Promotion of physical activity and fitness in sedentary patients with Parkinson’s disease: randomised controlled trial

Abstract

Objective To evaluate whether a multifaceted behavioural change programme increases physical activities in patients with Parkinson’s disease.

Design Multicentre randomised controlled trial.

Setting 32 community hospitals in the Netherlands, collaborating in a nationwide network (ParkinsonNet).

Participants 586 sedentary patients with idiopathic Parkinson’s disease aged between 40 and 75 years with mild to moderate disease severity (Hoehn and Yahr stage ≤3).

Intervention Patients were randomly assigned to the ParkFit programme or a matched general physiotherapy intervention. ParkFit is a multifaceted behavioural change programme, designed specifically to achieve an enduring increase in the level of physical activity (coaches using motivational strategies; ambulatory feedback).

Main outcome measures The primary endpoint was the level of physical activity, measured every six months with a standardised seven day recall (LASA physical activity questionnaire—LAPAQ). Secondary endpoints included two other measures of physical activity (activity diary and ambulatory activity monitor), quality of life (Parkinson’s disease questionnaire—PDQ-39), and fitness (six minute walk test).

Results 540 (92.2%) patients completed the primary outcome. During follow-up, overall time spent on physical activities (LAPAQ) was comparable between the groups (adjusted group difference 7%, 95% confidence interval −3 to 17%; P=0.19). Analyses of three secondary outcomes indicated increased physical activity in ParkFit patients, as suggested by the activity diary (difference 30%; P<0.001), the activity monitor (difference 12%; P<0.001), and the six minute walk test (difference 4.8 m; P=0.05). PDQ-39 did not differ between ParkFit patients and controls (difference −0.9 points; P=0.14). The number of fallers was comparable between ParkFit patients (184/299; 62%) and controls (191/287; 67%).

Conclusions The ParkFit behavioural change programme did not increase overall physical activity, as measured with the LAPAQ. The analysis of the secondary endpoints justifies further work into the possible merits of behavioural change programmes to increase physical activities in daily life in Parkinson’s disease.

Trial registration Clinical trials NCT00748488.

Introduction

Parkinson’s disease is a common neurodegenerative disease, characterised by motor symptoms and a wide variety of non-motor symptoms such as depression and apathy.^{1 2} Despite optimal medical treatment, it remains a progressive disease that negatively affects quality of life. Therefore, allied health interventions are increasingly deployed to treat both the motor and non-motor symptoms of Parkinson’s disease. The evidence to support the merits of these interventions is growing, and treatment guidelines (based partially on evidence and partially on practical clinical experience) for several allied healthcare interventions have been developed.^{3 4 5}

In recent years, several physiotherapy programmes have been tested in patients with Parkinson’s disease.^{6 7 8 9 10} Reviews and meta-analyses have generally found evidence to support “exercise” as being beneficial with regard to physical functioning, strength, balance, and gait speed.^{11 12 13 14 15} However, the physiotherapy programmes as tested in these studies were apparently insufficient to achieve an active lifestyle. Because of their combined physical limitations and mental changes, many patients with Parkinson’s disease lead a sedentary lifestyle.¹⁶ Reversing this lifestyle could have generic health benefits, including increased survival.^{17 18 19} Promoting physical activity may also improve specific symptoms of Parkinson’s disease, such as insomnia, depression, or constipation.²⁰ Furthermore, work in rodents suggests that physical activity might counter neurodegeneration in experimental parkinsonism.^{21 22}

An individually tailored, disease specific training programme is needed to improve physical activity in Parkinson’s disease.²³ We developed such an intervention (the ParkFit programme²⁴) based on models of behavioural change and containing established behavioural change techniques.^{25 26 27 28 29} To evaluate this programme, we designed a randomised controlled trial comparing ParkFit with a matched control intervention.²⁴

Methods

The ParkFit trial is a multicentre randomised controlled trial to increase physical activity levels over the course of two years in sedentary patients with Parkinson’s disease. The study design has been detailed elsewhere.²⁴

