Advertisement

Review of Utility Values for Economic Modeling in Type 2 Diabetes

Open ArchivePublished:May 17, 2014DOI:https://doi.org/10.1016/j.jval.2014.03.003

      Abstract

      Objectives

      Economic analysis in type 2 diabetes mellitus (T2DM) requires an assessment of the effect of a wide range of complications. The objective of this article was to identify a set of utility values consistent with the National Institute for Health and Care Excellence (NICE) reference case and to critically discuss and illustrate challenges in creating such a utility set.

      Methods

      A systematic literature review was conducted to identify studies reporting utility values for relevant complications. The methodology of each study was assessed for consistency with the NICE reference case. A suggested set of utility values applicable to modeling was derived, giving preference to studies reporting multiple complications and correcting for comorbidity.

      Results

      The review considered 21 relevant diabetes complications. A total of 16,574 articles were identified; after screening, 61 articles were assessed for methodological quality. Nineteen articles met NICE criteria, reporting utility values for 20 of 21 relevant complications. For renal transplant, because no articles meeting NICE criteria were identified, two articles using other methodologies were included. Index value estimates for T2DM without complication ranged from 0.711 to 0.940. Utility decrement associated with complications ranged from 0.014 (minor hypoglycemia) to 0.28 (amputation). Limitations associated with the selection of a utility value for use in economic modeling included variability in patient recruitment, heterogeneity in statistical analysis, large variability around some point estimates, and lack of recent data.

      Conclusions

      A reference set of utility values for T2DM and its complications in line with NICE requirements was identified. This research illustrates the challenges associated with systematically selecting utility data for economic evaluations.

      Keywords

      Introduction

      In 2010, an estimated 3.6 million people in England had diabetes, resulting in a direct health care cost of £8.8 billion and a total economic burden of £21.8 billion per year [
      • Hex N.
      • Bartlett C.
      • Wright D.
      • et al.
      Estimating the current and future costs of type 1 and type 2 diabetes in the UK, including direct health costs and indirect societal and productivity costs.
      ,
      • Holman N.
      • Forouhi N.G.
      • Goyder E.
      • Wild S.H.
      The Association of Public Health Observatories (APHO) Diabetes Prevalence Model: estimates of total diabetes prevalence for England, 2010–2030.
      ]. Prevalence is expected to increase by 28% by 2030, and at least 59 new treatments for diabetes are in late-stage testing or preregistration [
      • Hex N.
      • Bartlett C.
      • Wright D.
      • et al.
      Estimating the current and future costs of type 1 and type 2 diabetes in the UK, including direct health costs and indirect societal and productivity costs.
      ,

      IMS Knowledge Link version 5. Drugs used in diabetes currently in phase III or in pre-registration. IMS Health, Lifecycle R&D Focus Source. Available from: http://knowledgelink.imshealth.com/. [Accessed March 13, 2013].

      ]. Effective allocation of resources to manage diabetes is and will remain a key public health and economic imperative.
      The UK’s National Institute of Health and Care Excellence (NICE) appraises the effectiveness and cost-effectiveness of selected new medical technologies, and has considered a number of recent therapies for type 2 diabetes mellitus (T2DM). NICE methodology guidance recommends that effectiveness be assessed using quality-adjusted life-years, with health states measured using the EuroQol five-dimensional (EQ-5D) valuation questionnaire [

      National Institute for Health and Care Excellence. Guide to the methods of technology appraisal. 2008. Available from: http://www.nice.org.uk/media/B52/A7/TAMethodsGuideUpdatedJune2008.pdf. [Accessed March 13, 2013].

      ,
      • Papaioannou D.
      • Brazier J.
      • Paisley S.
      TSD 9 The Identification, Review and Synthesis of Health State Utility Values from the Literature.
      ,
      • Brazier J.
      • Longworth L.
      TSD 8 An Introduction to the Measurement and Valuation of Health for NICE Submissions.
      ,

      National Institute for Health and Care Excellence. Single technology appraisal (STA), Specification for manufacturer/sponsor submission of evidence. Rotterdam, The Netherlands: NICE, 2012.

      ].
      The EQ-5D questionnaire consists of the descriptive system and the EuroQol visual analogue scale (EQ-VAS). The EQ-5D questionnaire descriptive system is a self-administered questionnaire including five dimensions: mobility, self-care, usual activities, pain/discomfort, and anxiety/depression. The EQ-VAS consists of a vertical, visual analogue scale, where the end points are labeled “best imaginable health state” (100) and “worst imaginable health state” (0), on which the respondents rate their current health status [
      • Rabin R.
      • Oemar M.
      • Oppe M.
      EQ-5D-3L User Guide (Version 4.0).
      ]. The EQ-5D questionnaire descriptive system is preferred from an economic perspective because results can be translated into the EQ-5D questionnaire index values using scores from a set of preference weights measured with the time trade-off valuation technique on a sample from the general population. This difference implies that the index value can be regarded as a societal valuation of the patient’s health state whereas the EQ-VAS score is the patient’s own assessment of his or her health state.
      The effect of T2DM on utility is multifactorial and substantial [
      • Holmes J.
      • McGill S.
      • Kind P.
      • et al.
      Health-related quality of life in type 2 diabetes (TARDIS-2).
      ,
      • Wexler D.J.
      • Grant R.W.
      • Wittenberg E.
      • et al.
      Correlates of health-related quality of life in type 2 diabetes.
      ]. The choice of the utility assessment method can have a considerable effect on the predicted utility values and therefore on the outcome of economic evaluation [
      • Conner-Spady B.
      • Suarez-Almazor M.E.
      Variation in the estimation of quality-adjusted life-years by different preference-based instruments.
      ,
      • Kopec J.A.
      • Willison K.D.
      A comparative review of four preference-weighted measures of health-related quality of life.
      ]. This study aimed to identify a set of utility values that are consistent with the NICE reference case and that might be used in economic evaluations in TD2M, and to critically discuss and illustrate challenges in creating such a utility set.

