Frequency of and risk factors for potentially inappropriate medication use in Parkinson’s disease

Abstract

Background

impairments in neurotransmitter pathways put Parkinson’s disease (PD) patients at risk for drug–disease interactions and adverse medication events.

Objective

to determine the prevalence and risk factors for potentially inappropriate medication (PIM) prescriptions, as defined by the 2015 Beers List, in PD.

Methods

cross-sectional analysis was conducted on 2014 Medicare beneficiaries with PD who had parts A, B and D coverage. The prevalence of PIM prescriptions for older adults was determined overall, and specifically for medications that can exacerbate motor symptoms or cognitive impairment in PD. Logistic regression models were constructed to determine the association between age, sex, race, geography and poverty with PIM prescriptions.

Results

the final sample included 458,086 beneficiaries. In 2014, 35.8% of beneficiaries with PD filled a prescription for at least one PIM for older adults. In total, 8.7% of beneficiaries received a PIM that could exacerbate motor symptoms and 29.0% received a PIM that could worsen cognitive impairment. After adjustment, in all models, beneficiaries who were younger, female, white, urban-dwelling and eligible for Medicaid benefits were more likely to receive a PIM.

Conclusion

PIM prescriptions are not uncommon in PD, particularly for medications that can exacerbate cognitive impairment. Future research will examine underlying drivers of sex and other disparities in PIM prescribing. Additional studies are needed to understand the impact of PIMs on disease symptoms, healthcare utilisation and patient outcomes.

Key Points

• Potentially inappropriate medication (PIM) prescribing has not been well quantified among Parkinson’s disease (PD) patients.

• Over one-third of Medicare beneficiaries with PD received at least one Beers Criteria-defined PIM in 2014.

• <10% of patients received PIMs that could worsen PD but almost 30% received PIMs that could exacerbate cognitive impairment.

• Female PD patients are more likely to receive PIMs, compared to their male counterparts.

• Further research is needed to understand the drivers of PIM prescriptions and the impact of PIM prescriptions on outcomes in PD.

Introduction

Polypharmacy (concomitant use of ≥5 medications) [1] is common in Parkinson’s disease (PD) [2,3]. Medications are indicated in PD for motor and non-motor symptoms, common co-morbid conditions and adverse effects of dopaminergic treatment [2,4–6]. The literature on drug safety in PD is limited and focuses on drug–drug interactions [7,8], specific indications [9] and specific subpopulations [10].

PD motor and non-motor symptoms are associated with impairments in neurotransmitter pathways involving dopamine, acetylcholine, serotonin and norepinephrine [11–16]. This neurochemical disruption places PD patients at risk of adverse medication events that can ‘worsen’ PD symptoms. Such events are defined by Lindblad et al. [17] as drug–disease interactions—‘exacerbations by medications of pre-existing diseases, conditions, or syndromes’ [17]. However, the term ‘drug–disease interaction’ is not common in the PD literature, and such interactions have not been nationally quantified.

The Beers Criteria for Potentially Inappropriate Medication Use in Older Adults names potentially inappropriate medications (PIMs) that should be avoided in adults ≥65 years of age or may lead to drug–disease interactions [18]. Given that PD is a disease of ageing [19], the Beers Criteria can be used to quantify and understand PIMs in PD.

The objectives of this study were to quantify and determine risk factors for PIM prescriptions in older, Medicare beneficiaries diagnosed with PD.

Methods

Standard protocol approvals and patient consents

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

Data source and study sample

This cross-sectional study was conducted with CMS 2014 administrative claims data. Medicare is a government-run program that provides health insurance for 96% of Americans ≥65 years of age [20]. Specifically, we used the Medicare Carrier, Master Beneficiary Summary (MBSF) and Prescription Drug Event (PDE) Research Identifiable Files [21]. Carrier files include diagnosis and procedure codes documented in inpatient, outpatient and laboratory settings. MBSF files contain data on Medicare eligibility, demographics and 27 chronic medical conditions [22]. The PDE file contains prescription information for medications received from a pharmacy using Medicare coverage.

