Are Current Reporting Standards Used to Describe Health State Utilities in Cost-Effectiveness Models Satisfactory?

OBJECTIVE: The aims of this study were to examine current reporting standards of health state utilities (HSU) using a review of published cost-effectiveness analyses in cardiovascular disease and to explore the impact of variation in model inputs used in these on estimated quality-adjusted life-years (QALYs) and cost-effectiveness. METHODS: Key health/economics bibliographic databases were searched to identify relevant articles published after 2014. Any narrative or values relating to the HSU used in the model were extracted and reviewed. The HSUs were systematically applied to an existing model to explore the influence of different values on QALYs and the incremental cost-effectiveness ratio. RESULTS: Twenty-four peer-reviewed articles were identified. Only 2 studies referred to a literature review for the HSUs. Most (18 of 24) referenced previously published economic studies (as opposed to the original source) for at least 1 of the HSUs. Only 4 studies referenced the original sources and reported all of the HSUs accurately, and several did not provide all the HSUs. Little information was provided on the methods used to calculate QALYs, for example, the duration of time for acute HSUs, what the baseline HSU was, the method that was used to assign HSUs for subsequent different events, or how constant HSUs for clinical events were combined with age-adjusted baseline values. The huge differences in HSUs used in the studies produced substantial variations in the QALYs and incremental cost-effectiveness ratios generated from the cost-effectiveness model. CONCLUSION: Current standards are poor, and there is a need for greater transparency in reporting the HSUs used in cost-effectiveness models.

An Updated Systematic Review of Studies Mapping (or Cross-Walking) Measures of Health-Related Quality of Life to Generic Preference-Based Measures to Generate Utility Values.

BACKGROUND: Mapping is an increasingly common method used to predict instrument-specific preference-based health-state utility values (HSUVs) from data obtained from another health-related quality of life (HRQoL) measure. There have been several methodological developments in this area since a previous review up to 2007. OBJECTIVE: To provide an updated review of all mapping studies that map from HRQoL measures to target generic preference-based measures (EQ-5D measures, SF-6D, HUI measures, QWB, AQoL measures, 15D/16D/17D, CHU-9D) published from January 2007 to October 2018. DATA SOURCES: A systematic review of English language articles using a variety of approaches: searching electronic and utilities databases, citation searching, targeted journal and website searches. STUDY SELECTION: Full papers of studies that mapped from one health measure to a target preference-based measure using formal statistical regression techniques. DATA EXTRACTION: Undertaken by four authors using predefined data fields including measures, data used, econometric models and assessment of predictive ability. RESULTS: There were 180 papers with 233 mapping functions in total. Mapping functions were generated to obtain EQ-5D-3L/EQ-5D-5L-EQ-5D-Y (n = 147), SF-6D (n = 45), AQoL-4D/AQoL-8D (n = 12), HUI2/HUI3 (n = 13), 15D (n = 8) CHU-9D (n = 4) and QWB-SA (n = 4) HSUVs. A large number of different regression methods were used with ordinary least squares (OLS) still being the most common approach (used ≥ 75% times within each preference-based measure). The majority of studies assessed the predictive ability of the mapping functions using mean absolute or root mean squared errors (n = 192, 82%), but this was lower when considering errors across different categories of severity (n = 92, 39%) and plots of predictions (n = 120, 52%). CONCLUSIONS: The last 10 years has seen a substantial increase in the number of mapping studies and some evidence of advancement in methods with consideration of models beyond OLS and greater reporting of predictive ability of mapping functions.

Identification, Review, and Use of Health State Utilities in Cost-Effectiveness Models: An ISPOR Good Practices for Outcomes Research Task Force Report.

Cost-effectiveness models that present results in terms of cost per quality-adjusted life-year for health technologies are used to inform policy decisions in many parts of the world. Health state utilities (HSUs) are required to calculate the quality-adjusted life-years. Even when clinical studies assessing the effectiveness of health technologies collect data on HSUs to populate a cost-effectiveness model, which rarely happens, analysts typically need to identify at least some additional HSUs from alternative sources. When possible, HSUs are identified by a systematic review of the literature, but, again, this rarely happens. In 2014, ISPOR established a Good Practices for Outcome Research Task Force to address the use of HSUs in cost-effectiveness models. This task force report provides recommendations for researchers who identify, review, and synthesize HSUs for use in cost-effectiveness models; analysts who use the results in models; and reviewers who critically appraise the suitability and validity of the HSUs selected for use in models. The associated Minimum Reporting Standards of Systematic Review of Utilities for Cost-Effectiveness checklist created by the task force provides criteria to judge the appropriateness of the HSUs selected for use in cost-effectiveness models and is suitable for use in different international settings.

