Abstract
Background:
Parkinson disease (PD) patients are at increased risk of serious injury, such as fall-related fractures. Prescription medications are a modifiable factor for injury risk.
Objectives:
To determine the extent to which a serious injury requiring hospitalization affects prescribing of potentially inappropriate medications (PIMs) among older adults with PD.
Methods:
We conducted a quasi-experimental difference-in-difference (DID) study using 2013–2017 Medicare data. The cohort consisted of beneficiaries with PD hospitalized for injury versus for other reasons. PIMs were classified into PD and injury-relevant categories (CNS-active PIMs, PD motor symptom PIMs, PD non-motor symptom PIMs, PIMs that reduce bone mineral density). We estimated mean standardized daily doses (SDDs) of medications within each PIM category before and at 3, 6, and 12 months after hospitalization. We used generalized linear regression models to compare changes in mean SDDs for each PIM category between the injury and non-injury group at each timepoint, adjusting for biological, clinical and social determinants of health variables.
Results:
Both groups discontinued PIMs and/or reduced PIM doses after hospitalization. There were no between-group differences in mean SDD changes, after covariate adjustment, for any PIM category, except for the CNS-active PIMs category at 3 months (DID p-value = 0.00) and for the category of PIMs that reduce bone mineral density at all timepoints (DID p-values = 0.02, 0.04, 0.02 at 3, 6, and 12 months).
Conclusions:
Similar patterns of PIM among persons with PD after hospitalization for serious injury versus for other reasons may represent a missed opportunity to deprescribe high-risk medications during care transitions.
Keywords: Parkinson disease, Potentially inappropriate medication, Fall-risk-increasing drug, Unintentional traumatic injury, Fall-related fracture
1. Introduction
Parkinson disease (PD) is the second most prevalent neurodegenerative disease after Alzheimer dementia, affecting people ages 65 or older [1]. PD is characterized by dopaminergic and cholinergic neuronal death, leading to motor, autonomic, visual, and cognitive dysfunction. In addition to medications indicated for motor and non-motor symptoms, PD patients often require other preventative and symptomatic treatments for comorbid conditions common in older adult populations. As a result, polypharmacy (i.e., concurrent use of ≥5 different prescription medications) is common in PD [2,3]. The lack of a viable neuroprotective or neuropreventative strategy for PD has significantly increased researcher and patient interest in optimizing patient outcomes and identifying avoidable harms. Drug-drug and drug-disease interactions may significantly threaten PD patient health and outcomes.
Persons living with PD are at increased risk of injurious falls and fractures due to the progressive symptoms of the disease itself (gait shuffling and freezing, muscle bradykinesia and rigidity, visual dysfunction, orthostasis, and cognitive dysfunction) and also adverse effects (AEs) of medications, resulting in immobilization, poor quality of life, and reduced life expectancy [4,5]. Prescription medications represent a major modifiable factor for fall risk and fall-related fractures in the older adult population [6]. Previous studies have shown no or modest change in the prescribing of central nervous system (CNS)-active medications associated with falls or of medications that reduce bone density following a fall-related injury [6,7]. These studies did not include large numbers of PD patients, who represent a potentially more vulnerable population. In response to this knowledge gap and to inform the care of persons with PD, the objectives of our study were (1) to describe the use of medications categorizable as potentially inappropriate for use in the older adult PD population and (2) to determine the extent to which hospitalization for a fall-related fracture or an unintentional traumatic injury, as compared to hospitalization for other reasons, affects potentially inappropriate medication (PIM) use in PD.
2. Methods
2.1. Overview
We performed a quasi-experimental difference-in-difference (DID) study examining and comparing changes in the use of PIM use before and after hospitalization for fracture or unintentional traumatic injury versus hospitalization for other reasons, using health care claims and prescribing data from persons who receive insurance from the U.S. Medicare program and are receiving treatment for PD. 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).
2.2. Data source
The CMS Medicare program is the primary insurer of the adult population in the U.S. that is ages 65 and above. We used the following CMS 2013–2017 Research Identifiable Files to build our analytic datasets and test our study hypotheses: the Master Beneficiary Summary File (MBSF), Chronic Conditions Data Warehouse (CCW), Carrier, Medicare Provider Analysis and Review (MedPAR), and Prescription Drug Event (PDE) files. These files contained beneficiary demographics and Medicare enrollment information (MBSF), specific comorbid conditions (CCW), medical diagnoses and procedures (Carrier), inpatient diagnoses (MedPAR), and prescription drug claims (PDE).
2.3. 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.
2.4. Overview of study cohort inclusion and exclusion criteria
The study cohort consisted of Medicare beneficiaries who met insurance coverage criteria, had a qualifying diagnosis of PD, and experienced a hospitalization, and were community-dwelling before and after discharge. Below, we detail the steps to derive our analytical cohort.
Medicare eligibility.
We required beneficiaries to have complete Part A (inpatient), B (outpatient), and D (prescription) coverage with no other healthcare coverage (i.e., Medicare Advantage, full or partial dual eligibility, or Retiree Prescription Benefits) at the time of a qualifying PD diagnosis and for at least six months prior to a qualifying hospitalization.
Parkinson disease (PD) diagnosis.
A previously published algorithm [8] was used to identify Medicare beneficiaries with PD during the study period. Individuals were required to have an evaluation and management claim for PD, documented using International Classification of Diseases, 9th and 10th Revisions, Clinical Modification (ICD-9-CM and ICD-10-CM) diagnosis codes (i.e., 332.0, and G20), from a physician or advanced practice provider in the Carrier File. They were also required to have a prescription for an antiparkinsonian medication within six months before or after the qualifying PD claim (Supplemental Table 1). Within this PD cohort, we identified all persons who had a qualifying hospital admission, as detailed below.
Hospitalization.
