Predicting falls in elderly patients with chronic pain and other chronic conditions Aida LazkaniTiba DelespierreBernard BauduceauLinda Benattar-ZibiPhilippe Bertin • Gilles BerrutEmmanuelle CorrubleNicolas DanchinGenevieve DerumeauxJean Doucet • Bruno FalissardFrancoise ForetteOlivier HanonFlorence PasquierMichel Pinget • Rissane OurabahCeline PiedvacheLaurent Becquemont

Background The aim was to identify fall predictors in elderly suffering from chronic pain (CP) and to test their applicability among patients with other chronic conditions. Methods 1,379 non-institutionalized patients aged 65 years and older who were suffering from CP (S.AGE CP sub-cohort) were monitored every 6 months for 1 year. Socio-demographic, clinical and pain data and medication use were assessed at baseline for the association with falls in the following year. Falls were assessed retrospectively at each study visit. Logistic regression analyses were performed to identify fall predictors. The derived model was applied to two additional S.AGE sub-cohorts: atrial fibrillation (AF) (n = 1,072) and type-2 diabetes mellitus (T2DM) (n = 983). Results Four factors predicted falls in the CP sub-cohort: fall history (OR: 4.03, 95 % CI 2.79‐5.82), dependency in daily activities (OR: 1.81, 95 % CI 1.27‐2.59), age C75 (OR: 1.53, 95 % CI 1.04‐2.25) and living alone (OR: 1.73, 95 % CI 1.24‐2.41) (Area Under the Curve: AUC = 0.71, 95 % CI 0.67‐0.75). These factors were relevant in AF (AUC = 0.71, 95 % CI 0.66‐0.75) and T2DM (AUC = 0.67, 95 % CI 0.59‐0.73) sub-cohorts. Fall predicted probability in CP, AF and T2DM sub-cohorts increased from 7, 7 and 6 % in patients with no risk factors A. Lazkani (&) T. Delespierre C. Piedvache