The Identification, Review and Synthesis of Health State Utility Values from the Literature.

Systematic literature reviews of health-related quality of life (HRQoL) evidence that are to inform economic models can be challenging due to the volume of hits identified in searches using generic terms for HRQoL. Nevertheless, a robust review of the literature is required to ensure that the health state utility values (HSUVs) used in the economic model are the most appropriate available. This article provides a synopsis of literature relating to identifying, reviewing and synthesising HSUVs. The process begins with scoping the needs of the economic model, including the definitions of health states and the requirements of any reimbursement agencies. A sequence of searches may be required as the economic model evolves. The terminology used for HRQoL measures may be problematic, and as there is no robust HRQoL filter [equivalent to that applied for randomised control trial (RCTs)], sifting the results of sensitive searches can be resource intensive. Alternative approaches such as forward and backward citation searches may reduce the resources required, while maintaining the integrity of the search. Any included studies should be assessed in terms of quality using a recommended checklist, and insufficient detail in the primary studies should be noted as a short-coming in this exercise. Subject to homogeneity (similar populations, same measure and preference weights) evidence can be pooled in some way, although methodological research into the appropriateness of alternative techniques for meta-analysis is in its infancy. Reporting standards are key and as a minimum should include details on searches, inclusion/exclusion criteria (together with rationale for exclusion at each stage), assessment of quality and relevance of included studies, and justification for the choice of final HSUVs.

Estimating Health State Utility Values for Comorbidities.

A comorbidity is defined as the presence of at least one additional health condition co-occurring with a primary health condition. Decision analytic models in healthcare depict the typical clinical pathway of patients in general clinical practice and frequently include health states defined to represent comorbidities such as sequelae or adverse events. Health state utility values (HSUVs) are often not available for these and analysts generally estimate them. This article provides a summary of the methodological literature on estimating methods frequently used together with worked examples. The three main methods used (minimum, multiplicative and additive) can produce a wide range in the values estimated. In general, the minimum method overestimates observed HSUVs and the magnitude of error tends to increase as the observed values decrease. Conversely, the additive and multiplicative methods generally underestimate observed values and the magnitude of the errors is generally greater for the additive method. HSUVs estimated using the multiplicative method tend to decrease for lower HSUVs and the largest errors are in observed HSUVs >0.6. Differences in estimated values can produce substantial differences in the resulting incremental cost effectiveness ratio. Based on the current evidence, the multiplicative method is advocated but additional research is required to determine appropriate methods when estimating values for additional comorbidities.

A Review of Generic Preference-Based Measures for Use in Cost-Effectiveness Models.

Generic preference-based measures (GPBMs) of health are used to obtain the quality adjustment weight required to calculate the quality-adjusted life year in health economic models. GPBMs have been developed to use across different interventions and medical conditions and typically consist of a self-complete patient questionnaire, a health state classification system, and preference weights for all states defined by the classification system. Of the six main GPBMs, the three most frequently used are the Health Utilities Index version 3, the EuroQol 5 dimensions (3 and 5 levels), and the Short Form 6 dimensions. There are considerable differences in GPBMs in terms of the content and size of descriptive systems (i.e. the numbers of dimensions of health and levels of severity within these), the methods of valuation [e.g. time trade-off (TTO), standard gamble (SG)], and the populations (e.g. general population, patients) used to value the health states within the descriptive systems. Although GPBMs are anchored at 1 (full health) and 0 (dead), they produce different health state utility values when completed by the same patient. Considerations when selecting a measure for use in a clinical trial include practicality, reliability, validity and responsiveness. Requirements of reimbursement agencies may impose additional restrictions on suitable measures for use in economic evaluations, such as the valuation technique (TTO, SG) or the source of values (general public vs. patients).

Recommended Methods for the Collection of Health State Utility Value Evidence in Clinical Studies.

