Accepted Manuscript

Gender Inequality and Suicide Gender Ratios in the World

Qingsong Chang , Paul S.F. Yip , Ying-Yeh Chen

PII: S0165-0327(18)31326-0
DOI: https://doi.org/10.1016/j.jad.2018.09.032
Reference: JAD 10114

To appear in: Journal of Affective Disorders

Received date: 19 June 2018

Revised date: 15 August 2018
Accepted date: 15 September 2018

Please cite this article as: Qingsong Chang , Paul S.F. Yip , Ying-Yeh Chen , Gender Inequal-
ity and Suicide Gender Ratios in the World, Journal of Affective Disorders (2018), doi:
https://doi.org/10.1016/j.jad.2018.09.032

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service
to our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and
all legal disclaimers that apply to the journal pertain.
 ACCEPTED MANUSCRIPT

Highlights

 Global variations in suicide gender ratios were associated with gender

inequality.

 Male to female suicide rate ratios were higher in countries with more
egalitarian gender norms; whereas lower suicide gender ratios were generally
observed in countries with higher level of gender inequality.

T
IP
 Cultural-based indicator of gender inequality (i.e. natality inequality/son
preference) was more sensitive than economic-based indicator (i.e. the GII).

CR
 Strategies to eliminate culturally embedded discrimination against women
have the potentials to prevent suicides.

US
AN
M
ED
PT
CE
AC

1
 ACCEPTED MANUSCRIPT

Gender Inequality and Suicide Gender Ratios in the World

Qingsong Chang a, Paul S.F. Yip a, b, Ying-Yeh Chen c,d *

a
Department of Social Work and Social Administration, Faculty of Social Sciences,
University of Hong Kong, Hong Kong
b
Hong Kong Jockey Club Center for Suicide Research and Prevention, University of

T
Hong Kong, Hong Kong
c

IP
Taipei City Psychiatric Centre, Taipei City Hospital, Taipei City, Taiwan.
d
Institute of Public Health, School of Medicine, National Yang-Ming University,

CR
Taipei City, Taiwan

* Correspondence to Professor Ying-Yeh Chen

US
309 Songde Road, XinYi District, Taipei City, Taiwan
TEL: 886-2-27263141
AN
e-mail: ychen@tpech.gov.tw
M
ED
PT
CE
AC

2
 ACCEPTED MANUSCRIPT

Abstract

Background: To assess whether gender inequality determines the patterns of suicide

gender ratios.

Methods: Using suicide data obtained from the World Health Organization Statistical

Information System, 2012, suicide gender ratios were calculated and a world map of

T
the ratios constructed. Forest plots were utilized to assess whether gender inequality

IP
(indicated by the Gender Inequality Index [“GII”] and male to female sex ratios at

CR
birth) moderated the worldwide patterns of suicide gender ratios. Regression analyses

US
were then performed to estimate the extent to which gender inequality affects suicide

gender ratios before and after controlling for human development level (“HDI”).
AN
Results: Gradient relationships of suicide gender ratios across 3 tertiles of GII were
M

observed (ratios= 2.03, 2.54, 3.51, respectively for high, moderate and low GII,

P=0.03). High sex ratio at birth was significantly associated with lower suicide gender
ED

ratio (ratio=1.64 vs. 2.75, P=0.00). Regression analyses showed that highest tertile of
PT

GII and high sex ratios at birth were significantly associated with lower suicide

gender ratios (P=0.00 and P=0.00, respectively). When the level of human
CE

development level was controlled, high sex ratio at birth remained to be a significant
AC

determinant of suicide gender ratios (P=0.00), whereas the significance of GII

disappeared (P=0.19).

Limitations：The cross-sectional data do not allow for causal inferences.

3
 ACCEPTED MANUSCRIPT

Conclusions: Male to female suicide ratios were higher in countries with more

egalitarian gender norms. Strategies to eliminate culturally embedded gender

discrimination have the potentials to prevent suicides.

Key words: Suicide gender ratios; Gender inequality; Sex ratios at birth; Women‟s

resilience; Global analyses

T
IP
CR
US
AN
M
ED
PT
CE
AC

4
 ACCEPTED MANUSCRIPT

Introduction

It is well known that in most parts of the world, suicide rates in men are generally

higher than in women (World Health Organization, 2017a). According to the global

suicide report from the World Health Organization, the estimated overall male to

female suicide rate ratio was 1.9 in 2012 (male:15.0 per 100,000; female: 8.0 per

T
100,000). In high-income countries however, the male to female ratio of

IP
age-standardized suicide rate was 3.5 (male:19.9 per 100,000; female: 5.7 per

CR
100,000), whereas in low- and middle-income countries, the male-to-female ratio was

US
1.57 (male: 13.7 per 100,000, female: 8.7 per 100,000). The male to female suicide

rate ratios, nonetheless, vary greatly between countries. Among the 172 Member
AN
States, male to female suicide rate ratios ranged between 0.5 and 12.5 in 2012 (World
M

Health Organization, 2014). However, to the best of our knowledge, no previous

studies have explored the underlying factors in the relation of global variations in
ED

suicide gender ratios.

