A proposal to solve the comparability problem in cost-utility analysis

Comparing the Noncomparable: The Need for Equivalence Measures That Make Sense in Health-Economic Evaluations

BACKGROUND

The popularity of quality-adjusted life years (QALYs) has been resistant to concerns about validity and reliability. Utility-theoretic outcome equivalents are widely used in other areas of applied economics. Equivalence values can be derived for time, money, risk, and other metrics. These equivalence measures preserve all available information about individual preferences and are valid measures of individual welfare changes.

OBJECTIVE

The objective of this study was to derive alternative generalized equivalence measures from first principles and illustrate their application in an empirical comparative-effectiveness example.

METHODS

We specify a general-equilibrium model incorporating neoclassical utility functions, health production function, severity-duration preferences, and labor-market tradeoff function. The empirical implementation takes advantage of discrete-choice experiment methods that are widely accepted in other areas of applied economics and increasingly in health economics. We illustrate the practical significance of restrictive QALY assumptions using comparative-effectiveness results based on both QALYs and estimates of welfare-theoretic time-equivalent values for anti-tumor necrosis factor and prolonged corticosteroid treatments for Crohn's disease in three distinct preference classes.

RESULTS

The QALY difference between the two treatments is 0.2 months, while time-equivalent values range between 0.5 and 1.3 months for aggregate and class-specific differences. Thus, the QALY-based analysis understates welfare-theoretic values by 60%-85%.

CONCLUSION

These results suggest that using disease-specific equivalence values offer a meaningful alternative to QALYs to compare global outcomes across treatments. The equivalence values approach is consistent with principles of welfare economics and offers several features not represented in QALYs, including accounting for preference nonlinearities in disease severity and duration, inclusion of preference-relevant nonclinical healthcare factors, representing preferences of clinically-relevant patient subpopulations, and including utility losses related to risk aversion.

Hidden costs: The ethics of cost-effectiveness analyses for health interventions in resource-limited settings

Cost-effectiveness analysis (CEA) is an increasingly appealing tool for evaluating and comparing health-related interventions in resource-limited settings. The goal is to inform decision-makers regarding the health benefits and associated costs of alternative interventions, helping guide allocation of limited resources by prioritising interventions that offer the most health for the least money. Although only one component of a more complex decision-making process, CEAs influence the distribution of health-care resources, directly influencing morbidity and mortality for the world's most vulnerable populations. However, CEA-associated measures are frequently setting-specific valuations, and CEA outcomes may violate ethical principles of equity and distributive justice. We examine the assumptions and analytical tools used in CEAs that may conflict with societal values. We then evaluate contextual features unique to resource-limited settings, including the source of health-state utilities and disability weights, implications of CEA thresholds in light of economic uncertainty, and the role of external donors. Finally, we explore opportunities to help align interpretation of CEA outcomes with values and budgetary constraints in resource-limited settings. The ethical implications of CEAs in resource-limited settings are vast. It is imperative that CEA outcome summary measures and implementation thresholds adequately reflect societal values and ethical priorities in resource-limited settings.

Acceptability of less than perfect health states in rheumatoid arthritis: the patients' perspective

Some health problems are considered by many individuals as a 'normal' part of ageing. Our aim was to investigate whether patients with rheumatoid arthritis (RA) consider different types and levels of health losses as acceptable beyond a certain age. A multicenter cross-sectional survey was performed involving RA patients at the initiation of the first biological therapy. The EQ-5D and the Health Assessment Questionnaire Disability Index (HAQ-DI) questionnaires were used to describe domain-specific health states. Patients were asked to indicate for each domain from what age and onward (between ages 30 and 80 years in 10 year intervals) they considered moderate and severe problems acceptable or alternatively never acceptable. Seventy-seven RA patients (females 86%, mean age 50.3, disease duration 9.1 years) completed the questionnaire. Disease activity (DAS28), EQ-5D and HAQ-DI scores were mean 6.00 (SD 0.85), 0.35 (SD 0.36), 1.48 (SD 0.66), respectively. The majority of the patients considered age 70 and beyond as acceptable to have some health problems (EQ-5D: self-care 42%, pain/discomfort 34%, mobility 33%, usual activities 33%, anxiety/depression 27%), whilst at ages 30 and 40 as not acceptable. Severe health problems were mostly (57-69%) considered never acceptable, except the 'Usual activities' domain (acceptable from age 80 by 50.6%). The great majority of the patients (77-96%) were younger than what they indicated as the acceptability age limit. Similar results were found for the HAQ-DI. This small experimental study suggests that RA patients consider some health problems acceptable. This acceptability is age related and varies by health areas. Further larger studies are needed to explore explanatory variables and to compare with other diseases. Owing to the impact acceptability might have on RA patients' self-evaluation of current health state and decision-making, the topic deserves methodological improvement and further investigation.

