FormalPara Key Summary Points

Patients with advanced Parkinson’s disease require increased dopamine therapy (carbidopa–levodopa preparations) as well as other oral adjunctive therapies to control motor fluctuations.

Increased oral medications with their complex polypharmacy regimens often result in poor treatment adherence and discontinuation.

Patients uncontrolled on oral medications may benefit from device-aided therapies of carbidopa/levodopa enteral suspension (known as levodopa/carbidopa intestinal gel outside of the USA) or deep brain stimulation to reduce pill burden.

This real-world study of US Medicare beneficiaries (≥ 65 years) between 2014 and 2018 demonstrated significant and sustained reduction of average daily number of oral Parkinson’s disease medications in the 12 months after patients were initiated on endoscopic delivery of carbidopa/levodopa enteral suspension compared with patients initiated on deep brain stimulation.

Pill burden reduction could be considered a treatment goal to minimize disease burden on elderly patients with advanced Parkinson’s disease.

Digital Features

This article is published with digital features, including an infographic and a graphical abstract, to facilitate understanding of the article. To view digital features for this article, go to https://doi.org/10.6084/m9.figshare.21695339.

Introduction

Parkinson’s disease (PD) is a progressive neurodegenerative movement disorder affecting nearly one million people in the USA, which is estimated to exceed over 1.23 million people by 2030 [1]. Treatment to control motor symptoms relies on dopamine therapy with oral levodopa preparations administered as carbidopa/levodopa, as well as other oral adjunctive therapies [2,3,4]. As the disease progresses, even optimized carbidopa/levodopa treatment can become less effective due to erratic gastric emptying, narrowing therapeutic windows, and food–drug or drug–drug interactions [4,5,6]. Patients may experience intermittent symptom control, requiring multiple drug administrations per day [4, 7]. Higher doses of levodopa to control symptoms are associated with dyskinesias [8, 9]. Moreover, worsening of sensory, neuropsychiatric, and autonomic non-motor symptoms further contribute to disease and medication burden [4, 10]. Persistent fluctuations of off periods (e.g., poor mobility, slowness, stiffness) and on periods (e.g., good motor-system control with or without dyskinesia) require higher and more frequent dosing of oral levodopa to minimize the significant challenges to living with advanced Parkinson’s disease (APD) [8].

The Food and Drug Administration (FDA) Voice of the Patient captured these challenges for patients living with APD and relying on caretakers [11]. Some patients took 14 pills throughout the day and 15 pills at night, and described the planning involved in balancing food intake, medication use, and daily needs as tedious. Patients with APD emphasized the constant struggle with schedule and timing of medications, and frustrations with off periods and unpredictable exacerbation of symptoms [12]. Many patients shared consensus for PD treatments that would minimize dosages and administration burdens.

As PD progresses, most patients receive increasingly higher dosages of oral PD medications, and more complex multi-drug regimens that can result in suboptimal medication adherence [13]. Poor adherence and high discontinuation are particularly seen with levodopa equivalent daily dose (LEDD) regimens exceeding 1000 mg [14]. Patients uncontrolled on oral medications may benefit from device-aided therapies such as carbidopa/levodopa enteral suspension (CLES), also known as levodopa/carbidopa intestinal gel (LCIG) outside of the USA, or deep brain stimulation (DBS). Treatment with DBS of subthalamic nucleus (STN) and globus pallidus interna (GPi) has shown efficacy in managing symptoms of APD while reducing pill burden [4, 15,16,17], although smaller reductions have been reported for those undergoing GPi [18, 19]. Pill burden describes the number of tablets, capsules, or other dosage forms taken on a regular basis (https://clinicalinfo.hiv.gov/en/glossary/pill-burden#:~:text). CLES (Duopa, AbbVie, Inc, North Chicago, IL) [20] continuously delivers levodopa directly to the jejunum via a percutaneous endoscopic gastrojejunostomy tube (PEG-tube) and portable pump, bypassing the stomach and thus avoiding erratic gastric emptying and improving drug absorption. A clinical trial demonstrated reduction in pill burden, and significant improvements in motor and non-motor symptoms and quality of life (QoL), even when used as monotherapy [8, 17, 21]. Both therapies have shown to decrease off periods and increase on periods without dyskinesia.

