Systematic bias in cancer patient-reported outcomes: symptom 'orphans' and 'champions'.

INTRODUCTION: Patient-reported outcomes are an integral part of modern healthcare. We report a comparison of symptom item content from several validated cancer instruments to that of a published checklist, and identify the important differences this revealed.

METHODS: We defined orphans as any symptom not assessed in any of the six selected instruments; champions, in contrast, were present in all six. An empirically derived cancer multisymptom checklist was used.Three symptom categories were identified Orphan-absent from all 6,Champion-present in all 6,Intermediate-underreported.The contents of each validated instrument were cross-referenced against the checklist.

RESULTS: Eighteen (39%) checklist symptoms were orphans. Five (11%) were champions: fatigue, pain, anorexia, dyspnoea and nausea. Of the 46 checklist symptoms, 23 (50%) were inconsistently recorded. All 18 orphan symptoms were clinically important checklist symptoms. Common gastrointestinal and neuropsychological symptoms that pose great clinical management challenges were among the orphan symptoms.

CONCLUSIONS: Of 46 checklist cancer symptoms, over a third were orphans. All orphan symptoms were rated as clinically important and distressing by checklist. Only 5 checklist symptoms were champions: fatigue, pain, anorexia, dyspnoea and nausea. Important-but usually omitted-symptoms included early satiety (6 of 6) and weight loss (5 of 6). The bias appears to disproportionately affect gastrointestinal and neuropsychiatric symptoms. Symptom studies should specifically report the limitations and account for the inherent item bias of any instrument used.