Comparative Safety of Antimuscarinics versus Mirabegron for Overactive Bladder in Parkinson Disease

Abstract

Background:

Overactive bladder (OAB) is a common nonmotor symptom of Parkinson disease (PD), often treated with antimuscarinics or beta-3 agonists. There is lack of evidence to guide OAB management in PD.

Objectives:

To assess the comparative safety of antimuscarinics versus beta-3 agonists for OAB treatment in PD

Methods:

We employed a new-user, active-comparator cohort study design. We included Medicare beneficiaries age ≥65 years with PD who were new users of either antimuscarinic or beta-3 agonist. The primary outcome was any acute care encounter (i.e., non-elective hospitalization or emergency department visit) within 90 days of OAB drug initiation. The main secondary outcome was a composite measure of acute care encounters for anticholinergic related adverse events (AEs). Matching on high-dimensional propensity score (hdPS) was used to address potential confounding. We used Cox proportional hazards models to examine the association between OAB drug category and outcomes. We repeated analyses for 30- and 180-day follow-up periods.

Results:

We identified 27,091 individuals meeting inclusion criteria (mean age: 77.8 years). After hdPS matching, antimuscarinic users had increased risks for any acute care encounter (hazard ratio [HR] 1.23, 95% confidence interval [CI] 1.12-1.37) and encounters for anticholinergic related AEs (1.18, 1.04-1.34) compared to beta-3 agonist users. Similar associations were observed for sensitivity analyses.

Conclusions:

Among persons with PD, anticholinergic initiation was associated with a higher risk of acute care encounters compared with beta-3 agonist initiation. The long-term safety of anticholinergic vs. beta-3 agonist therapy in the PD population should be evaluated in a prospective study.

INTRODUCTION

Overactive bladder (OAB) is a potentially disabling non-motor symptom affecting more than one-third of persons living with Parkinson disease (PD).[1] Drugs used for OAB work by either blocking muscarinic acetylcholine receptors (i.e., antimuscarinics) or activating beta-3 adrenergic receptors (i.e., beta-3 agonist). Although several placebo-controlled[2-4] or pre-post designed trials[5] demonstrate similar efficacy between these classes, there is limited comparative safety data, particularly in vulnerable populations. Antimuscarinics, through their anticholinergic actions in the central nervous system, have the potential to produce adverse drug effects in older adults. Anticholinergic medication exposure is associated with an increased risk of falls, fracture, and delirium in the short-term, and in the long-term worsening of cognitive function or earlier onset of dementia symptoms.[6, 7] As a result of this real-world evidence, anticholinergic OAB medications are listed in the 2019 American Geriatrics Society Beers Criteria^® for Potentially Inappropriate Medication Use in Older Adults as having a drug-disease interaction for individuals with dementia. PD is associated with loss of cholinergic neurons in the nucleus basalis of Meynert,[8] beginning in the prodromal period of the disease, contributing to cognitive deficits and gait dysfunction in early disease, and freezing and dementia as the disease progresses.[8] Therefore, use of anticholinergics in PD persons with diminished cholinergic reserve or cholinergic dysfunction is more problematic compared to the general older adult population. Yet, there are no PD-specific guidelines or precautions relating to anticholinergic medications, largely due to lack of real-world evidence on drug-drug and drug-disease interactions in PD. To address this evidence gap for drug-disease interactions in PD, inform PD-specific drug safety clinical guideline development, and support clinical decision making for PD treatment, the aim of this study was to determine the short-term safety of antimuscarinics versus mirabegron (the only beta-3 agonist available in the U.S. at the time of this study) for the treatment of OAB in a representative sample of older adults with PD. We hypothesized that antimuscarinic new users would be at greater risk for acute healthcare events (i.e., all-cause non-elective hospitalization or emergency department [ED] visit) and anticholinergic related adverse events (AEs) as compared to new users of mirabegron. We chose to examine all-cause acute care encounter in addition to hospitalization or ED visit specifically for anticholinergic related AEs because more specific AE diagnoses might be under-recognized and under-coded in claims data.

