External validation of a 3-step falls prediction model in mild Parkinson’s disease

摘要: The 3-step falls prediction model (3-step model) that include history of falls, freezing gait and comfortable speed <1.1 m/s was suggested as a clinical fall tool in Parkinson’s disease (PD). We aimed to externally validate this well explore the value additional predictors 138 individuals with relatively mild PD. found discriminative ability identifying fallers be comparable previously studies [area under curve (AUC), 0.74; 95 % CI 0.65–0.84] better than single (AUC, 0.61–0.69). Extended analyses generated new for near 0.82; 0.75–0.89) including retropulsion according Nutt Retropulsion test (NRT) tandem (TG). This study confirms PD illustrates it outperforms use predictors. However, improve future outcomes, further are needed firmly establish scoring system risk categories based on model. influence methodological aspects data collection also needs scrutinized. A TG may considered an alternative model, but tested samples before being recommended. Taken together, our observations provide important additions evidence base