Anticholinergic Medication Use in Veterans Affairs Long‐Term Care Residents: Clinical Patterns and Opportunities for Deprescribing

ABSTRACT

Background

Anticholinergic exposure is common in nursing homes; however, the true burden may be underestimated, particularly due to over‐the‐counter agents such as first‐generation antihistamines that are not consistently captured in prior claims‐based studies. This gap limits accurate characterization of anticholinergic use and its clinical implications in older adults.

Participants and Setting

Veterans aged ≥ 65 years with stays ≥ 90 days (N = 45,183) residing in US Department of Veterans Affairs (VA) Community Living Centers (CLC). This study is a secondary data analysis of residents identified during fiscal years 2007–2019.

Methods

True anticholinergic exposure through barcode‐based medication administration dispensing records allowed daily measurement of anticholinergic exposure, with follow‐up extending from admission to discharge, death, or April 1, 2025. Anticholinergic medication use was defined as use ≥ 4 days per week during the CLC stay. Drugs were identified using VA classifications aligned with the 2023 American Geriatrics Society Beers Criteria and grouped into nine anticholinergic classes. Outcomes included prevalence, duration, and number of anticholinergic drug classes used; patterns by dementia status; and temporal trends over time.

Results

Among 45,183 CLC residents, 33.4% (n = 15,074) used anticholinergic medications. Residents with dementia (n = 20,254) showed substantial exposure to neuropsychiatric anticholinergics, including second‐generation antipsychotics (30.6%) and antiparkinsonian agents (10.1%). First‐generation antihistamine use remained high in this group (29.5%). Residents without dementia (n = 24,929) more frequently received antihistamines (34.4%), bladder antimuscarinics (23.7%), and skeletal muscle relaxants (11.1%). From FY2007–2019, anticholinergic use declined modestly, with decreases in antihistamines (12.1%–8.8%), antidepressants (6.8%–3.5%), and antiemetic/antivertigo agents (5.2%–1.2%). Most residents used a single anticholinergic class, increasing from 75.5% to 82.2% over time.

Conclusions

Although some anticholinergic classes may be clinically necessary, our findings highlight potentially underrecognized and modifiable sources of exposure, particularly first‐generation antihistamines, underscoring the need for deprescribing efforts.

Summary

• Key points

  • ○
    This study leverages large‐scale, real‐world medication administration data to characterize population‐level exposure patterns of anticholinergic medications.
  • ○
    One‐third of long‐stay Veterans in VA Community Living Centers used anticholinergic medications, with high exposure even among residents with dementia.
  • ○
    Barcode‐based medication administration data revealed substantial use of first‐generation antihistamines and other high‐burden anticholinergics that are often undercaptured in claims‐based studies.
  • ○
    Although overall anticholinergic use declined modestly from 2007 to 2019, several potentially modifiable drug classes remain prevalent in long‐term care.
• Why does this paper matter?

  • ○
    Anticholinergic medications with known cognitive risks remain widely used in long‐term care, even among residents with dementia and despite the availability of safer alternatives. These findings identify clear targets for deprescribing initiatives and support system‐level efforts to improve medication safety in aging populations.

1. Introduction

Currently, about 1.4 million people reside in nursing homes and an additional 800,000 reside in assisted living, with projections estimating these numbers will more than triple by 2050 [1, 2, 3]. Anticholinesterase drugs, which block acetylcholine in the central and peripheral nervous systems, are widely prescribed in older adults including in nursing homes and represent a modifiable risk factor for dementia. These include certain antihistamines, antidepressants, and medications for gastrointestinal and bladder disorders. In a retrospective analysis of Medicare claims from nursing home stays of at least 14 days, 29.6% of beneficiaries had a high anticholinergic burden (ACB score ≥ 3), often attributable to a single potent medication, while 35.2% had a low burden (ACB score 1–2) [4]. High burden anticholinergic use was associated with greater dementia severity [4]. In individuals without dementia, meta‐analytic evidence indicates that ≥ 3 months of anticholinergic use is associated with an increased risk of incident dementia [5] and all‐cause dementia [6]. In contrast, among individuals with pre‐existing dementia, evidence on cognitive worsening is mixed; however, anticholinergic exposure has been associated with adverse clinical outcomes, including higher all‐cause mortality. Despite these risks, prescribing to individuals with dementia remains common [7]. Despite these risks, prescribing to individuals with dementia remains common. A multicenter study by Hook et al. [8] found that patients with dementia are often discharged from hospitals with a higher anticholinergic burden than at admission, as measured by cumulative anticholinergic burden scales (e.g., ACB), even when concurrently prescribed cholinesterase. Given the projected growth of institutionalized older adults and the established risks, careful consideration of anticholinergic prescribing remains essential.

