Discordance between clinician- and patient-reported outcomes during radiotherapy for brain metastases: the...

Clinician-reported outcomes (CROs) remain a central component of side effect assessment in radiotherapy, while patient-reported outcomes (PROs) are increasingly being integrated into standard practice to better capture the subjective symptom burden experienced by patients. Agreement between CRO and PRO has not been prospectively studied in patients undergoing radiotherapy for brain metastases.

NeuroPRO was a prospective longitudinal cohort study enrolling patients receiving radiotherapy for brain metastases. CROs were assessed using predefined CTCAE (version 5.0) items, while PROs were collected using the EORTC QLQ-C30 and BN20 at baseline, end of treatment, and first follow-up. Upon completion of radiotherapy, decision regret was evaluated additionally using the Ottawa Decision Regret Scale. Chance-corrected agreement between CRO and PRO was quantified for each item using Gwet's second-order agreement coefficient (AC2). Longitudinal differences were additionally explored using two-way ANOVA and agreement was further characterized using Bland-Altman analyses with calculation of limits of agreement (LoA).

Sixty-three patients were included, completing 278 questionnaires and yielding 4448 matched CRO-PRO item comparisons. Across 27 of 32 shared symptoms (84%), PRO scores were higher than CRO scores, indicating systematic symptom underestimation by clinicians. Chance-corrected agreement (Gwet's AC2) was fair to moderate for psychological and neurological symptoms (worry AC2 = 0.30; depression AC2 = 0.39; drowsiness AC2 = 0.44) and substantial to almost perfect for gastrointestinal and treatment-related symptoms (vomiting AC2 = 0.96; seizures AC2 = 0.94; nausea AC2 = 0.92), corroborating the pattern observed with mean-based comparisons. The largest single-item discrepancies were observed for dyspnea (bias 1.00; 95% LoA 0.85-1.14; group p < 0.001), insomnia (bias 0.76; 95% LoA 0.22-1.29; group p < 0.001), and depression (bias 0.75; 95% LoA 0.39-1.12; group p < 0.001). Symptom clustering demonstrated higher agreement for directly treatment-related effects and physical functioning, whereas agreement was substantially lower for emotional and cognitive domains. Patient-reported global health status remained stable after radiotherapy, and decision regret was low.

Sole clinician assessment systematically underestimates symptom burden in patients undergoing radiotherapy for brain metastases, particularly within psychological domains and somatic symptoms not readily attributable to radiotherapy. These findings reveal a structural perception gap and support the routine integration of PROs into interdisciplinary neuro-oncologic care to enhance symptom recognition and promote patient-centered management (DRKS00031480).