Daily‐Life, Sensor‐Derived Tremor Measures Are Sensitive to Progression in Early Parkinson's Disease

Objective: Sensitive outcome measures are critical for evaluating the efficacy of novel treatments for Parkinson's disease (PD). In this study, we assess the sensitivity to change of sensor‐derived daily‐life tremor measures over 2 years in unmedicated and medicated persons with early PD.

Methods: We used 2‐year continuous wrist sensor data (median wear time: 22 hours/day) from the Personalized Parkinson Project (n = 462 medicated; n = 78 unmedicated at baseline), in combination with annual clinical evaluations of tremor severity. From the gyroscope data, we derived previously validated weekly measures for tremor time and power, which were smoothed over time using piecewise linear trend estimation. One‐ and 2‐year standardized response means (SRMs) were computed to compare the sensitivity to change between the sensor‐derived tremor measures and clinical tremor scores.

Results: In unmedicated participants with tremor, sensor‐derived tremor measures demonstrated a high sensitivity to progression (2‐year SRMs ranged from 0.67 to 1.09), which was significantly larger than clinical tremor scores (2‐year SRMs ranged from 0.21 to 0.41). In medicated participants, sensor‐derived tremor time decreased (2‐year SRM of −0.18), which was associated with both an increase in dopaminergic medication dose and higher disease duration. In contrast, the sensor‐derived tremor power measures and clinical rest tremor scores (measured in the OFF state) increased slightly (2‐year SRMs ranging from 0.11 to 0.27).

Interpretation: Before initiation of symptomatic treatment, sensor‐derived daily‐life tremor measures are substantially more sensitive to progression than clinical tremor scores, making them a promising tool to evaluate the efficacy of disease‐modifying treatments in early PD.