Study participants

Recruitment ran from September 2008 to January 2010. Patients treated in 32 community hospitals were invited to participate. Eligibility criteria were Parkinson’s disease according to UK Brain Bank criteria¹; age 40-75 years; sedentary lifestyle, defined as participation in physical activity of vigorous intensity less than three times a week and for less than 60 minutes in total per week or participation in moderate intensity physical activity less than three times a week and for less than 150 minutes in total per week²³; and Hoehn and Yahr stage 3 or lower. Exclusion criteria were mini-mental state examination score less than 24/30, inability to complete Dutch questionnaires, co-morbidity that interfered with daily functioning, daily institutionalised care, and previous deep brain surgery. All participants gave informed consent before the first assessment.

Study outcomes

Baseline characteristics

We scored disease stage according to the modified Hoehn and Yahr scale and assessed motor function by using the unified Parkinson’s disease rating scale (UPDRS part III, motor examination).

Primary endpoint

We made several amendments to the endpoints in the initial phase of the study, at a time when recruitment was under way for only two months. We here report our final selection of endpoints, as specified on ClinicalTrials.gov, in the adapted final research protocol that was accepted by the Ethical Committee (CMO) Arnhem Nijmegen, and in a recent design article.²⁴

The primary endpoint was the LASA physical activity questionnaire (LAPAQ)—a validated, interview based, seven day recall of physical activities. The LAPAQ was highly correlated with a seven day diary (r=0.68; P<0.001) and moderately correlated with a pedometer (r=0.56; P<0.001).³⁰ It asks patients about their daily amount of specific activities, allowing for calculation of total time spent on physical activities (expressed in hours per week). The LAPAQ covers the frequency and duration of the net sum of the following activities: walking outdoors, cycling, gardening, light and heavy household activities, and sport activities.³⁰ Consequently, higher scores on the LAPAQ (in hours per week) indicated more time spent on physical activity. We measured LAPAQ at baseline and after 6, 12, 18, and 24 months. At baseline and after 12 and 24 months, LAPAQ was completed during face to face interviews; after six and 18 months, it was completed by telephone. We assumed that patients would increase their level of physical activity during the first months of the intervention and would then maintain this level. Therefore, the main endpoint was the average of the level of physical activity during the entire follow-up period (that is, the average of 6, 12, 18, and 24 months). This approach has several advantages. Firstly, it provides a global assessment of the results of the intervention. Secondly, it provides maximal power. As the number of assessments taken into account increases, so does the power. We did not compare all individual time points (at 6, 12, 18, and 24 months) separately, because this leads to multiplicity.

Secondary endpoints

We defined four secondary endpoints²⁴: physical fitness, as measured with the six minute walk test at 12 and 24 months (average of all measurements)³¹; quality of life, as measured with the Parkinson’s disease questionnaire (PDQ-39³²) at 6, 12, 18, and 24 months (the average of all measurements); physical activity, measured subjectively every six months with a seven day activity diary (average of all measurements);³³ and physical activity, measured objectively every six months with an ambulatory activity monitor (average of all measurements).³⁴

The diary detailed the frequency and total duration (hours/week) spent on five specific activities: walking outdoors for more than 10 contiguous minutes, moderate intensity cycling for more than 10 contiguous minutes, high intensity cycling for more than 10 contiguous minutes, sport activities, and other strenuous activities (such as cutting wood). Participants wore the activity monitor (triaxial accelerometer³⁴) as a necklace, on the belt, or in the pocket. Data were collected during waking hours for 14 days and were stored minute by minute for each axis; output was expressed in kilocalories/minute. We included only completely observed days in the analysis.³⁵ Patients allocated to the ParkFit programme additionally used the monitor as a feedback tool by using light emitting diodes that reflected the amount of actually delivered daily physical activity. Control patients received no feedback from their activity monitor.

Safety and falls

We assessed safety by spontaneous reports of adverse events. We classified serious adverse events as events that caused death, were life threatening, or necessitated admission to hospital. We monitored falls monthly with an automated telephone system.³⁶ Information about adverse events was additionally collected at each physical assessment.

