Suicidal ideation and associated factors among school-going adolescents in rural Ghana

Suicidal ideation is a critical risk for attempted suicide and eventual suicide. Little is known about suicidal ideation among rural adolescents in most sub-Saharan African countries. We aimed to estimate the 12-month prevalence of suicidal ideation and to describe some of the common and gender-specific associated factors among in-school adolescents in rural Ghana. We conducted a cross-sectional survey involving a random sample of 1101 in-school adolescents aged 10–19 years in a rural district in Eastern Ghana. The Suicidal Behavior Questionnaire-Revised was used to assess suicidal ideation. Overall, 25.1% participants (95% CI = 22.5–27.7), representing 28.3% females (95% CI = 24.7–32.2) and 21.5% males (95% CI = 18.0–25.2) reported suicidal ideation during the previous 12 months. Females who experienced personal and interpersonal adversities mainly outside the family context were likely to report suicidal ideation, while suicidal ideation among males was associated with conflict with parents. Regardless of gender, adolescents who reported exposure to a friend’s attempted suicide were about two times more likely to report suicidal ideation. The prevalence of suicidal ideation among adolescents in rural Ghana compares with in-school estimates from other countries within sub-Saharan Africa, but also underscores the need for targeted and universal prevention programmes and intervention efforts to mitigate the potential transition from suicidal ideations to suicidal attempts and eventual deaths by suicide among rural adolescents.

Evidence of systematic reviews and meta-analyses shows variations in national and regional prevalence estimates of suicidal ideation among adolescents (Aggarwal, Patton, Reavley, Sreenivasan, & Berk, 2017;Lim et al., 2019;McKinnon et al., 2016;Uddin et al., 2019). The pooled 12-month prevalence estimates of suicidal ideation among adolescents have identified

Aims of the Study
It is noteworthy that the only available (household-based crosssectional) study reporting evidence on suicidal ideation among rural adolescents in Ghana was conducted in Ningo Prampram (Nyundo et al., 2020), a peri-urban community in the Greater Accra Region (Ghana Statistical Service -GSS, 2013). Thus, the current study sought to provide evidence from rural-dwelling adolescents in Eastern Ghana. Specifically, we aimed to: i. Estimate the 12-month prevalence of suicidal ideation among school-attending adolescents 3 in rural Ghana. ii. Identify the differences and commonalities in gender, personal factors, and social adversities associated with suicidal ideation among school-attending adolescents in rural Ghana.

Materials and Methods
We followed Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) recommendations for cross-sectional study, to design, conduct, and report this study (Von Elm et al., 2007).

Design, Setting, and Participants
Considering that this study is part of a larger study from Ghana, detailed information about the methods used has been reported elsewhere (Odame, Quarshie, Oti-Boadi, Andoh-Arthur, & Asante, 2020). We conducted a cross-sectional survey involving Junior High School 4 (JHS) adolescents in the Ayensuano rural district in the eastern region of Ghana (Ghana Statistical Service, 2014), using a self-report anonymous questionnaire. We obtained the list of schools and pupil statistics from the District Educational Directorate of Ayensuano. We predetermined a minimum sample size of 971 students, using Cochran's formula for calculating a sample size for proportions (Cochran, 1963). However, we increased the predetermined sample size by 25% for two analytical reasons: to obtain satisfactory precision and confidence interval widths, and to compensate for nonresponse (Naing, Winn, & Rusli, 2006). Figure 1 illustrates the participant recruitment process. In all, we approached and invited 1214 students to respond to the survey, but 1101 (female = 575; male = 526) provided complete data included in the studyrepresenting a response of rate of 90%.

Measures
The participants responded to a self-report anonymous questionnaire composed of questions related to socio-demographic characteristics, personal lifestyle factors, social adversity, and suicidal ideation.