A conceptual model framework and an initial literature review are invaluable when considering what health state utility values (HSUVs) are required to populate health states in decision models. They are the recommended starting point early within a research and development programme, and before development of phase III trial protocols. While clinical trials can provide an opportunity to collect the required evidence, their appropriateness should be reviewed against the requirements of the model structure taking into account population characteristics, time horizon and frequency of clinical events. Alternative sources such as observational studies or registries may be more appropriate when evidence describing changes in HSUVs over time or rare clinical events is required. Phase IV clinical studies may provide the opportunity to collect additional longitudinal real-world evidence. Aspects to consider when designing the collection of the evidence include patient and investigator burden, whom to ask, the representativeness of the population, the exact definitions of health states within the economic model, the timing of data collection, sample size, and mode of administration. Missing data can be an issue, particularly in longitudinal studies, and it is important to determine whether the missing data will bias inferences from analyses. For example, respondents may fail to complete follow-up questionnaires because of a relapse or the severity of their condition. The decision on the preferred study type and the particular quality of life measure should be informed by any evidence currently available in the literature, the design of data collection, and the exact requirements of the model that will be used to support resource allocation decisions (e.g. reimbursement).

The Use of Mapping to Estimate Health State Utility Values.

Mapping functions are estimated using regression analyses and are frequently used to predict health state utility values (HSUVs) in decision analytic models. Mapping functions are used when evidence on the required preference-based measure (PBM) is not available, or where modelled values are required for a decision analytic model, for example to control for important sociodemographic variables (such as age or gender). This article provides an overview of the latest recommendations including pre-mapping considerations, the mapping process including data requirements for undertaking the estimation of mapping functions, regression models for estimating mapping functions, assessing performance and reporting standards for mapping studies. Examples in rheumatoid arthritis are used for illustration. When reporting the results of mapping standards the following should be reported: a description of the dataset used (including distributions of variables used) and any analysis used to inform the selection of the model type and model specification. The regression method and specification should be justified, and as summary statistics may mask systematic bias in errors, plots comparing observed and predicted HSUVs. The final model (coefficients, error term(s), variance and covariance) should be reported together with a worked example. It is important to ensure that good practice is followed as any mapping functions will only be as appropriate and accurate as the method used to obtain them; for example, mapping should not be used if there is no overlap between the explanatory and target variables.

The Use of Health State Utility Values in Decision Models.

Methodological issues of how to use health state utility values (HSUVs) in decision models arise frequently, including the most appropriate evidence to use as the baseline (e.g. the baseline HSUVs associated with avoiding a particular health condition or event), how to capture changes due to adverse events and how to appropriately capture uncertainty in progressive conditions where the expected change in quality of life is likely to be monotonically decreasing over time. As preference-based measures provide different values when collected from the same patient, it is important to ensure that all HSUVs used within a single model are obtained from the same instrument where ever possible. When people enter the model without the condition of interest (e.g. primary prevention of cardiovascular disease, screening or vaccination programmes), appropriate age- and gender-adjusted HSUVs from people without the particular condition should be used as the baseline. General population norms may be used as a proxy if the exact condition-specific evidence is not available. Individual discrete health states should be used for serious adverse reactions to treatment and the corresponding HSUVs sourced as normal. Care should be taken to avoid double counting when capturing the effects for both less severe adverse reactions (e.g. itchy skin rash or dry cough) and more severe adverse events (e.g. fatigue in oncology). Transparency in reporting standards for both the justification of the evidence used and any 'adjustments' is important to increase readers' confidence that the evidence used is the most appropriate available.

The Role of Condition-Specific Preference-Based Measures in Health Technology Assessment.

A condition-specific preference-based measure (CSPBM) is a measure of health-related quality of life (HRQOL) that is specific to a certain condition or disease and that can be used to obtain the quality adjustment weight of the quality-adjusted life-year (QALY) for use in economic models. This article provides an overview of the role and the development of CSPBMs, and presents a description of existing CSPBMs in the literature. The article also provides an overview of the psychometric properties of CSPBMs in comparison with generic preference-based measures (generic PBMs), and considers the advantages and disadvantages of CSPBMs in comparison with generic PBMs. CSPBMs typically include dimensions that are important for that condition but may not be important across all patient groups. There are a large number of CSPBMs across a wide range of conditions, and these vary from covering a wide range of dimensions to more symptomatic or uni-dimensional measures. Psychometric evidence is limited but suggests that CSPBMs offer an advantage in more accurate measurement of milder health states. The mean change and standard deviation can differ for CSPBMs and generic PBMs, and this may impact on incremental cost-effectiveness ratios. CSPBMs have a useful role in HTA where a generic PBM is not appropriate, sensitive or responsive. However, due to issues of comparability across different patient groups and interventions, their usage in health technology assessment is often limited to conditions where it is inappropriate to use a generic PBM or sensitivity analyses.

