Practical problems with the collection and interpretation of serial quality of life assessments in patients with malignant glioma

Objectives: To assess the problems involved with the collection and interpretation of serial collected health related quality of life assessments in patients with malignant glioma. Patients and methods: One-hundred and fifty nine patients with malignant glioma from three Scottish neurosurgical centres in whom assessments of performance status, neurological impairment, mood, and quality of life had been recorded over a 6-month period were prospectively identified. The amount of missing data and the reasons for missing data were assessed. Characteristics of patients that were fully compliant with serial assessments were then compared with those that were not. Results: Compliance with serial assessments (both patient and observer reported) was poor, dropping to less than 50% at 6 months. Observer reported measures showed a similar pattern of decline as patient reported measures. The largest single cause of missing data (approximately 70%) was due to administrative failure. Compliant patients were found to have a significantly greater probability of survival compared to non-compliant patients and were also found to be younger and fitter relative to the rest of the study population. Conclusions: Studies utilising quality of life outcomes should give early consideration to minimising avoidable sources of missing data and recording the reasons for non-compliance. Quality of life studies basing conclusions on a complete case analysis should be wary of possible bias.