Impact statement
Our umbrella review provides the most comprehensive estimates for the prevalence of mental disorders and intentional self-harm in South Asia (SA) and highlights that large proportions of the population in the region (both general-adult and specific vulnerable groups) are affected by these adverse health conditions. Evidence is critically lacking beyond common mental disorders on several conditions including schizophrenia and psychotic disorders, behavioural syndromes, personality disorders and intellectual disabilities. Although limited by heterogeneity and methodological quality of included studies, our review findings show an urgent need for countries in SA to formulate and implement clinical and policy measures for the prevention and early treatment of mental disorders and intentional self-harm. The pooled prevalence estimated for depression and anxiety in the general-adult population could serve as a reference for policymakers to take necessary action for curbing the growing burden of mental disorders in SA.
Introduction
Mental disorders are recognised to be increasing globally, and contribute to a growing health, social and economic burden (World Health Organization, 2021). From 1990 to 2019, they have gone from the 13th to the 7th leading cause of disease burden in the world, with the number of disability-adjusted life-years (DALYs) due to mental disorders increasing from 80.8 million to 125.3 million; they also remain the second largest contributor to years lived with disability (GBD Mental Disorders Collaborators, 2022). Intentional self-harm accounts for a further 34.1 million DALYs, with their burden being greatest in low- and middle-income countries (Knipe et al., Reference Knipe, John, Padmanathan, Eyles, Dekel, Higgins, Bantjes, Dandona, Macleod-Hall and McGuinness2022). In South Asia (SA) (Afghanistan, Bangladesh, Bhutan, India, Maldives, Nepal, Pakistan and Sri Lanka) (The World Bank, 2019), rapid demographic and lifestyle changes are said to be associated with an exponential rise in mental and substance-use disorders, which health systems and services are unable to adequately meet (World Health Organization, 2016b; Ambekar et al., Reference Ambekar, Agrawal, Rao, Mishra, Khandelwal and Chadda2019). This has resulted in a considerable mental health treatment gap, with more than 75% of people affected in many countries not having access to the treatment they need (World Health Organization, 2016b; Gautham et al., Reference Gautham, Gururaj, Varghese, Benegal, Rao, Kokane, Chavan, Dalal, Ram and Pathak2020). Further, mental disorders have not been a policy priority among countries in the region, and their epidemiological and psychosocial burdens have been under-researched (Shidhaye et al., Reference Shidhaye, Sikander, Jordans, De Silva and Chatterjee2015). To address these issues and to improve the knowledge base for better planning and decision-making, an overall evaluation of the prevalence of mental disorders and intentional self-harm among countries in SA is needed.
Hossain et al. (Reference Hossain, Purohit, Sultana, Ma, McKyer and Ahmed2020) published an umbrella review, stating its advantages over a review of primary studies for understanding the population-level burden of mental disorders within the SA region. However, their inclusion criteria were limited to reviews solely conducted in SA (i.e., they excluded broader reviews, even if those reviews included some South Asian studies). We considered that expanding the scope of our umbrella review to identify all systematically conducted reviews, so long as they included evidence from at least one country in SA (whilst limiting our synthesis to South Asian studies), would provide a more complete picture of the prevalence of mental disorders and intentional self-harm in the region. In addition, a meta-analysis to provide an updated pooled estimate for the prevalence of mental disorders in the general adult population in SA would complement the overview provided by the umbrella review.
Methods
The review was registered with PROSPERO (CRD42021282957) (McDaid et al., Reference McDaid, Wright, Nasir, Siddiqi, Vidyasagaran, Huque, Muliyala and Nair2021). We followed the Joanna Briggs Institute (JBI) method for conducting the review (Aromataris et al., Reference Aromataris, Fernandez, Godfrey, Holly, Khalil and Tungpunkom2015) and the PRISMA guidelines for reporting (Page et al., Reference Page, McKenzie, Bossuyt, Boutron, Hoffmann, Mulrow, Shamseer, Tetzlaff, Akl and Brennan2021; Supplementary Appendix 1).