Socio-demographic Characteristics
We included seven items to assess the socio-demographic backgrounds of the participants: age, gender (female or male), grade, living arrangement, caretaker's employment status, family structure (measured by father's number of wives), and romantic relationship status.
Outcome Variable Suicidal ideation was the main outcome variable, even though we also assessed suicidal behaviour risk. We used the 4-item Suicide Behavior Questionnaire-Revised [SBQ-R] (Osman et al., 2001) to assess the outcome variables. Suicidal ideation was assessed with the question, How often have you thought about killing yourself in the past year? The response ratings ranged from never = 1, to very often (5 or more times) = 5. Answers were transformed into dichotomous responses (presence vs. absence of suicidal ideation, for analysis). The other three items are Have you ever thought about or attempted to kill yourself?, Have you ever told someone that you were going to commit suicide, or that you might do it?, and How likely is it that you will attempt suicide someday? The three items have different response options: 1-4, 1-3, and 0-6, respectively. The SBQ-R yields a total score between 3 and 18 points. For an adolescent sample, a total minimum score of 7 out of 18 is the cut-off (≥ 7) indicating high suicidal behaviour risk, whereas a total score between 3 and 6 indicates low suicidal behaviour risk (Osman et al., 2001). Thus, in the current study, suicidal behaviour risk was binary, high suicidal behaviour risk vs. low suicidal behaviour risk. In Ghana, the SBQ-R has a satisfactory internal consistency (Cronbach's alpha = 0.78) among both adolescents and adults (Adjorlolo, Anum, & Amin, 2020;Quarshie, Cheataa-Plange, Annor, Asare-Doku, & Lartey, 2019;Quarshie, Odame, & Annor, 2020). The Cronbach's alpha score in the current study was 0.71.

Exposure Variables
We included eight binary (No or Yes) response-rated personal lifestyle factors and variables assessing the experience of social or interpersonal adversities, adopted from the 2012 WHO-Global School-based Student Health Survey questionnaire used in Ghana (Owusu, 2012). Notably, even though many of the variables were assessed using questions that solicited continuous or frequency-based responses, we binarised the responses (as shown in e- Table 1) to facilitate analysis and meaningful interpretations of results. These variables included: weekly alcohol use (In a typical week, how many times do you have at least on alcoholic drink?), truancy (During the past 12 months, how many days did you miss classes or school without permission?), bullying victimisation (During the past 12 months, how many days were you bullied?), and sexual abuse victimisation (Has anyone forced you [i.e. physically or verbally] to engage in sexual activities against your will?). Additionally, we adopted one item from the 5-item Duke University Religion Index (DUREL) (Koenig & Büssing, 2010) to assess religious participation: How often do you attend church or other religious meetings?, with response options ranging from (1) never to (6) more than once per week. The supplementary material (e-Table 1) shows the variables included in this study and the specific questions used to assess them in the survey.

Procedure
The data collection took place between August 2019 and January 2020. The participants of the study were gathered in their school's assembly hall or a larger classroom designated for the survey, with sitting arrangement spaced by reasonable distance. After obtaining permission from the heads of the participating schools, informed consent of parents/guardians, and the participants' informed consent and assent, we gave each student a packet of the anonymous questionnaire to answer. Averagely, the completion of the questionnaire lasted between 22 and 35 min. Each student placed their answered questionnaire in an opaque box at the exit door.

