ABSTRACT
Purpose
To examine associations of current bladder antimuscarinic use with cognitive impairment in adults aged ≥ 75 years with urinary symptoms.
Methods
We analyzed National Health and Nutrition Examination Survey 2011–2014 data. Current use was identified from medications reported during the previous 30 days. Cognitive impairment was defined as CERAD delayed recall < 5, animal fluency < 14, or digit symbol substitution test < 34 among available tests. Survey‐weighted logistic models estimated overall and age‐specific odds ratios (ORs); the primary model included demographic factors and KDM biological age acceleration. Missing KDM data were multiply imputed.
Results
Among 537 participants representing 8 735 786 US adults, 34 reported current use and 375 met the cognitive‐impairment definition (weighted prevalence, 62.5%). The primary overall OR was 2.18 (95% confidence interval [CI], 0.93–5.07; p = 0.069). The use‐by‐age interaction was imprecise (interaction OR, 0.21; 95% CI, 0.03–1.33; p = 0.093). Among adults aged 75–79 years, 10 of 12 users and 130 of 210 nonusers were impaired (weighted, 82.03% vs. 52.37%); the adjusted OR was 7.18 (95% CI, 1.11–46.40; p = 0.039). Among adults in the public‐use group ≥ 80 years, the OR was 1.18 (95% CI, 0.44–3.12; p = 0.733).
Conclusions
The overall estimate was positive but imprecise, and the age‐specific contrast rested on sparse exposure. Because NHANES lacks prior‐use and discontinuation histories, results may reflect prescribing selection and do not support causal or age‐threshold treatment inferences.
Keywords: bladder antimuscarinics, cognitive impairment, NHANES, older adults, overactive bladder, urinary incontinence
Key Points
Among 537 US adults aged ≥ 75 years with urinary symptoms, only 34 reported current bladder antimuscarinic use.
The primary overall association with cognitive impairment was elevated but imprecise (OR, 2.18; 95% CI, 0.93–5.07).
Age‐specific ORs were 7.18 at ages 75–79 and 1.18 in the public‐use ≥ 80‐year group; the interaction remained imprecise.
The 75–79‐year estimate relied on 12 users; influence and continuous‐cognition analyses supported direction, not a precise magnitude.
Cross‐sectional current‐use data cannot separate treatment effects from prescribing selection, prior discontinuation, or reverse causation and should not define treatment age thresholds.
Plain Language Summary
Medicines that relax the bladder by blocking muscarinic receptors can help urgency and leakage but raise concerns about memory in older adults. We studied 537 US adults aged 75 years or older with urinary symptoms; 34 currently used a bladder antimuscarinic. Overall, current users had higher estimated odds of low cognitive‐test performance, but the estimate was uncertain. A much larger estimate appeared among adults aged 75–79 years, based on only 12 users, whereas no similar pattern appeared in the public‐use group aged 80 years or older. Because medication use and cognition were measured at the same time, and NHANES does not show who previously used or stopped these medicines, the age difference could reflect prescribing and discontinuation patterns rather than a drug effect. These results should not define an age cutoff for treatment.
Abbreviations
- BMI
body mass index
- CERAD
Consortium to Establish a Registry for Alzheimer's Disease
- CI
confidence interval
- CKD
chronic kidney disease
- CVD
cardiovascular disease
- KDM
Klemera–Doubal method
- MET
metabolic equivalent of task
- NHANES
National Health and Nutrition Examination Survey
- OAB
overactive bladder
- OR
odds ratio
- PHQ‐9
Patient Health Questionnaire‐9
1. Introduction
Urinary incontinence and overactive bladder symptoms are common in later life and can disrupt sleep, social participation, daily function, and quality of life [1, 2]. Bladder antimuscarinics remain established pharmacologic options, but their peripheral therapeutic action coexists with concern about central cholinergic effects on memory and attention [3, 4]. This concern is clinically salient in adults aged 75 years or older, among whom multimorbidity, polypharmacy, frailty, and reduced cognitive reserve are common. Guidelines and expert statements consequently emphasize individualized prescribing and attention to established anticholinergic cognitive concerns [3, 4, 5].
Cumulative anticholinergic exposure has been associated with incident dementia [6, 7, 8], while experimental evidence suggests that central nervous system penetration differs among bladder antimuscarinics [9, 10, 11]. Other observational studies link anticholinergic use with cognitive impairment, dementia, and brain vulnerability [12, 13, 14]. These questions differ from a 30‐day medication snapshot after earlier treatment decisions. Prescribing, tolerance, persistence, and discontinuation select current users, while nonusers include never and former users. Channeling, reverse causation, depletion of susceptible users, and sparse exposure may therefore distort cross‐sectional age‐specific estimates.
We therefore used National Health and Nutrition Examination Survey (NHANES) 2011–2014 data to estimate the cross‐sectional association between current bladder antimuscarinic use and survey‐defined cognitive impairment among adults aged 75 years or older with urinary incontinence or overactive bladder symptoms. The prespecified overall estimate was retained, but age was the sole primary effect‐modification analysis, comparing adults aged 75–79 years with the NHANES public‐use top‐coded group aged 80 years or older. We also examined raw age‐specific outcome distributions, influence diagnostics, alternative cognitive definitions, and continuous cognitive performance to distinguish the direction of the observed estimates from their stability and precision. Mortality linkage was evaluated only as a supplementary longitudinal analysis.
2. Methods
2.1. Study Design and Population
The cognitive analysis was cross‐sectional. NHANES is a nationally representative survey of the noninstitutionalized US population that uses a complex, multistage probability design [15]. We combined the 2011–2012 and 2013–2014 cycles because both included the relevant urology questionnaire and cognitive assessments [16, 17]. Participants were eligible if they were aged 75 years or older, had urinary incontinence or overactive bladder symptoms, and had ascertainable bladder antimuscarinic exposure and the primary cognitive outcome. NHANES public‐use age is top‐coded at 80 years; accordingly, the ≥ 80‐year group represents the released top‐coded category rather than exact ages above 80. Sequential exclusions are shown in Figure 1. Mortality linkage provided supplementary longitudinal follow‐up.
