Article Text

Download PDFPDF

Health state utility values in schizophrenia: protocol for a systematic review and meta-analysis
  1. David Aceituno1,2,
  2. Mark Pennington1,
  3. Barbara Iruretagoyena2,
  4. Matthew A Prina1,
  5. Paul McCrone1
  1. 1 Health Service and Population Research, Institute of Psychiatry Psychology and Neuroscience, London, UK
  2. 2 Department of Psychiatry, School of Medicine, Pontificia Universidad Católica de Chile, Santiago, Chile
  1. Correspondence to David Aceituno, Institute of Psychiatry Psychology and Neuroscience, Health Service and Population Research, London SE5 8AF, UK; david.aceituno_farias{at}


Introduction Cost-effectiveness analyses that use quality-adjusted life-years (QALYs) allow comparing the value for money of interventions across different health problems. Health state utility values (HSUVs) are crucial to calculate QALYs. These are weights attached to a given health state reflecting preferences in health-related quality of life (HRQoL). In schizophrenia, there is extensive evidence about the consequences of this condition on HRQoL. Besides, several interventions have claimed to be cost-effective in terms of QALYs gained. Despite this evidence, a systematic review of HSUVs has not been conducted. Therefore, we aim to synthesise the evidence about HSUVs in schizophrenia.

Methods and analysis We will conduct a systematic review of the literature about HSUVs in people with schizophrenia following the Preferred Reporting Items for Systematic review and Meta-Analysis and the International Society for Pharmacoeconomics and Outcomes Research task force recommendations. The submissions records of eight electronic peer-reviewed databases and three health technology assessment (HTA) agencies will be searched. Quantitative synthesis will be carried out in comparable studies, using random-effects meta-analysis. Heterogeneity will be explored using meta-regression if more than 10 studies per covariate are found. A narrative synthesis and methodological quality of included studies will be also reported.

Discussion This review will provide a synthesis of the HSUVs estimated for different states experienced by people with schizophrenia. This will inform analysts when calculating QALYs, using values in a more transparent and accountable manner. Finally, it will shed light on evidence gaps and limitations about this measure in mental health.

PROSPERO registration number CRD42019123582.

  • adult psychiatry

Statistics from



Schizophrenia is a chronic psychotic disorder that carries a substantial burden on patients, caregivers and society as a whole. For instance, having schizophrenia is associated with an average reduction in life expectancy of at least 15 years compared with the general population.1 2 It is, furthermore, one of the most disabling and costliest conditions according to the global burden of disease studies3 and several systematic reviews,4 5 respectively.

The recommended treatment for people with schizophrenia includes a combination of psychosocial interventions and pharmacotherapy.6 However, there are currently several options available not only in terms of medications but also in psychosocial interventions. For example, patients with schizophrenia might receive different antipsychotics with varied adverse effects, by different routes (oral, long-acting injections) and in combination with diverse psychosocial interventions, such as cognitive behavioural therapy, family therapy, psychoeducation and vocational rehabilitation.6 7

These alternatives differ in terms of acceptability, effectiveness and costs. As a result, in the context of a globally constrained mental health budget,8 decision makers face the dilemma of which alternatives to fund and which to not fund. Furthermore, every time a coverage decision is made, other programmes, interventions or services are not funded, which imposes an opportunity cost on other patients.9 10

Health systems have progressively adopted the evidence derived from cost-effectiveness analyses (CEAs) to support these difficult decisions. In order to accomplish this aim in an accountable manner, some countries have made explicit their preference for CEA that include quality-adjusted life-years (QALYs) as the outcome of interest.11 QALYs combine the quantity and quality of life derived from interventions in a single numeraire, which allow the comparison of therapeutic (or diagnostic) alternatives across diverse populations.12

In order to calculate QALYs, life-years gained are adjusted by a preference weight. This weight is obtained asking people to give their preference for different health states in terms of the impact on health-related quality of life (HRQoL). This score is presented in a scale where 1 means full health, 0 is as bad as being dead and negative values represent HRQoL worse than death. These values are known as health state utility values (HSUVs).10 12

There are different approaches to estimate HSUVs. These include direct methods, such as the Standard Gamble (SG), Time Trade-Off (TTO) and Visual Analogue Scale (VAS); indirect methods such as Multiattribute Health Status Classification Systems with preference scores (eg, EuroQoL 5-Dimensions (EQ-5D), Health Utilities Index) and mapping from non-preference-based measures onto generic preference-based measures of health (eg, mapping from Positive and Negative Syndrome Scale (PANSS) to EQ-5D13).

