Reliability and variability of physiotherapists scoring freezing of gait through video analysis

The "gold standard" marker for freezing of gait severity is percentage of time spent with freezing observed through video analysis.

This study examined inter- and intra-rater reliability and variability of physiotherapists rating freezing of gait severity through video analysis and explored the effects of experience.

Thirty physiotherapists rated 14 videos of Timed Up and Go performance by people with Parkinson's and gait freezing. Ten videos were unique, while four were repeated. Freezing frequency, total duration, and percentage of time spent with freezing were computed. Reliability and variability were estimated using ICC (2,1) and mean absolute differences. Between-group differences were calculated with the one-way ANOVA.

Inter- and intra-rater reliability ranged from moderate to good (ICC: inter-rater frequency = 0.63, duration = 0.78, percentage = 0.50; intra-rater frequency = 0.84, duration = 0.89, percentage = 0.50). Variability for freezing frequency was two episodes. Inter- and intra-rater variability for total freezing duration was 18.8 and 12.3  seconds, respectively. For percentage of time spent with freezing, this was 15.2% and 13.5%. Physiotherapy experience had no effect.

Physiotherapists demonstrated sufficient reliability, but variability was large enough to cause changes in severity classifications on existing rating scales. Percentage of time spent with freezing was the least reliable marker, supporting the use of freezing frequency or total duration instead.