A qualifying hospitalization was defined as a non-elective hospitalization occurring after PD diagnosis that did not result in inpatient death. The first-ever hospitalization that was for unintentional traumatic injury or fall-related fracture (excluding injuries/fractures with co-diagnosis of bone cancer, osteomyelitis or other pathological fractures) captured in the database was identified using previously published methods and is hereafter referred to as hospitalization for serious injury (see Supplemental Table 2 for diagnosis code algorithms) [6,7,9]. The admission date for the qualifying serious injury event was considered as the "index" date. Individuals with PD who were hospitalized for all other reasons were eligible to serve as the matched "non-injury" group, as described below. We only included the first-ever qualifying serious injury hospitalization for each person, and people in the "injury" group were not eligible to potentially serve as a non-injury match. To allow for at least a 6-month continuous healthcare before the first qualifying hospitalization event, the earliest allowable hospitalization index date was July 1, 2013.
Identifying community-dwelling individuals.
We used previously published methods [6] to identify persons in our PD cohort who were community-dwelling after their qualifying hospitalization, to ensure adequate capture of prescription drug events. We required patients to be community-dwelling for at least 45 of the first 180 days (25%) after discharge [6]. A person was designated as community-dwelling for each day there were no claims for hospice, inpatient rehabilitation, skilled nursing facility care. This definition allows for discharge to nursing home or a rehabilitation facility, as would be common in this population, but requires a return to the community. Persons enrolled in hospice in the 30 days prior to the index date of a qualifying hospitalization, or subsequently enrolled within 6 months of discharge were excluded.
Potentially inappropriate medication (PIM) exposure.
The Beers Criteria® [10] are developed by the American Geriatrics Society and contain medications that should be avoided in older adults, including those with specific diseases or risk factors. From this list, we extracted relevant PIMs for persons with PD, with a focus on (1) CNS-active medications that may increase the risk of falls and (2) medications that may result in a drug-disease interaction for persons with PD. We also extracted medications that are known to reduce bone mineral density (BMD) previously defined in a study by Munson and colleagues [6]. Using these data, we defined several study specific PIM categories: (1) "CNS-active PIMs" (CNS-active medications that are not recommended in older adults in general) [10,11], (2) "PD motor symptom PIMs" (medications that are not preferred for treatment of PD motor symptoms or persons with dopaminergic neuron dysfunction) [12], (3) "PD non-motor symptom PIMs" (medications that are not preferred for treatment of psychiatric, urinary, sleep disorders in older adults, plus high potency anticholinergics, which are not recommended for adults at risk of cognitive or cholinergic neuron dysfunction) [10–13], and "BMD PIMs" (medications with some evidence of reducing bone mineral density) [6,11,12,14]. There was expected partial overlap between categories, notably between the CNS-active and PD non-motor symptom PIM categories. The full medication list for each PIM category is included in Supplemental Table 3.
Exposure to medication within a given PIM category was determined from data contained in PDE claims, specifically using National Drug Codes (NDCs) from Lexicon Plus (Oracle Cerner: Austin, Texas), which distinguish all drugs in the U.S. by manufacturer, active ingredient, strength, dosage form, route of administration, and packaging. Only oral and specific non-injectable formulations (inhalation powder, transdermal patch) were considered. NDCs, start date, quantity dispensed, number of fills, and days’ supply variables were used to ascertain all medications a person possessed and was able to use as of the day before a qualifying hospitalization, and then at months 3, 6, and 12 after discharge. We did not allow grace periods or gaps between prescription fills.
We used previously published methods to estimate the standardized daily dose (SDD) [7] of each medication at each time point by dividing the total daily dose (in milligrams) by the minimum effective geriatric daily dose (in milligrams) based on Lexicomp® Drug Reference database. SDDs of individual drugs were summarized for each PIM category [7]. We also categorized the pre-hospitalization SDD into three groups using published recommended cutoffs: low (0 < SDD<1.0), medium (1.0 = SDD≤3.0), and high (SDD>3.0) [15].
2.5. Outcome
The primary outcome was the change in mean SDD from the day prior to index hospitalization to day 89 (3 months) and 179 (6 months) and 364 (12 months) following discharge. This outcome was estimated for each PIM category separately (Supplemental Fig. 1).
2.6. Covariates
All covariates were derived from data available during the 6 month period prior to a qualifying hospitalization, and were distributed as follows: biological – age, sex; social – race/ethnicity, a zip code based measure of social deprivation [16]; clinical – Charlson-Elixhauser co-morbidity score [17], claims-based frailty index (CFI) [18], prior dementia diagnosis, prescription claims for cognitive enhancing medications (more specifically medications indicated for the treatment of dementia or mild cognitive impairment by the U.S. Food and Drug Administration) and the levodopa equivalent daily dose (LEDD) of non-surgical PD motor therapies [19,20].
2.7. Matched non-injury group definition
To account for temporal trends in medication prescribing and consumption associated with hospitalization itself, we identified persons with PD who had a hospitalization for reasons other than injury (± 1 calendar year from index date of their corresponding injury match) [7] and who were also prescribed drugs from the PIM categories of interest. Within each PIM category of interest, persons with PD hospitalized for non-injury reasons were matched up to 4 to 1, without replacement, to persons hospitalized for serious injury on the following characteristics: age group (65–69, 70–74, 75–79, 80–84, ≥85 years), sex, race (White vs. Nonwhite), CFI category [nonfrail (CFI<0.10), prefrail (0.10–0.19), mildly frail (0.20–0.29), moderately frail (0.30–0.39), and severely frail (≥0.40)] [18], and baseline SDD category (low, medium, high).
