Global Health Action ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/zgha20 Macro-level mental health system indicators and cross-national suicide rates Johnny Andoh-Arthur & Samuel Adjorlolo To cite this article: Johnny Andoh-Arthur & Samuel Adjorlolo (2021) Macro-level mental health system indicators and cross-national suicide rates, Global Health Action, 14:1, 1839999, DOI: 10.1080/16549716.2020.1839999 To link to this article: https://doi.org/10.1080/16549716.2020.1839999 © 2021 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group. Published online: 19 Jan 2021. Submit your article to this journal Article views: 2098 View related articles View Crossmark data Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=zgha20 GLOBAL HEALTH ACTION 2021, VOL. 14, 1839999 https://doi.org/10.1080/16549716.2020.1839999 ORIGINAL ARTICLE Macro-level mental health system indicators and cross-national suicide rates Johnny Andoh-Arthur c and Samuel Adjorlolo a,b aDepartment of Mental Health, School of Nursing and Midwifery, College of Health Sciences, University of Ghana, Accra, Ghana; bResearch and Grant Institute of Ghana, Accra, Ghana; cDepartment of Psychology, School of Social Sciences, University of Ghana, Accra, Ghana ABSTRACT ARTICLE HISTORY Background: The relationship between macro-level mental health system indicators and Received 24 August 2020 population suicide rates is an area of contention in the literature, necessitating an analysis Accepted 15 October 2020 of current cross-national data to document any new trend in the relationship. Objective: This study investigated whether mental health system indicators are associated RESPONSIBLE EDITOR Jennifer Stewart Williams, with national suicide rates. Umeå University, Sweden Method: Using an ecological study design and multivariate non-parametric robust regression models, data on suicide rates and mental health system indicators of 191 countries retrieved KEYWORDS from WHOs 2017 Mental Health Atlas were compared. Suicide; mental health; Results: Findings revealed that the average suicide mortality rate was significantly higher in mental health governance; high- income countries, relative to low-income countries. High-income countries are signifi- mental health professionals; cantly more likely to have high number of mental health professionals, mental health policies suicide prevention and legislation, independent mental health authority and suicide prevention programs. These mental health system indicators demonstrated significant and positive association with suicide, suggesting that countries scoring high on these factors have higher odds of being categorized as high suicide risk countries. Conclusion: The findings have several implications for policy and practice, including the need to make existing mental health systems very responsive to suicide prevention. Background suicides in other contexts [13,14]. Indeed a systematic review from the Low-and Middle-Income countries Suicide accounts for about 800,000 deaths annually at (LMICs) has pointed to crucial role of non-psychiatric an estimated age standardized rate of 11.2 per 100,000 factors such as poverty in suicides [15]. It is possible [1]. Across the globe, suicide is the 15th most common that where mental health service is organized almost cause of death [1,2]. Suicide also reportedly accounted exclusively to focus on biomedical factors, it is most for 1.4% of premature deaths worldwide in 2015 [3]. likely that less priority will be given to non-psychiatric A recent study revealed that the age standardized mor- factors in suicides [16]. Impliedly, cross-national varia- tality rate for suicide has decreased by 32.7% worldwide tions in the role of psychiatric disorders and for that between 1990 and 2016, however, total number of matter suicide rates generally could be partly attributed deaths from suicide within the same period increased to variations in patterns, frequency, and meanings of by 6.7% globally over the 27-year-study period [4]. suicide, as well as variations in the organization and Suicide is notably a complex, multifactorial phenom- provision of mental health service [6,16–18]. Among enon that shows significant variations, particularly with the factors that are likely to influence suicide rates respect to the causal factors and mechanisms, as well as across regions and countries are national income, men- prevention and management strategies [5]. tal health governance system, and resources for mental Mental health problems such as mood, substance health, the so-called macro-level indictors of suicide use, psychotic or personality disorders have been because they require political commitment [19–21]. shown to play critical roles in suicide trajectory A number of ecological, country-level studies have [6–11]. Often repeated are some psychological autopsy examined the presumed link between suicide rate and studies from High-Income Countries that for instance, mental health system indicators, such as availability suggest that 90%–95% of individuals who die of suicide of mental health services [5], professional density had a diagnosable psychiatric disorders at the time of [2,22,23], mental health spending [24], mental health committing suicide [7,8,12]. Some authors have chal- legislations and policies [21] and antidepressant sales lenged this statistic by pointing either to methodological [23]. A study conducted in Finland found that the flaws that usually produces the statistic or to evidence prominence of outpatient services was associated suggesting no diagnosable psychiatric disorders in some CONTACT Samuel Adjorlolo sadjorlolo@ug.edu.gh Department of Mental Health, School of Nursing and Midwifery, College of Health Sciences, University of Ghana, P. O. Box LG 43, Legon, Accra, Ghana © 2021 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. 