METHODS

Ethical and institutional approval

This study was approved with a waiver of consent by the University of Pennsylvania Human Research Protections Office and the Centers for Medicare and Medicaid Services (CMS).

Data source

We used 2012-2017 CMS Research Identifiable File data, including the Master Beneficiary Summary (MBSF), Chronic Conditions Data Warehouse (CCW), Carrier, Medicare Provider Analysis and Review (MEDPAR), and Prescription Drug Event (PDE) files. These datasets contain information on insurance coverage and eligibility (MBSF), chronic and disabling medical conditions (CCW), diagnosis and treatment in the outpatient (Carrier) and inpatient (MEDPAR) settings, and prescription claims (PDE).

Data availability statement

Research Identifiable Files used in this study are available to researchers for purchase through CMS. Programming codes for the analyses can be shared with interested parties upon request.

Study design and sample

We employed a new-user, active-comparator study design, which allows observational studies to more closely approximate randomized controlled trials (RCTs) and thereby reduces bias and confounding by indication.[9] The “new-user” design principle allows for pre-treatment characteristics to be measured, and the “active comparator” principle compares the drug of interest to another efficacious treatment for the same clinical indication in the same population.[9]

Our study sample consisted of Medicare beneficiaries ages 65 and older who were new users of a qualifying OAB medication and met insurance eligibility and PD diagnosis criteria in the six months prior to the OAB prescription. We required beneficiaries to have Part A (inpatient), B (outpatient), and D (prescription) coverage with no other healthcare coverage (e.g., co-insurance with Medicare Advantage or Medicaid). We used a previously published algorithm for identifying individuals with PD in administrative claims datasets; this algorithm requires a physician or advanced practice provider encounter with a PD diagnosis, at least one antiparkinsonian medication prescription, and the absence of a diagnosis of secondary parkinsonism, atypical parkinsonism, bipolar disorder or schizophrenia (Supplemental Table 1).

Medication exposure

OAB medications were identified in the PDE file using by National Drug Codes from Lexicon Plus (Oracle Corporation: Austin, Texas). Drugs with an antimuscarinic mechanism of action (darifenacin, fesoterodine, oxybutynin, solifenacin, tolterodine, trospium) were studied collectively as one class. Mirabegron served as the beta-3 agonist active comparator. New users were identified as those having a prescription fill for an OAB drug, with no OAB prescription fills within the prior six months. We required the qualifying fill to be a single OAB drug, and therefore excluded individuals who received combination therapies or filled multiple distinct OAB prescriptions on the same day. We also excluded those who, based on MEDPAR data, were in an any inpatient setting at the time of the prescription fill. Mirabegron was approved for use by the U.S. Food and Drug Administration (FDA) on June 28, 2012. To ensure that both mirabegron and antimuscarinic treatment options were available, new user dates must have occurred on July 1, 2012 or later.

Study outcomes

The primary outcome was any acute care encounter (i.e., non-elective hospitalization or ED visit) within 90 days of OAB drug initiation. The main secondary outcome was a composite measure of all acute care encounters for anticholinergic related AEs. We also explored specific anticholinergic related AEs, as follows: (1) fracture of hip, thigh, or pelvis;[10] (2) unintentional traumatic injury;[10] and autonomic anticholinergic AEs,[11] including (3) constipation or urinary retention,[12] (4) fall,[13] and (5) dizziness/vertigo,[11] as these AEs might be under-diagnosed and under-documented compared to acute care encounters in general (Supplemental Table 1). Specific outcomes of interest were identified using International Disease Classification of Diseases, 9^th revision, clinical modification (ICD-9-CM) or ICD-10-CM diagnosis codes in any diagnostic position on inpatient discharge claims. Standard forward mapping using General Equivalence Mappings from CMS was used to convert ICD-9-CM to ICD-10-CM codes.

Follow-up and censoring

We followed individuals over a 90-day period that began on the day of new OAB drug prescription dispensing for the primary analysis. Patients were censored upon the earliest of: (1) death, (2) loss of eligibility, (3) OAB discontinuation (i.e., no new prescription after allowing for a grace period of 14 days at the end of the last prescription) or switching from one OAB drug class to another (i.e., antimuscarinic to beta-3 agonist or vice versa) , (4) end of study period (i.e., 12/31/2017), or (5) 90 days from their new-user date.

