Abstract
Background
Antipsychotic medications in people with dementia should be avoided or limited to treatment courses of fewer than 3–4 months. It is unknown whether discontinuation of long-term antipsychotic medications in this population can improve clinical outcomes. This study evaluated clinical outcomes associated with discontinuation of long-term antipsychotic medication use compared with continued use among individuals with dementia residing in nursing homes.
Methods
This cohort study used Medicare claims data (2013–2022) linked to Minimum Data Set clinical assessment data. The study population included nursing home residents aged ≥ 65 with dementia who had used antipsychotic medications for more than 90 days. Individuals who discontinued antipsychotic medications (gap ≥ 60 days) were matched to continuers using incidence density sampling based on time since first prescription of antipsychotic medications. Propensity score matching was applied to adjust for 86 covariates. Study outcomes included all-cause hospitalization, specific rehospitalization reasons, and all-cause mortality. Hazard ratios (HRs) were estimated using Cox proportional hazards model, and rate differences (RDs) per 1,000 person-years were estimated using generalized linear models.
Results
Two thousand one hundred eleven propensity score-matched pairs were included and followed for a median of 298 days. Antipsychotic medication discontinuation versus continuation was associated with lower risk of all-cause hospitalization (HR [95% CI]: 0.83 [0.72, 0.96]; RD [95% CI]: -55.85 [-97.82, -13.88]) and hospitalization for pneumonia (HR: 0.51 [0.29, 0.90]; RD: -12.85 [-23.26, -2.44]) or urinary tract infections (HR: 0.47 [0.26, 0.83]; RD: -14.22 [-24.62, -3.83]); non-significantly lower risks of hospitalization for falls (HR: 0.88 [0.70, 1.11]; RD: -13.19 [-37.78, 11.41]), delirium (HR: 0.76 [0.57, 1.02]; RD: -19.26 [-39.18, 0.66]), stroke (HR: 0.95 [0.42, 2.16]; RD: -0.31 [-7.06, 6.44]), and all-cause mortality (HR: 0.89 [0.79, 1.02]; RD: -38.23 [-81.01, 4.54]). Subgroup analyses showed the greatest reduction in hospitalization risk was among individuals with moderate-to-severe dementia (HR: 0.78 [95% CI: 0.65, 0.93]; RD: -96.76 [-163.75, -29.77]).
Conclusions
Among nursing home residents with dementia, discontinuing long-term antipsychotic medication use was associated with lower risks of hospitalization, pneumonia, and urinary tract infections, suggesting clinical benefits in this population.
Supplementary Information
The online version contains supplementary material available at https://doi.org/10.1186/s13195-026-02134-0.
Keywords: Dementia, Nursing home residents, Discontinuation, Antipsychotics
People with dementia often experience behavioral and psychological symptoms during the course of their illness. The occurrence and severity of these symptoms, such as agitation, aggression, psychosis, depression, and anxiety, can fluctuate over time and may not always align with the trajectory of cognitive decline. Such symptoms affect 32% to 68% of community-dwelling older adults with dementia [1, 2], with prevalence rising to as high as 92% among nursing home residents [3, 4]. Agitation and aggression are particularly common and are often managed with antipsychotic medications [1, 5]. However, the increased risk of serious adverse drug events, such as sedation, extrapyramidal symptoms, and stroke, may outweigh the modest effectiveness of antipsychotic medication use in this population [6]. Due to these safety concerns, clinical guidelines recommend that antipsychotic medication use in patients with dementia should be avoided [7] or at least restricted to fewer than 3–4 months [8–10].
Given the recommendations of clinical guidelines around antipsychotic medication use in patients with dementia, discontinuation of such drugs may result in important clinical benefits. But the largest randomized controlled trial to date examining antipsychotic medication discontinuation in long-term users with dementia, the Dementia Antipsychotic Withdrawal Trial (DART-AD), found no significant differences in cognitive and functional outcomes, or neuropsychiatric symptoms. Notably, this trial was underpowered, had a high dropout rate (38%), and approximately one-fourth of participants did not initiate the assigned treatment [11]. A systematic review including other smaller randomized trials assessing a similar question found insufficient evidence to evaluate the effectiveness and safety of antipsychotic medication discontinuation in this population due to small sample sizes and outcome heterogeneity [12]. A leading reason for such limited data is that enrolling patients with dementia in randomized trials has practical and ethical challenges, including recruitment barriers, high attrition, and limited generalizability to the more frail populations seen in routine clinical care [13, 14].
To understand the clinical implications of antipsychotic medication discontinuation in patients with dementia, we sought to overcome the inherent challenges of organizing clinical trials in this population by applying a target trial emulation framework [15, 16]. We used routinely collected data from a nationwide cohort of nursing home residents with dementia, a particularly vulnerable population frequently treated with long-term antipsychotic medications [9], to evaluate the patient-level impact of discontinuing versus continuing chronic antipsychotic medication use.
Methods
Database and study population
Medicare fee-for-service Part A (inpatient), B (outpatient), and D (prescription) data and the Minimum Data Set (MDS) were used in this study. Medicare is a federal program that provides health care coverage to individuals aged 65 and older, as well as those with disabilities. The dataset used for this study included all Medicare beneficiaries with dementia and a 16% random sample of those without a dementia diagnosis. The Minimum Data Set is a federally-mandated assessment by the Centers for Medicare & Medicaid Services (CMS) for residents in CMS-certified nursing homes. It collects standardized clinical assessment data on cognitive function, functional status, behavioral and psychological symptoms, and other clinical indicators from the care of nursing home residents. CMS-certified nursing facilities are required to complete these assessments at least quarterly, at admission and discharge, and following significant changes in clinical status.
We then emulated a hypothetical trial comparing discontinuation versus continuation of antipsychotic medications among patients with dementia living in nursing homes with long-term antipsychotic medication use (Supplementary Table 1). Key components of the trial design (i.e., eligibility, treatment strategies, assignment, follow-up, end points, causal contrast, and statistical analysis) and the approaches used to emulate each are described in Supplementary Table 1. In the emulation using observational data, we identified Medicare beneficiaries with dementia who had continuous antipsychotic medication use for more than 90 days between July 2013 and December 2022. Dementia was determined using a validated algorithm requiring a dementia diagnosis in at least two care settings, which has demonstrated a positive predictive value of 77% [17] and is among the highest reported for claims-based dementia algorithms [18, 19]. Continuous antipsychotic medication use was defined as having no prescription gap longer than 30 days between refills.