A challenge of accurately capturing anticholinergic medication use is that certain medications, particularly first‐generation antihistamines, are available over‐the‐counter (OTC) and are not captured in claims data. Prior claims‐based studies have reported that approximately 65% of nursing home residents have some level of anticholinergic burden (35.2% low; 29.6% high), with high burden often driven by a single potent medication [4]; however, these estimates likely underestimate antihistamine use due to unmeasured OTC exposure. Thus, prior research based on claims data has likely underestimated antihistamine medication use. Importantly, many anticholinergic medications are potentially avoidable, as first‐generation antihistamines like diphenhydramine have equally effective second‐generation substitutes that minimize anticholinergic load and reduce risk for adverse side effects such as cognitive impairment [9]. The Veteran Affairs (VA) nursing home population presents a unique opportunity to address this issue. Barcode Medication Administration (BCMA) data allow precise tracking of anticholinergic medication administrations, providing the foundation for evaluating patterns of use and identifying opportunities to reduce exposure. The objective of this study was to characterize population‐level patterns of anticholinergic medication use among Veterans receiving long‐term care in VA Community Living Centers (CLC), with particular attention to differences between residents with and without dementia. Specifically, we quantified the prevalence, duration, and complexity of anticholinergic prescribing across three mutually exclusive drug groups (first‐generation antihistamines, non‐antihistamine anticholinergics, and dual users), using facility‐based medication administration records to capture exposure intensity and number of agents over the entire CLC stay.

2. Methods

2.1. Cohort Description

We included veterans aged 65 years or older admitted between October 1, 2006 and September 30, 2019. The cohort was established in 2021 to serve as a research resource for examining medication use among residents in Veterans Affairs (VA) long‐term care settings. The initial cohort included residents admitted between 2006 and 2019. Cohort construction involved systematic identification of eligible residents and extensive curation and cleaning of linked administrative and clinical data elements. Subsequently, the dataset has been updated to incorporate additional follow‐up time, enabling longitudinal analyses of medication use and related outcomes among cohort participants.

We analyzed data from 487,401 residents who had nursing home stays without hospice. Sequential exclusions included residents with stays < 90 days (n = 415,365), those < 65 years old at admission (n = 22,374), and residents with acute care stays > 30 days during a nursing home stay (n = 14). After further excluding stays that occurred prior to a more recent stay (n = 4330) and stays without documented blood pressure assessment (n = 135), the final analytic sample included 45,183 unique residents with at least one blood pressure reading. We built a database for the CLC cohort by including veterans admitted to a CLC between October 1, 2006 and September 30, 2019 and stayed at least 90 days (long‐term stay). Of 45,183 residents, 29,038 had a single long‐term stay, and 16,145 had multiple long‐term stays (with the last stay included in the final cohort). Residents were excluded if they (1) had a stay < 90 days to exclude those undergoing post‐acute rehabilitation; (2) had an acute hospital stay > 30 days during CLC stay (n = 12); (3) were < 65 years at admission; or (4) had no blood pressure measurements (n = 125). Figure S1 displays more detailed information.

2.2. Anticholinergic Medication Use

Anticholinergic medications were identified in BCMA data by drug name among CLC residents during their nursing home stay, and use was defined as ≥ 4 days per week. The anticholinergic exposure definition of ≥ 4 days per week captures clinically meaningful use but does not fully account for variability in duration or cumulative exposure across the length of stay. As a result, this definition may introduce heterogeneity in exposure classification, ranging from short‐term to sustained use, which could attenuate observed associations.

Anticholinergic drugs were identified using the VA drug classification system available in the dataset, based on the 2023 American Geriatrics Society (AGS) Beers Criteria (AGS Beers Criteria) for potentially inappropriate medication (PIM) use in older adults [10]. The identified anticholinergics were grouped into nine categories: (1) first‐generation antihistamines, (2) anticholinergic antidepressants, (3) antiparkinsonian agents, (4) first‐generation antipsychotics, (5) second‐generation antipsychotics, (6) gastrointestinal antispasmodics, (7) skeletal muscle relaxants, (8) antiemetic/antivertigo agents, and (9) bladder antimuscarinics. Table S1 summarizes major classes of medications with anticholinergic activity, including common examples, typical clinical indications, and over‐the‐counter (OTC) availability.

In addition, residents were categorized into three mutually exclusive groups based on anticholinergic medication use over the entire CLC stay: (1) first‐generation antihistamine use only, (2) non‐antihistamine anticholinergic use only, or (3) dual use. Dual users were defined as residents who used both medication categories at any time during the stay, regardless of the timing or sequence of exposure.

