Abstract
OBJECTIVES: People with serious illness approaching end of life often end up in emergency and hospital care, frequently against expressed preferences. Consequently, oncology trials record care days as a measure of treatment burden called "time toxicity." However, this measure ignores the diminishing marginal utility of time: that the value of a day is higher when fewer remain. We aimed to incorporate this relationship into time toxicity.
METHODS: We derived a preference-independent time-toxicity metric sensitive to time remaining, end-weighted time toxicity. It has units of toxiles and is based on the fraction of the patient's remaining time taken up by a care visit. We trialed this metric in a cohort of Swedish cancer patients referred to palliative care.
RESULTS: There were 192 patients in the cohort; 146 (76%) eventually enrolled in palliative care. In the cohort, mean (SD, skewness) and median (IQR) end-weighted time toxicities were 0.63 (0.89, 1.77) and 0.23 (0.03-0.81) toxiles. End-weighted time toxicity was significantly less for the group eventually enrolled in palliative care (0.52 [0.75] and 0.22 [0.04-0.63] toxiles) than for the never enrolled group (0.97 [1.18] and 0.27 [0.27-1.76] toxiles) (P = .024). A time-series analysis showed that total toxicity in never- or not-yet-enrolled patients was 120.2 toxiles, whereas total toxicity in already-enrolled patients was 1.0 toxiles.
CONCLUSIONS: A healthcare metric that incorporates diminishing marginal utility of time can highlight the effect of palliative care and could improve resource allocation and patient satisfaction with care near end of life. Further research should explore stakeholder perspectives and practical applications.