetylcholine (Lane et al., 2005). Unlike cholinesterase inhibitors, which allow for the increased availability of acetylcholine, anticholinergic (AC) medications block the action of the neurotransmitter at cholinergic receptors (i.e., muscarinic and nicotinic receptors) present in the central and peripheral nervous systems. AC medications are used to manage or treat an array of medical (e.g., allergies, asthma, epilepsy, and Parkinson's disease) and psychiatric (e.g., anxiety, bipolar disorder, depression, schizophrenia spectrum disorders) conditions. Many commonly used prescription (e.g., antidepressants) and over-the-counter (e.g., antihistamines) medicines have AC effects (Gerretsen & Pollock, 2011).

In aging, AC medication use is associated with decrements in cognitive functioning. For example, greater AC medication use has been associated with cognitive impairment as measured by the Mini Mental State Examination (Sargent et al., 2020) and poorer cognitive functioning as assessed via performance-based measures of memory and executive functioning (Risacher et al., 2016). AC medication use is associated with both acute and chronic cognitive impairment (Boustani et al., 2008; Fox et al., 2011), and greater AC exposure is associated with elevated risk for delirium (Egberts et al., 2021), dementia (Cai et al., 2013; Mur et al., 2022), mild cognitive impairment (Cai et al., 2013), and AD (Gray et al., 2015). Collectively, these associations of AC medications with cognitive problems are referred to as anticholinergic cognitive burden (ACB).

Because of high rates of co-occurring medical (e.g., seizure disorders, gastrointestinal) (Hand et al., 2020; Weir et al., 2021) and psychiatric (e.g., anxiety, depression) (Lai et al., 2019) conditions across the lifespan, autistic individuals are likely to have elevated AC exposure relative to non-autistic individuals, including exposure earlier in the lifespan. Autistic individuals are also likely to have increased AC exposure as they age. A study of Medicare enrollees aged 65 and older found that after adjusting for sex, age, ethno-racial identity, and rurality, relative to non-autistic adults, autistic adults showed significantly greater odds of having metabolic (e.g., diabetes and thyroid disorders), respiratory (including asthma), circulatory (including hypertension and heart disease), neurological (including epilepsy and Parkinson's disease), gastrointestinal (including upper and lower gastrointestinal disorders, gastroenteritis, and constipation), and psychiatric conditions (e.g., schizophrenia and psychotic disorders, mood disorders, and anxiety disorders) (Hand et al., 2020). Importantly, these older age autistic adults also showed a higher prevalence of cognitive disorders, including delirium and dementia, compared to non-autistic adults (Hand et al., 2020). Additionally, converging evidence suggests that middle-aged and older autistic adults are at increased risk of (a) having early onset AD, delirium, and other cognitive disorders (Vivanti et al., 2021) and (b) self-reported experiences of cognitive decline (Klein et al., 2023).

Cross-sectional work using neuropsychological testing to investigate aging-related cognitive trajectories in autistic and non-autistic adults has presented discrepant results. Some show patterns of parallel cognitive aging (Torenvliet et al., 2022) between autistic and non-autistic adults. Other studies have shown accelerated cognitive aging (Geurts & Vissers, 2012), suggesting vulnerabilities to aging-related cognitive decline. Yet others show preserved cognitive aging, suggesting possible resilience to aging impacts in autistic adults relative to non-autistic adults (Lever & Geurts, 2016).

An emerging body of longitudinal work investigating cognitive aging in autistic versus non-autistic adults has also presented differing patterns of results. Some longitudinal work provides evidence for parallel cognitive aging across a range of measures, including verbal and visual recall, working memory, and prospective memory, among autistic and non-autistic adults, suggesting that autistic adults neither show increased nor decreased risk for age-related cognitive decline relative to non-autistic adults (Torenvliet et al., 2023). Other longitudinal studies, however, suggest that a subset of middle-aged and older autistic adults evidence steeper age-related decline (relative to non-autistic adults) in long-term visual memory (Walsh et al., 2022) and short-term verbal memory (Pagni et al., 2022). These differences in findings may be related to differences in sample size or methods across the studies. These differences, however, also point to the possibility that other factors could distinguish a subset of autistic adults at increased risk of experiencing cognitive decline.