Study sample

To be included in the analysis, we required beneficiaries to be ≥65 years of age and to have Medicare part A (inpatient), B (outpatient) and D (prescription) coverage for the full 2014 calendar year, along with a minimum of one prescription claim. Additionally, the sample was restricted to those with at least two PD diagnosis (ICD-9 code 332.0) claims in 2014 [23] but no diagnosis of atypical parkinsonism. We excluded beneficiaries who died in 2014 and/or had missing data for any of the primary study variables.

Variables

Beneficiary characteristics

Demographic data included sex, race and age. For low-income older adults with Medicare, additional insurance is provided through Medicaid; so, Medicaid eligibility was used as a proxy for lower socio-economic status. Beneficiary counties of residence were used to determine geography (urban, suburban or rural) [24].

Beers Criteria drug use and categorisation

We used the 2015 version of the Beers Criteria to identify claims for drugs that may be inappropriate in older adults with PD [18]. The 2015 version is based on evidence collected through 2014; consequently, we deemed it the most temporally relevant to our analysis. PIMs were categorised into three non-mutually exclusive groups (Supplementary data, Appendix 1). We excluded PIMs with high over-the-counter (OTC) availability and medications, such as sliding scale insulin, whose PIM status depended on certain requirements difficult to verify in 1 year of claims data (e.g. a specific clinical indication, a history that alternative medications had previously failed, a past health event justifying use, prolonged use, a required laboratory assessment or ongoing patient–physician reassessments of risk-benefit). We also excluded the dopamine-receptor antagonist antipsychotics clozapine and quetiapine from PIM designation, as evidence suggests that these drugs are the safest choices for individuals with PD [18]. Based on the literature, clinical experience and the classification change from the 2015 to 2019 Beers List, aripiprazole was categorised as a PIM [25–27].

The first grouping was the most broad and included Beers Criteria-defined PIMs that carried the highest level of recommendation for avoidance in any adult age 65 and older [18]. The Beers Criteria contains a recommendation against the use of certain medications in PD patients; these medications consist almost exclusively of those strongly capable of worsening the motor symptoms of PD [18]. These drug–PD interaction PIMs defined the second group. Third, the Beers Criteria also discourages the use of PIMs that may exacerbate cognitive impairment. PD patients are at high risk for cognitive impairment [28]; consequently, we lastly examined drug–cognitive impairment interaction PIMs.

Statistical analysis

The analysis was conducted in SPSS Statistics v25.0 (IBM Corp.). Statistical significance was determined by a two-sided α = 0.05. Descriptive analysis was conducted on beneficiary characteristics and PIMs utilisation. For each PIM grouping, bivariate analysis was used to compare beneficiary characteristics of those prescribed at least one PIM in 2014 and those without such a prescription.

Crude and adjusted logistic regression models were constructed to examine whether any beneficiary characteristics were associated with a prescription for each PIM grouping. Adjusted models included age, sex, race, Medicaid status and geography.

Evaluating PIMs that may exacerbate cognitive impairment in a mixed sample (with and without cognitive impairment) may overestimate the prevalence of such PIMs. Consequently, as a sensitivity analysis, the prevalence of this group of PIMs was examined only among those with a diagnosis of dementia per the MBSF.

Results

Sample characteristics

The final study sample included 458,086 beneficiaries with PD. In total, PD patients in the sample filled 31,232,420 prescriptions in 2014 (mean = 68, SD = 54.1). Part D beneficiaries with PD were predominantly between the age of 75 and 84 (46.5%) (Table 1). Beneficiaries with PD were primarily white (86.4%), female (54.7%) and living in an urban setting (80.8%). Approximately 30% of beneficiaries were dual Medicare/Medicaid eligible.

Table 1 .

