Real-world assessment of “OFF” episode–related healthcare resource utilization among patients with Parkinson’s disease in the United States

Abstract

Aims

Within 5 years of initiating carbidopa/levodopa, ∼50% of patients with Parkinson’s disease (PD) experience “OFF” episodes; little is known about the cost burden. We investigated the association of “OFF” episodes with patient characteristics, healthcare resource utilization (HCRU), and healthcare costs.

Methods

Analyses used neurologist-provided data from the US-specific 2017 and 2019 Adelphi Real World Disease Specific Programme for PD, including duration of “OFF” episodes and HCRU for 10–12 consecutive patients. Patients were grouped by presence/absence of “OFF” episodes and by average hours of daily “OFF” time. Between-group differences were assessed for demographics, personal circumstances, and clinical characteristics. Regression analyses modeled the relationship of “OFF” episodes with HCRU and costs.

Results

Of 1,309 patients, 41% experienced “OFF” episodes, 25% of whom were “OFF” ≥4 h/day. Patients having “OFF” episodes had more severe PD, were diagnosed for longer, and were younger than those without “OFF” (p < .0001). “OFF” episodes were associated with a greater number of prescribed PD drugs (p < .0001). Patients without “OFF” episodes were more likely to have full-time employment and less likely to be retired or unemployed because of PD (p < .001). Patients with and without “OFF” episodes had different living situations (p < .001): patients experiencing “OFF” were less likely to live alone and more likely to live in a nursing home and have a professional caregiver (p < .001). In the past 12 months, the number of hospitalizations, intensive care admissions, and emergency room visits; nights hospitalized; costs of consultations and hospitalizations; and total direct costs were all higher for patients experiencing “OFF” episodes (p < .05).

Conclusion

Patients with PD and “OFF” episodes had higher HCRU and costs than those without “OFF,” suggesting that “OFF” episodes contribute to the economic burden of PD. Further research is warranted to examine the extent that current PD treatments and treatment patterns impact HCRU and costs.

Keywords:

• Parkinson’s disease
• “OFF” episodes
• motor fluctuations
• real-world
• healthcare resource utilization
• cost

JEL CLASSIFICATION CODES:

• I10
• I11

Introduction

Parkinson’s disease (PD) is a chronic, progressive, and debilitating neurodegenerative disease. Globally, PD is the second most prevalent neurodegenerative disease, with only Alzheimer’s disease affecting more people^1,2. Worldwide, the number of cases of PD is growing rapidly, with 6.1 million individuals with PD in 2016 compared with 2.5 million in 1990². In the United States, there are approximately 1 million individuals with PD; annually, around 60,000 individuals are diagnosed³. As a result of aging, increased longevity, industrialization, and decreased smoking rates, the prevalence of PD has been forecast to more than double between 2015 and 2040⁴. It has been estimated that 1.2 million people aged ≥45 years in the United States will be diagnosed with PD by 2030⁵.

The underlying cause of PD is the degeneration of dopamine neurons in the substantia nigra, leading to reduced dopaminergic neurotransmission⁶. To counteract this reduction in neurotransmission, the dopamine precursor levodopa has been used for the management of PD since the 1960s⁷. Oral levodopa administered with carbidopa is effective in reducing motor symptoms in most patients and is the gold standard for PD⁸. With disease progression, most patients on chronic carbidopa/levodopa therapy develop motor complications, including motor fluctuations and dyskinesia⁹. Motor fluctuations consist of periods when symptoms improve as a result of the beneficial effect of a dose of carbidopa/levodopa (“ON”) and periods when symptoms reappear or worsen (“OFF” episodes)¹⁰. Motor fluctuations have been reported in 50% of patients with PD after 5 years of initiating carbidopa/levodopa and in 70% of patients beyond 9 years of treatment^11,12. The duration of “OFF” episodes that patients experience varies, but ∼65% of individuals reported spending at least 2 h of their day in “OFF” and >20% reported 4 h or more of “OFF” time per day in a survey conducted by the Michael J. Fox Foundation for Parkinson’s Research¹³.

The economic burden associated with PD is considerable and is forecasted to increase in conjunction with the expected increasing prevalence of PD. The total annual medical costs attributable to PD in the United States in 2017 was $51.9 billion, of which $25.4 billion was attributable to direct medical costs¹⁴. Direct medical costs arising from PD result from numerous medical consultations, hospital outpatient visits, and periods of hospitalization, as well as the costs of medication. Although the economic burden of PD has been reported, to the authors’ knowledge, there are no studies exploring the potential incremental healthcare resource utilization (HCRU) and cost burden specifically associated with the occurrence of “OFF” episodes.

Methods

Data collection

Data collected from the US-specific Adelphi Real World Disease Specific Programme (DSP) for PD were used in this analysis. The DSP is a real-world, cross-sectional survey of physicians and their consulting patients; details of DSP methodology have been published previously¹⁵. Surveys were conducted from May to August 2017 and August to November 2019, in full accordance with the United States Health Insurance Portability and Accountability Act of 1996. All data were collected following procedures with ethics committee approval, including the informed consent of patients, through local fieldwork partners.

