Cognitive interviewing of bereaved relatives to improve the measurement of health outcomes and care utilisation at the end of life in a mortality followback survey.

PURPOSE: The evaluation of end of life care via bereaved relatives is increasingly common. We aimed to improve the measurement of health outcomes and care utilisation at the end of life with this population.
METHODS: A cross-sectional study of 20 bereaved relatives of cancer patients was conducted. In phase I, nine underwent a cognitive interview for 65-150 min using five measures: Client Service Receipt Inventory, Life Before Death survey care satisfaction items, Palliative care Outcome Scale (POS), EQ-5D and Core Bereavement Items (CBI). In phase II, items/scales were revised and tested with a further 11 participants. Content analysis explored information processing and errors.
RESULTS: All 20 participants understood most questions despite finding them demanding and intense. Judging the accuracy of information and formulating final answers posed more difficulties than recalling facts. Uncertainty and missing data on care utilisation were often due to estimating averages; consequently categories were introduced. Participants perceived the care satisfaction rating scale to be positively biased; a very poor category was added. POS was seen to ask relevant but sometimes difficult questions, whilst the EQ-5D produced fewer missing data but lacked intermediate categories. CBI had terms that participants felt were unfamiliar, unrealistic, unclear or too strong; hence, we changed to the Texas Revised Inventory of Grief (with no problems identified).
CONCLUSIONS: Cognitive interviewing helped select measures and make changes that improved the measurement of health outcomes and care utilisation at the end of life with bereaved relatives. The use of both generic and specific health measures and cost measurement in bereavement is encouraged.