Next, we identified individuals who discontinued antipsychotic medications, defined as having a prescription gap of ≥ 60 days following the end of continuous exposure. The end of this 60-day gap was designated as the cohort entry date. This 60-day gap definition has a positive predictive value of 81% for discontinuation of atypical antipsychotic medications and 88% for discontinuation of haloperidol, based on discontinuation recorded in electronic health records [20]. Using an incidence density sampling approach, discontinuers were matched to continuers based on the duration from the first antipsychotic medication prescription to the cohort entry date (< 180, 180–239, 240–299, or ≥ 300 days). For each discontinuer, the matching pool consisted of individuals still using antipsychotic medications at the time the discontinuer stopped use. One continuer was randomly selected from this pool without replacement and assigned the same cohort entry date as the matched discontinuer.
Only long-term nursing home residents were further included in the study. We also excluded individuals with any history of psychiatric disorders, Parkinson’s disease and secondary parkinsonism, or end-of-life conditions (e.g., end-stage liver or renal disease, metastatic solid tumors, or receipt of hospice or palliative care) within the 180 days prior to the cohort entry date. We assumed that patients with dementia without other clinical indications for antipsychotic medication use were prescribed long-term antipsychotic medications primarily for the management of behavioral and psychological symptoms. Additionally, to avoid misclassification of discontinuers and continuers due to lack of medication information, those with hospital or skilled nursing facility stays within 60 days before the cohort entry date were excluded. The cohort selection flow is illustrated in Supplementary Figure 1, and detailed operational definitions are provided in Supplementary Method 1.
Study covariates and outcomes
We adjusted for 86 covariates measured during the 180-day period prior to and including the cohort entry date. These covariates included demographics, dementia-related characteristics, comorbidities, medication history, and use of health care resources. To capture overall comorbidity and frailty burden, we incorporated a validated comorbidity score predictive of mortality [21] and a claims-based frailty index validated against several clinical measures such as gait speed, grip strength, hospitalization, mortality, and dementia severity [22–27]. Moderate-to-severe dementia was defined as claims-based frailty index ≥ 0.28, an indicator that has been validated against Functional Assessment Staging Test stages 5–7 in Medicare beneficiaries with possible or probable dementia [26, 27]. Additionally, we assessed the most recent records from the Minimum Data Set within the 180-day baseline period to capture: (1) agitated behavioral (e.g., rejection of care or physical, verbal, or other behavioral symptoms) and psychotic (e.g., hallucinations or delusions) symptoms; (2) activities of daily living; (3) cognitive functioning levels based on daily activities (e.g., memory and decision-making abilities) measured by the Cognitive Functional Scale [28]; and (4) depression severity, assessed using the Total Severity Mood Score, which is calculated as the sum of individual item scores corresponding to specific symptoms of depression based on the Patient Health Questionnaire-9 [29]. These measures were derived from standardized MDS assessment instruments and reflect observed or reported symptoms and functional status within a 7–14 day lookback period at the time of assessment [30]. The median time (interquartile range [IQR]) from the most recent Minimum Data Set record to the cohort entry date was 51 days [24–80] for continuers and 54 days [25–79] for discontinuers. Detailed operational definitions are provided in Supplementary Method 2.
The primary outcome was all-cause hospitalization. Secondary outcomes included hospitalizations for delirium, pneumonia, urinary tract infections (UTIs), stroke, falls (including emergency department visits), as well as all-cause mortality (see validated outcome definitions in Supplementary Method 3). The study patients were followed from the cohort entry date until the occurrence of the study outcome, antipsychotic medication re-initiation for the discontinuers or discontinuation for the continuers, death (for non-fatal outcomes), disenrollment, maximum follow-up period of 365 days, or the end of the data availability.
Statistical analyses
Baseline characteristics between the two study groups were compared using standardized mean differences, with an absolute value of < 0.1 indicating adequate balance. 1:1 nearest neighbor PS matching, with a caliper of 1%, was applied to control for confounding. The probability of antipsychotic medication discontinuation versus continuation (i.e., PS) was estimated using a logistic regression model based on 86 baseline covariates (Supplementary Table 2).
The comparative effect of antipsychotic medication discontinuation versus continuation on study outcomes was estimated using the Cox proportional hazards model and presented as hazard ratios (HRs) with 95% confidence intervals (CIs). The proportional hazards assumption was checked by the scaled Schoenfeld residuals. Rate differences per 1000 person-years (RDs) of study outcomes between study groups were estimated using a generalized linear regression model with normal distribution and identify link [31].
Several sensitivity analyses were performed. First, we included individual components of the claims-based frailty index in the PS model to enhance confounding control. Second, Fine and Gray model was applied to account for competing risk of death. Third, discontinuing antipsychotic medications in individuals with severe behavioral problems can be challenging [8–10]. To further reduce potential confounding by indication for continued antipsychotic medication use, we excluded individuals with severe behavioral problems, defined as having an Agitated and Reactive Behavior Scale ≥ 3 [32], calculated by summing the frequency scores of four Minimum Data Set items related to behavioral problems (physical aggression, verbal aggression, other behavioral symptoms, and rejection of care). Each item is scored from 0 to 3, resulting in a total score ranging from 0 to 12. Fourth, hospitalizations and skilled nursing facility stays are common among older adults. However, individuals with recent hospital or skilled nursing facility stays were excluded from the main analysis, which may limit the generalizability of our findings. We removed this exclusion criterion in the sensitivity analysis, as the average length of these stays is less than 30 days [33, 34], making it unlikely that such admissions would account for the 60-day gap used to define antipsychotic medication discontinuation. Fifth, a 30-day and a 90-day prescription gap to define antipsychotic medication discontinuation, rather than the 60-day gap, was applied in the sensitivity analyses. Lastly, an intention-to-treat analysis without censoring patients who changed the study exposure status was conducted.
Subgroup analyses were performed based on age (< 85 vs. ≥85 years), sex (male vs. female), dementia severity (claims-based frailty index < 0.28 vs. ≥0.28), agitated behavioral symptoms and psychotic symptoms (absent vs. present), and cognitive impairment (intact or mild, moderate, and severe). PS estimation and matching were re-iterated within each subgroup. Heterogeneity between subgroups was assessed using Cochran’s Q-test, with p-values < 0.05 indicating significant heterogeneity.
All analyses were conducted using the Aetion Substantiate® (2025) with R (version 4.4.3) integration [35]. This study was approved by the Mass General Brigham Institutional Review Board (#2023P001111). Informed consent was deemed exempt as this study involved the use of secondary data, which were de-identified and did not involve direct interaction with participants. Our study followed the STROBE reporting guideline for observational studies.
