Bicycling for rehabilitation of persons with Parkinson Disease: A scoping review

Abstract

Background and Purpose:

Exercise is beneficial for Persons with Parkinson Disease (PwPD). The overarching purpose of this scoping review was to provide guidance to clinicians and scientists regarding current evidence for bicycling exercise for PwPD. A scoping review was conducted to examine the heterogeneous literature on stationary bicycling for PwPD to reduce motor symptoms and body function structure (BFS) impairments, improve activities and motor performance, and reduce disease severity.

Methods:

The PRISMA-ScR guidelines were followed. PubMed, CINAHL, and PEDRO were searched from inception to January 23, 2023. Articles reporting original data on relevant outcome measures were included. Search results were screened, and articles were extracted. Data were analyzed quantitatively with percentages of significant and clinically meaningful findings and qualitatively to extract themes.

Results:

Bicycling was categorized using bicycle types (assisted, non-assisted) and training modes (speed, aerobic, force). A high percent of the 34 studies showed statistical significance for reducing motor symptoms (83%), BFS impairments (78%), disease severity (82%) and improving activities (gait 72%, balance 60%). Clinically meaningful findings were achieved in 71% of the studies for reduction in disease severity and in 50% for improving gait.

Discussion and Conclusions:

The literature on bicycling for PwPD has evolved from speed to aerobic studies. The terminology describing types of bicycling was simplified. Of all the outcomes reported, reduction of disease severity achieved the highest frequency of clinical meaningful improvements. Bicycling was comparable to other forms of aerobic training for walking speed and endurance. Opportunities for translation to practice and research are presented.

INTRODUCTION

Over the course of the next 30 years, the worldwide prevalence of PD is expected to double from ~6 million to ~12 million.¹ The motor symptoms of PD (muscle rigidity, postural instability, bradykinesia, and resting tremor) along with body function structure (BFS) impairments (e.g., weakness, reduced vital capacity and fatigue) restrict motor performance, activity, and participation.^{2, 3} While pharmacological management reduces motor symptoms and improves motor performance, there are side effects, and the medication wears off. ³ Exercise as a tool to reduce motor symptoms and improve motor performance has been studied extensively in persons with PD (PwPD).⁴ Evidence demonstrated that exercise can complement the use of pharmacotherapy.^4–6 Exercise interventions have included treadmill training,⁶ aquatics,⁷ dancing⁸ and stationary bicycling in a variety of forms.⁹ Stationary bicycling is particularly appealing as it offers the potential of independent exercise in a safe environment. It is difficult to synthesize the existing literature on stationary bicycling and interpret findings to guide application to practice and research due to the variety of study designs, types of bicycling studied, and outcome measures used.

Stationary bicycling has been investigated using a variety of study designs to answer questions about different types of cycling, efficacy in single group studies and comparisons to other forms of exercise. Systematic reviews synthesizing this literature have either focused on a single type of bicycling (e.g. forced)⁹ or reported on single group and randomized controlled trials identifying statistical improvements in gait and balance.¹⁰ Each of the systematic reviews addressed a select aspect of the literature but did not represent the variety of bicycling approaches (bicycle type and training mode) nor did they interpret findings in the context of clinically meaningful changes. Further complicating synthesis, the terminology used in this literature to report outcomes of bicycling for PwPD comes from both the neurology literature (motor symptoms and motor performance) and the International Classification of Functioning (ICF) used in the rehabilitation literature (BFS, activities).¹¹

The overarching purpose of this scoping review was to unify the literature in order to provide guidance to clinicians and scientists in interpreting the current literature. A scoping review is particularly well suited for this purpose and does so by examining the extent (number of articles), range (variety), and nature (characteristics) of the evidence on bicycling for persons with PD and summarizing findings from this body of knowledge that is heterogeneous in methods and discipline.¹² The review began by describing the evidence of bicycling for PwPD, based on study design, by identifying both statistically significant and clinically meaningful changes. Data were presented related to reducing motor symptoms and BFS impairments, improving activities and motor performance, and reducing disease severity. Next, this review proposed unifying terminology in categorizing the bicycling type (e.g. assisted, non-assisted), training mode (e.g. speed, aerobic) and outcomes of bicycling for PwPD using the ICF. Implications of findings then were considered for clinical practice, and gaps in knowledge were identified that could be addressed in future rehabilitation studies.

METHODS

The study used the first five stages of a methodological framework consisting of: 1) Identifying the relevant research questions; 2) Searching for relevant studies; 3) Selecting studies; 4) Charting the data; 5) Collating and synthesizing the results.^{13, 14} Consulting with stakeholders to inform or validate study findings is optional and was omitted. The review followed the PRISMA-ScR (Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews) reporting guidelines.¹² The protocol was registered in the Open Science Framework on October 8, 2020 (https://osf.io/g6vef).

STAGE 1: Identifying the Relevant Research Questions

Three research questions were generated to address the first purpose, using a Population-Intervention-Comparison-Outcome (PICO) or PIO format to guide the scoping review. The terms selected were both from the neurology and rehabilitation literatures, specifically using the terminology from the ICF from the rehabilitation literature.¹¹ For PwPD:

Question 1: Is there evidence that bicycling reduces motor symptoms and BFS impairments?

Question 2: Is there evidence that bicycling improves motor performance and activities?

Question 3: Is there evidence that bicycling decreases disease severity?

STAGE 2: Searching for Relevant Studies

The search was executed in PubMed, the Cumulative Index to Nursing and Allied Health Literature (CINAHL), and the Physiotherapy Evidence Database (PEDro) by two researchers (LJ and JP). Databases were searched from inception to January 23, 2023. The PIO or PICO format guided search term selection. PubMed and CINAHL were searched using MeSH terms and key words, while only key words were searched in PEDro. Gray referencing and contact with key experts were not conducted. See Supplemental Digital Content 1 for full search strings.

STAGE 3: Selecting Studies

Articles published in peer-review journals meeting the inclusion and exclusion criteria in Supplemental Digital Content 2, Table 1 were selected.

Results from the searches were combined and duplicates removed. The titles and abstracts of the studies were screened for eligibility by two reviewers (LJ and JP). Results were compared. Studies were included in the next round of screening if there was a disagreement between the two reviewers. Inter-rater agreement was 91%.

Full-text articles were screened in the second round by the same reviewers. Inter-rater agreement at the full-text level was 95%. Disagreements about articles to include were resolved by the arbitrator (JED). For the PRISMA screening chart, see Figure 1.

Figure 1: PRISMA-ScR Flow Chart.

The article selection process following the PRISMA guidelines extension for scoping reviews.