2.8. Statistical analyses
SAS v9.4 (Cary, North Carolina) was used to build the analytic dataset and analyze the study cohort. We defined an alpha level of 0.05 for all statistical tests. Baseline biological, clinical, and social characteristics were compared between persons with PD patients hospitalized for injury versus for other reasons, using standardized mean difference or standardized difference in proportions, which is preferable to chi-square analyses and t-tests for large sample sizes. An absolute value of standardized difference greater than 0.10 signaled an imbalance between groups. The mean change in SDD in each medication category of interest at 3, 6, and 12 months following hospital discharge was estimated using linear regression, adjusting for year of cohort entry, clinical encounter type at index, the combined Charlson-Elixhauser comorbidity score, the social deprivation index, and the LEDD, then compared between groups using a generalized linear model at each time point.
We performed sensitivity analyses (1) excluding opioids from the CNS-active PIM category, as opioids could be used for pain management due to injury or fracture, and (2) excluding antidepressants from the CNS-active PIM category, as the decision to discontinue antidepressants generally requires either specialized assessments and/or specialist involvement due to complex aspects of mood disorder management.
Secondary analyses were conducted to explore potential discontinuation and medication tapering trends in the PD injury group and their matched PD non-injury group, including (1) evaluating the proportion of people who discontinued all medications from a given PIM category at 3 months (e.g., SDD of CNS-active PIM = 0) and who sustained discontinuation through one year following discharge, and (2) examining the change in SDD only among people who continued medications in a given PIM category (e.g., SDD of CNS-active PIM >0 throughout follow-up) to assess possible tapering (i.e., dose reduction and/or discontinuation of at least one medication within the PIM category) at 3, 6, and 12 months.
3. Results
3.1. Baseline characteristics of study cohort
Table 1 displays the baseline characteristics of the study cohort. We identified 14,646 individuals with PD meeting the inclusion/exclusion criteria with a hospitalization for serious injury in 2013–2017, of whom 9473 (64.7%) were able to be matched with up to 4 PD individuals hospitalized for other reasons (n = 32,487). The most common primary reasons for hospitalization among persons in the matched non-injury group were septicemia, urinary tract infection, PD, and pneumonia (Supplemental Table 4). On the day prior to the index date, 6586 (69.5%) individuals with PD in the injury group and 22,185 (68.2%) in the non-injury group were prescribed CNS-active PIMs; 2638 (27.9%) and 7934 (24.4%) were prescribed PD motor symptom PIMs; 3033 (32.1%) and 9772 (30.1%) were prescribed PD non-motor symptom PIMs; 5627 (59.4%) and 19,442 (59.8%) were prescribed BMD PIMs, respectively (Table 1). The non-injury group had a greater combined Charlson Elixhauser comorbidity score than the injury group (standardized difference = –0.26). Characteristics that were used in the matching process (age, sex, race, CFI category, baseline SDD category) as well as other baseline characteristics (e.g., social deprivation index, prior dementia diagnosis and prior cognitive enhancing medication use, mean LEDD) were well-balanced between groups (Table 1).
Table 1.
Baseline characteristics of PD injury and matched PD non-injury groups.
| Characteristics | Injury group (n = 9473) | Matcheda non-injury group (n = 32,487) | Standardized difference‡ |
|---|---|---|---|
| Age at index, mean (SD) | 79.6 (6.72) | 79.6 (6.79) | − 0.01 |
| Sex, n (col%) | |||
| Male | 4485 (47.3%) | 16,020 (49.3%) | − 0.04 |
| Female | 4988 (52.7%) | 16,467 (50.7%) | − |
| White race, n (col%) | 9159 (96.7%) | 31,696 (97.6%) | − 0.05 |
| Social deprivation index, mean (SD) | 39.2 (25.81) | 40.3 (26.08) | − 0.04 |
| Claims-based frailty index, mean (SD) | 0.2 (0.06) | 0.2 (0.06) | − 0.1 |
| Claims-based frailty index categories, n (col%) | |||
| 0−0.09 | 2 (0.0%) | 2 (0.0%) | − 0.1 |
| 0.1−0.19 | 3068 (32.4%) | 9648 (29.7%) | − |
| 0.2−0.29 | 5224 (55.1%) | 18,733 (57.7%) | − |
| 0.3−0.39 | 1158 (12.2%) | 4063 (12.5%) | − |
| 0.4+ | 21 (0.2%) | 41 (0.1%) | − |
| Charlson Elixhauser comorbidity score, mean (SD) | 2.2 (2.60) | 2.9 (2.95) | −0.26 |
| Prior dementia diagnosis, n (col%) | 4677 (49.4%) | 17,096 (52.6%) | − 0.07 |
| Prior dementia drug, n (col%) | 2189 (23.1%) | 7218 (22.2%) | 0.02 |
| LEDD, mean (SD) | 446.4 (407.18) | 411.5 (399.27) | 0.09 |
| SDD categories of CNS-active PIMs on day prior to index date | |||
| 0 | 2887 (30.5%) | 10,302 (31.7%) | 0.05 |
| 0 < SDD<1.0 (low) | 733 (7.7%) | 2278 (7.0%) | − |
| 1.0≤SDD≤3.0 (medium) | 2821 (29.8%) | 9596 (29.5%) | − |
| 3+ (high) | 3032 (32.0%) | 10,311 (31.7%) | − |
| SDD categories of PD motor symptom PIMs on day prior to index date | |||
| 0 | 6835 (72.2%) | 24,553 (75.6%) | 0.1 |
| 0 < SDD<1.0 (low) | 215 (2.3%) | 521 (1.6%) | − |
| 1.0≤SDD≤3.0 (medium) | 1054 (11.1%) | 3154 (9.7%) | − |
| 3+ (high) | 1369 (14.5%) | 4259 (13.1%) | − |
| SDD categories of PD non-motor symptom PIMs on day prior to index date | |||
| 0 | 6440 (68.0%) | 22,715 (69.9%) | 0.04 |
| 0 < SDD<1.0 (low) | 109 (1.2%) | 222 (0.7%) | − |
| 1.0≤SDD≤3.0 (medium) | 1628 (17.2%) | 5156 (15.9%) | − |
| 3+ (high) | 1296 (13.7%) | 4394 (13.5%) | − |
| SDD categories of BMD PIMs on day prior to index date | |||
| 0 | 3846 (40.6%) | 13,045 (40.2%) | 0.09 |
| 0 < SDD<1.0 (low) | 191 (2.0%) | 479 (1.5%) | . |
| 1.0≤SDD≤3.0 (medium) | 2919 (30.8%) | 10,016 (30.8%) | . |
| 3+ (high) | 2517 (26.6%) | 8947 (27.5%) | . |
| Year of cohort entry, n (col%) | |||
| 2013 | 959 (10.1%) | 2748 (8.5%) | 0.1 |
| 2014 | 2624 (27.7%) | 8378 (25.8%) | − |
| 2015 | 2544 (26.9%) | 8791 (27.1%) | − |
| 2016 | 1468 (15.5%) | 5494 (16.9%) | − |
| 2017 | 1878 (19.8%) | 7076 (21.8%) | − |
BMD: bone mineral density; CNS: central nervous system; LEDD: levodopa equivalent daily dose; PD: Parkinson disease; PIM: potentially inappropriate medication; SD: standard deviation; SDD: standardized daily dose.