Intervention

After baseline assessment, patients were randomly assigned to either the ParkFit programme or a matched physiotherapy intervention aimed at safety of movements. The investigators logged on to a protected website and entered the region, Hoehn and Yahr stage, age, sex, and current physical activity level of the patients. On the basis of a minimisation algorithm with these factors, the treatment was allocated and registered. Before inclusion, patients were informed that the trial compared two potentially beneficial interventions. We used “active” names for both interventions (the “ParkFit” and “ParkSafe” programmes). To ensure blinding, trained assessors who were unaware of group allocation examined patients. Patients were instructed not to discuss the nature of their physiotherapy with the assessors.

Experienced physiotherapists in the Dutch ParkinsonNet delivered both interventions.³⁷ In total, 154 physiotherapists were trained to deliver both interventions. This ascertained that differences in the personality or style of the physiotherapists could not bias the results. All patients were offered an equal maximum number of treatment sessions (35/year). The full study protocol has been described elsewhere.²⁴

ParkFit programme

We designed the ParkFit programme specifically to achieve a sustained increase in physical activity levels, on the basis of theories and models of behavioural change and on effective behavioural change techniques.^{25 26 27 28 29} Important elements were activity coaches who guided each patient towards a more active lifestyle during monthly personal coaching sessions; an educational brochure about the benefits of physical activity and suitable activities for patients with Parkinson’s disease; identifying and overcoming any perceived barriers to engaging in physical activity; systematic goal setting, using a health contract and logbook; stimulation to participate in group exercises; and an ambulatory monitor with automated feedback reflecting actually delivered physical activities.³⁴ Ambulatory monitor data were uploaded to a personalised website, where both the patient and coach could monitor progress.

The ParkFit programme also included regular physiotherapy sessions. On the basis of individual disabilities, the therapist and patient jointly formulated individually tailored aims of treatment aims, according to the evidence based guideline for physiotherapy in Parkinson’s disease.³

Control intervention

The control intervention consisted of a general physiotherapy programme aimed at promoting safety of movements, according to the evidence based guideline.³ Patients received an identical brochure to ParkFit patients, but with information about the benefits of physiotherapy and safety of movements. Patients were offered a maximum number of treatment sessions, similar to the ParkFit programme. An active lifestyle was not explicitly stimulated. Treatment aims were jointly formulated by therapist and patient, on the basis of perceived individual disabilities.

Statistical analysis

The main endpoint was the physical activity level during the entire follow-up (6, 12, 18, and 24 months). Because the physical activity level was skewed, we present medians and interquartile ranges and did analyses after logarithmic transformation. We evaluated differences between the two interventions by using a linear mixed model with random nested factors of “patient” and “exercise group.” We included region, Hoehn and Yahr stage, age, sex, and current physical activity level of the patient as covariables. We analysed results according to a modified intention to treat principle, whereby we excluded only patients with no follow-up measurements at all.

