How do you measure changes in quality of life for patients with cancer? Today we will review 3 different approaches: health state based, health state with age decrements, and health state with time-to-death (TTD) decrements.
Health state based. One common approach is just to use the quality of life of a given health state. For instance, in a simple 3 state model (i.e., stable disease or pre-progression, post-progression, death), one has a quality of life associated with each of these health states where health utilities prior to progression are higher than after progression and utilities for death are 0. This approach is simple and one can readily extrapolate the impact of a new treatment using this approach based on limited clinical trial data. The problem with this approach is that if people live longer, it assumes that there is not change in their utility as long they remain in that health state. In practice, as people age, quality of life will likely decline, however.
Health state with age decrements. An alternative approach would adjust the health state utility based on patient’s age. While many models adjust for background mortality, this approach adjusts quality of life while alive for “background morbidity”. One can readily calculate these utility decrements by regressing age on general utility measures (e.g., EQ-5D, SF-36, SF-12, SF-6D) and examining how utility changes over time. A number of papers have implemented this approach including Jakubowiak et al. 2016, May et al. 2017, Pil et al. 2016, Helou et al. 2017, Keller et al. 2016, and Andronis et al. 2017, among others. One problem with this approach, however, is that health utility likely does not decline in a linear manner. Instead, often certain events happen–often close to death–that significantly affect utility. Additionally, treatments that have longevity improvements are valued less in CEA models under this approach since when people live longer, the magnitude of the utility decrements increase due to higher rates of background morbidity.
Health state with time to death (TTD) utility decrements. In the TTD approach, the background utility decrements occur not as a function of age but as a function of time to death. Thus, if one lives longer, the utility in the last years of life is the same, but those disutilities are pushed out into the future. Increased quality of life gains occur based on the standard health state utilities adjusted for time to death. In fact, Gheorghe et al. 2015 argues that utility based on time to death better fits real-world utility patterns over time.
A paper by Versteegh et al. (2022) aims to estimate the relationship between quality of life decrements based on time until death. They use QoL and survival data from the Patient Reported Outcomes Following Initial Treatment and Long-Term Evaluation of Survivorship registry, which is linked to The Netherlands Cancer Registry. Using both linear regression models and beta regression models, the authors find that:
The mean QoL value in the SF-6D data set decreased from 0.778 if TTD is .60 months to 0.594 if TTD is 0 to 3 months, a 24% decrease. In the EQ-5D data set, the decrease is 18%, from 0.835 if TTD is .60 months to 0.682 if TTD is 0 to 3 months
The author found that utility did decline with age but the effect of TTD was a much better predictor of quality of life than age alone. Further, there were some differences in the impact of TTD on quality of life by tumor types, with the gradient being largest for multiple myeloma and shallowest for non-Hodgkin lymphoma. The full article is here.