2.3. Variables Used From Cohort and Definitions/Clinical

Baseline characteristics were defined at the time of CLC admission to provide a consistent reference point across residents. To characterize the timing of exposure, we assessed the distribution of time from admission to first qualifying anticholinergic use. Among residents meeting anticholinergic use criteria, the median time to exposure was 2 weeks (interquartile range: 0–16), with a mean (SD) of 18.7 (44.3) weeks; 6388 residents of the 45,183 (14.1%) met criteria during the admission week (week 0). Comorbidities include diagnoses ascertained via ICD‐9/10 codes anytime in the 1 year prior to CLC admission. Long‐term care stay was the duration between CLC admission and discharge, or date of death, or latest follow‐up until February 1, 2025 for those not discharged. The main data source was Veterans Health Administration's (VA) Corporate Data Warehouse (CDW), a standardized network of regional and national clinical databases [11]. In addition, we linked data with Centers for Medicare and Medicaid Services Minimum Data Set (MDS versions 2.0 and 3.0) that include variables on cognitive function scale (CFS) and activities of daily living (ADL) [12]. Cognitive function was assessed using the Minimum Data Set (MDS), which transitioned from version 2.0 to 3.0 on October 1, 2010. MDS 2.0 includes the cognitive performance scale (CPS) [12], whereas MDS 3.0 includes the Brief Interview for Mental Status (BIMS). To harmonize assessments across versions, we used the CFS, which integrates CPS and BIMS into a single 4‐level measure of cognitive function: cognitively intact (CFS = 1), mildly impaired (CFS = 2), moderately impaired (CFS = 3), and severely impaired (CFS = 4).

2.4. Dementia

The observation window for baseline characteristics was defined as the 1‐year period prior to CLC admission. Dementia status was ascertained during this period using qualifying ICD‐9 and ICD‐10 diagnostic codes. The code list was developed and is annually reviewed by a group of VA geriatricians and is routinely used for quality improvement and research within the VA. A participant was classified as having dementia if any qualifying diagnostic code appeared in the medical record during the baseline assessment window (ICD‐9 and ICD‐10 codes).

2.5. Statistical Analysis

Baseline characteristics of CLC residents were evaluated using mean (SD) and number (%), stratified by anticholinergics user status. T‐tests and chi‐square tests were used to compare between‐group differences for continuous and categorical variables respectively. We separated anticholinergic users into three mutually exclusive groups: first‐generation antihistamine users only, non‐antihistamine users only and dual users, and made comparisons of their use patterns (length of use and number of medications) stratified by dementia status. We then presented, in figures, the prevalent use of nine anticholinergics medication classes by dementia status, and use of three medication groups by fiscal year (FY) of CLC admission. Temporal trends in anticholinergic medication use from FY 2007 to 2019 were evaluated using linear tests of trend applied to annual proportions. Analyses were conducted for the full cohort and stratified by dementia status. Test statistics, standard errors (SE), and two‐sided p values were estimated to assess directional changes over time, with positive coefficients indicating increasing trends and negative coefficients indicating decreasing trends. For overall anticholinergic exposure, linear trend statistics were used to evaluate changes in the proportion of non‐users over time, reflecting reductions in anticholinergic use. Class‐specific trends were assessed separately for major anticholinergic drug categories, including first‐generation antihistamines, antidepressants with anticholinergic properties, and antiemetic/antivertigo agents. Additional analyses examined temporal changes in the complexity of anticholinergic regimens among users, including the proportion of residents using one versus two anticholinergic drug classes. These analyses were similarly conducted overall and stratified by dementia status.

All analyses were performed using SAS (Version 9.4) and Stata Release 18. p values < 0.05 were considered statistically significant. Figures were created using R version 4.2.3.

3. Results

3.1. Sample Description

Among 45,183 long‐term care residents, 33.4% (n = 15,074) were classified as anticholinergic users, defined as use for at least four days per week during their stay. Anticholinergic users differed from non‐users across multiple demographic, functional, and comorbidity profiles, as summarized in Table 1. Compared with non‐users, anticholinergic users were slightly younger and more likely to be female. Racial composition also differed (p < 0.001): a higher proportion of users were White (76.9% vs. 72.8%), whereas Black residents were less represented among users. Geographic and temporal variation was observed, with greater representation of users admitted during FY 2007–2010 and from Midwest and Southeast districts (all p < 0.001). Anticholinergic users had a slightly higher mean body mass index and better functional independence at admission, reflected by fewer ADL impairments (lower ADL scores; 13.5 ± 7.8 vs. 14.3 ± 7.7; p < 0.001). CFS distributions were comparable between groups, although users were marginally more likely to be mildly impaired (37.7% vs. 35.1%; p < 0.001). Anticholinergic users had higher rates of Parkinson's disease, Huntington's disease, and chronic obstructive pulmonary disease (p < 0.001). Dementia prevalence was similar but slightly more prevalent in the non‐user group (45.2% vs. 44.0%; p = 0.02).

TABLE 1.

Baseline characteristics of Long‐Term Care residents (N = 45,183) by anticholinergics use.