PT

Several studies pointed out that male to female suicide rate ratios in East Asian

countries were lower than in the Western countries, as suicide rates in women in those
CE

regions were higher (Chen, Chen, Lui, & Yip, 2017; Lester, 1982; Yip, Callanan, &
AC

Yuen, 2000). Lower women‟s status may be a potential explanation for the high

female suicide rates in East Asia (Chen et al., 2017; Phillips, Li, & Zhang, 2002;

Zhang, Jia, Jiang, & Sun, 2006,) as low women‟s status is related to violence against

women, feminization of poverty and women‟s shortage of resources to deal with life

predicament, lack of education and economic opportunities (Chant, 2008; Maselko &

5
 ACCEPTED MANUSCRIPT

Patel, 2008). However, whether women‟s status is associated with the patterns of

suicide gender ratios in the world has yet to be explored.

Recently, the United Nations Development Program (“UNDP”) compiled a new

global index of gender inequality – Gender Inequality Index (“GII”), to capture

women‟s disadvantage in the dimensions of empowerment, labor activity and

T
reproductive health (Gaye, Klugman, Kovacevic, Twigg, & Zambrano, 2010;

IP
Permanyer, 2013). This composite index presents a summary of gender disparity in

CR
each country and provides a simple measure to examine the associations between

US
global patterns of gender disparity and male to female suicide gender ratios. However,

GII has been criticized to have put too much emphasis on the economic and material
AN
dimensions of the disadvantages in women, neglecting the inherited cultural aspects
M

of gender inequality (Permanyer, 2013). In the current analyses, natality inequality,

i.e., son preference (Amartya Sen, 2001), has been added to represent the cultural
ED

aspect of gender inequality. Without any deliberate human interruption, laws of nature
PT

revealed that for every 100 girls, about 102-106 boys were being born (UNFPA Asia

and the Pacific Regional Office, 2012). Due to strong cultural attitude of son
CE

preference in some societies, especially in the low and middle income countries,
AC

owing to excess in female child mortality and sex-selective abortion, many girls went

missing (Das Gupta et al., 2002; A. Sen, 2003). Son preference represents culturally

rooted discrimination against women that women themselves internalize the practice,

this kind of gender bias self-perpetuates with little need of direct reinforcement from

the male world (Das Gupta et al., 2002; Kabeer, 1999). Combining GII and sex ratios

6
 ACCEPTED MANUSCRIPT

at birth would be better able to capture gender inequality in a more comprehensive

way, encompassing material, health as well as cultural dimensions of gender

inequality.

The aims of the current study are first, to provide descriptive analyses of suicide

gender ratios in the world, and second, to examine whether egalitarian gender norms

T
were negatively associated with male to female suicide rate ratios. As the level of

IP
human development may affect the associations between gender inequality and

CR
suicide gender ratios (Chen et al., 2017), it was controlled in the current study.

Methods
US
AN
Data
M

The suicide mortality data in 2012 were obtained from the WHO mortality database

(World Health Organization, 2017b), death records data from Taiwan‟s Department of
ED

Health (Ministry of Health and Welfare, 2017) and the Coroner's Court of Hong Kong
PT

(Coroner's Court of Hong Kong, 2013). Suicide mortality was identified using the

International Classification of Diseases, 10th revision (“ICD-10”) code of X60-X84

CE

(intentional self-harm). A total of 174 regions (172 WHO members plus Taiwan and
AC

Hong Kong) were assessed.

The Gender Inequality Index (“GII”) in 2012 was obtained from the United Nations

Development Programme (“UNDP”) which measures the human development cost of

gender inequality (United Nations Development Programme, 2017a). This composite

index measures three dimensions of gender inequality: (1) reproductive health:

7
 ACCEPTED MANUSCRIPT

measured by maternal mortality ratio and adolescent fertility rate; (2) empowerment:

measured by male and female secondary school educational attainment, and women‟s

level of parliamentary representation; and (3) labor market: measured by labor force

participation of men and women. The GII ranges between zero and one, with zero

being 0% inequality and one indicating 100% inequality. The GII data were available

T
for 157 regions that were examined, among them, Denmark had the lowest inequality,

IP
with the GII equals to 0.048 (indicating 4.8% potential human development loss due

CR
to gender inequality), and Yemen had the highest inequality (GII= 0.767). The

US
missing data on the GII were handled by Listwise Deletion method and countries with

missing information on the GII were not used in the analyses.

AN
Data on sex ratios at birth were obtained from the World Bank (The World Bank,

2017). Taiwan‟s Department of Statistics (Ministry of the Interior Department of

M

Statisitics, 2017) and the Census and Statistics Department (“C&SD”) of Hong Kong
ED

(Census and Statistics Department, 2017). Overall, in 2012, data were available from
PT

174 regions of the world, the values ranged between 102 in Rwanda and 116 in China.

Data on the Human Development Index (“HDI”) were obtained from the United
CE

Nations Development Programme (“UNDP”) (United Nations Development

AC

Programme, 2017b). The HDI measures a country‟s performance in life expectancies,

educational indicators, and the ability to lead a decent standard of living. HDI was

modelled as a control covariate in the current analyses. HDI was dichotomized to a

2-category variable, i.e., combining Very high HDI and High HDI countries; Medium

HDI and Low HDI countries.

