Edited by: Kevin Lu, University of South Carolina, United States
Reviewed by: Zafer Çaliskan, Hacettepe University, Turkey; Nan Yang, Sichuan University, China
This article was submitted to Health Economics, a section of the journal Frontiers in Public Health
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Schizophrenia is a severe and complex mental illness with early onset coupled with behavior or cognitive disorders that have a significant impact on patients' family and society. A systematic review reported that the global age-standardized prevalence of schizophrenia was 0.28% and the prevalence of cases rose from 13.1 million cases in 1990 to 20.9 million cases in 2016 (
Economic evaluations could generate evidence incorporating both costs and consequences for decision makers to clarify different uses for scarce resources (
Previous systematic reviews have evaluated studies published since 2000 (
The systematic review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement developed by Moher et al. (
The inclusion criteria were as follows: (a) economic evaluations adopting cost-effectiveness analysis (CEA) or cost-utility analysis (CUA) approach; (b) patients diagnosed with schizophrenia with no limitation on gender or age; (c) intervention including all antipsychotics; and (d) outcomes presented as incremental cost-effectiveness (ICER). Studies were excluded if they met the any of the following criteria: (a) not reported in English; (b) not related to economic evaluation; (c) cost of illness, health-related quality of life or budget impact analysis studies; (d) abstracts or studies with full-text unavailable; (e) not model-based studies; and (f) chose clinical effect as the only outcome.
An electronic literature search was performed in PubMed, Web of Science, EBSCO host, The Cochrane Library, ScienceDirect from January 2014 to December 2020. Search items included “schizophrenia,” “schizophrenic,” “pharmacoeconomics,” “economic evaluation,” “cost-effectiveness,” and “cost-utility.” The detailed strategy is provided in
The included studies were screened, extracted and double checked by two researchers independently. Disagreements were resolved by discussion or by consulting a third researcher. General information was collected including title, first author's surname, year of publication, country or region, intervention, and treatment sequences. To summarize the methods applied, characteristics such as perspectives, type of costs, outcomes, model structures, and necessary parameters were recorded. The results and conclusions of studies were included in the extracted form but were not reported as main outcomes due to the arguments regarding the extrapolation of evaluation results (
According to the review of quality assessment tools conducted by Walker et al. (
A total of 1,086 citations were retrieved from five electronic databases. After removing duplicates, 610 studies were eligible to enter the screening process and judgements were generated according to the titles and abstracts. Finally, 25 articles published in English were identified and included in the systematic review. A flow of the literature screening was provided in
Flow of information through different phases of the systematic review according to the PRISMA statement [#Detailed search results: PubMed Database (
The characteristics of the included studies were summarized in
Basic characteristics.
Einarson et al. ( |
Sweden | Schizophrenia patients with relapse | Paliperidone LAI, olanzapine LAI, risperidone LAI, haloperidol LAI, oral olanzapine | Health care sector perspective | Direct and indirect costs | QALYs | NA |
Lachaine et al. ( |
Canada | Moderate to severe schizophrenia patients above 40 | Asenapine, olanzapine | Health care sector perspective, societal perspective | Direct and indirect costs | QALYs | 5% |
Park et al. ( |
United States | Schizophrenia patients above 40 | Olanzapine, risperidone, quetiapine, ziprasidone | Health care sector perspective | Direct costs | QALYs | 3% |
Dilla et al. ( |
Spain | Schizophrenia patients in relapse due to low compliance | Olanzapine LAI, risperidone LAI | Health care sector perspective | Direct costs | QALYs, relapse averted, life years | 3% |
Anh et al. ( |
Vietnam | Schizophrenia patients above 15 | Chlorpromazine, haloperidol, levopromazine, risperidone, clozapine, olanzapine | Health care sector perspective | Direct costs | DALYs | 3% |
Lubinga et al. ( |
Uganda | Schizophrenia patients with average age of 25 | Chlorpromazine, haloperidol, risperidone, olanzapine, quetiapine | Societal perspective | Direct and indirect costs | DALYs | 3% |
Druais et al. ( |
France | Stable schizophrenia patients with average age of 38 | Paliperidone LAI, risperidone LAI, aripiprazole LAI, olanzapine LAI, haloperidol LAI, oral olanzapine | Payer perspective | Direct costs | QALYs, relapse averted | 4% |
Lin et al. ( |
Singapore | Schizophrenia patients with average age of 37 | Amisulpride, aripiprazole, chlorpromazine, olanzapine, paliperidone, quetiapine, risperidone, sulpiride, trifluoperazine, ziprasidone | Health care sector perspective | Direct costs | QALYs | 3% |
Rajagopalan et al. ( |
Scotland and Wales | Schizophrenia patients in relapse | Lurasidone, aripiprazole | Payer perspective | Direct costs | QALYs | 3.50% |
Einarson et al. ( |
Finland | Schizophrenia patients in relapse | Aripiprazole LAI, paliperidone LAI, olanzapine LAI, risperidone LAI | Health care sector perspective | Direct costs | QALYs, relapse averted | NA |
Einarson et al. ( |
Portugal | Schizophrenia patients | Paliperidone LAI, risperidone LAI, haloperidol LAI, oral olanzapine | Health care sector perspective | Direct costs | QALYs, relapse averted | NA |
Barnes et al. ( |
United Kingdom | Patients unresponsive to clozapine | Olanzapine, amisulpride | Societal perspective | Direct costs | QALYs | Not specified |
Einarson et al. ( |
Spain | Schizophrenia patients | PP3M, PP1M | Health care sector perspective | Direct costs | QALYs, relapse averted, hospitalization averted | NA |
Einarson et al. ( |