FIGURE 1.
Participant selection flow diagram. The flow diagram shows sequential exclusions used to construct the analytic sample of adults aged 75 years or older with urinary incontinence or overactive bladder symptoms in NHANES 2011–2014. NHANES, National Health and Nutrition Examination Survey.
2.2. Urinary Symptom Definitions
Stress urinary incontinence was defined by activity‐related leakage without concurrent urgency leakage; urgency incontinence by leakage with urge and inability to reach the toilet; and mixed incontinence by both features. Overactive bladder symptoms came from urgency‐related questionnaire items [1, 3]. The incontinence severity index combined leakage frequency and amount when both were valid and was used only in a restricted sensitivity analysis [18].
2.3. Bladder Antimuscarinic Exposure
Prescription medication data reflect medications participants reported using during the previous 30 days [19]. The exposure was current use of a bladder antimuscarinic identified from the harmonized generic medication name and NHANES drug‐code documentation. Included agents were oxybutynin, tolterodine, solifenacin, darifenacin, trospium, fesoterodine, and flavoxate; mirabegron was not classified as an antimuscarinic. Medication‐specific counts and reported duration of the current medication record are presented in Table S8. The small exposed sample precluded reliable drug‐specific effect estimation. Reported current‐record duration was not treated as cumulative exposure or modeled.
2.4. Cognitive Outcomes
Cognitive performance was assessed using the Consortium to Establish a Registry for Alzheimer's Disease (CERAD) word‐learning tests, animal fluency test, and digit symbol substitution test (DSST) [16, 17, 20, 21]. The primary survey‐based cognitive‐impairment definition was CERAD delayed‐recall score < 5, animal‐fluency score < 14, or DSST score < 34 among available eligible tests. Thus, a participant could meet the definition through low performance in any observed domain. This definition is not a clinical diagnosis of dementia. According to the NHANES cognitive documentation, participants requiring a proxy respondent were not administered the cognitive assessment [16, 17].
A sensitivity analysis required completion of all three component tests, thereby applying the same any‐low‐domain rule to a common testing set. Supportive analyses examined a global cognitive z score and the continuous CERAD immediate‐recall, CERAD delayed‐recall, animal‐fluency, and DSST scores. Because these tests measure related but distinct abilities and use different scales, coefficients were interpreted within each outcome rather than compared numerically across tests. These continuous outcomes were intended to evaluate directional consistency rather than replace the binary primary outcome (Table S9 and Figure S3). Cognitive‐outcome availability and weighted prevalence under alternative definitions are summarized in Table S10.
2.5. Klemera–Doubal Method (KDM) Biological Age and Covariates
The primary adjustment strategy was designed to remain parsimonious given 34 exposed participants. P0 included current bladder antimuscarinic use only. P1 included current use, age group, sex, race/ethnicity, and education. P2, the primary overall model, added KDM biological age acceleration [22]. KDM was used as a summary measure of biological aging while avoiding simultaneous inclusion of its overlapping biomarker and cardiometabolic components in the primary model. Component derivation and availability are detailed in Table S3; KDM adjustment cannot eliminate unmeasured or residual confounding.
Race/ethnicity was categorized as non‐Hispanic White, non‐Hispanic Black, Mexican American, or other, and education as less than high school, high school graduate, or college graduate or above. Table 1 presents the remaining demographic, behavioral, clinical, urinary, medication, frailty, KDM, and cognitive descriptors. These variables were not simultaneously entered in P2; definitions followed the prespecified NHANES workflow and established sources [23, 24, 25, 26, 27, 28, 29].
TABLE 1.
Baseline characteristics of adults aged 75 years or older with urinary incontinence or overactive bladder symptoms, by current bladder antimuscarinic use.
| Characteristic | Overall | No current use | Current use | p |
|---|---|---|---|---|
| Unweighted participants, n | 537 | 503 | 34 | |
| Weighted population, N | 8 735 786 | 8 251 692 | 484 094 | |
| Age, years | 78.69 (0.07) | 78.66 (0.08) | 79.27 (0.16) | 0.002 |
| Age group, n (%) | 0.161 | |||
| 75–79 years | 222 (42.2) | 210 (42.9) | 12 (31.4) | |
| ≥ 80 years | 315 (57.8) | 293 (57.1) | 22 (68.6) | |
| Sex, n (%) | 0.933 | |||
| Male | 253 (39.6) | 240 (39.6) | 13 (38.7) | |
| Female | 284 (60.4) | 263 (60.4) | 21 (61.3) | |
| Race/ethnicity, n (%) | 0.927 | |||
| Non‐Hispanic White | 353 (82.8) | 328 (82.8) | 25 (83.0) | |
| Non‐Hispanic Black | 103 (8.7) | 99 (8.8) | 4 (7.3) | |
| Mexican American | 22 (2.0) | 21 (2.1) | 1 (1.4) | |
| Other race/ethnicity | 59 (6.5) | 55 (6.4) | 4 (8.4) | |
| Education, n (%) | 0.613 | |||
| Less than high school | 77 (9.7) | 74 (10.0) | 3 (5.7) | |
| High school graduate | 202 (36.7) | 191 (36.9) | 11 (33.4) | |
| College graduate or above | 258 (53.6) | 238 (53.2) | 20 (60.9) | |
| Marital status, n (%) | 0.050 | |||