Neurologists from a broad geographical range across the Unites States were identified from publicly available lists of healthcare professionals (HCPs) and were invited to participate in the DSP. To be included, neurologists were required to meet the following eligibility criteria: they must have initially fulfilled licensure requirements between 1982 and 2015, been responsible for treatment decisions for patients with PD, and treated ≥10 patients with PD in a typical week. Participating neurologists completed a patient record form (PRF) for the next 10 to 12 consecutive consulting patients with PD. Questions in the PRF were framed such that answers were specifically PD-related (e.g. “How many hospitalizations, including surgery, has the patients had in the last 12 months in relation to their PD?”).

Measures and variables

Information recorded in the PRF included demographics, clinical characteristics, and personal circumstances (i.e. employment status, living situation, and need for a professional caregiver or respite care); current PD stage on the Hoehn & Yahr (H&Y) scale; details of initial healthcare consultations (i.e. visits to HCPs), referrals, and diagnoses; physician’s view of current disease severity and symptom control; and requirement for care and hospitalization. The PRF also included specific questions to capture whether patients were experiencing “OFF” episodes, and if they were, the average hours of daily “OFF” time.

Healthcare resource utilization information included the number of consultations with primary care physicians, neurologists, other specialists, and other HCPs; the number and type of PD-related hospitalizations; the hours of professional caregiving received per week; the institutionalization status; and details of drug treatment received in the past 12 months. Patient history and HCRU information were obtained retrospectively through review of the medical records held at the neurologist’s office.

In the regression analyses described below, dependent variables were the number of hospitalizations, intensive care unit (ICU) admissions, emergency room (ER) visits, and days spent in the hospital for PD in the past 12 months. The occurrence of “OFF” episodes was the main independent variable of interest, with analyses controlled for both the presence/absence of any “OFF” episodes and for the number of hours of daily “OFF” time. Other independent variables (age, sex, body mass index, current PD stage on the H&Y scale, number of concomitant conditions related to mobility, and number of concomitant conditions unrelated to mobility) were adjusted for (Supplementary Table 1). Time since diagnosis was not included as an independent variable; while this variable was significantly correlated with both age and H&Y stage, it was not completed for a substantial proportion of patients.

Analysis

Only patients receiving carbidopa/levodopa and with valid data available for all regression outcomes and covariates were included in the analyses.

Demographic and clinical characteristics, personal circumstances, PD treatment details, and HCRU were grouped by whether the patient experienced “OFF” episodes or not and by the average hours of daily “OFF” time. These data were analyzed descriptively. The significance threshold, alpha, was set at 0.05. The statistical significance of differences between patients experiencing “OFF” episodes and those not experiencing “OFF” episodes was assessed for these variables using the Student t test for continuous variables, a Chi-squared test for multicategorical variables, and the Fisher exact test for binary variables. The association of duration of “OFF” time with demographic and clinical characteristics, personal circumstances, and PD treatment details was analyzed by comparing patients experiencing 0 h of daily “OFF” time vs 1 h vs 2 h vs 3 h vs 4 + hours.

Regression analyses on the effect of “OFF” episodes vs no “OFF” episodes and the change/hour of daily “OFF” time on PD-related HCRU in the past 12 months were performed using negative binomial regression for HCP consultation rate, hospitalization rate, ER visit rate, and inpatient nights. Logistic regression was performed for ICU admission rate. Regression analysis on the effect of “OFF” episodes vs no “OFF” episodes and the change/hour of daily “OFF” time on PD-related costs in the past 12 months was performed by generalized linear modeling (GLM), with log link and gamma family.

Standard errors were adjusted, using the Huber-White sandwich estimator of variance or the robust estimator of variance¹⁶, to allow for intragroup correlation within reporting physician. The usual requirement that the observations be independent was relaxed. Adjusted predictions were produced for each regression (i.e. the predicted outcome for an outcome measure was produced for each hour of daily “OFF” time, assuming sample average values for other regression covariates). No statistical adjustments were made for multiple testing. McFadden’s R² values and mean variance inflation factor (VIF) values were calculated for all outcomes derived from regressions.

Direct medical costs (HCP consultations, PD-related hospitalizations, total direct costs) were calculated as US$/year, based on HCRU as recorded in the PRFs multiplied by unit costs, for patients in all analysis cohorts. All costs are reported in 2020 US dollars. Given the lack of a single central source of healthcare costs for the United States, unit costs were derived from a number of sources: HCP consultation costs were derived from data reported by the American Academy of Neurology, Centers for Disease Control and Prevention, and Centers for Medicare & Medicaid Services^17–19; the unit cost per hospital stay was derived from the Agency for Healthcare Research and Quality (AHRQ) website²⁰.

All analyses were conducted using Stata v16.1 (StataCorp LLC, College Station, TX)²¹.

Results

Patient demographics and clinical characteristics

In total, 210 neurologists provided data for 1,309 patients receiving carbidopa/levodopa. Patient demographic and clinical characteristics for the overall patient population are shown in Table 1. The majority of patients were White males. Almost 60% of patients had an H&Y score indicating fairly mild PD symptoms (i.e. scores of 1–2.5).

Table 1. Patient demographics and clinical characteristics.

