Accounting for the increasing value of time as the prognosis shortens

Background: Healthcare contact time outside the home – “time toxicity” – has been proposed to measure treatment burden in oncology clinical trials; however, this measure ignores patients’ increasing valuation of time as the prognosis shortens. An analogous intuition was formalized by Daniel Bernoulli in 1738 as the “logarithmic utility of money.” Losing 20 euros harms a pauper more than a millionaire. Similarly, the relative value of a day may be higher when a patient’s prognosis is 2-3 weeks than when it is 2-3 months.
Aim/Research question or hypothesis: To deepen the conceptualization of time toxicity, we aimed to display the increasing disutility (toxicity) of unplanned healthcare time as the prognosis shortens.
Methods: Our demonstration population is 81 high-utilization patients with mixed cancer diagnoses referred to a home-based palliative care team. Medical record review provided dates of emergency and unplanned hospital care. We implemented the Bernoulli-analogous time valuation by plotting each patient’s illness trajectory from diagnosis until death on a logarithmic timescale. Emergency and unplanned hospital care was marked with thick red lines (Figure 1, bottom). Because of the logarithmic timescale, the toxicity of an unplanned healthcare visit is the red line’s length. Stacking (adding) the toxicity lines produced a graph of the number of patients in unplanned healthcare versus time (Figure 1, top).
Results: 81 patient trajectories could be visualized for healthcare time toxicity.
The highest number of patients (in our demonstration population) are using unplanned healthcare in the 38 to 8 days before death.
Discussion: Palliative care involvement that reduces time toxicity could be particularly beneficial when patients’ prognosis is 2 to 3 months, shortly before our demonstration population shows the highest number of patients using unplanned healthcare. Relevant approaches would include home-based care and support for in-the-moment decision making. Further work with larger and differently selected populations would help judge the generality of this recommendation.