There are limited real-world studies directly assessing and comparing the impact of CLES versus DBS on oral PD-related pill burden and resultant LEDD. A prior study of commercial and supplemental Medicare insured patients showed benefit of these device-aided therapies to reduce oral pill burden [22]. We extend these findings to include assessment of LEDD and focus this work on the 100% Medicare fee-for-service (FFS) population, as PD predominately affects older adults. Our real-world study of Medicare beneficiaries provides a comprehensive evaluation of the comparative efficacies of CLES and DBS in the US APD population. We evaluated the effectiveness within and between these device-aided therapies to reduce oral pill burden and LEDD.

Methods

Study Design and Data Source

This retrospective cohort study used the 100% Medicare FFS claims database [23] from 2014 to 2018 linked to data from the CLES Patient Support Program (PSP) [20]. The 100% Medicare FFS claims was accessed through a research-focused data use agreement with the Centers for Medicare & Medicaid Services (CMS) and did not require Institutional Review Board approval due to use of retrospective de-identified patient data. Medicare is a national health insurance plan in the USA administered through the CMS [24] covering people aged ≥ 65 years, and younger individuals with disabilities and/or end-stage renal disease (ESRD) [25]. Medicare parts A and B cover medical events and part D covers prescription drug events. Medicare enrollment files provided patient demographic data. The CLES PSP collects data on patients enrolled and initiated (CLES-I) or not initiated (CLES-NI) on CLES treatment. Through a secure process, study participants from the CLES PSP were merged with Medicare claims using patient identifying information (PII) including first and last names, date of birth, gender, and zip code. PII was then removed through a secure de-identification process consistent with US federal regulations to ensure patient privacy [26].

Identification of Parkinson’s Disease and Treatment Cohorts

The study period was 1 July 2014–31 December 2018. Index dates were identified between 1 January 2015 and 31 December 2017 to allow for 6-month baseline and 12-month follow-up to permit longitudinal evaluation of drug treatment patterns. Patients were required to have a diagnosis of PD during the baseline period, defined as two or more claims with International Classification of Diseases (ICD) 9/10 Clinical Modification (CM) codes (ICD-9: 332.0x; ICD-10: G20). For CLES-I who had placement of the PEG-tube (identified through CLES PSP), index date was defined as first shipment date of CLES. CLES-I were required to have continuous CLES shipments during the 12-month follow-up and no claims for DBS. For CLES-NI without PEG-tube placement, the index date was defined as the first program enrollment date and no claims for DBS during the study period. For DBS, index date was defined as the first observed date of DBS surgery during the identification period, using the Current Procedural Terminology, fourth Edition (CPT-4) codes (61863, 61864, 61867, 61868, 0293).

Patient Demographic and Clinical Characteristics

Demographics included age, gender, race, US Census geographic region, reason for Medicare entitlement (age 65 years or disability/ESRD), and dual eligibility for Medicare and Medicaid. Medicaid is a state administered health insurance program that provides insurance to low-income adults and children. Clinical characteristics assessing the overall health status of PD patients included the Charlson Comorbidity Index (CCI) score [27] and other prevalent concomitant chronic conditions (i.e., hypertension, hyperlipidemia, rheumatoid arthritis, depression). Mobility was determined by the need for assistive ambulatory devices identified with the Healthcare Common Procedure Coding System.

Identification of Oral Parkinson’s Disease Medications

To determine PD-related pill burden, commonly prescribed medications to treat PD inclusive of levodopa preparations, dopamine agonists, monoamine oxidase (MAO-B) inhibitors, catechol-O methyl transferase (COMT) inhibitors, amantadine, anticholinergics, and other adjuvant medications were captured for each cohort during the baseline and follow-up periods using National Drug Codes. Pill burden was reported as the average number of daily oral PD pills across two measurement periods: monthly and 6-month intervals at baseline and follow-up. The pill burden for each measurement period was calculated as the cumulative quantity of oral PD pills dispensed during that period divided by number of days in the period. Prescription dates, days supplied, and quantity metrics were accounted for to capture the spillover of oral PD medications prescribed for overlap** months.