Data Analysis
We used the Statistical Package for Social Sciences (SPSS version 26.0 for Windows) to analyse the data. Suicidal ideation was the outcome variable in the main statistical analysis, while the exposure variables or correlates comprised personal factors and the variables of social or interpersonal adversity specified. The socio-demographic variables were included in the statistical modelling as covariates. As the loss of cases due to missing data was less than 5%, we used the list-wise deletion of missing data strategy (Graham, 2012). The coding of the variables included in the statistical analysis are shown in the supplementary material (e- Table 1). The data analysis proceeded in three stages; at each stage, we stratified the data by gender (female and male), guided by the aims of the study. Firstly, we performed descriptive analysis of the data by applying frequencies and proportions to assess the distribution of the participants across the socio-demographic variables and the 12-month prevalence estimates of suicidal ideation and suicidal behaviour risk. Secondly, given the categorical nature of the data, we applied the Pearson's Chi-square test (χ2) to explore the bivariate relationships between of 12-month Randomly selected classes in 12 schools that agreed to participate (n = 36) Schools randomly selected, approached and invited to participate (n = 18) Sample of schools that agreed to participate (n = 12) Junior High Schools (JHS) in Ayensuano rural district (n = 84) 6 Schools declined participation, because of end-of-term examinations commitment.
Total eligible students approached and invited to participate (n = 1,214) 21 excluded due to significant incomplete data.
Total eligible students who consented to participate in survey (n = 1,122) Students surveyed (n = 1,101) (Females = 575; Males = 526) Students excluded = 92: 34 Students did not return signed consent. 23 Absentees on day of survey. 9 Students withdrew consent prior to survey. 11 Students were unwell on day of survey. 15 Declined consent by parents/guardians on grounds of iatrogenic effects concerns. Fig. 1 Summary of participant recruitment process for self-report anonymous questionnaire survey suicidal ideation and each of the socio-demographic variables, personal factors and the variables of social adversity specified. We performed point-biserial correlation (r pb ) tests (Prematunga, 2012) to examine the possible bivariate relationship between religious participation and 12-month suicidal ideation. Statistically significant results were determined using the p value less than 0.05 (p < 0.05). Thirdly, we performed a multivariable logistic regression analysis, to examine the possible associations between the binary outcome variable (12-month suicidal ideation: No or Yes) and the specified correlates (personal factors, and variables of social adversity) and covariates (socio-demographic variables). We developed three models, one each for the overall sample, female adolescent sub-sample, and the male adolescent sub-sample. As recommended by leading statistical methodologists, the candidate correlates and covariates were entered in the multivariable logistic regression models regardless of the statistical significance of their bivariate relationships with the outcome variable (Babyak, 2004;Sun, Shook, & Kay, 1996). We reported the results of the logistic regression as odds ratios with 95% confidence intervals (CI) and associated p values.

Ethics
This study was approved by the Department of Psychology Research Ethics Committee, University of Ghana, Accra. We followed the ethical procedures of the Ghana Education Service for conducting research involving pupils in basic educational institutions in Ghana. The Ayensuano District Educational Directorate in Eastern Ghana, and heads of all the participating schools permitted this study. Each participant signed a written consent form prior to responding to the survey; parents/guardians of underage participants provided consent and we obtained the assent of participants aged 10-17 years, prior to taking part in the study. Table 1 shows the socio-demographic characteristics of the participants, stratified by gender. The participants in this study (n = 1101) were aged 10-19 years (mean = 15.3; modal = 16; SD = 1.9); and 575 (52.2%) were females. Most of the participants were older adolescents (53.1%), in JHS 1 (64.3%) and had a caretaker who was employed (89.5%). Most of the participants identified their family as monogamoustheir father had one wife (63.8%). More females (n = 270) than males (n = 199) self-identified as being in a romantic relationship (χ 2

Bivariate Associations
As shown in Table 2, overall, the bivariate analysis showed statistically significant associations between most of the socio-demographic variables, personal factors and social adversities included, but social adversities showed the strongest bivariate associations.
Furthermore, across the overall sample, there was a statistically non-significant negative correlation between religious participation and suicidal ideation during the previous 12 months (r pb = −0.031, n = 1101, p = 0.310). Similarly, when stratified by gender, there was no statistically significant association between religious participation and suicidal ideation among neither female (r pb = −0.043, N = 575, p = 0.300) nor male (r pb = −0.029, N = 526, p = 0.505) adolescents.
It is notable, thus, that across the overall sample and regardless of gender, having been exposed to a friend's suicide attempt appeared to be the strongest and only factor associated with the increased odds of suicidal ideation, while being an older adolescent emerged as the factor with a statistically significant association with reduced odds of suicidal ideation.