| Married/living with partner | 252 (51.6) | 241 (52.8) | 11 (30.7) | |
| Widowed/divorced/separated | 267 (45.4) | 245 (44.2) | 22 (67.0) | |
| Never married | 18 (3.0) | 17 (3.0) | 1 (2.3) | |
| Poverty income ratio | 2.56 (0.08) | 2.57 (0.09) | 2.39 (0.22) | 0.468 |
| Smoking status, n (%) | 0.184 | |||
| Nonsmoker | 291 (55.5) | 274 (55.9) | 17 (48.1) | |
| Former smoker | 217 (39.5) | 200 (38.8) | 17 (51.9) | |
| Current smoker | 29 (5.0) | 29 (5.3) | 0 (0.0) | |
| Alcohol use, n (%) | 0.447 | |||
| Nondrinker | 120 (23.0) | 114 (23.2) | 6 (18.4) | |
| Former drinker | 170 (29.6) | 159 (29.8) | 11 (26.4) | |
| Mild drinker | 198 (39.3) | 182 (38.5) | 16 (53.3) | |
| Moderate drinker | 33 (5.9) | 32 (6.1) | 1 (1.8) | |
| Heavy drinker | 16 (2.2) | 16 (2.3) | 0 (0.0) | |
| Physical activity, MET‐min/week | 1383.23 (99.04) | 1392.80 (100.61) | 1220.14 (202.06) | 0.379 |
| Body mass index, kg/m2 | 27.96 (0.33) | 27.92 (0.34) | 28.57 (0.96) | 0.512 |
| Urinary symptom phenotype, n (%) | 0.017 | |||
| Overactive bladder symptoms | 322 (60.2) | 295 (58.8) | 27 (83.6) | |
| Urgency urinary incontinence | 119 (22.5) | 114 (23.1) | 5 (11.1) | |
| Mixed urinary incontinence | 96 (17.3) | 94 (18.0) | 2 (5.3) | |
| Urinary incontinence severity index | 5.00 (0.16) | 4.88 (0.17) | 7.01 (0.62) | 0.004 |
| Hypertension, n (%) | 0.870 | |||
| No | 98 (18.9) | 91 (19.0) | 7 (17.7) | |
| Yes | 439 (81.1) | 412 (81.0) | 27 (82.3) | |
| Diabetes mellitus, n (%) | 0.257 | |||
| No | 345 (67.6) | 325 (68.0) | 20 (59.7) | |
| Yes | 192 (32.4) | 178 (32.0) | 14 (40.3) | |
| Cardiovascular disease history, n (%) | 0.534 | |||
| No | 348 (66.8) | 326 (67.1) | 22 (62.3) | |
| Yes | 189 (33.2) | 177 (32.9) | 12 (37.7) | |
| Chronic kidney disease, n (%) | 0.722 | |||
| No | 216 (42.0) | 202 (41.8) | 14 (45.8) | |
| Yes | 321 (58.0) | 301 (58.2) | 20 (54.2) | |
| Hyperlipidemia, n (%) | 0.475 | |||
| No | 108 (18.3) | 102 (18.6) | 6 (14.2) | |
| Yes | 429 (81.7) | 401 (81.4) | 28 (85.8) | |
| Cancer history, n (%) | 0.256 | |||
| No | 381 (66.2) | 362 (66.8) | 19 (56.2) | |
| Yes | 156 (33.8) | 141 (33.2) | 15 (43.8) | |
| Depression symptoms, n (%) | 0.039 | |||
| None‐to‐mild symptoms | 496 (94.3) | 469 (94.7) | 27 (86.6) | |
| Moderate‐to‐severe symptoms | 41 (5.7) | 34 (5.3) | 7 (13.4) | |
| PHQ‐9 score | 3.18 (0.21) | 3.10 (0.22) | 4.45 (0.66) | 0.067 |
| Number of prescription medications | 5.30 (0.17) | 5.22 (0.19) | 6.69 (0.41) | 0.007 |
| Polypharmacy, n (%) | 0.012 | |||
| No | 233 (44.1) | 226 (45.5) | 7 (19.4) | |
| Yes | 304 (55.9) | 277 (54.5) | 27 (80.6) | |
| Frailty index | 0.23 (0.01) | 0.22 (0.01) | 0.27 (0.02) | 0.079 |
| KDM biological age acceleration, years | −33.32 (0.73) | −32.99 (0.72) | −38.91 (2.25) | 0.008 |
| Cognitive impairment, n (%) | 0.108 | |||
| No | 162 (37.5) | 155 (38.2) | 7 (25.1) | |
| Yes | 375 (62.5) | 348 (61.8) | 27 (74.9) | |
| CERAD immediate recall score | 17.20 (0.35) | 17.18 (0.35) | 17.43 (0.93) | 0.779 |
| CERAD delayed recall score | 5.06 (0.12) | 5.03 (0.12) | 5.43 (0.39) | 0.344 |
| Animal fluency test score | 15.05 (0.27) | 15.10 (0.28) | 14.22 (0.96) | 0.370 |
| Digit symbol substitution test score | 41.06 (0.78) | 41.20 (0.81) | 38.68 (2.56) | 0.351 |
| Global cognitive z score | −0.29 (0.05) | −0.28 (0.05) | −0.30 (0.14) | 0.913 |
Note: Values are survey‐weighted means (standard errors) for continuous variables and unweighted n (survey‐weighted %) for categorical variables. Estimates account for NHANES strata, primary sampling units, and the combined 2011–2014 examination weight.
Abbreviations: BMI, body mass index; KDM, Klemera–Doubal method; NHANES, National Health and Nutrition Examination Survey.
2.6. Survey Weights and Statistical Analysis
Analyses incorporated NHANES examination weights, strata, and primary sampling units [15, 30]. Each 2‐year examination weight was divided by 2 for the combined 2011–2014 period. Continuous characteristics are presented as survey‐weighted means with standard errors and categorical characteristics as unweighted counts with survey‐weighted percentages. Baseline p values are descriptive and were not used to select covariates.
Survey‐weighted logistic regression estimated odds ratios (ORs) and 95% confidence intervals (CIs) for cognitive impairment comparing current users with nonusers. P0 was unadjusted; P1 adjusted for age group, sex, race/ethnicity, and education; primary P2 added KDM acceleration. The P2 interaction model added current use × age group; its OR compared the exposure OR at ≥ 80 with that at 75–79 years. Age‐specific models adjusted for sex, education, and KDM acceleration. Given sparse within‐stratum exposure, a sensitivity additionally used binary race/ethnicity adjustment. No alternative age thresholds were evaluated.
Primary KDM analyses handled missing KDM acceleration and feasible components using multiple imputation by chained equations with predictive mean matching, appropriate auxiliary predictors, 50 datasets, and 30 iterations. Survey‐design variables were included as predictors where specified; exposure, outcome, weights, strata, and primary sampling units were not imputed. Survey‐weighted estimates were pooled with Rubin's rules [31]. Table S1 gives full specifications. A separate procedure supported the secondary legacy sensitivity.
