Simplification of the Barthel scale for screening for frailty and severe dependency in polypathological patients.

Objectives: To analyse the sensitivity, specificity and positive predictive (PPV) and negative predictive (NPV) values of each measure of the Barthel index (BI) compared with the full questionnaire for polypathological patients (PPPs).

Methods: Multicentre cross-sectional study. We considered 2 cut-off points for the BI (≥90 points for screening frailty and <60 points for diagnosing severe dependence). For each measure and combination of 2 measures, we calculated the sensitivity, specificity, PPV and NPV with respect to the full BI.

Results: The mean BI of the 1,632 included PPPs (mean age, 77.9±9.8years; 53% men) was 69±31 (<90 for 58.7% and <60 for 31.4% of the patients). The "feeding" measure achieved the highest NPV, for a BI ≥60 and ≥90 points (87% and 99.6%, respectively). The "walking" and "going up and down stairs" measures achieved the highest PPV, for a BI ≥60 and ≥90 (99.2%/99.5% and 81%/92%, respectively. The combination of the 2 measures increased the PPV to 95% and 99.6%, respectively.

Conclusions: PPPs in hospital settings have a high rate of functional impairment. The measure for feeding achieved the highest NPV and can therefore be employed for diagnosing severe dependence. The combination of the measures for walking and going up and down stairs achieved the highest PPV and can therefore be employed to propose frailty screening for PPPs.