STAGE 4. Charting the Data

The data extraction categories were set a priori by two investigators (JED and JP) and revised after a preliminary search. A template for the data charting form was created before the extraction and included in the protocol registration. Data charting was conducted by two independent researchers (LJ & JP). Reliability of data extraction was assessed on 4 studies. Inter-rater agreement for charting the data was 86%. The arbitrator (JED) and the extractors reviewed and iterated the charting strategy. Any changes in charting strategy were applied to all the studies.

Extracted data were charted using a PIO or PICO format as follows:

1. Population included participant’s age, gender, disease severity, disease duration and study sample size.

2. Intervention included the type of stationary bicycling as stated by the author (forced- typically on an actuated bicycle (motor driven) or with tandem cycling, passive), mode of bicycling (interval or continuous), dose (frequency, duration, intensity), and training mode based on how target intensity was determined/measured (% of maximum heart rate for aerobic; bicycling cadence for speed; or watts for force). Whether the intervention was performed during the on/off state was also recorded.

3. Comparison group details were extracted for the bicycling comparisons and randomized controlled trials (RCTs) using an approach similar to the data extraction of the intervention group.

4. Outcomes were extracted separately for

   1. BFS impairments that reflect motor symptoms specific to PD (e.g., bradykinesia, rigidity),
   2. Other BFS impairments (e.g. strength, cardiovascular measures, postural control) to reflect non-PD specific effects,
   3. motor-performance-activities (e.g. balance and gait), and
   4. motor disease severity using the Unified Parkinson’s Disease Rating Scale Motor Section (UPDRS-III).¹⁵

Definitions of BFS and activity followed the ICF.¹¹ When the BFS was a specific motor symptom (e.g., rigidity) it was extracted into a separate Motor Symptom column, while non-specific BFS impairments, (e.g. strength, aerobic capacity) were entered in the BFS column. Outcomes were labeled as being statistically significant as well as if they met the minimal clinically importance difference (MCID) or minimal detectable change (MDC) when MCIDs were not available. MCIDs available for PwPD included: self-selected gait speed (small 0.06 m/s, medium 0.14 m/s and large 0.22 m/s),¹⁶ UPDRS-III −3.5 points,¹⁷ UPDRS-II- 3.05 points.¹⁸ MDCs for PwPD included: BBS 5 points,^{19, 20} functional reach test (FRT) - 9 inches,¹⁹ six-minute walk test (6MWT) 82 meters,^{19, 20} comfortable gait speed, 0.18m/s and fast gait speed 0.25 m/s.^{19, 20} (See Supplemental Digital Content 2, Table 2) It was noted if outcomes were collected during the on/off state of medication.

STAGE 5: Collating and Synthesizing the Results

Data were extracted by study design into 4 tables: 1) comparing PwPD to healthy adults; 2) comparing different modes/types of bicycling; 3) comparing pre-post outcomes in a single-group and 4) RCTs. This approach was used to capture the breadth of the literature and to reflect its evolution. The terms ‘assisted’ and ‘non-assisted’ bicycling were used to reduce the inconsistency of the terminology describing type of biking. Specifically, the authors’ terms were coded according to the following definitions. Assisted: cycling cadence is achieved by an external force applied (motorized or tandem by another rider) to the pedals. This force may augment the riders’ own efforts and result in cadences higher than what the cyclists can achieve on their own. Non-Assisted: cycling cadence is achieved solely by the riders’ efforts.^{9, 21} Other authors have used the terms motorized and non-motorized¹⁰ but this does not distinguish between tandem cycling which is non-motorized, but also assisted.

All studies were further organized based on the training mode of the intervention (e.g. how the study was dosed) as 1. Speed- bicycling cadence represented with revolutions per minute (RPM), 2. Aerobic- percent of maximum heart rate or heart rate reserve (HRR) or 3. Force –watts or peak workload. Three investigators independently reviewed the four tables to extract themes and patterns.

Frequency counts and percentages of articles that reported motor symptoms, BFS, motor performance-activity and disease severity across all the studies were calculated. Concept maps - graphical representations of study characteristics (nodes) and their relationships (links) - were generated in Figure 2.²² The nodes for the 4 study characteristics were: bicycling type (assisted or non-assisted), training mode (speed, aerobic, or force), exercise duration (hours), and outcomes by ICF category. The maps were generated to examine implications for practice, as they contain parameters clinicians would need to apply the findings, as well as to guide future research. The concept maps were generated using the iGraph package in R²³ for the bicycling comparison, single group and RCT studies. Frequency counts were used as weights for each node, and connections between nodes were represented as lines.

Figure 2: Concept Maps for Bicycling Comparison Studies, Single Group Studies, and RCTs.

Concept maps are shown for the bicycling comparison studies (Figure 2.a), the single group studies (Figure 2.b), and the RCTs (Figure 2.c). Nodes in the concept maps represent: with a circle training mode (speed, aerobic or force) with a triangle the type of bicycle (assisted or unassisted) with a star the duration of training in total hours, with a square the outcomes measured (e.g. motor symptoms, body function structure, activities and UPRDS). Node size represents frequency of studies, with a larger size indicating a higher frequency (e.g. In Figure 2.c the activity node was the largest outcome node). Connections between nodes are represented as lines in the concept map, where thicker lines represent connections that were more frequently reported. These lines are also color-coded based on frequency count following the color key (e.g. in Figure 2.c the connection between aerobic mode and BFS and activities has the highest frequency as noted by the purple line). Bicycling comparison studies (2.a) exclusively implemented speed-based protocols, using mostly assisted bicycles. Single-group and bicycling comparison studies (2.b) were primarily speed-based while RCTs (2.c) more often used aerobic bicycling approaches. Both single group and RCTs used non-assisted bicycles more than assisted bicycles. Only in the RCTs was study duration greater than 30 hours.

RESULTS

Overview

Thirty-four studies were included in this scoping review. There were 1037 participants with mild-moderate PD Hoehn & Yahr (primarily H&Y1.0 −3.0, with a few participants H&Y 4). Across all studies the mean age of participants ranged from 56–72 years and their mean disease duration ranged from 3–8 years. The frequency counts for each research question to describe the evidence are summarized in Supplemental Digital Content 2, Table 3. Studies, organized by design, identified the type of bicycle (assisted or non-assisted) and training mode (aerobic, speed and force). Sixteen studies used assisted bicycling (with terms such as: forced,^{24, 25} passive,²⁶ active assisted ²⁶ and dynamic cycling)²⁷ with either motorized or tandem bicycles; and 18 used non-assisted bicycling (also called voluntary,²¹ high speed low-resistance,²⁸ preferred cadence,²⁸ static cycling,²⁷ low-intensity).²⁹ Seventeen studies used speed,^{21, 26–41} 15 used aerobic^{5, 24, 25, 42–53} and 2 used force (resistance)^{54, 55} as the training mode. Most studies (n= 28) used continuous cycling, with the remaining using interval cycling (n = 6).^{28, 35, 38, 46, 48, 52} Dose of cycling varied from a single session of 40 minutes³⁷ to 78 sessions over 6 months for a total dose of 58.5 hours.⁵