Sensitivity analyses included observed‐KDM complete‐case analysis, urinary‐phenotype adjustment, restriction to participants with valid urinary‐severity data, the all‐three‐cognitive‐test‐complete outcome, and binary race‐adjustment of age‐specific models. For adults aged 75–79 years, a leave‐one‐exposed‐out (LOEO) analysis refitted the primary age‐specific model after omitting each exposed participant in turn; this was an influence/stability diagnostic, not a set of independent hypothesis tests. Continuous cognitive outcomes were analyzed with corresponding survey‐weighted linear models. Detailed results appear in Tables S5, S7, S9, and S11 and Figures S1, S3, and S4.
2.7. Supplementary Mortality Analysis
All‐cause and cardiovascular mortality were examined with survey‐weighted Cox regression using linked mortality follow‐up [32]. The preferred M2 model adjusted for age group, sex, race/ethnicity, education, and KDM acceleration. Proportional‐hazards diagnostics were evaluated using weighted residual methods [33]. Mortality sensitivities and diagnostics are reported in Tables S4 and S6 and Figure S2.
3. Results
3.1. Study Population
The analytic sample included 537 adults aged 75 years or older with urinary incontinence or overactive bladder symptoms, representing 8 735 786 US adults (Figure 1). Thirty‐four participants reported current bladder antimuscarinic use, representing 484 094 adults. Overall, 375 participants met the cognitive‐impairment definition; the survey‐weighted prevalence was 62.5%. The all‐three‐test‐complete weighted prevalence was 59.34%. In the broader age‐eligible population, weighted prevalence was 63.35% before the urinary‐symptom restriction and 62.50% after it (Table S10).
Baseline characteristics are shown in Table 1. Current users comprised 12 of 222 participants aged 75–79 years and 22 of 315 in the ≥ 80‐year top‐coded group. Users more often had overactive bladder symptoms and had greater mean urinary‐severity and prescription‐medication counts. These differences are descriptive and underscore the possibility of indication‐related and prescribing‐related selection; baseline p values were not interpreted as evidence of causal imbalance.
3.2. Overall Association With Cognitive Impairment
The unadjusted OR for cognitive impairment was 1.84 (95% CI, 0.88–3.85; p = 0.103) (Table 2 and Figure 2). After adjustment for age group, sex, race/ethnicity, and education, the P1 estimate was 1.98 (95% CI, 0.90–4.38; p = 0.087). In the primary KDM‐centered P2 model, the OR was 2.18 (95% CI, 0.93–5.07; p = 0.069). The adjusted point estimate remained above 1, but its CI was wide and included 1.
TABLE 2.
Survey‐weighted overall and age‐specific associations of current bladder antimuscarinic use with cognitive impairment.
| Panel A. Overall models | |||||
|---|---|---|---|---|---|
| Analysis | N | Exposed, n | OR | 95% CI | p |
| P0—unadjusted | 537 | 34 | 1.84 | 0.88–3.85 | 0.103 |
| P1—demographic/cognitive‐context adjusted | 537 | 34 | 1.98 | 0.90–4.38 | 0.087 |
| P2—primary KDM‐centered adjusted | 537 | 34 | 2.18 | 0.93–5.07 | 0.069 |
| Panel B. Interaction and age‐specific estimates | |||||
|---|---|---|---|---|---|
| Analysis | N | Exposed, n | Effect estimate | 95% CI | p |
| Exposure × age‐group interaction (interaction OR) | 537 | 34 | 0.21 | 0.03–1.33 | 0.093 |
| Age 75–79 years (exposure OR) | 222 | 12 | 7.18 | 1.11–46.40 | 0.039 |
| Age ≥ 80 years (exposure OR) | 315 | 22 | 1.18 | 0.44–3.12 | 0.733 |
Note: P0 included current antimuscarinic use only. P1 included use, age group, sex, race/ethnicity, and education. P2 additionally included KDM biological age acceleration and is the primary overall model. The interaction model added use × age group to P2; its interaction OR is the ratio of the exposure OR at age ≥ 80 years to that at age 75–79 years. Age‐specific models included use, sex, education, and KDM biological age acceleration. Age ≥ 80 years reflects the public‐use top‐code. KDM models used 50 multiply imputed datasets with Rubin pooling. All estimates incorporate strata, PSU, and WTMEC2YR/2.
Abbreviations: CI, confidence interval; KDM, Klemera–Doubal method; MI, multiple imputation; OR, odds ratio; PSU, primary sampling unit.
FIGURE 2.
Overall and age‐specific associations of current bladder antimuscarinic use with cognitive impairment. ORs and 95% CIs were estimated from survey‐weighted logistic regression models. P0 was unadjusted; P1 adjusted for age group, sex, race/ethnicity, and education; and primary P2 additionally adjusted for KDM biological age acceleration. Age‐specific exposure ORs are shown; the p value for the formal current‐use‐by‐age‐group interaction is annotated separately because the interaction OR represents a different estimand. Age ≥ 80 years reflects the NHANES public‐use top‐code. CI, confidence interval; KDM, Klemera–Doubal method; OR, odds ratio.
3.3. Age‐Specific Distributions and Estimates
The formal current‐use‐by‐age‐group interaction estimate was 0.21 (95% CI, 0.03–1.33; p = 0.093), with substantial uncertainty (Table 2). Raw distributions differed markedly by age (Table 3). Among adults aged 75–79 years, 10 of 12 current users and 130 of 210 nonusers met the cognitive‐impairment definition; the corresponding weighted proportions were 82.03% and 52.37%. The adjusted age‐specific OR was 7.18 (95% CI, 1.11–46.40; p = 0.039). This estimate rested on only 12 exposed participants.
TABLE 3.
