Serum IL-6 and IGF-1 improve clinical prediction of functional decline after hospitalization in older patients.

BACKGROUND AND AIMS Although inflammatory and hormonal markers have been associated with further functional adverse outcomes in community- dwelling seniors, these markers have not been studied from this perspective in acutely ill older patients. This prospective study was designed to determine whether biological markers can improve the predictive value of a clinical screening tool to assess the risk of functional decline in hospitalized older patients. METHODS Patients aged 75 years and over admitted for hip fracture, acute heart failure or infection (n=118) were recruited. The clinical screening tool SHERPA was filled in on admission, with concomitant measurement of interleukin-6 (IL-6), insulin-like growth factor 1 (IGF-1), C-reactive protein (CRP), white blood cells, urea, albumin, pre-albumin and total cholesterol. Functional decline was defined as a decrease of one point in the activities of daily living scale between pre-admission and 3-month post-discharge status. We compared the discrimination calibration of SHERPA vs SHERPA+, a logistic regression model including SHERPA and selected biomarkers. RESULTS Three months after discharge, functional decline had occurred in 46 patients. IL-6 and IGF-1 were selected, since their levels were significantly different between decliners and non-decliners, and were included in the new logistic regression model SHERPA+. Areas under the ROC curve [95% CI] for functional decline prediction were 0.73 [0.63-0.81] for SHERPA vs 0.79 [0.69-0.86] for SHERPA+ (p=0.14). However, SHERPA+ was better calibrated, as the average predicted risk of functional decline within subgroups matched the proportion which actually underwent functional decline (Brier score=0.185). Since functional decline was higher in patients with hip fracture, the SHERPA+ model was challenged by including the diagnosis. Only SHERPA, IGF-1 and diagnosis were significantly associated with functional decline. CONCLUSIONS Selected biological markers may marginally improve the clinical prediction of post-discharge functional decline in hospitalized patients, and may allow to stratify them appropriately. The predictive value of these biomarkers is not fully independent of disease status. Further studies are needed to confirm these results in a cohort representative of older patients admitted through the emergency department.