Nineteen studies reported interventions that were performed in the ON state (19/22); three studies^{33, 34, 37} reported on the short-term effects of bicycling that was performed in the OFF state (3/22); the remaining 12 studies did not specify if the interventions were performed in the ON or OFF state.^{5, 25, 30, 31, 36, 39, 41, 43, 44, 50, 52, 54} Seventeen studies measured outcomes during the ON state (17/28) with the remainder measured in the OFF state (11/28).^{5, 21, 25, 29–31, 33–35, 37, 46} Six of 34 studies did not specify if outcomes were measured in the ON or OFF state.^{38, 39, 41, 44, 50, 52} The majority of studies reporting outcomes in the OFF state measured disease severity with the UPDRS-III in the OFF state (6/10).^{5, 21, 29, 30, 34, 35} Tables 4–7 present the research findings by study design.

Table 4: Healthy Adults and Persons with Parkinson Disease Comparison.

• Studies of persons with PD compared to healthy adults and persons with PD organized using the PIO format.

Study	Population (Parkinson) 1	Population (Healthy)	Bicycle (Mode/Type)	Intervention	Motor Symptoms	BFS	UPDRS-III	Activities
Gallagher 201641	n=15 Age: 66.3 (9.6) H&Y:2.3 (0.5) UPDRS-III: Disease Duration 35.5 (14.2)	n=13 Age=66.7 (9.1)	Non-Assisted Upright	Continuous cycling 4 trials with auditory cues; 5 with Visual Cues RPM: 20% faster than baseline 9 minutes × 1 session Total time: 9 minutes HC matched intervention				RPM (Auditory Condition) PD HC Auditory Cues (AC) ↑ ↑ Virtual Environment (VE) ↑ NS AC + VE ↑ ↑ RPM (Visual Condition) VE ↑ NS VE+Visual Cues (VC) ↑ ↑ VE +20% faster VC ↑ ↑ 20% faster VC + Instruction ↑ ↑
Nadeau 201651 Nadeau 201850 Duchesne 201549	n= 19 Age: 59 (7.11) H&Y: 2.1 (0.2) UPDRS-III: 21.84 (6.16) Disease Duration 8.1(9.12)	n=20 Age= 64 (8.19)	Non-Assisted Recumbent	Continuous cycling HR: 60–80% HR Max (graded); RPM: 60 RPM 36 minutes × 3 × 12 weeks Total time: 21.5 hours HC matched intervention	Rigidity NS Tremor NS	PD HC Power Level ↑ ↑ VO2Max ↑ ↑ Single leg support ↓ NS Cadence ↑ NS Step Length, Step width, Double Support time: NS	NS	PD HC Walking Speed ↑▲ NS

¹the last entry of the population cell indicates disease duration in years; ↑, improvement (+); ↓, Improvement (−); △, exceeds MDC; ^{19, 20} ▲ exceeds MCID; ¹⁶ ◇, decrease in severity in UPDRS-3 (eg. moderate to mild); ☆, outcome measured or intervention performed in the OFF state; AC, auditory cues; Age, measured in years; BFS, body function structure; DM, dose-matched; H&Y, Hoehn & Yahr score; HC, healthy controls; HR, heart rate; NR, not-reported; NS, no significant difference; RPE, rating or perceived exertion; RPM, revolutions per minute; UPDRS-III, Unified Parkinson’s Disease Rating Scale Section 3 (motor exam); VC, visual cues; VE, virtual environment; VO2, maximal oxygen consumption

Table 7: Randomized Controlled Trials.

• Studies comparing bicycling to other forms of exercise using a randomized controlled design, organized using a PICO format.