Levodopa Equivalent Daily Doses

The average LEDD of only oral dopaminergic medications (e.g., carbidopa/levodopa, amantadine) were evaluated for the cohorts at baseline and at 0–6 months and 7–12 months follow-up. LEDD was defined as the milligrams required to produce the same level of symptom control as 100 mg of immediate release levodopa/carbidopa. The daily dosage of levodopa for each drug formulation was multiplied by a standardized conversion factor for each oral medication [9]. The average LEDD excluded CLES and accounted for spillover of oral PD medications prescribed for overlap** months.

Statistical Analysis

Descriptive statistics reported continuous variables as mean and standard deviation (SD) and categorical variables as counts and percentages. Bivariate analysis assessed differences in baseline patient demographics, clinical characteristics, and comorbidities by Student’s t-test for continuous variables and Chi-square test for categorical variables. Statistical significance was defined as p < 0.05.

CLES-I were matched 1:1 with CLES-NI and 1:3 with DBS patients using propensity score matching (PSM). Balancing of the PSM controlled for differences in patient demographic and clinical characteristics across the study cohorts (CLES-I versus CLES-NI and CLES-I versus DBS). Logistic regression was used to calculate the propensity score, which was defined as the probability of initiating CLES treatment (CLES-I). Covariates utilized in the PSM included age, gender, census region, race, dual eligibility status, and CCI. Greedy matching techniques used a caliper of width equal to 0.2 of the pooled standard deviation of the logit of the propensity score [28]. To ensure balance of covariates between the cohorts, baseline patient characteristics were evaluated pre- and post-PSM.

All outcomes were assessed based on the matched cohorts. Treatment efficacy was defined as the reduction in average daily pill burden. The number of prescribed pills was reported within and between cohorts at baseline and at 6- and 12-months follow-up. Total LEDD was similarly assessed. Unadjusted analysis of the difference in pill burden between the matched cohorts was evaluated by Mann–Whitney U tests. Analysis used generalized linear regression models, adjusting for age, gender, census region, dual eligibility status, and CCI to estimate the percentage differences in pill burden reductions between CLES-I versus CLES-NI and CLES-I versus DBS. All analyses performed used SAS 7.1 (SAS Institute; Cary, North Carolina).

Results

A total of 10,752 PD patients met the inclusion criteria (Table 1). Of those, 2.4% (262) patients were CLES-I, 2.2% (233) were CLES-NI, and 95.4% (10,257) received DBS (Table 2). At baseline, mean age was similar between CLES-I and CLES-NI (71.3 versus 72.0 years; p = 0.3162). Significantly fewer CLES-I were dual eligible for both Medicare and Medicaid compared with CLES-NI. No differences were noted in gender (> 50% male) or race (> 90% white), CCI scores, or use of ambulatory assistive devices. A similar pattern of patient characteristics was also observed between CLES-I and DBS patients. CLES-I and DBS patients were similar in age (71.3 versus 70.5 years; p = 0.0253, which is statistically significant, but not clinically meaningful) but had significantly lower CCI scores (1.3 versus 2.6; p < 0.0001) and greater use of ambulatory assistive devices (9.5% versus 6.2%; p = 0.0289). After PSM, baseline characteristics were predominately similar for CLES-I versus CLES-NI, while significant differences in census region, dual eligibility status, and CCI were noted between CLES-I versus DBS (Table 3). The sample size for CLES-I versus CLES-NI (1:1 match) was N = 188, and for CLES-I versus DBS (1:3 match) was N = 204 and N = 612, respectively.

Table 1 Attrition table for carbidopa/levodopa enteral suspension initiators, carbidopa/levodopa enteral suspension non-initiators, and deep brain stimulation
Full size table
Table 2 Baseline patient characteristics of carbidopa/levodopa enteral suspension initiators versus carbidopa/levodopa enteral suspension non-initiators and carbidopa/levodopa enteral suspension initiators versus deep brain stimulation prior to matching
Full size table
Table 3 Baseline patient characteristics of carbidopa/levodopa enteral suspension initiators versus carbidopa/levodopa enteral suspension non-initiators and carbidopa/levodopa enteral suspension initiators versus deep brain stimulation for propensity-matched cohorts
Full size table