Baseline covariates selection and measurement

Predefined baseline covariates were ascertained during the continuous 6-month baseline period. Demographic covariates included age, sex, race/ethnicity category, and a zip-code based social deprivation index (SDI)[14] score. Drug-related predefined covariates included the calendar year of OAB drug initiation and a count of prescription drugs that could potentially increase fall/fracture risk during the baseline period.[15] In addition to these demographic and predefined covariates, we employed a semi-automated, data-adaptive, high-dimensional propensity score (hdPS) approach[16] to identify potential confounders. Eleven data dimensions (inpatient, outpatient, and institutional diagnoses and procedures, prescriptions) measured during the 6-month baseline prior to OAB initiation were included in the algorithm, and the top 200 most prevalent covariates from each dimension were selected. The confounders were ranked based on their potential for bias using the Bross formula, and the top 500 empiric covariates were selected to be included in the PS estimation. We excluded specific diagnosis codes for PD and OAB, as well as medications for PD treatment (Supplemental Table 1) from the hdPS covariate selection algorithm, as they were markers for the study population or the OAB drug exposure. Four separate types of PS were generated, each included a specific set of covariates as follows: (1) demographic only, (2) demographic and predefined covariates, (3) demographic and hdPS selected empirical covariates, (4) demographic, predefined covariates, and hdPS selected empirical covariates. For each PS type, eligible beneficiaries were matched based on PS in a 4 (antimuscarinic):1 (mirabegron) ratio using nearest neighbor matching, as mirabegron is newer and not as commonly prescribed.

Statistical analyses

We used SAS v9.4 (SAS Institute: Cary, North Carolina) to construct and analyze the study sample. We assessed imbalances in demographic and predefined covariates between OAB medication groups (antimuscarinic versus mirabegron) using standardized differences, with an absolute value of standardized difference of 0.10 or greater signaling potentially important imbalances. Missing values for SDI (0.94%) were imputed by the median value. We employed Cox proportional hazards models to evaluate the association between study outcomes and use of antimuscarinic vs. mirabegron, respectively. Six models were fit for each study outcome: model 1 was unadjusted; models 2-5 included matching of beneficiaries based on PS.[16]

Sensitivity analyses

To examine safety immediately after initiation, we performed analyses that censored at 30 days after OAB drug initiation. On the other hand, we also repeated analyses that censored at 180 days after OAB drug initiation, to evaluate longer term safety of these medications.

RESULTS

Over the 2012-2017 study period, 27,091 individuals met PD, age, and insurance eligibility criteria and were new users of OAB medications. Among those eligible for the PS-matched analysis, the proportion of new users by medication are as follows: darifenacin: 347 (1.3%), fesoterodine: 995 (3.7%), oxybutynin: 11,115 (41.0%), solifenacin: 4,782 (17.7%), tolterodine: 2,678 (9.9%), trospium: 1,203 (4.4%) vs. mirabegron: 5,971 (22.0%). Table 1 shows baseline characteristics between mirabegron and antimuscarinic users, both before and after hdPS matching for the primary outcome. Mirabegron new users were more likely to be male (64.7% vs. 56.4%; standardized difference=−0.17); and there was also a difference in the year of OAB initiation with more mirabegron use in later years (standardized difference=0.66) before matching. Baseline characteristics were well-balanced after matching, except for year of OAB initiation that remained slightly imbalanced (standardized difference=0.25); however, we did not expect this slight imbalance to overly impact the analysis.

Table 1.