Sample size calculation

On the basis of the following power considerations, we aimed to include a total of 700 patients. In a small observational study on physical activity in Parkinson’s disease, patients scored 45% less than controls on the LAPAQ (unpublished data). The coefficient of variation was 110%. On the basis of a difference of 20% in hours per week (with a coefficient of variation of 110%) between the two treatment arms, the mixed model analysis will have at least 80% power (when the correlation between baseline and follow-up measurements is at least 0.50 and when the correlation between the various follow-up measurements is at most 0.75). The decision to define a 20% increase based on the LAPAQ activity as a clinically relevant difference was a pragmatic choice, because no earlier intervention studies had aimed to change activity behaviour in Parkinson’s disease patients. Moreover, previous behavioural change studies in other diseases (such as heart failure, diabetes, and chronic obstructive pulmonary disease) did not include the LAPAQ as an endpoint. In an earlier study by our group,¹⁶ we found that patients with Parkinson’s disease were 29% less active than controls (as measured with the LAPAQ): patients spent 12.9 hours per week on physical activity, whereas controls spent more than 17.5 hours. We deemed an increase in physical activity among Parkinson’s disease patients of more than four hours to be unrealistic and reasoned that an increase of two hours per week (that is, an increase of about 20%) would be feasible. We also considered a two hour increase in physical activity to be clinically relevant, for the following reasons. A dose-response relation exists between physical activity and cardiovascular disease or premature mortality.³⁸ Significant risk reductions have been observed with 45-150 minutes a week of brisk walking.³⁹ Additionally, women who walked or exercised vigorously for at least 2.5 hours a week had a 30% lower risk of coronary heart disease.³⁹ Conversely, the risk of cardiovascular disease was higher among women who spent more than 12 hours a day lying down or sleeping.³⁹ This suggests that a two hour increase in physical activities might help to prevent cardiovascular disease. The power is based on two sided 95% confidence intervals. We assumed that the clustering due to the fact that the intervention was carried out in training groups of approximately eight patients leads to an intraclass correlation coefficient of 0.1. On the basis of a previous trial of physical therapy in Parkinson’s disease,³⁷ we expected a dropout rate of 10%.

Results

Baseline characteristics

We included 586 patients (figure); 299 patients were randomly assigned to the ParkFit programme and 287 to the control intervention. The groups had comparable demographic and disease characteristics, although ParkFit patients tended to be less active than controls in daily life (that is, spent less time spent on physical activity according to LAPAQ) (table 1).

Screening, randomisation, and completion of primary outcome measure. *Vigorous intensity physical activity >3 times/week and >60 minutes/week or moderate intensity physical activity >3 times/week and >150 minutes/week. †Hoehn and Yahr stage >3, mini-mental state examination <24/30, severe comorbidity interfering with daily functioning, daily care in institution, or deep brain stimulation

Table 1.

 Baseline characteristics

	ParkFit (n=299)	Controls (n=287)
Demographics and clinical characteristics
Mean (SD) age (years)	65.1 (7.9)	65.9 (7.2)
Male sex	194 (65)	188 (65)
Mean (SD) body mass index	27.4 (4.5)	27.6 (4.0)
Mean (SD) disease duration (years)	5.0 (4.5)	5.5 (4.6)
Mean (SD) mini-mental state examination score	28.1 (1.7)	28.1 (1.7)
Modified Hoehn and Yahr stage (No (%)):
1	7 (2)	4 (1)
1.5	7 (2)	10 (3)
2	221 (74)	223 (78)
2.5	48 (16)	36 (13)
3	16 (5)	14 (5)
Mean (SD) unified Parkinson’s disease rating scale, part III	33.1 (11.3)	32.3 (9.5)
Mean (SD) daily levodopa equivalent dose (mg)	458 (362)	499 (414)
Level of physical activity
Median (interquartile range) LAPAQ (hours per week)	12.8 (8.3-20.3)	13.8 (8.3-23.9)

LAPAQ = LASA physical activity questionnaire.

Loss to follow-up

In total, 540 (92.2%) of the 586 participants completed the LAPAQ after 24 months. The proportion of patients lost to follow-up was comparable for ParkFit (8.7%) and controls (6.7%). Patients lost to follow-up were similar to those who completed the assessments, except for a higher age.

Compliance

Seventy five (12.8%) of the 586 participants did not complete the two year intervention (ParkFit n=44, control n=31). The main reasons were refusal to change from a regular physiotherapist to a ParkinsonNet physiotherapist, too great a burden, or dissatisfaction with the intervention. Reasons for drop out were similar between the two groups. The mean number of annual individual visits to the physiotherapist did not differ between ParkFit (13.6) and control patients (13.0). Patients in both groups were satisfied with the intervention and would recommend the intervention to others (186 (73%) ParkFit v 182 (71%) controls).

Endpoints

Primary endpoint

Compared with baseline, the overall time spent in physical activities was comparable between the two groups (adjusted group difference 7%, 95% confidence interval −3% to 17%; P=0.19) (table 2).