	Overall	No “OFF” episodes	“OFF” episodes	p Value^a	Average hours of daily “OFF” time				p Value^b
	(N = 1,309)	(n = 771)	(n = 538)		1	2	3	≥4
					(n = 90)	(n = 182)	(n = 131)	(n = 135)
Age,^c years
Mean (SD)	68.8 (11.3)	68.9 (10.5)	68.5 (12.3)	.5467	71.6 (9.4)	70.5 (10.4)	66.6 (13.5)	65.7 (14.2)	<.0001
Min, max	26, 90	26, 90	33, 90		44, 90	37, 90	37, 90	33, 90
Sex, n (%)
Male	806 (62)	470 (61)	336 (62)	.5846	53 (59)	121 (66)	76 (58)	86 (64)	.5231
Female	503 (38)	301 (39)	202 (38)		37 (41)	61 (34)	55 (42)	49 (36)
Race, n (%)
White	1,055 (81)	601 (78)	454 (84)	.0070	73 (81)	150 (82)	108 (82)	123 (91)	.0114
Black	137 (10)	91 (12)	46 (9)		13 (14)	19 (10)	9 (7)	5 (4)
Asian	58 (4)	44 (6)	14 (3)		2 (2)	5 (3)	5 (4)	2 (1)
Hispanic/Latino	41 (3)	27 (4)	14 (3)		1 (1)	7 (4)	4 (3)	2 (1)
Other	18 (1)	8 (1)	10 (2)		1 (1)	1 (<1)	5 (4)	3 (2)
Current H&Y score, n (%)
1–1.5	303 (23)	261 (34)	42 (8)	<.0001	4 (4)	22 (12)	8 (6)	8 (6)	<.0001
2–2.5	469 (36)	298 (39)	171 (32)		24 (27)	51 (28)	40 (31)	56 (41)
3	368 (28)	170 (22)	198 (37)		43 (48)	71 (39)	49 (37)	35 (26)
4–5	169 (13)	42 (5)	127 (24)		19 (21)	38 (21)	34 (26)	36 (27)
Physician-perceived patient disease severity, n (%)
Advanced PD	232 (18)	52 (7)	180 (33)	<.0001	24 (27)	49 (27)	52 (40)	55 (41)	<.0001
Intermediate PD	712 (54)	393 (51)	319 (59)		58 (64)	109 (60)	77 (59)	75 (56)
Early PD	365 (28)	326 (42)	39 (7)		8 (9)	24 (13)	2 (2)	5 (4)
Age at diagnosis of PD, years
N	1,030	628	402	<.0001	64	141	99	98	<.0001
Mean (SD)	63.2 (11.3)	64.6 (10.6)	61.1 (12.1)		64.3 (9.8)	64.4 (10.5)	58.3 (12.5)	57.1 (13.3)
Min, max	24, 90	25, 90	24, 88		41, 87	32, 88	36, 81	24, 83
Time since diagnosis of PD, years
N	1,030	628	402	<.0001	64	141	99	98	<.0001
Mean (SD)	4.4 (3.9)	3.6 (3.7)	5.8 (4.0)		6.5 (4.3)	5.0 (3.5)	6.6 (4.2)	5.6 (3.9)
Min, max	0.0, 25.1	0.0, 25.1	0.0, 20.9		0.3, 19.5	0.0, 19.2	0.0, 20.9	0.1, 19.4

^aDifference between patients experiencing “OFF” episodes and those not experiencing “OFF” episodes.

^bDifference between patients based on average hours of daily “OFF” time, including patients experiencing 0 daily hours of “OFF” time – 0 h “OFF” time vs 1 h vs 2 h vs 3 h vs 4+ hours.

^cPatients reported to be aged ≥90 years of age were assumed to be 90 years of age in this analysis.

Abbreviations. H&Y, Hoehn & Yahr; PD, Parkinson’s disease; SD, standard deviation.

Relationship of “OFF” episodes with demographic and clinical characteristics

Overall, 41.1% of patients were experiencing “OFF” episodes, of which 16.7%, 33.8%, 24.3%, and 25.1% were experiencing an average of 1, 2, 3, and ≥4 h, respectively, of daily “OFF” time (Table 1).

Patients experiencing “OFF” episodes were similar in age overall to those not experiencing “OFF” episodes. The proportions of male and female patients were similar regardless of the occurrence of “OFF” episodes. The proportion of patients who were White was numerically higher in those experiencing “OFF” episodes, while the proportion of patients who were Black was numerically higher in the group experiencing no “OFF” episodes (Table 1). There was a statistically significant association of race with the occurrence of “OFF” episodes (p = .0070).

While 73% of patients experiencing no “OFF” episodes had an H&Y score <3 (indicative of milder PD motor symptoms), only 40% of patients experiencing “OFF” episodes had an H&Y score <3. By contrast, an H&Y score of 4 or 5 (indicative of severe disability) was more common in patients experiencing “OFF” episodes than those experiencing no “OFF” episodes (p < .0001). Physician perception of disease severity was associated with the occurrence of “OFF” episodes (p < .0001), with fewer patients considered to have advanced PD in the group experiencing no “OFF” episodes than in the group experiencing “OFF” episodes. In addition, more patients experiencing no “OFF” episodes were considered to have early PD than those experiencing “OFF” episodes (Table 1).

The presence of “OFF” episodes was associated with age at diagnosis and time since diagnosis of PD. Patients who were younger at diagnosis and who had been diagnosed with PD for a longer period of time were more likely to have “OFF” episodes (p < .0001; Table 1).

Treatment for Parkinson’s disease

On average, patients experiencing “OFF” episodes were receiving more PD-related drugs than those who were not experiencing “OFF” episodes (p < .0001; Table 2). Patients experiencing “OFF” episodes were more likely to be receiving controlled-release carbidopa/levodopa (p = .0001), dopamine agonists (p < .0001), monoamine oxidase type B inhibitors (p < .0001), and catechol-O-methyl transferase inhibitors (p = .0023) than patients who were not experiencing “OFF” episodes. The proportion of patients receiving each of these drugs increased with increasing hours of daily “OFF” time.

