Is impact of fatigue an independent factor associated with physical activity in patients with idiopathic Parkinson's disease?

To investigate the longitudinal association between fatigue and physical activity in Parkinson's disease (PD) and determine whether this association is distorted by potential confounders. Data from baseline, 3, 6, and 12 week assessments in a single blind randomized clinical trial with cross-over design were used (N = 153). The Multidimensional Fatigue Inventory (MFI) was used to assess fatigue and an activity monitor to measure amount of physical activity (defined as % dynamic activity during each monitoring session). Time-independent and time-dependent factors were investigated for their possible univariate association with dynamic activity. Random coefficient analysis was applied. Candidate confounders were successively added to the longitudinal association model to determine if the association between dynamic activity and fatigue was distorted. A proportional change beyond 15% was considered to be significant. Fatigue was significantly associated with physical activity (beta = -0.099, SE = 0.032, P = 0.002). This association was not significantly distorted by type of intervention, age, gender, social support, disease duration, disease severity, motor impairment, cognition, anxiety, or medication intake. Depression caused proportional change of 22.2% in the regression coefficient of MFI. After controlling for depression, a significant association between MFI and dynamic activity remained (beta = -0.121, SE = 0.036, P = 0.000). The association found between fatigue and dynamic activity suggests that patients who experience higher levels of fatigue are less physically active. However, the total explained variance of dynamic activity by fatigue alone was small, suggesting that fatigue is only a minor factor in the complex of behavioral aspects that affect the amount of physical activity in patients with PD.