Cognitive impairment according to age group and current bladder antimuscarinic use.
| Age group | Current antimuscarinic use | Participants, n | Cognitively impaired, n | Weighted impairment, % |
|---|---|---|---|---|
| 75–79 years | No | 210 | 130 | 52.37 |
| 75–79 years | Yes | 12 | 10 | 82.03 |
| ≥ 80 years | No | 293 | 218 | 68.84 |
| ≥ 80 years | Yes | 22 | 17 | 71.59 |
Note: Counts are unweighted; percentages are NHANES survey‐weighted. Cognitive impairment was defined as CERAD delayed‐recall score < 5, animal‐fluency score < 14, or digit‐symbol substitution test score < 34 among available tests. Age ≥ 80 years reflects the NHANES public‐use age top‐code.
Abbreviations: CERAD, Consortium to Establish a Registry for Alzheimer's Disease; NHANES, National Health and Nutrition Examination Survey.
Among adults in the ≥ 80‐year public‐use top‐coded group, 17 of 22 users and 218 of 293 nonusers met the definition, corresponding to weighted proportions of 71.59% and 68.84%. The adjusted OR was 1.18 (95% CI, 0.44–3.12; p = 0.733). This stratum did not show a point estimate comparable with that in adults aged 75–79 years, but its CI was also compatible with both lower and higher odds.
3.4. Sensitivity and Supportive Cognitive Analyses
Overall sensitivity estimates remained above 1 but imprecise, including observed‐KDM complete cases (OR, 1.90; 95% CI, 0.75–4.86) and urinary‐phenotype adjustment (2.14; 0.92–4.96). Severity, all‐three‐test, and legacy results were similar in direction (Table S5).
For adults aged 75–79 years, estimates were 7.67 in the observed‐KDM complete‐case analysis, 7.36 with binary race adjustment, and 6.95 with the all‐three‐test‐complete outcome (Figure 3 and Table S7). Across the 12 LOEO fits, all coefficients remained positive and ORs ranged from 4.64 to 17.49; five of 12 CIs included 1 (Table S11 and Figure S4). Thus, the direction of the age‐75–79 estimate was preserved across influence analyses, although its magnitude and precision were sensitive to individual observations.
FIGURE 3.
Age‐specific robustness and influence analyses for current bladder antimuscarinic use and cognitive impairment. (A) Survey‐weighted age‐specific primary and sensitivity ORs with 95% CIs. (B) AGE75 LOEO influence analysis; detailed 12‐participant LOEO estimates appear in Figure S4 and Table S11. LOEO results are stability diagnostics, not independent hypothesis tests. CI, confidence interval; KDM, Klemera–Doubal method; LOEO, leave‐one‐exposed‐out; OR, odds ratio.
All five continuous estimates at ages 75–79 were negative, including global z score (β, −0.44), animal fluency (−3.18), and DSST (−7.35), but every CI included 0. No consistent pattern appeared at ≥ 80 years (Table S9 and Figure S3).
3.5. Supplementary Mortality Results
In the preferred M2 models, the HR was 1.04 (95% CI, 0.55–1.96; p = 0.891) for all‐cause mortality and 0.51 (95% CI, 0.15–1.72; p = 0.261) for cardiovascular mortality (Table S6 and Figure S2). Only four exposed participants experienced cardiovascular death. Mortality estimates were imprecise, and the cardiovascular proportional‐hazards diagnostic was not considered reliably assessable with so few exposed events (Table S4).
4. Discussion
In this cross‐sectional NHANES study, the primary adjusted estimate for current bladder antimuscarinic use and cognitive impairment was elevated but imprecise. Point estimates differed substantially by age: a large but highly uncertain estimate was observed among adults aged 75–79 years, whereas no comparable elevation was seen in adults in the public‐use group aged 80 years or older. The formal interaction was itself imprecise, and the estimate for adults aged 75–79 years rested on only 12 exposed participants. These features make the overall OR insufficient as a common summary of the age‐specific data and require cautious interpretation of both strata.
The unusual raw distribution among adults aged 75–79 years—10 of 12 exposed participants meeting the cognitive‐impairment definition—generated a large adjusted OR and an extremely wide CI. An approximately sevenfold point estimate would be difficult to reconcile with the broader literature if interpreted as a causal drug effect; the present data do not support that interpretation. LOEO analyses preserved the coefficient direction, but the OR varied from 4.64 to 17.49 and five CIs included 1, demonstrating sensitivity to individual observations. All five continuous cognitive estimates pointed toward poorer performance among users in this stratum, yet every CI included 0. Together, these findings support transparency about the observed direction while showing that the magnitude is not a reliable clinical effect‐size estimate.
Current‐user sampling is the central interpretive limitation. NHANES captures use during the previous 30 days but not dose, adherence, prior discontinued therapy, switching, cumulative exposure, total anticholinergic burden, treatment avoidance, or reversibility. The comparison is therefore not current users versus never users. Nonusers may include never users, former users, patients stopping after intolerance or cognitive concerns, and patients intentionally not offered treatment. These pathways can move susceptible patients out of the current‐user group. Prescribing channeling, depletion of susceptible users, reverse causation, and other selection processes are plausible when symptoms or early cognitive changes influence treatment [4, 5, 34]. Measured urinary, demographic, and biological‐aging adjustment cannot reconstruct these histories or identify the mechanism behind the age contrast.
The ≥ 80‐year result is not evidence of safety or lower biological susceptibility. It may be consistent with treatment avoidance or discontinuation among patients with recognized vulnerability, but NHANES cannot observe those prescribing decisions. Survival to advanced age, differences in the mix of current and former users, the public‐use age top‐code, chance, or unmeasured indication differences may also obscure heterogeneity. The imprecise interaction prevents a firm age‐difference conclusion.
The high impairment frequency warrants direct interpretation. The 375 impaired participants are an unweighted count; survey‐weighted prevalence was 62.5%. This advanced‐age cohort used a sensitive definition based on low performance in any available eligible domain, not a clinical dementia diagnosis. Such a rule is intentionally inclusive and can classify a participant from one low domain even when the other tests are not low. Weighted prevalence was similar before and after urinary‐symptom restriction (63.35% vs. 62.50%) and was 59.34% among all‐three‐test completers, arguing against the restriction as the main explanation. NHANES did not administer cognitive testing to participants requiring proxy respondents [16, 17], so caregiver proxy reporting is unlikely to explain valid included scores. Selective noncompletion or exclusion of people unable to complete testing nevertheless remains possible and may limit generalizability to the most impaired older adults.