Results
We identified 2,611 APM discontinuers (mean age [SD]: 84.37 [7.76], 36.1% male) and 2,623 continuers (mean age [SD]: 84.64 [7.56], 34.4% male) after the incidence density sampling (Supplementary Figure 2). Of the study cohort, 3,833 (73.2%) had exposure to antipsychotic medications for ≥ 180 days (Table 1). The comorbidity score (2.64 [2.35] vs. 2.75 [2.37]), claims-based frailty index (0.23 [0.08] vs. 0.23 [0.08]), history of comorbidities and comedication use were similar between discontinuers and continuers. For the clinical assessment data, the median time from the proximal Minimum Data Set assessment to the cohort entry date was 52 days (interquartile range: 26–78) for discontinuers and 50 days (IQR: 24–79) for continuers. Compared with continuers, discontinuers had a higher prevalence of disability in locomotion (66.6% vs. 60.7%), walking in a corridor (60.1% vs. 54.4%), and walking in a room (58.0% vs. 52.7%), and a lower prevalence of agitated behavioral symptoms (26.0% vs. 33.2%) and delusions (7.3% vs. 11.8%). After PS matching, no difference in baseline characteristics were observed based on the absolute standard mean difference values. Baseline characteristics are detailed in Supplementary Tables 2, and PS distributions pre- and post-matching are shown in Supplementary Figure 3.
Table 1.
Selected characteristics of the APM discontinuers and continuers before and after propensity score matching
| Baseline characteristicsA | Before propensity score matching | After propensity score matching | ||||
|---|---|---|---|---|---|---|
| Discontinuers (n = 2,611) | Continuers (n = 2,623) | aSMD | Discontinuers (n = 2,111) | Continuers (n = 2,111) | aSMD | |
| Demographics | ||||||
| Age, years | 84.37 (7.76) | 84.64 (7.56) | 0.036 | 84.66 (7.75) | 84.57 (7.61) | 0.012 |
| Gender | ||||||
| Male | 943 (36.1%) | 902 (34.4%) | 0.036 | 736 (34.9%) | 738 (35.0%) | 0.002 |
| Female | 1,668 (63.9%) | 1,721 (65.6%) | 0.036 | 1,375 (65.1%) | 1,373 (65.0%) | 0.002 |
| Exposure of APMs before cohort entry | ||||||
| Duration of APM exposure | ||||||
| <180 days | 699 (26.8%) | 702 (26.8%) | 0.000 | 539 (25.5%) | 544 (25.8%) | 0.005 |
| 180–239 days | 677 (25.9%) | 679 (25.9%) | 0.001 | 554 (26.2%) | 561 (26.6%) | 0.008 |
| 240–299 days | 471 (18.0%) | 475 (18.1%) | 0.002 | 395 (18.7%) | 382 (18.1%) | 0.016 |
| ≥300 days | 764 (29.3%) | 767 (29.2%) | 0.000 | 623 (29.5%) | 624 (29.6%) | 0.001 |
| Type of APMs | ||||||
| Haloperidol | 220 (8.4%) | 191 (7.3%) | 0.043 | 164 (7.8%) | 168 (8.0%) | 0.007 |
| Atypical APMs | 2,611 (100.0%) | 2,623 (100.0%) | NA | 2,111 (100.0%) | 2,111 (100.0%) | NA |
| Combined comorbidity score | 2.64 (2.35) | 2.75 (2.37) | 0.048 | 2.67 (2.34) | 2.68 (2.38) | 0.004 |
| Frailty index | 0.23 (0.08) | 0.23 (0.08) | 0.040 | 0.23 (0.08) | 0.23 (0.08) | 0.004 |
| Dementia-related characteristics | ||||||
| Dementia subtypes | ||||||
| Alzheimer’s disease | 1,519 (58.2%) | 1,577 (60.1%) | 0.040 | 1,249 (59.2%) | 1,242 (58.8%) | 0.007 |
| Others or unspecifiedB | 1,092 (41.8%) | 1,046 (39.9%) | 0.040 | 862 (40.8%) | 869 (41.2%) | 0.007 |
| Moderate-to-severe dementiaC | 1,359 (52.0%) | 1,388 (52.9%) | 0.017 | 1,106 (52.4%) | 1,110 (52.6%) | 0.004 |
| BPSD diagnosis | 834 (31.9%) | 961 (36.6%) | 0.099 | 727 (34.4%) | 721 (34.2%) | 0.006 |
| Comorbidities | ||||||
| Delirium | 513 (19.6%) | 542 (20.7%) | 0.025 | 418 (19.8%) | 422 (20.0%) | 0.005 |
| Insomnia | 78 (3.0%) | 100 (3.8%) | 0.046 | 69 (3.3%) | 66 (3.1%) | 0.008 |
| Pneumonia | 182 (7.0%) | 183 (7.0%) | 0.000 | 151 (7.2%) | 145 (6.9%) | 0.011 |
| Urinary tract infection | 635 (24.3%) | 666 (25.4%) | 0.025 | 518 (24.5%) | 525 (24.9%) | 0.008 |
| Hypertension | 1,698 (65.0%) | 1,730 (66.0%) | 0.019 | 1,376 (65.2%) | 1,360 (64.4%) | 0.016 |
| Hyperlipidemia | 867 (33.2%) | 888 (33.9%) | 0.014 | 698 (33.1%) | 699 (33.1%) | 0.001 |
| Diabetes | 655 (25.1%) | 695 (26.5%) | 0.032 | 538 (25.5%) | 533 (25.2%) | 0.005 |
| Heart failure | 405 (15.5%) | 401 (15.3%) | 0.006 | 320 (15.2%) | 332 (15.7%) | 0.016 |
| Ischemic heart disease | 526 (20.1%) | 504 (19.2%) | 0.023 | 413 (19.6%) | 419 (19.8%) | 0.007 |
| Atrial fibrillation | 362 (13.9%) | 436 (16.6%) | 0.077 | 314 (14.9%) | 326 (15.4%) | 0.016 |
| Stroke | 220 (8.4%) | 233 (8.9%) | 0.016 | 182 (8.6%) | 178 (8.4%) | 0.007 |
| Chronic kidney disease | 349 (13.4%) | 338 (12.9%) | 0.014 | 272 (12.9%) | 283 (13.4%) | 0.015 |
| COPD and bronchiectasis | 304 (11.6%) | 308 (11.7%) | 0.003 | 245 (11.6%) | 241 (11.4%) | 0.006 |
| Cancer | 106 (4.1%) | 93 (3.5%) | 0.027 | 80 (3.8%) | 82 (3.9%) | 0.005 |
| Falls | 216 (8.3%) | 224 (8.5%) | 0.010 | 176 (8.3%) | 175 (8.3%) | 0.002 |
| Hip or pelvic fracture | 115 (4.4%) | 129 (4.9%) | 0.024 | 94 (4.5%) | 95 (4.5%) | 0.002 |
| Gastrointestinal bleeding | 83 (3.2%) | 65 (2.5%) | 0.042 | 56 (2.7%) | 58 (2.7%) | 0.006 |
| Alcohol abuse or dependence | 36 (1.4%) | 57 (2.2%) | 0.060 | 32 (1.5%) | 33 (1.6%) | 0.004 |
| Healthcare utilization | ||||||
| Specialty visits | 813 (31.1%) | 833 (31.8%) | 0.013 | 649 (30.7%) | 657 (31.1%) | 0.008 |