Characteristic	Non‐users of anticholinergics (N = 30,109)	Users of anticholinergics (N = 15,074)	p
Age at CLC admission (years), mean (SD)	78.8 (8.6)	76.7 (8.2)	< 0.001
Female sex, n (%)	622 (2.1)	413 (2.7)	< 0.001
Race, n (%)			< 0.001
White	21,932 (72.8)	11,585 (76.9)
Black	5199 (17.3)	2121 (14.1)
Asian or Pacific Islander	376 (1.3)	176 (1.2)
American Indian/Alaska Native	156 (0.5)	69 (0.5)
Multiple Race	208 (0.7)	106 (0.7)
Unknown	2238 (7.4)	1017 (6.8)
Hispanic/Latino Ethnicity, n (%)	1405 (4.7)	648 (4.3)	0.1
Time of CLC Admission, n (%)			< 0.001
FY2007–2008	4520 (15.0)	2575 (17.1)
FY2009–2010	4293 (14.3)	2269 (15.1)
FY2011–2012	4259 (14.2)	2139 (14.2)
FY2013–2014	4469 (14.8)	2153 (14.3)
FY2015–2016	4783 (15.9)	2326 (15.4)
FY2017–2018	5099 (16.9)	2450 (16.3)
FY2019	2686 (8.9)	1162 (7.7)
District, n (%)			< 0.001
Continental	3768 (12.5)	1962 (13.0)
Midwest	7062 (23.5)	3854 (25.6)
North Atlantic	9396 (31.2)	4631 (30.7)
Pacific	5220 (17.3)	2176 (14.4)
Southeast	4663 (15.5)	2451 (16.3)
Body mass index, mean (SD)	27.9 (8.8)	28.6 (9.1)	< 0.001
Cognitive function scale (CFS), 1st week, n (%)			< 0.001
Cognitively intact	7356 (24.4)	3687 (24.5)
Mildly impaired	10,560 (35.1)	5676 (37.7)
Moderately impaired	5731 (19.0)	2581 (17.1)
Severely impaired	1802 (6.0)	793 (5.3)
Missing	4660 (15.5)	2337 (15.5)
Activities of daily living (ADL), 1st week (0–28), mean (SD)	14.3 (7.7)	13.5 (7.8)	< 0.001
Smoking status, n (%)			< 0.001
Never	4614 (15.3)	2185 (14.5)
Current	6960 (23.1)	4079 (27.1)
Former	12,536 (41.6)	5994 (39.8)
Missing/unknown	5999 (19.9)	2816 (18.7)
Comorbidities, n (%)
Dementia	13,616 (45.2)	6638 (44.0)	0.02
Parkinson's/Huntington's diseases	2249 (7.5)	1486 (9.9)	< 0.001
Hypertension	24,239 (80.5)	11,987 (79.5)	0.01
Diabetes	13,373 (44.4)	6784 (45.0)	0.2
Atrial fibrillation	7972 (26.5)	3508 (23.3)	< 0.001
Coronary heart disease	12,708 (42.2)	6082 (40.4)	< 0.001
Cerebrovascular disease	9763 (32.4)	4594 (30.5)	< 0.001
Peripheral vascular disease	7551 (25.1)	3535 (23.5)	< 0.001
Heart failure	10,258 (34.1)	4639 (30.8)	< 0.001
Depression	11,226 (37.3)	6769 (44.9)	< 0.001
Malignancy	9081 (30.2)	4703 (31.2)	0.02
Metastatic cancer	3118 (10.4)	1626 (10.8)	0.2
Osteoarthritis	7444 (24.7)	3996 (26.5)	< 0.001
Chronic obstructive pulmonary disease	5333 (17.7)	2912 (19.3)	< 0.001
Renal failure	9454 (31.4)	4247 (28.2)	< 0.001
Acute kidney injury	7430 (24.7)	3310 (22.0)	< 0.001
Coagulation defects or other hematological disorders	2311 (7.7)	1190 (7.9)	0.4

Note: This table displays demographic, clinical, and functional characteristics of long‐term care residents admitted to Community Living Centers (CLC), stratified by anticholinergic use (defined as ≥ 4 days per week during stay). Continuous variables are presented as mean (SD), and categorical variables as n (%). p values reflect comparisons between non‐users and users of anticholinergics, obtained using t‐tests for continuous variables and chi‐squared tests for categorical variables. Cognitive status was assessed using the cognitive function scale (CFS), and functional independence with the activities of daily living (ADL) scale (range: 0–28). Fiscal year of admission and district of residence are reported to account for temporal and regional variation.

Abbreviations: ADL, activities of daily living; CFS, cognitive function scale; CLC, Community Living Center; FY, fiscal year; SD, standard deviation.

3.2. Patterns of Antihistamine, Non‐Antihistamine Anticholinergic, and Dual Users

Patterns of anticholinergic use varied by dementia status and anticholinergic medication user status (Table 2). Among residents with dementia (n = 6638), non‐antihistamine anticholinergics were the most frequently used (n = 4678), followed by antihistamines only (n = 1230) and dual users (n = 730).

TABLE 2.

Anticholinergic use burden and medication counts by user status stratified by dementia.