Study	Population 1	Bicycle (Mode/Type)	Intervention	Comparison	Motor Symptoms	BFS	UPDRS-III	Balance	Gait & Activities
Speed
Laupheimer 201139	N = 44 (21,23) Age = 68 (8) H&Y = 2.7 (0.8) Disease Duration 7 (6) years	Assisted (actuated) Non-custom Recumbent	Forced Exercise (FE) cycling: passive or active RPM: 90 RPM or fastest tolerable 40 minutes × 5×10 weeks Total time: 33.3 hours	Usual Care (standard rehab) Dose type: NA Total time: NA	Tremor Cycle NS SoC DDK (right) Cycle > SoC				Gait Speed Cycle ▲ < SoC Upper limb (TMT) Cycle NS SoC
Ridgel 201940	N =16 (8,8) Age = 70 (7) H&Y = 1.4 (0.5) * Disease Duration 5 (2)	Assisted (actuated) Custom Recumbent	Forced Dynamic cycling RPM: 80 RPM 40 minutes × 3 × 2 weeks Total time: 4 hours	Stretching Dose type: NA Total time: dose matched	Cycle Stretch Bradykinesia ↑ NS Rigidity ↓ NS		Cycle Stretch ↓ NS	Cycle Stretch TUG ↓ NS	Cycle Stretch Gait Speed ↓ NS
Aerobic
Linder 202225	N = 28 (14,14) Age = 64.9 (6) UPDRS-III = 35.0 (10) Disease Duration NR	Assisted (actuated) Custom Recumbent	Forced exercise (FE) cycling HRR: 60–80% 50 minutes × 3 × 8 weeks Total time: 20 hours	Usual Care (Maintain regular activity levels) Dose type: NA Total time: NA		Cycle SoC Cadence ↑ > NS Step Length ↑ > NS Gait Kinetics ↑ NS NS Kinematics NS NS NS			Cycle SoC Gait Speed ↑▲ > NS
Segura 202047	N=13 (6,7) Age = 57.8 (NR) UPDRS-III = 57.7 Disease Duration 2–14 (NR)	Assisted (tandem) Non-custom Upright	FE cycling Conditioning Phase: 40 minutes × 1.5 × 8 weeks (8 hours) Conditioning HR: 50–60% HR Max Training Phase: 45 minutes × 3 × 16 weeks (36 hours) HR: 80% Max; RPM: ≥80 RPM Total time: 44 hours	Usual care (medication) Dose type: NA Total time: NA		Cycle SoC VO2 Max ↑ > NS	Cycle SoC NS ▲ NS NS		UPDRS-II Cycle NS SoC
Qutubuddin 201324	N = 23 (13, 10) Age = 68 (9) UPDRS-III = 16 (6) Disease Duration 7 (6)	Assisted (actuated) Non-custom Recumbent	FE cycling HR: 61–80% HR Max 30 minutes × 2× 8 weeks Total time: 8 hours	Usual care (standard rehab) Dose type: NA Total time: NA		Cycle SoC Finger Taps NS NS	Cycle SoC ↓ ▲ NS	Cycle SoC BBS NS NS
Cancela 202044	N=12 (7,5) Age = 68 (5) H&Y = 2.3 (0.8) Disease Duration NR	Assisted (actuated) Non-custom Recumbent	Continuous cycling + Tai Chi & Balance HR: 70% HRR 1.25 hours × 2 × 8 weeks (graded) Total time: 20 hours	Tai chi & balance Dose type: NA Total time: 9.3 hours		Cycle Tai Chi 2-min Step ↑ > NS	Cycle Tai Chi ↓▲ > NS▲	Cycle Tai Chi TUG NS NS NS TAT NS NS NS	Cycle Tai Chi 6MWT NS NS NS
Marusiak☆ 201946	N = 20 (10,10) Age = 72 (10) H&Y = 2.4 (0.6) Disease Duration 9 (5)	Non-Assisted Non-custom Upright	Interval cycling HR: (60–75% max, graded), RPM: 30% > voluntary 1 hour × 3 × 8 weeks Total time: 24 hours	Usual Care (standard rehab) Dose type: NA Total time: NA	Cycle SoC Bradykinesia ↓ < NS	Cycle SoC Grip Force ↑ > NS			Cycle SoC UPDRS-II ↓ > NS
vanderKolk ☆ 20195	N = 130 (65, 65) Age = 59 (8) H&Y =1.9 (0.2) Disease Duration 3 (1.3 – 7.3)	Non-Assisted Non-custom NR	Continuous cycling HR: 50–70% HRR (graded) 45 minutes × 3 × 26 weeks Total time: 58.5 hours	Stretching (flexibility, relaxation) Dose type: NA Total time: 9 hours		VO2 Max Cycle > Stretch Finger Taps Cycle. NS Stretch	On Cycle NS Stretch Off Cycle > Stretch	Mini-BESTest Cycle NS Stretch TUG Cycle. NS Stretch # falls Cycle. NS Stretch	6MWT Cycle NS Stretch Pegboard test Cycle NS Stretch
Burini 200643	N = 22 (11, 11) Age = 66 (7) H&Y = 2.8 (0.4) Disease Duration 11 (5)	Non-Assisted Non-custom NR	Continuous cycling HR: 50–60% max HRR 50 minutes × 3×7 weeks (20 sessions) Total time: 16.7 hours	Qigong (breathing, stretching) Dose type: NA Total time: dose matched		Cycle Qigong VO2 Peak ↓ NS DPP ↓ NS HR Peak NS NS	Cycle Qigong NS NS		Cycle Qigong 6MWT ↑ NS
Arcolin 201542	N =29 (16,13) Age = 69 (8) H&Y = 2.3 (0.5) Disease Duration 5 (3)	Non-Assisted Non-custom Upright	Continuous cycling + stretch & balance RPE: 11–14 on Borg Scale (graded) 2 hours × 5 ×3 weeks Total time: 30 hours	Treadmill (TM) + stretching & balance RPE: 11–14 Borg Scale (graded) Total time: dose matched		Cycle. TM Step Length ↑ ↑ Cadence (walk) ↑ NS	Cycle TM ↓▲ ↓▲	Cycle TM Mini-BEST ↑ ↑ TUG ↓ ↓	Cycle TM 6MWT ↑ ↑ Gait Speed ↑▲A↑▲
Tollar 201848	N =74 (25,25,24) Age = 71 (4) H&Y = 2.4 (0.5) Disease Duration 8 (2)	Non-Assisted Non-custom NR	Interval cycling (5:1) HR: 80% Max HR (110–140 bpm) 1 hour × 5 × 5 weeks Total time: 25 hours	Exergames (EXE) : Just Dance, Reflex Ridge, Space Pop HR: 80% Max HR (110–140 bpm) Total time: dose matched		CoP-EO Cycle NS EXE CoP- EC Cycle NS EXE		BBS Cycle < EXE △ BESTest Cycle NS EXE Tinetti Cycle NS EXE	6MWT Cycle△ NS EXEA DGI Cycle NS EXE PDQ-mobility Cycle NS EXE UPDRS-II Cycle▲ NS EXE▲
Ferraz 201845	N = 62 (20,22, 20) Age = 67 (64–71) H&Y = 2.5 (2 – 3) Disease Duration 6 (4–9)	Non-Assisted Non-custom NR	Continuous cycling RPE: 15, HR: 50–75% of max HR (graded) 50 minutes × 3 × 8 weeks Total time: 20 hours	Functional training (FT) = balance + mobility Exergame (EXE) = River Rush, Reflex Ridge, 20k leaks Dose type: NR Total time: dose matched		Cycle EXE FT Sit-Rise: ↓ ↓ ↓			Cycle EXE FT 6MWT ↑ ↑ ↑ Gait Speed (fast) NS ↓ NS
Force
Demonceau 201654	N = 46 (16,15,15) Age = 65 (8) H&Y = 1.5 (1.0 – 2.5) Disease Duration 5 (2.5 – 8.0)	Non-Assisted Non-custom NR	Continuous and interval cycling PWL: >50% PWL (graded) 1.25 hours × 2.5 × 12 weeks Total time: 37.5 hours	Strength training (full body) 1 Repetition Max: 50–90% (graded) Total time: dose matched		Cycle Strength VO2 Max ↑ NS PWL ↑ ↑ Torque NS ↑ HRR NS NS RER NS NS Stride Length NS NS Cadence (walk) NS NS		Cycle Strength TUG NS NS	Cycle Strength 6MWT NS ↑ Gait speed NS▲ NS▲

¹the last entry of the population cell indicates disease duration in years; ↑, within group improvement (+); ↓, within group worsening (−); ↓, within group improvement (−); △, exceeds MDC; ^{19, 20} ▲ exceeds MCID; ^16–18 <, between group favoring right group; >, between group favoring left group; ◇, decrease in severity in UPDRS-3 (eg. moderate to mild); ☆, outcome measured or intervention performed in the OFF state; 2-min step, 2 minute step test; 6MWT, 6 minute walk test; Age, measured in years; BBS, Berg Balance Scale; BESTest, Balance Evaluation Systems Test; BFS, body function structure; CoP, center of pressure; DDK, dysdiadokokinesia; DGI, dynamic gait index; DM, dose-matched; DPP, double peak product; Dyn, dynamic cycling; EC, eyes closed; EO, eyes open; EXE, exergaming; FE, forced exercise; FT, functional training; gait speed, comfortable gait speed; H&Y, Hoehn & Yahr score; HR, heart rate; HRR, heart rate reserve; mini-BESTest, mini Balance Evaluation Systems Test; NS, no significant difference; PDQ, Parkinson’s Disease Questionnaire; PWL, peak work load; RER, respiratory exchange ratio; RPE, rating of perceived exertion; Sit-Rise, sit rise test; SoC, Standard of Care (Usual Care); TAT, Tinetti Assessment Tool; TM, treadmill training; TMT, Timed Motor Test Battery; TUG, timed up and go; UPDRS-II, Unified Parkinson’s Disease Rating Scale Section 2 (motor experiences of daily living); UPDRS-III, Unified Parkinson’s Disease Rating Scale Section 3 (motor exam); VO2, maximal oxygen consumption

Question 1 For PwPD, is there evidence that bicycling reduces motor symptoms and BFS impairments?