KDM provided parsimonious biological‐aging adjustment while avoiding simultaneous inclusion of overlapping physiologic components [22], appropriate with only 34 exposed participants. Sensitivities were directionally similar but uncertain, and KDM cannot eliminate confounding by indication, medication history, frailty, sleep, or healthcare factors. Medication burden remains clinically relevant [35, 36], but the revised analysis supports no polypharmacy or physical‐activity subgroup claim.
Supplementary mortality estimates provide little leverage for interpreting cognition. Both preferred HRs had wide CIs, and only four exposed participants experienced cardiovascular death. They cannot support safety or harm inferences or establish cognition as a more sensitive safety endpoint; proportional‐hazards diagnostics also require sparse‐event caution.
Strengths were the nationally representative complex survey, symptomatic prescribing‐relevant population, explicit raw‐cell reporting, prespecified age question, parsimonious biological‐aging model, valid multiple imputation, and sensitivity, influence, and continuous‐outcome analyses. The revised presentation separates the prespecified overall result from the age‐specific evidence and reports both raw counts and survey‐weighted estimates. Medication distribution and reported duration were also presented transparently in Supplementary Table S8.
Several limitations are fundamental. Exposure and cognition were measured cross‐sectionally, so temporal ordering and causation cannot be established. There were only 34 current users overall and 12 among adults aged 75–79 years; consequently, the magnitude of the latter OR was highly unstable and imprecise, and reliable drug‐specific effects could not be estimated. The medication inventory lacked dose, adherence, cumulative lifetime exposure, prior discontinued use, switching, total anticholinergic burden, and cognitive reversibility after discontinuation. Current nonusers were therefore etiologically heterogeneous, and residual confounding by indication and treatment selection remains likely.
The cognitive outcome was based on survey screening tests rather than a clinical dementia diagnosis, used available domains, and may be affected by selective test noncompletion. The proxy administration rule reduces one proposed source of differential reporting but does not remove selection from outcome availability. Public‐use age was top‐coded at 80 years, preventing finer assessment of the oldest ages. Urinary symptoms were questionnaire‐defined, and unmeasured severity, sleep, functional status, and healthcare factors may remain. Multiple imputation depends on modeling and missing‐at‐random assumptions, even though results were compared with observed‐KDM complete cases. Finally, the mortality analysis was sparse—especially the four exposed cardiovascular deaths—and cannot support safety or harm conclusions.
These findings reinforce established anticholinergic cognitive concerns [3, 4, 5] but should not define an age threshold or generate screening, deprescribing, or switching instructions from this cross‐sectional study. The age contrast is a reason to investigate treatment selection, not to label one group susceptible and the other protected. Clinical decisions should follow established guidance, symptom burden, prior response, co‐medications, goals, and preferences. Prospective studies should capture initiation, dose, adherence, switching, discontinuation, cumulative exposure, reasons for stopping, and repeated cognition, including comparisons with non‐antimuscarinic options [37].
5. Conclusions
In NHANES adults aged 75 years or older with urinary symptoms, the overall adjusted association between current bladder antimuscarinic use and cognitive impairment was positive but imprecise. A large but highly uncertain estimate among adults aged 75–79 years was not seen in the public‐use group aged 80 years or older; sparse exposure, current‐use ascertainment, and the inability to observe prior use or discontinuation make causal interpretation inappropriate. Prospective studies with longitudinal medication histories and repeated cognitive assessment are needed to distinguish treatment effects from prescribing and discontinuation patterns.
Author Contributions
Renjie Huang: writing – original draft, methodology, investigation, formal analysis. Miao Huang: investigation, formal analysis, writing – review and editing. Rui Zhang: conceptualization, supervision, resources, writing – review and editing.
Funding
The authors have nothing to report.
Ethics Statement
The NHANES survey protocol was reviewed and approved by the National Center for Health Statistics Research Ethics Review Board. This secondary analysis used deidentified, publicly available data and was exempt from additional institutional review board approval.
Consent
All participants provided written informed consent.
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
Table S1: Missing‐data and multiple‐imputation specification.
Table S2: Full coefficient results for P1, P2, and the P2 interaction model.
Table S3: KDM component availability, derivation, and public‐use age top‐coding.
Table S4: Proportional‐hazards diagnostics.
Table S5: Robustness analyses for overall cognitive impairment.
Table S6: All‐cause and cardiovascular mortality models and sensitivity analyses.
Table S7: Age‐specific robustness estimates for cognitive impairment.
Table S8: Current antimuscarinic medication distribution and reported duration.
Table S9: Continuous cognitive outcomes overall and by age group.
Table S10: Cognitive‐outcome prevalence and availability.
Table S11: AGE75 leave‐one‐exposed‐out influence diagnostic.
Figure S1: Detailed age‐specific robustness analyses.
Figure S2: All‐cause and cardiovascular mortality analyses.
Figure S3: Continuous cognitive outcomes on outcome‐specific scales.
Figure S4: Detailed AGE75 leave‐one‐exposed‐out influence diagnostic.
Acknowledgments
We thank the National Center for Health Statistics for designing and administering NHANES, and we are grateful to all participants whose data made this research possible. The authors used AI‐assisted language and document‐formatting support during manuscript drafting. The authors reviewed and edited the content and take full responsibility for the submitted work.
Data Availability Statement
The NHANES datasets analyzed in this study are publicly available from the Centers for Disease Control and Prevention at https://wwwn.cdc.gov/nchs/nhanes/.
