Selection of ultimately ill cancer patients able to fulfill a questionnaire: Identification of inherent biases.

Aim: Physical or psychological well-being is an essential component of quality care assessment in palliative unit. This assessment is mainly based on self-assessment (questionnaires or interviews). The aim of this study is to compare the clinical characteristics of patients able to fulfill a questionnaire and those not able to do that.

Methods: The clinical characteristics of 166 cancer patients admitted in palliative care unit from December 2006 to February 2008 have been collected. Characteristics of patients able to fulfill a questionnaire (80, 48.2%) have been compared to other patients (86, 51.8%). Moreover, functional independence measure (FIM) had been evaluated by nurses.

Results: Median age (60 versus 62) and sex ratio (40/40 versus 42/44) are similar in both groups. Lung primaries are significantly less frequent in patients able to fulfill the questionnaire (4% versus 17%, P=0.005). Patients able to fulfill the questionnaire had had better performance status (Karnofsky Index≤30%: 54% versus 21%, P<0.0001). The total score of FIM (56.0 versus 91.5, P<0.00001) and the median overall survivals (2.3 weeks versus 6.6 weeks, P=0.0001) were significantly lower in the group of patients non able to fulfill the questionnaire.

Conclusions: Patients able to fulfill a questionnaire represent only 48.2% of all consecutive admitted patients. These patients are not representative of all patients since they had better performance status, they are less dependent and they display significant better survival. We have to think about new methods to avoid the biases generated by the use of patient-reported outcomes.