Many older adults take at least one prescription medication with anticholinergic (ACH) activity, which can impact the central nervous system and can lead to cognitive decline and impairment especially in an aging population susceptible to cognitive changes. We examined this relationship between ACH burden and cognitive function in middle-aged and older adults. We further determined if increased activity levels mitigated the relationships between ACH burden and cognition.
Data from The Reasons for Geographic and Racial Differences in Stroke project were used. We included 20,575 adults aged ≥45 years with longitudinal cognitive testing. The anticholinergic cognitive burden (ACB) scale was used to assess for ACH use and overall burden. Cognitive data included an overall composite score, a memory, and verbal fluency composites. Mixed effects models were conducted to determine if cognitive function worsened over time for participants with higher ACB (>3) scores. The full model adjusted for age, sex, race, education, diabetes, hypertension, cardiovascular disease, congestive heart failure, and dyslipidemia, self-reported physical activity (PA) and depressive symptoms.
A significant relationship between ACH burden and composite cognitive scores was found (p = <0.001), with those with higher ACB showing more rapid cognitive decline over time. There was an effect of age for participants with higher ACB (>3) scores and ACB as a continuous variable. Baseline PA level was associated with less cognitive decline over time and this effect was greater in older cohorts.
We observed an effect of ACHs on cognition in adults ≥45 years old that worsened with age. ACH users showed more cognitive effects, whereas PA emerged as a possible mitigating factor.
- Anticholinergics (ACHs) are associated with cognitive effects in adults ≥45 years of age.
- Cognitive effects are worsened with increasing age.
- Physical activity (PA) plays a mitigating factor in these drug-induced cognitive changes.
Why does this paper matter?
Findings highlight the association between ACH drugs and cognition as well as the likely role of PA in mitigating the relationship between these medications and cognitive decline in older adults. Our findings may help guide future clinical practice and prevention efforts.
Most adults between 62 and 85 years of age take multiple prescription medications,1 and 20%–50% take at least one drug with anticholinergic (ACH) properties, which are commonly prescribed for the treatment of urinary incontinence, Parkinson's disease, and neuropsychiatric conditions.2
The anticholinergic cognitive burden (ACB) scale3 is widely used to quantify ACH effects. Studies showed relationships between ACB scores and cognitive function,4 and cumulative use of ACH's was associated with increased risk for dementia.5 Studies showed an effect of physical activity (PA) on cognition and prevention of dementia.6 However, whether ACH effects on cognition vary according to age, ACB, and PA levels remains unknown.
We examined the relationship between scores on the ACB scale and cognitive function, and whether this relationship varied according to age, and if PA mitigated this relationship. We hypothesized that higher ACB would be associated with poorer cognitive performance and that this relationship would strengthen over time. We also hypothesized that the cognition of older adults would be particularly sensitive to ACHs, but that increased PA would modify this effect.
Participants were recruited from the Reasons for Geographic and Racial Differences in Stroke (REGARDS) study (previously described in detail).7 At baseline, a combination of telephone interviews and an in-home visits collected data using computer-assisted telephone interviewing including medical history, demographic information, medication inventory (prescription and non-prescription medications taken within the last 2 weeks) and self-reported PA levels. Similar assessments were performed at follow-up visits, a median of 9.4-years later. Cognitive assessments and evaluations for suspected stroke were conducted biannually via telephone interview. The current study included a subset of 20,575 REGARDS participants (see Figure S1) who met the following criteria: >45 years old, have cognitive data without evidence of self-reported stroke or prevalent cognitive impairment as assessed by the Six-Item Cognitive Screener.8 Participants provided written consent, and approval by all participating institutional review boards was secured.
Telephone-based cognitive assessments have been validated with older adults.9 Trained interviewers7 administered the test, which included, verbal fluency (Animal Naming and Phonemic/Letter Fluency) and learning and memory (Consortium to Establish a Registry for Alzheimer's Disease [CERAD]) Word List Memory test.10
Cognitive composite scores
A composite index was created by domain of cognitive function (fluency and memory), and overall. Methods to create the indices have been described previously.11, 12 Briefly, the fluency score averaged the z-scores from animal fluency and letter “F” assessments. The memory score was based on the aggregated z-scores for the immediate recall score and the delayed recall score from CERAD Word List Memory. The overall composite was derived by averaging z-scores for the fluency and memory scores.
