Longitudinal identification of sentinel symptoms across frailty dimensions in lung cancer patients undergoing radiotherapy.

PURPOSE: This study aimed to identify sentinel symptoms related to frailty in patients with lung cancer undergoing radiotherapy (RT) across the physical, psychological, and social domains at different treatment stages. This will provide a theoretical basis for targeted symptom management.

METHODS

This study used a prospective longitudinal design. The Chinese version of the Tilburg Frailty Indicator (TFI) was administered at five time points: T1 (baseline, 1 day before RT), T2 (after 10 sessions of RT), T3 (after 20 sessions of RT), T4 (at the completion of RT), and T5 (1 month after RT). The Apriori algorithm was applied to examine symptom associations within each frailty dimension and to identify sentinel symptoms.

RESULTS

This study included 286 patients with lung cancer undergoing radiotherapy. In the psychological frailty dimension, tension or anxiety before radiotherapy was identified as a sentinel symptom and remained prominent until the completion of radiotherapy, and poor memory was identified as the sentinel symptom at 1 month after radiotherapy. In the physical frailty dimension, fatigue emerged as the sentinel symptom at the 20th radiotherapy session and remained prominent until 1 month after radiotherapy. In the social frailty dimension, lack of adequate support was identified as the sentinel symptom at the 20th radiotherapy session and remained prominent until the end of radiotherapy. No sentinel symptom was identified in the social frailty dimension 1 month after radiotherapy.

CONCLUSION

Sentinel symptoms across different frailty dimensions showed dynamic changes throughout the radiotherapy period. These findings provide a new perspective for stage-specific and dimension-specific frailty assessment and precision nursing interventions during radiotherapy. Because the identified sentinel symptoms were derived from association rule mining, they should be interpreted as reflecting statistical associations and potential early indicators of frailty-related symptom burden rather than confirmed causal relationships. These findings may also inform the future development of early frailty warning models and intelligent intervention pathways.