2 S. ADJORLOLO AND J. ANDOH-ARTHUR with lower suicide rate [17] whereas others have made by member countries in achieving the WHO’s reported high suicide rates in countries with better Comprehensive Mental Health Action Plan by 2020. mental health services such as higher numbers of Among the core requirements of the Action Plan are psychiatric beds and availability of training in mental for member countries to, for instance, strengthen health for primary care professionals [5,18]. With mental health governance, render comprehensive and respect to professional density and suicide rate, find- integrated mental health care, and institute mental ings are mixed, with some studies reporting negative health promotion and prevention strategies [29]. The relationship [2,23,25,26], positive relationship Atlas, therefore, provides information on the state of [5,18,22], or no statistically significant relationship mental health across nations by unearthing the exis- [23]. An Austrian study, for instance, found that tence or otherwise of mental health-specific indices a high number of mental health professionals was pertinent to the Comprehensive Mental Health associated with lower rates of suicide, adjusting for Action Plan. These include policies, plans and laws per capita alcohol consumption and unemployment for mental health, human and financial resources; the rates [23]. In contrast, cross-national data from 191 type of mental health facilities, and availability of countries revealed that the number of psychiatrists mental health promotion and prevention programs. was significantly positively associated with population suicide rates [5]. In yet another study from Austria, no statistically significant relationship was observed Data collection procedure between psychiatrist density and suicide rate [23]. As noted previously, the Atlas is produced by the Analyses of cross-national data showed an increase WHO with data from member countries. For the in suicide rates and existence of mental health initia- 2017 Atlas, data were collected using a structured tives such as mental health policy, mental health questionnaire from 177 out of the 194 WHO member program [18,21]. More importantly, while suicide states, representing 97% of the world’s population. rates reportedly differ across countries [6], the direc- The questionnaires were developed in consultation tion of the difference is an area of contention. More with member states and internationals experts in the specifically, contrary to the finding that a vast major- area of mental health care measurement. In each ity, namely 78%, of suicides occur in the Low and member state, a focal person was identified to com- Middle-Income Countries [3], others have found plete the questionnaire by extracting data from multi- a high prevalence of suicide in high-income coun- ple sources, including from local team of experts, tries, relative to LMICs [27,28]. In their analysis of institutions such as Mental Health Authority/ cross-national data [18], found that countries with Commission and psychiatric and non-psychiatric high mental health expenditure tend to experience hospitals located across different parts of the country. high suicide rate. Once the focal person has completed the data- The inconsistent findings provide impetus for more gathering process, the questionnaires were sent back studies to examine suicide and mental health systems. to the WHO for processing and use. Where applic- There is the pressing need to subject current cross- able, the focal persons were re-contacted for further national data to empirical analysis to document any information and clarifications relating to the ques- new trend in the relationship between mental health tionnaire to ensure data quality. Detailed procedure system and suicide rates across geographical regions for data collection has been described elsewhere [30]. [5,22]. Consequently, this study is designed to first and foremost investigate the distribution of suicide rates across countries based on their national income, Measures/instrument and second adopt an ecological framework1 to inves- tigate the associations between suicide rate and mental Dependent variable health systems indicators (e.g. the number of profes- Suicide. The suicide mortality rate for the various sionals, availability of mental health policies).. countries were derived from the World Health Statistics data visualization dashboard, which is avail- able at http://apps.who.int/gho/data/ Methods Independent variables Data source The 2017 Atlas contain different macro-level data on The data were extracted from the mental health Atlas mental health service delivery. Some variables (e.g. 2017 published by the WHO for member states. The government’s total expenditure on mental health, Atlas is designed to monitor and evaluate progress number of mental health hospitals, and number of 1...........and second adopt an ecological framework to investigate the associations between suicide rate and mental health systems indicators (e.g., the number of professionals, availability of mental health policies). Ecological framework views problems i.e. suicide as the outcome of interaction among many factors at multiple levels – the individual, the relationship, the community, and the societal. GLOBAL HEALTH ACTION 3 psychiatric units in a general hospital) were missing variables under the mental