Anticholinergic cognitive burden
ACH use and burden were assessed using the ACB scale.3 Based on previous publications,3 medications included in the ACB scale were assigned a score of 1 (possible ACH effect), or 2 or 3 (definite ACH effect). Individual medication scores were then summed for an ACB total score. Scores ≥3 were considered clinically significant according to the ACB scale.3
A self-reported PA measure was collected at baseline; no activity, 1–3 sessions; or ≥4 sessions per week. PA was defined by responses to the question “How many times per week do you engage in intense physical activity, enough to work up a sweat?”. Validity of this widely used assessment of PA is well-established.13-15
Demographic and health information (Table S1) were collected at baseline and included age, race, sex, depressive symptoms, diabetes, hypertension, cardiovascular disease (CVD), congestive heart failure (CHF), dyslipidemia, PA, and education levels (See Table S3 for variable definitions).
Mixed effects models assessed whether cognition worsened over time with increasing ACB score. Three cognitive outcomes (fluency, memory, and overall) were analyzed. Each cognitive composite index was treated as the outcome in separate models that adjusted incrementally with ACB score as the main exposure of interest. In two separate sets of models, ACB score was modeled continuously and dichotomized, with ACB scores; >3 considered ACB+. Follow-up time was calculated from baseline, and random effects included the random intercept for each participant and a random slope for time. Model 0 was minimally adjusted for time, time,2 the interaction of time and ACB, number of medications, and indicators for first assessments, and random effects. Model 1 included additional adjustment for age, sex, race, and education. Adjustment for diabetes, hypertension, CVD, CHF, dyslipidemia and depressive symptoms were added in Model 2. Model 3 added adjustment for self-reported PA. Three-way interactions between ACB, time and potential modifiers of age and PA were individually tested in Model 3. To assess a burden-response relationship, we used total ACB score to divide the ACB group into 4 groups, each having an increase of 3.25 in ACB score (3.25–13). Statistical significance was set at p < 0.05.
We began with 25,913 potential participants. After our exclusion criteria, we included data for 20,575 participants with 19,488 contributing to fluency score and 19,227 contributing to memory score (Figure S1). Demographic information for low and high ACB are presented in Table S1.
ACB+ and cognition
In mixed models for each cognitive score, significant effects of ACB+ on cognitive scores were observed at baseline, so that participants with ACB+ were estimated to have composite scores between 0.08 and 0.12 standard deviations lower than ACB− participants across the different composite outcomes (Table 1, Figure 1). We observed significant associations between ACB+ and cognition over time for the fluency composite outcome (p = 0.0346), but a significant change in the slope associated with ACB+ was not observed in the overall composite (Figure 1) or memory composite scores.
|Model 1||Model 2||Model 3|
|Beta (SE)||t||p||Beta (SE)||t||p||Beta (SE)||t||p|
|ACB||−0.116 (0.016)||−7.07||<0.001||−0.115 (0.016)||−6.97||<0.001||−0.105 (0.016)||−6.40||<0.001|
|ACB * Time||−0.003 (0.002)||−1.50||0.134||−0.003 (0.002)||−1.55||0.122||−0.003 (0.002)||−1.54||0.124|
|ACB||−0.139 (0.022)||−6.29||<0.001||−0.136 (0.022)||−6.20||<0.001||−0.124 (0.022)||−5.66||<0.001|
|ACB * Time||−0.003 (0.002)||−1.18||0.239||−0.003 (0.002)||−1.22||0.221||−0.003 (0.002)||−1.22||0.222|
|ACB||−0.089 (0.019)||−4.70||<0.001||−0.091 (0.019)||−4.81||<0.001||−0.084 (0.019)||−4.41||<0.001|
|ACB * Time||−0.004 (0.002)||−2.10||0.036||−0.004 (0.002)||−2.13||0.034||−0.004 (0.002)||−2.11||0.035|
- Note: Model 1 = additional adjustment for age, sex, race, education; Model 2 = additional adjustment for diabetes, hypertension, cardiovascular disease, congestive heart failure, dyslipidemia, depressive symptoms; Model 3 = additional adjustment for physical activity.