Regardless of the method used to derive HSUVs, health technology assessment (HTA) agencies and consensus guidelines agree on the need for transparency and rigour when HSUVs are included in economic evaluations.12 14 15 The first step they recommend is to obtain a systematic review of the literature to inform estimates previously elicited.15 16

As far as we know, there is no published systematic review specifically about HSUVs in schizophrenia. In 2010, Mavranezouli17 reviewed the studies reporting utility values for schizophrenia that were utilised particularly in the economic analysis of the 2009 National Institute for Health and Clinical Excellence (NICE) clinical guideline. The review found seven studies reporting HSUVs, although the focus was on the suitability of current instruments to measure HRQoL in schizophrenia more than on the utility values for different health states in schizophrenia. More recently, Németh et al 18 systematically reviewed the evidence of economic models in schizophrenia, focusing on mapping utility algorithms, leaving out other methods to estimate HSUVs.


Considering the available evidence and the need for transparency in the use of HSUVs in HTA, we aim to review systematically the published utility values for adults with schizophrenia.

Methods and analysis

It has been recognised that some of the classic systematic review methods do not apply to HSUVs reviews.19–21 For instance, a Patients, Interventions, Comparators, and Outcomes (PICO) question is not the best framework to present the research question, given the absence of interventions and comparators. Also, the lack of MeSH terms challenges the implementation of a search with high specificity. We will use search filters specially developed for HSUVs22 in order to improve the precision of search results. Finally, publication bias appears to be less of a concern when seeking HSUV estimates, given that they are usually secondary outcomes, although this deserves more research.

This protocol seeks to meet transparency standards to inform HSUVs for use in economic models including HSUVs for schizophrenia. It has been developed based on the Preferred Reporting Items for Systematic review and Meta-Analysis Protocols (PRISMA-P)23 and published recommendations.19–21

This protocol was registered with Prospective Register of Systematic Reviews.

Eligibility criteria

We will include studies reporting HSUVs empirically estimated using a validated instrument in adults (>18 years old) with schizophrenia (defined according to structured criteria such as the Diagnostic and Statistical Manual (DSM) of mental disorders or the International Classification of Diseases (ICD)). Model- and trial-based economic evaluation will be included if they report novel HSUVs, allowing different study designs as recommended.19–21 No language restriction or time of follow-up will be applied. Studies reporting quality-of-life measures with no translation into utilities will be excluded. Finally, we will link the reports referring to the same sample and we will treat them as a single study.

Information sources

The search will include the following generic databases: Medline, PsycInfo, Embase, EconLit and Cochrane Library. In addition, the following specialised databases will be included: Cost-effectiveness Analysis Registry (formerly known as the Harvard Cost-Effectiveness Analysis database), Centre for Reviews and Dissemination, National Health System Economic Evaluation database, University of Sheffield School of Health and Related Research database and submissions to NICE, Canadian Agency for Drugs and Technologies in Health and Pharmaceutical Benefits Advisory Committee in Australia. This search will be supplemented by cross-referencing included studies and contacting authors in the field.

Search strategy

Our search strategy will include terms related to schizophrenia, such as ‘psychosis’ or ‘psychotic disorders’, as well as terms associated with HSUVs. We will use a published search filter22 and examples derived from previous recommendations,19 20 using multiple iterations for the same term. An example of the search strategy can be found in the online supplementary file 1.