health system governance data at a scale that does not allow meaningful com- domain were also recoded as 1 and 0, respectively, for parison. For example, in some instances, data are statistical purposes. available for high-income countries but not of other Next, χ2 analysis was conducted to determine the income groups. Consequently, a decision was reached association between the independent variables and to exclude some data from the analysis. Variables suicides. Phi and Cramer’s V coefficients were used with complete data across the various income groups to estimate the strength of the association, with coef- were considered in this study, as stated below; ficient values between .10 and .29, .30 to .49, and values at .50 and above interpreted as small, moderate Income group. The 2017 Atlas categorized WHO and large effects. This was followed by a binomial member countries into four different income groups logistic regression to determine the predictors of sui- using the gross national income (GNI) developed by cide categorization using odd ratios (OR) and the World Bank in 2016. These are low income (GNI adjusted odd ratio (AOR), where applicable. Income per capita of US$ 1,025 or less), low-middle income group and the variables under mental health system (GNI per capita between US$ 1,026 and US$ 4,035), governance and resource for mental health domains upper-middle income (GNI per capita between US$ were entered into the regression model indepen- 4,036 and US$ 12,475), and high income (GNI per dently. Only the variables demonstrating significant capita of US$ 12,476 or more). associations with suicide categorization based on χ2 analysis were included in this analysis. Last, because Mental health system governance. The following national income is a major determinant of several variables were extracted to index mental health sys- indicators of mental health, including the propensity tem governance: (1) Stand-alone policy or plan for to increase the number of mental health professionals mental health; (2) plan or strategy for child and/or and enact mental health legislations [5], the effect of adolescent mental health; (3) Stand-alone law for income group was controlled for to obtain the ‘true mental health; (4) existence of a dedicated authority effect’ of mental health legislation and professionals or independent mental health body. Responses to on suicide categorization. All analyses were per- these variables were present (Yes) or absent (No). formed using SPSS version 24. Resources for mental health. The variables extracted under resources for mental health were: (1) total Results number of mental health workers per 100,000 popu- lation; (2) number of psychiatrists; (3) number of National income and suicide mental health nurses; and (4) clinical psychologists The average suicide mortality rate for 155 countries per 100, 000 population. Responses to these variables was 8.96 (Range: .50–31.90). Income group-based were continuous. analyses revealed that low income (n = 27), low- middle income (n = 42), upper-middle income Suicide prevention programs. Data were collected on (n = 46) and high-income countries (n = 40) recorded whether the countries have suicide prevention pro- 7.06 (range: 3.7–12.20), 7.61 (range: 1.90–22.40), 8.49 grams, with Yes or No as the response options. (range: 1.70–31.00) and 12.18 (.50–31.90) average sui- cide mortality rate, respectively. Income group and suicide rate are significantly correlated, χ2 (3) = 14.47, Data analytic strategy p < .01, Phi & Cramer’s V = .31. Countries recording Data were summarized using descriptive statistics, at least 20 suicide rate per 100, 000 include Belgium, notably frequencies, percentages, and bar chart. This Latvia, Ukraine, Suriname, South Korea, Guyana, was followed by inferential statistics using chi square Russian Federation, and Lithuania. In contrast, (χ2) and binomial logistic regression, with alpha level Antigua and Barbuda, Barbados, Grenada, Bahamas set at .05. The principal outcome variable, national and Syrian Arab Republic were among the countries suicide rates, and the resources for mental health that reported less than two suicide rate per 100, 000. variables did not follow Gaussian distribution. From Table 1, national income and suicide rate Consequently, we proceeded with the inferential sta- evinced statistically significant relationship, χ2 tistical analysis by dichotomizing these variables [31]. (3) = 14.47, p < .05. Further analyses showed that First, they were numerically equalized and scaled the odds of being designated as high suicide risk did along a common standard metric by converting not differ significantly between low income (the refer- them into z-scores. The standardized scores were ence category) and low-middle and upper-middle subsequently recoded such that scores below and income countries (p > .05). In contrast, high-income above the mean were designated as low (0) and high countries have higher odds of being classified as high (1), respectively. The Yes/No responses to the suicide risk (b = 1.67, OR = 5.31). 