| Office visits, counts | 1.06 (2.29) | 0.97 (2.15) | 0.041 | 0.98 (2.01) | 0.98 (2.21) | 0.002 |
| Emergency department visits, counts | 1.54 (3.70) | 1.56 (3.46) | 0.005 | 1.52 (3.59) | 1.51 (3.49) | 0.001 |
| Hospitalization, days | 1.11 (4.05) | 1.30 (3.88) | 0.046 | 1.13 (3.73) | 1.12 (3.57) | 0.000 |
| Stay in skilled nursing facilities, days | 6.30 (15.39) | 8.14 (18.72) | 0.108 | 6.77 (16.22) | 6.67 (16.45) | 0.006 |
| Agitated behavioral symptomsD | 679 (26.0%) | 870 (33.2%) | 0.157 | 608 (28.8%) | 601 (28.5%) | 0.007 |
| Psychotic symptoms | ||||||
| Delusion | 191 (7.3%) | 309 (11.8%) | 0.152 | 179 (8.5%) | 172 (8.1%) | 0.012 |
| Hallucination | 60 (2.3%) | 91 (3.5%) | 0.070 | 55 (2.6%) | 59 (2.8%) | 0.012 |
| Dependence in activities of daily living | ||||||
| Bathing | 2,477 (94.9%) | 2,462 (93.9%) | 0.044 | 1,992 (94.4%) | 1,983 (93.9%) | 0.018 |
| Dressing | 2,291 (87.7%) | 2,241 (85.4%) | 0.068 | 1,832 (86.8%) | 1,831 (86.7%) | 0.001 |
| Eating | 941 (36.0%) | 822 (31.3%) | 0.100 | 710 (33.6%) | 697 (33.0%) | 0.013 |
| Grooming | 2,255 (86.4%) | 2,215 (84.4%) | 0.054 | 1,803 (85.4%) | 1,803 (85.4%) | 0.000 |
| Toileting | 2,216 (84.9%) | 2,161 (82.4%) | 0.067 | 1,764 (83.6%) | 1,765 (83.6%) | 0.001 |
| Transferring | 2,002 (76.7%) | 1,897 (72.3%) | 0.100 | 1,572 (74.5%) | 1,566 (74.2%) | 0.007 |
| Locomotion on unit | 1,739 (66.6%) | 1,593 (60.7%) | 0.122 | 1,346 (63.8%) | 1,337 (63.3%) | 0.009 |
| Locomotion off unit | 1,739 (66.6%) | 1,593 (60.7%) | 0.122 | 1,346 (63.8%) | 1,337 (63.3%) | 0.009 |
| Walking in corridor | 1,569 (60.1%) | 1,428 (54.4%) | 0.114 | 1,203 (57.0%) | 1,196 (56.7%) | 0.007 |
| Walking in room | 1,514 (58.0%) | 1,383 (52.7%) | 0.106 | 1,157 (54.8%) | 1,151 (54.5%) | 0.006 |
| Cognitive Functional ScaleE | ||||||
| Intact | 207 (7.9%) | 160 (6.1%) | 0.072 | 140 (6.6%) | 139 (6.6%) | 0.002 |
| Mild impairment | 478 (18.3%) | 430 (16.4%) | 0.051 | 375 (17.8%) | 369 (17.5%) | 0.007 |
| Moderate impairment | 1,515 (58.0%) | 1,596 (60.8%) | 0.058 | 1,267 (60.0%) | 1,258 (59.6%) | 0.009 |
| Severe impairment | 387 (14.8%) | 401 (15.3%) | 0.013 | 308 (14.6%) | 321 (15.2%) | 0.017 |
| Depression severityF | ||||||
| No symptoms/minimal | 1,731 (66.3%) | 1,712 (65.3%) | 0.022 | 1,402 (66.4%) | 1,390 (65.8%) | 0.012 |
| Mild | 188 (7.2%) | 184 (7.0%) | 0.007 | 150 (7.1%) | 150 (7.1%) | 0.000 |
| Moderate to severe | 94 (3.6%) | 74 (2.8%) | 0.044 | 67 (3.2%) | 66 (3.1%) | 0.003 |
Abbreviations: APM antipsychotic medication, aSMD Absolute standardized mean difference, NA Not applicable, BPSD Behavioral and psychological symptom of dementia, COPD Chronic obstructive pulmonary disease
AData are presented in number (proportion) or mean (standard deviation)
BInclude vascular dementia, Lewy body dementia, frontotemporal dementia, and unspecified subtypes
CModerate-to-severe dementia were defined using a claims-based frailty index of ≥ 0.280, which was validated against the Functional Assessment Staging Test Stage 5–7 in populations with possible, probable, and confirmed dementia
DAgitated behavioral symptoms were derived from Minimum Data Set (MDS) 3.0, including physical behavioral symptoms directed toward others, verbal behavioral symptoms directed toward others, and other behavioral symptoms not directed toward others during the 7-day look-back period
ECognitive Function Scale (CFS) was derived from MDS 3.0 cognitive assessment data, integrating Brief Interview for Mental Status (BIMS) results and staff-assessed cognitive performance items collected during the 7-day look-back period. Data were missing in 24 (0.9%) discontinuers and 36 (1.4%) continuers before propensity score matching, 21 (1.0%) discontinuers and 24 (1.1%) continuers after propensity score matching
FAssessed using MDS 3.0 mood assessment data (PHQ-9 or staff assessment), based on symptoms present during the 14-day look-back period. The score was missing in 598 (22.9%) discontinuers and 653 (24.9%) continuers before propensity score matching, 492 (23.3%) discontinuers and 505 (23.9%) continuers after propensity score matching
With a median follow-up of 298 days (Supplementary Table 3), reaching the maximum follow-up of 365 days (43.9%) was the most common reason for censoring, followed by a change in study exposure (31.5%). Antipsychotic medication discontinuers showed lower rates of hospitalization (HR [95% CI]: 0.83 [0.72, 0.96]; RD [95% CI]: -55.85 [-97.82, -13.88]), hospitalization for pneumonia (HR: 0.51 [0.29, 0.90]; RD: -12.85 [-23.26, -2.44]), and hospitalization for UTIs (HR: 0.47 [0.26, 0.83]; RD: -14.22 [-24.62, -3.83]) compared with continuers. No significant differences were observed in the rates of hospitalization for falls (HR: 0.88 [0.70, 1.11]; RD: -13.19 [-37.78, 11.41]), hospitalization for delirium (HR: 0.76 [0.57, 1.02]; RD: -19.26 [-39.18, 0.66]), stroke (HR: 0.95 [0.42, 2.16]; RD: -0.31 [-7.06, 6.44]), and all-cause mortality (HR: 0.89 [0.79, 1.02]; RD: -38.23 [-81.01, 4.54]) (Table 2). Results of sensitivity analyses were consistent with those of primary analysis (Supplementary Table 4), although some estimates did not reach statistical significance.
