How reliable are relatives' retrospective reports of terminal illness? Patients and relatives' accounts compared.

To assess the accuracy of relatives' recollections of patients' terminal illness 71 out of 77 caring relatives were re-interviewed about 4 months after they and the patients had given regular interviews throughout care. Current and retrospective ratings of problems and feelings have been compared for agreement, using the kappa index. Several volunteered symptoms showed poor agreement, notably pain, anorexia and depression (kappa = 0.03-0.21), but vomiting, dyspnoea and immobility ratings agreed moderately well (kappa = 0.43-0.68). Current ratings from patients' and relatives' were always in better agreement with each other than with the relatives retrospective ratings. Bias sometimes altered apparent prevalence; pain was described as more severe in retrospect, but weakness, malaise, depression and relatives' stress were under-rated later. Ratings of "discomfort only" became less common for all symptoms retrospectively. The regular current assessments of patients' and relatives' emotional state also agreed only slightly with relatives' follow-up accounts of depression but somewhat better for anxiety. Patients stated knowledge of diagnosis, awareness and acceptance of dying matched the relatives' retrospective assessments moderately well (kappa = 0.70, 0.50 and 0.41). This study and other available evidence indicate that relatives' retrospective reports of terminal illness, measured against current ratings, are moderately reliable for some items but can vary or be potentially misleading over other aspects, including pain. This could affect evaluations of care.