Search strategy
We searched multiple electronic databases and research repositories, covering published and grey literature, on 29 September 2021 (Supplementary Appendix 2). Our searches included index terms, synonyms, and alternative phrases to cover mental disorders, South Asian countries, prevalence or epidemiology, and review types. We used the search strategies for ‘prevalence’ and ‘South Asia’ from Uphoff et al. (Reference Uphoff, Newbould, Walker, Ashraf, Chaturvedi, Kandasamy, Mazumdar, Meader, Naheed, Rana, Wright, Wright, Siddiqi, Churchill and Impact2019), and for ‘mental disorders’ from Mishu et al. (Reference Mishu, Uphoff, Aslam, Philip, Wright, Tirbhowan, Ajjan, Al Azdi, Stubbs and Churchill2021), adapting them to include all ICD-10 categories of mental disorders and intentional self-harm (World Health Organization, 2016a; Supplementary Appendix 3). Searches were developed by an information specialist (JW) and peer-reviewed by a second, using the PRESS checklist (McGowan et al., Reference McGowan, Sampson, Salzwedel, Cogo, Foerster and Lefebvre2016). There were no limits for language or publication date. We also screened reference lists and forward citations of included studies. In addition, PROSPERO records were checked for any relevant ongoing or completed reviews. Retrieved records were de-duplicated in EndNote semi-automatically, using specified guidance (AUHE Information Specialists, 2016) and uploaded to COVIDENCE (www.covidence.org) for further evaluation.
Inclusion criteria and study selection
We included systematic reviews (with or without meta-analyses) that searched two or more databases, and provided keyword and/or search strategies, as per the quality criteria of the AMSTAR2 checklist (Shea et al., Reference Shea, Reeves, Wells, Thuku, Hamel, Moran, Moher, Tugwell, Welch and Kristjansson2017). Reviews reporting the prevalence or incidence of mental disorders in one or more countries in the World Bank-defined SA region were eligible. This included reviews that had data from countries beyond SA, but where we could extract the SA data on their own. All populations and settings were eligible, except studies of international military forces based in SA. Reviews on any mental, behavioural, and neurodevelopmental disorders (ICD-10, F-codes), or on suicide and intentional self-harm (ICD-10, X60-X84 codes) were eligible (Supplementary Appendix 4). Two authors independently evaluated all records at title and abstract and full-text screening stages. Discrepancies in screening were addressed through discussion with a third author.
Data extraction and synthesis
A pre-piloted data extraction tool was uploaded to COVIDENCE. Two authors independently extracted data and performed quality appraisals for 10% of included reviews, with good agreement; discrepancies were identified and resolved through consensus. All remaining extractions were performed by a single author and checked by a second. Extraction items included objective and type of review, year of publication, name and timeframe of databases, originating countries of primary studies, sample size and characteristics, as well as reported prevalence or incidence of mental disorders. We used the AMSTAR2 tool for evaluating the methodological quality of included reviews (Shea et al., Reference Shea, Reeves, Wells, Thuku, Hamel, Moran, Moher, Tugwell, Welch and Kristjansson2017).
Narrative synthesis was conducted according to the type of review (with or without meta-analysis) and mental disorders (ICD-10 categories), using tables and figures. For the reviews that went beyond SA, we only considered the pooled/range of estimates from the subgroup of studies that were relevant to the SA region. Next, we focused on the reviews with meta-analyses to summarise results for pooled prevalence of mental disorders in SA. Finally, to estimate prevalence for the common mental disorders (CMDs), depression and anxiety, we obtained data from primary studies in the included reviews. We limited this step to reviews with a pre-registered protocol (as a quality indicator), and those reporting on CMDs in the general adult population, given these conditions, which comprise the great majority of mental disorders, were the focus of the bulk of included reviews. Additional primary studies reporting CMD prevalence in SA were identified through forward citation screening of included reviews, to capture more recent studies.
Data extraction from primary studies was again performed by a single author and checked by a second on the following items: country, state or province of the study population, sample characteristics and sample size, and prevalence or incidence for each mental disorder. For quality assessment, we used the JBI Critical Appraisal Checklist for prevalence studies (Munn et al., Reference Munn, Moola, Riitano and Lisy2014), but did not exclude ones at high risk of bias from further analysis. We created a ‘summary of findings’ table for primary studies and carried out meta-analyses using Stata (2007), Version 17.0 to produce a pooled estimate of prevalence for depression and anxiety among the general population in SA. Heterogeneity was assessed using I 2 statistics, and subgroup analyses based on country and outcome ascertainment tools were conducted to explore the sources. Evidence of publication bias was assessed using funnel plots and Egger’s test.
