Abstract
Introduction: Comparing frailty models in different settings that predict inhospital mortality might modify patient disposition and treatment, but models are often complex. Methods: In the following study, we selected all acutely admitted adult patients in 2020–2021 to the three internal medicine departments at a regional 400-bed hospital. We attempt to determine (a) if a new scale (Laniado-4 scale) that includes only three yes/no questions derived from the Norton scale and the presence of a urinary catheter performs as well as the graded Norton scale (including all five domains), in predicting inhospital mortality and (b) to determine the predictive value of a simple frailty index that includes the new scale as well as categories of age, serum albumin, and creatinine values. We calculated odds ratios with 95% confidence intervals and c-statistics for the various models predicting inhospital mortality. Results: The mean patient age was 73 ± 19 years, and 49.1% (5,665/11,542) were males. A Laniado-4 scale performed better than the Norton scale for predicting inhospital mortality. A simple frailty index ranging from 0 to ≥8 points was associated with rates of inhospital mortality that increased from 0 to 37.7%, with an odds ratio of 2.13 (2.03–2.25) per 1 index point. The c-statistic was 0.887 (0.881–0.893). Conclusions: We conclude the Laniado-4 scale performed better than the Norton scale in predicting inhospital mortality and that a simple frailty index that included the 4-question scale and categories of age, serum creatinine, and serum albumin performed as well or better than more complicated models.