Table 2. Current treatment for PD.

	Overall (N = 1,309)	No “OFF” episodes (n = 771)	“OFF” episodes (n = 538)	p Value^a
Number of current PD drugs
Mean (SD)	1.7 (0.9)	1.5 (0.8)	1.9 (1.0)	<.0001
Min, max	1, 7	1, 7	1, 7
Current PD treatments, n (%)
Carbidopa/levodopa IR	885 (68)	573 (74)	312 (58)	<.0001
Carbidopa/levodopa CR	175 (13)	80 (10)	95 (18)	.0001
Carbidopa/levodopa/entacapone	167 (13)	79 (10)	88 (16)	.0011
Dopamine agonists	309 (24)	144 (19)	165 (31)	<.0001
MAO-B inhibitors	229 (17)	107 (14)	122 (23)	<.0001
COMT inhibitors	58 (4)	23 (3)	35 (7)	.0023
Others	155 (12)	57 (7)	98 (18)	<.0001
Initiated on advanced PD treatments, n (%)
Duopa gel pump^b	8 (1)	4 (1)	4 (1)	.7236
SC apomorphine^c	16 (1)	0 (0)	16 (3)	<.0001
DBS	35 (3)	14 (2)	21 (4)	.0212
	Overall (N = 1,309)	Average hours of daily “OFF” time					p Value^d
		0 (n = 771)	1 (n = 90)	2 (n = 182)	3 (n = 131)	≥4 (n = 135)
Number of current PD drugs
Mean (SD)	1.7 (0.9)	1.5 (0.8)	2.2 (1.0)	1.9 (0.9)	2.0 (1.2)	1.8 (1.0)	<.0001
Min, max	1, 7	1, 7	1, 6	1, 5	1, 7	1, 5
Current PD treatments, n (%)
Carbidopa/levodopa IR	885 (68)	573 (74)	54 (60)	109 (60)	79 (60)	70 (52)	<.0001
Carbidopa/levodopa CR	175 (13)	80 (10)	10 (11)	30 (16)	22 (17)	33 (24)	.0001
Carbidopa/levodopa/ entacapone	167 (13)	79 (10)	17 (19)	40 (22)	15 (11)	16 (12)	<.0001
Dopamine agonists	309 (24)	144 (19)	32 (36)	59 (32)	40 (31)	34 (25)	<.0001
MAO-B inhibitors	229 (17)	107 (14)	32 (36)	45 (25)	28 (21)	17 (13)	<.0001
COMT inhibitors	58 (4)	23 (3)	10 (11)	11 (6)	8 (6)	6 (4)	.0055
Others	155 (12)	57 (7)	17 (19)	36 (20)	21 (16)	24 (18)	<.0001
Initiated on advanced PD treatments, n (%)
Duopa gel pump^b	8 (1)	4 (1)	2 (2)	0 (0)	1 (1)	1 (1)	.1908
SC apomorphine^c	16 (1)	0 (0)	2 (2)	6 (3)	2 (2)	6 (4)	<.0001
DBS	35 (3)	14 (2)	6 (7)	1 (1)	9 (7)	5 (4)	.0005

^aDifference between patients experiencing “OFF” episodes and those not experiencing “OFF” episodes.

^bFor intestinal administration.

^cApomorphine hydrochloride injection pen.

^dDifference between patients based on average hours of daily “OFF” time, including patients experiencing 0 daily hours of “OFF” time – 0 h “OFF” time vs 1 h vs 2 h vs 3 h vs 4+ hours.

Abbreviations. COMT, catechol-O-methyl transferase; CR, controlled release; DBS, deep brain stimulation; IR, immediate release; MAO-B, monoamine oxidase-B; PD, Parkinson’s disease; SC subcutaneous; SD standard deviation.

Patients’ personal circumstances

The employment status of patients differed depending on whether they were experiencing “OFF” episodes (p < .001; Table 3). Overall, fewer patients experiencing “OFF” episodes were employed full-time compared with those not experiencing “OFF” episodes. Patients experiencing “OFF” episodes were more likely to be retired than those not experiencing “OFF” episodes. More patients experiencing “OFF” episodes were unemployed or retired as a result of their PD than those not experiencing “OFF” episodes (p < .0001; Table 3).

Table 3. Patients’ personal circumstances.

	No “OFF” episodes	“OFF” episodes	p Value^a
Employment status	n = 767	n = 535	<.001
Works full-time, n (%)	131 (17)	66 (12)
Works part-time, n (%)	54 (7)	61 (11)
Homemaker, n (%)	85 (11)	31 (6)
Retired, n (%)	461 (60)	345 (64)
Unemployed, n (%)	32 (4)	28 (5)
Other, n (%)	4 (1)	4 (1)
Unemployed or retired as a result of their PD	n = 434	n = 335	<.0001
Yes, n (%)	44 (10)	85 (25)
Living situation	n = 765	n = 532	<.001
Alone, n (%)	97 (13)	33 (6)
With spouse/partner, n (%)	575 (75)	404 (76)
With other family, n (%)	67 (9)	65 (12)
Nursing home, n (%)	21 (3)	27 (5)
Other, n (%)	5 (1)	4 (1)
Professional caregiver	n = 753	n = 527	<.0001
Yes, n (%)	45 (6)	72 (14)
Receiving respite care	n = 748	n = 517	<.001
Yes, n (%)	8 (1)	22 (4)
	Average hours of daily “OFF” time
	0	1	2	3	≥4	p Value^b
Professional caregiver	n = 753	n = 90	n = 179	n = 128	n = 130	<.0001
Yes, n (%)	45 (6)	6 (7)	23 (13)	22 (17)	21 (16)
Receiving respite care	n = 748	n = 90	n = 175	n = 128	n = 124	<.001
Yes, n (%)	8 (1)	6 (7)	6 (3)	8 (6)	2 (2)

^aDifference between patients experiencing “OFF” episodes and those not experiencing “OFF” episodes.