Bolded values indicate statistically significant imbalance between groups.
Up to 4:1 non-inury:injury matching without replacement within each PIM category based on age group, sex, race, claims-based frailty index category, and baseline SDD category.
3.2. Changes in SDDs over time
Table 2 shows changes and differences in mean SDDs over time in both groups. Both groups had reductions in mean SDDs of all PIM categories following discharge. There was an initial greater reduction in the mean SDD of CNS-active PIMs in the injury group at 3 months post-discharge after covariate adjustment compared to the non-injury group (DID -0.24, 95% confidence interval [CI] –0.36 to –0.11, p = 0.0002), but not thereafter. There were no statistically significant changes in mean SDDs of PD motor symptom PIMs and PD non-motor symptom PIMs at any time point. On the other hand, there was a greater reduction in mean SDDs of BMD PIMs in the injury group at all timepoints compared to the non-injury group (DID –0.19, 95% CI –0.36 to –0.03 at 3 months, p = 0.0237; DID –0.18, 95% CI –0.35 to –0.01 at 6 months, p = 0.0402; DID –0.24, 95% CI –0.43 to –0.05 at 12 months, p = 0.0156).
Table 2.
Adjusted mean and difference in SDD changes of PIM categories in injury and non-injury groups over time.
| Analyses | Medication category | Study group | Mean SDD (SD) at index | Average change in SDD from indexa (95%CI) |
||
|---|---|---|---|---|---|---|
| 3 months | 6 months | 12 months | ||||
| Primary analyses | CNS-active PIMs | Non-injury | 3.90 (4.49) | − 1.19 (− 1.25, − 1.12) | − 1.40 (− 1.46, − 1.34) | − 1.91 (− 1.97, − 1.84) |
| Injury | 3.86 (4.68) | − 1.42 (− 1.53, − 1.31) | − 1.47 (− 1.58, − 1.36) | − 1.94 (− 2.05, − 1.82) | ||
| Difference in SDD | NA | −0.24 (−0.36, −0.11) | − 0.07 (− 0.20, 0.06) | − 0.03 (− 0.16, 0.10) | ||
| p-value† | NA | 0.0002 | 0.2875 | 0.6603 | ||
| PD motor symptom PIMs | Non-injury | 5.31 (5.76) | − 1.01 (− 1.14, − 0.89) | − 1.07 (− 1.20, − 0.93) | − 1.24 (− 1.40, − 1.08) | |
| Injury | 5.07 (5.58) | − 1.21 (− 1.42, − 0.99) | − 1.06 (− 1.28, − 0.85) | − 1.25 (− 1.51, − 1.00) | ||
| Difference in SDD | NA | − 0.20 (− 0.45, 0.05) | 0.00 (− 0.25, 0.26) | − 0.01 (− 0.31, 0.29) | ||
| p-value† | NA | 0.1258 | 0.9820 | 0.9485 | ||
| PD non-motor symptom PIMs | Non-injury | 3.91 (4.63) | − 1.23 (− 1.33, − 1.12) | − 1.25 (− 1.36, − 1.14) | − 1.40 (− 1.52, − 1.27) | |
| Injury | 3.68 (4.69) | − 1.29 (− 1.48, − 1.10) | − 1.41 (− 1.60, − 1.22) | − 1.42 (− 1.64, − 1.21) | ||
| Difference in SDD | NA | − 0.06 (− 0.28, 0.16) | − 0.16 (− 0.38, 0.06) | − 0.03 (− 0.28, 0.23) | ||
| p-value† | NA | 0.5805 | 0.1538 | 0.8416 | ||
| BMD PIMs | Non-injury | 4.16 (4.71) | − 0.67 (− 0.75, − 0.59) | − 0.55 (− 0.64, − 0.47) | − 0.55 (− 0.65, − 0.46) | |
| Injury | 3.96 (4.48) | − 0.86 (− 1.01, − 0.72) | − 0.73 (− 0.88, − 0.58) | − 0.79 (− 0.96, − 0.62) | ||
| Difference in SDD | NA | −0.19 (−0.36, −0.03) | −0.18 (−0.35, −0.01) | ¡0.24 (−0.43, −0.05) | ||
| p-value† | NA | 0.0237 | 0.0402 | 0.0156 | ||
| Sensitivity analyses | CNS-active PIMs (excluding opioids) | Non-injury | 3.96 (4.41) | − 1.22 (− 1.29, − 1.16) | − 1.43 (− 1.50, − 1.37) | − 1.95 (− 2.02, − 1.88) |
| Injury | 3.92 (4.64) | − 1.44 (− 1.55, − 1.32) | − 1.47 (− 1.58, − 1.35) | − 1.96 (− 2.08, − 1.83) | ||
| Difference in SDD | NA | −0.21 (−0.34, −0.08) | − 0.03 (− 0.17, 0.10) | − 0.01 (− 0.15, 0.14) | ||
| p-value† | NA | 0.0015 | 0.6445 | 0.9431 | ||
| CNS-active PIMs (excluding antidepressants) | Non-injury | 3.53 (4.52) | − 1.35 (− 1.43, − 1.28) | − 1.55 (− 1.62, − 1.47) | − 1.99 (− 2.07, − 1.91) | |
| Injury | 3.42 (4.82) | − 1.59 (− 1.72, − 1.46) | − 1.70 (− 1.83, − 1.56) | − 2.06 (− 2.20, − 1.92) | ||
| Difference in SDD | NA | −0.24 (−0.39, −0.08) | − 0.15 (− 0.31, 0.00) | − 0.07 (− 0.23, 0.09) | ||
| p-value† | NA | 0.0026 | 0.0559 | 0.4065 | ||
BMD: bone mineral density; CI: confidence interval; CNS: central nervous system; LEDD: levodopa equivalent daily dose; NA: not applicable; PD: Parkinson disease; PIM: potentially inappropriate medication; SD: standard deviation; SDD: standardized daily dose.