Descriptive characteristics of eligible, incident new users of overactive bladder medications before and after matching for the primary outcome

	Before matching			After 1:4 (mirabegron:antimuscarinic) matching on hdPS†
Characteristic	Mirabegron (n=5,971)	Antimuscarinics (n = 21,120)	Standardized difference‡	Mirabegron (n=3,641)	Antimuscarinics (n =8,993)	Standardized difference‡
Age, median (IQR)	77.8 (72.8-83.0)	77.9 (72.6-83.1)	0.00	77.1 (72.1-82.4)	77.3 (72.2-82.5)	0.01
Sex			−0.17			−0.08
Male	3,863 (64.7%)	11,920 (56.4%)		2,387 (65.6%)	5,561 (61.8%)
Female	2,108 (35.3%)	9,200 (43.6%)		1,254 (34.4%)	3,432 (38.2%)
Race/Ethnicity			0.06			0.00
White	5,643 (94.5%)	19,783 (93.7%)		3,425 (94.1%)	8,420 (93.6%)
Black	105 (1.8%)	549 (2.6%)		63 (1.7%)	191 (2.1%)
Other/Unknown	144 (2.4%)	486 (2.3%)		99 (2.7%)	244 (2.7%)
Asian	49 (0.8%)	165 (0.8%)		33 (0.9%)	85 (0.9%)
Hispanic	25 (0.4%)	92 (0.4%)		19 (0.5%)	46 (0.5%)
North Am. Native	5 (0.1%)	45 (0.2%)		2 (0.1%)	7 (0.1%)
SDI, median (IQR)	35.0 (16.0-58.0)	36.0 (18.0-60.0)	0.05	35.0 (16.0-58.0)	36.0 (17.0-58.0)	0.02
Year of OAB medication initiation			0.66			0.25
2012*	453 (7.6%)	5,624 (26.6%)		309 (8.5%)	1,226 (13.6%)
2013*
2014	834 (14.0%)	4,001 (18.9%)		573 (15.7%)	1,662 (18.5%)
2015	1,160 (19.4%)	3,770 (17.9%)		755 (20.7%)	1,825 (20.3%)
2016	1,558 (26.1%)	3,977 (18.8%)		929 (25.5%)	2,172 (24.2%)
2017	1,966 (32.9%)	3,748 (17.7%)		1,075 (29.5%)	2,108 (23.4%)
Number of drugs with increased risk of fracture, median (IQR)	1.0 (1.0-2.0)	1.0 (1.0-2.0)	0.08	1.0 (0.0-2.0)	1.0 (0.0-2.0)	0.02

IQR=interquartile range; OAB=overactive bladder; SD=standard deviation; SDI=social deprivation index

^†hdPS includes demographics, predefined variables, and empirically selected variables

^‡Bolded values suggest imbalance between groups

^*Combined due to the Centers for Medicare & Medicaid Services cell suppression policy

After 4:1 matching on hdPS that included demographics, predefined variables, and empirically selected variables, the incidence rate of any all-cause acute care encounters was greater in people starting antimuscarinics compared to people starting mirabegron (963.7; 95 % confidence interval [CI] 915.2-1,014.7 vs. 777.4; 95% CI 712.4-848.3 per 1,000 person-years), as shown in Table 2. We observed similar trends for incidence rates for acute care encounters for any anticholinergic related AEs (424.0; 95 % CI 396.9-453.0 vs. 358.2; 95% CI 320.9-399.9 per 1,000 person-years), as well as for specific anticholinergic AEs (Table 2).

Table 2.

Incidence rate of acute care encounters and other outcomes for new users of overactive bladder medications after 4:1 (anticholinergic:mirabegron) matching on hdPS^†

Outcome	Exposure group	Number at risk	Number of events	Person-years at risk	Incidence rate‡ (95% CI)
Any acute care encounter¶	Antimuscarinics	8,993	1,443	1,497	963.7 (915.2-1,014.7)
	Mirabegron	3,641	504	648	777.4 (712.4-848.3)
Acute care encounters¶ for any anticholinergic related AEs§	Antimuscarinics	12,009	879	2,073	424.0 (396.9-453.0)
	Mirabegron	4,806	317	885	358.2 (320.9-399.9)
Constipation/Urinary retention	Antimuscarinics	13,834	457	2,439	187.4 (171.0-205.4)
	Mirabegron	5,427	126	1,021	123.4 (103.7-147.0)
Dizziness/Vertigo	Antimuscarinics	14,184	138	2,506	55.1 (46.6-65.1)
	Mirabegron	5,591	58	1,056	54.9 (42.5-71.1)
Fall	Antimuscarinics	14,346	117	2,534	46.2 (38.5-55.4)
	Mirabegron	5,738	48	1,084	44.3 (33.4-58.8)
Fracture of hip/thigh/pelvis	Antimuscarinics	14,435	45	2,558	17.6 (13.1-23.6)
	Mirabegron	5,784	15	1,097	13.7 (8.2-22.7)
Unintentional traumatic injury	Antimuscarinics	13,132	527	2,287	230.5 (211.6-251.0)
	Mirabegron	5,203	213	969	219.8 (192.2-251.4)