      Methods

      The published descriptions of five computer models that simulate long-term outcomes in T2DM were reviewed to identify diabetic complications that have an impact on patient utility. Models considered were the IMS CORE Diabetes Model [
      • Palmer A.J.
      • Roze S.
      • Valentine W.J.
      • et al.
      The CORE Diabetes Model: projecting long-term clinical outcomes, costs and cost-effectiveness of interventions in diabetes mellitus (types 1 and 2) to support clinical and reimbursement decision-making.
      ], the United Kingdom Prospective Diabetes Study (UKPDS) Outcomes Model [
      • Clarke P.M.
      • Gray A.M.
      • Briggs A.
      • et al.
      A model to estimate the lifetime health outcomes of patients with type 2 diabetes: the United Kingdom Prospective Diabetes Study (UKPDS) Outcomes Model (UKPDS no. 68).
      ], the Cardiff Diabetes Model [
      • McEwan P.
      • Peters J.R.
      • Bergenheim K.
      • Currie C.J.
      Evaluation of the costs and outcomes from changes in risk factors in type 2 diabetes using the Cardiff stochastic simulation cost-utility model (DiabForecaster).
      ], the Sheffield Diabetes Model [
      Mount Hood 4 Modeling Group
      Computer modeling of diabetes and its complications: a report on the Fourth Mount Hood Challenge Meeting.
      ], and the Centers for Disease Control and Prevention/Research Triangle Institute Type 2 Diabetes Model [16]. Health states used in the models were considered relevant if they described microvascular or macrovascular complications associated with T2DM, direct consequences of treatment (such as hypoglycemia), or were related to excess body weight.
      Relevant complications identified in the review of models were angina, heart failure, myocardial infarction, stroke, peripheral vascular disease, neuropathy, foot ulcer, microalbuminuria/protenuria, renal dialysis, renal transplant, cataract, diabetic retinopathy, vision loss, macular edema, hypoglycemia, and excess weight (defined as either presence vs. absence of obesity or increase in body mass index). When available, data were also extracted in the utility value for patients with T2DM without specific complications.
      A systematic literature review was conducted to identify articles reporting utility data for one or more of the complications in patients with diabetes. Each complication was searched separately in MEDLINE and Medline In-Process, Embase, EconLIT, and National Health Service Economic Evaluation Database, including articles from the earliest available date to May 2012 when the searches were run. The full search strategy can be found in Appendix 1 in Supplemental Materials found at doi: doi:10.1016/j.jval.2014.03.003.
      Articles were included in the review if they reported a study performed in adults and contained original data on the effect of diabetic complications on utility values. Exclusion criteria included studies conducted in pediatric populations, studies not reporting preference values or utility measures, or studies reporting utility values associated with an intervention. Only publication written in English were considered.
      Shortlisted articles were then reviewed against a set of criteria to identify studies that met the NICE reference case. These criteria included that the EQ-5D questionnaire value set or other established measures of preference should be reported, health states should be determined by patients, and valuation should be performed using a societal valuation algorithm [
      • Papaioannou D.
      • Brazier J.
      • Paisley S.
      TSD 9 The Identification, Review and Synthesis of Health State Utility Values from the Literature.
      ]. As NICE recommends utilities estimated from a sample of the general population, utilities estimated using the EQ-5D questionnaire index were considered more appropriate than the EQ-VAS [
      • Conner-Spady B.
      • Suarez-Almazor M.E.
      Variation in the estimation of quality-adjusted life-years by different preference-based instruments.
      ,
      • Lung T.W.
      • Hayes A.J.
      • Hayen A.
      • et al.
      A meta-analysis of health state valuations for people with diabetes: explaining the variation across methods and implications for economic evaluation.
      ]. Studies that met the NICE reference case criteria were considered as potential sources for the preferred input set. The following attributes were extracted from studies that met the inclusion criteria: country in which each study took place, year of publication, sample size and recruitment method, patients’ demographic characteristics, statistical methods, and measures of precision reported.
      For diabetic complications for which there were no index values that met the NICE reference case, an alternative source reviewing study methodology, complication definition, baseline population characteristics, and sample size was sought.
      The preferred input set was created using the following criteria: if only one measure was available that met NICE criteria, this was accepted. When more than one estimate was identified, studies reporting the marginal impact of T2DM complications relative to a baseline “no complication” state were preferred over those reporting disutilities alone; furthermore, we considered synthesizing health utility values if identified studies were sufficiently homogeneous [
      • Papaioannou D.
      • Brazier J.
      • Paisley S.
      TSD 9 The Identification, Review and Synthesis of Health State Utility Values from the Literature.
      ]. When an article presented utility values obtained using multiple statistical models, the best fitting statistical model or the one preferred by the authors was presented. When possible, disutility estimates calculated from statistical models were presented; otherwise, the difference between patients with and without the specific complication was presented.
      If these criteria were not sufficient to identify a preferred measure, the study that most precisely matched the definition of the preferred complication was selected and studies that reported counterintuitive results were excluded.
      For measures of utility in the preferred set, statistical methods to identify measures of uncertainty appropriate for use in economic evaluation were reviewed. When reported, the 95% confidence intervals were extracted; otherwise, they were estimated around each point value using reported mean and sample size values, assuming a normal distribution.