References
- 1. Irwin D. E., Milsom I., Hunskaar S., et al., “Population‐Based Survey of Urinary Incontinence, Overactive Bladder, and Other Lower Urinary Tract Symptoms in Five Countries: Results of the EPIC Study,” European Urology 50, no. 6 (2006): 1306–1314, 10.1016/j.eururo.2006.09.019. [DOI] [PubMed] [Google Scholar]
- 2. Coyne K. S., Sexton C. C., Thompson C. L., et al., “Prevalence and Effect on Health‐Related Quality of Life of Overactive Bladder in Older Americans: Results From the Epidemiology of Lower Urinary Tract Symptoms Study,” Journal of the American Geriatrics Society 59, no. 8 (2011): 1465–1470, 10.1111/j.1532-5415.2011.03492.x. [DOI] [PubMed] [Google Scholar]
- 3. Cameron A. P., Chung D. E., Dielubanza E., et al., “The AUA/SUFU Guideline on the Diagnosis and Treatment of Idiopathic Overactive Bladder,” Neurourology and Urodynamics 43, no. 8 (2024): 1742–1752, 10.1002/nau.25532. [DOI] [PubMed] [Google Scholar]
- 4. American Geriatrics Society Beers Criteria Update Expert Panel , “American Geriatrics Society 2023 Updated AGS Beers Criteria for Potentially Inappropriate Medication Use in Older Adults,” Journal of the American Geriatrics Society 71, no. 7 (2023): 2052–2081, 10.1111/jgs.18372. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Apostolidis A., Wagg A., Rahnama'i M. S., et al., “SUFU White Paper on Overactive Bladder Anticholinergic Medications and Dementia Risk,” Neurourology and Urodynamics 41, no. 8 (2022): 1928–1933, 10.1002/nau.25037. [DOI] [PubMed] [Google Scholar]
- 6. Gray S. L., Anderson M. L., Dublin S., et al., “Cumulative Use of Strong Anticholinergics and Incident Dementia: A Prospective Cohort Study,” JAMA Internal Medicine 175, no. 3 (2015): 401–407, 10.1001/jamainternmed.2014.7663. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Richardson K., Fox C., Maidment I., et al., “Anticholinergic Drugs and Risk of Dementia: Case‐Control Study,” BMJ 361 (2018): k1315, 10.1136/bmj.k1315. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Coupland C. A. C., Hill T., Dening T., Morriss R., Moore M., and Hippisley‐Cox J., “Anticholinergic Drug Exposure and the Risk of Dementia: A Nested Case‐Control Study,” JAMA Internal Medicine 179, no. 8 (2019): 1084–1093, 10.1001/jamainternmed.2019.0677. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Kay G., Crook T., Rekeda L., et al., “Differential Effects of the Antimuscarinic Agents Darifenacin and Oxybutynin ER on Memory in Older Subjects,” European Urology 50, no. 2 (2006): 317–326, 10.1016/j.eururo.2006.03.057. [DOI] [PubMed] [Google Scholar]
- 10. Staskin D., Kay G., Tannenbaum C., et al., “Trospium Chloride Has no Effect on Memory Testing and Is Assay Undetectable in the Central Nervous System of Older Patients With Overactive Bladder,” International Journal of Clinical Practice 64, no. 9 (2010): 1294–1300, 10.1111/j.1742-1241.2010.02433.x. [DOI] [PubMed] [Google Scholar]
- 11. Callegari E., Malhotra B., Bungay P. J., et al., “A Comprehensive Non‐Clinical Evaluation of the CNS Penetration Potential of Antimuscarinic Agents for the Treatment of Overactive Bladder,” British Journal of Clinical Pharmacology 72, no. 2 (2011): 235–246, 10.1111/j.1365-2125.2011.03961.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Fox C., Richardson K., Maidment I. D., et al., “Anticholinergic Medication Use and Cognitive Impairment in the Older Population: The Medical Research Council Cognitive Function and Ageing Study,” Journal of the American Geriatrics Society 59, no. 8 (2011): 1477–1483, 10.1111/j.1532-5415.2011.03491.x. [DOI] [PubMed] [Google Scholar]
- 13. Risacher S. L., McDonald B. C., Tallman E. F., et al., “Association Between Anticholinergic Medication Use and Cognition, Brain Metabolism, and Brain Atrophy in Cognitively Normal Older Adults,” JAMA Neurology 73, no. 6 (2016): 721–732, 10.1001/jamaneurol.2016.0580. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Welk B. and McArthur E., “Dementia Associated With Anticholinergic Drugs Used for Overactive Bladder: A Nested Case‐Control Study Using the French National Medical‐Administrative Database,” Journal of Urology 208, no. 4 (2022): 863–871, 10.1097/JU.0000000000002804. [DOI] [PubMed] [Google Scholar]
- 15. Centers for Disease Control and Prevention, National Center for Health Statistics , “National Health and Nutrition Examination Survey: Analytic Guidelines, 2011–2014 and 2015–2016,” accessed June 21, 2026, https://wwwn.cdc.gov/nchs/nhanes/analyticguidelines.aspx.
- 16. Centers for Disease Control and Prevention, National Center for Health Statistics , “NHANES 2011–2012 Cognitive Functioning Questionnaire Documentation,” accessed June 21, 2026, https://wwwn.cdc.gov/Nchs/Nhanes/2011‐2012/CFQ_G.htm.
- 17. Centers for Disease Control and Prevention, National Center for Health Statistics , “NHANES 2013–2014 Cognitive Functioning Questionnaire Documentation,” accessed June 21, 2026, https://wwwn.cdc.gov/Nchs/Nhanes/2013‐2014/CFQ_H.htm.
- 18. Sandvik H., Seim A., Vanvik A., and Hunskaar S., “A Severity Index for Epidemiological Surveys of Female Urinary Incontinence: Comparison With 48‐Hour Pad‐Weighing Tests,” Neurourology and Urodynamics 19, no. 2 (2000): 137–145, 10.1002/(SICI)1520-6777(2000)19:2<137::AID-NAU4>3.0.CO;2-G. [DOI] [PubMed] [Google Scholar]
- 19. Centers for Disease Control and Prevention, National Center for Health Statistics , “NHANES Prescription Medications Documentation, 2011–2012,” accessed June 21, 2026, https://wwwn.cdc.gov/Nchs/Nhanes/2011‐2012/RXQ_RX_G.htm.