Discussion
This study has shown three key findings: 1) approximately 2.5 out of 10 adolescents (2.8 in 10 females; 2.1 in 10 males) reported suicidal ideation during the past 12 months; 2) female adolescents who experienced personal and interpersonal adversities outside the family were more likely to report suicidal ideation, while suicidal ideation among males was strongly associated with conflict with parents; and 3) regardless of gender, rural adolescents who reported being exposed to a friend's suicidal attempt were about three times more likely to report suicidal ideation, compared to those who did not report exposure to a friend's attempted suicide.

Prevalence of Suicidal Ideation
The overall 12-month prevalence estimate of suicidal ideation in the current study (25.1%) is higher, compared to previous estimates from other rural contexts in sub-Saharan Africa (Kinyanda et al., 2011;Nyundo et al., 2020;Rudatsikira et al., 2007). This stark difference in estimates could be due to differences in methods (e.g., sampling and sample size, measures used etc.). As applied in the current study, nearly all the previous studies (Kinyanda et al., 2011;Nyundo et al., 2020;Rudatsikira et al., 2007) used a single-item measure to assess suicidal ideation. However, it must be clarified that the current study used a single-item contained in the multi-item scale of suicidal behaviour (SBQ-R) applied in the study; not a stand-alone single-item assessing suicidal ideation. Recent evidence suggests that the use of single-item measures could lead to misclassification of previous suicidal behaviour (Hom, Joiner Jr, & Bernert, 2016;Millner, Lee, & Nock, 2015). Interestingly, however, the estimates of the current study are comparable to those reported among in-school urban adolescents in Ghana ( As has been observed recently, probably, the precipitating factors of suicidal behaviour among in-school adolescents in Ghana may not be different between urban and rural school contexts; the general school climate (including peer relationships problems, unavailability of peer support, and bullying) could be presenting adolescents with increased risk of suicidal behaviour (Asante et al., 2017;Ohene, Johnson, Atunah-Jay, Owusu, & Borowsky, 2015;Owusu, Hart, Oliver, & Kang, 2011;. The finding that more female than male adolescents reported suicidal ideation is consistent with previous evidence (McKinnon et al., 2016;Page & West, 2011;Uddin et al., 2019), and could be reflective of the general fact that girls are more likely to report internalising (mental) health problems than boys (Van Droogenbroeck, Spruyt, & Keppens, 2018).