AE=adverse events; CI=confidence interval; hdPS=high-dimensional propensity score; REF=reference group

^†hdPS includes demographics, predefined variables, and empirically selected variables

^‡Events per 1,000 person-years

^§Composite of fracture of hip/thigh/pelvis, unintentional traumatic injury, constipation/urinary retention, fall, and dizziness/vertigo

^¶All-cause emergency department visit or non-elective hospitalization

After 4:1 matching on hdPS, which included demographics, predefined variables, and empirically selected variables, antimuscarinic use was associated with an increased hazard of any acute care events (HR=1.23; 95% CI 1.12-1.37), of acute care encounters for any anticholinergic related AEs (HR=1.18; 95% CI 1.04-1.34), and of constipation/urinary retention (HR=1.51; 95% CI 1.24-1.84) within 90 days of OAB drug initiation (Table 3). There was no association with fracture of hip/thigh/ pelvis (HR=1.30; 95% CI 0.72-2.32), unintentional traumatic injury (HR=1.05; 95% CI 0.89-1.23), fall (HR=1.04; 95% CI 0.74-1.46), nor dizziness/vertigo (HR=1.00; 95% CI 0.73-1.35) as shown in Table 3. Results for other hdPS models (demographic only, demographic + predefined, demographic + empiric) were similar and shown in Supplemental Table 2. Similar patterns were found in sensitivity analyses; interestingly, compared to the 90-day primary analysis, there was a stronger association between antimuscarinic use and adverse outcomes in the 30-day analysis (Table 3).

Table 3.

Hazard ratio^* of acute care encounters and other outcomes for new users of overactive bladder medications after 4:1 (anticholinergic:mirabegron) matching on hdPS scores^†

Outcome	Primary analysis	Sensitivity analyses
	90-day HR (95% CI)‡	30-day HR (95% CI)‡	180-day HR (95% CI)‡
Any acute care encounter¶	1.23 (1.12-1.37)	1.34 (1.15-1.56)	1.22 (1.12-1.33)
Acute care encounters¶ for any anticholinergic related AEs§	1.18 (1.04-1.34)	1.33 (1.10-1.61)	1.16 (1.04-1.30)
Fracture of hip/thigh/pelvis	1.30 (0.72-2.32)	2.23 (0.77-6.46)	1.05 (0.66-1.66)
Unintentional traumatic injury	1.05 (0.89-1.23)	1.07 (0.84-1.36)	1.12 (0.97-1.28)
Constipation/Urinary retention	1.51 (1.24-1.84)	2.05 (1.50-2.80)	1.38 (1.17-1.63)
Fall	1.04 (0.74-1.46)	1.34 (0.77-2.34)	1.08 (0.82-1.43)
Dizziness/Vertigo	1.00 (0.73-1.35)	1.21 (0.76-1.92)	1.05 (0.80-1.37)

CI=confidence interval, hdPS=high-dimensional propensity score, HR=hazard ratio

^*Reference group = mirabegron

^†hdPS includes demographics, predefined variables, and empirically selected variables

^‡Statistically significant values are highlighted in bold type

^§Composite of fracture of hip/thigh/pelvis, unintentional traumatic injury, constipation/urinary retention, fall, and dizziness/vertigo

^¶All-cause emergency department visit or non-elective hospitalization

DISCUSSION

Consistent with our hypothesis, among older Medicare beneficiaries initiating OAB treatment, the risks for any acute care events and for acute care encounters for anticholinergic related AEs, were greater among antimuscarinic initiators compared with mirabegron initiators after adjustment via hdPS matching. Guidelines for the treatment of OAB in PD are inconclusive due to a lack of randomized, placebo-controlled clinical trials.[1] Our real-world evidence findings suggest that mirabegron may be a safer choice than an antimuscarinic for the treatment of OAB in PD.