While most of the extant work on autistic adult samples has utilized performance-based measures of memory and cognition, another potentially effective approach is self-ratings of memory complaints and self-reported experiences of cognitive decline. In non-autistic samples of older adults, self-reported cognitive decline is correlated with performance-based assessments of cognitive decline (Passler et al., 2021) and is associated with MCI and dementia (Kasai et al., 2021). Furthermore, subjective memory complaints are associated with the development of MCI and dementia (Jessen et al., 2014; Mitchell et al., 2014). Though studies to date have included primarily non-autistic adults, this research suggests that self-report assessments of memory problems and of changes in cognition can be used to screen for and identify individuals who may be at risk for disorders of cognition such as MCI, dementia, and AD.

Taken together, while there is evidence for overall high rates of both general polypharmacy and psychotropic polypharmacy in autistic persons in childhood (Mandell et al., 2008; Ritter et al., 2021), adolescence (Esbensen et al., 2009; Frazier et al., 2011), and adulthood (Esbensen et al., 2009; Jobski et al., 2017; Lake et al., 2012; Vohra et al., 2016), to our knowledge, no research has characterized AC exposure in autistic individuals or examined associations between AC exposure and cognitive outcomes. Thus, the current study aimed to (1) characterize AC medication usage in a sample of middle-aged and older autistic adults (Study 1A); (2) examine associations between this AC medication use and concurrent self-reported memory problems (Study 1B); and (3) examine associations between AC medication use and self-reported cognitive decline at follow-up 2 years later (Study 2). We hypothesized that AC medication use would be: (1) common among autistic middle-aged and older adults (Study 1A); (2) positively correlated with self-reported memory problems (Study 1B); and (3) associated with self-reported cognitive decline at 2-year follow-up (Study 2).

METHODS

MEASURES

RESULTS

Study 1A results: Characterization of AC medication use and ACB

Table 1 presents the characteristics of Study 1A, Study 1B, and Study 2 participants. Study 1A revealed a high prevalence of AC medication use in autistic adults, with 62.9% reporting at least one AC medication according to the CALS and 48.2% according to the ACB Scale. The total range of ACB for participants was 0 to 11 with a mean of 1.8 (SD = 2.1) on the CALS and 0 to 10 with a mean of 1.2 (SD = 1.8) on the ACB Scale. The proportion of autistic adults reporting ACB ≥3 was 26.5% and 20.5% on the CALS and ACB Scale, respectively.

For the purposes of better understanding the types of AC medications reported by autistic adults in the Study 1A sample, Figure 1 presents the most common types of AC medications reported as indicated via the CALS (Figure 1a) and ACB Scale (Figure 1b). Psychotropic medications (particularly antidepressants, antipsychotics, and benzodiazepines) and antihistamines were the most commonly reported AC medication classes in the sample. The antidepressant medications reported included selective serotonin reuptake inhibitors (SSRIs; e.g., paroxetine), tricyclic antidepressants (TCAs; e.g., nortriptyline), and serotonin noradrenaline reuptake inhibitors (SNRIs; e.g., venlafaxine), as well as the serotonin modulator trazodone. Included among the antipsychotic medications reported were atypical (e.g., aripiprazole) as well as conventional (e.g., thorazine) antipsychotics. Reported antihistamines included medications such as diphenhydramine, hydroxyzine, and loratadine, and reported benzodiazepines included clonazepam, diazepam, and lorazepam. Also commonly reported were biguanides (e.g., metformin; note that biguanides are included only in the CALS and not the ACB Scale), beta-blockers (e.g., atenolol, metoprolol), opioids (e.g., morphine, oxycodone), H2-blockers (e.g., ranitidine, cimetidine), and skeletal muscle relaxants (e.g., baclofen, cyclobenzaprine). A list of AC medications reported in the sample can be found in the Data S1 (see Table S2).

