Multidisciplinary prognostic accuracy in severe frailty: assessing short-term survival predictions in hospitalized older adults

Abstract Background The increasing number of hospitalized older adults with severe frailty and limited life expectancy highlights the need for accurate short-term prognosis to inform care planning, effective communication, and resource allocation. However, survival prediction in this population is underexplored. This study aimed to assess (i) agreement among healthcare professionals on short-term prognosis, (ii) predictive accuracy of multiprofessional survival estimates, and (iii) factors affecting prognostic accuracy. Methods A prospective observational study was conducted on patients aged ≥ 65 years with severe frailty consecutively admitted to an Acute Geriatric Unit. All participants underwent Comprehensive Geriatric Assessment and screening for palliative care needs using the NECPAL tool. Healthcare professionals provided survival estimates at 1 and 3 months through a modified surprise question. Inter-rater agreement was assessed with Fleiss’ kappa. Kaplan-Meier curves were employed to compare survival across multiprofessional prognostic categories. The association between patient-related factors and prognostic accuracy was analyzed using logistic regression. Results For the 117 patients included (mean age: 86.4 years, 57.3% female), prognostic agreement among healthcare professionals was moderate at both 1 month (κ = 0.475) and 3 months (κ = 0.528). Multiprofessional estimates classified 39.3% of patients as having an expected survival of < 1 month, 18.8% as 1–3 months, and 41.9% as > 3 months. During the follow-up period, 47 patients (40.2%) died, 30% of whom tested negative for NECPAL screening. Estimates differentiated patients based on short-term survival (log-rank p < 0.001) with an overall accuracy of 60.7%; positive predictive value was highest for survival beyond 3 months. Prevalent delirium was significantly associated with reduced prognostic accuracy (OR 0.38; 95% CI: 0.16–0.88). Conclusions Reliable prognostic tools are needed to enhance care planning for severely frail older patients during the early stages of hospitalization. In this context, delirium assessment is crucial, as its presence significantly impacts the accuracy of short-term survival estimates, while the NECPAL tool demonstrated limited applicability.