ABSTRACT
Background: The Short Physical Performance Battery (SPPB) is an instrument for assessing physical performance widely used in research among the elderly in multiple settings. We did not find Brazilian longitudinal studies that aimed to analyze the predictive capacity and accuracy of the SPPB among community-dwelling older adults and no systematic reviews were found on the accuracy of the SPPB in predicting mortality in community- dwelling older adults. This study aimed to analyze the capacity and accuracy of the SPPB for predicting mortality in community-dwelling older adults, as well as to determine cut-off points for men and women. Method: Longitudinal observational study conducted with 411 (70.1 ± 7.25 years) community-dwelling older adults, between 2017 and 2020 (37.7 ± 6.24 months). Physical performance was evaluated using the SPPB and information on the all-cause mortality rate was also recorded. Multivariate Cox regression analyses and curves were performed using the Kaplan-Meier method. Receiver Operating Characteristic (ROC) curves were constructed, with the parameters of area under the ROC curve (AUC) to determine cutoff points for discriminating mortality, considering a significance level of 5% (p < 0.05) and 95% confidence interval (CI) 95%. Results: Older adults with very low and low physical performance in the SPPB, showed higher risks of mortality (HR = 9.67; 95% CI [1.20-77.65]; HR = 4.06; 95% CI [1.09-15.01]), respectively. In the subtest's analysis, older adults with low performance in the balance (HR = 0.54; 95% CI [0.36-0.81]) and gait speed tests (HR = 0.50; 95% CI [0.33-0.76]) showed greater risks of dying. The same was reproduced for categories in each test (participants that scored 2 points in the balance test had an HR = 5.86; 95% CI [1.84-18.61] and 2 points in the gait speed test, HR = 5.07; 95% CI [1.76-14.58]. The cutoff point ≤ 9 in the SPPB set the discriminator criterion for mortality in older people of both sexes. Conclusions: The SPPB, as well as the balance and gait speed subtests were predictors of mortality, and the SPPB is accurate in predicting mortality among community-dwelling older adults.