Predicting Productivity Losses from Health-Related Quality of Life Using Patient Data.

OBJECTIVE: This paper estimates productivity loss using the health of the patient in order to allow indirect estimation of these costs for inclusion in economic evaluation. METHODS: Data from two surveys of inpatients [Health outcomes data repository (HODaR) sample (n = 42,442) and health improvement and patient outcomes (HIPO) sample (n = 6046)] were used. The number of days off paid employment or normal activities (excluding paid employment) was modelled using the health of the patients measured by the EQ-5D, international classification of diseases (ICD) chapters, and other health and sociodemographic data. Two-part models (TPMs) and zero-inflated negative binomial (ZINB) models were identified as the most appropriate specifications, given large spikes at the minimum and maximum days for the dependent variable. Analysis was undertaken separately for the two datasets to account for differences in recall period and identification of those who were employed. RESULTS: Models were able to reflect the large spike at the minimum (zero days) but not the maximum, with TPMs doing slightly better than the ZINB model. The EQ-5D was negatively associated with days off employment and normal activities in both datasets, but ICD chapters only had statistically significant coefficients for some chapters in the HODaR. CONCLUSIONS: TPMs can be used to predict productivity loss associated with the health of the patient to inform economic evaluation. Limitations include recall and response bias and identification of who is employed in the HODaR, while the HIPO suffers from a small sample size. Both samples exclude some patient groups.

Cost-effectiveness modelling of telehealth for patients with raised cardiovascular disease risk: evidence from a cohort simulation conducted alongside the Healthlines randomised controlled trial.

OBJECTIVES: To investigate the long-term cost-effectiveness (measured as the ratio of incremental NHS cost to incremental quality-adjusted life years) of a telehealth intervention for patients with raised cardiovascular disease (CVD) risk. DESIGN: A cohort simulation model developed as part of the economic evaluation conducted alongside the Healthlines randomised controlled trial. SETTING: Patients recruited through primary care, and intervention delivered via telehealth service. PARTICIPANTS: Participants with a 10-year CVD risk ≥20%, as measured by the QRISK2 algorithm, and with at least 1 modifiable risk factor, individually randomised from 42 general practices in England. INTERVENTION: A telehealth service delivered over a 12-month period. The intervention involved a series of responsive, theory-led encounters between patients and trained health information advisors who provided access to information resources and supported medication adherence and coordination of care. PRIMARY AND SECONDARY OUTCOME MEASURES: Cost-effectiveness measured by net monetary benefit over the simulated lifetime of trial participants from a UK National Health Service perspective. RESULTS: The probability that the intervention was cost-effective depended on the duration of the effect of the intervention. The intervention was cost-effective with high probability if effects persisted over the lifetime of intervention recipients. The probability of cost-effectiveness was lower for shorter durations of effect. CONCLUSIONS: The intervention was likely to be cost-effective under a lifetime perspective. TRIAL REGISTRATION NUMBER: ISRCTN27508731; Results.

A model for measuring the health burden of classic congenital adrenal hyperplasia in adults.

AIM: Patients with classic congenital adrenal hyperplasia (CAH) have poor health outcomes. In the absence of a comprehensive observational study, this manuscript provides a model to estimate the lifetime disease burden of adults with classic CAH. METHODS: The model, built in Excel, comprises subdomains addressing the health consequences of CAH and synthesises evidence from clinical and epidemiological studies on health outcomes. RESULTS: The model estimates that adults with classic CAH will implement 'sick day rules' (doubling or tripling glucocorticoid and/or use of parenteral therapy) 171 times over their lifetime and attend hospital for adrenal crisis on 11 occasions. In a population of 1000, over 200 will die of a condition complicated by adrenal crisis resulting, on average, in a loss of 7 years of life. Patients with CAH may also suffer from excess CVD events. Treatment with glucocorticoids almost doubles the risk of bone fractures in patients with CAH compared to the general population, leading on average to an additional 0·8 fractures per patient with CAH over their lifetime. CONCLUSIONS: The disease burden model highlights gaps in evidence, particularly regarding intensity of care and adrenal crisis, and the relationship between control of CAH and risks of CVD, osteoporosis, diabetes and infertility. The model can be used for research on the impact of new clinical pathways and therapeutic interventions in terms of clinical events and cost.