DISCUSSION

That autistic children and adolescents grow up to be adults is a trivial statement, yet to date, research on autism has predominantly examined child and adolescent samples, while studies of adulthood, particularly middle and older adulthood, are limited. Thus, as a field, we know little about aging and cognition in autistic adults (Mason et al., 2022). To help address this critical knowledge gap, the current study examined one factor—AC medication exposure—that may particularly impact autistic adults, their cognition, and their aging-related cognitive trajectories.

This is the first study to characterize AC medication use and examine associations between these medications and self-reported cognitive complaints in autistic adults. We hypothesized that AC medication use would be common among middle-aged and older autistic adults. This hypothesis was confirmed: based on the ACB Scale and CALS, 48.2% and 62.9% of autistic adults in the current study, respectively, were taking at least one AC medication. Further, based on the ACB Scale and the CALS, 20.5%–26.5%, respectively, of autistic adults reported clinically meaningful levels of AC medication use (ACB score ≥3).

In the broader aging literature, the prevalence of AC medication use varies, depending on the scale(s) used to quantify AC potency, the country in which participants reside, and specific characteristics of the samples investigated (e.g., whether the study is of community-dwelling older adults, hospitalized older adults, or older adults residing in a memory care facility; or whether participants do or do not have a diagnosis of an aging-related cognitive disorder such as dementia). Additionally, differences in rates of AC medication use across studies are likely attributable, in part, to the time period during which data were collected, as AC medication use among older adults has increased over the past several decades (Grossi et al., 2020).

Estimates in a 2009 study, using an earlier version of the ACB Scale used in the current study (Boustani et al., 2008), found that ~20%–50% of U.S. adults >65 years old took at least one AC medication (Campbell et al., 2009). A UK population-based study of older adults aged ≥64 years indicated that 49.6% of the sample between 1990 and 1993 and 64.3% of the sample between 2008 and 2011 were taking at least one medication with anticholinergic effects (Grossi et al., 2020). Ramos et al. (2022) characterized AC medication use in a sample of 512 Spanish adults (aged 50–96 years) with subjective memory complaints, who were grouped according to those with and without cognitive decline, as determined by cognitive testing. ACB was characterized using the CALS and an earlier version of the ACB Scale (Boustani et al., 2008) than that used in the present study. The mean total ACB score based on the CALS was 2.1 (SD = 1.9) in the group with cognitive decline, and 1.6 (SD = 1.7) in the group without cognitive decline, and ACB ≥3 was found in 37.20% of the cognitive decline group and in 24.71% of the group without cognitive decline. For comparison, in the current study, the mean total ACB score based on the CALS was 1.8 (SD = 2.2), and ACB ≥3 was found in 26.5% of autistic adults. Importantly, the participants in the current study were younger on average (M = 52.0, SD = 9.1) than those in either the group with (M = 74.7, SD = 7.9) or the group without cognitive decline (M = 68.1, SD = 9.1) investigated by Ramos and colleagues.

In addition to finding that AC medication use is prevalent among autistic adults, we further found that higher levels of AC medication use were associated with concurrent self-reported prospective and retrospective memory problems. Based on the CALS, which provides a more updated, comprehensive list of AC medications relative to the ACB Scale, AC medication use was also associated with self-reported cognitive decline at 2-year follow-up.

Subjective memory impairment has been suggested to be the earliest clinical manifestation of dementia (Jessen et al., 2006), and among adults aged ≥65 years, subjective memory complaints predict future dementia at 4-year follow-up (Waldorff et al., 2012). Self-reported cognitive decline is correlated with performance-based metrics of cognitive decline (i.e., declines in word learning) (Passler et al., 2021) and is associated with disorders of cognition, including MCI and dementia (Kasai et al., 2021). Thus, future studies should examine whether autistic adults with subjective memory complaints and/or self-reported cognitive decline are at elevated risk for the development of disorders of cognition, including MCI, dementia, and AD, and whether such a potential elevated risk is associated with AC medication use.

AC medication use has been associated with subjective cognitive decline among community-dwelling adults (aged 47–91 years; M = 65.2 (8.6) at risk for AD (based on APOE ε4 carrier status and/or family history of dementia) (Margolis et al., 2021). Among adults ≥65 years with subjective cognitive decline, high ACB (≥3) is associated with a more than four-fold increased risk for dementia development (Naharci et al., 2017). Thus, the results reported here indicate that studies of aging and cognition in autism should take into consideration the potential impacts of AC medication use on self-report or performance-based measures of cognition.

