This is the first study that investigates how perceived walking difficulties evolve over time in people with PD, including identifying predictive factors. Our results showed that perceived walking difficulties increased significantly over 3 years. The first regression model (controlled for age) showed that concerns about falling was the strongest predictive factor for perceived walking difficulties, followed by perceived balance problems while dual tasking, being bothered by pain and having postural instability. The second model (controlled for age and baseline Walk-12G scores) showed that perceived balance problems while dual tasking was the strongest predictive factor for a change in perceived walking difficulties, followed by postural instability.
The mean Walk-12G score increased by 3.9 points over a 3-year period, which exceeds the measurement error (SEM) presented in this study as well as in a previous PD study 21. The level of perceived walking difficulties in our sample at baseline seems to be in line with previous PD studies: the Walk-12G mean score was 14.7 in the present study as compared to 13-15.5 in previous studies 21,29,35. However, one study reported a considerably lower Walk-12G score: median 8 (q1-q3 4.5–21). This might be explained by a shorter PD duration (mean 5 years vs. median 8 years in the present study) and less motor symptoms (UPDRS III median 13 vs. 28 in the present study; data not presented) 36.
All dichotomous independent variables showed an effect on perceived walking difficulties that exceed the measurement error. For concerns about falling, participants need to change their FES-I score at least 5 points in order to achieve an effect that exceeds the measurement error of Walk-12G. It should be noted that the possible total score of FES-I ranges from 16–64. Thus, a changed score of 5 points corresponds to 10% change in relation to the scoring range.
That concerns about falling was the strongest predictive factor of perceived walking difficulties is in line with prior cross-sectional studies, which described that fear of falling relates to walking difficulties in people with PD 29,36. In the current study, concerns about falling did not predict a change in perceived walking difficulties, i.e. when controlling for baseline Walk-12G scores. Controlling for a variable in a multivariable regression analysis means that the variable is included in all steps of the analysis, as well as in the final regression model. In the current study, FES-I scores were highly correlated with Walk-12G scores at baseline (rs =0.870), and this multicollinearity might explain why concerns about falling did not independently predict a change in perceived walking difficulties. Similarly, a previous study based on the same sample identified perceived walking difficulties as the strongest predictive factor of concerns about falling (FES-I), but it failed to predict a change in concerns about falling when controlling for baseline FES-I scores 19. Perceived walking difficulties and concerns about falling are interconnected and adjacent aspects, but they are not interchangeable constructs. Perceived walking difficulties focus on problems connected to walking ability, whereas FES-I targets concerns about falling while performing 16 different activities. In FES-I, five out of the 16 items explicitly mention walking and an additional item addresses stairclimbing. The remaining 10 items address concerns about falling while for example getting dressed or undressed, preparing simple meals, reaching for something above your head or on the ground, visiting a friend or relative, and going out to a social event 27.
Perceived balance problems while dual tasking was the second strongest predictor of perceived walking difficulties, and it was the strongest predictor for a change in perceived walking difficulties. Prior studies showed that dual tasking negatively affects objective gait measures (e.g. gait speed, gait variability, double support time and gait rhythmicity) in people with PD 37,38. On the other hand, dual task training can be used to increase the level of difficulty when training gait and balance. Some studies suggest that dual task training has a positive effect on objective gait measures (e.g. dual task gait velocity, stride length and cadence) in people with PD 39,40. Those who seems to benefit the most of dual task training are those with low initial gait speed when dual tasking and those with better cognitive functioning 39. Interestingly, a study of highly challenging balance training, which included dual tasking, showed statistical significant improvements in balance performance and gait speed, but detected no improvements in any of the used patient reported outcomes, i.e. Walk-12G and rating scales targeting balance confidence and health-related quality of life 35. Their findings might be due to that the intervention effect was considered clinically small, and the effect might therefore not transfer to perceived aspects such as Walk-12G scores. In other words, a potential explanation might be that objective changes in for example gait speed require large and/or long lasting changes in order to be reflected in patient-reported outcomes.
Pain and postural instability predicted perceived walking difficulties, and postural instability also predicted a change in perceived walking difficulties. Pain is a non-motor symptom that can affect up to 85% of people with PD, and it is most frequently located in the lower limbs 41. Pain in people with PD can result in activity limitations such as walking difficulties 42, and it has been shown to be negatively associated with quality of life 43. In a prior cross-sectional study based on the same larger project as the present study (i.e. “Home and Health in People Ageing with Parkinson’s disease”), both pain and perceived walking difficulties were shown to be associated with decreased life-space mobility 44. All considered, pain deserves attention at clinical follow-ups. Management of pain in people with PD may for example include medication, cognitive strategies or exercise 42,45.
In this study, postural instability was assessed by using item 30 of UPDRS part III; it is intended to assess the righting reflex and the reactive “response to a sudden posterior displacement produced by pull on shoulders” while standing erect with eyes open and feet slightly apart 24. That both item 30 and perceived balance problems while dual tasking affect perceived walking difficulties highlight the importance of addressing balance problems in people with PD. Not the least since anti-PD medication insufficiently affects postural instability 46. A meta-analysis found that physical exercise had a small positive effect on postural instability in people with PD, and “highly-challenging” balance training was advocated 47.
It needs to be noted that global cognitive functioning was significantly associated with perceived walking difficulties in univariable regression analysis, and close to significantly associated in model 2 (i.e. multivariable regression analysis). More studies are needed that address various cognitive domains in relation to perceived walking difficulties in order to further explore the potential impact of cognitive function on walking in people with PD.
At baseline, 243 participants completed the Walk-12G questionnaire and 61.3% of those completed the 3-year follow-up. There are dropouts in all longitudinal studies, which can affect the external validity of the findings. We have previously reported that those who completed the assessments at both time points were significantly younger and had shorter PD-duration than those who were lost for follow-up 19.
Twelve participants that did not manage the Five chair stands test were included in the group of participants categorized as having worse lower extremity function. However, there might be other reasons than poor lower extremity function that made them unable to complete the test. The categorization was done since excluding them might have rendered a final sample that was skewed towards having a better lower extremity functioning.
The independent variables in the present study were selected based on theoretical reasoning as well as on the results from previous cross-sectional studies 16,17. The regression models in our study explained 61.7% and 67.6% of the variance in the Walk-12G scores, respectively. This means that there are additional factors than those studied that can predict perceived walking difficulties. Future studies are needed to explore the effect of e.g. visual impairments on perceived walking difficulties in people with PD.