Are policy decisions on surgical procedures informed by robust economic evidence? A systematic review.

OBJECTIVES: The aim of this study was to examine the empirical and methodological cost-effectiveness evidence of surgical interventions for breast, colorectal, or prostate cancer. METHODS: A systematic search of seven databases including MEDLINE, EMBASE, and NHSEED, research registers, the NICE Web site and conference proceedings was conducted in April 2012. Study quality was assessed in terms of meeting essential, preferred and UK NICE specific requirements for economic evaluations. RESULTS: The seventeen (breast = 3, colorectal = 7, prostate = 7) included studies covered a broad range of settings (nine European; eight non-European) and six were published over 10 years ago. The populations, interventions and comparators were generally well defined. Very few studies were informed by literature reviews and few used synthesized clinical evidence. Although the interventions had potential differential effects on recurrence and mortality rates, some studies used relatively short time horizons. Univariate sensitivity analyses were reported in all studies but less than a third characterized all uncertainty with a probabilistic sensitivity analysis. Although a third of studies incorporated patients' health-related quality of life data, only four studies used social tariff values. CONCLUSIONS: There is a dearth of recent robust evidence describing the cost-effectiveness of surgical interventions in the management of breast, colorectal and prostate cancers. Many of the recent publications did not satisfy essential methodological requirements such as using clinical evidence informed by a systematic review and synthesis. Given the ratio of potential benefit and harms associated with cancer surgery and the volume of resources consumed by these, there is an urgent need to increase economic evaluations of these technologies.

Predicting preference-based utility values using partial proportional odds models.

BACKGROUND: The majority of analyses on utility data have used ordinary least square (OLS) regressions to explore potential relationships. The aim of this paper is to explore the benefits of response mapping onto health dimension profiles to generate preference-based utility scores using partial proportional odds models (PPOM). METHODS: Models are estimated using EQ-5D data collected in the Health Survey for England and the predicted utility scores are compared with those obtained using OLS regressions. Explanatory variables include age, acute illness, educational level, general health, deprivation and survey year. The expected EQ-5D scores for the PPOMs are obtained by weighting the predicted probabilities of scoring one, two or three for the five health dimensions by the corresponding preference-weights. RESULTS: The EQ-5D scores obtained using the probabilities from the PPOMs characterise the actual distribution of EQ-5D preference-based utility scores more accurately than those obtained from the linear model. The mean absolute and mean squared errors in the individual predicted values are also reduced for the PPOM models. CONCLUSIONS: The PPOM models characterise the underlying distributions of the EQ-5D data better than models obtained using OLS regressions. Additional research exploring the effect of modelling conditional responses and two part models could potentially improve the results further.

Association between body mass index and health-related quality of life, and the impact of self-reported long-term conditions - cross-sectional study from the south Yorkshire cohort dataset.

BACKGROUND: We sought to quantify the relationship between body mass index (BMI) and health-related quality (HRQoL) of life, as measured by the EQ-5D, whilst controlling for potential confounders. In addition, we hypothesised that certain long-term conditions (LTCs), for which being overweight or obese is a known risk factor, may mediate the association between BMI and HRQoL. Hence the aim of our study was to explore the association between BMI and HRQoL, first controlling for confounders and then exploring the potential impact of LTCs. METHODS: We used baseline data from the South Yorkshire Cohort, a cross-sectional observational study which uses a cohort multiple randomised controlled trial design. For each EQ-5D health dimension we used logistic regression to model the probability of responding as having a problem for each of the five health dimensions. All continuous variables were modelled using fractional polynomials. We examined the impact on the coefficients for BMI of removing LTCs from our model. We considered the self-reported LTCs: diabetes, heart disease, stroke, cancer, osteoarthritis, breathing problems and high blood pressure. RESULTS: The dataset used in our analysis had data for 19,460 individuals, who had a mean EQ-5D score of 0.81 and a mean BMI of 26.3 kg/m². For each dimension, BMI and all of the LTCs were significant predictors. For overweight or obese individuals (BMI ≥ 25 kg/m²), each unit increase in BMI was associated with approximately a 3% increase in the odds of reporting a problem for the anxiety/depression dimension, a 8% increase for the mobility dimension, and approximately 6% for the remaining dimension s. Diabetes, heart disease, osteoarthritis and high blood pressure were identified as being potentially mediating variables for all of the dimensions. CONCLUSIONS: Compared to those of a normal weight (18.5 < BMI < 25 kg/m²), overweight and obese individuals had a reduced HRQoL, with each unit increase in BMI associated with approximately a 6% increase in the odds of reporting a problem on any of the EQ-5D health dimensions. There was evidence to suggest that diabetes, heart disease, osteoarthritis and high blood pressure may mediate the association between being overweight and HRQoL.