Variable	With dementia (n = 6638)			Without dementia (n = 8436)
	First gen antihistamines only (n = 1230)	Non‐antihistamines only (n = 4678)	Dual users (n = 730)	First gen antihistamines only (n = 1859)	Non‐antihistamines only (n = 5536)	Dual users (n = 1041)
Weeks of use/total stay (%)
Mean (SD)	19.9 (25.2)	44.6 (37.6)	28.6 (32.8)	25.7 (28.9)	46.9 (37.4)	30.3 (32.6)
Median (IQR), range	7.4 (2.4–28.0) 0.15–100	34.4 (7.4–86.7) 0.18–100	11.1 (2.6–50.8) 0.16–100	12.5 (4.5–39.3) 0.13–100	38.5 (9.1–90.0) 0.15–100	15.0 (3.9–50.0) 0.16–100
# Anticholinergics—Mean (SD), median, range	1.10 (0.31), 1, 1–3	1.21 (0.50), 1, 1–6	2.51 (0.79), 2, 2–7	1.10 (0.32), 1, 1–3	1.22 (0.51), 1, 1–7	2.52 (0.84), 2, 2–10
Used 1 med, n (%)	1115 (90.7)	3837 (82.0)	NA	1683 (90.5)	4487 (81.1)	NA
Used 2 meds, n (%)	109 (8.9)	720 (15.4)	464 (63.6)	166 (8.9)	901 (16.3)	660 (63.4)
Used 3 meds, n (%)	*	106 (2.3)	189 (25.9)	*	122 (2.2)	271 (26.0)
Used 4+ meds, n (%)	NA	15 (0.3)	77 (10.5)	NA	26 (0.5)	110 (10.6)

Note: This table summarizes anticholinergic use patterns among nursing home residents, stratified by dementia status and type of user group: antihistamines only, non‐antihistamines only, and both antihistamines and non‐antihistamines (dual users). The weeks of use/total weeks of stay (burden) is expressed as mean (SD), median (IQR), and observed range. Medication counts are depicted as mean (SD), median, and range of distinct anticholinergic medications used during the nursing home stay. The distribution of medications used include the number and percentage of residents using 1, 2, or ≥ 3 anticholinergics (with more granular breakdowns presented when available). Asterisks (*) indicates a sample size less than 10 participants.

Non‐antihistamine users had the highest proportion of weeks on treatment relative to total stay (mean = 44.6%, SD = 37.6%), compared with 19.9% (SD = 25.2%) for antihistamine‐only users and 28.6% (SD = 32.8%) for dual users. The median percentage of use followed the same pattern (34.4% vs. 7.4% vs. 11.1%, respectively). Among residents without dementia (n = 8436), similar trends were observed, although the overall burden of use was slightly attenuated. Non‐antihistamine users again demonstrated the greatest exposure (mean = 46.9%, SD = 37.4%), followed by dual users (30.3%, SD = 32.6%) and antihistamine‐only users (25.7%, SD = 28.9%). Nearly two‐thirds of dual users received two anticholinergics, and the remaining received three or more.

3.3. Prevalence of Anticholinergic Medication Class by Dementia Status

Figure 1 summarizes the prevalence of specific anticholinergic medication classes among CLC residents by dementia status. Patterns of use differed between residents with and without dementia. Among residents with dementia (n = 6638), the most frequently prescribed anticholinergic classes were second‐generation antipsychotics (30.6%), first‐generation antihistamines (29.5%), and bladder antimuscarinics (20.2%). Among residents without dementia (n = 8436), the most commonly used anticholinergic medication classes were first‐generation antihistamines (34.4%), bladder antimuscarinics (23.7%), and antidepressants (17.7%). Neuropsychiatric anticholinergics showed marked differences by cognitive status: second‐generation antipsychotic use was nearly three times higher among residents with dementia compared with those without (30.6% vs. 11.0%), whereas skeletal muscle relaxants (11.1% vs. 5.1%) and antiemetic/antivertigo agents (11.9% vs. 4.2%) were more frequently used in residents without dementia. Use of antiparkinsonian agents was twice as common among those with dementia (10.1% vs. 5.3%). Overall, residents with dementia exhibited higher exposure to neuropsychiatric anticholinergics, while those without dementia were more often prescribed agents for musculoskeletal and gastrointestinal indications.

FIGURE 1.

Prevalence of anticholinergic medication classes by dementia status among veterans in long‐term care. Bar plots display the proportion of participants with and without dementia who were prescribed at least one medication from each anticholinergic class. Among residents with dementia (n = 6638), first‐generation antihistamines (29.5%), second‐generation antipsychotics (30.6%), and bladder antimuscarinics (20.2%) were the most commonly used classes. Among residents without dementia (n = 8436), first‐generation antihistamines (34.4%), bladder antimuscarinics (23.7%), antidepressants (17.7%), and antiemetic/antivertigo agents (11.9%) were most prevalent. Percentages were calculated using the total number within each dementia stratum. Class‐specific totals exceed 100% because individuals may receive medications from multiple anticholinergic classes.

3.4. Temporal Trends of Anticholinergic Use Categories

Figure 2 illustrates temporal trends in anticholinergic use among CLC residents from FYs 2007–2019. Across the full cohort (n = 45,183), overall anticholinergic use declined modestly over time. The proportion of antihistamine‐only users decreased from 7.4% in FY2007–2008 to 6.0% in FY2019, while dual users declined from 4.8% to 2.8%. In contrast, non‐antihistamine‐only use showed a consistent decrease from 24.2% to 21.4%. Consistent with these patterns, a linear test of trend demonstrated a significant increase in non‐use of anticholinergics over time in the full cohort (Statistic = 3.34 [SE = 1.45], p = 0.021). When stratified by dementia status, patterns were broadly similar. Among residents without dementia (n = 24,929), dual use declined from 5.2% to 3.5%, and antihistamine‐only use from 7.9% to 6.4%, with a corresponding increase in non‐use (Statistic = 3.77 [SE = 1.73], p = 0.029). Residents with dementia (n = 20,254) demonstrated comparable declines, with dual use falling from 4.2% to 1.9% and antihistamine‐only use from 6.7% to 5.6%, alongside a significant increase in non‐use over time (Statistic = 2.70 [SE = 1.17], p = 0.021).