^bDifference between patients based on average hours of daily “OFF” time, including patients experiencing 0 daily hours of “OFF” time – 0 h “OFF” time vs 1 h vs 2 h vs 3 h vs 4+ hours.

Abbreviation. PD, Parkinson’s disease.

Patients experiencing “OFF” episodes were living in different situations than those not having “OFF” episodes (p < .001; Table 3). The proportion of patients not experiencing “OFF” episodes who lived alone was numerically higher than for patients experiencing “OFF” episodes. A higher proportion of patients experiencing “OFF” episodes were cared for by a professional caregiver or received respite care than those not experiencing “OFF” episodes (p < .0001; Table 3), with the proportions increasing with increasing hours of daily “OFF” time (p < .001; Table 3).

Healthcare resource utilization

Patients experiencing “OFF” episodes had more neurologist consultations in the past 12 months than those not experiencing “OFF” episodes (p < .0001; Table 4). An increase in consultations was associated with increasing average hours of daily “OFF” time when 0 h was included in the analysis for neurologists (p < .0001) and movement disorder specialists (p < .05). By contrast, a decrease in primary care physician consultations was associated with increasing average hours of daily “OFF” time (p < .05).

Table 4. HCRU related to PD in the past 12 months.

	Overall (N = 1,309)	No “OFF” episodes (n = 771)	“OFF” episodes (n = 538)	p Value^a	Average hours of daily “OFF” time				p Value^b
					1 (n = 90)	2 (n = 182)	3 (n = 131)	≥4 (n = 135)
Number of PCP consultations
Mean (SD)	0.19 (0.78)	0.16 (0.63)	0.23 (0.96)	.1211	0.41 (1.37)	0.26 (0.86)	0.14 (0.73)	0.16 (0.95)	.0316
Min, max	0, 10	0, 5	0, 10		0, 10	0, 8	0, 6	0, 8
Number of neurologist consultations
Mean (SD)	3.00 (1.41)	2.74 (1.29)	3.37 (1.49)	<.0001	2.89 (1.46)	3.10 (1.30)	3.60 (1.63)	3.81 (1.45)	<.0001
Min, max	0, 10	0, 10	0, 9		0, 9	0, 8	0, 8	0, 7
Number of movement disorder specialist consultations
Mean (SD)	0.03 (0.34)	0.02 (0.29)	0.05 (0.40)	.1761	0.01 (0.11)	0.01 (0.07)	0.07 (0.50)	0.10 (0.60)	.0417
Min, max	0, 5	0, 5	0, 4		0, 1	0, 1	0, 4	0, 4
Number of PD nurse consultations
Mean (SD)	0.01 (0.22)	0.01 (0.14)	0.02 (0.29)	.6001	0.00 (0.00)	0.03 (0.44)	0.00 (0.00)	0.02 (0.26)	.6052
Min, max	0, 6	0, 2	0, 6		0, 0	0, 6	0, 0	0, 3
Number of psychiatrist consultations
Mean (SD)	0.05 (0.60)	0.04 (0.49)	0.06 (0.72)	.5685	0.00 (0.00)	0.00 (0.00)	0.08 (0.96)	0.17 (1.08)	.0947
Min, max	0, 11	0, 8	0, 11		0, 0	0, 0	0, 11	0, 10
Number of hospitalizations^c
Mean (SD)	0.08 (0.33)	0.04 (0.20)	0.13 (0.45)	<.0001	0.07 (0.29)	0.12 (0.43)	0.18 (0.49)	0.16 (0.52)	<.0001
Min, max	0, 3	0, 2	0, 3		0, 2	0, 3	0, 2	0, 3
0, n (%)	1,233 (94)	746 (97)	487 (91)	<.0001	85 (94)	166 (91)	114 (87)	122 (90)	<.0001
1, n (%)	55 (4)	23 (3)	32 (6)		4 (4)	11 (6)	11 (8)	6 (4)
2, n (%)	19 (1)	2 (<1)	17 (3)		1 (1)	4 (2)	6 (5)	6 (4)
≥3, n (%)	2 (<1)	0 (0)	2 (<1)		0 (0)	1 (1)	0 (0)	1 (1)
Number of ICU admissions
Mean (SD)	0.01 (0.08)	0.00 (0.0)	0.01 (0.12)	.0340	0.00 (0.00)	0.01 (0.07)	0.01 (0.09)	0.03 (0.21)	.0074
Min, max	0, 2	0, 1	0, 2		0, 0	0, 1	0, 1	0, 2
Number of visits to the ER
Mean (SD)	0.13 (0.56)	0.07 (0.43)	0.22 (0.70)	<.0001	0.09 (0.32)	0.18 (0.48)	0.24 (0.62)	0.32 (1.09)	<.0001
Min, max	0, 9	0, 9	0, 9		0, 2	0, 3	0, 4	0, 9
Nights spent in the hospital
Mean (SD)	0.27 (1.39)	0.10 (0.71)	0.52 (1.97)	<.0001	0.22 (1.06)	0.42 (1.85)	0.74 (2.11)	0.63 (2.41)	<.0001
Min, max	0, 17	0, 12	0, 17		0, 7	0, 15	0, 10	0, 17

^aDifference between patients experiencing “OFF” episodes and those not experiencing “OFF” episodes.