Descriptive characteristics of Medicare beneficiaries with Parkinson’s disease and at least one 2014 prescription fill (n = 458,086)

Beneficiary characteristic	n (Col %)
Age, years
65–69	21,681 (4.7)
70–74	81,297 (17.7)
75–79	105,891 (23.1)
80–84	107,376 (23.4)
85–89	87,484 (19.1)
≥90	54,357 (11.9)
Sex
Male	207,420 (45.3)
Female	250,666 (54.7)
Race
White	395,638 (86.4)
Black	29,361 (6.4)
Other/unknown	7,386 (1.6)
Asian	10,974 (2.4)
Hispanic	13,309 (2.9)
Native American	1,418 (0.3)
Geography
Urban	369,524 (80.8)
Suburban	61,936 (13.5)
Rural	25,874 (5.7)
Medicaid eligible
No	322,697 (70.4)
Yes	135,389 (29.6)

Beers Criteria PIM prescribing in PD

PIMs for older adults overall

In total, 35.8% (n = 164,117) of beneficiaries with PD received at least one PIM for older adults. The proportion of beneficiaries prescribed a PIM decreased with age from 43.5% among those 65–69 to 26.5% among those ≥90 years of age. Beneficiary characteristics significantly associated with receiving a PIM in adjusted models included age, sex, race, geography and Medicaid eligibility. After adjustment, the odds of receiving a prescription for a PIM declined with each 1-year increase in age over 65 (Adjusted odds ratio [AOR] = 0.97, 95%; CI: 0.97, 0.97) and was higher for women (AOR = 1.03; 95% CI: 1.02, 1.05) (Table 2). Minorities had a significantly lower odds of receiving a prescription for a PIM; dual Medicare/Medicaid eligibility was associated with a greater odds of PIM receipt (AOR 1.16, 95% CI: 1.15, 1.18).

Table 2 .

Factors associated with receipt of any prescription for a 2015 Beers Criteria potentially inappropriate medication (PIM) for any person ≥65 years of age among 2014 Medicare beneficiaries with Parkinson’s disease (n = 458,086)

Beneficiary characteristic	PIM prescriptionn (row %)	Crude odds ratio(95% CI)	AOR (95% CI)a
Age, per 1-year increase	—	0.97 (0.97, 0.97)	0.97 (0.97, 0.97)
Sex
Male	74,521 (35.9)	REF	REF
Female	89,596 (35.7)	0.99 (0.98, 1.00)	1.03 (1.02, 1.05)
Race
White	144,250 (36.5)	REF	REF
Black	8,929 (30.4)	0.76 (0.74, 0.78)	0.71 (0.69, 0.73)
Other/unknown	2,470 (33.4)	0.88 (0.83, 0.92)	0.81 (0.77, 0.85)
Asian	3,488 (31.8)	0.81 (0.78, 0.85)	0.77 (0.74, 0.81)
Hispanic	4,487 (33.7)	0.89 (0.85, 0.92)	0.86 (0.82, 0.89)
Native North American	493 (34.8)	0.93 (0.83, 1.04)	0.84 (0.75, 0.94)
Geography
Urban	131,962 (35.7)	REF	REF
Suburban	22,679 (36.6)	1.04 (1.02, 1.06)	1.00 (0.98, 1.01)
Rural	9,220 (35.6)	1.00 (0.97, 1.02)	0.96 (0.93, 0.98)
Medicaid eligible
No	113,096 (35.0)	REF	REF
Yes	51,021 (37.7)	1.12 (1.11, 1.14)	1.16 (1.15, 1.18)

^aAdjusted model includes age (continuous), sex, race, geography and Medicaid eligibility.

PIMs that can worsen PD motor symptoms

In total, 8.7% (n = 39,943) of the sample was prescribed a drug that can exacerbate PD motor symptoms. In adjusted models, all beneficiary characteristics were significantly associated with odds of receipt of these PIMs with the most striking increases in odds seen for female sex (AOR 1.27; 95% CI: 1.25, 1.30) and Medicaid benefits (AOR = 2.42, 95% CI: 2.37, 2.45) (Table 3).

