Prediction Models for Falls Risk Among Inpatients: A Systematic Review and Meta‐Analysis

ABSTRACT

Aim
To systematically review published studies on fall risk prediction models for inpatients.


Design
A systematic review and meta‐analysis of prognostic model studies.


Data Sources
A literature search was carried out in Web of Science, the Cochrane Library, PubMed, Embase, CINAHL, SinoMed, VIP Database, CNKI and Wanfang Database. The search covered studies on risk prediction models for falls in inpatients from inception to March 9, 2024.


Methods
The research question was formulated using the PICOTS framework. Data extraction was performed following the Critical Appraisal and Data Extraction for Systematic Reviews of Prediction Modelling Studies (CHARMS). The quality of studies related to risk prediction models was evaluated with the Prediction Model Risk of Bias Assessment Tool (PROBAST). Meta‐analysis was conducted using STATA 18.0 software.


Results
A total of 15 studies were included, with 13 eligible for meta‐analysis. Only 2 of these 15 studies had external validation. The reported AUC values ranged from 0.681 to 0.900. The overall risk of bias was high, mainly attributed to inappropriate data sources and improper processing in the analysis domain. The pooled AUC from the meta‐analysis was 0.799. After reviewing the predictors included in various models, FRIDs, fall history, age, gait, mental status, gender and incontinence were relatively common.


Conclusion
The fall risk prediction model for inpatients performs well overall, but it has a high risk of bias. Future development of risk prediction models should strictly adhere to the PROBAST, combine clinical reality, optimise study design and improve methodological quality.


Impact
This study provides medical professionals with a clear overview of constructing fall risk prediction models for inpatients. The fall‐related predictors in these models help healthcare providers identify high‐risk patients and implement preventive strategies. It also offers valuable insights for the development of future prediction models.


No Patient or Public Contribution
This study did not include patient or public involvement in its design, conduct, or reporting.