p-value corresponding to test of whether average change in SDD from index differs between those with injury and matched non-injury persons. Bolded values indicate significant difference.
Estimates for average change in SDD at each time point (relative to index) are based on regression models adjusted for the following covariates (as of index): year of cohort entry, clinical encounter type at index, combined Charlson-Elixhauser comorbidity score, social deprivation index, and LEDD.
3.3. Sensitivity and secondary analyses
When excluding opioids (2712/28,771 persons excluded) and antidepressants (9687/28,771 persons excluded) from the CNS-active PIM category, we noted a greater reduction in mean SDD only at 3 months in the injury group compared to the non-injury group (DID – 0.21, 95% CI – 0.34 to – 0.08, p = 0.0015 and DID – 0.24, 95% CI – 0.39 to – 0.08, p = 0.0026, respectively). There were no statistically significant reductions in SDD between 2 groups at the other timepoints (Table 2).
With regards to initial and sustained discontinuation of an entire PIM category, 40.6% people in the injury group and 42.8% people in the non-injury group discontinued all medications in the CNS-active PIM category at 3 months, of whom 61.8% and 76.3%, respectively, sustained discontinuation through 12 months (Table 3). A similar trend was observed for the BMD PIM category (injury group: 41.3%, non-injury: 42.3% at 3 months, and 60.5% vs. 74.6% at 12 months). Surprisingly, a lower proportion of people in the injury group had sustained discontinuation of PD motor symptom PIMs (injury 57.1% versus non-injury 73.2%), and of PD non-motor symptom PIMs (68.3% versus 77.4%) at 12 months (Table 3).
Table 3.
Proportions of people who discontinued PIM categories following discharge in injury and non-injury groups.
| Medication category | Study group | Proportion discontinuing entire PIM category at 3 months | Proportions sustaining discontinuation through 12 months |
|---|---|---|---|
| CNS-active PIMs | Non-injury | 9499/22,185 (42.8%) | 7250/9499 (76.3%) |
| Injury | 2675/6586 (40.6%) | 1653/2675 (61.8%) | |
| PD motor symptom PIMs | Non-injury | 3459/7934 (43.6%) | 2532/3459 (73.2%) |
| Injury | 1121/2638 (42.5%) | 640/1121 (57.1%) | |
| PD non-motor symptom PIMs | Non-injury | 5052/9772 (51.7%) | 3908/5052 (77.4%) |
| Injury | 1521/3033 (50.1%) | 1039/1521 (68.3%) | |
| BMD PIMs | Non-injury | 8224/19,442 (42.3%) | 6138/8224 (74.6%) |
| Injury | 2326/5627 (41.3%) | 1408/2326 (60.5%) |
BMD: bone mineral density; CI: confidence interval; CNS: central nervous system; PD: Parkinson disease; PIM: potentially inappropriate medication
In another secondary analysis to detect possible tapering of PIMs among those who stayed on a specific PIM category through one year (Table 4), we found a small reduction in the mean SDDs of PIM categories in both groups at each time point, although this difference was only statistically significant for the BMD PIM category, sporadically, at 3 and 12 months (p = 0.0050 and p = 0.0482, respectively).
Table 4.
Adjusted changes in medication use over time among those who stayed on PIM category of interest through one year.