4 S. ADJORLOLO AND J. ANDOH-ARTHUR Table 1. Relationship between national income, mental income countries with the highest number of mental health system indicators and suicide. health professionals include Finland (i.e. 250.55), USA Suicide (i.e. 271.28), Brazil (i.e. 317.45), Costa Rica (i.e. 341.94), Low, n High, n Total, n P/ and Monaco (i.e. 405.41). The least number of mental Variables (%) (%) (%) χ2 C National Income 14.47** .31 health professionals was recorded in sub-Saharan Low income 20(21.5) 7(11.3) 27(17.4) Africa (SSA) countries such as Chad (i.e. 0.04), Low middle income 28(30.1) 14(22.6) 42(27.1) Guinea (i.e. 0.05), Central African Republic (0.15), Upper middle income 31(33.3) 15(24.2) 46(29.7) High income 14(15.1) 26(41.9) 40(25.8) Mali (i.e. 0.16), and Kenya (i.e. 0.19). Similar trend Total 93(60) 62(40) 155(100) was observed when the analysis was focused on specific Mental Health Professionals mental health professionals, namely psychiatrists, men- Total mental health 73(90.1) 28(58.3) 101 17.92*** .37 tal health nurses and clinical psychologists. professionals 8(9.9) 20(41.7) (78.3) Low 81(62.8) 48(37.2) 28(21.7) As can be seen in Table 2, the odds of being High 129(100) designated as high risk for suicide is significantly Total Psychiatrists 72(90) 21(42.9) 93(72.1) 33.57*** .51 higher among countries with a high number of men- Low 8(10) 28(57.1) 36(27.9) tal health professionals (b = 1.88, OR = 6.52)., psy- High 80(62) 49(38) 129(100) chiatrists (b = 2.49, OR = 12.00), mental health nurses Total Mental health nurses 62(88.6) 19(50) 81(75) 19.54*** .43 (b = 2.05, OR = 7.75) and clinical psychologists Low 8(11.4) 19(50) 27(25) (b = 1.69, OR = 5.42). When the effect of national High 70(64.8) 38(35.2) 108(100) Total income was statistically controlled for, all the vari- Clinical psychologists 65(94.2) 27(75) 92(87.6) 8.04** .28 ables but clinical psychologists, retained their statis- Low 4(5.8) 9(25) 13(12.4) High 69(65.7) 36(34.3) 105(100) tical significance (p. < .05). Total Mental Health Governance System Mental health system governance and suicide Mental policies or 8(8.6) 6(9.7) 14(9) .05 .02 plans 85(91.4) 56(90.3) 141(91) A total of 162 countries provided data on mental No 93(60) 62(40) 155(100) Yes health system governance. A large proportion of the Total countries (n = 147, 90.7%) have instituted mental Mental health law 35(37.6) 21(33.9) 56(36.1) .04 .04 No 58(62.4) 41(66.1) 99(63.9) health policies or plans. More than half of the coun- Yes 93(60) 62(40) 155(100) tries (n = 105, 64.8%) reported that they have stand- Total Child/adolescent 61(65.6) 27(43.5) 88(56.8) 7.37** .22 alone mental health laws, whereas 83 countries mental health strategy 32(34.4) 35(56.5) 67(43.2) (51.2%) have a dedicated authority or independent No 93(60) 62(40) 155(100) Yes mental health commission that oversees mental Total health activities in their respective countries. With Mental health 53(57) 22(35.5) 75(48.4) 6.89** .21 respect to child/adolescent mental health, approxi- authority/commission 40(43) 40(64.5) 80(51.6) No 93(60) 62(40) 155(100) mately 57% (n = 92) of the countries indicated they Yes do not have plan or strategy for child/adolescent Total Suicide prevention 10.57** .26 mental health. A statistically significant association strategy was observed between income group and existence No 75(80.6) 35(56.5) 110(71) Yes 18(19.4) 27(43.5) 45(29) of plan or strategy for child/adolescent mental health, Total 93(60) 62(40) 155(100) χ2 (3) = 17.23, p = .001, Phi & Cramer’s V = .33, P/C = Phi and Cramer’s coefficient. dedicated mental health authority, χ2 (3) = 11.71, ** = p < .01; *** = p < .001. p = .008, Phi & Cramer’s V = .27, and stand-alone mental health law, χ2 (3) = 15.36, p = .002, Phi & Mental health professionals and suicide Cramer’s V = .31, but not with mental health plans/ policies χ2 (3) = 1.33, p = .722, Phi & The average number of mental health professionals per Cramer’s V = .09. 100, 000 for 137 countries that supplied complete data Suicide rate correlated significantly with the avail- was 37.61. When analyzed against income group, it was ability of child/adolescent mental health strategies, χ2 observed that the average number of mental health (1) = 7.37, p < .01, Phi & Cramer’s V = .22, and the professionals differ across the income groups; low existence of independent mental health authority or income (i.e. 1.58), lower-middle income (i.e. 7.33), commission, χ2 (1) = 6.89, p < .01, Phi& Cramer’s upper-middle income (i.e. 43.61), and high income V = .21], but not with the availability of mental health (i.e. 90.12) countries. Chi square analysis revealed that plans/policies and mental health laws (p > .05). high-income countries are significantly likely to have Further analyses reveal that, the odds of suicide rate more mental health professionals per 100,000, relative is significantly higher in countries with child/adoles- to low middle and low-income countries, χ2 cent mental health strategies (b = .91, OR = 2.47) and (3) = 45.07, p < .001, Phi& Cramer’s V = .58. High- independent mental health authority (b = .88, GLOBAL HEALTH ACTION 5 Table 2. Logistic regression of suicide on income group and mental health indicators. 95% CI Odd ratio Variables b (Unadjusted) B 95% CI Adjusted Odd ratio Income Group Low versus low-middle income .36 1.43 (.49, 4.18) - - Low versus upper-middle income .32 1.38 (.48, 3.99) - - Low versus high income 1.67** 5.31 (1.81, 3.60) - - Mental Health Personal (per 100,000) Total mental health professionals 1.88*** 6.52 (2.56, 5.49) 1.50** 4.49 (1.47, 4.73) Psychiatrists 2.49*** 12.00 (4.76, 6.23) 2.71*** 15.04 (4.23, 7.41) Nurses 2.05*** 7.75 (2.93, 4.50) 1.31** 6.10 (1.93, 3.29) Psychologists 1.69** 5.42 (1.54, 3.10) 1.25 3.50 (.90, 3.69) Mental health professionals Child/adolescent strategy .91** 2.47 (1.28, 4.78) .68 1.98 (.97, 4.05) Authority/Commission .88** 2.41 (1.24, 4.67) .73* 2.07 (1.01, 4.21) Suicide prevention programs 1.17** 3.21(1.57, 6.60) 1.06** 2.87(1.31, 5.31) ** = p < .01; *** = p < .001. OR = 2.41). When the effect of national income was Suicide rates in low and middle income and controlled for, only the existence