8
 ACCEPTED MANUSCRIPT

Analytic Strategies

Suicide rates were age-standardized, using the world standard population structure

in 2012. Male to female suicide gender ratios were then calculated and a world map

constructed based on different levels of gender ratios – <1.0, 1.0-1.9, 2.0-2.9, 3.0-3.9,

4.0-4.9 and >=5. Graphical presentation of predictors, i.e., GII and sex ratios at birth,

T
and suicide gender ratios, stratified by HDI are presented in the figures in the

IP
Appendix (Appendix Figure 1 and Appendix Figure 2).

CR
In order to analyze the impact of gender inequality on suicide gender ratios, forest

US
plots were utilized to show how indicators of gender inequality i.e., the GII and sex

ratios at birth, moderated male to female suicide rate ratios. The GII was grouped into
AN
tertiles – high (>0.49), medium (0.29-0.49) and low (<0.29). Sex ratios at birth were

modelled as a dichotomized variable – normal and high, i.e., for every 100 girls more
M

than 106 boys were being born. The total between-group variance (Total Qb) was
ED

calculated to assess whether differential moderating effects existed in different

PT

subgroups (Borenstein & Higgins, 2013). Pooled suicide gender ratios with 95%

confidence interval (Morris & Gardner, 1988) were calculated using Comprehensive
CE

Meta-Analysis software program. The software program takes population size of each
AC

analyzed region into account. A series of sensitivity analyses were then conducted to

examine whether the point estimate was driven by a region with a large population

size. One country was taken out one at a time to examine whether the estimated

suicide gender ratios of the remaining 173 regions were statistically different from the

pooled estimates including all 174 regions. The sensitivity analyses indicated that

9
 ACCEPTED MANUSCRIPT

China was the only region that had substantial impact on the pooled estimate. The

suicide gender ratio was estimated to be 2.01 (95%CI [1.73, 2.33]) with all 174

regions included, excluding China would result in a statistically significant increase in

the pooled ratio to 2.36 (95% CI [1.76, 2.37]). Hence, the reported results included

with and without China.

T
Finally, having controlled for the level of human development, i.e., the HDI,

IP
linear regression analyses were conducted to examine the associations between gender

CR
inequality and suicide gender ratios. The outcome variable was log-transformed as

US
according to the Kline‟s rule, i.e., skew index absolute value <3; kurtosis index

absolute values <10 (Kline, 2005), the distribution of suicide gender ratios was not
AN
normal. In addition to examining the rate ratios, whether gender inequality affected
M

female suicide rates is also explored by conducting an additional regression analysis

with female suicide rates as the outcome variables, and indicators of gender inequality
ED

as the independent variables, controlling for male suicide rates and HDI. This
PT

additional analysis is presented in Appendix Table 1.

CE

Results
AC

As can be seen from Figure 1, suicide gender ratios varied markedly across

different countries in the world. Overall, highest male to female suicide gender ratios

were found in East European countries and the former Soviet Republics; lowest ratios

however appeared in the Asian countries. A scatter plot of GII and suicide gender

ratios stratified by HDI was presented in Appendix Figure 1. A boxplot presenting the

10
 ACCEPTED MANUSCRIPT

distribution of sex ratios at birth and suicide gender ratios was shown in Appendix

Figure 2.

Figure 2 presents the forest plots of suicide gender ratios in different levels of GII.

The impact of GII on suicide gender ratios was not observed (P=0.42). However,

when a separate analysis excluding China was performed (the lower part of Figure 2),

T
a gradient relationship of suicide gender ratios across high, medium and low GII

IP
countries (P=0.03) was found. In other words, when China was not included in the

CR
analyses, countries in the highest tertile of GII (i.e. countries with the largest gender

US
inequality) had the lowest suicide gender ratio (2.03); whereas countries in the lowest

tertile of GII (i.e. countries with better performance in gender equality) had the
AN
highest gender ratio (3.51).
M

Referring to the forest plots of suicide gender ratios in countries with normal versus

high sex ratios at birth (Figure 3), the suicide gender ratio was significantly lower in
ED

countries with high sex ratios at birth (1.64 vs. 2.75, P=0.00). The analysis excluding
PT

China (the lower part of Figure 3) still revealed this pattern, although to a lesser extent

(2.08 vs. 2.75, P=0.09).

CE

Table 1 shows the regression analyses estimating the impacts of gender inequality
AC

on suicide gender ratios before and after controlling for the HDI. Overall, indicators

of gender inequality i.e., high sex ratios at birth and high GII, were associated with

lower level of male to female suicide gender ratios. However, in the HDI adjusted

model, sex ratios at birth remained to be significantly associated with suicide gender

ratios (P=0.00); whereas the significance of GII has disappeared (P>0.05) (Table 1).

11
 ACCEPTED MANUSCRIPT

The regression analyses of gender inequality and female suicide rates are presented

in Appendix Table 1. The results are generally consistent with the analyses modelling

the male to female suicide rate ratios as outcome variables.