Table 2.

 Effect of intervention (in hours/week) on level of physical activity measured with LASA physical activity questionnaire (LAPAQ; primary outcome)

Time point	ParkFit			Controls		Estimated difference (95% CI)*	P value
	No	Median (IQR)		No	Median (IQR)
Baseline	299	12.8 (8.3-20.3)		287	13.8 (8.3-23.9)	7% (−3% to 17%)	0.19
6 months	285	13.2 (9.2-20.5)		277	14.2 (8.5-22.0)
12 months	281	12.5 (7.2-21.1)		277	12.4 (7.3-17.9)
18 months	277	12.3 (7.0-19.0)		271	12.3 (6.8-19.1)
24 months	273	12.5 (6.3-18.4)		267	12.0 (7.0-18.3)

IQR=interquartile range.

*Estimated relative difference, based on mixed model analysis.

Secondary endpoints

Both the activity diary and the activity monitor data suggested increased levels of physical activity in ParkFit patients (table 3). Additionally, ParkFit patients increased their physical fitness compared with controls (4.8 m, 0.1 to 9.6; P=0.05). Quality of life did not differ between the groups (−0.9 points, −2.1 to 0.3; P=0.14).

Table 3.

 Effect of intervention on secondary outcome measures

	ParkFit			Controls		Estimated difference (95% CI)*	P value
	No	Effect of intervention		No	Effect of intervention
Activity diary—median (interquartile range) hours/week
Baseline	297	5.5 (3.1-10.3)		282	6.3 (3.3-10.5)
6 to 24 months	276	7.6 (4.7-12.4)		276	6.9 (4.2-10.8)
Mean change	275	1.3		273	0.5	30% (17% to 45%)	<0.001
Activity monitor—median (interquartile range) kcal/day
Baseline	273	453 (368-618)		269	462 (346-604)
6 to 24 months	269	504 (390-667)		269	440 (355-582)
Mean change	254	38.7		258	−14.2	12% (7% to 16%)	<0.001
Quality of life—mean (SD) Parkinson’s disease questionnaire (PDQ-39)
Baseline	297	26.0 (13.7)		286	26.2 (13.1)
6 to 24 months	278	26.4 (13.7)		277	27.7 (12.7)
Mean change	278	0.1		276	1.7	−0.9 (−2.1 to 0.3)	0.14
Physical fitness—mean (SD) six minute walk test (6MWT)
Baseline (distance in m)	298	391.6 (87.5)		283	392.9 (84.5)
12 and 24 months	256	404 (95.1)		256	394.4 (86.5)
Mean change	255	8.4		253	−1.6	4.8 (0.1 to 9.6)	0.05

*Estimated (relative) difference, based on analysis of covariance.

Safety and falls

Eight patients died during follow-up because of cardiovascular problems, cancer, or medical complications (ParkFit n=5, control n=3). These deaths were unrelated to exercise sessions. Controls reported eight hip fractures, and ParkFit patients reported two. The frequency and severity of all other adverse events were similar in the two groups (ParkFit n=221, control n=242). The number of patients with one or more falls was comparable in the two groups: 184 (62%) in ParkFit and 191 (67%) in controls.

Discussion

This randomised controlled trial showed that a multifaceted behavioural change programme does not promote overall physical activities, as measured with the primary outcome (LASA physical activity questionnaire—LAPAQ), in sedentary patients with Parkinson’s disease. Two of our secondary outcomes focused on other measures of physical activity and did suggest improvements for patients allocated to the ParkFit programme. This was shown both subjectively (with activity diaries) and objectively (with an ambulatory activity monitor). Moreover, physical fitness (an indirect reflection of greater physical activity) increased in ParkFit patients. Quality of life did not differ between the two study arms. The ParkFit group did not have more falls.