      Results

      The literature search identified 16,574 records as presented in Figure 1. A total of 339 full-text articles were retrieved of which 61 included utility values that were assessed against NICE reference case criteria. The number of studies screened, assessed for eligibility, and included in the review along with reasons for exclusions at each stage, are presented in Table 1. The studies were mainly conducted in Europe, but publications from Asia, America, and Australia were also identified. Most studies were based on patients with T2DM. The recruitment methods and environment varied greatly, which may affect the profile of the selected patients. Sample size varied from 17 up to 4641 patients.
      Table 1Study appropriateness versus the NICE reference case.
      ReferenceCountryDiabetes typeRespondent recruitmentNMean age (y)Valuation instrumentTariff usedStatistical method
      Bagust and Beale
      • Bagust A.
      • Beale S.
      Modelling EuroQol health-related utility values for diabetic complications from CODE-2 data.
      Belgium, Italy, Spain, UK, and The Netherlands2CODE-2 study patients464167EQ-5D questionnaireUK
      • Dolan P.
      Modeling valuations for EuroQol health states.
      Multiple regression analysis model
      Clarke et al.
      • Clarke P.
      • Gray A.
      • Holman R.
      Estimating utility values for health states of type 2 diabetic patients using the EQ-5D (UKPDS 62).
      UK2UKPDS study patients319262EQ-5D questionnaireUK
      • Dolan P.
      Modeling valuations for EuroQol health states.
      Tobit model
      Currie et al.
      • Currie C.J.
      • Morgan C.L.
      • Poole C.D.
      • et al.
      Multivariate models of health-related utility and the fear of hypoglycaemia in people with diabetes.
      UK2Postal Survey in Cardiff, UK130562EQ-5D questionnaireUK
      • Dolan P.
      Modeling valuations for EuroQol health states.
      Multivariate analysis
      Fenwick et al.
      • Fenwick E.K.
      • Xie J.
      • Ratcliffe J.
      • et al.
      The impact of diabetic retinopathy and diabetic macular edema on health-related quality of life in type 1 and type 2 diabetes.
      Australia1 and 2Specialized eye clinics in Melbourne, Australia57766EQ-5D questionnaireUK
      • Dolan P.
      Modeling valuations for EuroQol health states.
      Multivariate quantile regression model
      Glasziou et al.
      • Glasziou P.
      • Alexander J.
      • Beller E.
      • Clarke P.
      Which health-related quality of life score? A comparison of alternative utility measures in patients with type 2 diabetes in the ADVANCE trial.
      Australia2ADVANCE study patients97867EQ-5D questionnaireUK
      • Dolan P.
      Modeling valuations for EuroQol health states.
      Mean values
      Kiberd and Jindal
      • Kiberd B.A.
      • Jindal K.K.
      Screening to prevent renal failure in insulin dependent diabetic patients: an economic evaluation.
      Studies reporting utility values for renal transplant in diabetic patients.
      CanadaHealth care workers17TTOMean values
      Kontodimopoulos et al.
      • Kontodimopoulos N.
      • Pappa E.
      • Chadjiapostolou Z.
      • et al.
      Comparing the sensitivity of EQ-5D, SF-6D and 15D utilities to the specific effect of diabetic complications.
      Greece2Routine visit to the diabetes outpatient clinic31965EQ-5D questionnaireUK
      • Dolan P.
      Modeling valuations for EuroQol health states.
      Linear multivariate regression model
      Langelaan et al.
      • Langelaan M.
      • de Boer M.R.
      • van Nispen R.M.
      • et al.
      Impact of visual impairment on quality of life: a comparison with quality of life in the general population and with other chronic conditions.
      The NetherlandsNot specifiedRehabilitation center for visually impaired adults12842EQ-5D questionnaireThe Netherlands
      • Lamers L.M.
      • Stalmeier P.F.
      • McDonnell J.
      • et al.
      Measuring the quality of life in economic evaluations: the Dutch EQ-5D tariff [in Dutch].
      Mean values
      Laupacis et al.
      • Laupacis A.
      • Keown P.
      • Pus N.
      • et al.
      A study of the quality of life and cost-utility of renal transplantation.
      Studies reporting utility values for renal transplant in diabetic patients.
      CanadaNot specifiedPatients on a transplant waiting list in two hospitals16842TTOMean values
      Lee et al.
      • Lee W.J.
      • Song K.H.
      • Noh J.H.
      • et al.
      Health-related quality of life using the EuroQol 5D questionnaire in Korean patients with type 2 diabetes.
      Korea2Outpatient clinic of university hospital (Korea)85858EQ-5D questionnaireSouth Korea
      • Jo M.W.
      • Yun S.C.
      • Lee S.I.
      Estimating quality weights for EQ-5D health states with the time trade-off method in South Korea.
      Univariate model
      Lloyd et al.
      • Lloyd A.
      • Nafees B.
      • Gavriel S.
      • et al.
      Health utility values associated with diabetic retinopathy.
      UK1 and 2Five clinical sites in the UK12262EQ-5D questionnaireUK
      • Dolan P.
      Modeling valuations for EuroQol health states.
      Univariate model
      Marrett et al.
      • Marrett E.
      • Stargardt T.
      • Mavros P.
      • Alexander C.M.
      Patient-reported outcomes in a survey of patients treated with oral antihyperglycaemic medications: associations with hypoglycaemia and weight gain.
      USA2An annual, cross-sectional Internet-based survey198458EQ-5D questionnaireUS
      • Shaw J.W.
      • Johnson J.A.
      • Coons S.J.
      US valuation of the EQ-5D health states: development and testing of the D1 valuation model.
      Mix linear regression model
      Matza et al.
      • Matza L.S.
      • Yurgin N.
      • Boye K.S.
      • et al.
      Obese versus non-obese patients with type 2 diabetes: patient-reported outcomes and utility of weight change.
      UK and Scotland2Advertisement in newspapers12956EQ-5D questionnaireUK
      • Dolan P.
      Modeling valuations for EuroQol health states.
      Unadjusted data
      O’Reilly et al.
      • O’Reilly D.J.
      • Xie F.
      • Pullenayegum E.
      • et al.
      Estimation of the impact of diabetes-related complications on health utilities for patients with type 2 diabetes in Ontario, Canada.
      Canada2Community-dwelling patients114764EQ-5D questionnaireUS
      • Shaw J.W.
      • Johnson J.A.
      • Coons S.J.
      US valuation of the EQ-5D health states: development and testing of the D1 valuation model.
      Ordinary least square mean regression
      Quah et al.
      • Quah J.H.M.
      • Luo N.
      • Ng W.Y.
      • et al.
      Health-related quality of life is associated with diabetic complications, but not with short-term diabetic control in primary care.
      Singapore2A polyclinic laboratory69963EQ-5D questionnaireUK
      • Dolan P.
      Modeling valuations for EuroQol health states.
      Multiple regression model
      Redekop et al.
      • Redekop W.K.
      • Koopmanschap M.A.
      • Stolk R.P.
      • et al.
      Health-related quality of life and treatment satisfaction in Dutch patients with type 2 diabetes.
      The Netherlands2Dutch general practitioners113665EQ-5D questionnaireUK
      • Dolan P.
      Modeling valuations for EuroQol health states.
      Multivariate analysis using ordinary least squares linear regression
      Smith et al.
      • Smith D.H.
      • Johnson E.S.
      • Russell A.
      • et al.
      Lower visual acuity predicts worse utility values among patients with type 2 diabetes.
      USA2Diabetes registry population207466EQ-5D questionnaireUS
      • Shaw J.W.
      • Johnson J.A.
      • Coons S.J.
      US valuation of the EQ-5D health states: development and testing of the D1 valuation model.
      Linear multivariate regression model
      Solli et al.
      • Solli O.
      • Stavern K.
      • Kristiansen I.S.
      Health-related quality of life in diabetes: the associations of complications with EQ-5D scores.
      Norway1 and 2Norwegian Diabetes Association members35664EQ-5D questionnaireUK
      • Dolan P.
      Modeling valuations for EuroQol health states.
      Linear multivariate regression model
      Sullivan et al.
      • Sullivan P.W.
      • Ghushchyan V.H.
      • Ben-Joseph R.
      The impact of obesity on diabetes, hyperlipidemia and hypertension in the United States.