of independent high-income countries mental health authority significantly increased the odds of being labelled as high risk for suicide While suicide remains one of the global challenges (b = .73 AOR = 2.07). confronting nations, the study found that the average suicide mortality rate was significantly higher in high-income countries, relative to low-income coun- Suicide prevention programs and suicide tries. This finding, which is largely consistent with previous studies [18,27,28], contradicts the widely Of the 162 countries providing data on suicide, held view that LMICs tend to experience the greatest majority (n = 116, 71.6%) indicated they do not proportion of suicides [3]. The supposedly low pre- have stand-alone, government initiated national sui- valence of suicide in LMIC could be accounted for by cide prevention programs/plan. Income group corre- several factors, including the relatively high social lated significantly with the existence of suicide support, religious commitment and/or involvement, prevention strategies, χ2 (3) = 15.34, p < .01, Phi & and better family cohesion, which are potential pro- Cramer’s V = .31, with high-income group more tective factors against suicide [32,33]. likely to have national suicide prevention strategies. The foregoing notwithstanding, there is also the Indeed, when the analysis was disaggregated by possibility that real-suicide figures are obscured in income group, it was observed that only four low LMIC partly due to the fundamental problem of income (n = 27; i.e. Afghanistan, Mozambique, inaccurate or lack of data pertaining to suicide. Chad and Uganda), five low-middle income (n = 43; Indeed, suicide rates depend not only on the effi- e.g. Philippines, Vanuatu, Timor-Lesta, Bhutan and ciency of civil registration systems, which are gener- Nicaragua), 18 upper middle income (n = 49; e.g. ally poor in LMIC, but also on the reporting of deaths Iran, Panama, Malaysia, Ecuador and Turkey) and which is in turn heavily influenced by the social, 19 high-income countries (n = 43; e.g. Monaco, cultural and legal consequences of suicide [34]. Israel, Italy and Spain) have national suicide preven- Despite the ongoing campaign and advocacy to decri- tion plans/strategies. minalize suicide across countries, suicide continues to As shown in Table 1, availability of national sui- exist as a legal term that is often accompanied by cide prevention plans was significantly associated legally prescribed punitive measures in several sub- with suicide rate, χ2 (1) = 10.57, p < .01, Phi & Saharan African countries, including Ghana [35], Cramer’s V = .26. Further analysis revealed that Nigeria, Botswana, Gambia, Kenya, Malawi, countries with suicide prevention strategies are sig- Tanzania, Zambia and Ugandan [36]. In some nificantly more likely to have higher odds of being LMICs such as Ghana, suicide is viewed as a taboo categorized as high suicide risk country (b = 1.17, and unnatural death. Given this orientation, indivi- OR = 3.21), even after controlling for national duals expressing suicidal tendencies or persons income groupings (b = 1.06, AOR = 2.87). deceased by suicides as well as their families are socially sanctioned. This observation incentivize families and communities to conceal or Discussion misreport suicidal behaviors so as to protect and The study primarily investigated the macro-level fac- preserve the sanctity of the family name [37,38]. tors influencing suicide mortality rates across coun- Under-reporting of suicide is therefore highly preva- tries using data from the WHO’s 2017 Mental Health lent in LMICs [28,39] owing to the prevailing social, Atlas. cultural and legal proscriptions against suicides [5]. 6 S. ADJORLOLO AND J. ANDOH-ARTHUR In contrast, the reported high prevalence of suicide in predominantly in high-income countries implies that high-income countries could be due to an artifact of greater attention is paid to urgency-driven curative more efficient death registration systems and their medical solutions, thereby belittling the importance case finding effects [34]. Likewise, the implementa- of associated sociocultural and economic factors tion of suicide prevention programs and the asso- [5,41]. Indeed, the observation that suicide can ciated awareness and hypervilgilance could occur in the absence of psychological and mental culminate into case-finding effect where there is the health problems [42] implies that programs mainly general tendency to label deaths as suicides [21]. designed to promote and improve mental health in The foregoing has enormous implication on the general may have contributed little to reducing the use of population suicide rate as a proxy indicator of prevalence of suicide rates. Some major risk factors the effectiveness of a country’s mental health services. that have been found to have exhibited direct and That is, although LMIC reportedly have low suicide indirect relationship with suicide include unemploy- rates, it will be problematic to attribute this to better ment, alcohol consumption and substance misuse in mental health system in these countries, as discussed general, social inequalities, loss of social cohesion, previously. Indeed, LMICs are characterized by low and financial difficulties emanating from high indebt- number of mental health professionals, relative to edness and bankruptcy [43–45]. An earlier study, for high-income countries; a development that has been instance, found that the relationship between unem- attributed to low investment in training mental health ployment and suicide was statistically significant even professionals and