Factors Associated with Suicidal Ideation
This study has shown that more female than male adolescents reported personal lifestyle problems and interpersonal adversities mainly outside the family circle; compared with males, females who experienced personal lifestyle problems and interpersonal adversities were more likely to report suicidal ideation. Specifically, relative to boys, girls are more likely to experience the negative effects of polygamous families, including poor schoolwork, corporal punishment, and poverty (Al-Sharfi, Pfeffer, & Miller, 2016); and shortly after a romantic breakup, girls are more likely to experience suicidal ideation (Paul, 2018). Evidence from Ghana suggests that, predominantly, girls (than boys) are more vulnerable to various forms of sexual abuse (Quarshie, Osafo, Akotia, Peprah, & Andoh-Arthur, 2017;Quarshie et al., 2018); thus, it is not surprising that the current study suggests that sexual abuse victimisation is a strong associated factor of suicidal ideation among girls. The evidence that schoolwork problems is associated with suicidal ideation is to be expected, as evidence from Ghana suggests that teachers and parents tend to place high academic performance expectations on their wards (Ansong, Okumu, Bowen, Walker, & Eisensmith, 2017); a situation which often results in test anxietywhich in turn leads to poor exam performance among students. Poor schoolwork tends to increase the risk of self-harm and suicidal behaviour in young people (Evans & Hurrell, 2016). Put together, these findings may be pointing to the general psychosocial β beta value, aOR adjusted odds ratio, CI Confidence Interval, GOF goodness of fit; statistically significant results are in bold face difficulties experienced mostly by girls in low-resource contexts, particularly, rural communities within Africa wherecompared to boys and mengirls and women are at increased vulnerability to social exclusion and discrimination, unsupportive cultural norms, and multiple psychosocial challenges (Kabiru, 2015;Kabiru et al., 2013). The evidence that conflict with parents was strongly associated with suicidal ideation among males was counterintuitive and unexpected. Future studies could explore plausible explanations for this association among boys. However, the extant literature suggests that self-derogation and self-blame mediate the association between parent-adolescent conflict and adolescent suicidal ideation (Shagle & Barber, 1993). In this context, adolescent suicidal ideation may serve the purpose of escaping or avoiding the self-derogation and selfblame experience resulting from the conflict with parents (Shagle & Barber, 1993). Perhaps, future studies using robust designsincluding carefully planned qualitative approachescould explore this possible explanation of suicidal ideation as a 'coping mechanism' resulting from parent-adolescent conflict.
Consistent with the global literature, regardless of gender, rural adolescents in the current study who reported being exposed to a friend's suicide attempt were about three times more likely to report suicidal ideation (Hill et al., 2020;Insel & Gould, 2008;Jarvi, Jackson, Swenson, & Crawford, 2013). Although a worrying situation, the direct exposure to or witnessing or having knowledge of a friend's attempt at suicide to manage distress could serve as a model for adolescent peers experiencing challenging circumstances.
The finding that older adolescence is associated with reduced odds of suicidal ideation could be consistent with the evidence of longitudinal studies that self-harm and suicidal behaviour peak in adolescence, around mid-ages 15-16 years, and reduce towards late-ages 18-19 years to early adulthood (Plener, Schumacher, Munz, & Groschwitz, 2015). The natural challenges of interpersonal difficulties and increased risky health related behavioursincluding alcohol and substance misuseduring early adolescence could be responsible for the increased odds of suicidal ideation among younger adolescents (Hawton et al., 2012;Viner et al., 2012).
An unexpected finding of the present study is the significant association of alcohol with suicidal ideation among females (the association did not reach the desired statistical threshold among males), although more males (23.8%) than females (21.4%) reported alcohol use. This is surprising because previous studies from Ghana have consistently shown higher estimates of alcohol use among adolescent males than females (Adu-Mireku, 2003;Asante & Kugbey, 2019). Considering that this is the first study from rural Ghana, further studies are needed to clarify this association. While it is possible that the association between alcohol use and suicidal ideation may be stronger among female than male adolescents in rural Ghana, the weaker association observed among males in this study could be due to underpowered male sample size in the analysis. Beyond these possibilities, the evidence underscores the need for government to strengthen the enforcement of the country's alcohol policy to prevent access and sale of alcoholic beverages to underage persons (Ministry of Health Ghana, 2016). A recent systematic review suggests that strengthening policies targeted at reducing harmful alcohol consumption could contribute to reducing suicidal behaviours (Kõlves et al., 2020).
Clearly, several unexpected findings of the current study will benefit from future research towards expanding our understanding of suicidal behaviours among rural-dwelling adolescents in Ghana. For example, the global and regional literature is replete with evidence to suggest that bullying victimisation, parental divorce, and truancy are strongly associated with escalated risk of suicidal behaviours among school-going young people (Asante et al., 2017;Evans, Hawton, & Rodham, 2004;Fuller-Thomson & Dalton, 2011;Wagner, Silverman, & Martin, 2003;Wilson, Dunlavy, Viswanathan, & Bovet, 2012); however, the current study found no statistically significant associations between these factors and suicidal ideation. While these factors exist in families, schools, and among students in Ghana, we strongly suspect that the variables were underpowered in the analysis of our study.

Implications and Recommendations
The evidence of this study supports the socio-ecological framework of suicidal behaviour among adolescents, due to the interrelated and multi-level nature of the associated factors (Ayyash-Abdo, 2002;Perkins & Hartless, 2002;Shagle & Barber, 1995;WHO, 2014). This implies that intervention, prevention and health promotion programmes should be designed and approached holisticallyto include the individual adolescent, their school, family, and other contexts outside the family and school (Perkins & Hartless, 2002;WHO, 2014). At the population level, this study supports recommendations by recent studies that the Ministry of Education and the Ghana Education Service should consider developing and adopting an adolescent mental health policy, and assessment and management protocols to support students who may be (at risk of) engaging in suicidal behaviour (Asante et al., 2017;Baiden et al., 2019;Quarshie, Waterman, & House, 2020a, b). At the local and individual level, social skills, stress management and problem-solving skills programmes for in-school adolescents could be helpful in encouraging help-seeking and preventing suicidal behaviour among adolescents. Rural schools, local communities, parent-teacher groups, and families could consider working together towards providing social support to adolescents, encouraging parent-child dialogue to resolve family conflict, and clamping down on sexual abuse of children and adolescents. Potentially, these school-community-family collaborations could culminate in supportive family climate, healthpromoting school environment, and enhanced adolescent (mental) wellbeing.

