The Global Burden of Disease (GBD) study, which was established by the Institute of Health Metrics and Evaluation (IHME), measures the burden of diseases and injuries in over 200 countries and territories. Although schizophrenia is a relatively common mental problem, its burden has not been quantified across all global regions. Therefore, this study presents an assessment of the burden of schizophrenia from 1990 to 2019 for all countries in MENA. There are 21 countries in MENA, which are: Afghanistan, Algeria, Bahrain, Egypt, Iran, Iraq, Jordan, Kuwait, Lebanon, Libya, Morocco, Oman, Palestine, Qatar, Saudi Arabia, Sudan, the Syrian Arab Republic, Tunisia, Turkey, the United Arab Emirates and Yemen. A full description of the methodology utilised by IHME to model the burden of disease has been previously described7,16,18. The GBD 2019 estimates, which cover the period 1990–2019, are available at the following links: http://ghdx.healthdata.org/gbd-results-tool and https://vizhub.healthdata.org/gbd-compare/.
Case definition and data sources
Schizophrenia is a serious mental disorder which is characterised by a large number of symptoms, including: delusions, hallucinations, diminished interest, flat affect, thought disorders, and emotional withdrawal. The GBD disease modelling process only included data from studies that diagnosed schizophrenia using either the Diagnostic and Statistical Manual of Mental Disorders (DSM) criteria (DSM-IV-TR: 295.10-295.30, 295.60, 295.90) or the International Classification of Diseases (ICD) criteria (ICD 10: F20). The diagnostic criteria encompass the following key elements: (1) Presence of at least two of the following symptoms, each enduring for a substantial part of a one-month period (a shorter duration if effectively treated): (i) Delusions, (ii) Hallucinations, (iii) Disorganised speech (e.g., frequent incoherence or derailment), (iv) Markedly disorganised or catatonic behavior, (v) Negative symptoms (i.e., affective flattening, alogia, or avolition); (2) Dysfunction at work and socially; (3) Persistence of the disorder’s signs and symptoms for a duration of six months or more; (4) Exclusions included substance abuse, schizoaffective and mood disorders, and/or general medical conditions, as well as any connection to pervasive developmental disorders7.
IHME conducted a systematic review for schizophrenia, which encompassed searching the scientific literature (i.e., PsycInfo, Embase, and PubMed), examining the grey literature, and consultation with an expert. As part of the GBD project, the electronic databases are searched biennially for mental disorders, including schizophrenia. The last systematic review for schizophrenia was carried out in GBD 2017, with the next review being due in GBD 2020. However, consulting the expert and searching the grey literature produced new data sources in GBD 20197.
The inclusion criteria applied were as follows: (1) published after 1980; (2) cases were defined using DSM or ICD criteria; (3) inclusion of sufficient methodological details and sample characteristics for assessing study quality; and (4) samples that represented the general population. Specifically excluded were samples from inpatients or pharmacological treatments, case studies, veterans, or refugee cases. There were no constraints placed on the publication language. The data sources utilised to model the schizophrenia burden are accessible at this website: https://ghdx.healthdata.org/gbd-2019/data-input-sources7.
Data processing and disease model
When necessary, the data extraction process involved three different age and sex splitting procedures: (1) The available estimates were divided into specific five-year age groups by sex. For example, in studies which reported the prevalence in broad age ranges separately for males and females (e.g., 15–65 year old men and women individually), and in cases where studies had smaller age groups without sex separation (e.g., prevalence among 15 to 29 year olds, then in 30 to 70 year olds, for both sexes combined), the sex ratios reported and uncertainty ranges were used to divide the age specific estimates by sex. (2) Meta-Regression with Bayesian priors, Regularisation, and Trimming (MR-BRT) was used to split the remaining data. This method involved matching sex-specific estimates for each parameter, according to location, age, and year. MR-BRT regression was then employed to model the pooled sex ratios, along with their associated uncertainty bounds. These pooled sex ratios were then utilised to split the estimates in the dataset. The prevalence ratio between males and females was 1.17 (95% uncertainty interval (UI) 0.60–1.75). 3. For prevalence estimates covering age categories spanning 25 years or more, the age pattern estimated by DisMod-MR 2.1 was used to split the data into five-year age groups. It’s important to note that the DisMod-MR model used for estimating the age pattern did not contain any previously age split data7.
IHME utilised DisMod MR 2.1, using the standard GBD 2019 decomposition structure, to estimate the data related to schizophrenia. At each stage of the decomposition process, IHME compared the new model with the best model from GBD 2017 and the best model from the previous stage. If substantial differences were observed between models, these variances were thoroughly explored and elucidated. In cases where it was deemed necessary, adjustments were implemented to the dataset or the model priors. When outliers were identified, they were included or excluded based upon a re-examination of their quality and methodology.
Initially, all epidemiological parameters were integrated into the modelling process. It was believed, based on the literature on schizophrenia and discussion with the expert that no cases of schizophrenia occurred before the age of 10 or after the age of 80. Furthermore, the remission rate was restricted to a maximum of 0.04, in line with the data in the dataset. In areas lacking available data, prevalence estimates were informed by location-level covariates. Only one location-level covariate, lag distributed income (LDI), was utilised to model the prevalence of schizophrenia.
Compilation of results
The two sequelae (acute and residual) of schizophrenia, along with their corresponding disability weights (DWs), can be found in Table S1. To calculate the years lived with disability (YLDs), the prevalence estimates for each sequela were multiplied by their respective DWs. The YLDs and DALYs were the same, since there was no mortality due to schizophrenia. All estimates were standardised using the GBD standard population. 95% uncertainty intervals (UIs) were included with all estimates and were generated by producing 1000 iterations at each stage of the estimation process. The final estimates represented the mean values over the 1000 iterations, and the 95% UIs were indicated as the 25th and 975th values among the numerically ordered iterations.
Smoothing Spline models19 was employed to investigate the relationship the socio-demographic index (SDI) has with the burden of schizophrenia. The SDI is a composite model that contains per capita income, mean number of years attending school (aged 15 and above), and the fertility rate in women aged 25 or less. The SDI ranges from 0 to 1, representing the spectrum from the lowest to the highest development level7. The estimates for the point prevalence and annual incidence were obtained from the GBD website (http://ghdx.healthdata.org/gbd-results-tool) and all visual representations were created with R software (Version 3.5.2).
Ethics approval and consent to participate
The present study was approved by Ethics Committee of Shahid Beheshti University of Medical Sciences, Tehran, Iran (IR.SBMU.RETECH.REC.1401.387).