the relocation of mental health after controlling for mental illness [46]. The fore- professionals to high-income countries for better going presupposes that appropriate macroeconomic conditions of service. For instance, on the latter and social welfare policies that ensure and promote point, an earlier study reported more Ghanaian psy- basic human rights, social security, gender equality chiatrists working in the USA than those working in and equitable development may contribute to redu- Ghana [40]. cing population suicide rates [5]. Perhaps, it is against this background that others have renewed calls for situating suicide research and prevention within Correlates of suicides rate social and cultural contexts [47]. The purpose is to The study also found that high-income countries are deepen the understanding of suicides in accordance significantly more likely to have child and adolescent with postulation of the stress-diathesis model sug- mental health strategies, dedicated mental health gesting that suicide risk is multi-factorial [48] and authority and suicide prevention programs, relative should therefore not be reduced to only psychiatric to LMIC. More importantly, given the robust link illness [6]. Furthermore, although specific suicide between mental health problems and tendency for prevention programs may exist particularly in high- suicide particularly in the high-income countries income countries, there is the possibility they have [9,11], it is plausible to reason that programs and not been implemented [49], or even if they are imple- initiatives designed to promote and restore mental mented, they could be riddled with implementation health functioning will contribute to a reduction in issues that could thwart or reduce their effectiveness. suicide rate. In contrast, however, the study found that macro-level mental health indicators examined in this study (e.g. number of mental health profes- Limitations sionals, dedicated mental health authority and The findings of the study should be reviewed in light of national suicide prevention programs) are associated the following limitations. First, all ecological studies with an increase in suicide rate, even after controlling have the potential limitation of ecological fallacy, for national income. This is largely consistent with which can occur as an association observed between previous studies [5,21]. Although the finding is some- the study variables on the aggregate level will not what counterintuitive and so could discourage further necessarily represent the association existing at the investment in mental health, on the other hand, it individual level. Causal associations cannot be assumed presents another opportunity for policy makers and since this study analyzed data collected using a cross- relevant stakeholders to reexamine the various sectional study design. The study could also be con- national mental health initiatives within the ambit founded by potential variables such as cultural and of suicide prevention [21]. More specifically, the religious differences between countries which were study highlighted several possibilities, including the not captured in the WHO’s 2017 Mental Health Atlas. view that the aforementioned national mental health There is also the possibility of under or over-reporting initiatives may not be designed with a focus on sui- of data which may occur for several reasons, including cide prevention, notwithstanding that they are an attempt to present a good picture of mental health intended to promote general mental health and well- situations. The application of advanced data mining being. Moreover, the over-medicalization of suicide, techniques may uncover patterns that might have GLOBAL HEALTH ACTION 7 been missed with the data analytic strategy employed in common risk factors contribute to suicides in some this study. Also, because the quality of training and people but not in others. focus of psychiatrists might differ across countries [22], the lack of data on these critical variables could affect the findings of the study. Last, the findings Acknowled reported in this study could have been different with We are grateful to Mr. Godfred Danso for extracting the data from all WHO member countries. data from the WHO Mental Health Atlas Notwithstanding the limitations, the findings reported in this study are largely consistent with previous ecolo- gical studies. Author contributions SA conceived the idea and extracted data; JAA drafted the manuscript, SA performed the analysis and wrote result, Conclusion JAA wrote the discussion and conclusion; SA and JAA This ecological study examined the association between proofread the manuscript. population suicide rates and key mental health indica- tors based on the most recent WHO Mental Health Disclosure statement Atlas which reported suicide rates from 191 countries. No conflict of interest was reported by the authors Findings generally revealed significant and positive relationship between macro-level mental health indica- tors and suicides. Although the findings appear some- Ethics and consent what counterintuitive in the light of the robust link The study used data generated by the WHO between suicide and mental illnesses in many coun- tries, nonetheless, they prompt action towards improv- ing suicide reporting and recording systems across the Funding information globe generally and mapping of locally relevant factors contributing to suicides as a guide for formulating and There is no external funding to declare implementing context sensitive prevention and inter- vention measures. . Paper context Noteworthy