Netherlands | Schizophrenia patients | PP3M, PP1M, haloperidol LAI, risperidone microspheres, oral olanzapine | Payer perspective | Direct costs | QALYs, relapse (hospitalization treated or out-patient treated) | NA |
Wiwat et al. ( |
Thailand | Stable schizophrenia patients above 15 | Aripiprazole, risperidone | Societal perspective | Direct costs | QALYs | 3% |
Nuhoho et al. ( |
United Arab Emirates | Stable schizophrenia patients | Paliperidone LAI, other oral antipsychotics | Payer perspective | Direct costs | QALYs, rate of hospitalization, relapse, emergency | NA |
Aigbogun et al. ( |
United States | Stable schizophrenia patients | Brexpiprazole, cariprazine, lurasidone | Payer perspective | Direct costs | QALYs, relapse averted, hospitalization averted | NA |
Németh et al. ( |
Hungary | Patients with negative symptoms of schizophrenia with average age of 40 | Cariprazine, risperidone | Payer perspective | Direct costs | QALYs | 4% |
Zhao et al. ( |
China | Schizophrenia patients | Olanzapine ODT, olanzapine SOT, aripiprazole SOT | Payer perspective | Direct costs | QALYs, averaged annual relapse | NA |
Abdall-Razak et al. ( |
United Kingdom | Not specified | Paliperidone, amisulpride | Payer perspective | Direct costs | QALYs | NA |
Dutina et al. ( |
Serbian | Adult patients about to receive for the second-line treatment | Aripiprazole, olanzapine | Payer perspective | Direct costs | QALYs | 3% |
Arteaga et al. ( |
France | Adult chronic schizophrenic patients stabilized on PP1M with baseline age of 38.75 | PP3M, PP1M | Payer perspective | Direct costs | QALYs | 4% |
Yi et al. ( |
China | Schizophrenia patients with starting age of 35 | Amisulpride, olanzapine | Payer perspective | Direct costs | QALYs | 3% |
Lin et al. ( |
China | Not specify | Aripiprazole ODT, aripiprazole SOT, olanzapine SOT | Payer perspective | Direct costs | QALY | NA |
Jin et al. ( |
United Kingdom | Individuals referred to secondary care mental health services with mean age of 23.5 | Amisulpride, aripiprazole, haloperidol, olanzapine, quetiapine, risperidone, placebo, clozapine | Payer perspective | Direct costs | QALYs | 3.5% |
Due to the diverse efficacy and low compliance rate, therapeutic changes are common among schizophrenia patients, which makes treatment sequences worth consideration. Nineteen studies (76%) (
Treatment sequences
Einarson et al. ( |
Yes | Paliperidone LAI or olanzapine LAI or risperidone LAI or oral olanzapine or haloperidol LAI | Olanzapine LAI or paliperidone LAI or haloperidol LAI or oral olanzapine |
Clozapine |
Lachaine et al. ( |
Yes | Asenapine or olanzapine | Aripiprazole or ziprazidone or risperidone or quetiapine | NA |
Park et al. ( |
Yes | Olanzapine or risperidone or quetiapine or ziprasidone | Olanzapine or risperidone or quetiapine or ziprazidone |
Clozapine |
Dilla et al. ( |
Yes | Olanzapine LAI or risperidone LAI | Other antipsychotics | NA |
Anh et al. ( |
No | Chlorpromazine or haloperidol or levopromazine or risperidone or clozapine or olanzapine | NA | NA |
Lubinga et al. ( |
Yes | Chlorpromazine or haloperidol or risperidone or olanzapine or quetiapine | Risperidone or haloperidol |
Perphenazine |
Druais et al. ( |
Yes | Paliperidone LAI or risperidone LAI or aripiprazole LAI or olanzapine LAI or haloperidol LAI or oral olanzapine | 25%paliperidoneLAI+25%risperidone LAI+25%aripiprazoleLAI+25%olanzapine LAI |
NA |
Lin et al. ( |
Yes | Amisulpride or aripiprazole or chlorpromazine or haloperidol or olanzapine or paliperidone or quetiapine or risperidone or sulpiride or trifluoperazineor ziprazidone | The other drugs excluded the first-line drugs |
Clozapine |
Rajagopalan et al. ( |
Yes | Lurasidone or aripiprazole | Amisulpride | Clozapine |
Einarson et al. ( |
Yes | Aripiprazole or paliperidone or olanzapine or risperidone | Olanzapine or risperidone |
Clozapine |
Einarson et al. ( |
Yes | Paliperidone or risperidone or haloperidol or olanzapine | Haloperidol or olanzapine |
Clozapine |
Barnes et al. ( |
No | Olanzapine or amisulpride | NA | NA |
Einarson et al. ( |
Yes | PP3M or PP1M | Aripiprazole | Clozapine |
Einarson et al. ( |
Yes | PP3M or PP1M or haloperidol or risperidone or olanzapine | Haloperidol or oral olanzapine | Clozapine |
Wiwat et al. ( |
Yes | Aripiprazole or risperidone | Clozapine | NA |
Nuhoho et al. ( |
Yes | Paliperidone LAI or paliperidone LAI plus oral antipsychotics | Risperidone or paliperidone or aripiprazole or olanzapine or quetiapine |
Risperidone or paliperidone or aripiprazole or olanzapine or quetiapine |
Aigbogun et al. ( |
Yes | Brexpiprazole or cariprazine or lurasidone | Olanzapine or risperidone or quetiapine or ziprazidone or aripiprazole |
NA |
Németh et al. ( |
No | Cariprazine or risperidone | NA | NA |
Zhao et al. ( |
Yes | Olanzapine ODT or olanzapine SOT or aripiprazole SOT | Aripiprazole or amisulpride or ziprazidone or clozapine | NA |
Abdall-Razak et al. ( |
No | Paliperidone or amisulpride | NANA | |
Dutina et al. ( |
Yes | Aripiprazole or olanzapine | Clozapine | NA |
Arteaga et al. ( |
Yes | PP3M or PP1M | Paliperidone, olanzapine, aripiprazole, risperidone |
NA |
Yi et al. ( |
No | Amisulpride or olanzapine | NA | NA |
Lin et al. ( |
Yes | Aripiprazole ODT or aripiprazole SOT or olanzapine SOT | Not specify | NA |
Jin et al. ( |
No | Amisulpride, aripiprazole, haloperidol, olanzapine, quetiapine, risperidone, placebo, clozapine | NA | NA |
Various terms were used to define the study perspectives of the reviewed articles. Therefore, the terms were classified into three categories in this review, which were defined as health care sector perspective (including healthcare system, ministry of health, national health service, and government), payer perspective (including payer, third-party payer, and health insurance), and societal perspective (including societal, modified societal, and broadly societal perspective) based on the report recommended by ISPOR (
Among the eight studies (32%) choosing the health care sector perspective (
The characteristics of the models and related health states are summarized in
Basic characteristics of the models in the included economic evaluations.