Table 2.
Safety outcomes of antipsychotic medication discontinuers versus continuers from the propensity score-matched analysis
| Discontinuers | Continuers | Discontinuers vs. Continuers | ||||
|---|---|---|---|---|---|---|
| Event number | Event rate /1,000 PYs | Event number | Event rate /1,000 PYs | HR (95% CI) | RD (95% CI) | |
| All-cause hospitalization | 334 | 266.21 | 421 | 322.06 | 0.83 (0.72, 0.96) | -55.85 (-97.82, -13.88) |
| Falls | 130 | 99.10 | 154 | 112.29 | 0.88 (0.70, 1.11) | -13.19 (-37.78, 11.41) |
| Delirium | 82 | 61.31 | 113 | 80.58 | 0.76 (0.57, 1.02) | -19.26 (-39.18, 0.66) |
| Pneumonia | 18 | 13.26 | 37 | 26.12 | 0.51 (0.29, 0.90) | -12.85 (-23.26, -2.44) |
| Urinary tract infection | 17 | 12.56 | 38 | 26.78 | 0.47 (0.26, 0.83) | -14.22 (-24.62, -3.83) |
| Stroke | 11 | 8.10 | 12 | 8.41 | 0.95 (0.42, 2.16) | -0.31 (-7.06, 6.44) |
| All-cause mortality | 427 | 313.56 | 503 | 351.80 | 0.89 (0.79, 1.02) | -38.23 (-81.01, 4.54) |
Abbreviations: PYs Person-years, HR Hazard ratio, CI Confidence interval, RD Rate difference
Across subgroups, hospitalization rates were lower with antipsychotic medication discontinuation among individuals with moderate-to-severe dementia (HR: 0.78 [0.65, 0.93]; RD: -96.76 [-163.75, -29.77]), without agitated behavioral symptoms (0.84 [0.71, 0.99]; RD: -53.97 [-103.94, -3.99]), and without psychotic symptoms (HR: 0.86 [0.74, 0.99]; RD: -48.89 [-94.12, -3.66]). No significant associations were observed in other subgroups, including individuals aged < 85 or ≥ 85 years, males, females, those without moderate-to-severe dementia, those with behavioral or psychological symptoms, and across varying levels of cognitive impairment (Table 3). A significant heterogeneity in the RD of all-cause hospitalization by dementia severity was detected (p = 0.03), with a greater risk reduction associated with antipsychotic medication discontinuation among those with moderate-to-severe dementia (RD: -96.76 [-163.75, -29.77]) compared to those without (RD: -1.15 [-53.84, 51.54]).
Table 3.
Risk of all-cause hospitalization associated with antipsychotic medication discontinuation versus continuation in subgroup analyses
| PS-matched pairs | Event number (rate/1,000 PYs) | Discontinuers vs. Continuers | |||||
|---|---|---|---|---|---|---|---|
| Discontinuers | Continuers | HR (95% CI) | P hetero−geneity | RD (95% CI) | P hetero−geneity | ||
| Age | 0.44 | 0.45 | |||||
| <85 | 915 | 133 (245.82) | 166 (286.48) | 0.86 (0.68, 1.08) | -40.66 (-101.03, 19.71) | ||
| ≥85 | 1,123 | 187 (281.65) | 202 (290.15) | 0.97 (0.80, 1.19) | -8.50 (-65.34, 48.34) | ||
| Gender | 0.88 | 0.67 | |||||
| Male | 659 | 121 (335.81) | 155 (391.42) | 0.86 (0.68, 1.09) | -55.61 (-141.50, 30.28) | ||
| Female | 1,362 | 204 (242.81) | 239 (276.66) | 0.88 (0.73, 1.06) | -33.86 (-82.23, 14.52) | ||
| Moderate-to-severe dementiaA | 0.14 | 0.03 | |||||
| No | 900 | 105 (199.03) | 116 (200.18) | 0.99 (0.76, 1.29) | -1.15 (-53.84, 51.54) | ||
| Yes | 1,059 | 216 (330.67) | 276 (427.43) | 0.78 (0.65, 0.93) | -96.76 (-163.75, -29.77) | ||
| Agitated behavioral symptomsB | 0.36 | 0.29 | |||||
| No | 1,462 | 237 (265.40) | 289 (319.36) | 0.84 (0.71, 0.99) | -53.97 (-103.94, -3.99) | ||
| Yes | 525 | 75 (258.45) | 88 (260.95) | 0.99 (0.73, 1.35) | -2.50 (-82.46, 77.46) | ||
| Psychotic symptomsB | 0.33 | 0.22 | |||||
| No | 1,895 | 312 (279.47) | 382 (328.36) | 0.86 (0.74, 0.99) | -48.89 (-94.12, -3.66) | ||
| Yes | 115 | 18 (215.64) | 20 (179.60) | 1.19 (0.63, 2.25) | 36.04 (-90.92, 163.00) | ||
| Cognitive Functional ScaleB | 0.66 | 0.68 | |||||
| Intact or mild impairment | 447 | 75 (271.21) | 93 (324.20) | 0.84 (0.62, 1.14) | -52.99 (-143.04, 37.06) | ||
| Moderate impairment | 1,219 | 203 (279.64) | 228 (296.48) | 0.95 (0.79, 1.15) | -16.84 (-71.25, 37.57) | ||
| Severe impairment | 252 | 34 (229.82) | 39 (293.82) | 0.79 (0.50, 1.25) | -64.00 (-184.30, 56.29) | ||
Abbreviations: PS Propensity score, PYs Person-years, HR Hazard ratio, CI Confidence interval, RD Rate difference
AModerate-to-severe dementia were defined using a claims-based frailty index of ≥ 0.280, which was validated against the Functional Assessment Staging Test Stage 5–7 in populations with possible, probable, and confirmed dementia
BAgitated behavioral symptoms (including physical, verbal, and other behavioral symptoms), psychotic symptoms (including delusion and hallucination), and Cognitive Functional Scale were derived from Minimum Data Set (MDS) 3.0 assessments during the 7-day look-back period
Discussion
In a nationwide US cohort of nursing home residents with dementia, we found that discontinuing long-term antipsychotic medication use – consistent with clinical guidelines – was associated with lower risks of all-cause hospitalization, as well as hospitalization for pneumonia or urinary tract infection, compared to continued use. This study showed the potential of target trial emulation as a complementary tool that can be used to address inherent limitations in randomized trials and demonstrated some of the meaningful clinical advantages of policies encouraging broader de-prescribing of chronic antipsychotic medications among patients with dementia.