Table 3 .

Factors associated with receipt of any prescription for a 2015 Beers List medication with potential drug–disease interactions among 2014 Medicare beneficiaries with Parkinson’s disease (n = 458,086)

Beneficiary characteristic	PIMs that can worsen PD motor symptoms			PIMs that can worsen cognitive impairment
	PIM prescription n (row %)	Crude odds ratio (95% CI)	AORa (95% CI)	PIM prescription n (row %)	Crude odds ratio (95% CI)	AORa (95% CI)
Age, per 1-year increase	—	0.96 (0.96, 0.97)	0.96 (0.96, 0.97)	—	0.97 (0.97, 0.97)	0.97 (0.97, 0.97)
Sex
Male	15,481 (7.5)	REF	REF	57,168 (27.6)	REF	REF
Female	24,462 (9.8)	1.34 (1.31, 1.37)	1.27 (1.25, 1.30)	75,783 (30.2)	1.14 (1.12, 1.15)	1.19 (1.17, 1.20)
Race
White	34, 106 (8.6)	REF	REF	117,586 (29.7)	REF	REF
Black	3,127 (10.7)	1.26 (1.21, 1.31)	0.93 (0.89, 0.97)	6,771 (22.6)	0.71 (0.69, 0.73)	0.64 (0.62, 0.66)
Other/unknown	532 (6.7)	0.82 (0.75, 0.90)	0.66 (0.61, 0.73)	1,919 (26.0)	0.83 (0.79, 0.87)	0.76 (0.72, 0.80)
Asian	734 (6.7)	0.76 (0.70, 0.82)	0.50 (0.44, 0.54)	2,659 (24.2)	0.76 (0.72, 0.79)	0.69 (0.66, 0.73)
Hispanic	1,298 (9.8)	1.15 (1.08, 1.21)	0.76 (0.72, 0.81)	3,631 (27.3)	0.89 (0.85, 0.92)	0.82 (0.79, 0.86)
Native N. American	146 (10.3)	1.22 (1.02, 1.44)	0.87 (0.73, 1.04)	384 (27.2)	0.88 (0.78, 0.99)	0.77 (0.69, 0.87)
Geography
Urban	32,055 (8.7)	REF	REF	106,769 (28.9)	REF	REF
Suburban	5,723 (9.2)	1.07 (1.04, 1.10)	1.01 (0.98, 1.04)	18,497 (29.9)	1.07 (1.04, 1.10)	1.00 (0.98, 1.02)
Rural	2,109 (8.2)	0.93 (0.89, 0.98)	0.88 (0.84, 0.93)	7,481 (28.9)	0.93 (0.89, 0.98)	0.95 (0.93, 98)
Medicaid eligible
No	20,772 (6.4)	REF	REF	89,928 (27.9)	REF	REF
Yes	19,171 (14.2)	2.40 (2.35, 2.45)	2.42 (2.37, 2.45)	43,023 (31.8)	1.21 (1.19, 1.22)	1.25 (1.23, 1.26)

^aAdjusted model includes age (continuous), sex, race, geography and Medicaid eligibility.

PIMs that can worsen cognitive impairment

Among beneficiaries with PD, irrespective of cognitive status, 29.0% (n = 132,951) were prescribed a PIM that can worsen cognitive impairment. In adjusted models, all beneficiary characteristics were associated with receipt of one of these PIM prescriptions (Table 3). Again, women (AOR = 1.19, 95% CI: 1.17, 1.20) and dual Medicare/Medicaid-eligible beneficiaries (AOR = 1.25, 95% CI: 1.23, 1.26) had increased odds of receiving a PIM that could worsen cognition. Minority race/ethnicity was associated with lower odds of receiving a PIM in this category. In a sensitivity analysis restricted to those with a pre-existing diagnosis of dementia (n = 159,392), 31.2% filled a prescription for a PIM that can worsen cognition.