^bDifference between patients based on average hours of daily “OFF” time, including patients experiencing 0 daily hours of “OFF” time – 0 h “OFF” time vs 1 h vs 2 h vs 3 h vs 4+ hours.

^cIncluding surgery.

Abbreviations. ER, emergency room; HCRU, healthcare resource utilization; ICU, intensive care unit; PCP, primary care physician; PD, Parkinson’s disease; SD, standard deviation.

The number of hospitalizations and the number of ICU admissions, ER visits, and nights spent in the hospital in the past 12 months as a result of their PD were all higher for those experiencing “OFF” episodes than those not experiencing “OFF” episodes (p < .0001 for all except ICU admissions: p < .05; Table 4). For all of these healthcare resource uses, an increase was noted with increasing average hours of daily “OFF” time when 0 h was included in the analysis (p < .0001 for all except ICU admissions: p < .01; Table 4).

Regression analyses on mean PD-related HCRU showed that patients experiencing “OFF” episodes (vs no “OFF” episodes) had significantly higher rates of mean HCP consultations (4.11 vs 3.28; p < .0001), hospitalizations (0.08 vs 0.06; p < .0001), and ER visits (0.16 vs 0.10; p = .001) in the past 12 months (Table 5). Regression analyses did not confirm a difference between patients having “OFF” episodes and those not experiencing “OFF” episodes for ICU admissions or nights in the hospital in the past 12 months (Table 5).

Table 5. Regression analysis on mean PD-related HCRU in the past 12 months per patient.

Mean (SE), 95% CI	No “OFF” episodes	“OFF” episodes	p Value
Number of HCP^a consultations^b	3.28 (0.12)	4.11 (0.43)	<.0001
	3.05, 3.52	3.27, 4.95
Number of hospitalizations^b	0.06 (0.02)	0.08 (0.01)	<.0001
	0.03, 0.09	0.06, 0.11
Number of ICU admissions^c	0.00 (0.00)	0.01 (0.00)	.283
	0.00, 0.00	0.00, 0.01
Number of ER visits^b	0.10 (0.03)	0.16 (0.04)	.001
	0.04, 0.16	0.09, 0.23
Number of nights spent in the hospital^b	0.62 (0.55)	0.58 (0.39)	.258
	–0.45, 1.70	–0.19, 1.36

^aIncludes PCPs, neurologists, movement disorder specialists, PD nurses, and psychiatrists.

^bNegative binomial regression analysis.

^cLogistic regression analysis.

Abbreviations. CI, confidence interval; ER, emergency room; HCP, healthcare professional; HCRU, healthcare resource utilization; ICU, intensive care unit; PCP, primary care physician; PD, Parkinson’s disease; SE, standard error.

Regression analyses on the effect of “OFF” episodes on PD-related HCRU showed a significant positive association of rates of HCP consultations in the past 12 months with the occurrence of “OFF” episodes (incident rate ratio [IRR] 1.25; 95% CI 1.04, 1.51; p = .017) and with increasing average hours of daily “OFF” time (IRR 1.08; 95% CI 1.01, 1.15; p = .025) (Table 6). The number of ICU admissions in the past 12 months also increased with increasing average hours of daily “OFF” time (odds ratio 1.48; 95% CI 1.23, 1.77; p < .001) (Table 6).

Table 6. Regression analysis on the effect of “OFF” episodes on PD-related HCRU in the past 12 months.

	“OFF” episodes vs no “OFF” episodes	p Value	Change/hour of daily “OFF” time^a	p Value
HCP consultation rate, IRR^b,c	1.25	.017	1.08	.025
(95% CI)	(1.04, 1.51)		(1.01, 1.15)
Hospitalization rate, IRR^b,c	1.40	.170	1.10	.100
(95% CI)	(0.87, 2.26)		(0.98, 1.23)
ICU admission rate, OR^d,e	4.37	.135	1.48	<.001
(95% CI)	(0.63, 30.17)		(1.23, 1.77)
ER visit rate, IRR^b,c	1.62	.208	1.18	.087
(95% CI)	(0.77, 3.42)		(0.98, 1.43)
In-patient nights, IRR^b,c	0.94	.878	1.06	.544
(95% CI)	(0.43, 2.07)		(0.87, 1.30)

^aEffect associated with the hourly change in “OFF” time (i.e. 0 daily hours of “OFF” time vs 1 h vs 2 h vs 3 h vs 4+ hours).

^bNegative binomial regression analysis.

^cData are multiplicative on the prediction of the outcome (e.g. 1.25 is a 25% greater total cost and 0.94 is a 6% reduction).

^dLogistic regression analysis.

^eData are multiplicative on the odds of the event occurring.