Einarson et al. ( |
Decision tree, cohort | Incorporating clinical events including discontinuation, exacerbation, compliance, hospitalization | 1 year | NA |
Lachaine et al. ( |
Decision tree combined with 9-state Markov model, cohort | Diabetes, stroke, CHDs, hypertension, no comorbidity, 2/3/4 comorbidities, death | 5–10 years | 1 year |
Park et al. ( |
9-state Markov model, cohort | First line treatment with/without irreversible SE, 2nd line treatment with/without irreversible SE, clozapine treatment with/without irreversible SE, uncontrolled state with/without SE, death | 10 years | 18 weeks |
Dilla et al. ( |
Discrete event simulation, microsimulation | Treatment, treatment emergent adverse events, relapse, doctor-initiated treatment re-evaluation, patient-initiated treatment discontinuation | 5 years | NA |
Anh et al. ( |
3-state Markov model, cohort | Schizophrenia patients, recovery patients, schizophrenia-specific and other causes of deaths | Lifetime | 1 year |
Lubinga et al. ( |
10-state Markov model, cohort | Residual on/off 1st line AP, acute on/off 1st line AP, residual on/off 2nd line AP, acute on/off 2nd line AP, residual on 3rd line | Lifetime | 1 year |
Druais et al. ( |
4-state Markov model, cohort | Stable treated, stable non-treated, relapse, death | 5 years | 3 months |
Lin et al. ( |
4-state Markov model, cohort | Stable treated, stable non-treated, relapse, death | Lifetime | 1 year |
Rajagopalan et al. ( |
5-state Markov model, cohort | Non-stable/relapse trial of antipsychotic agents, stable/adherent, stable/non-adherent, relapse, death | 10 years | 6 weeks |
Einarson et al. ( |
decision tree, cohort | Incorporating clinical events including discontinuation, exacerbation, compliance, hospitalization | 1 year | NA |
Einarson et al. ( |
decision tree, cohort | Incorporating clinical events including discontinuation, exacerbation, compliance, hospitalization | 1 year | NA |
Barnes et al. ( |
3-state Markov model, cohort | Symptom response, SEs, death | 1–10 years | 3 months |
Einarson et al. ( |
decision tree, cohort | Incorporating clinical events including discontinuation, exacerbation, compliance, hospitalization | 1 year | NA |
Einarson et al. ( |
decision tree, cohort | Incorporating clinical events including stable, intolerant, relapse treated as out-patient, relapse requiring hospitalization and dropout | 1 year | NA |
Wiwat et al. ( |
4-state Markov model, cohort model | Remission with 1st antipsychotics, relapse, remission with clozapine, death | Lifetime | 4 weeks |
Nuhoho et al. ( |
Decision tree, cohort | Incorporating clinical events including adherence, exacerbation, hospitalization | 1 year | 3 months |
Aigbogun et al. ( |
Decision-analytic model, cohort | Incorporating clinical events including treatment discontinuation, relapse/impending relapse, AEs | 1 year | NA |
Németh et al. ( |
8-state Markov model, cohort | Constructed according to both severity of symptoms and disease types | 1–10 years | 1/12 weeks |
Zhao et al. ( |
Decision-analytic model, microsimulation | Incorporating adherence levels, relapse with/without hospitalization, treatment discontinuation, AEs suicide risk | 1 year | 3 months |
Abdall-Razak et al. ( |
Decision tree, cohort | Incorporating relapse, remission, AEs, diabetes complications | 1 year | NA |
Dutina et al. ( |
5-state Markov model, cohort | Remission without AEs, remission with AEs, relapse, second response, death | 10 years | 3 months |
Arteaga et al. ( |
5-state Markov model, cohort | 1st-line treatment, no active treatment, 2nd-line treatment, relapse, death | 5 years | 1 month |
Yi et al. ( |
5-state Markov model, cohort | Acute phase, remission, relapse, death | Lifetime | 1 year |
Lin et al. ( |
Discrete event simulation, microsimulation | Incorporating adherence levels, relapse with/without hospitalization, stable and adverse events | 1 year | NA |
Jin et al. ( |
Discrete event simulation, microsimulation | Incorporating 4 module for different pathway with relevant interventions | Lifetime | NA |
As summarized in
The cycle lengths of Markov models in the included studies varied from 4 weeks to 1 year, where 3 months (
Adverse events (AEs) could impact adherence, efficacy and therapy changes as well as health-related quality of life, especially for schizophrenia. Thus, it is meaningful to take relevant AEs into account in the models. The AEs selected in the evaluations are listed in
Summary of the adverse events considered in the included economic evaluations.