| Medication category | Study group | Mean SDD (SD) at index | Average change in SDD from indexa (95% CI) |
||
|---|---|---|---|---|---|
| 3 months | 6 months | 12 months | |||
| CNS-active PIMs | Non-injury | 4.62 (5.05) | − 0.20 (− 0.30, − 0.10) | − 0.14 (− 0.25, − 0.04) | − 0.20 (− 0.31, − 0.10) |
| Injury | 4.39 (5.46) | − 0.35 (− 0.53, − 0.17) | − 0.34 (− 0.53, − 0.16) | − 0.37 (− 0.55, − 0.18) | |
| p-value† | NA | 0.1365 | 0.0696 | 0.1302 | |
| PD motor PIMs | Non-injury | 6.01 (6.04) | − 0.10 (− 0.28, 0.07) | − 0.06 (− 0.23, 0.12) | − 0.06 (− 0.26, 0.14) |
| Injury | 5.72 (6.05) | − 0.32 (− 0.62, − 0.02) | − 0.27 (− 0.57, 0.02) | − 0.13 (− 0.47, 0.21) | |
| p-value† | NA | 0.2140 | 0.2102 | 0.7401 | |
| PD non-motor PIMs | Non-injury | 4.56 (5.47) | − 0.22 (− 0.41, − 0.03) | − 0.17 (− 0.37, 0.03) | − 0.37 (− 0.59, − 0.16) |
| Injury | 4.08 (6.21) | − 0.22 (− 0.55, 0.11) | − 0.47 (− 0.83, − 0.12) | − 0.27 (− 0.64, 0.11) | |
| p-value† | NA | 0.9919 | 0.1435 | 0.6350 | |
| BMD PIMs | Non-injury | 4.57 (4.88) | − 0.08 (− 0.19, 0.03) | − 0.01 (− 0.13, 0.10) | − 0.02 (− 0.14, 0.10) |
| Injury | 4.49 (5.44) | − 0.41 (− 0.62, − 0.21) | − 0.13 (− 0.35, 0.09) | − 0.27 (− 0.50, − 0.05) | |
| p-value† | NA | 0.0050 | 0.3547 | 0.0482 | |
BMD: bone mineral density; CI: confidence interval; CNS: central nervous system; LEDD: levodopa equivalent daily dose; NA: not applicable; PD: Parkinson disease; PIM: potentially inappropriate medication; SD: standard deviation; SDD: standardized daily dose.
p-value corresponding to test of whether average change in SDD from index differs between those with injury and matched non-injury persons. Bolded values indicate significant difference.
Estimates for average change in SDD at each time point (relative to index) are based on regression models adjusted for the following covariates (as of index): year of cohort entry, clinical encounter type at index, combined Charlson-Elixhauser comorbidity score, social deprivation index, and LEDD.
4. Discussion
Up to 70% of the PD population falls annually, and two-thirds of these report recurrent falls [21]. In additional to being very frequent, falls are a driver of disability, health care utilization, and through serious injury such as hip fracture, mortality [22,23]. Recently, the World Falls Guidelines (WFG) Task Force developed updated World Guidelines for Falls Prevention and Management for Older Adults, which includes medication review, assessment for polypharmacy, screening and identification of PIMs, and deprescribing of fall-risk-increasing drugs as part of a multi-domain falls prevention intervention [5]. Previous studies have suggested that the majority of individuals with PD have significant polypharmacy and PIM use, particularly PIMs that may increase the risk of falls with serious injury [24–26]. However, our study of community-dwelling PD patients found no meaningful differences PIM reduction among individuals hospitalized for serious injury versus other reasons, except for the CNS-active PIM category at 3 months post-discharge and for the BMD PIM category throughout follow-up. In the context of the recent WFG guidelines, our findings highlight a missed opportunity for a complete medication review and deprescribing to prevent recurrent serious injuries in persons with PD.
It is not very surprising we found no differences (except at 3 months post-discharge) in prescribing when measured using the broadest PIM category, CNS-active PIMs, given that this category contains many of the most widely prescribed PIMs (e.g., paroxetine, meclizine, zolpidem). However, prescribing in our more selective PIM categories (the PD motor symptom and PD non-motor symptom PIM categories), defined based on the potential for drug-disease interactions in persons with dopaminergic and cholinergic neuronal dysfunction, also did not meaningfully change. Possible explanations for this result are that discontinuation of disease-specific medications were left to neurologists, who are currently not routinely involved in post-hospitalization care, or, that there is under-recognition of the potential drug-disease interactions that may occur in PD. In contrast, the statistically significant reductions in BMD PIM category in the injury group may signal a deprescribing in response to growing general knowledge about fracture risk and routine screening for bone mineral density; a spurious association cannot be ruled out.
Despite there being no evidence of forethought, deprescribing (or medication discontinuation) appeared to be common. Over 40% of persons in our sample had discontinued PIMs across categories of interest three months post-discharge, and up to 3/4 of these individuals sustained discontinuation through one year. Prior pharmacoepidemiology studies have found similar prescribing trends after hospitalization for multiple conditions (e.g. fracture, heart failure, myocardial infarction) the Medicare population [6,27–30], perhaps reflecting high base-line variability in medication use, or the presence of common factors driving medication discontinuation after a hospitalization (such as re-admission, rehabilitation) in this population, which also prevented us from being able to detect between group differences. We also observed possible tapering of medications, through SDD reductions in specific PIM categories; however, whether the tapering is clinically significant remains to be investigated. For example, an average change of 1.5 SDD for the PIM category of PD non-motor symptoms could translate into clinically meaningful changes in the doses of aripiprazole (minimum effective geriatric daily dose of 10 mg; reduction in total daily dose from 15 mg to 0 mg), zolpidem (minimum effective geriatric daily dose of 5 mg; reduction in total daily dose from 10 mg to 2.5 mg), and oral oxybutynin (minimum effective geriatric daily dose of 5 mg; reduction in total daily dose from 10 mg to 2.5 mg) but not for lorazepam (minimum effective geriatric daily dose of 0.5 mg; reduction in total daily dose from 2 mg to 1.25 mg) or diazepam (minimum effective geriatric daily dose of 1 mg; reduction in total daily dose from 20 mg to 18.5 mg). Future studies that focus on specific PIMs are needed to further guide the interpretation and acting on this data.
Our use of data from a large, population representative U.S. insurance program allowed for the study of a cohort of older adults with PD with diversity that is not found in academic center patient populations, as over 93% of adults aged 65 and older were covered through the Medicare program in 2021 [31]. Having claims data for each individual that spanned health care settings and crossed time allowed us to employ contemporary pharmacoepidemiology methods to measure medication use and trends before and after a sentinel event. Despite these strengths, our study had several limitations. Our PIM list, even though guided by current published clinical evidence, is not exhaustive and may miss medications with emerging relevance. Our definition of community-dwelling, although previously used, can result in medication measurements that may be impacted by re-admission or other acute care use, delayed refilling of prescriptions. Although prescriptions are well captured in administrative claims data, we are unable to determine whether the medications were actually ingested, although this limitation may have more relevance for comparative safety (as opposed to comparative deprescribing) studies. Lastly, our database ended in 2017, thus our findings might not reflect the increase in awareness and promotion of deprescribing practices in more recent years.