Discussion

T
Main Findings

IP
The present results indicate that international variations in suicide gender ratios

CR
were associated with gender inequality. Male to female suicide rate ratios were higher

US
in countries with more egalitarian gender norms; whereas lower suicide gender ratios

were generally observed in countries with higher level of gender inequality.

AN
Cultural-based indicator of gender inequality i.e., natality inequality/son preference,
M

was more sensitive than economic-based indicator, i.e., the GII. After controlling for

levels of human development i.e., HDI, associations between natality inequality and
ED

suicide gender ratios remained, while the impacts of GII on suicide gender ratios
PT

disappeared.

Interpretation and comparison with previous studies

CE

1. Suicide gender ratios in the world

AC

This is the first study which demonstrates that gender inequality, particularly

culturally rooted inequality, to be a potent determinant of suicide gender ratios in the

world. Previous studies tend to indicate that countries with low male to female suicide

rate ratios generally stem from high female suicide rates rather than the results of low

male rates (Canetto & Lester, 1995a; Lester, 1982). The present findings however,

12
 ACCEPTED MANUSCRIPT

suggest that gender inequality, particularly in the form of natality inequality, may

result in higher female suicide rates and consequently lower male to female suicide

rate ratios. Relative suicide gender ratios have been presented, and absolute female

rates included in the table in the Appendix because relative rates are better able to

take into account the „national tradition‟ of suicides with different base rate (Hansen

T
& Pritchard, 2008).

IP
2. Women‟s strengths and resilience

CR
Even though women face a variety of social disadvantages, suicide rates are

US
generally lower in women than in men. This could be an indication of women‟s

resilience in responding to stress and crises. Canetto points out that women‟s suicidal
AN
behaviors tend to be labeled pejoratively, although women are in general more

adaptive and are survivors (Canetto & Lester, 1995b). For example, „failed suicides‟
M

„attention-seeking‟ and „not determined to take one‟s own life‟ are usually used to
ED

describe women‟s suicidal behaviors; whereas men‟s suicidal behaviors are

PT

sometimes termed positively (e.g. “successful” suicides), even when those behaviors

led to self-destruction (Canetto & Lester, 1995b). This kind of gender bias in
CE

suicidology may prevent researchers from looking into the strengths and resilience in
AC

women. The present results however, suggest that future research should place

emphasis on how women overcome social adversities and how gender inequality

thwarts their capabilities to survive.

3. Cultural roots of gender inequality

Gender inequality in the form of son preference has prominent impact on the

13
 ACCEPTED MANUSCRIPT

distribution of suicide gender ratios in the world. Although most societies show some

degrees of son preference, some are mild and may not virtually be detectable;

however, others are strong enough to cause excess in female infanticides and even

sex-selective feticides (Chung & Gupta, 2007; Das Gupta et al., 2002; A. Sen, 2003).

Son preference is particularly intense in East and Southeast Asia where patrilineal

T
kinship is rigidly practiced (Das Gupta et al., 2002). In those patrilineal societies,

IP
assets and properties are passed to sons, the daughters‟ are absorbed into their

CR
husbands‟ lineage after marriage. In India, the need to pay high dowries can be a

US
further disincentive to have a daughter. In the patrilineal kinship system, aged parents

rely on the support of their sons, not on their daughters; daughters, however, have to
AN
take care of their husbands‟ parents. In East Asia, the practice of ancestor worship
M

further fuels the need to have male offspring. It is believed that the rituals of ancestor

worships can ensure secured afterlife for ancestors, and ancestors in turn, bring good
ED

luck to their descendants (Das Gupta et al., 2002). The tradition dictates the male line
PT

to (and only the male line can) practice the rituals; without bearing a son to worship

the ancestors are regarded as failing to perform filial duty in those societies (Das
CE

Gupta et al., 2002; Deuchler, 1992). This kind of rigid patrilineal practice cannot be
AC

easily changed by raising women‟s status at the individual level. In Asia, women‟s

educational attainment and employment opportunities have also increased immensely

in the past few decades, nevertheless, the intensity of son preference still prevails.

The case of China provides an excellent example to illustrate how culturally

rooted gender inequality is resistant to change, and hence may be more sensitive in

14
 ACCEPTED MANUSCRIPT

detecting the correlation between gender inequality and suicide gender ratios. The

present analyses on GII and suicide gender ratios did not reveal a correlation until

China is excluded from the analyses (Figure 2). In a very short time span, China has

risen to an economic superpower, women‟s health performance, such as maternal

mortality, educational attainment and employment opportunities have simultaneously

T
increased unprecedentedly along with its economic boom. In fact, China is ranked in

IP
the lowest tertile of the GII, indicating that the country has achieved fairly good

CR
gender equality when measured by the economic/material aspects and the

US
reproductive health dimensions of women‟s status. China‟s sex ratio at birth however,

is among the highest in the world, for every 100 girls, 116 boys were being born. At
AN
first sight, this may seem to be contradictory. China has made prominent
M

improvement in women‟s empowerment, on the other hand, it is deeply rooted in

patriarchy system which still reinforces the practice of son preference. In fact, women
ED

in relatively higher socioeconomic status may have more resources to perform

PT

perinatal sex selection. The phenomenon of this „China paradox‟ clearly indicates that

achieving gender equality should go beyond improving opportunities in health,

CE

education, employment or political participation, moreover, it is also important to

AC

scrutinize inherited beliefs and challenge traditional priorities. This way, the deeply

rooted cultural prescription of gender can possibly be altered.