Estimating production costs in the economic evaluation of health-care programs

We propose a method for calculating the production costs of an intervention in a manner that accounts for differences in productive 'effort.' This method could be used within a cost-effectiveness analysis framework in the evaluation of new medical technologies, pharmaceuticals, treatment programs, or public health interventions. We apply it to show evidence in favor of implementing a newborn screening program to detect congenital hearing impairment.

Explaining the characteristics of the power (CRRA) utility family

The power family, also known as the family of constant relative risk aversion (CRRA), is the most widely used parametric family for fitting utility functions to data. Its characteristics have, however, been little understood, and have led to numerous misunderstandings. This paper explains these characteristics in a manner accessible to a wide audience.

The effect of age, race and gender on preference scores for hypothetical health states

PURPOSE

Preferences are known to vary by individuals' personal experience with a health state, but variation among respondents' scoring of the same hypothetical state is unproven but relevant to the use of community-perspective preference scores. This research explored the systematic contribution of respondents' age, race and gender to variability in community perspective preferences for hypothetical health states.

METHODS

Data from four community samples were pooled for the analysis. Linear regression modeling was used to test for the effect of respondent age, race and gender on preference scores while controlling for health state severity.

RESULTS

In this sample of 956 preference scores from 390 individuals across 4 studies, older respondents provided lower preference scores for the same hypothetical health state compared with younger respondents (regression coefficient for 1 year of age = -0.002, p < 0.001), and white individuals provided higher preference scores for the same states compared with non-white individuals (regression coefficient = 0.056, p = 0.014).

CONCLUSION

Preferences for hypothetical health states may vary by the age and race of the respondent providing the score. Community-perspective preferences should thus be elicited from large, random samples of the relevant population to ensure variation on these as well as other yet-unidentified characteristics that may affect scores.

Acceptability of less than perfect health states

Health normally deteriorates beyond a certain age. This means, in Amartya Sen's terms, that one's health capabilities decline beyond a certain age, making it more difficult to achieve functionings such as mobility or sexual activity. In this paper, we investigate whether this normal reduction in quality of life also induces less than perfect health states to be considered acceptable at advanced stages of life. In other words, we investigate whether it is considered acceptable that health capabilities decline over time. In this study, we use domain-specific descriptions of health (mostly following the EQ-5D domains) in order to investigate whether the acceptability of less than perfect health states is similar for all types of health losses. Besides a theoretical consideration of this issue, we present some empirical evidence based on the answers of 226 respondents to a web-based survey. The results show that often individuals do indeed consider less than perfect health states acceptable, especially at more advanced stages of life. Mild health problems are more often considered acceptable than severe health problems. The acceptability of health states is related to the quality of life score of these states, i.e., worse states are considered less acceptable. This may have implications for the allocation of scarce health care resources.

Validity and feasibility of the use of condition-specific outcome measures in economic evaluation

BACKGROUND

Usually, generic questionnaires such as the EQ-5D or Health Utility Index (HUI) are used to obtain utility scores for computing QALYs. Sometimes, however, application of these instruments is not possible, or the responsiveness is doubted. An alternative strategy is to attribute utility scores to health states of a condition-specific outcomes measure (CSOM). We explored the validity and feasibility of this strategy.

RESEARCH DESIGN

Our samples determined utility scores for the health states of the International Index of Erectile Function (IIEF) using time tradeoff (TTO). To reduce costs and time, the general population (n = 169) was interviewed in groups. We tested the validity of the group sessions in students. To test the extent of agreement between values obtained using the group and those obtained through individual administration, 63 students were interviewed individually and 54 in groups.

RESULTS

The utility scores for the disease-specific health states showed good construct validity. Also, the criterion validity of the adapted TTO was confirmed.

DISCUSSION

Disease-specific utility scores can be used in QALY analysis by converting them to a generic scale. Efforts should be undertaken to prevent response spreading. Administrating TTO in groups could reduce the time and costs of TTO administration and render the strategy of determining utilities for condition-specific health states more feasible.