Discussion

PD is a common neurodegenerative disease that currently cannot be prevented and has no disease-modifying therapy [29]. These factors highlight the need to optimise patient care and health outcomes. In this population-based study of older adults diagnosed with PD in the United States, we examined the use of medications that could be inappropriate in any older adult and then considered two potential types of drug–disease interactions—medications that could worsen the motor symptoms of PD and medications that could worsen one key non-motor symptom of PD—cognition.

In our study, 35.8% of beneficiaries with PD received a PIM that the Beers Criteria strongly recommends against using in any older adult. Two prior studies conducted in the general population using the more restrictive 2012 Beers Criteria found an even higher prevalence of PIM prescriptions among older adults (between 40.8 and 55.7%) [30,31]. The lower observed prevalence of PIM prescriptions in our study may be related to several factors. Most importantly, our study excluded some Beers Criteria PIMs that we decided a priori would be difficult to study using prescription claims data. We were also unable to examine in-hospital or OTC prescriptions, such as non-steroidal anti-inflammatory drugs (NSAIDs). NSAIDs were found to be the most common PIM used by Medicare beneficiaries in a prior study [30]. OTC PIM utilisation in PD should be examined in a future study. PIM prescribing in the general population appears to be on the decline [30,31]. Our findings in PD patients using more recent data could reflect larger declines in PIM prescribing. Lastly, specialty care is associated with fewer PIM prescriptions [31]. Although neurologist care is underutilised in PD, neurologists are involved in the care of 58% of Medicare beneficiaries with PD [32], which may reduce the risk of PIM prescriptions [32]. Future studies will explore whether PIM prescriptions differ by provider specialty for PD. Additionally, the prevalence of PIM prescriptions among patients with PD in the United States should be compared to countries that use a multidisciplinary PD care model led by geriatricians, who have greater training and expertise in PIM avoidance. Although the prevalence of PIM prescriptions in older adults with PD may be lower than the general older adult population, over one-third of beneficiaries with PD are at risk of negative health outcomes due to PIMs, which is considerable.

In our study, 8.7% of the study sample was prescribed PIMs that could worsen PD motor symptoms by disrupting dopaminergic transmission. PIMs in this category include antipsychotics and antiemetics. Nausea/vomiting and psychosis are common non-motor symptoms and complications of PD therapy, with psychosis reported in up to 60% of PD patients [33, 34]. Consequently, although the potential to receive a PIM is high based on clinical need, actual prescribing was <10%. This finding is not surprising given that there have been scientific data [35] and clinical guidelines recommending against the use of dopamine receptor antagonists in persons with PD [36–38]. This level of prescribing likely reflects safer prescribing practices that are consistent with providers following evidence-based guidelines in PD. These data also suggest that evidence for safer prescribing can integrate into clinical decision making and clinical practice.

Although the evidence that supports the avoidance of dopamine receptor blocking agents in PD is extensive and longstanding, there is limited evidence and clinician acceptance of what medications should be avoided to protect cognition in PD, even though this non-motor symptom has emerged as a risk factor for institutionalisation, hospitalisation and longer hospitalisation durations [39–41]. Perhaps reflecting this lack of evidence, we found that PIM prescriptions capable of exacerbating cognitive impairment were common and prescribed to 29.0% of individuals in our overall sample. Not all persons with PD have cognitive impairment or dementia, but all are vulnerable to this disorder due to the disease-related disruption of cholinergic pathways in the brain. The potential population-level burden of this drug–cognitive impairment interaction is therefore considerable. We found, contrary to a prior study [42], that individuals with pre-existing dementia diagnoses were more likely to receive these PIMs (31.2%). However, Sheu et al. found that extensive anticholinergic use is associated with dementia in PD [43]. One prior study demonstrated that a subset of these PIMs have detrimental impacts on healthcare outcomes, such as fractures, in PD [44]; however, additional research is needed to examine the full impact of these PIMs on outcomes in PD and inform care guidelines.

