Maksat Jumamyradov

Objectives: In 2013, the EQ-5D-Y-3L valuation study conducted by Craig and colleagues (ie, the original study) of child health-related quality of life (HRQoL) revealed that U.S. respondents often found it burdensome and guilt-inducing to choose between hypothetical health problems of children. This study introduces an alternative approach where respondents sequentially relieve the health problems of a 10-year-old child for 1 week. Methods: We conducted a discrete choice experiment (DCE) survey (N=631) with paired comparisons and kaizen tasks. Each kaizen task displayed a single profile of a child’s HRQoL using the EQ-5D-Y-3L descriptive system and asked respondents to select first, second, and third improvements for the child’s problems. Combining the preference evidence, a conditional logit model was estimated to produce EQ-5D-Y-3L values on an “experience” scale, where positive values signify experiences better than “being in a coma” and negative values worse. Results: All 10 main effects were statistically significant (P<0.01), with the highest value placed on alleviating pain and discomfort. The worst-case scenario (33333) had a value of −0.337 on the experience scale, indicating it is worse than a coma. These new estimates highly correlate with the original U.S. EQ-5D-Y-3L values (Pearson correlation=0.726; Spearman correlation=0.794). Conclusion: This innovative approach to child health valuation replaces paired comparisons with Kaizen tasks, reducing respondent burden and study costs. Its use of experience scaling, instead of QALYs, aligns with U.S. guidelines (eg, the Inflation Reduction Act of 2022) and summarizes child HRQoL gains for health technology assessment.

Background A key challenge in value assessment is how to summarize effectiveness, particularly the impact of interventions on patient health-related quality of life (HRQoL). One approach is to quantify the gains in HRQoL and life expectancy together as quality-adjusted life years (QALYs); however, this approach has faced various criticisms regarding its potential discriminatory aspects toward persons with disabilities, older adults, and the most vulnerable individuals in society. Methods Instead of QALYs, we provide an alternative approach that summarizes HRQoL gains from the perspective of its stakeholders (e.g., patients, parents, and caregivers) using an “experience” scale. On an experience scale, a positive value signifies an experience better than having no experience at all, while a negative value indicates an experience worse than having no experience. To illustrate the merits of this approach, we examine US preferences on the relief of child health problems, namely a discrete choice experiment (DCE) with kaizen tasks and alternatives described using the EQ-5D-Y-3L. Results Using this approach, we demonstrate the differences in perspectives between parents (N = 179), mothers (N = 99), and fathers (N = 80) of children younger than 18 years of age, as well as the feasibility of this patient-centered approach using a brief DCE survey of less than 100 respondents each (and without QALYs). Specifically, we found that mothers place a higher value on the child’s feelings than fathers. The results also suggest other differences between the perspectives of mothers and fathers, but these differences were not statistically significant (p-values < .05). Conclusions We put forth that future value assessments may summarize gains in HRQoL on a patient experience scale (i.e., experience scale from the patient perspective) to inform decision-making.

Objectives: To estimate values on a quality-adjusted life year (QALY) scale using individual preference evidence, choice analyses typically include ancillary parameters, such as scale factors and discount rates. These parameters potentially differ among respondents. In this study, we investigated how allowing heterogeneity in scale and rate affects the estimation of EQ-5D-5L values. Methods: Using the first wave of the 2016 EQ-5D-5L valuation study (N = 1017), we estimated a conditional logit (CL) model and three mixed logit models: random scale, random rate, and bivariate. Prior to the exploratory study, we hypothesized that scale and rate are correlated and that allowing heterogeneity in both parameters decreases the number of insignificant incremental effects. We confirmed the exploratory findings by re-estimating these models using paired comparison responses from a second wave (N = 1229). Results: Scale and rate exhibited significant heterogeneity and were positively correlated. As hypothesized, allowing this heterogeneity improved the face validity of the EQ-5D-5L value set by reducing the number of insignificant incremental effects (from 6 to 2 p-values > 0.05; out of 20). Nevertheless, the CL and bivariate mixed logit estimates are highly correlated and concordant (Pearson correlation coefficient of 0.897, Spearman correlation coefficient of 0.888, Lin’s concordance coefficient of 0.763). Conclusions: Allowing this heterogeneity adds three parameters to the estimation (two variances and a correlation) and improves the face validity of the EQ-5D-5L values. This finding may influence experimental design and choice analysis in health valuation more generally.