4. Modernization and suicide gender ratios

Earlier studies indicated that modernization (e.g. urbanization, industrialization)

could bring about female emancipation, yet, it might at the same time aggravate

15
 ACCEPTED MANUSCRIPT

women‟s deviant behavior, such as suicide (Stack and Danigelis, 1985; Norris, 2003).

For example, Stack and Danigelis analyzed the data between 1919 and 1972 in 17

industrialized countries; they found an association between modernization and higher

female suicide rates in 14 out of the 17 countries (Stack and Danigelis, 1985). Forty

years after this earlier study, one recent paper by Chen et al demonstrated that the

T
relationship between gender equality (indicated by women‟s labor force participation

IP
rates) and suicide rates in both genders was dependent on the level of human

CR
development. In countries where human developmental capabilities are strong, higher

US
level of gender equality was beneficial for women; while the male suicide rates were

not affected by gender equality. However, the advantages of gender equality reach a
AN
plateau in countries where the level of human development is at the very top, such as
M

the North Scandinavian countries. Nonetheless, in countries where human capabilities

are relatively less advanced, gender equality was associated with an increase in
ED

suicide rates in both genders (Chen et al., 2017). In the current analyses, the
PT

correlation coefficient between HDI and GII was -0.89 (Appendix Table 2), indicating

a strong correlation between human development and women‟s emancipation. Overall,

CE

the relationship between gender equality and suicide rates is not static; it changes over
AC

time, varies by the social context, and depends on which genders are being examined,

and also the indicators being used.

Policy implications

The present study results indicate that gender equality may influence the

distribution of suicide gender ratios through reducing relative suicide rates in women.

16
 ACCEPTED MANUSCRIPT

Thus, eradicating gender inequality would have the potential to promote the general

well-being of the population, especially for women. However, policies to improve

gender equality are highly context dependent.

Countries differ in their cultural and religious backgrounds, and in their

developmental stages, therefore, priorities and policies need to vary widely depending

T
on the settings. For example, in many developing countries, the key issues are to

IP
maintain basic functioning, policies related to access to food, rights to education and

CR
health care are thus crucial.

US
For countries in East and Southeast Asia however, where imbalanced sex ratios at

birth prevails, legislations to penalize non-medical sex-selective abortions can be one

AN
of the policy options (Nie, 2011). In China, the recent policy of relaxing the one-child
M

policy may further help to reduce the sex ratio imbalance, as sex-selective abortion is

highly correlated with fertility decline (Nie, 2011; Zhou, Wang, Zhou, & Hesketh,
ED

2012).
PT

However, in more developed countries, appropriate policies may include equal

opportunities to employment and payment, reduced feminization of poverty and the

CE

increase of the number of women in decision-making roles (Pan American Health

AC

Organization, 2017). Furthermore, enacting laws to prevent violence against women

and support to the victims, enforcing legislations in relation to pregnancy protection,

maternity and family leave policies, to name a few, are also important policy elements

(Pan American Health Organization, 2005).

Gender equality action plans should also involve the educational sectors. It is

17
 ACCEPTED MANUSCRIPT

important to design gender equality and gender sensitive curricula collaborating with

the media and women‟s rights groups. Moreover, civil forums and regular campaigns

can further be organized to raise the awareness of the public on gender equality.

Although culture-based gender discrimination may not be shaped overnight, new

concepts and norms nonetheless can be diffused and gradually be accepted through

T
legislations, education and media campaigns. Lastly, gender equality policies should

IP
also be regularly reviewed and scrutinized, then their inadequacies can be timely

CR
addressed to and modified.

Strengths and Limitations

US
The strength of this study lies in its global perspective in the examination of the
AN
relationships between gender inequality and suicide gender ratios. It was estimated
M

that 73% suicides in the world occurred in developing countries; these nations,

however, accounted for only a small part of research on suicide (Vijayakumar, 2005;
ED

Yip and Law, 2010). The current paper however, brought these nations which were
PT

typically left out into the analyses. Through a gender lens, potential policies are

advanced to eliminate gender inequality and hopefully, suicide rates can drop
CE

accordingly.
AC

Nonetheless, the results of the present study should be interpreted in light of the

following limitations. Firstly, validity of the suicide mortality data should be

interpreted with a degree of caution. Underreporting of suicides is particularly

problematic in developing countries (World Health Organization, 2014). However,

underreporting would have driven the present results towards null, as the rates of

18
 ACCEPTED MANUSCRIPT

underreporting were presumably to be higher in women than in men (Canetto &

Lester, 1995a). Secondly, this is a cross-sectional ecological study, the observed

correlations do not necessarily indicate causality. Thirdly, indicators used to represent

gender inequality i.e., the GII and sex ratios at birth, are not exhaustive. For example,

domestic-level gender inequality (e.g. time used or decision-making in the domestic

T
sphere) was not assessed. Furthermore, UNDP presented GII as a composite index,

IP
the values on individual dimensions of the GII however, were not available. Thus this

CR
calls for future research to disentangle the elements of GII that are especially

predictive of suicide gender ratios.