Results
Our searches yielded 1,048 records, with 770 remaining after deduplication (Figure 1). Following title and abstract screening, 548 records were excluded, and all but one of the remaining 222 papers were obtained. Full-text screening resulted in the exclusion of 94 records (see Figure 1 and Supplementary Appendix 5 for details). 124 reviews (127 records) met our eligibility criteria and were included in the narrative synthesis. Three reports covering one review were merged (Barua et al., Reference Barua, Ghosh, Kar and Basilio2010; Barua et al., Reference Barua, Ghosh, Kar and Basilio2011a, Reference Barua, Ghosh, Kar and Basilio2011b); for another review, we merged and extracted data from both the original and updated reports (Oram et al., Reference Oram, Stockl, Busza, Howard and Zimmerman2012; Ottisova et al., Reference Ottisova, Hemmings, Howard, Zimmerman and Oram2016).
Figure 1. PRISMA flow chart for included reviews.
For the meta-analyses of primary studies, we found only two reviews with pre-registered protocols, which reported on the prevalence of CMDs in the general adult population (Naveed et al., Reference Naveed, Waqas, Chaudhary, Kumar, Abbas, Amin, Jamil and Saleem2020; Zuberi et al., Reference Zuberi, Waqas, Sadiq, Hossain, Rahman, Saeed and Fuhr2021). These provided 22 primary studies. Three additional studies were identified through forward citation screening of included reviews, resulting in 25 distinct primary studies for our meta-analyses (14 depression-only, three anxiety-only, and eight both) (see Supplementary Appendix 6 for flow chart of primary studies).
Characteristics of reviews included in the review of reviews
Table 1 provides the summary characteristics of all included reviews. Twenty-five reviews had conducted meta-analyses providing pooled estimates for mental disorders in SA. A further 99 reviews did not provide pooled estimates, either because no meta-analysis was conducted (n = 61), no pooled values limited to SA countries were presented (n = 37), or pooled prevalence was not estimated (n = 1).
Table 1. Summary characteristics of included reviews
Abbreviations: ADHD, attention deficit hyperactivity disorder; ASD, autism spectrum disorder; AUD, alcohol use disorder; CI, confidence interval; CMD, common mental disorder; CUD, cannabis use disorder; ED, eating disorder; GAD, generalised anxiety disorder; HCW, health care worker; HIVP, HIV positive; IDP, internally displaced population; MDD, major depressive disorder; NCD, non-communicable disease; NR, not reported; OCD, obsessive compulsive disorder; PTSD, post-traumatic stress disorder; PWA, people with AIDS; SA, South Asia; SUD, substance use disorder.
a Reporting discrepancy noted.
The earliest review was published in 2004 (Mirza and Jenkins, Reference Mirza and Jenkins2004), with the majority (n = 116) published after 2010. The number of databases searched ranged from two to fourteen, and the majority of reviews presented evidence from India (n = 90). The number of South Asian primary studies ranged from one to 149, and sample size ranged from 109 to 863,657 (not reported in 15 reviews). Reviews covered diverse populations, with participants recruited from a range of clinical and community-based settings. Only ten reviews were rated as ‘high’ quality, while most (61) were rated as ‘critically low’ (details in Supplementary Appendix 7).
A total of 65 reviews presented the prevalence of mood (affective) disorders including depressive and bipolar disorders, followed by 45 on anxiety disorders, and 10 on a combination of mood and anxiety disorders, grouped together as CMDs. A further nine reviews reported the prevalence of substance use disorders (SUDs), while others covered a range of other mental disorders: seven on behavioural and emotional disorders with usual onset in childhood and adolescence, including conduct disorder and attention-deficit hyperactivity disorder (ADHD), four on pervasive developmental disorders including autism spectrum disorder (ASD), three each on dementia, schizophrenia and psychotic disorder, personality disorder, and intellectual disabilities, and two on eating disorders. Of these, only one included a meta-analysis providing a pooled estimate of ASD prevalence among children in India. We also found six reviews that reported the prevalence of ‘any mental disorder’ and 23 that reported on suicide and intentional self-harm. Many identified reviews covered mental disorders in specific population subgroups including older people, perinatal women, students, healthcare workers (HCWs), and persons with comorbidities. Twenty-two reviews focused on the impact of COVID-19 on the psychosocial health of various population groups (Supplementary Appendix 8).