      USANot specifiedA nationally representative survey of the US population203945EQ-5D questionnaireUS
      • Shaw J.W.
      • Johnson J.A.
      • Coons S.J.
      US valuation of the EQ-5D health states: development and testing of the D1 valuation model.
      Censored least absolute deviations (CLAD) estimator regression model
      Vexiau et al.
      • Vexiau P.
      • Mavros P.
      • Krishnarajah G.
      • et al.
      Hypoglycaemia in patients with type 2 diabetes treated with a combination of metformin and sulphonylurea therapy in France.
      France2Questionnaires during the primary care office visit40062EQ-5D questionnaireUK
      • Dolan P.
      Modeling valuations for EuroQol health states.
      Linear multivariate regression model
      Wasserfallen et al.
      • Wasserfallen J.B.
      • Halabi G.
      • Saudan P.
      • et al.
      Quality of life on chronic dialysis: comparison between haemodialysis and peritoneal dialysis.
      SwitzerlandNot specified19 dialysis centers of western Switzerland45564EQ-5D questionnaireUK
      • Dolan P.
      Modeling valuations for EuroQol health states.
      Unadjusted data
      ADVANCE, Action in Diabetes and Vascular Disease: Preterax and Diamicron Modified Release Controlled Evaluation; CODE-2, The Cost of Diabetes in Europe - Type II; EQ-5D, EuroQol five-dimensional; TTO, time trade-off; UKPDS, United Kingdom Prospective Diabetes Study.
      Not a diabetic population.
      low asterisk Studies reporting utility values for renal transplant in diabetic patients.
      A total of 19 studies met the NICE criteria, and at least one estimate was identified for each of the relevant complications apart from renal transplant. No studies were identified for renal transplant, so the two studies using alternative methodology were considered for inclusion in the preferred set. The characteristics of the 21 studies considered for the preferred set are shown in Table 1. Twenty of 21 studies reported EQ-5D questionnaire data; 5 presented values based on a UK population, and 13 applied a UK value set to EQ-5D questionnaire responses collected elsewhere.
      Figure 1 The index values presented for T2DM without complication ranged from 0.711 to 0.94 [
      • Currie C.J.
      • Morgan C.L.
      • Poole C.D.
      • et al.
      Multivariate models of health-related utility and the fear of hypoglycaemia in people with diabetes.
      ,
      • Smith D.H.
      • Johnson E.S.
      • Russell A.
      • et al.
      Lower visual acuity predicts worse utility values among patients with type 2 diabetes.
      ]. The estimated utility values for each complication are shown in Table 2.
      Figure thumbnail gr1
      Fig. 1PRISMA flow diagram. EQ-5D, EuroQol five-dimensional; NICE, National Institute for Health and Care Excellence; PRISMA: Preferred Reporting Items for Systematic Reviews and Meta-Analyses.
      Table 2Candidate utility values by health state.
      ReferenceBaseline T2DM without complication(Dis)utility associated with complicationVariability around the disutility valueDifference statistically significant? (P < 0.05)
      Myocardial infarction
       O’Reilly et al.
      • O’Reilly D.J.
      • Xie F.
      • Pullenayegum E.
      • et al.
      Estimation of the impact of diabetes-related complications on health utilities for patients with type 2 diabetes in Ontario, Canada.
      0.760−0.059SE: 0.017Yes
       Clarke et al.
      • Clarke P.
      • Gray A.
      • Holman R.
      Estimating utility values for health states of type 2 diabetic patients using the EQ-5D (UKPDS 62).
      0.785−0.05595% CI (–0.067 to –0.042)
       Glasziou et al.
      • Glasziou P.
      • Alexander J.
      • Beller E.
      • Clarke P.
      Which health-related quality of life score? A comparison of alternative utility measures in patients with type 2 diabetes in the ADVANCE trial.
      0.843−0.041Not reported
       Lee et al.
      • Lee W.J.
      • Song K.H.
      • Noh J.H.
      • et al.
      Health-related quality of life using the EuroQol 5D questionnaire in Korean patients with type 2 diabetes.
      Not reported−0.007SE: 0.015
      Ischemic heart disease
       Clarke et al.
      • Clarke P.
      • Gray A.
      • Holman R.
      Estimating utility values for health states of type 2 diabetic patients using the EQ-5D (UKPDS 62).
      0.785−0.09095% CI (–0.126 to –0.054)
       Glasziou et al.
      • Glasziou P.
      • Alexander J.
      • Beller E.
      • Clarke P.
      Which health-related quality of life score? A comparison of alternative utility measures in patients with type 2 diabetes in the ADVANCE trial.
      0.843−0.068Not reported
       Quah et al.
      • Quah J.H.M.
      • Luo N.
      • Ng W.Y.
      • et al.
      Health-related quality of life is associated with diabetic complications, but not with short-term diabetic control in primary care.
      0.910−0.050Not reportedYes
       Solli et al.
      • Solli O.
      • Stavern K.
      • Kristiansen I.S.
      Health-related quality of life in diabetes: the associations of complications with EQ-5D scores.
      Impaired vision.
      0.850−0.03795% CI (–0.103 to 0.030)
       Lee et al.
      • Lee W.J.
      • Song K.H.
      • Noh J.H.
      • et al.
      Health-related quality of life using the EuroQol 5D questionnaire in Korean patients with type 2 diabetes.
      Not reported−0.027SE: 0.011Yes
      Heart failure
       Clarke et al.
      • Clarke P.
      • Gray A.
      • Holman R.
      Estimating utility values for health states of type 2 diabetic patients using the EQ-5D (UKPDS 62).
      0.785−0.10895% CI (–0.169 to –0.048)
       Lee et al.
      • Lee W.J.
      • Song K.H.
      • Noh J.H.
      • et al.
      Health-related quality of life using the EuroQol 5D questionnaire in Korean patients with type 2 diabetes.
      Not reported−0.051SE: 0.0154Yes
      Stroke
       Clarke et al.
      • Clarke P.
      • Gray A.
      • Holman R.
      Estimating utility values for health states of type 2 diabetic patients using the EQ-5D (UKPDS 62).
      0.785−0.16495% CI (–0.222 to –0.105)
       Solli et al.
      • Solli O.
      • Stavern K.
      • Kristiansen I.S.
      Health-related quality of life in diabetes: the associations of complications with EQ-5D scores.
      0.850−0.13595% CI (−0.247 to −0.023)Yes
       Glasziou et al.
      • Glasziou P.
      • Alexander J.
      • Beller E.
      • Clarke P.
      Which health-related quality of life score? A comparison of alternative utility measures in patients with type 2 diabetes in the ADVANCE trial.
      0.843−0.104Not reported
       O’Reilly et al.
      • O’Reilly D.J.
      • Xie F.
      • Pullenayegum E.
      • et al.
      Estimation of the impact of diabetes-related complications on health utilities for patients with type 2 diabetes in Ontario, Canada.
      0.760−0.046SE: 0.023Yes
       Quah et al.
      • Quah J.H.M.
      • Luo N.
      • Ng W.Y.
      • et al.
      Health-related quality of life is associated with diabetic complications, but not with short-term diabetic control in primary care.
      0.910−0.07Not reportedYes
      Peripheral vascular disease
       Bagust and Beale
      • Bagust A.
      • Beale S.
      Modelling EuroQol health-related utility values for diabetic complications from CODE-2 data.
      0.790−0.061SE: 0.015Yes
       Kontodimopoulos et al.
      • Kontodimopoulos N.
      • Pappa E.
      • Chadjiapostolou Z.
      • et al.
      Comparing the sensitivity of EQ-5D, SF-6D and 15D utilities to the specific effect of diabetic complications.
      0.770−0.186SE: 0.054Yes
       Glasziou et al.
      • Glasziou P.
      • Alexander J.
      • Beller E.
      • Clarke P.
      Which health-related quality of life score? A comparison of alternative utility measures in patients with type 2 diabetes in the ADVANCE trial.
      0.843−0.083Not reported
       Quah et al.
      • Quah J.H.M.
      • Luo N.
      • Ng W.Y.
      • et al.
      Health-related quality of life is associated with diabetic complications, but not with short-term diabetic control in primary care.
      0.910−0.080Not reportedYes
      Proteinuria
       Bagust and Beale
      • Bagust A.
      • Beale S.
      Modelling EuroQol health-related utility values for diabetic complications from CODE-2 data.
      0.790−0.048SE: 0.022Yes
      Hemodialysis
       Wasserfallen et al.