Conclusions
US
AN
Gender inequality contributed to global variations in suicide gender ratios, lower
M

ratios were found in places where gender-based discrimination was more prominent.

It has long been known that even though women face a variety of social disadvantages,
ED

their suicide rates are lower than their male counterparts. How women develop
PT

strengths to survive hardships are areas that need in-depth exploration, and this kind

of inquiry would complement suicide prevention in men.

CE

Gender inequality is not a uni-dimensional construct, it involves a variety of

AC

dimensions and is presented in different ways. Inequality in the form of health and

objective social status may be better eliminated through general improvement in

economic conditions. However, culturally embedded discrimination cannot easily be

removed by way of economic prosperity. Thus, understanding the deeply rooted

cultural component of gender inequality is one of the keys to unveil suicide gender

19
 ACCEPTED MANUSCRIPT

ratios in the world and it will provide important information for formulating effective

suicide prevention measures at a global level.

Author Statement

Contributors

QS Chang: study design, data collection, data analysis, figures, data interpretation,

T
writing of the first draft.

IP
PSF Yip: study design, data collection, data analysis, data interpretation, critically

CR
revise the first draft of the manuscript.

US
YY Chen: literature search, study design, data collection, data interpretation, finalized

the manuscript, supervised the study.

AN
M

Role of the Funding source

The current study is supported by Taiwan's Ministry of Science and Technology

ED

(MOST 105-2314-B-532-004-MY3), the Department of Health Taipei City

PT

Government (10701-62-007), and a Strategic Public Policy Research (SPPR) of the

University Grant Council at the University of Hong Kong (SPPR-HKU-12).

CE
AC

The funders have no role in the study design; in the collection, analysis and

interpretation of the data; in the writing of the report; and in the decision to submit the

paper for publication.

20
 ACCEPTED MANUSCRIPT

Acknowledgements

None.

References
Borenstein, M., & Higgins, J.P.T., 2013. Meta-Analysis and Subgroups. Prev Sci, 14,

T
134–143.

IP
Canetto, S. S., & Lester, D., 1995a. The epidemiology of women's suicidal behavior.
In S. S. Canetto & D. Lester (Eds.), Women and suicidal behavior (pp. 35-57).

CR
New York, NY: Springer.
Canetto, S. S., & Lester, D., 1995b. Women and suicidal behavior: issues and
dilemmas. In S. S. Canetto & D. Lester (Eds.), Women and suicidal behavior
(pp. 3-8). New York, NY: Springer.

sex. Retrieved from

US
Census and Statistics Department, 2017. Demographic characteristics-Live births by

https://www.censtatd.gov.hk/hkstat/sub/gender/demographic/index.jsp.
AN
Chant, S., 2008. The „feminisation of poverty‟ and the „feminisation‟ of anti-poverty
programmes: room for revision? J Dev Stud, 44(2), 165-197.
Chen, Y. Y., Chen, M., Lui, C. S. M., & Yip, P. S. F., 2017. Female labour force
M

participation and suicide rates in the world. Soc Sci Med, 195, 61-67.
doi:10.1016/j.socscimed.2017.11.014.
Chung, W., & Gupta, M. D., 2007. The decline of son preference in South Korea: the
ED

roles of development and public policy. Popul Dev Rev, 33(4), 757-783.
Coroner's Court of Hong Kong, 2013. Coroners’ report on suicide in 2012. Hong
Kong, HK: Coroner's Court of Hong Kong.
PT

Das Gupta, M., Zhenghua, J., Bohua, L., Zhenming, X., Chung, W., & Hwa-Ok, B.,
2002. Why is son preference so persistent in East and South Asia? A
CE

cross-country study of China, India, and the Republic of Korea. Washington,

DC: The World Bank.
Deuchler, M., 1992. The confucian transformation of Korea: a study of society and
AC

ideology. Cambridge, MA: Harvard University, Harvard University Asia

Center.
Gaye, A., Klugman, J., Kovacevic, M., Twigg, S., & Zambrano, E., 2010. Measuring
key disparities in human development: the Gender Inequality Index. New York,
NY: United Nations Development Programme.
Hansen, L., & Pritchard, C., 2008. Consistency in suicide rates in twenty-two
developed countries by gender over time 1874-78, 1974--76, and 1998-2000.
Arch Suicide Res, 12(3), 251-262. doi:10.1080/13811110802101153.
Kabeer, N., 1999. Resources, agency, achievements: reflections on the measurement
of women's empowerment. Development and Change, 30(3), 435-464.
21
 ACCEPTED MANUSCRIPT

Kline, R. B., 2005. Principles and Practice of Structural Equation Modeling ( 2nd
ed.). New York, NY: Guilford Press.
Lester, D., 1982. The distribution of sex and age among completed suicides: a
cross-national study. Int J Soc Psychiatry, 28(4), 256-260.
doi:10.1177/002076408202800403.
Maselko, J., & Patel, V., 2008. Why women attempt suicide: the role of mental illness
and social disadvantage in a community cohort study in India. J Epidemiol
Community Health, 62(9), 817-822. doi:10.1136/jech.2007.069351.
Ministry of Health and Welfare, 2017. Ministry of Health and Welfare. Retrieved from
http://www.mohw.gov.tw/mp-2.html.