Abbreviations. CI, confidence interval; ER, emergency room; HCP, healthcare professional; HCRU, healthcare resource utilization; ICU, intensive care unit; IRR, incident rate ratio; OR, odds ratio; PD, Parkinson’s disease.

Healthcare costs

Direct costs in the 12 months prior to data collection arising from PD-related HCP consultations and hospitalizations increased with increasing daily hours of “OFF” time, taking into account patients with 0 h of “OFF” time (all p < .0001; Table 7).

Table 7. Direct PD-related healthcare costs in the past 12 months by average hours of daily “OFF” time.

Mean (SD) USD^a	Overall (N = 1,309)	Average hours of daily “OFF” time					p Value
		0 (n = 771)	1 (n = 90)	2 (n = 182)	3 (n = 131)	≥4 (n = 135)
HCP consultations	$397 (208)	$355 (182)	$400 (215)	$412 (181)	$484 (247)	$524 (250)	<.0001
Hospitalizations^b	$946 (4,644)	$346 (2,366)	$753 (3,524)	$1,517 (6,156)	$2,527 (7,059)	$2,192 (8,003)	<.0001
Total direct costs	$11,338 (38,602)	$8,124 (39,614)	$13,130 (28,424)	$11,742 (30,175)	$22,130 (49,446)	$17,489 (34,286)	<.001

^a2020.

^bInclusive of ER and ICU visits.

Abbreviations. ER, emergency room; HCP, healthcare professional; ICU, intensive care unit; PD, Parkinson’s disease; SD, standard deviation; USD, United States dollars.

Regression analyses on mean PD-related healthcare costs showed that patients experiencing “OFF” episodes (vs no “OFF” episodes) had significantly higher mean costs for HCP consultations ($428 vs $372; p < .001) and total direct costs ($12,748 vs $9,895; p < .001) in the past 12 months (Figure 1). Patients experiencing “OFF” episodes vs no “OFF” episodes had numerically lower costs for hospitalizations; however, this trend was not statistically significant ($11,741 vs $23,361; p = .565).

Figure 1. GLM regression analysis on mean PD-related healthcare costs in the past 12 months. Abbreviations. GLM, generalized linear model; HCP, healthcare professional; PD, Parkinson’s disease; USD, United States dollars. Costs are reported in 2020 USD.

Generalized linear model regression analyses showed a significant positive association of the cost of HCP consultations in the 12 months prior to the survey with the presence vs absence of “OFF” episodes (estimated effect 1.15; 95% CI 1.07, 1.24; p < .001) and with increasing daily hours of “OFF” time (1.05; 95% CI 1.03, 1.08; p < .001) (Table 8). A nonsignificant trend toward decreased costs of hospitalizations was observed for the presence vs absence of “OFF” episodes (0.50; 95% CI 0.21, 1.23; p = .132). All mean VIF values for all regression analyses (Tables 5, 6, 8 and Figure 1) were low (between 1 and 2), indicating a lack of substantial multicollinearity, and all R² values ranged between 0.003 and 0.247.

Table 8. GLM regression analysis on the effect of “OFF” episodes on PD-related healthcare costs in the past 12 months.

	“OFF” episodes vs no “OFF” episodes	p Value	Change/hour of daily “OFF” time^a	p Value
Costs of HCP consultations (US$)^b	1.15	<.001	1.05	<.001
	(1.07, 1.24)		(1.03, 1.08)
Costs of hospitalizations (US$)^b	0.50	.132	0.98	.807
	(0.21, 1.23)		(0.80, 1.19)
Total direct costs (US$)^b	1.29	.137	1.08	.100
	(0.92, 1.80)		(0.98, 1.20)

Costs are in 2020 United States dollars.

^aEffect associated with the hourly change in “OFF” time (i.e. 0 daily hours of “OFF” time vs 1 h vs 2 h vs 3 h vs 4+ hours).

^bData represent the exponential of coefficient (95% CI) from GLM regression analysis and are multiplicative on the prediction of the outcome (e.g. 1.15 is a 15% greater total cost and 0.50 is a 50% reduction).

Abbreviations. CI, confidence interval; GLM, generalized linear model; HCP, healthcare professional; PD, Parkinson’s disease.

Discussion

In this analysis using real-world data, 41% of 1,309 patients with PD currently receiving carbidopa/levodopa were experiencing an average of at least 1 h of daily “OFF” time, with 25% having ≥4 h of daily “OFF” time. Having more severe PD and having been diagnosed with PD longer and at a younger age were all associated with experiencing “OFF” episodes, while age and sex were not.

Other studies have also reported that being diagnosed for longer and having a longer duration of levodopa treatment are associated with the occurrence of “OFF” episodes^22–24. These studies also indicate that “OFF” episodes are more common in patients with younger onset of PD^23–25. The frequency of “OFF” episodes was similar in males and females in one published study²⁴; in another study, patients experiencing “OFF” episodes were more likely to be female and have longer duration of levodopa treatment²². H&Y stage was similar between those experiencing “OFF” episodes and those not having “OFF” episodes in one large multicentre study in Japan²².

Patients who had “OFF” episodes were less likely to be employed full-time and were more likely to be unemployed or retired as a result of their PD. These patients were also less likely to live alone and more likely to be living in a nursing home and being cared for by a professional caregiver than those who had no “OFF” episodes. This suggests an increasing caregiver burden associated with “OFF” episodes, as was observed in other studies²⁶.

