Utility Scores for Comorbid Conditions: Methodological Issues and Advances

Abstract:

Utility scores quantify health-related quality of life (HRQOL) along a continuum that typically ranges from 0.0 (dead) to 1.0 (full health), and are essential in developing summary measures of population health (SMPH), as well as performing cost-effectiveness analysis (CEA) of different treatments and intervention strategies. A key methodological issue is that traditionally, utility scores have been developed primarily for single health conditions, even though comorbidities are common in both general and patient populations.

Inaccuracies in health measurement are likely to occur when comorbidity is ignored in the estimation of utility scores. In this chapter, methodological issues and advances with regard to deriving utility scores for comorbid health conditions are reviewed.

Direct utility elicitation protocols such as the standard gamble (SG) or time trade-off (TTO) are the most theoretically desirable approaches, but are cognitively burdensome for raters. With population survey data, scores from utility-based HRQOL instruments (e.g., the Health Utilities Index) can often be computed for self-reported comorbidities, but this strategy is often constrained by the limited number of conditions queried, as well as potentially compromised by self-report bias. Another suggested, yet little-used strategy is to map the expected impact of a given comorbidity into the descriptive system of a generic, multiattribute utility instrument, and then compute the corresponding utility score with the scoring algorithm.

Convenient mathematical models (e.g., additive, multiplicative, minimum) for combining single-condition utility scores have also been proposed, but the empirical evidence for their performance is mixed, as well as difficult to assess due to a lack of standardization in utility instrumentation and analytical procedures used. An “encompassing” mathematical model that subsumes traditional models as special cases appears to be more accurate, but has only been examined with respect to directly elicited utilities in the prostate cancer context. A crucial next step is evaluating its performance with respect to a wider variety of health conditions and data sources.

In future work on evaluating and refining methods for obtaining comorbidity-related utilities, cross-study comparability can be enhanced by striving for more consistency in utility instrumentation and analytical techniques.