Einarson et al. ( |
× | × | × | × | × | × | NA |
Lachaine et al. ( |
√ | √ | √ | √ | × | √ | NA |
Park et al. ( |
√ | √ | × | × | √ | √ | Agranulocytosis |
Dilla et al. ( |
√ | √ | √ | × | × | × | Somnolence, sexual dysfunction, postinjection syndrome, suicide |
Anh et al. ( |
√ | √ | √ | × | × | × | Agranulocytosis |
Lubinga et al. ( |
√ | √ | √ | √ | × | × | Ischemic heart disease |
Druais et al. ( |
√ | √ | √ | √ | × | × | NA |
Lin et al. ( |
√ | √ | √ | √ | × | × | NA |
Rajagopalan et al. ( |
√ | √ | √ | √ | × | × | NA |
Einarson et al. ( |
× | × | × | × | × | × | NA |
Einarson et al. ( |
× | × | × | × | × | × | NA |
Barnes et al. ( |
√ | √ | √ | × | × | √ | Sexual dysfunction, aversive subjective experience, cardiac symptoms |
Einarson et al. ( |
× | × | × | × | × | × | NA |
Einarson et al. ( |
× | × | × | × | × | × | NA |
Wiwat et al. ( |
√ | √ | √ | √ | √ | × | NA |
Nuhoho et al. ( |
× | × | × | × | × | × | NA |
Aigbogun et al. ( |
√ | √ | √ | × | × | √ | NA |
Németh et al. ( |
√ | √ | × | × | × | × | NA |
Zhao et al. ( |
√ | √ | √ | √ | × | √ | NA |
Abdall-Razak et al. ( |
√ | √ | √ | √ | × | × | Diabetes complications: amputation, MI, stroke, IHD, HF |
Dutina et al. ( |
√ | √ | × | × | × | √ | Neutropenia |
Arteaga et al. ( |
√ | √ | √ | √ | √ | × | NA |
Yi et al. ( |
√ | √ | √ | × | √ | √ | Liver function damage |
Lin et al. ( |
√ | √ | √ | √ | √ | √ | NA |
Jin et al. ( |
√ | √ | √ | √ | √ | Neutropenia |
However, some studies did not fully describe the consideration of AEs (
Medication compliance (also known as adherence) can be defined as the extent to which the medication-taking of a patient matches that defined by the prescriber while medication persistence (also known as continuous adherence or discontinuation rate) refers to the act confirming to the recommended continuing treatment for the duration of time from the prescriber (
Information about the adoption of compliance or persistence in the included studies is listed in
Summary of the methods used to integrate medication compliance.
Einarson et al. ( |
√ | √ | (a) Act as branches behind chance nodes, (b) influence the probabilities for events | √ | √ | √ | Simple average | |
Lachaine et al. ( |
× | × | NA | |||||
Park et al. ( |
× | √ | Lead to therapy changes | √ | Kaplan-Meier discontinuation curves | |||
Dilla et al. ( |
× | √ | Lead to therapy changes | √ | ||||
Anh et al. ( |
√ | × | Not specified | √ | Simple average | |||
Lubinga et al. ( |
× | √ | Lead to relapse | √ | NA | |||
Druais et al. ( |
× | √ | (a) Act as transition probabilities, (b) influence relapse of the disease | √ | NA | |||
Lin et al. ( |
× | √ | Lead to change or discontinuation of the therapy | √ | NA | |||
Rajagopalan et al. ( |
× | √ | Act as transition probabilities | √ | √ | Regression analysis, partial assumption | ||
Einarson et al. ( |
√ | √ | (a) Act as branches behind chance nodes, (b) influence the probabilities of events | √ | √ | √ | Partial assumption, simple average | |
Einarson et al. ( |
√ | √ | (a) Act as branches behind chance nodes, (b) influence the probabilities of events | √ | √ | √ | Partial assumption, simple average | |
Barnes et al. ( |
Not specified | NA | ||||||
Einarson et al. ( |
× | √ | (a) Act as branches behind chance nodes, (b) influence the probabilities of events | √ | NA | |||
Einarson et al. ( |
× | √ | (a) Lead to change or discontinuation of the therapy, (b) influence the relapse of the disease | √ | √ | NA | ||
Wiwat et al. ( |
√ | √ | Influence relapse of the disease | √ | NA | |||
Nuhoho et al. ( |
√ | √ | (a) Act as branches behind chance nodes, (b) influence the probabilities of events | √ | NA | |||
Aigbogun et al. ( |
× | √ | Lead to change or discontinuation of the therapy | √ | Indirect comparison based on data from clinical trials | |||
Németh et al. ( |
× | × | NA | |||||
Zhao et al. ( |
√ | √ | Classified different types of patients | √ | √ | √ | Assumption | |
Abdall-Razak et al. ( |
× | × | NA | |||||
Dutina et al. ( |
Not specified | NA | ||||||
Arteaga et al. ( |
× | √ | Act as transition probabilities | √ | √ | NA | ||
Yi et al. ( |
× | √ | Act as transition probabilities | NA | ||||
Lin et al. ( |
√ | √ | Classified different types of patients | √ | √ | √ | NA | |
Jin et al. ( |
√ | √ | Persistence rate acting as transition probabilities while non-compliance seen as reason for non-persistence | √ | √ | NA |
Various sources of reference for compliance or persistence were adopted in the evaluations, including retrospective studies, clinical trial studies, and naturalistic studies. Simple averaging was the major method adopted for multisource data. Most discontinuation rates were collected from clinical trials. Even though there might be differences between the definitions of compliance and persistence, most of the studies did not explain or discuss this issue. Considering the definition of compliance, data from real-world studies might be superior as no intervention has been implemented to influence the medication behaviors of patients.