Thirty-two of 34 studies reported motor symptoms or BFS measures. For motor symptoms, 83.3% (10/12) of studies reported significant improvements. For BFS, 78% (21/27) of studies reported significant improvements. MCIDs and MDCs were not available on motor symptoms or BFS outcomes for PwPD. Motor symptoms decreased after active assisted bicycling^{21, 26, 27} but not after non-assisted^{21, 27} or passive (completely assisted) cycling.²⁶ Cycling cadence during assisted passive cycling at 60, 70 and 80 RPM had a similar effect on reducing tremor and bradykinesia.³³

Question 2 For PwPD, is there evidence that bicycling improves motor performance and activities?

For gait, 72.0% (13/18) of studies reported statistically positive findings. Nine studies reported clinically meaningful results.^{25, 26, 28, 38, 39, 42, 48, 51, 54} The clinically meaningful improvements in comfortable gait speed ranged from a small MCID change of 0.06⁵⁴ to a moderate MCID change of 0.20²⁶ to a large MCID change of 0.37³⁹ meters per second.¹⁶ Clinically meaningful improvements, in this instance measured with the MDC for 6MWT were found in one study, reporting an increase in walking distance of 142 meters.⁴⁸ For balance, 60.0% (9/15) of studies reported statistically significant positive findings with none reporting clinically meaningful results. For other activities (e.g. UPDRS-II, 9-Hole Peg Test (9HPT), 44.4% (4/9) of studies reported statistically positive findings with 1 study reporting clinically meaningful results, a decrease of 3.2 points in the UPDRS-II.⁴⁸

Question 3 For PwPD, is there evidence that bicycling reduces disease severity?

Seventeen of 34 studies reported disease severity. Of these studies, 76% (13/17) reported statistically positive findings and 12 (71%) reported clinically meaningful decreases on the UPDRS-III, ranging from 3.53³⁸ to 22.3³⁴ points.

DISCUSSION

This scoping review describes the evidence of stationary bicycling for PwPD to reduce motor symptoms, BFS impairments, and disease severity and to improve motor performance and activities. Synthesizing the literature based on study design (Tables 4–7) and illustrating it with concept maps (Figure 2) provided an overview of research breadth and evolution as well as illustrated the complexity of this literature. Comparison between PwPD and healthy adults suggested that PwPD responded differently to bicycling. Many of the findings of the effects of bicycling were statistically significant as well as clinically meaningful. Disease severity had the highest frequency of reported clinically meaningful changes, followed by activity measures. BFS findings could not be interpreted for clinical meaning as MCIDs have not been established for those measures.

A scoping review was indicated considering the high degree of heterogeneity of the bicycling literature. Heterogeneity was evident in choices related to study designs, bicycling types, training parameters (e.g. duration and dosing emphasis) and outcomes measured. For example, study designs ranged from single group to RCTs. Bicycling occurred on assisted and unassisted bikes with doses ranging from 45 minutes to 58.5 hours. Outcomes ranged from disease severity with UPDRS scores to body function structure changes such as VO2 max to activities such as balance. A scoping review, in contrast to a systematic review provides a mechanism for examining this heterogeneity. However, careful attention must be paid to the context of each study to interpret the findings.

Further, a scoping review captured the evolution over time of the literature about bicycling for PwPD. The evolution from speed-based studies to aerobic studies illustrates that emphasis has moved away from addressing bradykinesia to focusing on fitness - as indicated by measuring maximum oxygen uptake as a study outcome in addition to disease mitigation.^5,47 As evidenced in the concept maps (see Figure 2), bicycling comparison and single-group studies measured BFS outcomes while RCTs measured more activity outcomes. Disease severity was measured in all studies. Given the aerobic focus of the RCTs, the total training time of those studies was greater than the speed studies.

The overarching goal of this work was to provide guidance to clinicians and scientists regarding the literature on bicycling for PwPD. To accomplish this purpose, we first discuss the evidence related to the effect of bicycling on motor symptom/BFS, activities, and disease. To simplify interpretation for practice and research, we next proposed unifying terminology categorizing bicycling types, training mode and outcomes. Using these analyses, we then discussed implications for practice and research.

Evidence

1: For PwPD, is there evidence that bicycling reduces motor symptoms and BFS impairments?

Findings from this scoping review indicated that BFS improvements were frequently reported in the literature. Individual motor symptoms specific to PD were more often reported in the early literature on assisted bicycling in which speed was the focus of the training. Generally, the findings suggested that speed-based bicycling might decrease individual motor symptoms, primarily bradykinesia and rigidity. The findings were observed regardless of whether the bicycle was assisted or non-assisted. However, it is important to note that in the absence of MCID values, these changes could not be interpreted as clinically meaningful. The authors of these early studies speculated that several mechanisms may explain the findings of these speed-based studies such as an upregulation of central pattern generators (central adaptation)²⁸ and higher-level neural adaptation³⁸ or in the case of bradykinesia as improved motor timing following sustained exercise.⁴⁰

Studies reporting BFS outcomes typically chose outcomes that were the focus of training such as maximum VO₂ for aerobic, strength for force generation, and acceleration or changes in time stride variables of gait for speed studies. Specificity of training modality was supported. For example, significant lower extremity strength outcomes were reported in a force-generation study⁵⁵ while no significant strength improvements were reported in the speed^{28, 38} or aerobic⁵² studies.

Findings from two studies hint that bicycling may have important central nervous system effects for PwPD, reporting un-trained BFS improvements in the UE.^{30, 35} Jansen et al hypothesized that changes in the basal ganglia and improvements in connectivity between cortical and subcortical regions of the brain following exercise could explain these effects on the untrained UE. ³⁰ Importantly BFS outcomes such as cognitive changes and structural brain changes such as brain connectivity and brain-derived neurotrophic factor (BDNF) have also been reported. Notably, BDNF increases were reported in response to tandem (assisted) bicycling in two studies.^{35, 47} Neural imaging findings were consistent with plasticity⁴⁷ specifically, increased connectivity between the primary motor cortex and ipsilateral thalamus was interpreted as strengthening sensorimotor brain regions as a result of training.⁵⁶ Additionally, based on fMRI data it was suggested that acute effects of exercise may share similar pathways to symptomatic relief from medication, substantiating the possibility of greater use of exercise and reduced use of medication.³⁴

2: For PwPD, is there evidence that bicycling improves motor performance and activities?