In a prior systematic review of randomized trials evaluating clinical outcomes following the discontinuation versus continuation of long-term antipsychotic medication use in patients with dementia, 10 trials were identified covering a total of 632 participants [12]. The review concluded that discontinuation may not impact overall neuropsychiatric symptoms, as measured by various clinical scales, but the quality of the evidence leading to this conclusion was low due to the randomized trials have a substantial risk of bias, small sample sizes, and limited data on adverse events and mortality. By leveraging a population-based database, our study included an around 25-times larger sample size (4,222 matched individuals versus 165 participants in the largest randomized trial [DART-AD]), allowing for the evaluation of clinically meaningful outcomes associated with antipsychotic medication discontinuation versus continuation and helping to address gaps in existing evidence. Lower rates of falls, delirium, pneumonia, and urinary tract infections, which are common adverse events associated with antipsychotic medication use [36–40], may help explain the reduced risk of all-cause hospitalization observed among antipsychotic medication discontinuers in our study. A similar pattern was reported in a previous study of post-discharge older adults, in which antipsychotic medication discontinuation was associated with a lower risk of rehospitalization and antipsychotic medication-associated adverse effects, compared with continuation [41].
We observed various clinical benefits of antipsychotic medication discontinuation, some of which were especially pronounced among the moderate-to-severe dementia subgroup: in particular, a 22% reduction in hospitalization risk, amounting to a staggering 97 hospitalizations per 1,000 person years. Individuals with advanced dementia, who often have a higher baseline risk of adverse outcomes and greater susceptibility to adverse drug events, may therefore derive important benefits from discontinuing antipsychotic medications.
By contrast, reduced risk of hospitalizations was not observed in the subgroup with agitated behavioral and psychotic symptoms. Previous studies have reported that individuals with more severe neuropsychiatric symptoms may experience behavioral benefits from continued antipsychotic medication use [11, 12, 42]. Ballard et al. reported that individuals with high baseline Neuropsychiatric Inventory scores (> 14) were more likely to develop pronounced behavioral problems following antipsychotic medication discontinuation [42]. Similarly, in the DART-AD trial, participants with high baseline Neuropsychiatric Inventory scores showed a mean Neuropsychiatric Inventory change of -16.9 points favoring antipsychotic medication continuation compared to discontinuation, although the interaction between baseline symptom severity (i.e., Neuropsychiatric Inventory ≤ 14 or > 14) and treatment group was not statistically significant [11]. These subgroup results therefore raise questions about the relative benefits and risks of discontinuing antipsychotic medications among individuals with agitated behavioral and psychotic symptoms. However, these findings based on the subgroup analyses were under-powered and should be viewed as exploratory. Further studies are warranted to corroborate these observations.
Several limitations should be acknowledged. First, although this study emulated a hypothetical trial, it does not overcome limitations of data sources and lack of randomization. Despite that we incorporated clinical assessment variables from the Minimum Data Set (including agitated behavioral and psychotic symptoms, functional status, cognitive impairment, and depression severity), the severity of neuropsychiatric symptoms may not have been fully captured. The intention and reasons for antipsychotic medication discontinuation could not be measured from claims data. The lack of adjustment for these characteristics may lead to potential residual confounding. Second, although we made efforts to use MDS-derived measures in the analysis to adjust for residual confounding and to identify clinically meaningful patient subgroups, a key limitation is that these measures reflect symptom presence and functional status within a 7–14 day lookback period at the time of assessment. Therefore, these measures could only provide a time-limited snapshot of clinical status and may not fully capture fluctuating symptoms occurring outside the assessment window, even though assessments are repeated at quarterly intervals or following clinically significant changes. This may lead to underestimation of the longer-term prevalence of certain conditions. Third, while the gap-based approach used to define discontinuation of antipsychotic medications has been validated in Medicare claims, misclassification of study exposure remains possible. Also, with such gap-based definition, our study included only discontinuers and continuers who survived the specified gap period (i.e., 60 days). Therefore, the findings may not be generalizable to individuals who experienced withdrawal effects or other outcomes within the gap period.
Fourth, because behavioral and psychological symptoms are often under-detected or under-diagnosed in routine clinical practice [43, 44], we assumed that patients with dementia with no documented history of psychiatric disorders or other clinical indications received long-term antipsychotic medications primarily for managing behavioral and psychological symptoms. Fifth, dementia was identified using a validated claims-based algorithm (positive predictive value of 77%) rather than direct clinical assessment, which may introduce diagnostic misclassification. For example, some ICD codes included in the algorithm are non-specific and may reflect age-related cognitive or functional decline rather than a confirmed clinical diagnosis of dementia. Moreover, some individuals with clinically significant cognitive impairment may not have been captured. As a result, the study population primarily reflects nursing home residents with dementia documented in administrative claims, and findings may not be fully generalizable to individuals with undiagnosed or unrecorded dementia. In addition, we lacked reliable information on dementia subtypes, as there are no well-validated claims-based definitions beyond Alzheimer’s disease, and other etiologies (e.g., Lewy body, frontotemporal dementia) could not be consistently reported due to CMS cell size suppression requirements. Sixth, our findings may be limited to nursing home residents with dementia and long-term antipsychotic medication exposure, a population with an urgent need for deprescribing antipsychotic medications due to the known risks associated with prolonged antipsychotic medication use.
In addition, we excluded individuals with selected psychiatric conditions (depression, bipolar disorder, and schizophrenia) and Parkinson’s disease to reduce the likelihood that antipsychotic use reflected indications other than dementia-related behavioral and psychological symptoms and to improve internal validity by minimizing confounding by treatment indication. While this approach strengthened cohort specificity, it may limit generalizability to residents with comorbid psychiatric conditions as well as those with Parkinson’s disease or Parkinsonian syndrome. Lastly, the limited number of events for specific causes of hospitalization precluded subgroup analyses to evaluate whether the observed associations (e.g., pneumonia, UTI) differed across patient subgroups.
Conclusion
Discontinuation of antipsychotic medications was associated with reduced risks of hospitalization, as well as severe pneumonia and urinary tract infections, among patients with dementia living in nursing homes. These findings show that meaningful clinical benefits can result from re-evaluating long-term antipsychotic medication use regularly and encouraging efforts to deprescribe antipsychotic medications among patients with dementia.
Supplementary Information
Acknowledgements
Not applicable.
Authors’ contributions
CTY conducted literature search, wrote the first draft of the manuscript, and performed all analyses. All authors contributed to the study’s conception, design, methodology, result interpretation, and manuscript drafting. All authors approved the final manuscript and had final responsibility for the decision to submit. KJL acquired the data. CTY and KJL had full access to the data as authorized under the data use agreements. KJL and DHK supervised the study and are the guarantors. The corresponding author attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted.