Summary of pooled prevalence from systematic reviews with meta-analysis
We now focus on the 25 reviews with meta-analyses on the prevalence of various mental disorders in SA. Eleven were exclusively of studies conducted in India, four in Pakistan, and the remaining 10 covered multiple countries in the region. The population comprised all adults (including perinatal women and older people, n = 1), general adults (n = 3), adults with specific conditions such as alcohol use disorders (AUD) or non-communicable disease (NCD) (n = 6), women (n = 6), older people (n = 4), children and adolescents (n = 2), HCWs (n = 2), and university students (n = 1). In general, these reviews reported high pooled prevalence of mental disorders among both general-adult (up to 33.0% for depression) (Naveed et al., Reference Naveed, Waqas, Chaudhary, Kumar, Abbas, Amin, Jamil and Saleem2020), and specific population subgroups (up to 55.0% for depression among stroke survivors) (Patra et al., Reference Patra, Nitin, Devi, Surya, Lewis and Kamalakannan2021). The pooled prevalence of suicidal behaviours among adults was 6.4% (95% CI = 3.1–12.4) (Naveed et al., Reference Naveed, Waqas, Chaudhary, Kumar, Abbas, Amin, Jamil and Saleem2020), and among children and adolescents was 17.1% (5.0–35.4) (Ganesan et al., Reference Ganesan, Radhakrishnan and Rajamanickam2020), whereas the pooled prevalence of any mental disorder among victims of suicide was 90.4% (71.8–97.2) (Cho et al., Reference Cho, Na, Cho, Im and Kang2016).
We identified 19 pooled estimates for mood disorders (17 studies), followed by six for anxiety disorders (5 studies), and three for CMDs (Figure 2). The pooled prevalence (range) for depressive disorders in the general population was 10.0% (4.0–25.0) to 33.0% (7.0–75.0). Estimates were generally higher for specific population subgroups, including older people (21.9% (11.6–31.1) to 42.0% (38.0–46.0)), perinatal women (22.0% (19.0–25.0) to 37.0% (30.0–44.0)), peri-menopausal women (42.5% (28.7–57.5)), university students (42.7% (34.8–50.9), HCWs = 31.7% (18.7–48.3) to 34.1% (28.9–39.4)), and adults with comorbidities (18.0% (5.6–45.1) to 55.0% (43.0–65.0)). Similarly, pooled prevalence (range) for anxiety disorders in the general population was 4.0% (0.0–27.0) to 25.8% (19.4–33.5); for adults with comorbidities, it was 2.4% (0.9–5.8) to 29.0% (22.0–36.0), and among all adults and HCWs during the COVID-19 pandemic it was 41.3% (34.7–48.1). Based on reviews covering multiple countries in SA, the pooled prevalence of CMDs in the general adult population alone was estimated to be 19.8% (10.3–34.7) (Steel et al., Reference Steel, Marnane, Iranpour, Chey, Jackson, Patel and Silove2014), whereas it was higher (28.4% (13.9–49.3)) among adult populations that included older people and perinatal women (Naveed et al., Reference Naveed, Waqas, Chaudhary, Kumar, Abbas, Amin, Jamil and Saleem2020). One review from India reported a pooled value of 21.9% (17.5–26.3) for prevalence of CMDs among antenatal women (Kalra et al., Reference Kalra, Tran, Romero, Chandra and Fisher2021).
Figure 2. Pooled estimates of CMDs, depression and anxiety from meta-analytic reviews.
Note: Two studies (2021 Atif and 2021 Zuberi) provided two relevant estimates each for different population groups; the vertical dotted line denotes a pooled prevalence of 14.0% (drawn to correspond with the Figure 3 forest plot).
