The UPDRS under the microscope: how reliable is it as a measure of Parkinson’s?

Statistical analysis of 5 years’ worth of Unified Parkinson’s Disease Rating Scale (UPDRS) from over 400 individuals has shown it not to be uniformly reliable as a tool for the measurement of Parkinson’s. There is substantial variability particularly in terms of assessing movement in the ON state and mood, behaviour and thought.

These findings highlight strengths as well as important limitations of the most important tool for the measurement of Parkinson’s in the clinic as well as in trial settings. They could feed into a modified or new scale for the measurement of Parkinson’s.

One of the most important aspects of monitoring Parkinson’s, in order to track its progression and measure any effects of therapies that may stop, slow or reverse it, relates to how symptoms are measured. The UPDRS is the tool traditionally used in neurological assessments, which take place in clinic and form the basis of primary outcome measures for clinical trials assessing the efficacy of new therapies.

But how reliable is the UPDRS? Apart from the usual difficulties associated with differences as a result of the rater or clinician, Parkinson’s itself is highly variable, with some symptoms changing rapidly over the course of a few hours. How good is the UPDRS at capturing the intrinsic, slowly evolving state of a person with a highly changeable condition in a reliable manner? In other words, how consistent within any one person are UPDRS measurements? Reliability refers to the extent of measurement error.

The authors analysed data from the PPMI dataset, a multicenter cohort study designed to identify progression biomarkers, launched in 2010. The UPDRS data of a total of 423 people with Parkinson’s aged 62 years on average, were included. Multiple assessments were completed both ON and OFF medications, over a maximum follow up period of 5 years. The authors analysed data from all 3 sections of the UPDRS.

They found that Parts II (activities of daily living) and III (movement) assessed in the OFF state performed well in terms of reliability. However, Parts I (mood, behaviour and thought) and III in the ON state showed lower reliability within any one individual assessed. Out of the individual main factors of the whole scale that emerged, tremor at rest and gait/mobility performed better than other items related to slowness of movement, rigidity, kinetic/postural tremor, and non-motor symptoms.

The next challenge is to develop an improvement tool for the assessment of Parkinson’s, and such findings could feed directly into these efforts.