Pill Burden

CLES-I versus CLES-NI: During the 6-month baseline, the mean number of daily pills prescribed per month remained relatively stable for CLES-I and CLES-NI, with a higher pill burden observed in CLES-I (Fig. 1A). One month prior to index, CLES-I had a greater pill burden versus CLES-NI (11.9 versus 10.6; p = 0.0474). However, by 2-months post-index, a significant decline in pill burden was seen for CLES-I versus CLES-NI (7.6 versus 11.1; p < 0.0001). By 6 months, CLES-I required significantly fewer pills than CLES-NI (4.7 versus 11.4; p < 0.0001), with further pill reduction at 12 months (3.5 versus 11.1; p < 0.0001). For CLES-I at 0–6 months and 7–12 months, there was a −46.3% (Δ −5.5 pills/day) and −68.3% (Δ −8.2 pills/day) reduction in average pill burden compared with baseline, respectively (p < 0.0001) (Fig. 2A). Compared with CLES-NI, CLES-I had a reduction of 6.6 pills at 0–6 months and 9.0 pills at 7–12 months (Table 4A). For CLES-NI, a significant increase in average pill burden of 10.5% (Δ +1.1; p < 0.05) and insignificant increase of 8.2% (Δ +0.90; p > 0.05) was observed at respective follow-up periods compared with baseline. Adjusted analysis of average daily pill burden estimated that CLES-I take 57.3% fewer pills at 0–6 months and 74.1% fewer pills at 7–12 months compared with CLES-NI (p < 0.0001) (Fig. 3).

Fig. 1
figure 1

Unadjusted 30-day mean pill burden for matched cohorts at baseline and follow-up. The line graph depicts the average number of daily oral PD-related pills per month for matched cohorts. A Carbidopa/levodopa enteral suspension initiators versus carbidopa/levodopa enteral suspension non-initiators and B carbidopa/levodopa enteral suspension initiators versus deep brain stimulation at baseline and through 12-month follow-up

Full size image
Fig. 2
figure 2

Unadjusted mean change in pill burden for matched cohorts from baseline to follow-up. The bar graph depicts the short-term (0–6 months) and long-term (7–12 months) unadjusted reduction in pill burden for matched cohorts. A Carbidopa/levodopa enteral suspension initiators versus carbidopa/levodopa enteral suspension non-initiators and B carbidopa/levodopa enteral suspension initiators versus deep brain stimulation at baseline through 0–6-months and 7–12 months follow-up

Full size image
Table 4 Unadjusted mean difference in reduction in pills per day among matched cohorts, A Carbidopa/levodopa enteral suspension initiators versus carbidopa/levodopa enteral suspension non-initiators and B carbidopa/levodopa enteral suspension initiators versus deep brain stimulation
Full size table
Fig. 3
figure 3

Adjusted mean pill burden reduction for matched cohorts at follow-up. The bar graph depicts the comparison of adjusted reduction in pill burden for matched cohorts: carbidopa/levodopa enteral suspension initiators versus carbidopa/levodopa enteral suspension non-initiators and carbidopa/levodopa enteral suspension initiators versus deep brain stimulation at baseline through 0–6-months and 7–12 months follow-up

Full size image

CLES-I versus DBS: During the 6-month baseline period, the mean number of daily pills per month remained relatively stable for CLES-I and DBS patients, with a significantly higher pill burden in CLES-I (12.3 versus 7.7; p < 0.0001) (Fig. 1B). By 3 months post-index, CLES-I required significantly fewer pills than DBS (5.9 versus 7.3; p < 0.0001), with further pill reductions at 6 months (4.8 versus 7.4; p < 0.0001) and 12 months (3.6 versus 7.0; p < 0.0001). For CLES-I, these findings represented a −46.5% and −67.7% significant difference in reduction (p < 0.0001) (Fig. 2B) in average pill burden at 0–6 months and 7–12 months follow-up compared with baseline, or a difference of −5.4 and −7.8 pills compared with DBS, respectively (Table 4B). For DBS, an insignificant decrease in average pill burden of 3.9% (Δ −0.3) and 6.1% (Δ −0.5) was observed at the respective follow-up periods. Adjusted analysis of average daily pill burden estimated that CLES-I take 49.6% fewer pills at 0–6 months and 70.1% fewer pills at 7–12 months compared with DBS (p < 0.0001) (Fig. 3).