We also found one general-adult, population-based estimate for the pooled prevalence of any mental disorder, covering all countries in SA except Maldives and presented as a rate per 1000 (95% CI): 122.0 (8.0–252.0) (Ranjan and Asthana, Reference Ranjan and Asthana2017). In addition, we found two meta-analyses reporting SUDs prevalence of 0.0% (0.0–1.0) to 32.0% (6.0–78.0) (Naveed et al., Reference Naveed, Waqas, Chaudhary, Kumar, Abbas, Amin, Jamil and Saleem2020; Zuberi et al., Reference Zuberi, Waqas, Sadiq, Hossain, Rahman, Saeed and Fuhr2021), one on dementia prevalence (2.0% (2.0–3.0)) (Choudhary et al., Reference Choudhary, Ranjan and Asthana2021) and one on ASD prevalence (0.1% (0.0–0.2)) (Chauhan et al., Reference Chauhan, Sahu, Jaiswal, Kumar, Agarwal, Kaur, Singh and Singh2019). These are not presented in Figure 2.
Pooled prevalence of depression and anxiety in the general adult population from primary studies
We identified 25 primary studies reporting the prevalence of CMDs in the general adult population (Table 2): 16 from India, three from Nepal, one each from Pakistan, Sri Lanka, and Afghanistan, and three large, population-based studies that covered multiple countries in SA. Study quality overall was high. Meta-analyses found a pooled prevalence of 16.0% (95% CI = 11.0–22.0, I 2 = 99.9%) for depression, 12.0% (5.0–21.0, I 2 = 99.9%) for anxiety, and 14.0% (10.0–19.0, I 2 = 99.9%) for depression and anxiety combined (Figure 3).
Table 2. Summary characteristics of primary studies included in meta-analyses
Figure 3. Forest plot of primary studies on the prevalence of depression and anxiety in South Asia.
The pooled prevalence (95% CI) of depression varied notably by country, from 5.0% (4.0–6.0) in Afghanistan, 5.0% (5.0–6.0) in Sri Lanka and 6.0% (5.0–6.0) in Pakistan to 16.0% (10.0–24.0) in India, 25.0% (6.0–52.0) in Nepal, and 25.0% (24.0–25.0) in Bangladesh. Similarly, the pooled prevalence of anxiety varied between 3.0% (2.0–3.0) in Afghanistan, 4.0% (3.0–4.0) in Pakistan and 6.0% (2.0–14.0) in India, to 19.0% (16.0–23.0) in Nepal, 21.0% (20.0–22.0) in Bangladesh, and 65.0% (64.0–66.0) in Sri Lanka. The pooled values for both conditions also varied markedly according to whether (and which) diagnostic or screening tools were used to ascertain the presence of depression and/or anxiety. For depression, the pooled prevalence from estimates based on diagnostic tools (e.g., Composite International Diagnostic Interview (CIDI) and Mini International Neuro-psychiatric Interview (MINI)) was 5.0% (3.0–6.0), whereas it was 27.0% (13.0–44.0) based on screening measures. Similarly, the pooled prevalence for anxiety from estimates based on diagnostic tools was 1.0% (0.0–3.0), whereas it was 26.0% (19.0–34.0) based on screening measures. Funnel plot asymmetry was observed and Egger’s test for meta-analysis of depression was statistically significant indicating publication bias. Forest plots for subgroup analyses and funnel plots can be found in Supplementary Appendix 9.
Discussion
This umbrella review has identified many reviews covering a range of mental disorders in SA, with the majority focusing on the prevalence of CMDs among different population groups. Our findings suggest a high prevalence of these conditions in the region, with greater burden among specific population groups, including perinatal women, older people, people with chronic physical illnesses, refugees, and other vulnerable groups. More than 20 reviews were identified on the prevalence of CMDs during COVID-19 and suggest a high burden of mental disorders among healthcare workers, teachers, and students in SA during the pandemic. In common with Hossain et al. (Reference Hossain, Purohit, Sultana, Ma, McKyer and Ahmed2020) we found that most studies were from India, while evidence from Afghanistan, Bhutan, and Maldives was particularly limited. The advantages and novelty of this review are in providing a more complete and updated picture of the prevalence of mental disorders in the region. But despite the broader inclusion criteria and the updated searches, we found no reviews with pooled estimates of prevalence for many conditions, including severe mental disorders such as schizophrenia and psychotic disorders, behavioural syndromes, personality disorders, or intellectual disabilities. Reviews without meta-analyses for these conditions were also limited. Further, most reviews scored ‘low’ or ‘critically low’ on quality assessment, with very few assessed as providing an accurate and comprehensive summary of available studies on the topic.
