Pre-frailty represents a transitional risk status between health and frailty. Early identification and intervention during this stage can delay or even reverse the frailty development. However, there is a lack of standardized pre-frailty assessment tool tailored for community-dwelling older adults.

To compare the effectiveness of Fried's Frailty Phenotype (FP) and FRAIL Scale in assessing pre-frailty among community-dwelling older adults undergoing health examinations, thereby providing evidence for the selection of appropriate assessment tools.

A cross-sectional study was conducted using convenience sampling to recruit older adults aged 60 years and above undergoing health examinations at five community health centres in Beijing from December 1, 2024, to March 20, 2025. Demographic data were collected, and frailty status was assessed using the FP and the FRAIL scale. Activities of daily living (ADL) were evaluated using the Modified Barthel Index (MBI), and quality of life was assessed using the 36-Item Short Form Survey (SF-36). Spearman rank correlation and Kappa statistics were used to analyze the consistency and correlation between the two scales. Using MBI and SF-36 scores as validity criteria, the validity of both tools was evaluated via Spearman rank correlation, receiver operating characteristic (ROC) curve analysis, and Bayes discriminant analysis.

The prevalence of pre-frailty detected by the FP was higher than that by the FRAIL scale (36.3% vs. 25.3%). The two scales showed a moderate positive correlation (r_s=0.713, P<0.001) and moderate agreement (Kappa=0.606, P<0.001), with consistent classification in 81.2% of participants. Frailty severity assessed by both scales was positively correlated with ADL decline and negatively correlated with SF-36 total, Physical Component Summary (PCS), and Mental Component Summary (MCS) scores. Both scales demonstrated associations with ADL decline, with ROC curve areas under the curve (AUC) of 0.736 and 0.735, respectively (P<0.001). Bayes discriminant analysis indicated that the cross-validation accuracy for ADL decline was higher for the FP (86.3%) than the FRAIL scale (85.1%). ROC analysis revealed that the FP had higher sensitivity (74.0% vs. 64.4%), while the FRAIL scale had superior specificity (80.1% vs. 65.8%) for predicting ADL decline. The optimal cutoff value for both scales in predicting ADL decline was 0.5.

The FP and FRAIL scale demonstrate moderate correlation and consistency, and both are negatively associated with quality of life. Both tools possess moderate validity in verifying ADL decline and are suitable for assessing pre-frailty in community-dwelling older adults. The FP, with its higher sensitivity and inclusion of objective indicators, is more suitable for pre-frailty screening in health examination settings aiming for "early detection and intervention". Conversely, the FRAIL scale, due to its simplicity and high specificity, serves as a viable alternative for rapidly identifying high-risk individuals in resource-limited settings. These findings suggest practical value in integrating frailty screening into routine community health examinations and initiating interventions based on a 0.5 cutoff value.