According to the health states defined in the model, the most commonly used utilities in the studies were those for stable schizophrenia, relapse without hospitalization, relapse with hospitalization and the disutilities of EPS, weight gain, and diabetes. Utilities and references are summarized in
Summary of the utilities for health states of schizophrenia.
Einarson et al. ( |
0.89 | 0.659 | 0.49 | ( |
Lachaine et al. ( |
0.75 | NA | NA | ( |
Park et al. ( |
0.856 | NA | −0.358 | ( |
Dilla et al. ( |
0.77 | — | −0.18 | From SOHO data |
Druais et al. ( |
0.919 | 0.762 | 0.604 | ( |
Lin et al. ( |
0.8 | NA | 0.67 | ( |
Rajagopalan et al. ( |
0.799 | NA | 0.67 | ( |
Einarson et al. ( |
0.89 | 0.659 | 0.49 | ( |
Einarson et al. ( |
0.89 | 0.659 | 0.49 | ( |
Barnes et al. ( |
0.696 | NA | NA | AMICUS trial |
Einarson et al. ( |
0.7/0.65 |
0.485/0.469 |
0.27 | ( |
Einarson et al. ( |
0.890/0.840/0.795/0.790 |
0.690/0.665/0.643/0.640 |
0.49 | ( |
Wiwat et al. ( |
0.69 | NA | 0.58 | ( |
Nuhoho et al. ( |
0.89 | 0.659 | 0.49 | ( |
Aigbogun et al. ( |
0.88 | 0.74 | 0.53 | ( |
Németh et al. ( |
Not reported | Not reported | Not reported | ( |
Zhao et al. ( |
0.88 | 0.74 | 0.53 | ( |
Abdall-Razak et al. ( |
0.799 | 0.67 | 0.67 | ( |
Dutina et al. ( |
0.919 | 0.604 | 0.604 | ( |
Arteaga et al. ( |
0.916/0.865 |
−0.358 | −0.358 | ( |
Yi et al. ( |
0.92 | 0.74 |
0.60 | ( |
Lin et al. ( |
0.88/0.75/0.75 |
0.74/0.63/0.63 |
0.53/0.53/0.42 |
( |
Jin et al. ( |
0.80 | 0.67 | NA | ( |
Summary of the utilities of adverse events or complications.
Einarson et al. ( |
NA | ||||
Lachaine et al. ( |
−0.074 | −0.031 | −0.06/−0.05 |
Hypertension: −0.02 CHD: −0.07/−0.06 |
( |
Park et al. ( |
−0.256 | NA | −0.151 | Hyperprolactinemia: −0.089 CHD: −0.151 | ( |
Dilla et al. ( |
−0.054 | −0.003 | NA | Sexual disfunction: −0.066 | From SOHO data |
Druais et al. ( |
−0.197 | −0.094 | −0.15 | ( |
|
Lin et al. ( |
0.72 |
0.77 |
0.77 |
myocardial infarction: 0.74 |
( |
Rajagopalan et al. ( |
89% |
96% |
−0.15 | NA | ( |
Einarson et al. ( |
NA | ||||
Einarson et al. ( |
NA | ||||
Barnes et al. ( |
NA | NA | NA | Adverse events-0.006 | AMICUS trial |
Einarson et al. ( |
NA | NA | NA | NA | NA |
Einarson et al. ( |
NA | NA | NA | NA | NA |
Wiwat et al. ( |
0.62 |
0.66 |
0.66 |
Hyperprolactinemia: 0.62 | ( |
Nuhoho et al. ( |
NA | ||||
Aigbogun et al. ( |
−0.099 | −0.036 | NA | Akathisia: −0.09 pathoglycemia: −0.067 dyslipidemia: −0.099 sedation: −0.084 | ( |
Németh et al. ( |
Not reported | Not reported | Not reported | Not reported | ( |
Zhao et al. ( |
88.8% |
95.9% |
NA | NA | ( |
Abdall-Razak et al. ( |
88.8% |
95.9% |
0.76 |
Amputation: −0.109, non-fatal myocardial infarction: −0.129, non-fatal stroke: −0.181, heart failure: −0.108, ischemic heart disease: −0.132 | ( |
Dutina et al. ( |
−0.256 | NA | NA | Metabolic syndrome: −0.132 | ( |
Arteaga et al. ( |
−0.256 | −0.089 | −0.151 | Prolactin-related syndrome: −0.089 | ( |
Yi et al. ( |
0.72 |
0.83 |
NA | increased blood glucose level: 0.77 |
( |
Lin et al. ( |
88.8% |
95.9% |
88.8% |
hyperlipidemia/hyperprolactinemia: 88.8% |
( |
Jin et al. ( |
−0.07 | −0.03 | −0.09 | NA | ( |
It was noteworthy that the utilities for the same disease states varied greatly. Possible reasons included varieties in the classification of the researched states, selection of the population, and different methods or instruments applied among referenced quality of life studies.