Antimuscarinic use was associated with an increased risk of some, but not all specific anticholinergic related AEs, compared to mirabegron. Antimuscarinics were associated with 1.5 times the risk of an ED visit or hospital admission for constipation/urinary retention, which are known adverse effects of antimuscarinics. A prior meta-analysis found that, compared with placebo, OAB treatment with antimuscarinics was significantly associated with constipation, urinary retention, and dry mouth.[11] In our study, we did not examine ED or hospital admissions for dry mouth given individuals with dry mouth would likely present to outpatient (if at all), vs. acute care settings. Additionally, sialorrhea is an autonomic symptom seen in PD,[1] which may plausibly make dry mouth symptoms less problematic in this population.

We found no association between antimuscarinic use and risks of unintentional traumatic injury, fractures, falls, and dizziness/vertigo, compared to mirabegron use after adjustment via hdPS matching. Our findings are somewhat different from Crispo et al. (2016), who found that anticholinergics were significantly associated with fractures among individuals with PD; however, this study focused on overall anticholinergic burden.[6] Our short-term findings were similar to findings from Welk and colleagues, who found no difference in the 30-day odds ratio for fall/fracture in oxybutynin users and newer anticholinergic users (i.e., tolterodine, solifenacin, darifenacin, fesoterodine, trospium) compared to mirabegron users among older adults in Ontario, Canada.[17] However, these authors observed an increased HR for fall/fracture among oxybutynin users with continuous usage, [17] which we did not find when extending follow-up to 180 days in our PD sample. As noted previously, OAB symptoms are a known risk factor for falls in PD.[18] As a result, in addition to serving as a safety outcome, falls may also serve as a marker of OAB treatment effectiveness, and individuals with PD might benefit more from appropriate OAB treatment compared to the general population in the long term. In individuals with PD experiencing OAB, fall risk due to OAB must be balanced with drug-related fall risk due to OAB treatment, particularly antimuscarinics.

On the other hand, when all anticholinergic related AEs were pooled together as a composite measure, we found that antimuscarinic users had 1.2 times greater risk of AEs than mirabegron users. This finding is likely due to improved power. A similar magnitude of association was found for any acute care encounter. A prior study in PD found anticholinergic use to be associated with an increased risk of ED visits.[6] A higher number of acute care events in antimuscarinic new users may simply reflect differences in underlying characteristics and healthcare utilization. However, given our use of hdPS, which controls for unmeasured factors that are correlated with measured factors,[16] our findings suggest that this association may reflect a greater risk of AEs due to antimuscarinic use not captured in our a priori selected diagnostic codes. Additionally, it is very likely that certain specific anticholinergic related health events or symptoms may not be recognized as AEs in clinical settings.

OAB medications are often discontinued.[19-21] OAB medication persistence is particularly poor when examining other chronic medications such as statins or oral antidiabetics.[21] AEs are a frequent reasons for OAB treatment discontinuation.[11] Consequently, a treatment option with an improved safety profile, as was found in our study for mirabegron, may improve treatment persistence. Although many sociodemographic and clinical factors could impact OAB treatment persistence and adherence,[19, 20] prior studies found that individuals with OAB taking mirabegron had higher treatment persistence, longer time to discontinuation, and better adherence.[19, 22] Interestingly, among persons with OAB, those with PD are more likely to continue and adhere to OAB treatment.[19, 20] In one OAB treatment study, individuals were more likely to stop mirabegron due to a lack of efficacy or high brand-name cost, not due to adverse events.[5] There has been a general shift towards prescribing more mirabegron and less antimuscarinics for OAB management.[22] We found this same promising shift in our study.