T
Ministry of the Interior Department of Statisitics, 2017. Interior national indicators.

IP
Retrieved from https://www.moi.gov.tw/stat/chart.aspx.
Morris, J. A., & Gardner, M. J. 1988. Statistics in medicine: Calculating confidence

CR
intervals for relative risks (odds ratios) and standardised ratios and rates.
British MedicalJjournal (Clinical research ed.), 296, 1313-16.
Nie, J. B., 2011. Non-medical sex-selective abortion in China: ethical and public
policy issues in the context of 40 million missing females. Br Med Bull, 98,
7-20. doi:10.1093/bmb/ldr015.
US
Norris, N., 2003. Rising tide: gender equality and cultural change around the world.
Cambridge: Cambridge University Press.
AN
Pan American Health Organization, 2005. Gender equality policy. Washington D.C.:
Pan American Health Organization.
Pan American Health Organization, 2017. Gender equality policy. Retrieved from
M

http://www.paho.org/hq/index.php?option=com_content&view=article&id=26
80&Itemid=4017&lang=en.
Permanyer, I., 2013. A critical assessment of UNDP's Gender Inequality Index. Fem
ED

Econ, 19(2), 1-37.

Phillips, M. R., Li, X., & Zhang, Y., 2002. Suicide rates in China, 1995-99. The
Lancet, 359(9309), 835-840. doi:S0140-6736(02)07954-0 [pii]
PT

10.1016/S0140-6736(02)07954-0.
Sen, A., 2001. The many faces of gender inequality. The New Republic, 17, 35-39.
CE

Sen, A., 2003. Missing women-revisited. BMJ, 327(7427), 1297-1298.

doi:10.1136/bmj.327.7427.1297.
Stack, S., & Danigelis, N., 1985. Modernization and the sex differential in suicide,
AC

1919-1972. Comparative Social Research, 8, 203-216.

The World Bank, 2017. World Bank data-sex ratio at birth (male births per female
births). Retrieved from
https://data.worldbank.org/indicator/SP.POP.BRTH.MF?end=2015&start=201
5&view=map&year=2012.
UNFPA Asia and the Pacific Regional Office, 2012. Sex imbalances at birth: current
trends, consequences and policy implications. Bangkok, Thailand: UNFPA
Asia and the Pacific Regional Office.
United Nations Development Programme, 2017a. Gender Inequality Index (GII).
Retrieved from http://hdr.undp.org/en/content/gender-inequality-index-gii.
22
 ACCEPTED MANUSCRIPT

United Nations Development Programme, 2017b. Human development data-Human

Development Index. Retrieved from http://hdr.undp.org/en/data.
Vijayakumar, L., 2005. Suicide and mental disorders in Asia. International Review of
Psychiatry, 17(2), 109-114.
World Health Organization, 2014. Preventing suicide: a global imperative. Geneva,
Switzerland: World Health Organization.
World Health Organization, 2017a. Global Health Observatory (GHO) data-suicide
rates (per 100 000 population). Retrieved from
http://www.who.int/gho/mental_health/suicide_rates_male_female/en/.
World Health Organization, 2017b. Health statistics and information

T
systems-download the raw data files of the WHO Mortality Database.

IP
Retrieved from http://www.who.int/gho/mental_health/suicide_rates/en/.
Yip, P. S. F., Callanan, C., & Yuen, H. P., 2000. Urban/rural and gender differentials in

CR
suicide rates: east and west. J Affect Disord, 57(1-3), 99-106.
doi:S0165032799000580 [pii].
Yip, P.S.F, Law Y.W. 2010. Towards Evidence-based Suicide Prevention Programmes.
World Health Organization, West Pacific Region.

US
Zhang, J., Jia, S., Jiang, C., & Sun, J., 2006. Characteristics of Chinese suicide
attempters: an emergency room study. Death Stud, 30(3), 259-268.
doi:10.1080/07481180500493443.
AN
Zhou, C., Wang, X. L., Zhou, X. D., & Hesketh, T., 2012. Son preference and
sex-selective abortion in China: informing policy options. Int J Public Health,
57(3), 459-465. doi:10.1007/s00038-011-0267-3.
M
ED
PT
CE
AC

23
 ACCEPTED MANUSCRIPT

T
IP
CR
Table 1: Regression analyses on the associations between the indicators of gender inequality and male to female suicide rate ratios