Our meta-analysis of primary studies provides pooled estimates for the prevalence of depression and anxiety in the general adult population in SA. We had originally planned to use a 2001 cut-off for the primary studies, set to correspond with the World Health Report on Mental Health (World Health Organization, 2001), but revised this to post-2009 studies, to keep in line with the search period followed by one of the reviews from which we harvested primary studies (Naveed et al., Reference Naveed, Waqas, Chaudhary, Kumar, Abbas, Amin, Jamil and Saleem2020). Similarly, whilst our protocol mentioned meta-analyses for all mental disorders, we limited this step to reviews on CMDs, given these conditions were the focus of the bulk of identified reviews. Both the reviews from which we harvested primary studies had also previously reported pooled estimates for these conditions in SA, but one included studies in all adult populations, including higher-risk perinatal women and older people (Naveed et al., Reference Naveed, Waqas, Chaudhary, Kumar, Abbas, Amin, Jamil and Saleem2020), while the other was limited to studies in Afghanistan and Pakistan (Zuberi et al., Reference Zuberi, Waqas, Sadiq, Hossain, Rahman, Saeed and Fuhr2021). The inclusion of populations with greater disease burden in the former likely explained its higher prevalence compared to our estimates for both depression (26.4% vs. 16.0%) and anxiety (25.8% vs. 12.0%). With regard to the latter review, while reported country-specific pooled estimates are comparable to ours for Pakistan, its estimates are considerably higher for Afghanistan for both conditions (33.0% vs. 5.0% for depression and 25.0% vs. 3.0% for anxiety). This difference may be explained by the inclusion of two studies reporting high prevalence, which were excluded in our review on the basis of publication date (Scholte et al., Reference Scholte, Olff, Ventevogel, de Vries, Jansveld, Cardozo and Crawford2004; Mufti et al., Reference Mufti, Naeem, Ayub, Saifi, Haroon and Kingdon2005). On the other hand, our searches identified results from a recent national survey on depression and anxiety disorders in Afghanistan, which we included in our meta-analyses (Kovess-Masfety et al., Reference Kovess-Masfety, Keyes, Karam, Sabawoon and Sarwari2021), while the addition of the excluded primary studies from Afghanistan and Pakistan (Nisar et al., Reference Nisar, Billoo and Gadit2004) does not considerably change the region-specific pooled estimates for depression or anxiety (Supplementary Appendix 9).
In addition to mental disorders, our umbrella review included 23 reviews on suicide and intentional self-harm, including one review with meta-analysis among adults in SA, which reported a 6.4% pooled prevalence of suicidal behaviours (Naveed et al., Reference Naveed, Waqas, Chaudhary, Kumar, Abbas, Amin, Jamil and Saleem2020). Other reviews found adult suicide rates ranging from 0.43 to 331.0 per 100,000 population, which varied greatly across countries in the region, and in some cases are likely to be gross underestimations of actual rates (Jordans et al., Reference Jordans, Kaufman, Brenman, Adhikari, Luitel, Tol and Komproe2014). An even higher prevalence of suicidal behaviours was found among specific population groups, including perinatal women (Fuhr et al., Reference Fuhr, Calvert, Ronsmans, Chandra, Sikander, De Silva and Patel2014; Amiri and Behnezhad, Reference Amiri and Behnezhad2021), people with HIV/AIDS (Collins et al., Reference Collins, Holman, Freeman and Patel2006; Das and Leibowitz, Reference Das and Leibowitz2011), female sex workers (Somrongthong et al., Reference Somrongthong, Panza, Alam, Sultana, Jiayu, Huda, Zaman and Wahed2019) and tribal populations (Devarapalli et al., Reference Devarapalli, Kallakuri, Salam and Maulik2020). Three reviews on suicidal behaviours among children and adolescents were identified, all from India (Aggarwal and Berk, Reference Aggarwal and Berk2015; Aggarwal et al., Reference Aggarwal, Patton, Reavley, Sreenivasan and Berk2017; Ganesan et al., Reference Ganesan, Radhakrishnan and Rajamanickam2020). Further, we found three reviews among suicide and self-harm populations, which reported a high prevalence of mental disorders, particularly depressive disorders (Cho et al., Reference Cho, Na, Cho, Im and Kang2016; Ahmed et al., Reference Ahmed, Hossain, Aftab, Soron, Alam, Chowdhury and Uddin2017; Knipe et al., Reference Knipe, Williams, Hannam-Swain, Upton, Brown, Bandara, Chang and Kapur2019).
