Is patient self-reporting more accurate than clinician reporting of symptoms for predicting survival in patients with cancer? Meta-analysis of 30 closed EORTC randomized controlled trials
9597 Background: This study investigated whether patient self-reporting of symptoms improved prediction of survival as compared to clinician reporting or whether it provided an additive value when taken together with clinician assessment of the same symptoms. Methods: Patients with advanced cancer from 30 European Organisation for Research and Treatment of Cancer (EORTC) Randomized Controlled Trials were included in this retrospective pooled analysis. Clinician [Common Toxicity Criteria (CTC)] and patient (EORTC QLQ-C30) symptom assessment were reported at entry into the study. Data were obtained for six symptoms: pain, fatigue, vomiting, nausea, diarrhea and constipation. The prognostic accuracy for survival was assessed by modeling the contrast in reporting using the Harrell's discrimination c-index (c). Results: Data were available from patient and clinician assessment for pain [number of trials (t) =8, number of patients (n) =1214], fatigue [t=5, n=1237], vomiting [t=5, n=824], nausea [t=6, n=1393], diarrhea [t=6, n=815] and constipation [t=4, n=751]. Fatigue (c=0.59 vs 0.55, p<.01) and constipation (c=0.57 vs 0.52, p=0.03) as reported by patients (vs clinicians) were significantly higher in predicting survival. Patient reported pain (c=0.59 vs 0.58, p=0.17), nausea (c=0.54 vs 0.52, p=0.51), vomiting (c=0.55 vs 0.52, p=0.21) and diarrhea (c=0.51 vs 0.52, p=0.49) did not predict survival any more accurately than clinician assessment. Patient and clinician assessment combined (vs clinicians alone) improved the prognostic accuracy for fatigue (c=0.61 vs 0.55, p=0.01), pain (c=0.60 vs 0.58, p<0.01), nausea (c=0.54 vs 0.52, p=0.04), vomiting (c=0.56 vs 0.52, p=0.04) and constipation (c=0.5 vs 0.52, p=0.01), but not for diarrhea (c=0.52 vs 0.52, p=0.44). Conclusions: Our results suggest that patients’ ratings of their own fatigue and constipation have more prognostic value than clinicians’ ratings of these symptoms. Further, the prognostic value of clinicians's ratings can be improved by combining them with patients’ assessments for the symptoms pain, fatigue, constipation, nausea and vomiting. No significant financial relationships to disclose.