Unadjusted Adjusted
Variables
B 95% CI p-value B 95% CI p-value
China included
Gender Inequality Index
Medium vs. Low
High vs. Low
Human Development Index
-0.16
-0.42
[-0.33, 0.02]
[-0.61, -0.23]
US 0.09
0.00
-0.07
-0.19
[-0.26, 0.12]
[-0.48, 0.10]
0.48
0.19
AN
Higher vs. Lower 0.25 [0.01, 0.49] 0.04
Sex ratios at birth
High vs. normal -0.35 [-0.58, -0.12] 0.00 -0.38 [-0.60, -0.17] 0.00
Human Development Index
Higher vs. Lower 0.35 [0.21, 0.48] 0.00
M

China excluded
Gender Inequality Index
Medium vs. Low -0.18 [-0.36, 0.01] 0.04 -0.10 [-0.28, 0.09] 0.32
ED

High vs. Low -0.45 [-0.63, -0.27] 0.00 -0.22 [-0.50, 0.06] 0.13
Human Development Index
Higher vs. Lower 0.25 [0.02, 0.48] 0.04
Sex ratios at birth
High vs. normal -0.29 [-0.53, -0.06] 0.01 -0.32 [-0.54, -0.11] 0.00
PT

Human Development Index

Higher vs. Lower 0.36 [0.23, 0.49] 0.00
Note. Indicators of gender inequality (i.e. Gender Inequality Index (“GII”) and sex ratios at birth) were the independent variables and
log-transformed suicide gender ratios were the dependent variables. GII and sex ratios at birth were analyzed in two separate models.
CE

24
AC
 ACCEPTED MANUSCRIPT

T
IP
CR
US
AN
M
ED
PT

Figure 1. World map of male to female suicide rate ratios, 2012

CE

25
AC
 ACCEPTED MANUSCRIPT

T
IP
CR
US
AN
M
ED

Figure 2. Forest plots of male to female suicide rate ratios and 95% Confidence Intervals (95% CI) among countries with high, medium and low Gender Inequality
PT

Index (“GII”)
CE

26
AC
 ACCEPTED MANUSCRIPT

T
IP
CR
US
AN
M
ED

Figure 3. Forest plots of male to female suicide rate ratios and 95% Confidence Intervals (95% CI) among countries with normal versus high sex ratios at birth
PT
CE

27
AC
 ACCEPTED MANUSCRIPT

T
IP
CR
Appendix Table 1. Regression analyses on the associations between the indicators of gender inequality and female suicide rates

Variables Unadjusted Adjusted

B 95% CI p-value B 95% CI p-value
China included
Gender Inequality Index
Medium vs. Low
High vs. Low
Human Development Index
Higher vs. Lower
Male Suicide Rates
-0.39
0.07
US
[-0.71, 0.08]
[-0.26, 0.40]
0.02
0.68
-0.22
-0.19

-0.48
0.06
[-0.46, 0.02]
[-0.53, 0.16]

[-0.76, -0.19]
[0.05, 0.07]
0.07
0.30

0.00
0.00
AN
Sex ratios at birth
High vs. normal 0.38 [0.00, 0.77] 0.05 0.37 [0.12, 0.61] 0.00
Human Development Index
Higher vs. Lower -0.39 [-0.55, -0.05] 0.00
Male Suicide Rates 0.06 [0.05, 0.07] 0.00
China excluded
M

Gender Inequality Index

Medium vs. Low -0.38 [-0.70, 0.06] 0.02 -0.19 [-0.43, 0.04] 0.11
High vs. Low 0.08 [-0.25, 0.41] 0.63 -0.16 [-0.50, 0.18] 0.36
Human Development Index
Higher vs. Lower -0.48 [-0.76, -0.20] 0.00
ED

Male Suicide Rates 0.06 [0.05, 0.07] 0.00

Sex ratios at birth
High vs. normal 0.35 [-0.04, 0.75] 0.08 0.31 [0.06, 0.56] 0.02
Human Development Index
Higher vs. Lower -0.40 [-0.56, -0.24] 0.00
PT

Male Suicide Rates 0.06 [0.05, 0.07] 0.00

Note. Indicators of gender inequality (i.e. Gender Inequality Index (“GII”) and sex ratios at birth) were the independent variables and
log-transformed female suicide rates were the dependent variables. HDI and male suicide rates were controlled in the adjusted model. GII and
sex ratios at birth were analyzed in two separate models.
CE

28
AC
 ACCEPTED MANUSCRIPT

Appendix Table 2. Pearson's correlations among variables used in the analyses

Pearson's Correlations Suicide Gender Sex ratios
gender ratios Inequality Index at birth

Gender Inequality Index -0.27*

Sex ratios at birth -0.16* -0.12
Human Development Index 0.30* -0.89* 0.09
Note. * P < 0.05; valuables in the matrix are continuous.

T
IP
CR
US
AN
M
ED
PT
CE
AC

29
 ACCEPTED MANUSCRIPT

T
IP
CR
US
AN
M
ED
PT
CE

Appendix Figure 1. Scatter plot of gender inequality index and male-to-female suicide rate ratios, 2012, stratified by human development levels

30
AC
 ACCEPTED MANUSCRIPT

T
IP
CR
US
AN
M
ED
PT

Appendix Figure 2. Boxplot of male to female suicide rate ratios among countries with normal sex ratios at birth versus higher sex ratios at birth,
2012.
CE

31
AC