FIGURE 2.

Anticholinergic prescribing patterns among veterans in long‐term care. Stacked bar charts display the distribution of anticholinergic use categories by FY (FY2007–2019) for the full cohort (N = 45,183), residents without dementia (N = 24,929), and residents with dementia (N = 20,254). Categories include non‐users (gray), users of only antihistamines (purple), users of only non‐antihistamine anticholinergics (orange), and dual users of both classes (light blue).

3.5. Temporal Trends of Anticholinergic Medication Class

Figure 3, Table S2 presents temporal trends in the use of specific anticholinergic drug classes among CLC residents from FY 2007–2019, overall and stratified by dementia status. Across the full cohort (n = 45,183), first‐generation antihistamines were the most frequently prescribed anticholinergic class, though their prevalence declined steadily from 12.1% in FY 2007–2008 to 8.8% in FY 2019 (Statistic = −1.71 [SE = 0.78], p = 0.027). Similar downward trends were observed for antidepressants with anticholinergic properties (6.8% → 3.5%; Statistic = −1.76 [SE = 0.78], p = 0.024) and antiemetic/antivertigo agents (5.2% → 1.2%; Statistic = −2.61 [SE = 1.11], p = 0.018). In contrast, prescriptions for second‐generation antipsychotics modestly increased over time (6.4% → 7.2%), while use of antispasmodics, skeletal muscle relaxants, and bladder antimuscarinics remained relatively stable. Among residents with dementia (n = 20,254), second‐generation antipsychotics consistently represented the most prevalent anticholinergic class, increasing from 9.9% to 10.8% over the study period. Antihistamine and antidepressant use declined notably (10.9% → 7.5%; Statistic = −1.76 [SE = 0.86], p = 0.041 and 6.0% → 2.5%; Statistic = −1.79 [SE = 0.86], p = 0.037, respectively). Antiemetic/antivertigo use also declined (Statistic = −1.19 [SE = 0.52], p = 0.024). Use of antiparkinsonian and bladder antimuscarinic agents remained relatively unchanged. In residents without dementia (n = 24,929), the overall pattern was similar but with higher relative use of antihistamines (13.1% → 9.9%; Statistic = −1.59 [SE = 0.75], p = 0.034) and antidepressants (Statistic = −1.61 [SE = 0.71], p = 0.023), and a marked decline in antiemetic/antivertigo agents (Statistic = −3.74 [SE = 1.60], p = 0.019). Skeletal muscle relaxant use was slightly higher (2.8% → 4.1%), while neuropsychiatric agent use remained lower compared to residents with dementia.

FIGURE 3.

Temporal trends in anticholinergic medication use by drug class across fiscal years among veterans in long‐term care. Percentages reflect the proportion of all participants (users and non‐users) within each fiscal year who received at least one medication from the specified anticholinergic drug class. Counts represent the total number of individuals prescribed medications in that class during each 2‐year cycle. Drug classes are not mutually exclusive; therefore, individuals may appear in multiple categories if prescribed medications from more than one anticholinergic class. Results are presented for the full cohort and stratified by dementia status.

3.6. Temporal Trends of the Number of Anticholinergic Medications

Table S3 summarizes the distribution of the number of anticholinergic drug classes used per resident from FY 2007–2019, in the full cohort and stratified by dementia status. The percentage distributions were calculated from the cohort of non‐users and users. Across the full cohort (n = 15,074 anticholinergic users), most residents were prescribed only one anticholinergic drug class, and this proportion increased over time, from 75.5% in FY 2007–2008 to 82.2% in FY 2019. The proportion using two classes declined modestly (19.5%–15.0%), and use of three or more classes remained rare (< 4% across all periods). Among residents with dementia (n = 6638 anticholinergic users), patterns were similar: single‐class use rose from 77.5% to 84.6% (Statistic = 3.80 [SE = 2.18], p| = 0.081), while multi‐class use decreased slightly over time. Use of three or more classes was infrequent (< 5%), and use of four or more classes was nearly absent after FY 2015. For residents without dementia (n = 8436 anticholinergic users), the proportion of residents using one drug class increased from 74.0% to 80.2% (Statistic = 2.57 [SE = 1.47], p = 0.081) and those using two classes decreasing from 21.0% to 16.6% (Statistic = −1.76 [SE = 1.03], p| = 0.088).

Table S4 presents linear tests of temporal trends in anticholinergic use, including overall exposure, class‐specific use, and regimen complexity, stratified by dementia status.

