Purpose: Adverse events in chemotherapy clinical trials are assessed and reported by clinicians, yet clinician accuracy in assessing symptoms has been questioned. We compared patient reporting of eight symptoms using a validated instrument, the European Organization for the Research and Treatment of Cancer Quality-of-Life Questionnaire C30 (QLQ-C30 or QLQ) with physicians' reporting of the same symptoms in the study's adverse events log.
Patients and methods: Thirty-seven men with metastatic, androgen-independent prostate cancer enrolled onto a phase II trial of weekly calcitriol and docetaxel completed the QLQ every 4 weeks for up to 28 weeks. A patient-reported symptom was defined as an increase in a QLQ symptom score by at least 10 points (0 to 100 scale), sustained for at least 4 weeks. A physician-reported symptom was considered present if it was ever documented in the adverse event log.
Results: Forty-nine (new or worsened) symptoms were detected by both physician and QLQ, 48 symptoms were detected by the physician alone, and 55 symptoms were detected by the QLQ alone. They agreed on the absence of a symptom in 102 instances of 254 possible opportunities. Their uncorrected agreement was 59.4%, but Cohen's kappa, a coefficient of agreement that corrects for chance, was 0.15, indicating only slight agreement. Using the QLQ as the standard, overall physician sensitivity and specificity was 47% and 68%, respectively, although it varied considerably among symptoms.
Conclusion: Even in a tightly controlled clinical trial, physician reporting was neither sensitive nor specific in detecting common chemotherapy adverse effects. Tools for collecting patient-reported adverse event data in chemotherapy clinical trials should be developed.