ACB and cognition
To determine if cognition was associated with overall ACB score, we modeled ACB score continuously and visualized the predicted models at five age values that spanned the range in our sample (Figure 2). Results (Table S4) showed that increased ACB was associated with both lower cognition scores at baseline and a sharper rate of decline over time across all three composite outcomes (all p < 0.01).
ACB, cognition, and age
Figure 2 shows fluency composite scores (Panel A) and memory composite (Panel B) scores across varying ages by ACB+. While the trajectories of fluency and memory composite scores differed across age, there were no significant three-way interactions between age and ACB+ on cognition over time (memory, p = 0.62; fluency, p = 0.56). At each age, the gap at baseline between ACB+ and ACB− groups was consistent with main effect models for both fluency (Panel A) and memory (Panel B) with no significant difference in the rate of decline. All ages showed decline in fluency composite score except age 45 whose fluency score improved marginally with time. Unlike verbal fluency scores, memory scores showed practice effects. We did not identify significant three-way interactions between age, ACB+, and cognition (p = 0.73) in the overall composite scores.
ACB, cognition, and PA
To determine if self-reported PA at baseline mitigated the effects of ACB on the cognitive composites, we examined whether self-reported frequency of PA modified the impact of ACB over time (Model 3; Table S5). No significant interaction between time, ACB+ and PA was evident for the fluency (p = 0.74) or memory (p = 0.08) composite scores. However, a significant interaction emerged in the overall composite (p = 0.013). Compared with those who did not report PA, those who engaged in 1–3 sessions per week had a weaker relationship between ACB and cognition, which was not demonstrated for participants who reported 4+ weekly sessions.
We examined the effects of ACH's on cognition in middle aged and older adults. Cognition was worse for the ACB+ group at baseline and worsened over time for our fluency composite outcome. PA mitigated the negative effects of ACB+ on cognition over time. Prior studies showed a positive association between cognitive impairments and ACB scores.16-18 Our results extend previous findings16, 17, 19 by demonstrating that increased ACB has a negative effect on cognition, and that PA has a moderating effect on overall composite scores.
Our results extend previous findings18 that reported an effect of ACH on scores on a brief cognitive screening instrument. The current investigation provided a more detailed assessment of cognitive function and found that compared with the ACB− group, fluency, and memory scores were lower at baseline for the ACB+ group. Memory scores showed improvement with time, which dissipated for those 81 years and older. The conflicting results between fluency scores and improved memory scores are perplexing but may be a result of practice effects.19
Our current study also extends findings from studies that examined the effects of ACH on cognition. For example, Cai and colleagues20 reported a relationship between chronic (2–3 months) high ACB scores and cognitive function in older adults. We attempted to expand these findings by dividing our ACB+ group into four groups, each having an increase of 3.25 in ACB score (3.25–13) and found that participants with higher ACB were more likely to experience negative cognitive effects at baseline and a more pronounced rate of decline over time.