Patients experiencing “OFF” episodes had higher HCRU than those not experiencing “OFF” episodes and were more likely to have more neurologist consultations, hospitalizations, ICU admissions, and ER visits as a result of PD in the 12 months before the study than those not experiencing “OFF” episodes. The finding of increased HCRU in patients with “OFF” episodes was observed even when correcting for potentially confounding variables, such as age and concomitant conditions related and unrelated to mobility. A number of analyses from the International Multicenter National Parkinson Foundation Quality Improvement study have showed that ER visits and hospitalizations are more likely in patients experiencing “OFF” episodes^27,28. Despite several low R² values, we observed statistically significant relationships, and the conclusions drawn from these models can be valuable.

Costs data for this study were obtained from HCUPnet, which is sponsored by AHRQ²⁰. While costs may not be exact for patients with PD, there are studies that have utilized AHRQ as a source to derive costs data in other disease states^29–31. In our study, the higher level of HCRU reported for patients experiencing “OFF” episodes was reflected in higher healthcare costs than those in patients not experiencing “OFF” episodes. The costs of PD-related HCP consultations and hospitalizations in the 12 months before data collection all increased with increasing average hours of daily “OFF” time. Other studies have shown that the occurrence of “OFF” episodes is associated with increased healthcare costs³². Interestingly, GLM regression analyses demonstrated a possible trend toward decreased cost of hospitalizations with the presence vs absence of “OFF” episodes, suggesting that the impact of experiencing “OFF” episodes may be variable across HCRU outcomes. High coefficients for the effect of H&Y scores (not reported) were observed. This suggests that this covariate may influence hospitalizations more than other HCRU outcomes; however, this finding requires further investigation and confirmation.

Throughout the literature, the DSP methodology has been used to assess the relationships between diseases and/or symptoms of disease with HCRU^33–35; nonetheless, there are some limitations to this specific study that should be highlighted. As PRFs were completed by neurologists for the next 10 to 12 consecutively consulting patients with PD, the sample collected was pseudo-random, rather than a truly random sample. This was a cross-sectional rather than a longitudinal survey; therefore, data may be used to assess the association between factors but not to assess causality. As with all studies of this type, the methodology relies on accurate reporting by neurologists. For example, it was not possible to assess whether all patients included in the survey were seen by neurologists while in the “ON” or “OFF” state, and the reporting of daily hours of “OFF” time was not collected longitudinally by patient diaries or other means. Daily “OFF” hours were reported by the neurologist based on information gathered through neurologist/patient encounters.

Time since diagnosis was not included as a confounder in regression analyses, as the amount of missing data would have resulted in loss of analysis power. However, H&Y stage was included as an independent variable and is highly correlated with time since diagnosis. While there was adjustment for concomitant conditions in the analysis, not all conditions that were PD related (e.g. dementia, psychosis, orthostatic hypotension, freezing) were collected. These conditions may have increased the overall HCRU and cost burden. The associations reported herein could have been the result of an underlying feature of PD driving the effect on the outcome other than “OFF” episodes. Care should be taken with interpreting any outcome in which HCRU counts were particularly small (e.g. ICU admissions, PD nurse consultations, etc.), as this may have impacted statistical significance. Lastly, no adjustment for multiple testing was made in this study; therefore, caution should be used in interpreting the results, particularly those that are counterintuitive or associated with a P value that indicates statistical nonsignificance. While acknowledging these limitations, a substantial body of data from a representative population of patients with PD was included in the analysis.

Conclusions

Using real-world data from patients with PD, these findings provide valuable insight into the association of “OFF” episodes in PD with demographics, clinical characteristics, personal circumstances, treatment, HCRU, and healthcare costs. Further research is needed to examine the extent to which established PD treatments and treatment patterns may impact HCRU and costs and whether newly approved or novel treatments might result in additional reductions in the need for resources and increased cost savings.

Transparency

Declaration of funding

This study was supported by funding from Sunovion Pharmaceuticals Inc. (Marlborough, MA, USA).

Declaration of financial/other relationships

AT and EP are employees of Sunovion Pharmaceuticals Inc. (Marlborough, MA, USA). EJ, JP, JW, and AG are employees of Adelphi Real World (Manchester, UK).

JME peer reviewers on this manuscript have received an honorarium from JME for their review work, but have no other relevant financial relationships to disclose.

Author contributions

All authors (AT, EJ, EP, JP, JW, and AG) contributed to conception and design and have been involved in drafting the manuscript and revising it for critically important intellectual content. All authors (AT, EJ, EP, JP, JW, and AG) read and approved the final manuscript and agree to be accountable for all aspects of the work.

Medical writing support under the guidance of the authors was provided by Carole Evans, PhD, on behalf of Adelphi Real World (Manchester, UK) and Robert Schupp, PharmD, CMPP, on behalf of The Lockwood Group (Stamford, CT, USA) and was supported by funding from Sunovion Pharmaceuticals Inc. (Marlborough, MA, USA) in accordance with Good Publication Practice (GPP3) guidelines.

Previous presentations

Data were presented in part at the 2nd Pan American Parkinson’s Disease and Movement Disorders Congress (MDS-PAS), June 22–24, 2018, Miami, FL, and the International Congress of Parkinson’s Disease and Movement Disorders (MDS), October 5–9, 2018, Hong Kong.

Acknowledgements

The authors would like to thank all patients and physicians who participated in the Adelphi Real World Disease Specific Programme for Parkinson’s Disease in the United States.