In discrete choice experiments (DCEs), differences between respondents’ preferences may be associated with observable or unobservable factors. Unobservable heterogeneity, related to latent factors associated with the choices of individuals, may be modelled using correlated (i.e. informative heterogeneity) or uncorrelated (i.e. uninformative heterogeneity) individual-specific parameters of a logit model. In this study, we simulated unobservable heterogeneity among DCE respondents and compared the results of the maximum simulated likelihood (MSL) estimation of the mixed logit model when correctly specified and mis-specified. These results show that the MSL estimates are biased and can differ greatly from the true parameters, even when correctly specified. Before estimating a mixed logit model, we highly recommend that choice modellers conduct simulation analyses to assess the potential extent of biases before relying on the MSL estimates, particularly their variances and correlations, and then ultimately determine which model specification produces the least bias.

In a recent study, it was demonstrated that the maximum simulated likelihood (MSL) estimator produces significant biases when applied to the bivariate normal and bivariate Poisson-lognormal models. The study’s conclusion suggests that similar biases could be present in other models generated by correlated bivariate normal structures, which include several commonly used specifications of the mixed logit (MIXL) models. This paper conducts a simulation study analyzing the MSL estimation of the error components (EC) MIXL. We find that the MSL estimator produces significant biases in the estimated parameters. The problem becomes worse when the true value of the variance parameter is small and the correlation parameter is large in magnitude. In some cases, the biases in the estimated marginal effects are as large as 12% of the true values. These biases are largely invariant to increases in the number of Halton draws.

Health preference research (HPR) is the subfield of health economics dedicated to understanding the value of health and health-related objects using observational or experimental methods. In a discrete choice experiment (DCE), the utility of objects in a choice set may differ systematically between persons due to interpersonal heterogeneity (e.g., brand-name medication, generic medication, no medication). To allow for interpersonal heterogeneity, choice probabilities may be described using logit functions with fixed individual-specific parameters. However, in practice, a study team may ignore heterogeneity in health preferences and estimate a conditional logit (CL) model. In this simulation study, we examine the effects of omitted variance and correlations (i.e., omitted heterogeneity) in logit parameters on the estimation of the coefficients, willingness to pay (WTP), and choice predictions. The simulated DCE results show that CL estimates may have been biased depending on the structure of the heterogeneity that we used in the data generation process. We also found that these biases in the coefficients led to a substantial difference in the true and estimated WTP (i.e., up to 20%). We further found that CL and true choice probabilities were similar to each other (i.e., difference was less than 0.08) regardless of the underlying structure. The results imply that, under preference heterogeneity, CL estimates may differ from their true means, and these differences can have substantive effects on the WTP estimates. More specifically, CL WTP estimates may be underestimated due to interpersonal heterogeneity, and a failure to recognize this bias in HPR indirectly underestimates the value of treatment, substantially reducing quality of care. These findings have important implications in health economics because CL remains widely used in practice.

This paper finds that the maximum simulated likelihood (MSL) estimator produces substantial biases when applied to the bivariate normal distribution. A specification of the random parameter bivariate normal model is considered, in which a direct comparison between the MSL and maximum likelihood (ML) estimators is feasible. The analysis shows that MSL produces biased results for the correlation parameter. This paper also finds that the MSL estimator is biased for the bivariate Poisson-lognormal model, developed by Munkin and Trivedi (1999. “Simulated Maximum Likelihood Estimation of Multivariate Mixed-Poisson Regression Models, with Application.” The Econometrics Journal 2: 29–48). A simulation study is conducted, which shows that MSL leads to serious inferential biases, especially large when variance parameters in the true data generating process are small. The MSL estimator produces biases in the estimated marginal effects, conditional means and probabilities of count outcomes.