Definitions or classifications of the health states were usually developed based on a literature review (
Layperson, patients with schizophrenia and caregivers are common responders in research on health-related quality of life. The differentiation of responders could induce heterogeneity among the results. For example, Briggs et al. (
Due to the specialty of mental disease as well as the choice of population such as laypersons, caregivers, or psychiatrists, the majority of the methods generated utilities using the standard gambling or time trade off approaches. The EuroQol-5 dimensions (EQ-5D) questionnaire is preferred by The National Institute for Health and Care Excellence (NICE). However, the sensitivity of the EQ-5D index to capture both social and psychological well-being for patients with schizophrenia is still controversial (
The included studies compared the cost-effectiveness of commonly used SGAs (including long-acting injections). However, due to economic, political, cultural diversities among the different regions, the results of one economic evaluation may not be applicable beyond the defined setting (
The widely used QHES and CHEERS lists economic evaluation checklists were applied for a quantitative and qualitative review.
Assessed with the QHES list, scores ranged from 60 to 93 for 25 studies, where 21 (84.0%) studies scored between 75 and 93, and 4 (16.0%) studies scored between 60 and 74, indicating the relatively high quality of the majority studies. As summarized in
Quality assessment of economic evaluations with the QHES checklist.
Einarson et al. ( |
√ | × | × | √ | √ | √ | √ | × | √ | √ | √ | √ | √ | √ | √ | √ | 81 |
Lachaine et al. ( |
√ | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | × | √ | √ | √ | 93 |
Park et al. ( |
√ | √ | √ | √ | √ | √ | √ | √ | √ | √ | × | √ | √ | √ | √ | √ | 93 |
Dilla et al. ( |
√ | √ | √ | √ | √ | √ | × | √ | √ | √ | × | √ | √ | √ | √ | × | 85 |
Anh et al. ( |
√ | √ | × | √ | √ | √ | √ | √ | √ | √ | √ | √ | × | × | √ | × | 76 |
Lubinga et al. ( |
√ | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | × | × | √ | √ | 87 |
Druais et al. ( |
√ | √ | √ | √ | √ | √ | √ | × | √ | √ | √ | √ | √ | √ | √ | √ | 93 |
Lin et al. ( |
√ | √ | √ | √ | √ | √ | √ | √ | × | √ | √ | √ | × | √ | √ | √ | 85 |
Rajagopalan et al. ( |
√ | √ | × | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | × | √ | √ | 86 |
Einarson et al. ( |
√ | √ | √ | √ | √ | √ | × | × | × | √ | √ | √ | × | √ | √ | √ | 73 |
Einarson et al. ( |
√ | √ | √ | √ | √ | √ | × | × | √ | √ | √ | √ | × | √ | √ | √ | 81 |
Barnes et al. ( |
√ | √ | √ | √ | √ | √ | √ | × | √ | √ | √ | √ | × | √ | √ | √ | 86 |
Einarson et al. ( |
√ | √ | √ | √ | √ | √ | √ | × | √ | √ | √ | √ | × | √ | √ | √ | 86 |
Einarson et al. ( |
√ | √ | √ | √ | √ | √ | √ | × | × | √ | √ | √ | × | √ | √ | √ | 78 |
Wiwat et al. ( |
√ | √ | √ | √ | √ | √ | √ | √ | × | √ | √ | √ | √ | × | √ | √ | 86 |
Nuhoho et al. ( |
√ | √ | × | √ | √ | √ | √ | × | × | √ | √ | √ | √ | √ | √ | √ | 77 |
Aigbogun et al. ( |
√ | √ | √ | √ | √ | √ | √ | × | √ | √ | √ | √ | × | √ | √ | √ | 86 |
Németh et al. ( |
√ | × | √ | √ | √ | √ | × | × | √ | × | √ | × | √ | √ | √ | √ | 70 |
Zhao et al. ( |
√ | √ | × | √ | √ | √ | √ | × | √ | √ | × | √ | √ | √ | √ | √ | 78 |
Abdall-Razak et al. ( |
√ | √ | × | √ | × | √ | × | × | √ | √ | √ | √ | × | √ | √ | √ | 64 |
Dutina et al. ( |
√ | √ | × | √ | √ | √ | × | √ | × | √ | √ | √ | × | × | √ | √ | 60 |
Arteaga et al. ( |
√ | × | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | 96 |
Yi et al. ( |
√ | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | × | × | √ | √ | 87 |
Lin et al. ( |
√ | × | √ | √ | √ | √ | × | × | √ | √ | √ | √ | √ | × | √ | √ | 78 |
Jin et al. ( |
√ | × | × | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | × | √ | √ | 82 |
Qualified studies | 25/25 | 20/25 | 17/25 | 25/25 | 24/25 | 25/25 | 18/25 | 12/25 | 19/25 | 23/25 | 22/25 | 24/25 | 12/25 | 17/25 | 25/25 | 23/25 |
Results of the assessment with the QHES list.
The quality reports evaluated with the CHEERS checklists were showed in
Quality assessment report of economic evaluations with the CHEERS checklist.