Possible explanations for findings

The ParkFit study is therefore a negative trial, showing no difference for the primary outcome (LAPAQ questionnaire) between the study arms. We selected the LAPAQ as primary outcome because it closely reflected the goals of the ParkFit intervention—namely, promotion of physical activities. We regarded an actual increase in physical activity levels as a necessary intermediate and prerequisite to eventually obtain health benefits, including improvements in quality of life. The LAPAQ questionnaire is a validated instrument to measure habitual physical activity in large populations.^{30 40} LAPAQ covers a wide range of daily life activities, and we previously showed that patients with Parkinson’s disease are 29% less active than controls, as measured with the LAPAQ.¹⁶ Our study was powered to defect a 20% increase based on the LAPAQ, which would equate to an increase in physical activities of two hours a week. The ParkFit programme did not achieve this, suggesting that more robust interventions are needed to promote physical activities in daily life.

Our choice of control intervention might have obscured greater differences on the LAPAQ between ParkFit patients and controls. We chose to refer patients in the control arm to a physiotherapist who aimed to improve the safety of movements, but without emphasising the volume of physical activities. This approach helped to maintain blinding of patients with respect to treatment allocation. An additional reason for having a physiotherapy programme as control intervention was that abstaining control patients from physiotherapy for two years was considered unethical, in the light of growing evidence for the effectiveness of specific physiotherapy interventions.^{11 13 14 15} Furthermore, the ParkFit study took place in the “real world,” and physiotherapy in Parkinson’s disease is “usual care,” not only in the Netherlands (where at least 60% of Parkinson’s disease patients receive physiotherapy annually)⁴ but also in the United Kingdom.⁴¹

Although we found no effect on the primary outcome, two of our secondary outcomes did pick up an increase in physical activities, as measured both subjectively (activity diary) and objectively (activity monitors). According to the diary, ParkFit patients spent almost 1.5 hours a week extra on physical activity, compared with baseline. This differed significantly from controls, who increased their level of physical activity by 30 minutes compared with baseline. This amount of increase in physical activity, as observed with the diary, is comparable to findings in older populations and patients with other chronic conditions.^{42 43 44 45} For example, behavioural counselling for older patients in primary care yielded a one hour increase in moderate intensity physical activity.⁴² In addition, pedometer based counselling programmes increased the total physical activity of cardiac patients by almost 1.5 hours a week.^{43 44} Both the LAPAQ and the diary are subjective instruments, but only the diary showed increased activity levels. One possible explanation for this discrepancy is the fact that the diary merely includes strenuous activities, whereas the LAPAQ questionnaire reflects the net sum of all physical activities (including household activities). Therefore, we cannot exclude that a possible increase in (strenuous) outdoor and sporting activities for ParkFit patients was offset by a concurrent decrease in household activities. The LAPAQ cannot capture such differential effects on specific physical activities, as it merely measures the net sum of all physical activities. We therefore regard our decision to select overall physical activity as primary outcome as a shortcoming in the study design, and this aspect should be included in future research in this area.

Objective assessment of physical activity using a triaxial accelerometer showed an increase in physical activities for ParkFit patients, with a 12% increase in time spent on physical activity after 24 months. Generally, accelerometers underestimate total energy expenditure, because some activities are difficult to detect. This includes upper body movements, specific activities such as cycling, and relatively static movements such as gardening or strength training.³³ On the other hand, accelerometers as used in our study can reliably measure activities such as indoor and outdoor walking.^{34 46} The accelerometers thus measured a different aspect of physical activity than the LAPAQ, and this could explain the difference in outcome from the LAPAQ. Compliance with use of the accelerometers was good, suggesting that it is a feasible surrogate outcome in future studies.