Levodopa Equivalent Daily Doses Levels

CLES-I versus CLES-NI: At baseline, the mean (± SD) oral LEDD (mg) was higher in CLES-I 935 (± 697) than in CLES-NI 795 (± 579) (Fig. 4A). For CLES-I, the post-index oral LEDD declined to 531 (± 513) at 0–6 months and to 237 (± 367) at 7–12 months follow-up. In contrast, the post-index oral LEDD for CLES-NI increased to 1140 (± 762) at 0–6 months and 1112 (± 707) at 7–12 months. Thus, oral LEDD was 2.1× higher among CLES-NI patients at 0–6 months and 4.7× higher at 7–12 months compared with CLES-I.

Fig. 4
figure 4

Unadjusted mean levodopa equivalent daily doses (LEDD) for matched cohorts at baseline and follow-up. The bar graph depicts the unadjusted mean LEDD per month for matched cohorts. A Carbidopa/levodopa enteral suspension initiators versus carbidopa/levodopa enteral suspension non-initiators and B carbidopa/levodopa enteral suspension initiators versus deep brain stimulation at baseline and 0–6 months and 7–12 months follow-up

Full size image

CLES-I versus DBS: At baseline, the mean (± SD) oral LEDD (mg) was higher in CLES-I 944 (± 675) than in DBS 511 (± 512) (Fig. 4B). For CLES-I, the post-index oral LEDD declined to 536 (± 473) at 0–6 months follow-up and 236 (± 359) at 7–12 months. In contrast, the post-index oral LEDD for DBS increased slightly with respective mean levels of 617 (± 581) and 594 (± 583). Thus, oral LEDD was 2.3× higher among DBS patients at 0–6 months and 2.5× higher at 7–12 months compared with CLES-I.

Discussion

In comparison to CLES-NI and DBS, CLES-I was associated with a significantly greater and sustained reduction in the average daily number of oral PD medications with a concomitant reduction in LEDD. No significant decline in pill reduction was observed in the DBS cohort, and LEDD remained relatively stable at baseline through follow-up. There was a significant decline in pill burden among CLES-I compared with CLES-NI, a comparator control with similar patient demographic and clinical characteristics and similar disease progression based on eligibility for CLES.

The significant and sustained reduction of pill burden for CLES-I likely reflects the ability to deliver a continuous infusion of levodopa directly to the jejunum for more rapid and consistent absorption to the bloodstream versus the short half-life of intermittent pulsatile delivery of orally administered levodopa [29, 30], which is the gold standard for PD treatment. CLES provides the same 1:4 ratio of carbidopa to levodopa as the oral preparation [27]. Our study limited LEDD assessment to only oral levodopa and dopamine agonists to better assess the reduction in pill burden for device-aided therapies, since LEDD may also increase soon after initiating CLES [17] Therefore, the findings may contrast with other real-world studies that show an increase in LEDD due to the inclusion of levodopa doses in CLES and concomitant antiparkinsonian medications. Initiation and titration performed in the outpatient setting provides patients with personalized and customized therapeutic regimens. Adjustments to dose delivery can result in improved and sustained symptom control that is not possible with oral levodopa administration. The observed reduction in pill burden in CLES-I may result in improved QoL and activities of daily living (ADL) in these patients [8, 17, 21], particularly for those who achieve pill-free status [20].