      • Wasserfallen J.B.
      • Halabi G.
      • Saudan P.
      • et al.
      Quality of life on chronic dialysis: comparison between haemodialysis and peritoneal dialysis.
      Not reported0.621SD: 0.299
      Peritoneal dialysis
       Wasserfallen et al.
      • Wasserfallen J.B.
      • Halabi G.
      • Saudan P.
      • et al.
      Quality of life on chronic dialysis: comparison between haemodialysis and peritoneal dialysis.
      Not reported0.581SD: 0.323
      Renal transplant
      TTO values because no EQ-5D questionnaire values were identified.
       Kiberd and Jindal
      • Kiberd B.A.
      • Jindal K.K.
      Screening to prevent renal failure in insulin dependent diabetic patients: an economic evaluation.
      0.8380.762Not reported
       Laupacis et al.
      • Laupacis A.
      • Keown P.
      • Pus N.
      • et al.
      A study of the quality of life and cost-utility of renal transplantation.
      Not reported0.820Not reported
      Moderate nonproliferative background diabetic retinopathy
       Fenwick et al.
      • Fenwick E.K.
      • Xie J.
      • Ratcliffe J.
      • et al.
      The impact of diabetic retinopathy and diabetic macular edema on health-related quality of life in type 1 and type 2 diabetes.
      0.800−0.040IQR: 0.31
      Cataract
       Lee et al.
      • Lee W.J.
      • Song K.H.
      • Noh J.H.
      • et al.
      Health-related quality of life using the EuroQol 5D questionnaire in Korean patients with type 2 diabetes.
      Not reported−0.016SE: 0.0076Yes
      Moderate macular edema
       Fenwick et al.
      • Fenwick E.K.
      • Xie J.
      • Ratcliffe J.
      • et al.
      The impact of diabetic retinopathy and diabetic macular edema on health-related quality of life in type 1 and type 2 diabetes.
      0.800−0.040IQR: 0.31
      Vision-threatening diabetic retinopathy
       Fenwick et al.
      • Fenwick E.K.
      • Xie J.
      • Ratcliffe J.
      • et al.
      The impact of diabetic retinopathy and diabetic macular edema on health-related quality of life in type 1 and type 2 diabetes.
      0.800−0.070IQR: 0.38
       Solli et al.
      • Solli O.
      • Stavern K.
      • Kristiansen I.S.
      Health-related quality of life in diabetes: the associations of complications with EQ-5D scores.
      Impaired vision.
      0.850−0.01295% CI (−0.074 to 0.051)
      Severe vision loss
       Lloyd et al.
      • Lloyd A.
      • Nafees B.
      • Gavriel S.
      • et al.
      Health utility values associated with diabetic retinopathy.
      0.830−0.490Not reported
       Smith et al.
      • Smith D.H.
      • Johnson E.S.
      • Russell A.
      • et al.
      Lower visual acuity predicts worse utility values among patients with type 2 diabetes.
      0.940−0.15095% CI (–0.190 to –0.110)
       Clarke et al.
      • Clarke P.
      • Gray A.
      • Holman R.
      Estimating utility values for health states of type 2 diabetic patients using the EQ-5D (UKPDS 62).
      0.7850−0.07495% CI (–0.124 to –0.025)
       Langelaan et al.
      • Langelaan M.
      • de Boer M.R.
      • van Nispen R.M.
      • et al.
      Impact of visual impairment on quality of life: a comparison with quality of life in the general population and with other chronic conditions.
      Not reported0.64SD: 0.240
      Neuropathy
       Kontodimopoulos et al.
      • Kontodimopoulos N.
      • Pappa E.
      • Chadjiapostolou Z.
      • et al.
      Comparing the sensitivity of EQ-5D, SF-6D and 15D utilities to the specific effect of diabetic complications.
      0.770−0.247SE: 0.084Yes
       Solli et al.
      • Solli O.
      • Stavern K.
      • Kristiansen I.S.
      Health-related quality of life in diabetes: the associations of complications with EQ-5D scores.
      0.850−0.18795% CI (−0.316 to −0.057)Yes
       Bagust and Beale
      • Bagust A.
      • Beale S.
      Modelling EuroQol health-related utility values for diabetic complications from CODE-2 data.
      0.790−0.084SE: 0.014Yes
       Quah et al.
      • Quah J.H.M.
      • Luo N.
      • Ng W.Y.
      • et al.
      Health-related quality of life is associated with diabetic complications, but not with short-term diabetic control in primary care.
      0.910−0.050Not reportedYes
      Active ulcer
       Kontodimopoulos et al.
      • Kontodimopoulos N.
      • Pappa E.
      • Chadjiapostolou Z.
      • et al.
      Comparing the sensitivity of EQ-5D, SF-6D and 15D utilities to the specific effect of diabetic complications.
      0.770−0.206SE: 0.069Yes
       Bagust and Beale
      • Bagust A.
      • Beale S.
      Modelling EuroQol health-related utility values for diabetic complications from CODE-2 data.
      0.790−0.170SE: 0.019Yes
       Solli et al.
      • Solli O.
      • Stavern K.
      • Kristiansen I.S.
      Health-related quality of life in diabetes: the associations of complications with EQ-5D scores.
      0.850−0.01695% CI (−0.134 to 0.101)
      Amputation event
       Clarke et al.
      • Clarke P.
      • Gray A.
      • Holman R.
      Estimating utility values for health states of type 2 diabetic patients using the EQ-5D (UKPDS 62).
      0.785−0.28095% CI (–0.389 to –0.170)
       O’Reilly et al.
      • O’Reilly D.J.
      • Xie F.
      • Pullenayegum E.
      • et al.
      Estimation of the impact of diabetes-related complications on health utilities for patients with type 2 diabetes in Ontario, Canada.
      0.760−0.063SE: 0.059Yes
      Major hypoglycemia event
       Vexiau et al.
      • Vexiau P.
      • Mavros P.
      • Krishnarajah G.
      • et al.
      Hypoglycaemia in patients with type 2 diabetes treated with a combination of metformin and sulphonylurea therapy in France.
      3-month period.
      0.820
      Patients without hypoglycemia.
      −0.270Not reportedYes
       Marrett et al.
      • Marrett E.
      • Stargardt T.
      • Mavros P.
      • Alexander C.M.
      Patient-reported outcomes in a survey of patients treated with oral antihyperglycaemic medications: associations with hypoglycaemia and weight gain.
      TTO values because no EQ-5D questionnaire values were identified.
      0.860
      Patients without hypoglycemia.
      −0.160Not reportedYes
       Currie et al.
      • Currie C.J.
      • Morgan C.L.
      • Poole C.D.
      • et al.
      Multivariate models of health-related utility and the fear of hypoglycaemia in people with diabetes.
      3-month period.
      0.711−0.047Not reported
      Minor hypoglycemia event
       Vexiau et al.
      • Vexiau P.
      • Mavros P.
      • Krishnarajah G.
      • et al.
      Hypoglycaemia in patients with type 2 diabetes treated with a combination of metformin and sulphonylurea therapy in France.
      3-month period.
      0.820
      6-mo period.
      −0.070Not reportedYes
       Marrett et al.
      • Marrett E.
      • Stargardt T.
      • Mavros P.
      • Alexander C.M.
      Patient-reported outcomes in a survey of patients treated with oral antihyperglycaemic medications: associations with hypoglycaemia and weight gain.
      6-mo period.
      0.860
      6-mo period.
      −0.050Not reportedYes
       Currie et al.
      • Currie C.J.
      • Morgan C.L.
      • Poole C.D.
      • et al.
      Multivariate models of health-related utility and the fear of hypoglycaemia in people with diabetes.
      3-month period.
      0.711−0.014Not reported
      Excess BMI
       Matza et al.
      • Matza L.S.
      • Yurgin N.
      • Boye K.S.
      • et al.
      Obese versus non-obese patients with type 2 diabetes: patient-reported outcomes and utility of weight change.
      , Obese vs. nonobese
      0.800−0.080Not reported
       Redekop et al.
      • Redekop W.K.
      • Koopmanschap M.A.
      • Stolk R.P.
      • et al.
      Health-related quality of life and treatment satisfaction in Dutch patients with type 2 diabetes.
      , Obese vs. nonobese
      0.810−0.044Not reportedYes
       Sullivan et al.
      • Sullivan P.W.
      • Ghushchyan V.H.
      • Ben-Joseph R.
      The impact of obesity on diabetes, hyperlipidemia and hypertension in the United States.
      , Obese vs. nonobese
      0.761−0.059Not reportedYes
       Marrett et al.
      • Marrett E.
      • Stargardt T.
      • Mavros P.
      • Alexander C.M.
      Patient-reported outcomes in a survey of patients treated with oral antihyperglycaemic medications: associations with hypoglycaemia and weight gain.