We also found that effects of ACH on cognition increased with age. Taken together, our findings and those from previous studies suggest a potential age threshold after which there is progressively more impact of ACB on cognition. These findings are consistent with research reporting an increased risk of AD and brain atrophy in adults 50–65 years of age taking ACH drugs across a 20-year follow-up period.21 Furthermore, others reported an absence of cognitive effects of ACH in children and middle-aged adults,22 whereas age-related pathophysiological changes may predispose older adults to the negative side effects of ACH.23 Increased susceptibility of older adults to the negative cognitive effects of ACHs has been attributed to increased BBB permeability,24 and decreased hepatic and renal drug clearance, which result in increased drug bioavailability and risk of toxicity.25
Self-reported frequency of PA at baseline was associated with a decreased effect of ACB on cognition. The protective effects of PA held for both the memory and fluency composites but was especially striking for memory. Cognition was most affected for those who did not report PA in the ACB+ and ACB− groups at baseline and throughout the duration of the study. Supporting evidence from a previous study showed a relationship between PA and ACB burden using accelerometer data (available in a subsample of participants) and cognitive function.26
Limitations of our analyses merit discussion. First, it is possible that more comprehensive and in-person cognitive assessments might yield better delineation of rates of impairment and neuropsychological pattern. Second, while data on prescribed ACH's at baseline were recorded, measures of adherence to drug schedules were not collected, limiting inferences that can be made. Third, we used a Likert-scale self-report measure of activity levels that has not yet been validated with older adults. Retrospective self-reports are inherently prone to inaccuracies and response biases, and therefore sometimes show limited reliability and validity. A more objective longitudinal assessment of PA is warranted, and future studies should aim to collect data to help identify the physiological effects of reported activity levels. Lastly, it is not possible to eliminate confounding as an explanation of the findings given the differences in low versus high ACH users at baseline.
We demonstrated the effects of ACB on cognition across time in middle-aged and older adults and highlighted an effect of ACHs on cognition regardless of age. Older adults appeared more susceptible to ACHs, whereas PA emerged as a modifying factor that may protect older adults from the negative cognitive effects of ACHs. Our results have several clinical implications. First, cognition of those with moderate to high ACH burden should be monitored and referrals for cognitive assessments made when appropriate. Second, antidepressants have ACH properties yet are widely prescribed to middle aged and older adults. Clinicians should consider psychotherapy as adjunct treatment or replacement to antidepressant medications whenever possible. Finally, recommendations to PA should be provided to patients who are prescribed ACH's, especially because PA has been associated with reductions in depressive symptoms,27-29 which may allow clinicians to lessen ACB.
Adam Gerstenecker, Amani M. Norling, Aleena Bennett, Michael Crowe, and Leann Long contributed to the study concept and design. Aleena Bennett, Michael Crowe, Leann Long, Adam Gerstenecker, and Amani M. Norling contributed to the data analyses. Adam Gerstenecker, Amani M. Norling, Aleena Bennett, Michael Crowe, and Leann Long carried out results interpretations. Amani M. Norling and Adam Gerstenecker drafted the manuscript. Aleena Bennett, Michael Crowe, Leann Long, Sara A. Sims, Terina Myers, Victor A. Del Bene and Ronald M. Lazar critically revised the manuscript.
The authors thank the other investigators, the staff, and the participants of the REGARDS study for their valuable contributions. This research project is supported by a cooperative agreement U01 NS041588 from the National Institute of Neurological Disorders and Stroke, National Institutes of Health, Department of Health and Human Services, and by the McKnight Brain Institute at UAB. The authors thank the other investigators, the staff, and the participants of the REGARDS study for their valuable contributions. A full list of participating REGARDS investigators and institutions can be found at http://www.regardsstudy.org. This study was also supported by the UAB McKnight Brain Institute.
This research project is supported by cooperative agreement U01 NS041588 co-funded by the National Institute of Neurological Disorders and Stroke (NINDS) and the National Institute on Aging (NIA), National Institutes of Health, Department of Health and Human Services, and the UAB McKnight Brain Institute.
CONFLICT OF INTEREST
|jgs18279-sup-0001-Supinfo.docxWord 2007 document , 147.9 KB||
Table S1: Baseline demographic information
Figure S1: Flow diagram for sample identification
Table S3: Variable definitions
Table S4: Mixed model analyses (continuous ACB)
Table S5: Effect of ACB and PA levels on cognition
Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.
- 1, , , , . Changes in prescription and over-the-counter medication and dietary supplement use among older adults in the United States, 2005 vs 2011. JAMA Intern Med. 2016; 176(4): 473-482.
- 2, . Drugs with anticholinergic properties: a current perspective on use and safety. Expert Opin Drug Saf. 2011; 10(5): 751-765.