The DBS cohort started with significantly fewer pills (7.7 versus 12.2) and lower baseline LEDD (511 versus 944 mg) compared with CLES-I. DBS pill burden remained unchanged from baseline throughout follow-up. At 12 months after index, DBS had significantly higher number of pills per day (7.0 versus 3.6) and higher LEDD (594 versus 236 mg) compared with CLES-I. Moreover, the finding that pill reduction and LEDD was essentially unchanged between baseline and follow-up DBS surgery is surprising based on reports from other investigators, particularly for patients undergoing DBS-STN [4, 15,16,17]. Ng et al. found about a 30% reduction in LEDD in DBS-STN patients, which was sustained over a 10-year period with reduced medication and simpler dosing regimens [31]. DBS-GPi is associated with smaller reductions in medication [18, 19]. Our study was unable to determine a target site from our data source. The lower baseline pill burden for DBS patients may reflect less disease severity prior to the intervention. It has been demonstrated that DBS with medical therapy at an earlier stage of PD resulted in improved QoL, ADL, motor disability, levodopa-induced motor complications, and time with good mobility and no dyskinesia when compared with those randomized to receive medical therapy only [15, 16]. However, DBS has limited impact on improvements of non-motor symptoms [32]. While most studies have shown that LEDD can be reduced in patients following STN-DBS [32,33,34], some patients may develop non-motor dopamine withdrawal syndrome, such as apathy or depression [33, 35]. The combination of DBS with dopaminergic medications may be necessary to control the spectrum of motor and non-motor symptoms in patients with APD [32]. This may explain the unchanged pill burden and LEDD in the 12 months after DBS initiation observed in our study.

Limitations of this study include those inherent to the use of administrative data and those specific to our study. First, the 100% Medicare FFS claims data is mainly for billing and reimbursement purposes and not for capturing medical conditions and outcomes. Thus, we were unable to determine certain medical information about our study populations such as age of onset and Parkinson’s disease duration. The study population includes US Medicare patients, so findings may have limited generalizability to other populations. Second, the lack of a standard clinical definition of APD may not control for severity among PD patients included in the study. An international panel of experts, however, recently proposed that taking oral levodopa at least five times daily, having 2 hours of off symptoms daily, or having 1 hour with troublesome dyskinesia daily suggests APD [36]. Third, the reported monthly PD pills reflect prescription fills, which may not all have been taken by patients. Fourth, certain details for the device-aided therapies are not available. Index date for CLES-I may represent the titration period and not optimization of treatment or infusion times, or 16- versus 24-hour use. DBS placement and number of leads for STN and GPi are not available. Target sites have a different effect on pill burden and may have blunted medication reduction findings. Since some of our DBS Medicare patients are aged ≥ 70 years, and DBS is typically performed in patients < 70 years, it is possible that the index surgery represents a second procedure. Finally, as pill burden was the focus of this study due to the extraordinary multi-drug challenges (e.g., drug–drug, drug–food interactions; pill size and quantity; rigid scheduling times) faced by elderly patients with APD, medical device burden was not studied and not available in the dataset for evaluation, although use of devices is not without the possibility of adverse events. Despite these limitations, this study includes the largest sample to date for measuring pill burden in PD patients in the USA, though limited by the number of CLES patients in the PSP database. Since a large majority of APD patients are older than age 65 years and covered by Medicare, the study population is representative of most APD patients in the USA. Moreover, the use of a retrospective observational design to compare the efficacy of CLES and DBS eliminates attrition associated with cross-sectional and case-control study designs. Importantly, the robustness of the analytic approach coupled with real-world data lends novel insights on the comparative effectiveness of the two approved device-aided therapies in the USA.

Conclusions

Findings from this real-world observational study suggest greater pill burden reduction in CLES-I compared with DBS or CLES-NI patients. A significant pill reduction was observed early after CLES initiation and continued to decline throughout the 12-month follow-up. In contrast, pill burden remained unchanged in patients who received DBS, and CLES-NI patients had increases in both quantity and dosages of levodopa during follow-up. This finding demonstrates that pill burden reduction could be considered a treatment goal to minimize the burden on patients and their caretakers, who must keep track of medication schedules among many other responsibilities [37]. Pill burden reduction offers not only an opportunity to simplify treatment regimens but could afford relief to patients to break from a constant cycle of taking multiple pills daily that may impact ADL and QoL, along with possible drug–drug or drug–food interactions. Information on pill burden associated with device-aided therapies, such as CLES or DBS, may further inform patients, caregivers, and clinicians in selection of the optimal therapy to meet patient needs in management of their APD. Further research comparing healthcare resource utilization, PD-related adverse events (e.g., hospitalizations, falls), cost in APD populations treated with device-aided therapies, and longer follow-up would provide further insights into the comparative effectiveness of the two current device-aided therapies available for APD.