Falls are common in Parkinson's disease (PD) resulting in increased disability and morbidity. Rehabilitation strategies and modification of falling risk factors can improve outcomes in PD.

Our main aim was characterization of a PD population's subset at primary care level (Tables 1 and 2) and study possible associations with occurrence of falls.

This study is cross-sectional and retrospective. The sample comprised every patient of a family health unit with diagnosis of PD. Sociodemographic factors, medical history, participation on rehabilitation programs (RP), physical activity, active recreational activities, occurrence of falls in the last year, Activities-Specific Balance Confidence (ABC) scale score, Barthel Index score and potential risk factors for falls were obtained through questionnaire application by phone and complemented with clinical records. A binomial logistic regression was performed to ascertain the effect of some of those variables on the likelihood of falls occurrence.

Thirty-four participants were elected, of which 14.7% were physically active, 52.9% performed recreational activities and 44.1% participated in a RP in the last year. Fifty-four falls were reported in 18 patients in the last year. An ABC score cutoff of 55.0 to distinguish fallers from non-fallers was determined (sensibility 0.78, specificity 0.81) (Fig. 1). A logistic regression model explained 72.6% of falls occurrence (P<0.005, sensibility 0.83, specificity 0.87) (Table 3). Non-participation in RP had 33.2 times higher odds of falling occurrence in the last year.

It's important to address and manage the fall risk at primary care level that requires a multidisplinary approach. Lower ABC scores and non-participation on rehabilitation programs were associated with occurrence of falls in PD. The quantification of fall risk may be used for selection of PD patients, which benefit the most from rehabilitation strategies.