To develop a frailty risk prediction model for elderly inpatients based on Meta-analysis, providing a basis for early clinical screening of frail older patients.

A systematic literature search was conducted to identify studies analyzing factors influencing frailty. Meta-analysis was then performed to pool the odds ratios (OR) or relative risks (RR) for each factor as summary statistics. The combined OR value was used to calculate β to construct a logistic regression equation, and the β was used to assign scores to each influencing factor to build a frailty risk scoring table. The FRAIL scale served as the gold standard for frailty diagnosis, and the model was externally validated in a cohort of elderly inpatients. The total score of the model was used as the test indicator for the diagnostic test, and the area under the curve (AUC) was used as the comprehensive evaluation index of the model.

A total of 57 studies were included in the meta-analysis. The constructed frailty risk scoring model assigns scores as follows: age (≥60 years old, 2 points), comorbidities (clinical diagnosis≥5, 10 points), physical activity (daily activity volume < 6000 steps, 17 points), hypertension (6.5 points), diabetes (6.5 points), renal insufficiency (5.5 points), NRS 2002≥3 points (13 points), BNP≥100 pg/mL (1 point), CRP≥5 mg/L (2 points). The external validation trial showed that the area under the receiver operating characteristic curve of this predictive model was 0.973 (95%CI: 0.955~0.991, P < 0.0001); when the frailty risk scale score was≥25 points, it could be judged as frail, and the sensitivity of this critical value was 0.925, and the specificity was 0.925.

The frailty risk prediction model constructed through meta-analysis in this study has excellent stability and predictive performance. The indicators collected by the model are highly accessible in clinical settings and are suitable for clinical application and promotion.