- 3, , , , . Impact of anticholinergics on the aging brain: a review and practical application. Aging Health. 2008; 4(3): 311-320. doi:10.2217/1745509X.4.3.311
- 4, , , et al. Anticholinergic drugs and incident dementia, mild cognitive impairment and cognitive decline: a meta-analysis. Age Ageing. 2020; 49(6): 939-947.
- 5, , , et al. Cumulative use of strong anticholinergics and incident dementia: a prospective cohort study. JAMA Intern Med. 2015; 175(3): 401-407.
- 6, , . A review of the effects of physical activity and exercise on cognitive and brain functions in older adults. J Aging Res. 2013; 2013:657508. doi:10.1155/2013/657508
- 7, , , et al. The reasons for geographic and racial differences in stroke study: objectives and design. Neuroepidemiology. 2005; 25(3): 135-143.
- 8, , , , . Six-item screener to identify cognitive impairment among potential subjects for clinical research. Med Care. 2002; 40(9): 771-781.
- 9, , , et al. Cognitive assessment via telephone: a scoping review of instruments. Arch Clin Neuropsychol. 2020; 35(8): 1215-1233.
- 10, , , et al. Consortium to establish a registry for Alzheimer's disease (CERAD): the first twenty years. Alzheimers Dement. 2008; 4(2): 96-109.
- 11, , , et al. C-reactive protein and risk of cognitive decline: the REGARDS study. PLoS One. 2020; 15(12):e0244612.
- 12, , , et al. Fast-food for thought: retail food environments as resources for cognitive health and wellbeing among aging Americans? Health Place. 2020; 64:102379.
- 13, , , . The validity of self-reported exercise-induced sweating as a measure of physical activity. Am J Epidemiol. 1990; 132(1): 107-113.
- 14, , . Physical activity assessment for epidemiologic research: the utility of two simplified approaches. Prev Med. 1987; 16(5): 636-646.
- 15, , , , . A mail survey of physical activity habits as related to measured physical fitness. Am J Epidemiol. 1988; 127(6): 1228-1239.
- 16, , , et al. The cognitive impact of anticholinergics: a clinical review. Clin Interv Aging. 2009; 4: 225-233.
- 17, , , et al. Anticholinergic medication use and cognitive impairment in the older population: the medical research council cognitive function and ageing study. J Am Geriatr Soc. 2011; 59(8): 1477-1483. doi:10.1111/j.1532-5415.2011.03491.x
- 18, , , , , . Disparities in cognitive impairment with anticholinergic drug use: a population-based study. Neurol Clinic Pract. 2021; 11(3): e277-e286.
- 19, , , , . CERAD practice effects and attrition bias in a dementia prevention trial. Int Psychogeriatr. 2013; 25(7): 1115-1123.
- 20, , , , . Long-term anticholinergic use and the aging brain. Alzheimers Dement. 2013; 9(4): 377-385.
- 21, , , . Midlife anticholinergic drug use, risk of Alzheimer's disease, and brain atrophy in community-dwelling older adults. Alzheimers Dement. 2017; 3(3): 471-479.
- 22, , , et al. The effects of anticholinergic medications on cognition in children: a systematic review and meta-analysis. Sci Rep. 2021; 11(1): 1-15.
- 23. The problems of anticholinergic adverse effects in older patients. Drugs Aging. 1993; 3(4): 335-348.
- 24, . Blood–brain barrier: ageing and microvascular disease–systematic review and meta-analysis. Neurobiol Aging. 2009; 30(3): 337-352.
- 25, . Age-related changes in pharmacokinetics. Curr Drug Metab. 2011; 12(7): 601-610.
- 26, , , , , . Objectively measured physical activity and cognitive function in older adults. Med Sci Sports Exerc. 2017; 49(1): 47-53. doi:10.1249/MSS.0000000000001079
- 27, , , . Exercise as a treatment for depression: a meta-analysis. J Affect Disord. 2016; 202: 67-86.
- 28, . Exercise training and depression in older adults. Neurobiol Aging. 2005; 26(1): 119-123.
- 29, , , . The relationship between physical activity and depressive symptoms in healthy older women. Gerontol Geriat Med. 2016; 2: 2333721415626859.