The report that 50% of activity measures showed clinically meaningful changes supports that bicycling may improve motor performance and activity, primarily for gait. Similar results were reported in recent meta-analysis that included single group studies and RCTs.¹⁰ We interpret this result as specificity of training with respect to the exercise modality. That is, speed training transferred to faster gait speed;⁴⁸ improvements in gait speed exceeded the MCID in speed-based studies.^{26, 39} The link between faster cycling and transfer to gait speed most recently has been specifically shown in a single study.²⁵ For the gait speed outcomes, participants exercised with assisted bicycles with doses ranging from 24²⁶ to 33 hours.³⁹

There also are instances in which there is transfer to an untrained activity. For example, improvements in walking speed were reported with speed-based bicycling.^{26, 28, 36, 38} This suggests that speed-based bicycle training transferred to the untrained task of walking. These results were achieved when speed was the focus of training irrespective of whether the bike was assisted motorized,²⁶ assisted tandem,³⁶ or non-assisted.^{28, 38}

The most perplexing finding was the reporting of statistically improved standing balance scores after bicycling. This appears to be a positive result regardless of the training mode (speed, aerobic or force). In fact, even when bicycling was compared to other types of training, such as treadmill walking⁴² and exergames⁴⁸ positive results, on the mini-BESTest and BESTest were reported for bicycling that were comparable to the training in standing. Transfer to standing balance after training sitting on a bicycle was attributed to a “general fitness effect”⁴⁸ or specifically to improved lower extremity strength and power generation.³⁶

3: For PwPD, is there evidence that bicycling reduces disease severity?

The most consistent finding was the clinically meaningful reduction in disease severity. Seven speed^{21, 27, 29, 30, 34, 35, 38} and 5 aerobic studies^{24, 42, 44, 47, 53} reported clinically meaningful changes in the UPDRS-III with as low as 1 hour (acute effect) in assisted speed-based study³⁴ to 30 hours (long-term effect) of training in a non-assisted aerobic study.⁴² In the randomized controlled trials, bicycling was superior to standard of care,^{24, 47} and comparable to treadmill walking⁴² and tai chi⁴⁴ in producing clinically meaningful reductions in disease severity.

Unifying Terminology and Categorization

This review identified inconsistency in use of terms to describe similar types of bicycling. Consistent reporting on how cycling was dosed speed, aerobic and strength and whether the cycling was voluntary or assisted could facilitate comparisons across studies. For this reason, we proposed new terminology to improve communication across researchers and consumers of the literature. Specifically describing bicycle types as assisted or non-assisted reduced the number of terms (e.g. motorized, actuated, forced) used to describe studies. Categorization of studies by training mode as aerobic, speed, or force-based provided more detail about a training approach and its potential efficacy. Finally, reporting outcome measures according to ICF terminology¹¹ also may facilitate interpretation of study findings. For example, outcomes such as strength,⁵⁵ step length,²⁹ aerobic capacity,⁵ and rigidity were grouped as measures of BFS eliminating the term motor symptoms. Additionally, using ICF terminology to organize outcomes illustrated where the greatest clinically meaningful benefits of cycling were reported, which in this review were greatest for health condition severity and unclear for motor symptoms and BFS. Investigators of future studies are encouraged to use these categories of bicycle type, training mode and ICF-based outcome measures. This consistency in reporting could facilitate interpretation of findings across studies and may enhance application to practice.

Identifying Gaps in the Literature with Applications to Practice & Research

Applications to Practice

Findings from this scoping review, although not definitive, support the use of bicycling in treatment of people with PD. Using a holistic approach, both clinically meaningful and statistically significant findings across all study types may inform clinical practice. Further, the mode of training can be linked to differences in body function structure, activity and disease severity outcomes (See Figure 2). When compared to other active interventions, bicycling was shown to be roughly equivalent at improving gait activities and measures of BFS except for maximum VO₂ which favored the bicycling group. Of importance, when compared to other forms of aerobic exercise, such as exergames or treadmill training, there were remarkably similar outcomes (see Table 7). These findings offer the clinician and the patient options when selecting interventions that may consider patient preferences and availability of equipment. It is worth noting that of the 34 studies only 8 compared different forms of bicycling and 11 compared bicycling to active control groups including standard of care, stretching, and other exercises including exergaming and treadmill walking.

The studies reported in this review used assisted and non-assisted bicycles. This finding has practical implications as assisted bicycles are motorized, expensive and not easily accessible. Non-assisted bicycling, may need to be supplemented with motivational apps,⁵ musical beat, a metronome, verbal cueing and even virtual environments⁴¹ to ensure that participants achieve the desired training cadence or target heart rate. This raises the question of how sustainable training parameters are in practice. While this question has been investigated to some extent, it merits further study.⁵⁷

The studies reported in this review were conducted in different settings. The most common was a laboratory setting, but several investigators used group approaches^{28, 36, 38} that were in some instances community based and did not always require specialized bicycles.^{36 58} Studies are beginning to evaluate the use of a home program ⁵ increasing the options for PwPD to exercise. Several published protocols indicate that more home-based studies are in progress.⁵⁹

Applications to clinical practice may differ based on whether the intervention was performed in the ON or OFF state. Exercising in an ON state is beneficial as participants exercise closer to their maximum capacity and is consistent with the guidelines for exercise of PwPD. It is worth noting that cycling in the OFF state, did show some short-term benefits of bicycling for improving motor symptoms and disease severity.^{33, 34, 37} These short-term improvements in disease severity and symptoms may be helpful for participants experiencing OFF states during gaps between medications.

Applications to Research

In addition to extending the training to the home there are other gaps in knowledge raised by this review. Durability of the findings was lacking as most of the studies included in this review did not have follow-up measures. Further, questions about adherence, motivation, and preference were not consistently reported. Bicycling has been studied both in an individual and group setting, and it was unclear which is better. There are several other un-answered questions about which are the best training parameters as the studies reported here had a wide range of training durations and intensities. It is also not clear what mode of bicycling (continuous or interval) will yield a better result. A definitive comparison between assisted (forced) and unassisted (voluntary) bicycling is currently being tested in the Cyclical Lower Extremity Exercise (CYCLE) trial.⁶⁰ The contributions of cycling to CNS plasticity also merit further study. Given the surprising improvements in balance more work is required to replicate and explain the findings and to determine if those balance improvements in any way reduce fall risk. Finally, investigators are encouraged to use consistent language facilitating literature synthesis and interpretation across studies.