5. Conclusions
In conclusion, we found clinically negligible differences in PIM use among persons with PD hospitalized for serious injury versus other diagnoses. Future studies will examine the extent to which PIM is a modifiable risk factor for recurrent falls, injury, hospitalization, or death, and may identify medication review and deprescribing as a means of improving the trajectory of disease-related disability for PD.
Supplementary Material
Funding sources
This study is funded by grants from the National Institute of Neurological Diseases and Stroke [#R01NS099129] and the National Institute on Aging [#K24AG075234] of the National Institutes of Health.
Footnotes
Ethical compliance statement
This study was approved with a waiver of consent by the University of Pennsylvania Human Research Protections Office and CMS. We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this work is consistent with those guidelines.
Declaration of competing interest
All other authors declared no competing interests for this work.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.org/10.1016/j.parkreldis.2023.105793.
References
- [1].Aarsland D, Batzu L, Halliday GM, Geurtsen GJ, Ballard C, Ray Chaudhuri K, Weintraub D, Parkinson disease-associated cognitive impairment, Nat. Rev. Dis. Prim 7 (1) (2021) 47. [DOI] [PubMed] [Google Scholar]
- [2].McLean G, Hindle JV, Guthrie B, Mercer SW, Co-morbidity and polypharmacy in Parkinson’s disease: insights from a large Scottish primary care database, BMC Neurol 17 (1) (2017) 126. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [3].Leoni O, Martignoni E, Cosentino M, Michielotto D, Calandrella D, Zangaglia R, Riboldazzi G, Oria C, Lecchini S, Nappi G, Frigo G, Drug prescribing patterns in Parkinson’s disease: a pharmacoepidemiological survey in a cohort of ambulatory patients, Pharmacoepidemiol. Drug Saf 11 (2) (2002) 149–157. [DOI] [PubMed] [Google Scholar]
- [4].Fasano A, Canning CG, Hausdorff JM, Lord S, Rochester L, Falls in Parkinson’s disease: a complex and evolving picture, Mov. Disord 32 (11) (2017) 1524–1536. [DOI] [PubMed] [Google Scholar]
- [5].Montero-Odasso M, van der Velde N, Martin FC, Petrovic M, Tan MP, Ryg J, Aguilar-Navarro S, Alexander NB, Becker C, Blain H, Bourke R, Cameron ID, Camicioli R, Clemson L, Close J, Delbaere K, Duan L, Duque G, Dyer SM, Freiberger E, Ganz DA, Gomez F, Hausdorff JM, Hogan DB, Hunter SMW, Jauregui JR, Kamkar N, Kenny RA, Lamb SE, Latham NK, Lipsitz LA, Liu-Ambrose T, Logan P, Lord SR, Mallet L, Marsh D, Milisen K, Moctezuma-Gallegos R, Morris ME, Nieuwboer A, Perracini MR, Pieruccini-Faria F, Pighills A, Said C, Sejdic E, Sherrington C, Skelton DA, Dsouza S, Speechley M, Stark S, Todd C, Troen BR, van der Cammen T, Verghese J, Vlaeyen E, Watt JA, Masud T, Task A, Force on Global Guidelines for Falls in Older, World guidelines for falls prevention and management for older adults: a global initiative, Age Ageing 51 (9) (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
- [6].Munson JC, Bynum JP, Bell JE, Cantu R, McDonough C, Wang Q, Tosteson TD, Tosteson AN, Patterns of prescription drug use before and after fragility fracture, JAMA Intern. Med 176 (10) (2016) 1531–1538. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [7].Hart LA, Walker R, Phelan EA, Marcum ZA, Schwartz NRM, Crane PK, Larson EB, Gray SL, Change in central nervous system-active medication use following fall-related injury in older adults, J. Am. Geriatr. Soc 70 (1) (2022) 168–177. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [8].Butt DA, Tu K, Young J, Green D, Wang M, Ivers N, Jaakkimainen L, Lam R, Guttman M, A validation study of administrative data algorithms to identify patients with Parkinsonism with prevalence and incidence trends, Neuroepidemiology 43 (1) (2014) 28–37. [DOI] [PubMed] [Google Scholar]
- [9].Dawwas GK, Hennessy S, Brensinger CM, Acton EK, Bilker WB, Chung S, Dublin S, Horn JR, Manis MM, Miano TA, Oslin DW, Pham Nguyen TP, Soprano SE, Wiebe DJ, Leonard CE, Signals of muscle relaxant drug interactions associated with unintentional traumatic injury: a population-based screening study, CNS Drugs 36 (4) (2022) 389–400. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [10].The 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 67 (4) (2019) 674–694. [DOI] [PubMed] [Google Scholar]
- [11].Seppala LJ, Petrovic M, Ryg J, Bahat G, Topinkova E, Szczerbinska K, van der Cammen TJM, Hartikainen S, Ilhan B, Landi F, Morrissey Y, Mair A, Gutierrez-Valencia M, Emmelot-Vonk MH, Mora MAC, Denkinger M, Crome P, Jackson SHD, Correa-Perez A, Knol W, Soulis G, Gudmundsson A, Ziere G, Wehling M, O’Mahony D, Cherubini A, van der Velde N, STOPPFall (screening tool of older persons prescriptions in older adults with high fall risk): a delphi study by the EuGMS Task and finish group on fall-risk-increasing drugs, Age Ageing 50 (4) (2021) 1189–1199. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [12].Pazan F, Gercke Y, Weiss C, Wehling M, Raters F, The U.S.-FORTA (fit fOR the aged) list: consensus validation of a clinical tool to improve drug therapy in older adults, J. Am. Med. Dir. Assoc 21 (3) (2020) 439 e9–e439 e13. [DOI] [PubMed] [Google Scholar]