      0.860−0.020Not reportedYes
       Bagust and Beale
      • Bagust A.
      • Beale S.
      Modelling EuroQol health-related utility values for diabetic complications from CODE-2 data.
      (BMI unit above 25 kg/m2)
      0.790−0.006SE: 0.001Yes
       Kontodimopoulos et al.
      • Kontodimopoulos N.
      • Pappa E.
      • Chadjiapostolou Z.
      • et al.
      Comparing the sensitivity of EQ-5D, SF-6D and 15D utilities to the specific effect of diabetic complications.
      (per BMI unit)
      0.770−0.006SE: 0.002Yes
       Solli et al.
      • Solli O.
      • Stavern K.
      • Kristiansen I.S.
      Health-related quality of life in diabetes: the associations of complications with EQ-5D scores.
      (per BMI unit)
      0.850−0.00295% CI (–0.126 to –0.054)
      BMI, body mass unit; CI, confidence interval; EQ-5D, EuroQol five-dimensional; IQR, interquartile range; SE, standard error; TTO, time trade-off; T2DM, type 2 diabetes mellitus.
      * TTO values because no EQ-5D questionnaire values were identified.
      Impaired vision.
      3-month period.
      § 6-mo period.
      ǁ Patients without hypoglycemia.
      Table 3 presents the preferred set of values, alongside 95% confidence intervals, whereas Figure 2 presents the same data with a summary of the range of values extracted at the previous analysis stage. The uncertainty around the point estimate was important. With the expectation of the amputation event, all the complications’ 95% confidence interval overlapped with the 95% confidence interval of the “T2DM without complication health state.”
      Table 3Preferred utility values for modeling complications associated with T2DM.
      ParameterProposed referenceProposed utility value95% CIRange of candidate values
      T2DM without complicationClarke et al.
      • Clarke P.
      • Gray A.
      • Holman R.
      Estimating utility values for health states of type 2 diabetic patients using the EQ-5D (UKPDS 62).
      0.7850.681–0.8890.690–0.940
      Myocardial infarctionClarke et al.
      • Clarke P.
      • Gray A.
      • Holman R.
      Estimating utility values for health states of type 2 diabetic patients using the EQ-5D (UKPDS 62).
      −0.055–0.067 to −0.042−0.059 to −0.007
      Ischemic heart diseaseClarke et al.
      • Clarke P.
      • Gray A.
      • Holman R.
      Estimating utility values for health states of type 2 diabetic patients using the EQ-5D (UKPDS 62).
      −0.090–0.126 to −0.054−0.090 to −0.027
      Heart failureClarke et al.
      • Clarke P.
      • Gray A.
      • Holman R.
      Estimating utility values for health states of type 2 diabetic patients using the EQ-5D (UKPDS 62).
      −0.108–0.169 to −0.048−0.108 to −0.051
      StrokeClarke et al.
      • Clarke P.
      • Gray A.
      • Holman R.
      Estimating utility values for health states of type 2 diabetic patients using the EQ-5D (UKPDS 62).
      −0.164–0.222 to −0.105−0.164 to −0.070
      Severe vision lossClarke et al.
      • Clarke P.
      • Gray A.
      • Holman R.
      Estimating utility values for health states of type 2 diabetic patients using the EQ-5D (UKPDS 62).
      −0.074–0.124 to −0.025−0.070 to −0.012
      Amputation eventClarke et al.
      • Clarke P.
      • Gray A.
      • Holman R.
      Estimating utility values for health states of type 2 diabetic patients using the EQ-5D (UKPDS 62).
      −0.280–0.389 to −0.170−0.280 to −0.063
      Peripheral vascular diseaseBagust and Beale
      • Bagust A.
      • Beale S.
      Modelling EuroQol health-related utility values for diabetic complications from CODE-2 data.
      −0.061−0.090 to −0.032
      Estimated from the standard error values provided.
      −0.186 to −0.061
      ProteinuriaBagust and Beale
      • Bagust A.
      • Beale S.
      Modelling EuroQol health-related utility values for diabetic complications from CODE-2 data.
      −0.048−0.091 to −0.005
      Estimated from the standard error values provided.
      One reference identified
      NeuropathyBagust and Beale
      • Bagust A.
      • Beale S.
      Modelling EuroQol health-related utility values for diabetic complications from CODE-2 data.
      −0.084−0.111 to −0.057
      Estimated from the standard error values provided.
      −0.247 to −0.050
      Active ulcerBagust and Beale
      • Bagust A.
      • Beale S.
      Modelling EuroQol health-related utility values for diabetic complications from CODE-2 data.
      −0.170−0.207 to −0.133
      Estimated from the standard error values provided.
      −0.206 to −0.016
      Excess BMI (each unit above 25 kg/m2)Bagust and Beale
      • Bagust A.
      • Beale S.
      Modelling EuroQol health-related utility values for diabetic complications from CODE-2 data.
      −0.006−0.008 to −0.004
      Estimated from the standard error values provided.
      −0.006 to −0.002
      HemodialysisWasserfallen et al.
      • Wasserfallen J.B.
      • Halabi G.
      • Saudan P.
      • et al.
      Quality of life on chronic dialysis: comparison between haemodialysis and peritoneal dialysis.
      −0.164−0.274 to −0.054
      Estimated from the standard error values provided.
      One reference identified
      Peritoneal dialysisWasserfallen et al.
      • Wasserfallen J.B.
      • Halabi G.
      • Saudan P.
      • et al.
      Quality of life on chronic dialysis: comparison between haemodialysis and peritoneal dialysis.
      −0.204−0.342 to −0.066
      Estimated from the standard error values provided.
      One reference identified
      Renal transplantKiberd and Jindal
      • Kiberd B.A.
      • Jindal K.K.
      Screening to prevent renal failure in insulin dependent diabetic patients: an economic evaluation.
      0.7620.658–0.8660.762–0.820
      CataractLee et al.
      • Lee W.J.
      • Song K.H.
      • Noh J.H.
      • et al.
      Health-related quality of life using the EuroQol 5D questionnaire in Korean patients with type 2 diabetes.
      −0.016−0.031 to −0.001
      Estimated from the standard error values provided.
      One reference identified
      Moderate nonproliferative background diabetic retinopathyFenwick et al.
      • Fenwick E.K.
      • Xie J.
      • Ratcliffe J.
      • et al.
      The impact of diabetic retinopathy and diabetic macular edema on health-related quality of life in type 1 and type 2 diabetes.
      −0.040−0.066 to −0.014
      Estimated from the interquartile range values provided.
      One reference identified
      Moderate macular edemaFenwick et al.
      • Fenwick E.K.
      • Xie J.
      • Ratcliffe J.
      • et al.
      The impact of diabetic retinopathy and diabetic macular edema on health-related quality of life in type 1 and type 2 diabetes.
      −0.040−0.066 to −0.014
      Estimated from the interquartile range values provided.
      One reference identified
      Vision-threatening diabetic retinopathyFenwick et al.
      • Fenwick E.K.
      • Xie J.
      • Ratcliffe J.
      • et al.
      The impact of diabetic retinopathy and diabetic macular edema on health-related quality of life in type 1 and type 2 diabetes.
      −0.070−0.099 to −0.041
      Estimated from the interquartile range values provided.
      −0.070 to –0.012
      Major hypoglycemia eventCurrie et al.
      • Currie C.J.
      • Morgan C.L.
      • Poole C.D.
      • et al.
      Multivariate models of health-related utility and the fear of hypoglycaemia in people with diabetes.
      −0.047−0.012
      Disutilities converted into annual values.
      −0.020–0.005
      Disutilities converted into annual values.
      Minor hypoglycemia eventCurrie et al.
      • Currie C.J.
      • Morgan C.L.
      • Poole C.D.
      • et al.
      Multivariate models of health-related utility and the fear of hypoglycaemia in people with diabetes.
      −0.014−0.004
      Disutilities converted into annual values.
      −0.031 to −0.001
      Disutilities converted into annual values.
      CI, confidence interval; T2DM, type 2 diabetes mellitus.
      low asterisk Estimated from the standard error values provided.
      Estimated from the interquartile range values provided.
      Disutilities converted into annual values.
      Figure thumbnail gr2
      Fig. 2Preferred utility values for modeling complications associated with T2DM and 95% confidence interval. BMI, body mass index; CI, confidence interval; DR, diabetic retinopathy; T2DM, type 2 diabetes mellitus.