Good practice guidelines for the use of statistical regression models in economic evaluations.

Decision-analytic models (DAMs) used to evaluate the cost effectiveness of interventions are pivotal sources of evidence used in economic evaluations. Parameter estimates used in the DAMs are often based on the results of a regression analysis, but there is little guidance relating to these. This study had two objectives. The first was to identify the frequency of use of regression models in economic evaluations, the parameters they inform, and the amount of information reported to describe and support the analyses. The second objective was to provide guidance to improve practice in this area, based on the review. The review concentrated on a random sample of economic evaluations submitted to the UK National Institute for Health and Clinical Excellence (NICE) as part of its technology appraisal process. Based on these findings, recommendations for good practice were drafted, together with a checklist for critiquing reporting standards in this area. Based on the results of this review, statistical regression models are in widespread use in DAMs used to support economic evaluations, yet reporting of basic information, such as the sample size used and measures of uncertainty, is limited. Recommendations were formed about how reporting standards could be improved to better meet the needs of decision makers. These recommendations are summarised in a checklist, which may be used by both those conducting regression analyses and those critiquing them, to identify what should be reported when using the results of a regression analysis within a DAM.

Estimating health state utility values for joint health conditions: a conceptual review and critique of the current evidence.

BACKGROUND: Analysts frequently estimate the health state utility values (HSUVs) for joint health conditions (JHCs) using data from cohorts with single health conditions. The methods can produce very different results, and there is currently no consensus on the most appropriate technique. OBJECTIVE: To conduct a detailed critical review of existing empirical literature to gain an understanding of the reasons for differences in results and identify where uncertainty remains that may be addressed by further research. RESULTS: Of the 11 studies identified, 10 assessed the additive method, 10 the multiplicative method, 7 the minimum method, and 3 the combination model. Two studies evaluated just 1 of the techniques, whereas the others compared results generated using 2 or more. The range of actual HSUVs can influence general findings, and methods are sometimes compared using descriptive statistics that may not be appropriate for assessing predictive ability. None of the methods gave consistently accurate results across the full range of possible HSUVs, and the values assigned to normal health influence the accuracy of the methods. CONCLUSIONS: Within the limitations of the current evidence base, we would advocate the multiplicative method, conditional on adjustment for baseline utility, as the preferred technique to estimate HSUVs for JHCs when using mean values obtained from cohorts with single conditions. We would recommend that a range of sensitivity analyses be performed to explore the effect on results when using the estimated HSUVs in economic models. Although the linear models appeared to give more accurate results in the studies we reviewed, these models require validating in external data before they can be recommended.

Using health state utility values in models exploring the cost-effectiveness of health technologies.

BACKGROUND: To improve comparability of economic data used in decision making, some agencies recommend that a particular instrument should be used to measure health state utility values (HSUVs) used in decision-analytic models. The methods used to incorporate HSUVs in models, however, are often methodologically poor and lack consistency. Inconsistencies in the methodologies used will produce discrepancies in results, undermining policy decisions informed by cost per quality-adjusted life-years. OBJECTIVE: To provide an overview of the current evidence base relating to populating decision-analytic models with HSUVs. FINDINGS: Research exploring suitable methods to accurately reflect the baseline or counterfactual HSUVs in decision-analytic models is limited, and while one study suggested that general population data may be appropriate, guidance in this area is poor. Literature describing the appropriateness of different methods used to estimate HSUVs for combined conditions is growing, but there is currently no consensus on the most appropriate methodology. While exploratory analyses suggest that a statistical regression model might improve accuracy in predicted values, the models require validation and testing in external data sets. Until additional research has been conducted in this area, the current evidence suggests that the multiplicative method is the most appropriate technique. Uncertainty in the HSUVs used in decision-analytic models is rarely fully characterized in decision-analytic models and is generally poorly reported. CONCLUSIONS: A substantial volume of research is required before definitive detailed evidence-based practical advice can be provided. As the methodologies used can make a substantial difference to the results generated from decision-analytic models, the differences and lack of clarity and guidance will continue to lead to inconsistencies in policy decision making.