is the lower rate of suicides in indivi- dual countries within the LMICs, which is as revealing Despite the robust association between mental health status and risk for suicide at the individual level, ecological studies as it is intriguing. Suicide and suicide prevention within exploring mental health systems and suicide rates are not most LMIC countries are not national priorities [50]. only scarce but have produced inconsistent results. The study Apart from the social, cultural and legal barriers that found that (high income) countries with proper mental hinder accurate reporting and recording of suicides, it health system and governance are more likely to record is possible that the burden of disease attributable to high suicide rates. There is the need to make existing mental health systems very responsive to suicide prevention. communicable and infectious diseases in LMICs tended to sway concerted national health responses away from non-communicable and relatively ‘rare’ ORCID health problems such as suicides. Where modest atten- Johnny Andoh-Arthur http://orcid.org/0000-0002-7036- tion was given to the latter, it was largely focused on 1835 cardiovascular diseases, diabetes, cancer and chronic Samuel Adjorlolo http://orcid.org/0000-0001-9308-6031 pulmonary diseases [51,52]. However, given that some countries in the LMICs have or are undergoing ‘epidemiological transition’ whereby the relative References importance of infectious diseases is gradually becoming [1] WHO. Preventing suicide: global imperative. Geneva; less than that of chronic diseases [53], there is a need 2014. [cited 2020 Oct 8]. Available from: https://www. for national governments within the LMICs to commit who.int/mental_health/suicide-prevention/world_ more resources to non-infectious causes of mortality report_2014/en/ and morbidity such as suicides. Such efforts will be in [2] Kawaguchi H, Koike S. Association between the den-sity of physicians and suicide rates in Japan: nation- keeping with the target 3.4 of the Sustainable wide ecological study using a spatial Bayesian model. Development Goals, which is to reduce by one-third PLos One. 2016;11:e0148288. premature mortality from non-communicable diseases [3] Bachmann S. Epidemiology of suicide and the psy- by the year 2030 [54]. In the light of the above, we chiatric perspective. Int J Environ Res Public Health. suggest the need to increase qualitative and quantitative 2018;15:1425.[4] Naghavi M. Global, regional, and national burden of research within individual countries that will help dee- suicide mortality 1990 to 2016: systematic analysis for pen understanding of the meaning of suicide and iden- the Global Burden of Disease Study 2016. BMJ. tifiable risk factors for people and how and why 2019;364. DOI:10.1136/bmj.l94 8 S. ADJORLOLO AND J. ANDOH-ARTHUR [5] Rajkumar A, Brinda E, Duba A, et al. National suicide [23] Kapusta ND, Niederkrotenthaler T, Etzersdorfer E, rates and mental health system indicators: an ecologi- et al. Influence of psychotherapist density and anti- cal study of 191 countries. Int J Law Psychiatry. depressant sales on suicide rates. Acta Psychiatr 2013;36:339–342. Scand. 2009;119:236–242. [6] Brådvik L. Suicide risk and mental disorders. [24] Ross JM, Yakovlev PA, Carson F. Does state spending Int J Environ Res Public Health. 2018;15:2028. on mental health lower suicide rates? J Socio Econ. [7] Bertolote JM, Fleischmann A. Suicide and psychiatric 2012;41:408–417. diagnosis: a worldwide perspective. World Psychiatry. [25] Korosec Jagodic H, Rokavec T, Agius M, et al. 2002;1:181. Availability of mental health service providers and [8] Arsenault-Lapierre G, Kim C, Turecki G. Psychiatric suicide rates in Slovenia: a nationwide ecological diagnoses in 3275 suicides: a meta-analysis. BMC study. Croat Med J. 2013;54:444–452. Psychiatry. 2004;4:37. [26] Tondo L, Albert MJ, Baldessarini RJ. Suicide rates in [9] Harris EC, Barraclough B. Suicide as an outcome for relation to health care access in the USA: an ecological mental disorders. A meta-analysis. Br J Psychiatry. study. J Clin Psychiatry. 2006;67:517–523. 1997;170:205–228. [27] Khazaei S, Armanmehr V, Nematollahi S, et al. [10] Andoh-Arthur J, Hjelmeland H, Osafo J, et al. Suicide rate in relation to the Human Development Substance use and suicide among men in Ghana: Index and other health related factors: a global ecolo- a qualitative study. Curr Psychol. 2020;1–13. gical study from 91 countries. J Epidemiol Glob DOI:10.1007/s12144-020-00644-0 Health. 2017;7:131–134. [11] Furqan Z, Sinyor M, Schaffer A, et al. “I can’t crack [28] Blasco-Fontecilla H, Artieda-Urrutia P, de Leon J. the code”: what suicide notes teach us about experi- A proposal for using the ratio of attempted to com- ences with mental illness and mental health care. Can pleted suicides across several countries worldwide. J Psychiatry. 2019;64:98–106. Epidemiol Psychiatr Sci. 2018;1–5. DOI:10.1017/ [12] Cavanagh JT, Carson AJ, Sharpe M, et al. S2045796018000628 Psychological autopsy studies of suicide: a systematic [29] WHO. Mental health action plan 2013–2020. Geneva; review. Psychol Med. 2003;33:395–405. 2013. [Cited 2020 Oct 8]. Available from: https://apps. [13] Hjelmeland H, Dieserud G, Dyregrov K, et al. who. int / i r i s /b i t s t ream/handle/10665/89966/ Psychological autopsy studies as diagnostic tools: are 9789241506021_eng.pdf?sequence=1 they methodologically flawed? Death Stud. [30] WHO. Mental Health Atlas. Geneva; 2017. [Cited 2012;36:605–626. 2020 Oct 8]. Available from: https://apps.who.int/ [14] Shahtahmasebi S. Suicide prevention: politics or iris/bitstream/handle/10665/272735/9789241514019- conspiracy. Dyn Hum Behav (DHH). 