Einarson et al. ( |
× | √ | √ | × | √ | × | √ | × | √ | √ | √ | × | √ | √ | √ | × | √ | √ | √ | √ | √ | √ | √ | √ | 18/24 |
Lachaine et al. ( |
√ | √ | √ | √ | √ | √ | √ | √ | √ | √ | × | × | √ | √ | √ | × | √ | √ | √ | √ | √ | √ | √ | √ | 21/24 |
Park et al. ( |
√ | √ | √ | × | √ | √ | √ | √ | √ | √ | √ | × | √ | √ | × | √ | √ | √ | √ | √ | √ | √ | √ | √ | 21/24 |
Dilla et al. ( |
√ | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | × | √ | √ | √ | × | √ | √ | √ | √ | √ | √ | × | √ | 21/24 |
Anh et al. ( |
× | × | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | × | √ | √ | √ | × | √ | √ | × | × | 18/24 |
Lubinga et al. ( |
× | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | × | √ | √ | × | √ | √ | √ | √ | √ | √ | √ | √ | √ | 21/24 |
Druais et al. ( |
√ | √ | √ | √ | √ | √ | √ | × | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | 23/24 |
Lin et al. ( |
√ | × | √ | √ | √ | √ | √ | √ | × | √ | √ | × | × | √ | √ | × | √ | √ | √ | √ | √ | √ | √ | √ | 19/24 |
Rajagopalan et al. ( |
√ | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | × | √ | √ | √ | √ | √ | √ | 23/24 |
Einarson et al. ( |
√ | √ | √ | √ | √ | √ | √ | × | √ | √ | √ | × | × | √ | √ | × | √ | × | √ | × | × | √ | √ | × | 16/24 |
Einarson et al. ( |
√ | √ | √ | √ | √ | √ | √ | × | √ | √ | √ | × | √ | √ | √ | × | √ | × | √ | √ | × | √ | √ | √ | 19/24 |
Barnes et al. ( |
√ | √ | √ | × | √ | √ | √ | × | √ | √ | √ | √ | × | √ | × | × | × | × | √ | × | × | √ | √ | √ | 15/24 |
Einarson et al. ( |
√ | √ | √ | √ | √ | √ | √ | × | × | √ | √ | √ | √ | √ | √ | × | √ | √ | √ | √ | √ | √ | √ | √ | 21/24 |
Einarson et al. ( |
√ | √ | √ | × | √ | √ | √ | × | √ | √ | √ | √ | × | √ | × | × | √ | × | √ | √ | √ | √ | √ | √ | 18/24 |
Wiwat et al. ( |
√ | √ | √ | √ | √ | × | √ | √ | √ | √ | × | × | √ | × | √ | √ | √ | √ | √ | √ | √ | √ | √ | × | 19/24 |
Nuhoho et al. ( |
√ | √ | √ | × | √ | √ | √ | × | √ | √ | × | × | √ | × | √ | × | √ | × | √ | √ | √ | √ | √ | √ | 17/24 |
Aigbogun et al. ( |
√ | √ | √ | √ | √ | √ | √ | × | √ | √ | √ | √ | √ | √ | × | √ | √ | √ | √ | √ | √ | √ | √ | √ | 22/24 |
Németh et al. ( |
√ | √ | × | √ | × | √ | √ | × | √ | √ | × | √ | √ | √ | × | √ | √ | √ | √ | √ | √ | √ | √ | √ | 19/24 |
Zhao et al. ( |
√ | √ | √ | × | √ | √ | √ | × | √ | √ | × | √ | √ | √ | √ | √ | √ | × | √ | √ | √ | √ | √ | √ | 20/24 |
Abdall-Razak et al. ( |
√ | √ | √ | × | √ | √ | √ | × | √ | √ | × | × | √ | √ | × | × | × | × | √ | × | √ | √ | √ | √ | 15/24 |
Dutina et al. ( |
√ | √ | √ | × | √ | √ | √ | √ | √ | √ | × | × | × | √ | √ | × | × | × | √ | √ | × | √ | √ | √ | 16/24 |
Arteaga et al. ( |
√ | √ | × | √ | × | √ | √ | × | √ | √ | √ | × | √ | × | √ | √ | √ | √ | √ | √ | √ | × | √ | √ | 18/24 |
Yi et al. ( |
√ | √ | √ | × | × | √ | √ | √ | × | √ | × | × | √ | × | √ | × | √ | √ | √ | √ | √ | √ | √ | √ | 17/24 |
Lin et al. ( |
√ | √ | × | √ | × | √ | √ | × | √ | √ | × | × | √ | × | √ | √ | × | √ | √ | √ | √ | × | √ | √ | 17/24 |
Jin et al. ( |
√ | √ | √ | × | × | √ | √ | √ | √ | √ | √ | × | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | √ | 22/24 |
Qualified studies | 22/25 | 23/25 | 23/25 | 15/25 | 20/25 | 23/25 | 25/25 | 12/25 | 22/25 | 25/25 | 16/21 | 9/25 | 20/25 | 20/25 | 18/25 | 11/25 | 21/25 | 16/25 | 25/25 | 21/25 | 20/25 | 24/25 | 23/25 | 22/25 |
Results of the assessment with CHEERS checklist.
Unlike the CHEERS checklist acting as a recommendation of the report format, the QHES list was developed to appraise the quality of economic evaluation (
To compare the quality among the included studies, model types, regions, and time horizons were used as the indicators to classify the studies. The average QHES score of Markov model studies was 84.46 (13 studies) which was higher than that of decision tree model studies (77.14, 7 studies). The average score of the studies applying microsimulation was 80.75. The average numbers of the items consistent with the CHEERS recommendations were 19.23, 17.71, and 20 for Markov model studies, decision tree model studies, and microsimulation model studies. As a result, the Markov model and microsimulation model rather than the decision tree model are more appropriate model types for the study of schizophrenia. The quality of studies among different regions was also slightly different. The average QHES scores of the studies from North America, Asia, and European countries were 90.67, 81.00, and 80.07, respectively. It should be noticed that certain discrepancies exist among the scores of the studies from European countries where the maximum and minimum scores were 96 and 64. The numbers of the items consistent with the CHEERS recommendations (21.3, 18.14, and 18.86, respectively) were similar. However, even though the studies from North America seemed to be of higher quality, the numbers of the studies from the three regions differed a lot (three studies from North America, 7 studies from Asia, and 14 studies from European countries) and this might introduce bias when assessing the qualities.