Our searches identified three reports based on the Global Burden of Disease studies, which we excluded on the basis of study design (Baxter et al., Reference Baxter, Charlson, Cheng, Shidhaye, Ferrari and Whiteford2016; Liu et al., Reference Liu, Lunde, Jia and Qin2020; Sagar et al., Reference Sagar, Dandona, Gururaj, Dhaliwal, Singh, Ferrari, Dua, Ganguli, Varghese and Chakma2020), and because analyses were either limited to just India or estimated annual percentage change in the burden of depression across the region, not directly comparable to the results of our analyses. Similarly, three reviews (Reddy and Chandrashekar, Reference Reddy and Chandrashekar1998; Ganguli, Reference Ganguli2000; Arora and Aeri, Reference Arora and Aeri2019) included in the Hossain et al. (Reference Hossain, Purohit, Sultana, Ma, McKyer and Ahmed2020) umbrella review did not meet our eligibility criteria on study design, but those topics were covered in other included reviews. Our review includes all other reviews they included, but by going beyond geographically limited reviews and summarising the evidence from multi-country reviews that included at least one South Asian country, we have identified many more reviews, providing a more complete picture of the evidence regarding the prevalence of mental disorders in the region. Diverse terms were used to describe the reviews that were included (systematic, scoping, narrative, etc.), but we screened for studies that met our criteria to be considered systematic reviews, and thereby ensured consistency in our inclusions (Haddaway et al., Reference Haddaway, Lotfi and Mbuagbaw2022). In addition, our meta-analyses of primary studies on depression and anxiety provide important new information on the prevalence of these conditions among the general adult population in the region.
Some key limitations of the research should be acknowledged. First, our approach for identifying primary studies was through harvesting studies from included reviews and forward citation screening, rather than a systematic search and screening of databases. This may have missed studies and introduced a selection bias, but our pre-defined strategy on having a registered protocol likely protected against this. In addition, there are possibilities of publication bias, which our funnel plots suggested were likely. Our meta-analyses also found high heterogeneity, which could be explained to some extent by differences between countries and assessment tool used, demonstrated by subgroup analyses. The finding that studies using screening tools report higher prevalence than those using diagnostic interviews has been previously reported, which may have overestimated the prevalence of mental disorders (Zuberi et al., Reference Zuberi, Waqas, Sadiq, Hossain, Rahman, Saeed and Fuhr2021). In the methodological literature on clinical trials, developing and adopting ‘core outcome sets’ has been advocated to address the heterogeneity that precludes meaningful synthesis of evidence across studies. Core outcomes sets mandate the inclusion of key outcomes to be measured in all trials of interventions for particular conditions and may also define the tools to be used to measure them (Chiarotto et al., Reference Chiarotto, Ostelo, Turk, Buchbinder and Boers2017). A similar agreed set of defined measures for observational studies of various mental ill health conditions may be a way forward for better synthesis.
Next, although the majority of primary studies received overall high ratings, few were nationally representative surveys of the general adult population. Nonetheless, there were primary studies from most countries in the region, apart from Bhutan and Maldives. In contrast to the quality of primary studies in our meta-analyses, our narrative synthesis is largely based on reviews that scored ‘low’ or ‘critically low’. We therefore limited our presentation of prevalence estimates solely to the meta-analytical reviews, while the overall narrative summary provides a broader mapping of identified evidence from all reviews by type of review and mental disorder. Finally, there is the possibility that our umbrella review may have missed some relevant reviews on mental disorders in SA, but we searched a large number of (including region-specific) databases and reviewed the literature as comprehensively as possible.