Supplemental material

Selection process

Two reviewers will screen titles and abstract independently and will compare them against inclusion criteria. It has been noted that abstracts may not clearly report important terms in HSUVs reviews.21 Hence, studies will be excluded at this stage only if there is a clear agreement between both reviewers. This will imply the inclusion of a higher number of full texts as recommended.19 21 After reviewing the full text of selected abstracts, disagreements regarding final inclusion will be solved by discussion and the opinion of a third reviewer.

Data extraction and management

Data from selected studies will be extracted by two authors (DA and BI) independently and entered into a prepiloted electronic spreadsheet (online supplementary file 2A). All data entries will be compared between author and disagreement will be solved by discussion with a third author. In the case of incompleteness or doubts about any data, principal investigators will be contacted. Search strategy, records and selected studies will be handled using Rayyan, a free web-based software designed to accelerate the systematic review process.24 Data used will be those publicly available; therefore, no specific strategy to keep data confidential will be used.

Data items

We will extract information about the selected studies in terms of general characteristics, methods and quantitative results. A detailed list of data items might be found in the online supplementary file 2A.

Risk of bias of individual studies

There is no instrument specifically designed to assess the risk of bias of HSUV studies.16 Applying current instruments might be inappropriate, as the classical pyramid of evidence does not apply to the risk of bias in HSUV estimates. Observational studies may be a good source of evidence, considering that usually HSUVs seek to estimate general population values. Furthermore, this is an area where available recommendations do not agree.19–21

Consequently, this systematic review will provide a descriptive picture of methodological quality for every study (online supplementary file 2B), following the recommendations of the NICE decision support unit25 and the ISPOR Good Practices for Outcome Research task force.16

Data synthesis

There is some debate about the importance of a pooled estimate of HSUVs. Detractors consider that pooling results from heterogeneous studies might give a biased estimate of health states values, and meta-regression techniques might result in a considerable amount of false negatives and positives.26 To others, as long as a sufficient number of comparable studies are included, mixed-effect models are useful to explore reasons for heterogeneity and might be ideal to inform decision models, which ought to reflect all the available evidence.19 20

This review aims to inform decision models in schizophrenia; therefore, a pooled estimate of the HSUVs is considered useful. However, we will perform different random-effects meta-analyses for direct and indirect valuation methods. This approach has been suggested given the incommensurability of the different methods to estimate HSUVs.19 Such analyses will use the inverse of variance as mean-weighting method.27

After that, a mixed-effect meta-regression will be applied if more than 10 studies per covariates are available.19 28 Predictor variables included will be age, sex, responder type (eg, patients, caregivers, health professionals) and valuation method (eg, SG, TTO, VAS). We will explore the inclusion of other covariates and will use a stepwise procedure of model selection in order to reduce the likelihood of errors.


As previously mentioned, publication bias is not perceived to be a serious risk in HSUV reviews. As a result, there is no mention of publication bias in previous HSUV systematic reviews. Consequently, tests for funnel plot asymmetry (eg, Egger test) will not be carried out.


Decision models need to be populated with transparent and systematically obtained parameters. This has been acknowledged as a crucial factor that affects cost-effectiveness estimates29 and eventually policy decisions.

HSUVs are intended to capture people’s preference for certain health states, and they are likely to have a large impact on cost-effectiveness estimates when QALYs are the outcomes of interest. Therefore, analysts must be aware of the available literature to justify values included in their models.

We will systematically review the evidence about HSUVs in schizophrenia in order to provide published estimates for those modelling this disabling and costly condition. We will describe the strengths and limitations of the current literature, and we will attempt to apply a meta-regression if enough studies are obtained. This will allow us to account for heterogeneity and other covariates in utility estimates, which should enable more precise CEA.

We believe this protocol adds transparency to the systematic review process. It has been designed according to PRISMA-P recommendations. Similar work has been published in eye disease,30 and we think this is a desirable approach to take, in order to increase accountability of decision models in HTA.

Supplemental material



  • Contributors DA and MP participated in the conception, drafting, revising and final approval of this manuscript.BI participated in the conception, revising and final approval. MAP and PMC participated in the drafting, revising and final approval.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Patient consent for publication Not required.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.