Funding
This study is funded by National Institute on Aging (R01AG075335 and R01AG081412). This work is supported by a grant from the Alzheimer’s Association (AARF-25-1485126) (CTY). ASK’s research is supported by Arnold Ventures. DHK is supported by the grant K24AG073527 from the National Institute on Aging. The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Data availability
A data use agreement is required for each of the data sources in this study. These data use agreements do not permit the authors to share patient level source data or data derivatives with individuals and institutions not covered under the data use agreements. The databases used in this study are accessible to other researchers by contacting the data providers and acquiring data use agreements or licenses.
Declarations
Ethical approval and consent to participate
This study was approved by the Mass General Brigham Institutional Review Board (#2023P001111). Informed consent was deemed exempt as this study involved the use of secondary data, which were de-identified and did not involve direct interaction with participants.
Consent for publication
Not applicable.
Competing interests
KJL has received research grants from Takeda, AbbVie, and UCB for projects unrelated to this study. DHK received personal fees from Pfizer, Alosa Health (ended on 12/31/2022) and VillageMD (ended on 12/13/2022) unrelated to this study. ASK reports serving as an expert witness on behalf of a class of individual plaintiffs in a case against Gilead relating to its tenofovir-containing products, on behalf of a payor in a case against Johnson & Johnson regarding patents and biosimilar Stelara entry, on behalf of the Federal Trade Commission in a case against pharmacy benefits managers related to insulin, and on behalf of a group of state attorneys general and payors in a case against multiple generic manufacturers related to alleged price fixing.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Dae Hyun Kim and Kueiyu Joshua Lin contributed equally to this work as senior authors
References
- 1.Kwon CY, Lee B. Prevalence of Behavioral and Psychological Symptoms of Dementia in Community-Dwelling Dementia Patients: A Systematic Review. Front Psychiatry. 2021;12:741059. 10.3389/fpsyt.2021.741059. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Vaingankar JA, Chong SA, Abdin E, Picco L, Jeyagurunathan A, Seow E, et al. Behavioral and psychological symptoms of dementia: prevalence, symptom groups and their correlates in community-based older adults with dementia in Singapore. Int Psychogeriatr. 2017;29(8):1363–76. [DOI] [PubMed] [Google Scholar]
- 3.Björk S, Juthberg C, Lindkvist M, Wimo A, Sandman PO, Winblad B, et al. Exploring the prevalence and variance of cognitive impairment, pain, neuropsychiatric symptoms and ADL dependency among persons living in nursing homes; a cross-sectional study. BMC Geriatr. 2016;16(1):154. 10.1186/s12877-016-0328-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Bakker TJ, Duivenvoorden HJ, van der Lee J, Trijsburg RW. Prevalence of psychiatric function disorders in psychogeriatric patients at referral to nursing home care–the relation to cognition, activities of daily living and general details. Dement Geriatr Cogn Disord. 2005;20(4):215–24. [DOI] [PubMed] [Google Scholar]
- 5.Kelleher JE, Weedle P, Donovan MD. The Prevalence of and Documented Indications for Antipsychotic Prescribing in Irish Nursing Homes. Pharm (Basel). 2021;9(4):160. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Liperoti R, Pedone C, Corsonello A. Antipsychotics for the treatment of behavioral and psychological symptoms of dementia (BPSD). Curr Neuropharmacol. 2008;6(2):117–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.By the 2023 American Geriatrics Society Beers Criteria® Update Expert Panel. American Geriatrics Society 2023 updated AGS Beers Criteria for potentially inappropriate medication use in older adults. J Am Geriatr Soc. 2023;71(7):2052–81. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Zuidema SU, Johansson A, Selbaek G, Murray M, Burns A, Ballard C, et al. A consensus guideline for antipsychotic drug use for dementia in care homes. Bridging the gap between scientific evidence and clinical practice. Int Psychogeriatr. 2015;27(11):1849–59. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Lind KE, Raban MZ, Georgiou A, Westbrook JI. Duration of Antipsychotic Medication Use by Aged Care Facility Residents With Dementia. Alzheimer Dis Assoc Disord. 2019;33(4):331–8. [DOI] [PubMed]
- 10.Reus VI, Fochtmann LJ, Eyler AE, Hilty DM, Horvitz-Lennon M, Jibson MD, et al. The American Psychiatric Association Practice Guideline on the Use of Antipsychotics to Treat Agitation or Psychosis in Patients With Dementia. Am J Psychiatry. 2016;173(5):543–6. [DOI] [PubMed] [Google Scholar]
- 11.Ballard C, Hanney ML, Theodoulou M, Douglas S, McShane R, Kossakowski K, et al. The dementia antipsychotic withdrawal trial (DART-AD): long-term follow-up of a randomised placebo-controlled trial. Lancet Neurol. 2009;8(2):151–7. [DOI] [PubMed] [Google Scholar]
- 12.Van Leeuwen E, Petrovic M, van Driel ML, De Sutter AI, Vander Stichele R, Declercq T, et al. Withdrawal versus continuation of long-term antipsychotic drug use for behavioural and psychological symptoms in older people with dementia. Cochrane Database Syst Rev. 2018;3(3):CD007726. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Mody L, Miller DK, McGloin JM, et al. Recruitment and retention of older adults in aging research. J Am Geriatr Soc. 2008;56(12):2340–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Kennedy-Martin T, Curtis S, Faries D, Robinson S, Johnston J. A literature review on the representativeness of randomized controlled trial samples and implications for the external validity of trial results. Trials. 2015;16:495. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Wang SV, Schneeweiss S, Initiative RCT-DUPLICATE, et al. Emulation of Randomized Clinical Trials With Nonrandomized Database Analyses: Results of 32 Clinical Trials. JAMA. 2023;329(16):1376–85. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Hernán MA, Wang W, Leaf DE. Target Trial Emulation: A Framework for Causal Inference From Observational Data. JAMA. 2022;328(24):2446–7. [DOI] [PubMed] [Google Scholar]