Overall, the findings of our research show a high burden of mental disorders among the general-adult population in SA, with even higher prevalence among specific population subgroups. These findings are also supported by reviews published since our searches were carried out (Manna et al., Reference Manna, Tripathy, Sah, Padhi, Kaur, Nowrouzi-Kia and Chattu2022; Palfreyman and Gazeley, Reference Palfreyman and Gazeley2022; Al-Mamun et al., Reference Al-Mamun, Hasan, Quadros, Kaggwa, Mubarak, Sikder, Hossain, Muhit, Moonajilin and Gozal2023; Javan Biparva et al., Reference Javan Biparva, Raoofi, Rafiei, Masoumi, Doustmehraban, Bagheribayati, Vaziri Shahrebabak, Noorani Mejareh, Khani and Abdollahi2023). Our results highlight an urgent need for countries in SA to formulate and implement both clinical and policy measures for the prevention and early treatment of mental disorders and intentional self-harm. The mapping of evidence according to the type of review and mental disorder (Supplementary Appendix 8) shows that population-level prevalence estimates are generally lacking beyond CMDs, including for schizophrenia and psychotic disorders, behavioural syndromes, personality disorders, and intellectual disabilities. These identified gaps are supported by other recent reviews (Russell et al., Reference Russell, Nagaraj, Vengadavaradan, Russell, Mammen, Shankar, Viswanathan, Earnest, Chikkala and Rebekah2022; Bastien et al., Reference Bastien, Ding, Gonzalez-Valderrama, Valmaggia, Kirkbride and Jongsma2023), and should be a focus of future research, along with the strengthening of epidemiological surveillance systems to better capture morbidity, mortality, and economic burden of all mental disorders and intentional self-harm in the region.
Open peer review
To view the open peer review materials for this article, please visit http://doi.org/10.1017/gmh.2023.72.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/gmh.2023.72.
Data availability statement
The details of data searches and extractions from the included studies are provided in the Supplementary Material. The review protocol, including the analysis plan, can be accessed freely from the PROSPERO database, using the registration number mentioned. We do not have any additional data to share.
Author contribution
Conceptualization: D.M.D., N.S.; Conducting literature searches: J.W.; Data analysis: A.L.V., M.R.F.; Data extraction and quality appraisal: A.L.V., D.M.D., M.N., K.P.M., S.T., M.R.F., R.H., J.W., S.B., N.S.; Data interpretation: A.L.V., D.M.D., M.R.F., N.S.; Design of literature search strategy: J.W.; Manuscript writing: A.L.V., M.R.F., S.B.; Revision of manuscript and editing: A.L.V., D.M.D., M.N., K.P.M., S.T., M.R.F., R.H., J.W., S.B., N.S.; Study design: D.M.D., N.S. All authors had full access to all the data in the study and had final responsibility for the decision to submit for publication.
Financial support
This research is funded by the National Institute for Health Research (NIHR) – Grant 17/63/130 NIHR Global Health Research Group: Improving Outcomes in Mental and Physical Multi-morbidity and Developing Research Capacity (IMPACT) in South Asia at the University of York, using UK aid from the UK Government to support global health research. The views expressed in this publication are those of the author(s) and not necessarily those of the NIHR or the UK government.
Competing interest
The authors declare none.
Open access
Comments
To,
The Editorial Team
Global Mental Health
17 April 2023
Subject: A systematic umbrella review and meta-analysis on ‘Prevalence of mental disorders in South Asia’ for your consideration
Dear Editorial Team,
We hereby submit our article titled “Prevalence of mental disorders in South Asia: A systematic review of reviews” to Global Mental Health for your consideration. To our knowledge, this study provides the most comprehensive and up-to-date estimates for the prevalence of mental disorders and intentional self-harm in South Asia.
Through extensive searching and rigorous screening processes, we have identified a large number of systematic reviews reporting the prevalence of mental disorders and/or intentional self-harm in South Asian countries. We have followed the recommended best practices for conducting and reporting our research, and the manuscript is drafted following GMH’s formatting guidelines. In addition to narratively synthesising findings from the reviews themselves, we identified primary studies of common mental disorders among adults from a subset of selected reviews and their forward citations and pooled them using random-effect models. Our findings highlight the need for South Asian countries to formulate and implement both clinical and policy measures for the prevention and early treatment of these conditions within the region.
This research was funded by the UK National Institute for Health Research (NIHR) – Grant 17/63/130 NIHR Global Health Research Group: Improving Outcomes in Mental and Physical Multi-morbidity and Developing Research Capacity (IMPACT) in South Asia at the University of York.
Please let us know if any further information is required.
With many thanks for your consideration,
Yours sincerely,
Aishwarya (on behalf of the review team)