There was a single model that was significant for Study 2: when characterizing ACB based on the CALS, the total ACB score was associated with the self-reported measure of cognitive decline (AD8) cut-off of ≥1 after controlling for age and sex assigned at birth. In contrast, for Study 1B, all models reached statistical significance. The sample of Study 2 (N = 197) is smaller relative to the sample of Study 1B (N = 382). Thus, Study 2 may have been underpowered to detect effects. No a priori power analysis was computed, and we do not report a post hoc power analysis given known problems with post-hoc power calculations (Althouse, 2021; Goodman & Berlin, 1994; Hoenig & Heisey, 2001). Thus, the examination of AC medication and associations with retrospective self-reported cognitive decline in a larger sample of autistic adults is an important direction for future research.

Importantly, extant studies on AC medication use and its associations with cognitive functioning in non-autistic individuals typically characterize samples older than the sample of autistic adults reported here. The mean age of autistic adults in the current study was 52.0 years (SD = 9.1 years); in contrast, most research has characterized AC medication use and its cognitive effects in samples of participants aged ≥65 years. In comparing participants in the current Study 1A sample aged <65 (n = 368) with those ≥65 years (n = 47), there were no significant differences in the two age groups for AC medication prevalence on any of the metrics used to characterize the full Study 1A sample (N = 415) (Table S6). Thus, the substantial levels of AC medication use and its associations with subjective cognitive complaints in this comparatively younger sample indicate that consideration of AC medication use—including consideration of use at younger ages than typically examined—may be important in this population, which may be at elevated risk for AC medication exposure (e.g., antipsychotic medication exposure (Mandell et al., 2008)) earlier in the lifespan.

The strengths of the current study include the characterization of AC medication use in a large sample of middle-aged and older autistic adults. AC medication characterization was robust, using two scales. One of the two scales was also relatively more conservative: the ACB Scale, developed in 2012, does not include many AC medications reported in the sample given that these medications were not yet developed or in use at the time of the scale's publication.

The results reported here must be viewed in light of the limitations of the current study. The study's sample does not include autistic persons with co-occurring intellectual disabilities (ID). Autistic individuals with co-occurring ID have high rates of polypharmacy, including psychotropic polypharmacy rates (Esbensen et al., 2009; Tsakanikos et al., 2007; Vohra et al., 2016), which suggests that exposure to AC medications among these autistic persons may be even higher than rates reported here. Understanding the prevalence of AC medication and its potential impacts on cognition among these individuals is a needed area of research.

Another limitation of the current study is that the data reported were self-report, including self-reported cognitive decline and memory problems. Future studies should conduct medical record reviews and consider examining AC/ACB effects on performance-based measures of cognition (e.g., working memory, attention, verbal fluency) in addition to subjective cognitive complaints. Given that current medication use was captured via self-report, it is possible that participants did not list all of the medications they were taking. Moderating this concern, however, was our use of conservative inclusionary/exclusionary criteria, including only those participants whose reported medications were complete and unambiguous, while excluding those individuals for whom medication use was unclear, either because they indicated they were taking too many medications to list or because the medications listed were not named (e.g., participants indicated they were taking antihistamines or antidepressants but did not provide either the generic or brand name of the medication). Thus, it is likely that, if anything, the prevalence of AC medication use is underestimated in the current sample, indicating that the risk could be even greater. Also moderating concerns regarding the use of self-report medication data, the high rates of psychotropic medications in the current study are consistent with those in other studies of autistic adults (Lake et al., 2012; Vohra et al., 2016), including studies using large databases (Houghton et al., 2017; Vohra et al., 2016).

In the current study, cognitive decline was operationalized via a self-report retrospective measure collected at a single timepoint. Longitudinal studies that examine change over time (i.e., change across two or more time points) in self-reported memory and other cognitive challenges are needed. The present study did not specifically query additional details of participants' current medication use, such as the dosage and frequency of each medication, adherence to each medication, or the length of time participants had been using each medication. Future studies examining frequency, dosage, adherence, and medication history would allow for analyses that also examine possible dose-dependent effects and/or cumulative AC medication exposure as these relate to cognitive outcomes.

AC medications may impact the transition from normative trajectories of aging to mild cognitive impairment (MCI) (Campbell et al., 2018), and among older persons (>70 years) with normal cognitive functioning, those taking more potent anticholinergic medications (medications with an ACB score ≥2) are at increased risk for the development of various markers of cognitive impairment at follow-up 6 years later (Campbell et al., 2010). Promisingly, however, deprescribing AC medications may reduce adverse outcomes associated with AC medication use in older adults (Carrière et al., 2009; Lupu et al., 2021), although more research is needed concerning how deprescribing impacts cognitive functioning and how best to implement deprescribing (Soiza et al., 2021).

Given the results reported here, alongside emerging evidence that middle-aged and older autistic adults demonstrate elevated rates of self-reported cognitive decline (Klein et al., 2023) and delirium or dementia (Hand et al., 2020), including early-onset dementia and other cognitive disorders (Vivanti et al., 2021), AC medication use may be an important modifiable risk factor in aging-related outcomes among autistic adults. To promote healthy aging, autistic adults may benefit from prescribers' careful attention to and consideration of recommendations concerning prescribing practices as these relate to AC medications (Institute of Medicine, 2015). These practices include minimizing the use of certain AC medications in older persons, where possible, in order to reduce the risks conferred by these medications (2019 American Geriatrics Society Beers Criteria® Update Expert Panel, 2019).

Although we did not ask for additional details concerning participants' current medication use, some participants provided this information unprompted, including the reason they were taking each medicine they reported. Many reported taking some of the most commonly found AC medications in the sample (i.e., antihistamines, benzodiazepines, and antipsychotic medications) for insomnia. Given that autistic adults experience greater sleep-related problems compared to non-autistic adults (Morgan et al., 2020), and among middle-aged and older autistic adults, sleep difficulties are common and associated with increased anxious symptomatology (Charlton et al., 2023), it may be particularly important to understand why autistic adults are prescribed or are taking AC medications in order to identify alternative interventions where possible. Additionally, many of the high-potency AC medications frequently reported in the sample are over-the-counter medications (e.g., diphenhydramine and doxylamine succinate). Thus, providers may wish to provide guidance concerning the potential harms of common over-the-counter medications so that autistic adults are able to make informed decisions about their use of these medicines (Institute of Medicine, 2015).

Cognitive effects represent one type of adverse effect associated with AC medication exposure. There are other adverse effects of AC medications in aging persons, however, including elevated risk for falls (Stewart et al., 2021; Wong et al., 2023) and mortality (Ali et al., 2020; Graves-Morris et al., 2020; Ruxton et al., 2015). Thus, in addition to probing the cognitive-related effects of AC medication use, future work should also investigate other health-related adverse effects associated with AC medications in autistic persons. The identification of such effects and ways to minimize them may hold promise for improving aging-related quality of life across cognitive and physical health domains in autistic adults.

CONCLUSIONS

The current study revealed that a large proportion of middle-age and older autistic adults take at least one AC medication, and a substantial minority take clinically-relevant levels of AC medication. Use of these medications was in turn associated with both greater contemporaneous subjective memory complaints and self-reported cognitive decline based on retrospective reporting. This study provides preliminary evidence for an association between relatively high AC medication usage and cognitive challenges in autistic adults.

FUNDING INFORMATION

This research was supported by start-up funds from the George Washington University to G.L.W. Support for writing this manuscript was provided by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (under grant R21HD106164) to G.L.W. Support for writing this manuscript was provided by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (under grants R21HD100997; R21HD106164) to N.R.L. Support for writing this manuscript was provided by the National Institute of Mental Health (under grant K01MH129622) to G.A.M.

ETHICS STATEMENT

The authors assert that all procedures contributing to this work adhere to the ethical standards of the relevant national and institutional committees and are in accordance with the Declaration of Helsinki.