To analyze the quality differences of studies with different time horizons, we classified the studies into two categories: short-term studies, i.e., the time horizon was 1 year or less, and long-term studies, i.e., the time horizon was longer than 1 year. The averaged QHES scores of the short-term studies (78.2, 10 studies) were lower than that of the longer-term studies (84.3, 15 studies). However, the number of the items consistent with the CHEERS recommendations are similar (18.3 for short-term studies vs. 19.53 for long-term studies). As mentioned above, the QHES list was developed to appraise the quality of economic evaluation while the CHEERS checklist was developed to recommend the report format. It can be inferred from the description of the time horizon from the two lists that the QHES list (Did the analytic horizon allow time for all relevant and important outcomes?) was more subjective and focused on the relationship between the item assessed and model outcomes. While the CHEERS checklist [State the time horizon(s) over which costs and consequences are being evaluated and say why appropriate] highlighted the fact that the statement of relevant aspect was provided and with no requirement to judge the appropriateness for the evaluation. Thus, attention should be paid when choosing checklists and interpreting the results.
With the increasing number of publications on health economic studies in recent years, systematic reviews in this filed have caught the attention of decision makers as useful tools to generate evidence (
Despite challenges in generating results of economic evaluations via systematic reviews, it was suggested that systematic reviews should focus on methods of model development, sources of both efficacy and utility data, and resources used for specific diseases. It might be valuable for researchers and decision makers to identify the differences among studies (
Among the included studies, most compared the cost-effectiveness of SGAs or long-acting injections. Fifteen studies considered the treatment sequences in model development, which could enlighten the future model development.
Despite the burden and productivity loss due to schizophrenia, few studies chose the societal perspective covering the indirect costs. Due to the early onset of schizophrenia, the average age of these patients is younger than that of patient with other chronic diseases. It would be necessary to include indirect costs in the evaluation. A retrospective study in the US concluded that indirect and non-health care costs were strong contributors and could be more than 70% of the total burden (
Description of the treatment sequences from most of the included studies could improve the model design and reflect the clinical prescription especially for chronic diseases or patients with high rate of therapy change. However, there remains challenges considering treatment sequences in economic evaluation. Medication treatment may vary among individuals due to the genotypes, metabolism, comorbidities, adverse events and so on (
According to the summary of the models applied, the Markov model was the most frequently used and treatment sequences, relapse, remission, and adverse events were the important health state elements in model development. The time horizons varied from 5 years to lifetime for the Markov models. While for the decision tree models, a 1 year time horizon was preferred. Due to the uncertainty in the time frame of treatment, it is recommended the impact of the time horizon be explored in a sensitivity analysis. Adverse events such as EPS, weight gain, and diabetes should be considered in the models for schizophrenia since these factors have a recognized influence on the treatment effect. Also, consideration of different types of AEs should be properly defined to estimate the impact on health outcomes and costs in the economic evaluation.
Based on the criteria studies, compliance and persistence were not clearly classified, thus definition is recommended for economic evaluations since it might determine the choice of appropriate data source. When integrating compliance or persistence, data are required on both health outcomes and costs for patients who are non-compliant or discontinued treatment (
Utility values, derived from the literature may contribute to the heterogeneity among results when they are applied to the same health states in different studies. Differences in the classifications of health states, survey responders, elicitation methods, and regions were the main factors influencing the utility values. Thus, it is recommended that researchers choose proper sources based on the factors above, as well as the publication year or update of methods.
Certain limitations to study quality have been identified, such as the description of appropriate time horizon selection, discount rate, statement, and justification of the choice of model type, assumptions and limitations to the evaluations. For the reporting quality of the studies, time horizon, preference-based outcome measurement, and assumptions were the major missing parts. To improve both the quality of the study and the quality of the report, it is suggested that researchers conduct the evaluation and generate the manuscript under the respective guidance and checklists.
Though there exist studies reviewed the economic evaluation of treatment for schizophrenia (
There remain some limitations of this review. First, the review only included studies published after 2014 and does not represent the economic methods used in earlier years. Studies published recently may be more valuable for analysis, considering the relatively high quality, most recent treatment options and updated clinical evidence. Second, this review only included model-based economic evaluations. Even though trial-based economic evaluations for schizophrenia are also important evidence, this study aims to generate summaries and suggestions for model methodology rather than synthesizing economic evidence. In addition, trial-based studies may not provide long-term clinical outcomes and source consumption, especially for chronic diseases.
Based on the results of this review, it is suggested that future research focus on methods to integrate compliance or persistence data for chronic diseases. Due to the diverse utilities cited in the models, characteristics of study groups and measuring approach of preference-based health outcomes from the health-related quality of life research should be explained to provide appropriate options for the studies. Publications of economic evaluations should be designed and reported according to applicable gelines and checklists to improve study quality and provide both scientific and valuable evidence for decision makers. Future research could pay more attention to the economic evaluation of long-acting injection antipsychotics.
The original contributions presented in the study are included in the article/
LW and HL contributed to the study design, analysis, and writing. FS, XG, and HX contributed to the review work. JL contributed to the manuscript revise. All authors contributed to the article and approved the submitted version.
This study received funding from Sumitomo Pharma (Suzhou) Co., Ltd.
This study received funding from Sumitomo Pharma (Suzhou) Co., Ltd. The funder had the following involvement with the study: decision to published. JL was employed by company Sumitomo Pharma. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
The Supplementary Material for this article can be found online at: