This cohort study investigates changes in use of common medications after an incident dementia diagnosis in a national sample of Medicare beneficiaries.
Key Points
Question
Is an incident dementia diagnosis associated with changes in use of long-term medications?
Findings
In this cohort study of 266 675 Medicare Part D beneficiaries with incident dementia and 266 675 matched control individuals, dementia diagnosis was associated with a modestly greater rate of decline of cardiometabolic and anticholinergic medication use but increased central nervous system–active medication use and overall mean medication counts.
Meaning
The findings suggest that an incident dementia diagnosis may be an underused clinical opportunity to reduce polypharmacy by deprescribing medications with high safety risks, with limited likelihood of benefit, or that may be associated with risk of further impairing cognition.
Abstract
Importance
Dementia is a life-altering diagnosis that may affect medication safety and goals for chronic disease management.
Objective
To examine changes in medication use following an incident dementia diagnosis among community-dwelling older adults.
Design, Setting, and Participants
In this cohort study of adults aged 67 years or older enrolled in traditional Medicare and Medicare Part D, patients with incident dementia diagnosed between January 2012 and December 2018 were matched to control patients based on demographics, geographic location, and baseline medication count. The index date was defined as the date of first dementia diagnosis or, for controls, the date of the closest office visit. Data were analyzed from August 2021 to June 2023.
Exposure
Incident dementia diagnosis.
Main Outcomes and Measures
The main outcomes were overall medication counts and use of cardiometabolic, central nervous system (CNS)–active, and anticholinergic medications. A comparative time-series analysis was conducted to examine quarterly changes in medication use in the year before through the year following the index date.
Results
The study included 266 675 adults with incident dementia and 266 675 control adults; in both groups, 65.1% were aged 80 years or older (mean [SD] age, 82.2 [7.1] years) and 67.8% were female. At baseline, patients with incident dementia were more likely than controls to use CNS-active medications (54.32% vs 48.39%) and anticholinergic medications (17.79% vs 15.96%) and less likely to use most cardiometabolic medications (eg, diabetes medications, 31.19% vs 36.45%). Immediately following the index date, the cohort with dementia had a greater increase in mean number of medications used (0.41 vs −0.06; difference, 0.46 [95% CI, 0.27-0.66]) and in the proportion of patients using CNS-active medications (absolute change, 3.44% vs 0.79%; difference, 2.65% [95% CI, 0.85%-4.45%]) owing to an increased use of antipsychotics, antidepressants, and antiepileptics. The cohort with dementia also had a modestly greater decline in use of anticholinergic medications (quarterly change in use, −0.53% vs −0.21%; difference, −0.32% [95% CI, −0.55% to −0.08%]) and most cardiometabolic medications (eg, quarterly change in antihypertensive use: –0.84% vs –0.40%; difference, –0.44% [95% CI, –0.64% to –0.25%]). One year after diagnosis, 75.2% of the cohort with dementia were using 5 or more medications (2.8% increase).
Conclusions and Relevance
In this cohort study of Medicare Part D beneficiaries, following an incident dementia diagnosis, patients were more likely to initiate CNS-active medications and modestly more likely to discontinue cardiometabolic and anticholinergic medications compared with the control group. These findings suggest missed opportunities to reduce burdensome polypharmacy by deprescribing long-term medications with high safety risks or limited likelihood of benefit or that may be associated with impaired cognition.
Introduction
Over 5 million US individuals have been diagnosed with Alzheimer disease or related dementias, with hundreds of thousands of patients newly diagnosed annually.1 Dementia is a life-changing diagnosis that may alter older adults’ ability to manage medications and engage in medical treatment decisions. The majority of older adults with dementia also face multiple other chronic conditions,2 and worsening cognitive impairment may alter the risk-benefit balance of medications taken for these conditions.
Clinical practice guidelines typically recommend tailoring treatments by patient comorbidity, likelihood of benefit, and patient goals.3,4,5,6 A new dementia diagnosis can substantially alter goals of chronic disease management because of shortened life expectancy, an increase in the potential risks associated with medications, and for some patients, shifts in their goals of care, with maximizing quality of life taking precedence over life-extending therapies. Thus, recognition of dementia in the outpatient setting should prompt reconsideration of the risks and benefits of all medications. In particular, polypharmacy, the use of potentially inappropriate medications, and the use of central nervous system (CNS)–active medications are common in older adults with dementia and may be associated with further impairment of cognition, leading to worse patient outcomes.7,8,9,10 Little is known, however, about the association between an incident dementia diagnosis and medication use patterns. Understanding the chronology of medication changes following a first dementia diagnosis may identify targets for deprescribing interventions to reduce preventable medication-related harms. Medications that can affect cognition (eg, CNS-active medications and anticholinergic medications) and those for which dosing errors may lead to significant harm (eg, insulins and anticoagulant medications) are potentially high-yield targets for interventions, as is polypharmacy in general because patients with cognitive impairment may have more difficulty managing complex regimens.11,12,13
Available data on medication use among persons with dementia are largely cross-sectional7,8,9,10 or focused on individuals residing in nursing homes.14,15,16,17 The few studies that have assessed medication use patterns directly following a dementia diagnosis have focused on individual conditions.18,19 Given the pervasiveness of polypharmacy among individuals with dementia in the US, this analysis sought to assess prescribing patterns in a national sample of traditional Medicare beneficiaries, comparing prescription fill patterns of common CNS-active, anticholinergic, and cardiometabolic medications following an incident dementia diagnosis with fill patterns of a matched control group of older adults without dementia. We hypothesized that an incident dementia diagnosis would be associated with increased rates of deprescribing of medications overall and CNS-active medications in particular.
Methods
In this cohort study, we analyzed a 20% national sample of Centers for Medicare & Medicaid Services (CMS) administrative and pharmacy claims for the period from January 1, 2010, to December 31, 2019. For each year, the sample included all persons at least 67 years of age who were continuously enrolled in Medicare Parts A, B, and D in the 24 months prior to and 12 months following cohort entry (or until death). We excluded beneficiaries enrolled in Medicare Advantage plans, for whom diagnoses from claims were not available. This project was approved by the CMS privacy board and the Harvard Medical School institutional review committee, which waived the requirement for obtaining informed consent because all administrative claims data were deidentified. The study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.
Incident Dementia Cohort
Incident dementia was identified using validated International Classification of Diseases, Ninth Revision, Clinical Modification and International Statistical Classification of Diseases, Tenth Revision, Clinical Modification diagnosis codes from the CMS Chronic Conditions Data Warehouse (CCW) algorithm for Alzheimer disease, related disorders, or senile dementia.20,21,22 While the CCW algorithm includes diagnosis codes from any outpatient or inpatient claim, prior studies have found that dementia codes billed in the inpatient setting may be inaccurate.23 Furthermore, codes assigned only to services (eg, laboratory tests) may reflect testing to rule out conditions. Thus, we included only diagnoses that were associated with ambulatory office visits or neurology-psychiatry testing visits identified by Current Procedural Terminology codes (eTable 1 in Supplement 1). The date of the incident dementia diagnosis (index date) was defined as the date of the first dementia diagnosis code, requiring 24 previous months without a dementia diagnosis.
Matched Control Cohort
We matched patients with an incident dementia diagnosis to control patients using coarsened exact matching, a nonparametric method that matches based on bins of characteristics.24 Matching variables included age (in 2-year bins); sex; race and ethnicity (Black, Hispanic, White, or other [included American Indian/Alaska Native, Asian/Pacific Islander, unspecified, and unknown]); Medicaid dual eligibility; medication count (0-2, 3-5, 6-8, 9-11, 12-14, 15-17, 18-20, or ≥21) at the index date, which was defined for controls as the date of the closest visit; geographic region; and calendar year. Race and ethnicity were included given the racial and ethnic disparities in dementia care and diagnosis25 and were identified using the Research Triangle Institute race code. We chose not to match on individual comorbidities, as many comorbidities are risk factors for dementia (eg, cardiometabolic conditions) while others may represent diagnoses initially suspected prior to the diagnosis of dementia (eg, psychiatric conditions). As an alternative, we used the total number of prescribed medications at cohort entry as a summary measure of comorbidity burden, which has been shown to be a valid proxy measure of comorbidities.26,27 Patients were matched by most granular possible geographic region, first by zip code, then by county, and then by state. Patients who developed dementia after the index date were eligible to be included in the control cohort at an earlier time point if time intervals did not overlap.
Medication Use
We calculated medication counts including all oral, inhaled, injected, and transdermal medications and excluding medical supplies and intravenous and topical treatments. We examined medication use overall and in 3 broad categories: cardiometabolic, CNS-active, and anticholinergic (eTable 2 in Supplement 1). Cardiometabolic classes included antihypertensives, lipid-lowering medications, prescription antiplatelets, oral anticoagulants, insulins, and noninsulin diabetes medications. These classes were chosen as among the most commonly used classes for prevention and treatment of cardiovascular disease and diabetes with variable medication risk profiles. Central nervous system–active medications included antidepressant, antipsychotic, antiepileptic, benzodiazepine, nonbenzodiazepine benzodiazepine receptor agonist hypnotic, and opioid medications.10 These CNS-active classes were chosen based on their inclusion in the Beers Criteria6 and other guidelines of high-risk medications associated with impaired cognition as well as prior pharmacoepidemiology studies of patients with dementia.9,10 Anticholinergic medications were identified based on the Beers Criteria.6
We assessed medication use in 90-day intervals (quarters) beginning 360 days prior to the index date and extending 360 days following the index date. For each time point, medication use was defined as at least 1 pharmacy fill with at least a 28-day supply for any medication in the class in the preceding 90 days. A minimum 28-day supply was required to avoid inclusion of brief 1-time prescriptions, and a single-fill requirement was chosen to allow for patients filling 90-day supply prescriptions. Patients were censored if they died during the quarter.
Statistical Analysis
We compared descriptive characteristics for both cohorts using standardized mean differences (SMDs). We then determined mean medication counts of overall medications, antidementia medications, CNS-active medications, and all other medications at each time point as well as the proportion of beneficiaries prescribed each medication class. We assessed changes in polypharmacy using thresholds of more than 5 and more than 10 medications.
We first graphically examined trends in medication use. We then conducted a comparative interrupted time series (CITS) analysis.28 The CITS design assumed that the preintervention to postintervention change in both the intercept and slope observed in the control group would have been observed in the cohort with incident dementia, absent the diagnosis of dementia (ie, interruption). Unlike a difference-in-differences framework, the CITS design does not require parallel trends in treatment and control groups prior to the index date but does require a linear model to capture the preintervention to postintervention change in the outcome in the control group. We estimated ordinary least-squares regressions with Newey-West SEs to account for autocorrelation29 and included variables for study time point, an indicator for whether the time point was before or after the index date, an indicator for the cohort, and interactions between each variable. The preperiod (quarter −3 to quarter 0) and postperiod (quarter 1 to quarter 4) were defined relative to the index date. Differences between cohorts are reported for baseline medication class use, preintervention and postintervention trends (slopes), the immediate level change following the index date, and the change in postperiod slopes compared with the preperiod slopes.
Finally, because changes in proportions of medication use can be attributable to different trajectories of use (eg, patients starting or stopping medications), we examined use of any CNS-active medication and use of each specific medication class by patients at the index date and the final study quarter, classifying medication use patterns as never used, continued, new starts, or stops. Data were analyzed from August 2021 to June 2023. All estimates are presented with 95% CIs and were calculated using SAS, version 9.2 (SAS Institute Inc) or Stata, version 17.1 (StataCorp LLC).
Results
From January 2012 through December 2018, we identified 271 070 patients with incident dementia, of whom 266 675 (98.4%) were matched to controls. Most patients with incident dementia were 80 years or older (65.1%; mean [SD] age, 82.2 [7.1] years). A total of 67.8% were female; 32.2%, male; 8.0%, Black; 6.0%, Hispanic; 82.0%, White; and 4.0%, other race and ethnicity. One-third (32.2%) were Medicaid dual-eligible prior to diagnosis (Table). By study design, demographic characteristics and baseline medication counts were identical for the matched control group. There were minimal differences in baseline medical comorbidities between cohorts, with all SMDs below 0.10 except for epilepsy and stroke or transient ischemic attack. There were larger differences in mental health and neurologic comorbidities, with a greater proportion of patients with incident dementia having a history of stroke (11.9% vs 6.3%; SMD, 0.19), depression (31.5% vs 18.6%; SMD, 0.30), or schizophrenia (10.3% vs 2.8%; SMD, 0.31). By the end of follow-up, 11.4% of the cohort with dementia and 6.0% of the control cohort were censored due to death.
Table. Cohort Characteristics.
| Characteristic | Cohort, % | Standardized mean difference | |
|---|---|---|---|
| Incident dementia (n = 266 675) | Matched control (n = 266 675)a | ||
| Age category, y | |||
| 67-69 | 1.7 | 1.8 | 0 |
| 70-74 | 13.2 | 13.0 | |
| 75-79 | 19.9 | 20.1 | |
| 80-84 | 23.8 | 23.6 | |
| ≥85 | 41.5 | 41.5 | |
| Race and ethnicity | |||
| Black | 8.0 | 8.0 | 0 |
| Hispanic | 6.0 | 6.0 | |
| White | 82.0 | 82.0 | |
| Otherb | 4.0 | 4.0 | |
| Sex | |||
| Female | 67.8 | 67.8 | 0 |
| Male | 32.2 | 32.2 | |
| Medicaid eligible | 33.2 | 33.2 | 0 |
| Region | |||
| New England | 19.7 | 19.7 | 0 |
| Midwest | 23.3 | 23.3 | |
| South | 40.9 | 40.9 | |
| West | 16.0 | 16.0 | |
| Year of study entry | |||
| 2012 | 11.5 | 11.5 | 0 |
| 2013 | 13.7 | 13.7 | |
| 2014 | 12.5 | 12.5 | |
| 2015 | 14.1 | 14.1 | |
| 2016 | 16.8 | 16.8 | |
| 2017 | 16.0 | 16.0 | |
| 2018 | 15.4 | 15.4 | |
| Baseline medication count | |||
| 0-2 | 4.2 | 4.2 | 0 |
| 3-5 | 12.3 | 12.3 | |
| 6-8 | 19.3 | 19.3 | |
| 9-11 | 19.8 | 19.8 | |
| 12-14 | 16.0 | 16.0 | |
| 15-17 | 11.4 | 11.4 | |
| 18-20 | 7.2 | 7.2 | |
| >20 | 9.9 | 9.9 | |
| Comorbidities | |||
| Medical conditions | |||
| Hypertension | 76.0 | 78.0 | 0.05 |
| Hyperlipidemia | 57.1 | 61.3 | 0.09 |
| Rheumatoid arthritis or osteoarthritis | 45.8 | 47.6 | 0.04 |
| Ischemic heart disease | 43.5 | 43.8 | 0.01 |
| Anemia | 37.8 | 35.0 | 0.06 |
| Diabetes | 35.8 | 36.7 | 0.02 |
| Chronic kidney disease | 31.3 | 30.2 | 0.02 |
| Congestive heart failure | 26.3 | 26.0 | 0.01 |
| Hypothyroidism | 23.3 | 23.3 | 0.00 |
| COPD | 16.9 | 18.0 | 0.03 |
| Atrial fibrillation | 15.8 | 16.6 | 0.02 |
| Osteoporosis | 13.6 | 12.7 | 0.03 |
| Stroke or TIA | 11.9 | 6.3 | 0.19 |
| BPH | 9.3 | 8.9 | 0.01 |
| Asthma | 6.0 | 7.7 | 0.07 |
| Epilepsy | 4.3 | 2.0 | 0.14 |
| Mental health conditions | |||
| Depression | 31.5 | 18.6 | 0.30 |
| Anxiety | 23.3 | 16.1 | 0.18 |
| Schizophrenia | 10.3 | 2.8 | 0.31 |
Abbreviations: BPH, benign prostatic hyperplasia; COPD, chronic obstructive pulmonary disease; TIA, transient ischemic attack.
Groups were matched on age (in 2-year bins), sex, race and ethnicity, Medicaid dual eligibility, baseline medication count, geographic region, and calendar year.
Included American Indian/Alaska Native, Asian/Pacific Islander, unspecified, and unknown.
Trends in Medication Counts
Prior to the index date, both cohorts demonstrated quarterly growth in mean overall medication counts, with modestly greater growth in the cohort with dementia (0.18 vs 0.10 medications per quarter; difference, 0.07 [95% CI, 0.00-0.14]) (Figure 1A and eTable 3 in Supplement 1). This difference was attributable to increasing use of both antidementia and CNS-active medications in the cohort with dementia prior to diagnosis (Figure 1B and C). Following the index date, the cohort with dementia had a greater immediate mean increase in overall medications (0.41 vs −0.06; difference, 0.46 [95% CI, 0.27-0.66]), the proportion using 5 or more medications (absolute difference, 2.77%; 95% CI, 1.24%-4.30%), and the proportion using 10 or more medications (absolute difference, 2.13%; 95% CI, 0.57%-3.69%) (eFigure 1 and eTable 4 in Supplement 1). These increases were attributable to a greater mean number of antidementia medications (0.23; 95% CI, 0.20-0.26), CNS-active medications (0.11; 95% CI, 0.05-0.17), and other medications (0.13; 95% CI, 0.02-0.24) (Figure 1D). In the postintervention period, overall medication use declined in both groups, but the relative change in slope was greater in the cohort with dementia (relative change in mean medications per quarter, −0.29 vs −0.13; difference, −0.16 [95% CI, −0.26 to −0.15]). One year after diagnosis, 75.2% of the cohort with dementia was using 5 or more medications (increase of 2.8%) and 30.8% was using 10 or more medications (increase of 1.3%).
Figure 1. Trends in Overall Medication Use.
Each quarter represents the preceding 90 days. The vertical dotted line represents the last quarter of pharmacy fills prior to the index date. Corresponding time series analyses are presented in eTable 3 in Supplement 1. CNS indicates central nervous system.
Trends in Use of CNS-Active and Anticholinergic Medications
At baseline, a greater proportion of the cohort with dementia used CNS-active medications (54.32% vs 48.39%; difference, 5.93% [95% CI, 4.88%-6.97%]) and anticholinergics (17.79% vs 15.96%; difference, 1.83% [95% CI, 1.69%-1.97%]; quarterly change in use, −0.53% vs −0.21%; difference, −0.32% [95% CI, −0.55% to −0.08%]) (Figure 2 and eFigure 2 and eTable 5 in Supplement 1). In the preperiod, the cohort with dementia had a significantly greater slope of use of any CNS-active, antipsychotic, antidepressant, antiepileptic, benzodiazepine, and anticholinergic medications compared with the control cohort. Immediately following the index date, the cohort with dementia had greater increases in any CNS-active medications use (absolute change, 3.44% vs 0.79%; difference, 2.65% [95% CI, 0.85%-4.45%]) owing to a comparatively increased use of antipsychotics, antidepressants, and antiepileptics. However, the cohort with dementia had modest decreases in anticholinergic use (absolute change, −0.26% vs 0.32%; difference, −0.58% [95% CI, −1.15% to −0.01%]) and opioid use (absolute change, −0.54% vs 0.53%; difference, −1.07% [95% CI, −1.79% to −0.36%]). In the postperiod, the use of any CNS-active medications declined in both groups, but the relative change in slope was greater in the cohort with dementia (−1.99% vs −0.42%; difference, −1.21% [95% CI, −1.99% to −0.42%]) owing to greater relative reductions in use of antipsychotics, antidepressants, antiepileptics, and benzodiazepines. The use of anticholinergics also declined to a greater extent in the cohort with dementia (relative change in slope, −0.66% vs −0.19%; difference, −0.46% [95% CI, −0.75% to −0.17%]). In the cohort with dementia, prescribing patterns varied; for example, by the last study quarter, 11.0% of the cohort had newly initiated any CNS-active medications while 6.3% had discontinued all CNS-active medications (Figure 3).
Figure 2. Trends in Use of Central Nervous System–Active and Anticholinergic Medications.
Each quarter represents the preceding 90 days. The vertical dotted line represents the last quarter of pharmacy fills prior to the index date. Opioids and benzodiazepines are shown in eFigure 3 in Supplement 1; nonbenzodiazepine benzodiazepine receptor agonists are not shown due to low use in both cohorts. Corresponding time series analyses are presented in eTable 5 in Supplement 1.
Figure 3. Change in Use of Central Nervous System (CNS)–Active and Anticholinergic Medications From the Index Date to the End of the Study Period.
BZRA indicates nonbenzodiazepine benzodiazepine receptor agonist.
Trends in Use of Cardiometabolic Medications
At baseline, a smaller proportion of the cohort with dementia used each cardiometabolic medication class except insulin (eg, diabetes medications, 31.19% vs 36.45%) (Figure 4 and eFigure 3 and eTable 6 in Supplement 1). Immediately following the index date, the only significant change was greater use of antihypertensives and insulins in the cohort with dementia (absolute change in proportion using antihypertensives, 0.85% vs 0.28%; difference, 0.57% [95% CI, 0.17%-0.96%]; quarterly change in proportion using antihypertensives, –0.84% vs –0.40%; difference, –0.44% [95% CI, –0.64% to –0.25%]; absolute change in proportion using insulins, 0.19% vs −0.01%; difference, 0.21% [95% CI, 0.04%-0.37%]). In the postperiod, the cohort with dementia had declining use of all medication classes except insulin. Compared with preperiod slopes, the relative change in postperiod slopes was greater in the cohort with dementia for antihypertensives, antiplatelets, anticoagulants, and insulins. Prescribing patterns were variable within the cohort with dementia; for example, by the last study quarter, 5.3% of the cohort had newly initiated lipid-lowering medications while 6.9% had discontinued lipid-lowering medications (eFigure 4 in Supplement 1).
Figure 4. Trends in Use of Cardiometabolic Medications.
Each quarter represents the preceding 90 days. The vertical dotted line represents the last quarter of pharmacy fills prior to the index date. Diabetes medication trends are shown in eFigure 3 and the corresponding time series analyses in eTable 4 in Supplement 1.
Discussion
In a national sample of Medicare beneficiaries, leading up to and immediately following an incident dementia diagnosis, beneficiaries were exposed to a greater number of overall medications, including increased use of both antidementia medications and CNS-active medications. In the subsequent year, beneficiaries with dementia had a modestly greater rate of decline of overall medication use owing to declining use of cardiometabolic medications and anticholinergics; however, overall, patients with dementia were receiving more medications 1 year after the diagnosis of dementia than in the year prior to the diagnosis, with similar patterns of use of medications in classes other than CNS-active and antidementia drugs.
Our results provide a baseline to inform efforts to rethink the clinical approach to medication use at the time of a new dementia diagnosis. A new dementia diagnosis should prompt a conversation about the consequences this diagnosis has for the benefits and harms of current medications for all patients, with the goal of aligning medication use with patient priorities and goals. In many cases, the consequences of dementia for life expectancy and function are likely to reduce the benefit profile of long-term medications and warrant deprescribing, particularly in the case of patients facing burdensome polypharmacy or using medications associated with impaired cognition. However, we observed low rates of deprescribing and a paradoxical increase in the use of CNS-active medications, suggesting a gap between scientific evidence and current clinical practice.
Increased use of CNS-active medications leading up to and following dementia diagnosis is contrary to professional guidelines and likely reflects off-label use for behavioral and psychological symptoms of dementia despite limited evidence to support this practice.30,31 These findings are largely consistent with prior studies examining medication use after acetylcholinesterase inhibitor initiation.32,33 Many CNS-active medications have been identified as potentially inappropriate due to their association with increased risk of falls, worsening cognitive function, and other adverse drug events.6,33,34,35,36 However, physicians and caregivers have identified the need to meet patient-oriented goals and barriers to nonpharmacologic treatment options for behavioral and psychological symptoms of dementia as reasons for continued use of these classes as well as a perception that these medications are generally safe.37,38,39,40 Moreover, the trend prior to the incident diagnosis likely reflects attempts to treat psychological manifestations of early dementia that may be difficult to distinguish from conditions such as depression prior to firmly establishing a diagnosis of dementia, which often emerges over the course of several visits. In contrast, we observed a greater decline in use of anticholinergics than CNS-active medications. This difference, along with the observed variability in initiation and cessation of CNS-active medications, suggests that efforts to raise awareness of potentially inappropriate medications through the Beers List,6 Choosing Wisely Campaign,41 and other educational interventions have had variable uptake.
The use of some cardiometabolic medications decreased to a greater extent for patients with an incident dementia diagnosis, which may reflect efforts to reduce medication burden. However, the medication classes with the steepest rate of discontinuation, lipid-lowering and antihypertensive medications, were classes with low risks for adverse drug events, while higher-risk classes, such as insulins, antiplatelet medications, and anticoagulant medications, had smaller or no reductions in use. Numerous studies have documented high rates of diabetes overtreatment in older adults and associations between hypoglycemia and worsening cognitive function42,43,44,45; thus, this represents an area for improved care delivery and consideration of deprescribing.4,46 In contrast, the benefits and safety of discontinuing low-risk cardiovascular medications, such as statins and antihypertensives, remain unsettled, with conflicting data from observational studies and trials with short-term follow-up.47,48,49,50,51,52
Ultimately, improving prescribing safety and quality following an incident dementia diagnosis requires care coordination between patients, caregivers, primary care physicians, pharmacists, and specialists. While life trajectories for patients with dementia are variable, newer mortality prediction models may help patients, clinicians, and caregivers make informed decisions about the likelihood of medication benefit.53,54 Addressing medication safety proactively requires financial models to support care coordination as well as evidence-based interventions to support medication optimization strategies.55,56 More work is needed in both domains. In 2017, Medicare introduced a billing code for cognitive assessment and care plan services, which was envisioned to provide funding for a separate visit focused on care coordination when a patient shows signs of cognitive impairment. While ideally suited for supporting medication interventions, early uptake of the care coordination visit has been limited to less than 2% of beneficiaries with diagnosed dementia.57,58 Interventions to reduce polypharmacy and potential inappropriate medications have been modestly successful for patients without dementia.59,60 Of note, in the recent OPTIMIZE trial, an educational intervention aimed at primary care clinicians and patients with cognitive impairment did not reduce polypharmacy or use of potentially inappropriate medications.61 In a recent evaluation, models of decision-making that were aligned with patient priorities for older adults without dementia led to reductions in overall medication use, suggesting a path forward that remains to be tested in populations with dementia.62
Limitations
This study has several limitations. First, the study population included beneficiaries with traditional Medicare and Medicare Part D prescription coverage, which limits generalizability. Second, billing diagnosis codes were used to identify incident dementia; these have been demonstrated to have good criterion validity but are less accurate than medical interview and might also lag behind the consideration of a dementia diagnosis by clinicians. Although the onset of cognitive impairment symptoms preceded clinical diagnosis, we used ambulatory diagnosis codes to represent the first time that dementia was recognized and coded in clinical practice and thus the most likely time point for medication changes to be made in response to a new condition. Third, prescription medication claims may overestimate medication exposure if prescriptions were filled but not used and may underestimate exposure if medication expenses were paid outside insurance or by hospice benefit, although hospice enrollment was uncommon. Fourth, patients with incident dementia and control patients were matched by demographics, geographic region, and medication counts but not comorbidity, which led to imbalances in baseline psychiatric and neurologic conditions but should not bias the comparative time series estimate of the change in slopes comparing the preperiod and postperiod without requiring parallel outcomes in the preperiod.28 Fifth, we assessed medication use only in the year immediately following diagnosis; further study of longer-term trajectories of medication use is warranted.
Conclusions
In this cohort study, Medicare beneficiaries with an incident dementia diagnosis were more likely to initiate CNS-active medications and slightly more likely to have cardiometabolic and anticholinergic medications discontinued compared with control patients. These findings highlight opportunities to reduce burdensome polypharmacy by deprescribing long-term medications with high safety risks or limited likelihood of benefit or that may be associated with impaired cognition.
eTable 1. Visits Examined for Dementia Diagnosis
eTable 2. Study Medications
eTable 3. Comparative Interrupted Time Series Analysis of Overall Medication Counts
eFigure 1. Trends in Polypharmacy
eTable 4. Comparative Interrupted Time Series Analysis of Polypharmacy
eFigure 2. Trends in Use of Additional Central Nervous System–Active Medications
eTable 5. Comparative Interrupted Time Series Analysis of Central Nervous System–Active and Anticholinergic Medications
eFigure 3. Trends in Use of Diabetes Medications
eTable 6. Comparative Interrupted Time Series Analysis of Cardiometabolic Medication
eFigure 4. Change in Use of Cardiometabolic Medications Comparing Index Date and End of Study Period
Data Sharing Statement
References
- 1.Matthews KA, Xu W, Gaglioti AH, et al. Racial and ethnic estimates of Alzheimer’s disease and related dementias in the United States (2015-2060) in adults aged ≥65 years. Alzheimers Dement. 2019;15(1):17-24. doi: 10.1016/j.jalz.2018.06.3063 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Clague F, Mercer SW, McLean G, Reynish E, Guthrie B. Comorbidity and polypharmacy in people with dementia: insights from a large, population-based cross-sectional analysis of primary care data. Age Ageing. 2017;46(1):33-39. [DOI] [PubMed] [Google Scholar]
- 3.Boyd C, Smith CD, Masoudi FA, et al. Decision making for older adults with multiple chronic conditions: executive summary for the American Geriatrics Society Guiding Principles on the Care of Older Adults With Multimorbidity. J Am Geriatr Soc. 2019;67(4):665-673. doi: 10.1111/jgs.15809 [DOI] [PubMed] [Google Scholar]
- 4.Lipska KJ, Krumholz H, Soones T, Lee SJ. Polypharmacy in the aging patient: a review of glycemic control in older adults with type 2 diabetes. JAMA. 2016;315(10):1034-1045. doi: 10.1001/jama.2016.0299 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Qaseem A, Wilt TJ, Rich R, et al. ; Clinical Guidelines Committee of the American College of Physicians and the Commission on Health of the Public and Science of the American Academy of Family Physicians . Pharmacologic treatment of hypertension in adults aged 60 years or older to higher versus lower blood pressure targets: a clinical practice guideline from the American College of Physicians and the American Academy of Family Physicians. Ann Intern Med. 2017;166(6):430-437. doi: 10.7326/M16-1785 [DOI] [PubMed] [Google Scholar]
- 6.2019 American Geriatrics Society Beers Criteria Update Expert Panel . American Geriatrics Society 2019 updated AGS Beers Criteria for potentially inappropriate medication use in older adults. J Am Geriatr Soc. 2019;67(4):674-694. [DOI] [PubMed] [Google Scholar]
- 7.Growdon ME, Gan S, Yaffe K, Steinman MA. Polypharmacy among older adults with dementia compared with those without dementia in the United States. J Am Geriatr Soc. 2021;69(9):2464-2475. doi: 10.1111/jgs.17291 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Fralick M, Bartsch E, Ritchie CS, Sacks CA. Estimating the use of potentially inappropriate medications among older adults in the United States. J Am Geriatr Soc. 2020;68(12):2927-2930. doi: 10.1111/jgs.16779 [DOI] [PubMed] [Google Scholar]
- 9.Maust DT, Strominger J, Bynum JPW, et al. Prevalence of psychotropic and opioid prescription fills among community-dwelling older adults with dementia in the US. JAMA. 2020;324(7):706-708. doi: 10.1001/jama.2020.8519 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Maust DT, Strominger J, Kim HM, et al. Prevalence of central nervous system-active polypharmacy among older adults with dementia in the US. JAMA. 2021;325(10):952-961. doi: 10.1001/jama.2021.1195 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Scott IA, Hilmer SN, Reeve E, et al. Reducing inappropriate polypharmacy: the process of deprescribing. JAMA Intern Med. 2015;175(5):827-834. doi: 10.1001/jamainternmed.2015.0324 [DOI] [PubMed] [Google Scholar]
- 12.Krishnaswami A, Steinman MA, Goyal P, et al. ; Geriatric Cardiology Section Leadership Council, American College of Cardiology . Deprescribing in older adults with cardiovascular disease. J Am Coll Cardiol. 2019;73(20):2584-2595. doi: 10.1016/j.jacc.2019.03.467 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Bjerre LM, Farrell B, Hogel M, et al. Deprescribing antipsychotics for behavioural and psychological symptoms of dementia and insomnia: evidence-based clinical practice guideline. Can Fam Physician. 2018;64(1):17-27. [PMC free article] [PubMed] [Google Scholar]
- 14.Tjia J, Rothman MR, Kiely DK, et al. Daily medication use in nursing home residents with advanced dementia. J Am Geriatr Soc. 2010;58(5):880-888. doi: 10.1111/j.1532-5415.2010.02819.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Tjia J, Briesacher BA, Peterson D, Liu Q, Andrade SE, Mitchell SL. Use of medications of questionable benefit in advanced dementia. JAMA Intern Med. 2014;174(11):1763-1771. doi: 10.1001/jamainternmed.2014.4103 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Maust DT, Kim HM, Chiang C, Kales HC. Association of the Centers for Medicare & Medicaid Services’ National Partnership to Improve Dementia Care with the use of antipsychotics and other psychotropics in long-term care in the United States from 2009 to 2014. JAMA Intern Med. 2018;178(5):640-647. doi: 10.1001/jamainternmed.2018.0379 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Briesacher BA, Tjia J, Field T, Peterson D, Gurwitz JH. Antipsychotic use among nursing home residents. JAMA. 2013;309(5):440-442. doi: 10.1001/jama.2012.211266 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Sarkar U, Lyles C, Steinman M, et al. Changes in medication use after dementia diagnosis in an observational cohort of individuals with diabetes mellitus. J Am Geriatr Soc. 2017;65(1):77-82. doi: 10.1111/jgs.14429 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Bangerter LR, Clark CN, Jarvis MS, et al. Medication fill patterns for cognitive and behavioral or psychological symptoms of Alzheimer disease and related dementias. JAMA Netw Open. 2022;5(6):e2215678. doi: 10.1001/jamanetworkopen.2022.15678 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Centers for Medicare & Medicaid Services . Chronic Conditions Data Warehouse. Accessed May 1, 2022. https://www2.ccwdata.org/web/guest/condition-categories
- 21.Taylor DH Jr, Fillenbaum GG, Ezell ME. The accuracy of medicare claims data in identifying Alzheimer’s disease. J Clin Epidemiol. 2002;55(9):929-937. doi: 10.1016/S0895-4356(02)00452-3 [DOI] [PubMed] [Google Scholar]
- 22.Fujiyoshi A, Jacobs DR Jr, Alonso A, Luchsinger JA, Rapp SR, Duprez DA. Validity of death certificate and hospital discharge ICD codes for dementia diagnosis: the Multi-Ethnic Study of Atherosclerosis. Alzheimer Dis Assoc Disord. 2017;31(2):168-172. doi: 10.1097/WAD.0000000000000164 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.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: 10.1001/jamahealthforum.2022.0653 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Ho DE, Imai K, King G, Stuart EA. Matching as nonparametric preprocessing for reducing model dependence in parametric causal inference. Polit Anal. 2007;15(3):199-236. doi: 10.1093/pan/mpl013 [DOI] [Google Scholar]
- 25.Kornblith E, Bahorik A, Boscardin WJ, Xia F, Barnes DE, Yaffe K. Association of race and ethnicity with incidence of dementia among older adults. JAMA. 2022;327(15):1488-1495. doi: 10.1001/jama.2022.3550 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Jacobs PD, Kronick R. The effects of coding intensity in Medicare Advantage on plan benefits and finances. Health Serv Res. 2021;56(2):178-187. doi: 10.1111/1475-6773.13591 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Radomski TR, Zhao X, Hanlon JT, et al. Use of a medication-based risk adjustment index to predict mortality among veterans dually-enrolled in VA and Medicare. Healthc (Amst). 2019;7(4):S2213-0764(18)30230-6. doi: 10.1016/j.hjdsi.2019.04.003 [DOI] [PMC free article] [PubMed]
- 28.Fry CE, Hatfield LA. Birds of a feather flock together: comparing controlled pre-post designs. Health Serv Res. 2021;56(5):942-952. doi: 10.1111/1475-6773.13697 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Linden A. Conducting interrupted time-series analysis for single- and multiple-group comparisons. Stata J. 2015;15(2):480-500. doi: 10.1177/1536867X1501500208 [DOI] [Google Scholar]
- 30.Sink KM, Holden KF, Yaffe K. Pharmacological treatment of neuropsychiatric symptoms of dementia: a review of the evidence. JAMA. 2005;293(5):596-608. doi: 10.1001/jama.293.5.596 [DOI] [PubMed] [Google Scholar]
- 31.Dudas R, Malouf R, McCleery J, Dening T. Antidepressants for treating depression in dementia. Cochrane Database Syst Rev. 2018;8(8):CD003944. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Efjestad AS, Ihle-Hansen H, Hjellvik V, Engedal K, Blix HS. Drug use before and after initiating treatment with acetylcholinesterase inhibitors. Dement Geriatr Cogn Dis Extra. 2019;9(1):196-206. doi: 10.1159/000497307 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Seppala LJ, van de Glind EMM, Daams JG, et al. ; EUGMS Task and Finish Group on Fall-Risk-Increasing Drugs . Fall-risk-increasing drugs: a systematic review and meta-analysis, III: others. J Am Med Dir Assoc. 2018;19(4):372.e1-372.e8. doi: 10.1016/j.jamda.2017.12.099 [DOI] [PubMed] [Google Scholar]
- 34.Gadzhanova S, Roughead E, Robinson M. Use of medicines with anticholinergic and sedative effect before and after initiation of anti-dementia medications. Drugs Real World Outcomes. 2015;2(1):53-60. doi: 10.1007/s40801-015-0012-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Tannenbaum C, Paquette A, Hilmer S, Holroyd-Leduc J, Carnahan R. A systematic review of amnestic and non-amnestic mild cognitive impairment induced by anticholinergic, antihistamine, GABAergic and opioid drugs. Drugs Aging. 2012;29(8):639-658. [DOI] [PubMed] [Google Scholar]
- 36.Porsteinsson AP, Drye LT, Pollock BG, et al. ; CitAD Research Group . Effect of citalopram on agitation in Alzheimer disease: the CitAD randomized clinical trial. JAMA. 2014;311(7):682-691. doi: 10.1001/jama.2014.93 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Kerns JW, Winter JD, Winter KM, Boyd T, Etz RS. Primary care physician perspectives about antipsychotics and other medications for symptoms of dementia. J Am Board Fam Med. 2018;31(1):9-21. doi: 10.3122/jabfm.2018.01.170230 [DOI] [PubMed] [Google Scholar]
- 38.Green AR, Lee P, Reeve E, et al. Clinicians’ perspectives on barriers and enablers of optimal prescribing in patients with dementia and coexisting conditions. J Am Board Fam Med. 2019;32(3):383-391. doi: 10.3122/jabfm.2019.03.180335 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Kerns JW, Winter JD, Winter KM, Kerns CC, Etz RS. Caregiver perspectives about using antipsychotics and other medications for symptoms of dementia. Gerontologist. 2018;58(2):e35-e45. doi: 10.1093/geront/gnx042 [DOI] [PubMed] [Google Scholar]
- 40.Reeve E, Bell JS, Hilmer SN. Barriers to optimising prescribing and deprescribing in older adults with dementia: a narrative review. Curr Clin Pharmacol. 2015;10(3):168-177. doi: 10.2174/157488471003150820150330 [DOI] [PubMed] [Google Scholar]
- 41.Tip sheet: ten things clinicians and patients should question. Health in Aging Foundation. Accessed July 6, 2023. https://www.healthinaging.org/tools-and-tips/tip-sheet-ten-things-physicians-and-patients-should-question
- 42.Feil DG, Rajan M, Soroka O, Tseng CL, Miller DR, Pogach LM. Risk of hypoglycemia in older veterans with dementia and cognitive impairment: implications for practice and policy. J Am Geriatr Soc. 2011;59(12):2263-2272. doi: 10.1111/j.1532-5415.2011.03726.x [DOI] [PubMed] [Google Scholar]
- 43.Lee AK, Rawlings AM, Lee CJ, et al. Severe hypoglycaemia, mild cognitive impairment, dementia and brain volumes in older adults with type 2 diabetes: the Atherosclerosis Risk in Communities (ARIC) cohort study. Diabetologia. 2018;61(9):1956-1965. doi: 10.1007/s00125-018-4668-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.McCoy RG, Lipska KJ, Yao X, Ross JS, Montori VM, Shah ND. Intensive treatment and severe hypoglycemia among adults with type 2 diabetes. JAMA Intern Med. 2016;176(7):969-978. doi: 10.1001/jamainternmed.2016.2275 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Arnold SV, Lipska KJ, Wang J, Seman L, Mehta SN, Kosiborod M. Use of intensive glycemic management in older adults with diabetes mellitus. J Am Geriatr Soc. 2018;66(6):1190-1194. doi: 10.1111/jgs.15335 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Farrell B, Black C, Thompson W, et al. Deprescribing antihyperglycemic agents in older persons: evidence-based clinical practice guideline. Can Fam Physician. 2017;63(11):832-843. [PMC free article] [PubMed] [Google Scholar]
- 47.Kutner JS, Blatchford PJ, Taylor DH Jr, et al. Safety and benefit of discontinuing statin therapy in the setting of advanced, life-limiting illness: a randomized clinical trial. JAMA Intern Med. 2015;175(5):691-700. doi: 10.1001/jamainternmed.2015.0289 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Thompson W, Morin L, Jarbøl DE, et al. Statin discontinuation and cardiovascular events among older people in Denmark. JAMA Netw Open. 2021;4(12):e2136802. doi: 10.1001/jamanetworkopen.2021.36802 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Orkaby AR, Driver JA, Ho YL, et al. Association of statin use with all-cause and cardiovascular mortality in US veterans 75 years and older. JAMA. 2020;324(1):68-78. doi: 10.1001/jama.2020.7848 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Sheppard JP, Burt J, Lown M, et al. ; OPTIMISE Investigators . Effect of antihypertensive medication reduction vs usual care on short-term blood pressure control in patients with hypertension aged 80 years and older: the OPTIMISE randomized clinical trial. JAMA. 2020;323(20):2039-2051. doi: 10.1001/jama.2020.4871 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51.Moonen JE, Foster-Dingley JC, de Ruijter W, et al. Effect of discontinuation of antihypertensive treatment in elderly people on cognitive functioning—the DANTE Study Leiden: a randomized clinical trial. JAMA Intern Med. 2015;175(10):1622-1630. doi: 10.1001/jamainternmed.2015.4103 [DOI] [PubMed] [Google Scholar]
- 52.Williamson JD, Pajewski NM, Auchus AP, et al. ; SPRINT MIND Investigators for the SPRINT Research Group . Effect of intensive vs standard blood pressure control on probable dementia: a randomized clinical trial. JAMA. 2019;321(6):553-561. doi: 10.1001/jama.2018.21442 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53.Haaksma ML, Eriksdotter M, Rizzuto D, et al. Survival time tool to guide care planning in people with dementia. Neurology. 2020;94(5):e538-e548. doi: 10.1212/WNL.0000000000008745 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 54.Deardorff WJ, Barnes DE, Jeon SY, et al. Development and external validation of a mortality prediction model for community-dwelling older adults with dementia. JAMA Intern Med. 2022;182(11):1161-1170. doi: 10.1001/jamainternmed.2022.4326 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 55.Anderson TS, Goyal P, Marcum ZA. Implementing a proactive deprescribing approach to prevent adverse drug events. J Gen Intern Med. 2020;35(12):3694-3696. doi: 10.1007/s11606-020-05886-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- 56.Keller MS, Vordenberg SE, Steinman MA. Moving deprescribing upstream. J Gen Intern Med. 2022;37(12):3176-3177. doi: 10.1007/s11606-022-07537-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 57.Li J, Andy C, Mitchell S. Use of Medicare’s new reimbursement codes for cognitive assessment and care planning, 2017-2018. JAMA Netw Open. 2021;4(9):e2125725. doi: 10.1001/jamanetworkopen.2021.25725 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 58.Agarwal SD, Basu S, Landon BE. The underuse of Medicare’s prevention and coordination codes in primary care: a cross-sectional and modeling study. Ann Intern Med. 2022;175(8):1100-1108. doi: 10.7326/M21-4770 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 59.Martin P, Tamblyn R, Benedetti A, Ahmed S, Tannenbaum C. Effect of a pharmacist-led educational intervention on inappropriate medication prescriptions in older adults: the D-PRESCRIBE randomized clinical trial. JAMA. 2018;320(18):1889-1898. doi: 10.1001/jama.2018.16131 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 60.Bloomfield HE, Greer N, Linsky AM, et al. Deprescribing for community-dwelling older adults: a systematic review and meta-analysis. J Gen Intern Med. 2020;35(11):3323-3332. doi: 10.1007/s11606-020-06089-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 61.Bayliss EA, Shetterly SM, Drace ML, et al. Deprescribing education vs usual care for patients with cognitive impairment and primary care clinicians: the OPTIMIZE pragmatic cluster randomized trial. JAMA Intern Med. 2022;182(5):534-542. doi: 10.1001/jamainternmed.2022.0502 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 62.Tinetti ME, Naik AD, Dindo L, et al. Association of patient priorities–aligned decision-making with patient outcomes and ambulatory health care burden among older adults with multiple chronic conditions: a nonrandomized clinical trial. JAMA Intern Med. 2019;179(12):1688-1697. doi: 10.1001/jamainternmed.2019.4235 [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
eTable 1. Visits Examined for Dementia Diagnosis
eTable 2. Study Medications
eTable 3. Comparative Interrupted Time Series Analysis of Overall Medication Counts
eFigure 1. Trends in Polypharmacy
eTable 4. Comparative Interrupted Time Series Analysis of Polypharmacy
eFigure 2. Trends in Use of Additional Central Nervous System–Active Medications
eTable 5. Comparative Interrupted Time Series Analysis of Central Nervous System–Active and Anticholinergic Medications
eFigure 3. Trends in Use of Diabetes Medications
eTable 6. Comparative Interrupted Time Series Analysis of Cardiometabolic Medication
eFigure 4. Change in Use of Cardiometabolic Medications Comparing Index Date and End of Study Period
Data Sharing Statement