Long-term Effect of Regular Physical Activity and Exercise Habits in Patients With Early Parkinson Disease

Abstract

Background and objectives: Owing to the lack of long-term observations or comprehensive adjustment for confounding factors, reliable conclusions regarding long-term effects of exercise and regular physical activity in Parkinson disease (PD) have yet to be drawn. Here, using data from the Parkinson's Progression Markers Initiative study that includes longitudinal and comprehensive evaluations of many clinical parameters, we examined the long-term effects of regular physical activity and exercise habits on the course of PD.

Methods: In this retrospective, observational cohort study, we primarily used the multivariate linear mixed-effects models to analyze the interaction effects of their regular physical activity and moderate to vigorous exercise levels, measured with the Physical Activity Scale for the Elderly questionnaire, on the progression of clinical parameters, after adjusting for age, sex, levodopa equivalent dose, and disease duration. We also calculated bootstrapping 95% confidence intervals (CIs) and conducted sensitivity analyses using the multiple imputation method and subgroup analyses using propensity score matching to match for all baseline background factors.

Results: Two hundred thirty-seven patients with early PD (median [interquartile range] age, 63.0 [56.0-70.0] years, male 69.2%, follow-up duration 5.0 [4.0-6.0] years) were included. Regular physical activity and moderate to vigorous exercise levels at baseline did not significantly affect the subsequent clinical progression of PD. However, average regular overall physical activity levels over time were significantly associated with slower deterioration of postural and gait stability (standardized fixed-effects coefficients of the interaction term [β_interaction] = -0.10 [95% CI -0.14 to -0.06]), activities of daily living (β_interaction = 0.08 [95% CI 0.04-0.12]), and processing speed (β_interaction = 0.05 [95% CI 0.03-0.08]) in patients with PD. Moderate to vigorous exercise levels were preferentially associated with slower decline of postural and gait stability (β_interaction = -0.09 [95% CI -0.13 to -0.05]), and work-related activity levels were primarily associated with slower deterioration of processing speed (β_interaction = 0.07 [95% CI 0.04-0.09]). Multiple imputation and propensity score matching confirmed the robustness of our results.

Discussion: In the long term, the maintenance of high regular physical activity levels and exercise habits was robustly associated with better clinical course of PD, with each type of physical activity having different effects.

Trial registration information: ClinicalTrials.gov Identifier: NCT01176565.

Classification of evidence: This study provides Class II evidence that sustained increase in overall regular physical activity levels in patients with early PD was associated with slower decline of several clinical parameters.

Figure 1. Temporal Changes in Overall Regular Physical Activity Levels and Proportion of Participants With Appropriate Exercise Habits

Linear regression lines showing temporal changes in overall regular physical activity levels quantified by the total score of the Physical Activity Scale for the Elderly (PASE) questionnaire (A) and line graphs showing temporal changes in the percentage of participants meeting the recommendation from quality metrics published by the American Academy of Neurology (AAN) (B). Note that in the control group, the PASE total score appears to have an increasing trend over time, but this trend did not reach statistical significance. Gray areas represent the 95% confidence intervals of the regression lines. PD = Parkinson disease.

Figure 2. Summary of the Interaction Effect of Each Regular Physical Activity Level on the Decline of Each Function in Patients With PD

Heatmaps showing the degree of interaction effect of the overall level of regular physical activity and level of different types of physical activity on the progression of each clinical parameter, as determined from the t value of the fixed-effects interaction term in our multivariate linear mixed-effects model. Note that there were no statistically significant interaction effects between the baseline regular physical activity levels and progression of any clinical parameters (A). However, the average regular physical activity levels over the follow-up period had statistically significant interaction effects on the temporal progression of several clinical parameters (B). ESS = Epworth Sleepiness Scale; GDS = Geriatric Depression Scale; HVLT = Hopkins Verbal Learning Test; JLO = Judgment of Line Orientation; LNS = Letter-Number Sequencing; MDS-UPDRS = Movement Disorders Society–sponsored revision of the Unified Parkinson's disease rating scale; MOCA = Montreal Cognitive Assessment; MSE-ADL = Modified Schwab & England Activities of Daily Living scale PASE = Physical Activity Scale for Elderly; PD = Parkinson disease; PIGD = Postural Instability and Gait Disturbance; RBDSQ = REM Sleep Behavior Disorder Screening Questionnaire; SCOPA-AUT = Scales for Outcomes in Parkinson's Disease–Autonomic Dysfunction; SDMT = Symbol Digit Modalities Test. *Significant association after the Bonferroni correction (Bonferroni-corrected p < 0.05).

Figure 3. Distribution of Propensity Scores and Balance Measures After Propensity Score Matching

At first, patients with Parkinson disease (PD) were dichotomized according to the median of the average Physical Activity Scale for the Elderly (PASE) total score. After propensity score matching, both the higher and lower regular physical activity groups consisted of 86 patients with PD (A) and were matched such that standardized mean differences between all background factors fell within a strict cutoff of 0.1 (B). ESS = Epworth Sleepiness Scale; GDS = Geriatric Depression Scale; HVLT = Hopkins Verbal Learning Test; JLO = Judgment of Line Orientation; LNS = Letter-Number Sequencing; MDS-UPDRS = Movement Disorders Society–sponsored revision of the Unified Parkinson's disease rating scale; MOCA = Montreal Cognitive Assessment; MSE-ADL = Modified Schwab & England Activities of Daily Living scale; PIGD = Postural Instability and Gait Disturbance; RBDSQ = REM Sleep Behavior Disorder Screening Questionnaire; SCOPA-AUT = Scales for Outcomes in Parkinson's Disease–Autonomic Dysfunction; SDMT = Symbol Digit Modalities Test.

Figure 4. Interaction Effects of Different Type of Regular Physical Activity Levels on Declines in Postural and Gait Function, ADL, and Processing Speed After Propensity Score Matching

In propensity score–matched groups with higher and lower overall levels of regular physical activity, we plotted temporal changes in Movement Disorders Society–sponsored revision of the Unified Parkinson's Disease Rating Scale (MDS-UPDRS) Postural Instability/Gait Disturbance (PIGD) subscore (A) and Modified Schwab & England Activities of Daily Living (MSE-ADL) score (B). We also plotted temporal changes in the MDS-UPDRS PIGD subscore in propensity score–matched groups with higher and lower moderate to vigorous exercise levels (C) and in the Symbol Digit Modalities Test (SDMT) score in propensity score–matched groups with higher and lower work-related activity levels (D). Note that the temporal changes in these scores were visually and statistically different between 2 groups. Solid lines represent regression lines, and shaded areas represent 95% confidence intervals of regression lines.