4. Discussion

The objective of this study was to characterize population‐level patterns of anticholinergic prescribing among VA long‐term care residents and evaluate differences by dementia status, drug class, and temporal period over the CLC stay. Across all analyses, several consistent themes emerged. Residents with dementia had substantially greater exposure to neuropsychiatric anticholinergics, particularly second‐generation antipsychotics and antiparkinsonian agents. Whereas residents without dementia were more often prescribed agents for bladder, musculoskeletal, or gastrointestinal indications. Despite longstanding concerns about cognitive harm, first‐generation antihistamine use remained common in both groups, although it was modestly lower among residents with dementia. Patterns of anticholinergic intensity also varied: non‐antihistamine users had the highest cumulative exposure, while antihistamine‐only users had shorter but still meaningful durations of use. Dual users represented a smaller but clinically important subgroup, often receiving three or more anticholinergic agents concurrently. Temporal trends showed gradual reductions in overall anticholinergic use over the 12‐year period, with declines in antihistamines, antidepressants with anticholinergic properties, and antiemetic/antivertigo agents. This gradual decline may likely be due to the increased awareness and deprescribing efforts regarding the use of certain anticholinergics and their adverse side effects, including cognitive impairment. In contrast, prescribing of second‐generation antipsychotics increased modestly, and the distribution of the number of anticholinergic classes per resident shifted toward single‐class exposure. Taken together, these findings underscore persistent and clinically relevant anticholinergic exposure in long‐term care among residents with dementia alongside slow but measurable shifts toward lower complexity and fewer drug classes over time.

Strong anticholinergics, particularly first‐generation antihistamines (e.g., diphenhydramine, hydroxyzine, chlorpheniramine), are generally discouraged in individuals with dementia due to risks for cognitive worsening, delirium, falls, and functional decline [13, 14, 15, 16]. Although antihistamine‐only users had the lowest proportion of weeks spent as users, approximately 10% of residents with dementia were exposed to these medications, either alone or in combination with other anticholinergic agents. These patterns may reflect both clinical necessity and variability in prescribing practices.

Potential alternatives with lower anticholinergic burden have been described in the AGS Beers Criteria and related guidance [17], including second‐generation antihistamines for allergic conditions and non‐pharmacologic or lower‐risk pharmacologic options for sleep, mood, and urinary symptoms [18, 19, 20]. However, the appropriateness of these alternatives depends on individual patient characteristics, symptom severity, prior treatment response, and tolerability. In certain clinical contexts, such as Parkinson's disease, where agents like trihexyphenidyl or benztropine may still be used, clinicians must balance symptom management with potential cognitive risks [21]. While our analysis focused on overall anticholinergic use, it is important to note that some anticholinergic medications may be clinically appropriate for short‐term indications, and our binary classification does not distinguish between appropriate and potentially inappropriate use, which is dependent on clinical context, patient characteristics, and goals of care.

These findings highlight the need to balance clinical necessity with medication safety in dementia care. Rather than applying uniform deprescribing strategies, medication decisions should be individualized, incorporating indication, duration of use, cumulative burden, and patient‐specific goals. Careful medication review may help identify opportunities to reduce unnecessary exposure while maintaining appropriate symptom management.

Anticholinergic prescribing patterns observed in this study may also reflect broader system‐level influences within the VA, including Criteria for Use (CFU) and Prior Authorization (PA) policies implemented during the study period. These policies may have promoted the use of non‐anticholinergic alternatives in certain clinical contexts. For example, mirabegron has been supported within the VA as an alternative for managing overactive bladder, particularly among Veterans with dementia, in whom minimizing anticholinergic burden is a priority [22]. While the present analysis was not designed to evaluate the impact of specific policies, these findings should be interpreted in the context of physician‐, facility‐, and system‐level factors that may influence medication use. Future studies are needed to directly assess the effects of policy changes on prescribing patterns and patient outcomes.

This study draws on a large, national cohort of VA‐CLC residents and distinguishes three mutually exclusive exposure groups. Use of BCMA medication administration records is a major strength of this study and substantially reduces exposure misclassification. Unlike pharmacy dispensing files or insurance claims, which capture what was prescribed or filled, BCMA documents what residents received. This distinction is critical in pharmacoepidemiologic research, where discrepancies between prescribed and administered medications are common. BCMA data also capture over‐the‐counter anticholinergics, including first‐generation antihistamines, which are almost never represented in claims data or self‐reported medication histories used by cohorts such as Alzheimer's Disease Research Centers or large observational aging studies. This allows for a more complete and accurate assessment of anticholinergic exposure, particularly for agents most likely to be undercounted in traditional data sources. Lastly, our cohort consisted of long‐term care residents, and to prevent spurious prescribing patterns, medications needed to have been taken for 4 or more days consecutively.