2016;3. eng.pdf?ua=1 [15] Iemmi V, Bantjes J, Coast E, et al. Suicide and poverty [31] Lumley T, Diehr P, Scott Emerson A, et al. The in low-income and middle-income countries: Importance of the normality assumption in large pub- a systematic review. Lancet Psychiatry. lic health data sets. Annu Rev Public Health. 2016;3:774–783. 2002;23:151–169. [16] Boldt M. The meaning of suicide: implications for [32] Rushing NC, Corsentino E, Hames JL, et al. The research. Crisis. 1988;9:93–108. relationship of religious involvement indicators and [17] Pirkola S, Sund R, Sailas E, et al. Community social support to current and past suicidality among mental-health services and suicide rate in Finland: depressed older adults. Aging Ment Health. a nationwide small-area analysis. Lancet. 2013;17:366–374. 2009;373:147–153. [33] Šedivy NZ, Podlogar T, Kerr DC, et al. Community [18] Shah A, Bhandarkar R, Bhatia G. The relationship social support as a protective factor against suicide: A between general population suicide rates and mental gender-specific ecological study of 75 regions of 23 health funding, service provision and national policy: European countries. Health Place. 2017;48:40–46. a cross-national study. Int J Social Psychiatry. [34] Joseph A, Abraham S, Muliyil J, et al. Evaluation of 2010;56:448–453. suicide rates in rural India using verbal autopsies, [19] Homan C, Johar R, Liu T, et al., editors. Toward 1994–9. BMJ. 2003;326:1121–1122. macro-insights for suicide prevention: analyzing [35] Adinkrah M. Criminal prosecution of suicide attempt fine-grained distress at scale. Proceedings of the survivors in Ghana. Int J Offender Ther Comp Workshop on Computational Linguistics and Criminol. 2013;57:1477–1497. Clinical Psychology: From Linguistic Signal to [36] Adinkrah M. Anti-suicide laws in Nine African coun- Clinical Reality; 2014. [cited 2020 Oct 8]. Available tries: criminalization, prosecution and penalization. from: https://www.aclweb.org/anthology/W14-3213. Afr J Criminol Justice Stud. 2016;9:279–291. pdf [37] Andoh-Arthur J, Hjelmeland H, Osafo J, et al. [20] Lee J. Municipal-level determinants of suicide rates in Walking a tightrope: reflections on police gatekeeping South Korea: exploring the role of social capital and roles in suicide research in Ghana. Int J Soc Res local government policies. J Asian Public Policy. Methodol. 2018;21:289–301. 2019;1–18. DOI:10.1080/17516234.2019.1565336 [38] Osafo J, Hjelmeland H, Akotia C, et al. Social injury: [21] Burgess P, Pirkis J, Jolley D, et al. Do nations’ mental an interpretative phenomenological analysis of the health policies, programs and legislation influence attitudes towards suicide of lay persons in Ghana. their suicide rates? An ecological study of 100 coun- Int J Qual Stud Health Well-being. 2011;6:8708. tries. Aust N Z J Psychiatry. 2004;38:933–939. [39] Reynders A, Scheerder G, Van Audenhove C. The [22] Sher L. Are suicide rates related to the psychiatrist reliability of suicide rates: an analysis of railway sui- density? A cross-national study. Front Public Health. cides from two sources in fifteen European countries. 2016;3. DOI:10.3389/fpubh.2015.00280 J Affect Disord. 2011;131:120–127. GLOBAL HEALTH ACTION 9 [40] Doku VCK, Mallett MR. Collaborating with develop- suicide. Death Stud. 2018;1–11. DOI:10.1080/ ing countries in psychiatric research. Br J Psychiatry. 07481187.2018.1527796 2018;182:188–189. [48] Mann JJ, Waternaux C, Haas GL, et al. Toward [41] Pridmore S. Medicalisation of suicide. Malays J Med a clinical model of suicidal behavior in psychiatric Sci. 2011;18:78. patients. Am J Psychiatry. 1999;156:181–189. [42] Milner A, Sveticic J, De Leo D. Suicide in the absence [49] Jacob K, Sharan P, Mirza I, et al. Mental health sys- of mental disorder? A review of psychological autopsy tems in countries: where are we now? Lancet. studies across countries. Int J Social Psychiatry. 2007;370:1061–1077. 2013;59:545–554. [50] Jacob K. Suicide in india: part perceptions, partial [43] Andoh-Arthur J, Knizek BL, Osafo J, et al. Suicide insights, and inadequate solutions. Natl Med J India. among men in Ghana: the burden of masculinity. 2017;30:155–158. Death Stud. 2018;42:658–666. [51] Vos T, Barber RM, Bell B, et al. Global, regional, and [44] Asante KO, Kugbey N, Osafo J, et al. The prevalence national incidence, prevalence, and years lived with and correlates of suicidal behaviours (ideation, plan disability for 301 acute and chronic diseases and inju- and attempt) among adolescents in senior high ries in 188 countries, 1990–2013: a systematic analysis schools in Ghana. SSM Popul Health. 2017;3:427–434. for the Global Burden of Disease Study 2013. Lancet. [45] Haw C, Hawton K, Gunnell D, et al. Economic reces- 2015;386:743–800. sion and suicidal behaviour: possible mechanisms and [52] Boutayeb A. The double burden of communicable and ameliorating factors. Int J Social Psychiatry. non-communicable diseases in developing countries. 2015;61:73–81. Trans R Soc Trop Med Hyg. 2006;100:191–199. [46] Goldman-Mellor SJ, Saxton KB, Catalano RC. [53] Phillips M. Suicide prevention in developing countries: Economic contraction and mental health: a review of where should we start? World Psychiatry. 2004;3:156. the evidence, 1990–2009. Int J Ment Health Syst. [54] UNDP. Sustainable development goals. n.d. [cited 2010;39:6–31. 2020 Apr 03]. Available from: https://www.undp.org/ [47] Hjelmeland H, Loa Knizek B. The emperor’s new content/undp/en/home/sustainable-development- clothes? A critical look at the interpersonal theory of goals.html