Predicting risk of declining functional ability in community-dwelling older people

The frailty phenotype (FP) proposed by Fried and colleagues has been shown to be strongly associated with incident disability, but its discriminative capacities remain suboptimal, with good specificity but a sensitivity of only 10-20%. The objective of the present study was to evaluate whether the addition to the FP of other biological and social variables may improve the prediction of declining functional ability in community-dwelling older people.

Prospective observational study.

Community-dwelling older subjects.

We used data from the InChianti (N 897) and the SHARE (N 444) studies to derive and validate a scoring system consisting of FP components along with age, perceived health status and markers of socio-economic disadvantage. Backward stepwise logistic regressions were used to obtain a parsimonious model, able to predict the loss of ability to perform instrumental or basic activities of daily living over time.

A scoring system derived from a model only including age, low physical activity level, exhaustion and perceived health status had an area under the receiver operating characteristic curve (AUROC) of 0.846 in the training cohort (InChianti), and 0.745 in the testing cohort (SHARE). By applying the cut-off of 33 and 25 in the InChianti and SHARE, respectively, sensitivity raised to 0.70 and 0.62 with specificity of 0.83 and 0.70, respectively.

A simple score based on anamnestic variables may be more sensitive than the FP towards worsening functional ability, while retaining good specificity. Further studies are needed to confirm its performance.