A key limitation is that we did not assess prior exposure to alternative medications before initiation of first‐generation antihistamines. Evaluating treatment sequencing would require detailed individual‐level health record review, which is not feasible for a study of this scale. As such, we cannot determine whether safer or guideline‐recommended options were attempted before anticholinergic use or whether these agents were selected due to inadequate response or specific clinical indications. Future studies should evaluate treatment sequencing and switching patterns in more granular datasets to better understand clinical decision‐making and the appropriateness of anticholinergic prescribing. In addition, this was a descriptive and observational study design; therefore, we did not evaluate causal relationships. We did not account for dose‐ and potency‐related differences in anticholinergic activity within pharmacologic classes (e.g., quetiapine vs. risperidone vs. olanzapine). As a result, our measures reflect medication counts rather than pharmacologic intensity, and overall anticholinergic burden may vary even when the number of medications remains stable. Additional limitations of this study include: (1) lack of evaluation of clinical indications (e.g., behavioral symptoms, urinary symptoms, or sleep disturbance) and severity of underlying conditions; (2) inability to assess subtype‐specific prescribing or identify the most frequently used medications within each drug subclass; and (3) limited generalizability beyond the predominantly older, male Veteran population residing in VA nursing homes.

4.1. Conclusions and Future Directions

In this large cohort of VA CLC residents, anticholinergic use remained common across every drug class, including among residents with dementia. The overall prevalence was only slightly lower in residents with dementia, which highlights a persistent pattern of high‐risk prescribing in a population that is highly susceptible to cognitive and functional harms. Dual users of antihistamine and non‐antihistamine anticholinergic medications used a greater number of anticholinergic agents and had longer durations of exposure, which may reflect both symptom management and the presence of underlying clinical conditions (e.g., Parkinson's disease or psychiatric disorders) requiring non‐antihistamine anticholinergic therapies for which safer alternatives may be limited. These findings underscore the potential value of anticholinergic stewardship interventions in long‐term care settings. Evidence from prior deprescribing studies provides important context. A recent systematic review reported that, although interventions can reduce anticholinergic exposure, there is insufficient high‐certainty evidence to demonstrate consistent improvements in cognitive outcomes [23]. Similarly, recent implementation work has shown that pharmacist‐led strategies can achieve reductions in anticholinergic burden, but the magnitude of change is often modest and influenced by practical constraints such as clinical complexity and limited time for medication optimization [24]. Collectively, these findings suggest that while deprescribing is achievable, its impact may vary across patients and care settings. Future research should focus on identifying drivers of persistent exposure, incorporating dose and potency into burden metrics, and linking these patterns to clinically meaningful outcomes including cognition, behavioral symptoms, falls, and hospitalization.

4.2. Clinical Practice Implications for Deprescribing

Our results support several immediate clinical actions for reducing anticholinergic exposure in older adults with and without dementia. First‐generation antihistamines should be avoided whenever possible, and safer therapeutic alternatives such as loratadine or mirabegron should be considered for allergic or bladder symptoms. Non‐pharmacologic strategies should be integrated into routine care, particularly for sleep, agitation, and urinary symptoms. Medication reviews should be performed regularly and should account for the cumulative burden, duration of use, and therapeutic necessity of each anticholinergic medication. A patient‐centered approach is essential to balance symptom control with cognitive safety and quality of life.

4.3. Key Clinical Priorities Include

1. Dual and multi‐class users represent individuals with the highest anticholinergic burden and should be prioritized for careful review and deprescribing when appropriate.

2. Deprescribing initiatives should address both central nervous system and peripheral anticholinergic agents because both contribute to overall burden.

3. Ongoing monitoring is needed to support behavioral management while reducing unnecessary exposure and preserving cognitive and functional outcomes.

Author Contributions

Cellas A. Hayes: study concept and design, interpretation of data, preparation of manuscript, funding acquisition. Yongmei Li: study concept and design, analysis, preparation of manuscript. Bocheng Jing: preparation of manuscript. Laura A. Graham: study concept and design, analysis. Chintan Dave: study concept and design, interpretation of data. Michelle C. Odden: study concept and design, interpretation of data, preparation of manuscript, funding acquisition.

Funding

This research was funded by National Institutes of Health, R01AG062568‐03S1 (awarded to M.C.O.). C.A.H. is funded by Burroughs Wellcome Fund Postdoctoral Enrichment Program (PDEP) 1267001, and the NIH NIA Health Enhancement Scientific Program (HESP) U19AG078109.

Disclosure

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: Cohort flow diagram for CLC residents.

Table S1: Categories of medications with anticholinergic properties and their clinical uses based on the 2023 American Geriatrics Society (AGS) Beers Criteria.

Table S2: Temporal trends in anticholinergic medication use by drug class across fiscal years among veterans in long‐term care.

Table S3: Distribution of anticholinergic medication burden: number of anticholinergic drug classes used across fiscal years, stratified by dementia status.

Table S4: Linear tests of temporal trends in anticholinergic use (FY 2007–2019).

Hayes C. A., Li Y., Jing B., Graham L. A., Dave C., and Odden M. C., “Anticholinergic Medication Use in Veterans Affairs Long‐Term Care Residents: Clinical Patterns and Opportunities for Deprescribing,” Journal of the American Geriatrics Society 74, no. 8 (2026): 2285–2296, 10.1111/jgs.70570.