- 20. Morris J. C., Heyman A., Mohs R. C., et al., “The Consortium to Establish a Registry for Alzheimer's Disease (CERAD). Part I. Clinical and Neuropsychological Assessment of Alzheimer's Disease,” Neurology 39, no. 9 (1989): 1159–1165, 10.1212/WNL.39.9.1159. [DOI] [PubMed] [Google Scholar]
- 21. Carone D. A. E., Strauss E. M. S., Sherman O., and Spreen , “A Compendium of Neuropsychological Tests: Administration, Norms, and Commentary,” Applied Neuropsychology 14, no. 1 (2007): 62–63, 10.1080/09084280701280502. [DOI] [Google Scholar]
- 22. Klemera P. and Doubal S., “A New Approach to the Concept and Computation of Biological Age,” Mechanisms of Ageing and Development 127, no. 3 (2006): 240–248, 10.1016/j.mad.2005.10.004. [DOI] [PubMed] [Google Scholar]
- 23. Piercy K. L., Troiano R. P., Ballard R. M., et al., “The Physical Activity Guidelines for Americans,” Journal of the American Medical Association 320, no. 19 (2018): 2020–2028, 10.1001/jama.2018.14854. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Whelton P. K., Carey R. M., Aronow W. S., et al., “2017 ACC/AHA/AAPA/ABC/ACPM/AGS/APhA/ASH/ASPC/NMA/PCNA Guideline for the Prevention, Detection, Evaluation, and Management of High Blood Pressure in Adults,” Hypertension 71, no. 6 (2018): e13–e115, 10.1161/HYP.0000000000000065. [DOI] [PubMed] [Google Scholar]
- 25. American Diabetes Association Professional Practice Committee , “2. Diagnosis and Classification of Diabetes: Standards of Care in Diabetes‐2024,” Diabetes Care 47 (2024): S20–S42, 10.2337/dc24-S002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults , “Executive Summary of the Third Report of the National Cholesterol Education Program Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III),” Journal of the American Medical Association 285, no. 19 (2001): 2486–2497, 10.1001/jama.285.19.2486. [DOI] [PubMed] [Google Scholar]
- 27. Kroenke K., Spitzer R. L., and Williams J. B. W., “The PHQ‐9: Validity of a Brief Depression Severity Measure,” Journal of General Internal Medicine 16, no. 9 (2001): 606–613, 10.1046/j.1525-1497.2001.016009606.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Levey A. S., Stevens L. A., Schmid C. H., et al., “A New Equation to Estimate Glomerular Filtration Rate,” Annals of Internal Medicine 150, no. 9 (2009): 604–612, 10.7326/0003-4819-150-9-200905050-00006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Rockwood K. and Mitnitski A., “Frailty in Relation to the Accumulation of Deficits,” Journals of Gerontology. Series A, Biological Sciences and Medical Sciences 62, no. 7 (2007): 722–727, 10.1093/gerona/62.7.722. [DOI] [PubMed] [Google Scholar]
- 30. Lumley T., “Analysis of Complex Survey Samples,” Journal of Statistical Software 9, no. 8 (2004): 1–19, 10.18637/jss.v009.i08. [DOI] [Google Scholar]
- 31. van Buuren S. and Groothuis‐Oudshoorn K., “Mice: Multivariate Imputation by Chained Equations in R,” Journal of Statistical Software 45, no. 3 (2011): 1–67, 10.18637/jss.v045.i03. [DOI] [Google Scholar]
- 32. Centers for Disease Control and Prevention, National Center for Health Statistics , “Public‐Use Linked Mortality Files,” accessed June 21, 2026, https://www.cdc.gov/nchs/data‐linkage/mortality‐public.htm.
- 33. Grambsch P. M. and Therneau T. M., “Proportional Hazards Tests and Diagnostics Based on Weighted Residuals,” Biometrika 81, no. 3 (1994): 515–526, 10.1093/biomet/81.3.515. [DOI] [Google Scholar]
- 34. Musa M. K., Saga S., Blekken L. E., Harris R., Goodman C., and Norton C., “Toileting Difficulties in Older People With and Without Dementia Receiving Formal In‐Home Care: A Longitudinal Study,” Nursing Open 6, no. 3 (2019): 1055–1066, 10.1002/nop2.289. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Maher R. L., Hanlon J., and Hajjar E. R., “Clinical Consequences of Polypharmacy in Elderly,” Expert Opinion on Drug Safety 13, no. 1 (2014): 57–65, 10.1517/14740338.2013.827660. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Hilmer S. N., Mager D. E., Simonsick E. M., et al., “A Drug Burden Index to Define the Functional Burden of Medications in Older People,” Archives of Internal Medicine 167, no. 8 (2007): 781–787, 10.1001/archinte.167.8.781. [DOI] [PubMed] [Google Scholar]
- 37. Griebling T. L., Campbell N. L., Mangel J., et al., “Effect of Mirabegron on Cognitive Function in Elderly Patients With Overactive Bladder: MoCA Results From a Phase 4 Randomized, Placebo‐Controlled Study (PILLAR),” BMC Geriatrics 20, no. 1 (2020): 109, 10.1186/s12877-020-1474-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table S1: Missing‐data and multiple‐imputation specification.
Table S2: Full coefficient results for P1, P2, and the P2 interaction model.
Table S3: KDM component availability, derivation, and public‐use age top‐coding.
Table S4: Proportional‐hazards diagnostics.
Table S5: Robustness analyses for overall cognitive impairment.
Table S6: All‐cause and cardiovascular mortality models and sensitivity analyses.
Table S7: Age‐specific robustness estimates for cognitive impairment.
Table S8: Current antimuscarinic medication distribution and reported duration.
Table S9: Continuous cognitive outcomes overall and by age group.
Table S10: Cognitive‐outcome prevalence and availability.
Table S11: AGE75 leave‐one‐exposed‐out influence diagnostic.
Figure S1: Detailed age‐specific robustness analyses.
Figure S2: All‐cause and cardiovascular mortality analyses.
Figure S3: Continuous cognitive outcomes on outcome‐specific scales.
Figure S4: Detailed AGE75 leave‐one‐exposed‐out influence diagnostic.
Data Availability Statement
The NHANES datasets analyzed in this study are publicly available from the Centers for Disease Control and Prevention at https://wwwn.cdc.gov/nchs/nhanes/.