- [13].Seppi K, Ray Chaudhuri K, Coelho M, Fox SH, Katzenschlager R, Perez Lloret S, Weintraub D, Sampaio C, C. The collaborators of the Parkinson’s disease update on non-motor symptoms study group on behalf of the movement disorders society evidence-based medicine, update on treatments for nonmotor symptoms of Parkinson’s disease-an evidence-based medicine review, Mov. Disord 34 (2) (2019) 180–198. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [14].Woolcott JC, Richardson KJ, Wiens MO, Patel B, Marin J, Khan KM, Marra CA, Meta-analysis of the impact of 9 medication classes on falls in elderly persons, Arch. Intern. Med 169 (21) (2009) 1952–1960. [DOI] [PubMed] [Google Scholar]
- [15].Hanlon JT, Boudreau RM, Roumani YF, Newman AB, Ruby CM, Wright RM, Hilmer SN, Shorr RI, Bauer DC, Simonsick EM, Studenski SA, Number and dosage of central nervous system medications on recurrent falls in community elders: the Health, Aging and Body Composition study, J Gerontol A Biol Sci Med Sci 64 (4) (2009) 492–498. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [16].Butler DC, Petterson S, Phillips RL, Bazemore AW, Measures of social deprivation that predict health care access and need within a rational area of primary care service delivery, Health Serv. Res 48 (2 Pt 1) (2013) 539–559. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [17].Gagne JJ, Glynn RJ, Avorn J, Levin R, Schneeweiss S, A combined comorbidity score predicted mortality in elderly patients better than existing scores, J. Clin. Epidemiol 64 (7) (2011) 749–759. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [18].Kim DH, Schneeweiss S, Glynn RJ, Lipsitz LA, Rockwood K, Avorn J, Measuring frailty in Medicare data: development and validation of a claims-based frailty index, J Gerontol A Biol Sci Med Sci 73 (7) (2018) 980–987. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [19].Tomlinson CL, Stowe R, Patel S, Rick C, Gray R, Clarke CE, Systematic review of levodopa dose equivalency reporting in Parkinson’s disease, Mov. Disord 25 (15) (2010) 2649–2653. [DOI] [PubMed] [Google Scholar]
- [20].Schade S, Mollenhauer B, Trenkwalder C, Levodopa equivalent dose conversion factors: an updated proposal including opicapone and safinamide, Mov Disord Clin Pract 7 (3) (2020) 343–345. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [21].Canning CG, Paul SS, Nieuwboer A, Prevention of falls in Parkinson’s disease: a review of fall risk factors and the role of physical interventions, Neurodegener. Dis. Manag 4 (3) (2014) 203–221. [DOI] [PubMed] [Google Scholar]
- [22].Kerr GK, Worringham CJ, Cole MH, Lacherez PF, Wood JM, Silburn P, Predictors of future falls in Parkinson disease, Neurology 75 (2) (2010) 116–124. [DOI] [PubMed] [Google Scholar]
- [23].Matinolli M, Korpelainen J, Sotaniemi K, Myllylä V, Korpelainen R, Recurrent falls and mortality in Parkinson’s disease: a prospective two-year follow-up study, Acta Neurol. Scand 123 (3) (2011) 193–200. [DOI] [PubMed] [Google Scholar]
- [24].Csoti I, Herbst H, Urban P, Woitalla D, Wüllner U, Polypharmacy in Parkinson’s disease: risks and benefits with little evidence, J. Neural. Transm 126 (2019) 871–878. [DOI] [PubMed] [Google Scholar]
- [25].Crispo JA, Willis AW, Thibault DP, Fortin Y, Hays HD, McNair DS, Bjerre LM, Kohen DE, Perez-Lloret S, Mattison DR, Associations between anticholinergic burden and adverse health outcomes in Parkinson disease, PLoS One 11 (3) (2016), e0150621. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [26].Mantri S, Fullard M, Gray SL, Weintraub D, Hubbard RA, Hennessy S, Willis AW, Patterns of dementia treatment and frank prescribing errors in older adults with Parkinson disease, JAMA Neurol 76 (1) (2019) 41–49. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [27].Lund BC, Schroeder MC, Middendorff G, Brooks JM, Effect of hospitalization on inappropriate prescribing in elderly Medicare beneficiaries, J. Am. Geriatr. Soc 63 (4) (2015) 699–707. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [28].Booth JN III, Colantonio LD, Chen L, Rosenson RS, Monda KL, Safford MM, Kilgore ML, Brown TM, Taylor B, Dent R, Statin discontinuation, reinitiation, and persistence patterns among Medicare beneficiaries after myocardial infarction: a cohort study, Circulation: Cardiovascular Quality and Outcomes 10 (10) (2017), e003626. [DOI] [PubMed] [Google Scholar]
- [29].Yun H, Curtis JR, Guo L, Kilgore M, Muntner P, Saag K, Matthews R, Morrisey M, Wright NC, Becker DJ, Patterns and predictors of osteoporosis medication discontinuation and switching among Medicare beneficiaries, BMC Muscoskel. Disord 15 (1) (2014) 1–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [30].Sueta CA, Rodgers JE, Chang PP, Zhou L, Thudium EM, Kucharska-Newton AM, Stearns SC, Medication adherence based on part D claims for patients with heart failure after hospitalization (from the Atherosclerosis Risk in Communities Study), Am. J. Cardiol 116 (3) (2015) 413–419. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [31].Keisler-Starkey KB, L.N, Health Insurance Coverage in the United States: 2021, Current Population Reports, U.S. Census Bureau, Washington, DC, 2022. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
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.