Prescriptions for all PIMs categories we studied were higher for women. This sex difference may simply reflect the higher PIM burden found in older women in general [45]. Consistent with our finding, a prior study found that inappropriate co-prescribing for dementia was more common for women with PD [46]. Women are less likely to receive neurologist care, a healthcare process associated with improvements in other health outcomes for PD [32,47]. Consequently, women could receive lower quality prescribing due to reduced specialist care. However, minorities are also less likely to see a specialist after being diagnosed with PD [32]. These beneficiaries were less, not more, likely to receive a PIM, suggesting that a lack of specialty care does not wholly explain the observed sex difference. The PIMs on the Beers List are primarily medications used for psychiatric conditions and other non-motor symptoms in PD. In general, women have more non-motor symptoms than men [48,49]. Racial/ethnic differences in non-motor symptoms of PD have not been well studied, but older blacks in the general population are less likely to receive treatment for depression, a common psychiatric illness in PD [49,50]. Our data may reflect actual differences in the prevalence of certain non-motor symptoms or disparities/bias in provider screening, detection and action for clinical indications for PIMs in PD. More research is needed to examine the prevalence and treatment of conditions that have PIM options in PD, particularly studies that also examine differences in screening, diagnosis and treatment by sex and race.

In this study, we used Medicare data to examine national estimates of PIM prescriptions in beneficiaries with PD. Strengths of our study include sample size and diversity that is not possible with health system- or disease registry-based studies, which often have limited information on rural-dwelling, very old and minority PD patients. With respect to limitations, most importantly, as discussed previously, this study likely underestimated overall PIM prescriptions by excluding several Beers Criteria PIMs. Second, administrative data, like medical records, cannot confirm PD pathology; so, it is possible that some of our PD cases have mixed pathology or an incorrect diagnosis. We were unable to examine clinical characteristics that may impact prescribing patterns (e.g. cognitive and motor function test scores) that are unavailable in administrative data. Finally, the Beers Criteria does not provide certainty about which drugs are inappropriate; it contains those that ‘may be’ inappropriate, based on the current evidence [51]. Expert opinion may differ with regard to which drugs should be considered inappropriate [17]. Despite these limitations, we provide evidence that PIM prescriptions are common among Medicare beneficiaries with PD, particularly for drugs that can exacerbate cognitive impairment. Additional research is needed to determine the drivers of observed differences in PIM exposure, and to quantify the impact, if any, PIMs have on PD disease course, healthcare utilisation and outcomes.

Declaration of Conflicts of Interest

None.

Declaration of Funding

This work was supported by a grant from the National Institute of Neurological Diseases and Stroke of the National Institutes of Health [R01 NS099129]. The funding source 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. D.S.A. received support from the National Institutes of Health [grant number R01 NS099129] and the Parkinson’s Foundation. S.H. received support from the National Institutes of Health [grant numbers R01 AG025152, R01 AG060975 and R01 NS099129]. D.W. received support from the National Institutes of Health [grant number R01 NS099129]. D.W. has also received research funding or support from Michael J. Fox Foundation for Parkinson’s Research, Alzheimer’s Therapeutic Research Initiative (ATRI), Alzheimer’s Disease Cooperative Study (ADCS) and the International Parkinson and Movement Disorder Society (IPMDS); honoraria for consultancy from Acadia, Aptinyx, Biogen, Bracket, CHDI Foundation, Clintrex LLC, Enterin, F. Hoffmann-La Roche Ltd, Ferring, Promentis, Sunovion and Takeda and license fee payments from the University of Pennsylvania for the QUIP and QUIP-RS. S.L.G. received support from the National Institutes of Health [grant numbers R01 AG056326, U01 AG006781 and R01 NS099129] as well as the Centers for Disease Control and Prevention [grant number U01CE002967]. D.X. received support from National Institutes of Health [grant number R01 NS099129]. A.W.W. received financial support from National Institutes of Health [grant number R01 NS099129], the Parkinson’s Foundation and the University of Pennsylvania.