      Discussion

      This article presents a range of utility values identified by doing a systematic literature review of diabetes and its complications and recommends a proposed reference case in accordance with the NICE reference case for use in economic analyses. Relevant values elicited with the EQ-5D questionnaire and in patients with T2DM have been identified for all prespecified T2DM complications, except renal transplant after diabetic nephropathy. Values elicited in the UK population could be identified for most complications using the value set published by Dolan et al. [
      • Dolan P.
      Modeling valuations for EuroQol health states.
      ].
      For the preferred utility value set, most utility values were extracted from Clarke et al. [
      • Clarke P.
      • Gray A.
      • Holman R.
      Estimating utility values for health states of type 2 diabetic patients using the EQ-5D (UKPDS 62).
      ] because of its large sample size, T2DM-specific nature, recognized strong methodological quality, and use of the EQ-5D questionnaire in a UK population [
      • Clarke P.
      • Gray A.
      • Holman R.
      Estimating utility values for health states of type 2 diabetic patients using the EQ-5D (UKPDS 62).
      ]. An advantage of selecting as many utility values as possible from a single study is to retain internal consistency. It should, however, be noted that for some of the complications, the sample size was relatively small. When values were not reported by Clarke et al., the utilities presented in Bagust and Beale [
      • Bagust A.
      • Beale S.
      Modelling EuroQol health-related utility values for diabetic complications from CODE-2 data.
      ] were selected whenever possible and appropriate. Although this study did not include UK patients, the large sample size, use of the EQ-5D questionnaire to elicit utility values associated specifically with T2DM-related complications, and the robust methodology made it a suitable alternative.
      For health states not selected from the two aforementioned articles, other references were used. For the utility values associated with dialysis, the only article identified using the EQ-5D questionnaire and a diabetic population was by Wasserfallen et al. [
      • Wasserfallen J.B.
      • Halabi G.
      • Saudan P.
      • et al.
      Quality of life on chronic dialysis: comparison between haemodialysis and peritoneal dialysis.
      ] who conducted a study including all chronic hemodialysis and peritoneal dialysis patients in 19 centers in western Switzerland. These values are proposed for use in the absence of more robust data specific to the diabetes population. Out of the two articles reporting the impact of diabetic retinopathy, the article from Fenwick et al. [
      • Fenwick E.K.
      • Xie J.
      • Ratcliffe J.
      • et al.
      The impact of diabetic retinopathy and diabetic macular edema on health-related quality of life in type 1 and type 2 diabetes.
      ] was preferred because it presented values for different degrees of severity for diabetic retinopathy and was conducted with appropriate methodology. Out of the three articles reporting utility values associated with minor and major hypoglycemia episodes, the article by Currie et al. [
      • Currie C.J.
      • Morgan C.L.
      • Poole C.D.
      • et al.
      Multivariate models of health-related utility and the fear of hypoglycaemia in people with diabetes.
      ] is recommended. Several articles present a disutility value for patients experiencing one or more events in a given period of time, but this was the only one that presents a disutility value per event for different levels of episode severity. No article presenting the disutility value associated with renal transplant after diabetic nephropathy elicited with the EQ-5D questionnaire and/or from the general UK population could be identified. Laupacis et al. [
      • Laupacis A.
      • Keown P.
      • Pus N.
      • et al.
      A study of the quality of life and cost-utility of renal transplantation.
      ] present a utility value of 0.820 for diabetic patients 12 months after transplant. This value is higher than the suggested baseline for diabetes without complication utility value (0.785) and was consequently rejected [
      • Laupacis A.
      • Keown P.
      • Pus N.
      • et al.
      A study of the quality of life and cost-utility of renal transplantation.
      ]. Therefore, the suggested utility value for renal transplant after diabetic nephropathy was 0.763 as published by Kiberd and Jindal [
      • Kiberd B.A.
      • Jindal K.K.
      Screening to prevent renal failure in insulin dependent diabetic patients: an economic evaluation.
      ].
      Several inherent difficulties were encountered in deriving a set of utilities from multiple and heterogeneous studies. First, the choice of a baseline utility value was problematic. For example, in the studies of interest that all had similar objectives and used the same instruments in comparable populations, the baseline utility values elicited varied between 0.711 and 0.94 [
      • Currie C.J.
      • Morgan C.L.
      • Poole C.D.
      • et al.
      Multivariate models of health-related utility and the fear of hypoglycaemia in people with diabetes.
      ,
      • Smith D.H.
      • Johnson E.S.
      • Russell A.
      • et al.
      Lower visual acuity predicts worse utility values among patients with type 2 diabetes.
      ]. The choice of the baseline utility value clearly exerts a considerable influence on predicted cost-effectiveness because it affects the incremental weight of a given complication on health-state utility. For the proposed utility value set, it was deemed most appropriate to use the value of 0.785 elicited by Clarke et al. [
      • Clarke P.
      • Gray A.
      • Holman R.
      Estimating utility values for health states of type 2 diabetic patients using the EQ-5D (UKPDS 62).
      ]. This was because it lies in the range of identified values for T2DM without complications and was taken from the article in which most complication values were selected. The baseline utility value should ideally be obtained through a meta-analysis if sufficient data meeting the NICE recommendation are available (i.e., collected from the same patient population using the EQ-5D questionnaire and valued using the same UK value set), and sensitivity analyses using the limits of the confidence interval should be performed.
      Second, there was a great uncertainty around the disutility associated with each individual complication. Although most studies report measures of uncertainty, some assumptions were required to estimate confidence intervals, which may not reflect the full range of uncertainty in the underlying studies. Most utility values selected in the suggested utility value set were adjusted for age, sex, and presence of multiple complications, but values from Wasserfallen et al. [
      • Wasserfallen J.B.
      • Halabi G.
      • Saudan P.
      • et al.
      Quality of life on chronic dialysis: comparison between haemodialysis and peritoneal dialysis.
      ] as well as from Kiberd and Jindal [
      • Kiberd B.A.
      • Jindal K.K.
      Screening to prevent renal failure in insulin dependent diabetic patients: an economic evaluation.
      ] were not. Unadjusted disutility values should be interpreted with caution, and sensitivity analyses should be performed to test the uncertainty around these values, especially for the most frequent diabetes complications.
      Third, there was considerable heterogeneity across the identified studies, particularly regarding study country, study size and type, and inclusion of diabetes type. Consequently, we found limited justification for attempting to pool estimates of disutility.
      Another challenge in creating a utility value set for use in modeling is the use of different value sets. In the present review, all index values were elicited using a UK value set except the index value associated with the presence of cataracts. None of the articles reported the values with two different value sets. A publication in the literature of diabetes, however, reveals that although there might be a difference in mean calculated index values, UK, US, and Japanese value sets showed equivalent psychometric properties [
      • Sakthong P.
      • Charoenvisuthiwongs R.
      • Shabunthom R.
      A comparison of EQ-5D index scores using the UK, US, and Japan preference weights in a Thai sample with type 2 diabetes.
      ]. It also revealed that the variation in estimated EQ-5D questionnaire index values has a marginal impact on the projected incremental cost utilization ratio. Nevertheless, when creating a utility value set for use in modeling, it would be preferable to select values using the same EQ-5D questionnaire value set or to get access to the original data to convert the data using the appropriate EQ-5D questionnaire value set.
      Another area of uncertainty concerns the variation in utility over time associated with the evolution of medical practice as well as among different subpopulations. The two main studies contributing to the proposed data set—CODE-2 and United Kingdom Prospective Diabetes Study—although still the best sources identified, reported on data collected some years ago. The authors of this article would recommend not selecting publications merely because they are contemporary but instead selecting them on how robust their methodology is, unless there have been recent progresses in the treatment of a complication.
      Finally, the impact of clinical events on health-related quality of life may vary over the course of the disease. Consequently, one might want to vary the disutility values associated with a complication depending on the time since onset. For the present set of diabetes preferred utility values, Clarke et al. reported the relationship between health state utilities and clinical events occurring in the previous year or prior to the previous year. The authors did not find a greater disutility during the year of the event versus the disutility for events that occurred during the previous years.
      Despite these above-mentioned limitations, the suggested data set for T2DM was derived from the available literature by transparent methods and might be used as a common input for economic evaluation of different technologies in line with the NICE reference case. The preferred parameter set is presented with confidence limits and ranges of available estimates to facilitate sensitivity analysis. The readers can also refer to Table 1, which presents the sample size and the demographic characteristics of the patients interviewed for each study, and may help to guide the choice toward a specific cohort when planning a cost-effectiveness study. There may be specific characteristics of novel interventions that justify additional or different values from those presented, but the preferred data set may be of help to future economic evaluations by providing a starting point for such considerations.
      Consistency in the use of statistical models and reporting would improve the comparability of utility-related research. In addition, further research surrounding the appropriate estimation of utility values for patients experiencing several complications would improve the current evidence base. This is likely to be of increasing importance for patients with T2DM with advanced disease because patients typically develop additional complications over time.

      Conclusions

      This systematic literature review generated a set of utility values for T2DM and its complications in line with NICE requirements. This research also illustrates, however, the difficulties associated with systematically selecting utility data for economic evaluations, mainly due to the heterogeneity between quality-of-life studies and the uncertainty around the elicited utility values. The applicability of the identified utility value set should be considered before applying it to a modeling study and appropriate sensitivity analyses should be conducted.
      Source of financial support: This literature review was initiated and supported by IMS Health and Eli Lilly.

      Supplementary materials

      References

        • Hex N.
        • Bartlett C.
        • Wright D.
        • et al.
        Estimating the current and future costs of type 1 and type 2 diabetes in the UK, including direct health costs and indirect societal and productivity costs.
        Diabet Med. 2012; 29: 855-862
        • Holman N.
        • Forouhi N.G.
        • Goyder E.
        • Wild S.H.
        The Association of Public Health Observatories (APHO) Diabetes Prevalence Model: estimates of total diabetes prevalence for England, 2010–2030.
        Diabet Med. 2011; 28: 575-582
      1. IMS Knowledge Link version 5. Drugs used in diabetes currently in phase III or in pre-registration. IMS Health, Lifecycle R&D Focus Source. Available from: http://knowledgelink.imshealth.com/. [Accessed March 13, 2013].

      2. National Institute for Health and Care Excellence. Guide to the methods of technology appraisal. 2008. Available from: http://www.nice.org.uk/media/B52/A7/TAMethodsGuideUpdatedJune2008.pdf. [Accessed March 13, 2013].

        • Papaioannou D.
        • Brazier J.
        • Paisley S.
        TSD 9 The Identification, Review and Synthesis of Health State Utility Values from the Literature.
        Decision Support Unit, School of Health and Related Research. University of Sheffield, Sheffield, UK2010
        • Brazier J.
        • Longworth L.
        TSD 8 An Introduction to the Measurement and Valuation of Health for NICE Submissions.
        Decision Support Unit, ScHARR, University of Sheffield, Sheffield, UK,2011
      3. National Institute for Health and Care Excellence. Single technology appraisal (STA), Specification for manufacturer/sponsor submission of evidence. Rotterdam, The Netherlands: NICE, 2012.

        • Rabin R.
        • Oemar M.
        • Oppe M.
        EQ-5D-3L User Guide (Version 4.0).
        Rotterdam, The Netherlands: EuroQol Group. 2011;
        • Holmes J.
        • McGill S.
        • Kind P.
        • et al.
        Health-related quality of life in type 2 diabetes (TARDIS-2).
        Value Health. 2000; 3: 47-51
        • Wexler D.J.
        • Grant R.W.
        • Wittenberg E.
        • et al.
        Correlates of health-related quality of life in type 2 diabetes.
        Diabetologia. 2006; 49: 1489-1497
        • Conner-Spady B.
        • Suarez-Almazor M.E.
        Variation in the estimation of quality-adjusted life-years by different preference-based instruments.
        Med Care. 2003; 41: 791-801
        • Kopec J.A.
        • Willison K.D.
        A comparative review of four preference-weighted measures of health-related quality of life.
        J Clin Epidemiol. 2003; 56: 317-325
        • Palmer A.J.
        • Roze S.
        • Valentine W.J.
        • et al.
        The CORE Diabetes Model: projecting long-term clinical outcomes, costs and cost-effectiveness of interventions in diabetes mellitus (types 1 and 2) to support clinical and reimbursement decision-making.
        Curr Med Res Opin. 2004; 20: S5-26
        • Clarke P.M.
        • Gray A.M.
        • Briggs A.
        • et al.
        A model to estimate the lifetime health outcomes of patients with type 2 diabetes: the United Kingdom Prospective Diabetes Study (UKPDS) Outcomes Model (UKPDS no. 68).
        Diabetologia. 2004; 47: 1747-1759
        • McEwan P.
        • Peters J.R.
        • Bergenheim K.
        • Currie C.J.
        Evaluation of the costs and outcomes from changes in risk factors in type 2 diabetes using the Cardiff stochastic simulation cost-utility model (DiabForecaster).
        Curr Med Res Opin. 2006; 22: 121-129
        • Mount Hood 4 Modeling Group
        Computer modeling of diabetes and its complications: a report on the Fourth Mount Hood Challenge Meeting.
        Diabetes Care. 2007; 30: 1638-1646
        • Lung T.W.
        • Hayes A.J.
        • Hayen A.
        • et al.
        A meta-analysis of health state valuations for people with diabetes: explaining the variation across methods and implications for economic evaluation.
        Qual Life Res. 2011; 20: 1669-1678
        • Currie C.J.
        • Morgan C.L.
        • Poole C.D.
        • et al.
        Multivariate models of health-related utility and the fear of hypoglycaemia in people with diabetes.
        Curr Med Res Opin. 2006; 22: 1523-1534
        • Smith D.H.
        • Johnson E.S.
        • Russell A.
        • et al.
        Lower visual acuity predicts worse utility values among patients with type 2 diabetes.
        Qual Life Res. 2008; 17: 1277-1284
        • Dolan P.
        Modeling valuations for EuroQol health states.
        Med Care. 1997; 35: 1095-1108
        • Clarke P.
        • Gray A.
        • Holman R.
        Estimating utility values for health states of type 2 diabetic patients using the EQ-5D (UKPDS 62).
        Med Decis Making. 2002; 22: 340-349
        • Bagust A.
        • Beale S.
        Modelling EuroQol health-related utility values for diabetic complications from CODE-2 data.
        Health Econ. 2005; 14: 217-230
        • Wasserfallen J.B.
        • Halabi G.
        • Saudan P.
        • et al.
        Quality of life on chronic dialysis: comparison between haemodialysis and peritoneal dialysis.
        Nephrol Dial Transplant. 2004; 19: 1594-1599
        • Fenwick E.K.
        • Xie J.
        • Ratcliffe J.
        • et al.
        The impact of diabetic retinopathy and diabetic macular edema on health-related quality of life in type 1 and type 2 diabetes.
        Invest Ophthalmol Vis Sci. 2012; 53: 677-684
        • Laupacis A.
        • Keown P.
        • Pus N.
        • et al.
        A study of the quality of life and cost-utility of renal transplantation.
        Kidney Int. 1996; 50: 235-242
        • Kiberd B.A.
        • Jindal K.K.
        Screening to prevent renal failure in insulin dependent diabetic patients: an economic evaluation.
        BMJ. 1995; 311: 1595-1599
        • Sakthong P.
        • Charoenvisuthiwongs R.
        • Shabunthom R.
        A comparison of EQ-5D index scores using the UK, US, and Japan preference weights in a Thai sample with type 2 diabetes.
        Health Qual Life Outcomes. 2008; 6: 1
        • Glasziou P.
        • Alexander J.
        • Beller E.
        • Clarke P.
        Which health-related quality of life score? A comparison of alternative utility measures in patients with type 2 diabetes in the ADVANCE trial.
        Health Qual Life Outcomes. 2007; 5: 21
        • Kontodimopoulos N.
        • Pappa E.
        • Chadjiapostolou Z.
        • et al.
        Comparing the sensitivity of EQ-5D, SF-6D and 15D utilities to the specific effect of diabetic complications.
        Eu J Health Econ. 2012; 13: 111-120
        • Langelaan M.
        • de Boer M.R.
        • van Nispen R.M.
        • et al.
        Impact of visual impairment on quality of life: a comparison with quality of life in the general population and with other chronic conditions.
        Ophthalmic Epidemiol. 2007; 14: 119-126
        • Lamers L.M.
        • Stalmeier P.F.
        • McDonnell J.
        • et al.
        Measuring the quality of life in economic evaluations: the Dutch EQ-5D tariff [in Dutch].
        Ned Tijdschr Geneeskd. 2005; 149: 1574-1578
        • Lee W.J.
        • Song K.H.
        • Noh J.H.
        • et al.
        Health-related quality of life using the EuroQol 5D questionnaire in Korean patients with type 2 diabetes.
        J Korean Med Sci. 2012; 27: 255-260
        • Jo M.W.
        • Yun S.C.
        • Lee S.I.
        Estimating quality weights for EQ-5D health states with the time trade-off method in South Korea.
        Value Health. 2008; 11: 1186-1189
        • Lloyd A.
        • Nafees B.
        • Gavriel S.
        • et al.
        Health utility values associated with diabetic retinopathy.
        Diabet Med. 2008; 25: 618-624
        • Marrett E.
        • Stargardt T.
        • Mavros P.
        • Alexander C.M.
        Patient-reported outcomes in a survey of patients treated with oral antihyperglycaemic medications: associations with hypoglycaemia and weight gain.
        Diabetes Obes Metab. 2009; 11: 1138-1144
        • Shaw J.W.
        • Johnson J.A.
        • Coons S.J.
        US valuation of the EQ-5D health states: development and testing of the D1 valuation model.
        Med Care. 2005; 43: 203-220
        • Matza L.S.
        • Yurgin N.
        • Boye K.S.
        • et al.
        Obese versus non-obese patients with type 2 diabetes: patient-reported outcomes and utility of weight change.
        Curr Med Res Opin. 2007; 23: 2051-2062
        • O’Reilly D.J.
        • Xie F.
        • Pullenayegum E.
        • et al.
        Estimation of the impact of diabetes-related complications on health utilities for patients with type 2 diabetes in Ontario, Canada.
        Qual Life Res. 2011; 20: 939-943
        • Quah J.H.M.
        • Luo N.
        • Ng W.Y.
        • et al.
        Health-related quality of life is associated with diabetic complications, but not with short-term diabetic control in primary care.
        Ann Acad Med Singapore. 2011; 40: 276-286
        • Redekop W.K.
        • Koopmanschap M.A.
        • Stolk R.P.
        • et al.
        Health-related quality of life and treatment satisfaction in Dutch patients with type 2 diabetes.
        Diabates Care. 2002; 25: 458-463
        • Solli O.
        • Stavern K.
        • Kristiansen I.S.
        Health-related quality of life in diabetes: the associations of complications with EQ-5D scores.
        Health Qual Life Outcomes. 2010; 8: 18
        • Sullivan P.W.
        • Ghushchyan V.H.
        • Ben-Joseph R.
        The impact of obesity on diabetes, hyperlipidemia and hypertension in the United States.
        Qual Life Res. 2008; 17: 1063-1071
        • Vexiau P.
        • Mavros P.
        • Krishnarajah G.
        • et al.
        Hypoglycaemia in patients with type 2 diabetes treated with a combination of metformin and sulphonylurea therapy in France.
        Diabetes Obes Metab. 2008; 10: 16-24