- 17.Festa N, Price M, Moura LMVR, et al. Evaluation of Claims-Based Ascertainment of Alzheimer Disease and Related Dementias Across Health Care Settings. JAMA Health Forum. 2022;3(4):e220653. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.McCarthy EP, Chang CH, Tilton N, Kabeto MU, Langa KM, Bynum JPW. Validation of Claims Algorithms to Identify Alzheimer’s Disease and Related Dementias. J Gerontol Biol Sci Med Sci. 2022;77(6):1261–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Lee E, Gatz M, Tseng C, Schneider LS, Pawluczyk S, Wu AH, Deapen D. Evaluation of Medicare Claims Data as a Tool to Identify Dementia. J Alzheimers Dis. 2019;67(2):769–78. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Yang CT, Ngan K, Kim DH, Yang J, Liu J, Lin KJ. Establishing a Validation Framework of Treatment Discontinuation in Claims Data Using Natural Language Processing and Electronic Health Records. Clin Pharmacol Ther. 2025;118(1):138–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Gagne JJ, Glynn RJ, Avorn J, Levin R, Schneeweiss S. A combined comorbidity score predicted mortality in elderly patients better than existing scores. J Clin Epidemiol. 2011;64(7):749–59. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Kim DH, Glynn RJ, Avorn J, et al. Validation of a Claims-Based Frailty Index Against Physical Performance and Adverse Health Outcomes in the Health and Retirement Study. J Gerontol Biol Sci Med Sci. 2019;74(8):1271–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Kim DH, Patorno E, Pawar A, Lee H, Schneeweiss S, Glynn RJ. Measuring Frailty in Administrative Claims Data: Comparative Performance of Four Claims-Based Frailty Measures in the U.S. Medicare Data. J Gerontol Biol Sci Med Sci. 2020;75(6):1120–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Kim DH, Schneeweiss S, Lipsitz LA, Glynn R, Rockwood K, Avorn J. Measuring Frailty in Medicare Data: Development and Validation of a Claims-Based Frailty Index. J Gerontol Biol Sci Med Sci. 2018;73(7):980–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Sison SDM, Shi SM, Oh G, Jeong S, McCarthy EP, Kim DH. Claims-Based Frailty Index and Its Relationship With Commonly Used Clinical Frailty Measures. J Gerontol Biol Sci Med Sci. 2024;79(7):glae094. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Park CM, Sison SDM, McCarthy EP, et al. Claims-Based Frailty Index as a Measure of Dementia Severity in Medicare Claims Data. J Gerontol Biol Sci Med Sci. 2023;78(11):2145–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Park CM, McCarthy EP, Jang J, Sison SDM, Kim DH. Validation of Claims-Based Frailty Index for Identifying Moderate-to-Severe Dementia in Medicare Beneficiaries. J Am Med Dir Assoc. 2024;25(10):105176. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Thomas KS, Dosa D, Wysocki A, Mor V. The Minimum Data Set 3.0 Cognitive Function Scale. Med Care. 2017;55(9):e68–72. [DOI] [PMC free article] [PubMed]
- 29.Kroenke K, Spitzer RL, Williams JB. The PHQ-9: validity of a brief depression severity measure. J Gen Intern Med. 2001;16(9):606–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Centers for Medicare & Medicaid Services. Long-Term Care Facility Resident Assessment Instrument 3.0 User’s Manual. Accessed on May 31. 2026 from https://www.cms.gov/files/document/finalmds-30-rai-manual-v1191october2024.pdf.
- 31.Xu Y, Cheung YB, Lam KF, Tan SH, Milligan P. A simple approach to the estimation of incidence rate difference. Am J Epidemiol. 2010;172(3):334–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.McCreedy E, Ogarek JA, Thomas KS, Mor V. The minimum data set agitated and reactive behavior scale. Measuring behaviors in nursing home residents with dementia. J Am Med Dir Assoc. 2019;20(12):1548–52. [DOI] [PMC free article] [PubMed]
- 33.Centers for Medicare & Medicaid Services. Medicare skilled nursing facility (SNF) transparency data (CY2013). Accessed on May 6. 2025 from https://www.cms.gov/newsroom/fact-sheets/medicare-skilled-nursing-facility-snf-transparency-data-cy2013.
- 34.Biniek JF, Cubanski J, Neuman T. Amid the COVID-19 Pandemic, Medicare Spending on Skilled Nursing Facilities Increased More than 4% Despite an Overall Decline in Utilization. Kaiser Family Foundation. Accessed on May 6, 2025 from https://www.kff.org/medicare/issue-brief/amid-the-covid-19-pandemic-medicare-spending-on-skilled-nursing-facilities-increased-more-than-4-despite-an-overall-decline-in-utilization/.
- 35.Wang SV, Verpillat P, Rassen JA, Patrick A, Garry EM, Bartels DB. Transparency and Reproducibility of Observational Cohort Studies Using Large Healthcare Databases. Clin Pharmacol Ther. 2016;99(3):325–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Seppala LJ, Wermelink AMAT, de Vries M, et al. Fall-Risk-Increasing Drugs: A Systematic Review and Meta-Analysis: II. Psychotropics. J Am Med Dir Assoc. 2018;19(4):e37111–7. [DOI] [PubMed] [Google Scholar]
- 37.Fraser LA, Liu K, Naylor KL, et al. Falls and fractures with atypical antipsychotic medication use: a population-based cohort study. JAMA Intern Med. 2015;175(3):450–2. [DOI] [PubMed] [Google Scholar]
- 38.Nosè M, Recla E, Trifirò G, Barbui C. Antipsychotic drug exposure and risk of pneumonia: a systematic review and meta-analysis of observational studies. Pharmacoepidemiol Drug Saf. 2015;24(8):812–20. [DOI] [PubMed] [Google Scholar]
- 39.van Strien AM, Souverein PC, Keijsers CJPW, Heerdink ER, Derijks HJ, van Marum RJ. Association Between Urinary Tract Infections and Antipsychotic Drug Use in Older Adults. J Clin Psychopharmacol. 2018;38(4):296–301. [DOI] [PubMed] [Google Scholar]
- 40.Faure Walker N, Brinchmann K, Batura D. Linking the evidence between urinary retention and antipsychotic or antidepressant drugs: A systematic review. Neurourol Urodyn. 2016;35(8):866–74. [DOI] [PubMed] [Google Scholar]
- 41.Yang CT, Wilkins JM, DiCesare E, Pritchard KT, Chen Q, Zhang Y, et al. Health Outcomes of Discontinuing Antipsychotics After Hospitalization in Older Adults. JAMA Psychiatry. 2025;82(7):671–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Ballard C Thomas A, Fossey J Lee L. A 3-month, randomized, placebo-controlled, neuroleptic discontinuation study in 100 people with dementia: the Neuropsychiatric Inventory median cut-off is a predictor of clinical outcome. J Clin Psychiatry. 2004;65(1):114–9. [DOI] [PubMed] [Google Scholar]
- 43.Anantapong K, Jiraphan A, Aunjitsakul W, Sathaporn K, Werachattawan N, Teetharatkul T, et al. Behavioural and psychological symptoms of people with dementia in acute hospital settings: a systematic review and meta-analysis. Age Ageing. 2025;54(1):afaf013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Buhagiar K, Afzal N, Cosgrave M. Behavioural and psychological symptoms of dementia in primary care: a survey of general practitioners in Ireland. Ment Health Fam Med. 2011;8(4):227–34. [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
Data Availability Statement
A data use agreement is required for each of the data sources in this study. These data use agreements do not permit the authors to share patient level source data or data derivatives with individuals and institutions not covered under the data use agreements. The databases used in this study are accessible to other researchers by contacting the data providers and acquiring data use agreements or licenses.