Given the importance of medication state on outcome assessment scores, it is notable that six studies did not report whether outcomes were measured in the ON or OFF state.^{38, 39, 41, 44, 50, 52} While measurement in an OFF state may be preferable as it can help control for an unpredictable response to medications,⁵ measures of cardiovascular fitness (VO2 max) are better recorded in the ON state allowing participants to reach their full fitness capacity in a graded maximal exercise test.^{5, 21} Therefore, the decision to assess outcomes in the ON or OFF state may depend on the outcome being measured. Regardless, authors are encouraged to provide justification for measuring ON or OFF to help readers better understand and interpret findings.

Limitations

The study has several limitations. Following the scoping review guidelines, we did not appraise the quality of the studies.¹² Therefore, we retained two studies that might have been excluded based on quality.^{31, 43} This review focused only on motor performance, and non-motor benefits of cycling (e.g. cognitive or affective) are also important to consider. Finally adopting a holistic approach to this scoping review, we included all study designs. As many of the studies were not RCTs, causal inference of outcome improvement needs to be made with caution. However, the findings were not substantially changed if the single groups studies were eliminated from the interpretation.

CONCLUSIONS

This review reported that bicycling as a tool for rehabilitation for PwPD, with a mean age of 58–72, HY 1–3 and mean disease duration of 3–8 years, has been studied using different types of bicycles (assisted and non-assisted) as well as training modes (speed, aerobic and force). The positive improvements for PwPD in motor symptoms, primarily in speed-based studies and reduction in BFS impairments (in both speed and aerobic studies) could not be interpreted as clinically meaningful. Improved activities, in particular gait, were seen in both speed and aerobic studies. When compared to other forms of aerobic training, bicycling yielded comparable results for walking speed and endurance. The most consistent clinically meaningful finding was the reduction in disease severity. The scoping review offered the terms assisted (cycling cadence is achieved by another external force applied to the pedals) and non-assisted (cadence achieved solely from cyclist’s voluntary effort) as a way to unify the terminology in this literature. Further it categorized training mode based on how the intervention was dosed as speed, aerobic, or force. This permitted a more consistent reporting of the evidence, which could be incorporated in future synthesis of the literature to facilitate organization and interpretation of findings. Training on assisted bicycles may be difficult to transfer to practice based on their cost; however, training on non-assisted bicycles may require supplementation with coaching, cues or some form of external pacing. Durability of the findings, details on bicycling mode (continuous and interval) and dose remain to be elucidated.

Supplementary Material

1

Supplemental Digital Content # 1: Search strings for each database.

Supplemental Digital Content # 2: Tables 1–3

Table 5: Bicycling Comparison Studies.

• Studies comparing different types or modes of bicycling organized using a PICO format. For all studies, the comparison group was dose-matched for total time and used the same bicycle as the intervention group.

Study	Population 1	Bicycle (Mode & Type)	Intervention	Comparison	Motor Symptoms	BFS	UPDRS-III	Activities
Ridgel ☆ 200921	N = 10 (5,5) Age = 58 (2) UPDRS-III = 48 (13) Disease Duration 8 (7)	Assisted (tandem) Custom, upright Non-Assisted Non-custom, upright	Forced Exercise (FE) Cycling HR: 60–80% HRR (graded) RPM: 80–90 RPM or 30% > voluntary 1 hour × 3 × 8 weeks Total time: 24 hours	VE Cycling HR: 60–80% HRR (graded) RPM: self-selected	FE VE Bradykinesia ↓ NS Rigidity ↓ NS Tremor ↓ NS	FE VE Dexterity ↑ NS VO2 Max NS NS Measured in the on state	FE VE ↓▲0 NS
Ridgel☆ 201133	N = 32 (20,12) Age = 63 (9) H&Y = 2 (0.8) Disease Duration 5 (3)	Assisted (actuated) Non-custom, recumbent	Passive Cycling ☆ RPM: (60, 70, or 80 RPM) 40 minutes × 1 × 3 weeks Total time: 2 hours		Bradykinesia NSD Tremor NSD
Ridgel 201527	N = 47 (24,23) Age = 67 (2) H&Y = 2 (0.2) Disease Duration 7 (1)	Assisted (actuated) Custom, recumbent Non-Assisted Custom, recumbent	Dynamic Cycling HR: 50–80% HRR RPM: 75–85 RPM 50 minutes × 3 × 1 week Total time: 2.5 hours	Static Cycling HR: 50–80% HRR RPM: self-selected	Dynamic. Static Rigidity ↓ NS LE ↓ NS UE ↓ NS		Dynamic Static ↓ ▲ NS	Dynamic Static TUG NS NS
Miller-Koop 201931	N = 59 (31,28) Age = 63 (8) H&Y = 2 (0.4) Disease Duration 3 (4)	Assisted (actuated) Custom, recumbent Non-Assisted Custom, recumbent	FE Cycling HR: 60–80% HRR RPM: 30% > voluntary 50 minutes × 3 × 8 weeks Total time: 20 hours	VE Cycling HR: 60–80% HRR RPM: self-selected		FE VE Acceleration Unclear from reporting of results		FE VE TUG NS NS ↓ Turn Vel ↑ > ↑
Stuckenschneider 201526	N =20 (10,10) Age = 71 (5) H&Y = 3 (2.5–4) Disease Duration NR	Assisted (actuated) Non-custom, recumbent Assisted (actuated) Non-custom, recumbent	Active Assisted Cycling RPM: 30% > max (cap at 90 rpm) 40 minutes × 3 × 12 weeks 24 hours	Passive Cycling RPM: same (30% > max)	Active Passive Tremor ↓ NS	Active Passive Stride Length ↑ ↑	Active Passive NS NS	Active Passive Gait Speed ↑▲ ↑▲
Penko 202132	N=90 (35,35,20) Age: 63 (8) UPDRS -III = 36 (9) Disease Duration 3 (1–4)	Assisted (actuated) Custom, recumbent Non-Assisted Custom, recumbent	FE Cycling HR: 60–80% HRR RPM: 35% > voluntary 55 minutes × 3 × 8 weeks Total time: 22 hours	VE Cycling HR: 60–80% HRR RPM: self-selected		FE VE VO2 Max ↑ NS ↑
Jansen☆ 202130	N=29 (15,14) Age: 64 (6) UPDRS -III = 33 (10) Disease Duration 4 (5)	Assisted (actuated) Custom, recumbent Non-Assisted Custom, recumbent	FE Cycling HR: 60–80% HRR RPM: 35% > voluntary 55 minutes × 3 × 8 weeks Total time: 22 hours	VE Cycling HR: 60–80% HRR RPM: self-selected		FE VE Dexterity ↑ > ↑	FE VE ↓▲◇ NS ↓▲◇
Uygur 201428	N=10 Age: 65 (6) H&Y: 2 (0.7) Disease Duration 4 (4)	Non-Assisted Non-custom, recumbent Non-Assisted Non-custom, recumbent	High-Speed Low Resistance (HSLR) Interval Cycling RPM : self-selected fast 30 minutes × 1 × 2 weeks Total time: 1 hour	Preferred Cadence (PC) Continuous Cycling RPM: self-selected comfortable		HSLR PC Grip Strength NS NS		HSLR PC 4SST ↓ NS TUG NS NS Gait Speed ↓▲ NS 9HPT NS NS Shirt Button NS NS