Strengths and Limitations
This study is partly in response to the call by the most recent systematic reviews for expansive studies on self-harm and suicidal behaviour among young people in sub-Saharan Africa (Aggarwal et al., 2017;Lim et al., 2019;Quarshie, Waterman, & House, 2020c). This study represents the first attempt from Ghana to provide systematic evidence on the prevalence of suicidal ideation and some personal and interpersonal correlates among rural in-school adolescents. Nonetheless, some plausible limitations of the study are noteworthy. Given that suicidal behaviour is criminalised and stigmatised in Ghana (Osafo, Akotia, Andoh-Arthur, & Quarshie, 2015), it is possible that our participants might have provided socially desirable and guarded responses to questions related to suicidal thoughts and acts. However, we believe that social desirability bias might be minimal in this study, as participants were allowed enough privacy in responding to the surveywe used an anonymous selfreport questionnaire, participants sat far apart from one another, and teachers were kept in the background, during the survey. It is also notable that the cross-sectional design used does not allow for causal interpretation of our findings.
Furthermore, our study failed to include absentees and out-of-school rural-dwelling adolescents. This implies that whereas our findings may be applied to non-clinical inschool adolescents in rural Eastern Ghana (and, possibly, other rural contexts of the country), it may not be necessarily generalisable across their peers who are out-ofschool. Our study also failed to consider some known potential covariates of suicidal ideation among adolescents, for example, personality, depression, anxiety, among other personal, family, and locality factors (Kinyanda et al., 2011;Nyundo et al., 2020;Rudatsikira et al., 2007;Shilubane et al., 2014). Also, longitudinal studies using robust designs are needed to understand the risks and protective factors, and patterns of suicidal ideation across adolescence in rural Ghana. Lastly, in keeping with recent recommendations regarding adolescent suicidal behaviour research in sub-Saharan Africa (Quarshie, Waterman, & House, 2020a, c), future studies could employ qualitative methods to explore the individualised, gender-specific, and shared meanings and motivations of rural adolescents who report suicidal ideationsthe findings may be informative for (targeted) intervention and prevention programmes (White, Marsh, Kral, & Morris, 2016).

Conclusion
Overall, 25.1% participants (95% CI = 22.5-27.7), representing 28.3% females (95% CI = 24.7-32.2) and 21.5% males (95% CI = 18.0-25.2) reported suicidal ideation during the previous 12 months. The prevalence of suicidal ideation among adolescents in rural Ghana compares with estimates among school-going adolescents from other countries within sub-Saharan Africa, but also underscores the need for targeted and universal prevention programmes and intervention efforts to mitigate the potential transition from suicidal ideation to suicidal attempt and eventual suicide among adolescents in rural Ghana.
Acknowledgements We wish to thank the Ayensuano District Educational Directorate and heads of all the participating schools who granted permissions for this study. We also thank parents and guardians who consented for their underage wards to participate in this study. Finally, we thank all the students who participated in this study to contribute data about their suicidal behaviour for this study.
Author Contributions ENBQ and SKO contributed to the study concept and design; ENBQ performed statistical analysis of the data; SKO and ENBQ drafted the manuscript and critiqued the manuscript for important intellectual content. All the authors contributed to the interpretation of results and revision of the manuscript and approved the final version. ENBQ serves as guarantor for the contents of this paper.

Data Availability
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations
Conflict of Interest The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.