The two remaining secondary outcomes aimed at finding possible health benefits. Physical fitness showed a small but significant difference in favour of ParkFit, but quality of life did not differ between the ParkFit and control intervention. The ParkFit intervention had no major adverse effects. We were concerned about possibly increased fall rates, because amount of physical activity is associated with a greater risk of falling.⁴⁷ However, the ParkFit programme was not associated with more falls or injuries. In fact, controls reported eight hip fractures, whereas ParkFit patients reported only two. However, these numbers are very small and this finding is coincidental as we did not include hip fractures as a primary or secondary outcome. Therefore, further research should investigate whether this difference in hip fractures is related to the intervention. Another concern was cardiovascular complications due to more strenuous activities. All participants received a sports health assessment before participation. We observed two cardiovascular deaths in the ParkFit group, but these were unrelated to exercise. Other adverse effects were comparable between the groups. Taken together, this suggest that ParkFit was a safe intervention but that the programme needs to be adjusted to achieve more substantial increases in physical activity that translate into tangible health improvements.

Implications for trial design

Our experience with this ParkFit study was a lesson in trial design in this newly emerging field. Although the primary outcome was negative, we have shown the possibility of an exercise based trial in disabled people. Several features set the ParkFit study apart from previous exercise studies: the prolonged follow-up, showing that patients in both arms were able to comply with the intervention for two years; the careful matching of treatment intensity between the study arms; the large sample size, making the ParkFit trial by far the largest study on physical activity in Parkinson’s disease and other chronic diseases^{11 20 43 44}; and the excellent follow-up rate. The feasibility of the study was supported by the ParkinsonNet infrastructure, a nationwide network of allied health professionals who are specialised in Parkinson’s disease.⁴⁸ A generic challenge for trials aiming to evaluate the merits of allied health treatment is the lack of expertise among therapists who deliver the trial intervention, creating undesirable variability and insufficient contrast with the control arm. Having expert therapists within ParkinsonNet greatly facilitates the delivery of a relatively uniform intervention according to treatment guidelines.^{37 49}

As discussed above, our study also highlights the challenges of selecting the appropriate outcomes for a complex intervention such as a behavioural change programme. Physical activity is a complex behaviour: it includes sports as well as non-sports activities, and it can be characterised by purpose (occupational or leisure), type (cycling, fitness, or soccer), intensity (light, moderate, or vigorous), and duration. Further research should focus on comprehensive, valid, and reliable instruments to accurately measure all these aspects of physical activity behaviour. This is a specific challenge in patients with chronic diseases, as they perform more light and moderate activities that are easily overestimated when using questionnaires and are difficult to detect with activity monitors.

Furthermore, our trial provided new insights in the risk of selection bias. Our participants were on average less sedentary than patients who declined to participate.²⁴ Hence, those who needed to promote their physical activities were most likely to refuse participation. Whether the effects found here can be generalised to more sedentary Parkinson’s disease patients therefore remains unclear. Nor can we extend our findings to patients with severe apathy, severe cognitive impairment, or depression, because these were excluded. Finally, the ParkFit programme was a multifaceted intervention, with coaches using behavioural change techniques, ambulatory feedback devices, and peer pressure from group exercises. Future work should decide which of these components is most effective and if any component is also effective when used alone.

Conclusion

ParkFit, a multifaceted behavioural change programme, did not change the overall volume of physical activities in older, sedentary patients with Parkinson’s disease. However, analysis of the secondary outcomes suggested greater participation in specific elements of physical activity and showed an improved fitness among ParkFit patients. These results for the secondary outcomes suggest that replicating a similar behavioural change study—for example, with the secondary outcomes as primary parameters—may be worthwhile. Such a trial might also put more focus on the quality of life and cost aspects.

What is already known on this topic

• Many patients with Parkinson’s disease adopt a sedentary lifestyle because of motor and non-motor disability

• Patients with Parkinson’s disease might benefit from physical activity and exercise, but how they can be motivated to change their sedentary lifestyle and persistently increase their level of physical activities remains unclear

What this study adds

• The ParkFit behavioural change programme did not increase the overall volume of all physical activities

• Greater participation in specific elements of physical activity and an improved fitness were seen among ParkFit patients, without more falls and without a change in quality of life

• This study shows the possibility of an exercise based trial in disabled people and highlights several challenges for future studies, in particular with regard to the choice of outcome measures of physical activity

Cite this as: BMJ 2013;346:f576