¹the last entry of the population cell indicates disease duration in years; ℥, Both groups used same bike; ↑, within group improvement (+); ↓, within group worsening (−); ↓ = within group improvement (−); △, exceeds MDC; ^{19, 20} ▲ exceeds MCID; ^{16, 17} <, between group favoring right group; >, between group favoring left group; ◇, decrease in severity in UPDRS-3 (eg. moderate to mild); ☆, outcome measured or intervention performed in the OFF state; 4SST, 4 square test; 9HPT, 9 hole peg test; Act, active-assisted cycling; Age, measured in years; BFS, body function structure; Brady, bradykinesia; DM, dose-matched; Dyn, dynamic cycling; FE, forced exercise; H&Y, Hoehn & Yahr score; HR, heart rate; HRR, heart rate reserve; HSLR, high-speed low-resistance; LE, lower extremity; NS, no significant difference; Pass, passive cycling; PC, preferred cadence; RPM, revolutions per minute; Stat, static cycling; TUG, timed up and go; Turn Vel, turning velocity; UPDRS-III, Unified Parkinson’s Disease Rating Scale Section 3 (motor exam); UE, upper extremity; VE, voluntary exercise; VO2, maximal oxygen consumption

Table 6: Single Group Studies.

• Studies using single group pre-post design organized using a PIO format.

Study	Population1	Bicycle (Mode & Type)	Intervention	Motor Symptoms	BFS	UPDRS -III	Balance	Gait & Activities
Speed
Ridgel☆ 201237	n=10 Age: 64(2) H&Y: 1.8 (0.3) Disease Duration 6.5 (1.9)	Assisted (Actuated) Non-custom NR	Continuous☆ 80–85 rpm; motor at 75 rpm 40 minutes × 1 Total time: 40 minutes	Bradykinesia ↓ Tremor NS
Alberts☆ 201634	n=9 Age: 61 (10) H&Y: NR Disease Duration 3.4 (1.8)	Assisted (Tandem) Custom Upright	Continuous☆ 80–90 RPM; 65–80% MaxHR 1 hour × 1 × 1 week Total Time: 1 hour			↓▲
McGough 201636	n=41 Age: 63 (9) H&Y: 1.9 (0.2) Disease Duration 5.4(5.3)	Assisted (Tandem) Custom Recumbent/ Upright	Continuous 80–90 RPM 1 hour × 3 × 10 weeks Total Time: 25 hours		Cadence (ss) ↑ Cadence (f) NS 5×STS ↓		SPPB ↑ BBS ↑ TUG ↓	Gait Speed (ss) ↑ Gait Speed (f) NS
Marusiak ☆ 201535	n=11 Age: 71 (10) H&Y=2.3 (0.7) Disease Duration 8(4)	Non-Assisted Non-custom Upright	Interval (HIIT) 80–90 rpm (cool-down: 60 rpm) 1 hour × 3 × 8 weeks Total Time: 24 hours	Rigidity ↓	Stiffness (Biceps) ↓ Frequency (Biceps ↓	↓▲
Uygur 201738	n=14 Age: 63 (9) H&Y: 2.5 (0.6) Disease Duration 3.3(2.4)	Non-Assisted Non-custom Recumbent	Interval (HIIT) Fastest speed possible 30 minutes × 2 × 6 weeks Total Time: 6 hours	Bradykinesia ↓	Grip Strength NS Steps in 10mWT ↓	↓ ▲	FRT ↑ 4SST ↓ ABC ↑ TUG ↓	Gait Speed ↓▲ - - - - - - 9HPT ↓
Chang☆ 201829	n=13 Age: 59 (7) H&Y: 1.8; 2.0☆ Disease Duration 6.44 (4.0)	Non-Assisted Non-custom NR	Continuous 40 RPM min, max HR: 50–55% (15–40 minutes) 2 × 8 weeks (graded) Total Time: 8.5 hours	Rigidity NS Tremor NS Akinesia ↓	Step Length NS Step Width NS Step Time NS Double Limb Support NS	↓ ▲	TUG ↓	Gait Speed ↑
Aerobic
Lauhoff 201353	n=23 Age: 71(7) H&Y: 2 (0.8) Disease Duration 6.4 (5.2)	Non-Assisted Non-custom NR	Continuous 60–80% max HR ~40 minutes × 1 × 6 weeks Total time: ~ 4 hours		Max HR NS Average HR NS	↓▲	BBS ↑ TUG ↓	6MWT NS Gait Efficiency NS - - - - - - - - UPDRS-II ↓
Haas 201652	n=6 Age: 63 (7) H&Y: 2(2) Disease Duration NR	Non-Assisted Non-custom Upright	Interval HRpeak/ HRmax 30 minutes × 6 sessions in 3 weeks Total Time: 3 hours		Quad Strength NS Hamstring Strength NS VO2Max NS Cycle Endurance ↑		TUG ↓
Force
Dibble 200655	n=10 Age: 64 (10) H&Y: 2.5 (0.5) Disease Duration 6.1(3.9)	Non-Assisted Custom Recumbent	Continuous RPE 9–15 (graded) 3×12 weeks Total Time: NR		Work ↑ Quad Strength ↑

¹the last entry of the population cell indicates disease duration in years; ↑, improvement (+); ↓= Improvement (−); △, exceeds MDC; ^{19, 20} ▲ exceeds MCID; ^16–18 ◇, decrease in severity in UPDRS-3 (eg. moderate to mild); ☆, outcome measured or intervention performed in the OFF state; 4SST, 4 square test, 5×STS, 5 time sit to stand test; 6MWT, 6 minute walk test; 9HPT, 9 hole peg test; ABC, activities-specific balance confidence scale; Age, measured in years; BBS, Berg Balance Scale; BFS, body function structure; FRT, functional reach test; H&Y, Hoehn & Yahr score; HIIT=High-Intensity Interval Training; HR, heart rate; NR, not-reported; NS, no significant difference; Quad, quadriceps muscle; RPE, rating of perceived exertion; RPM, revolutions per minute; SPPB, short physical performance battery; TUG, timed up and go; UPDRS-II, Unified Parkinson’s Disease Rating Scale Section 2 (motor experiences of daily living); UPDRS-III, Unified Parkinson’s Disease Rating Scale Section 3 (motor exam); VO2, maximal oxygen consumption