Given the aging of our population, the prevalence of both prostate cancer and other age-associated diseases is rapidly increasing. Measuring the impact of comorbidities on quality of life (QOL) will be crucial for assessing cancer treatments in the future. QOL in the form of utility-assessment for use in cost-effectiveness analysis (CEA) is of central importance in decision-making for prostate cancer. However, direct elicitation of utilities for comorbidities, or joint health states (JS), may pose substantial respondent burden. A prediction model for JS utilities based on single health state (SS) utilities would be extremely valuable. Current prediction models for estimating JS utilities from SS utilities are inconsistent with our data for men at risk for prostate cancer. We developed and validated a straightforward linear index model in a prior study, which improves upon other current models. Considering utilities in terms of utility losses l(.) relative to prefect health, our linear index is: E{l(JS)} = 0.05 + 0.72max{l(SS1),l(SS2)} + 0.33min{l(SS1),l(SS2)} - 0.18l(SS1)l(SS2). Its parameters have a theoretical psychological basis in more heavily weighting the more severe component of a JS, suggesting the model may be quite general. Testing the extension of this model for predicting JS utilities is needed. Using individual-level data on utilities for prevalent health states associated with men at risk for prostate cancer, we propose to test the generalizability of our new model in three ways. First, we will test across both disease-specific and non-specific comorbidities prevalent in older men. Second, we will test across a wider range of severities in each comorbid condition. Third, we test in a more general male population to include those without prostate cancer but still in the age range to be at risk for the conditions we will ask about. These additional scenarios include three different health domains and two severities within each domain. More specifically, the new comorbidities include metastatic disease spread (biochemical cancer recurrence and painful metastatic disease), stroke (mild and severe), and functional losses (dependence in the single activity of daily living (ADL) of transferring and dependence in all 6 standard ADLs), each of which will be added to the common prostate cancer outcome of impotence. They will provide an important test of the generalizability of our model. We propose an important test of the extension of our prediction model for establishing better QOL measurement for CEA for men at-risk for prostate cancer. If this model proves more general, it could next be tested in an even broader context across cancer sites and other prevalent non-cancer comorbidities. If it is more limited, it will still guide us regarding the incorporation of comorbidities into CEA for prevalent diseases. Either way, we will be far more able to incorporate QOL associated with comorbidities into CEA.
Prostate cancer treatment decisions require weighing the impact of damages to quality of life from the disease, treatment side-effects, and related or unrelated comorbidities that are prevalent in older men at risk for the disease. How to measure the effects of simultaneous factors on quality of life for incorporation into cost- effectiveness analysis is not yet clear. We have developed a straight-forward prediction model for estimating quality of life for men in various states of having and treating prostate cancer and various comorbidities. This study proposes to test the robustness of this estimation methodology across common comorbidities.