Although we did not examine the comparative safety of individual drugs within the antimuscarinic group due to their limited use, there are differences in antimuscarinic properties, particularly with respect to blood-brain barrier permeability, which may impact the risk for AEs related to the central nervous system.[23] Of note, oxybutynin crosses the blood-brain barrier most readily.[23] We found in our study that oxybutynin was the most frequently prescribed OAB treatment overall, which may account for a substantial proportion of AEs in the antimuscarinic group. In addition, extended-release formulations of oxybutynin may have more favorable adverse effect profiles.[24] Among older Medicare beneficiaries who were new users of antimuscarinics, those started on oxybutynin immediate-release formulations were most likely to discontinue treatment.[20] Future studies should examine differences in safety within the antimuscarinic group, especially between theoretically safer drugs such as trospium, darifenacin, or fesoterodine compared to oxybutynin and tolterodine.[25]

As noted previously, we were unable to confirm if AEs were caused by OAB treatment. However, an even stronger association between antimuscarinic, compared with mirabegron use, and acute care event for anticholinergic related AEs, was found in the 30-day analysis. This finding suggested a more plausible relationship between anticholinergics and AEs. Future studies should examine the relationship between OAB treatment and AEs leveraging medical record data. As an additional limitation, we did not examine cognitive outcomes in this study. Anticholinergics are associated with dementia in older adults,[26] cognitive decline in PD,[7] and Alzheimer disease pathology in PD.[27] Another study by Bishara and colleagues among individuals with dementia also found that OAB drugs with high central anticholinergic burden (i.e., tolterodine and oxybutynin) could potentially lead to increased mortality risk and faster cognitive decline.[25] However, our study focused on short-term, compared with long-term, adverse effects of drug utilization. We chose this time frame due to the low persistence of OAB treatment[19-21] and the fact that most randomized controlled trials of OAB medications only lasted for 12 weeks.[28] Studying more acute cognitive changes is difficult in claims data due to a lack of cognitive testing, the under-coding of delirium in claims data,[29] and the plausibility of protopathic bias (i.e., bias arises when exposure to a drug occurs in response to a symptom of a specific disease that is not yet diagnosed).[6, 26] Such a study should be conducted in data with access to cognitive testing measures.

Our use of Medicare claims data allowed us to examine the comparative safety of OAB medications among a more generalizable sample of persons with PD who may not be captured in specialty care center studies. Our use of an incident new-user, active-comparator design eliminated many methodological challenges. Confounding by indication is a major challenge to studies of the negative impacts of anticholinergics.[6] Studying an indication (OAB) with clear anticholinergic and non-anticholinergic treatment options allowed us to isolate the effect of anticholinergics from OAB. OAB treatment discontinuation and adherence differs between those who have received prior OAB medication treatment and those who are treatment-naïve.[19] Our use of an incident new-user design eliminated the need to consider differences in prevalent new-users who may have switched medication types.[30] There were a few noted differences between new users of antimuscarinics and new users of mirabegron. Our use of a hdPS approach allowed us to reduce residual confounding and control for unmeasured confounders that are correlated with measured factors.[16]

OAB medications account for a substantial proportion of anticholinergics prescribed to older adults.[26] Older adults with PD are vulnerable to anticholinergics, yet are still often prescribed them.[31] Overall, our study suggests that if both mirabegron and antimuscarinics are effective in treating OAB in PD, mirabegron may be a safer treatment option. Newer treatment guidelines for OAB also include combination therapy of solifenacin and mirabegron.[32] Vibegron, another beta-3 agonist, was approved by the FDA in 2020. As treatment options shift, additional real-word evidence studies of OAB treatment safety should be conducted, especially in at-risk populations like those with neurodegenerative diseases. Such studies can help inform treatment guidelines for autonomic symptoms in PD.

HIGHLIGHTS.

• Overactive bladder (OAB) is a common nonmotor symptom of Parkinson disease (PD).

• There is lack of safety data of antimuscarinics vs. mirabegron for OAB treatment.

• Antimuscarinics were associated with greater acute care event risks vs. mirabegron.

• Long-term safety of these agents in PD should be evaluated in a prospective study.