Childhood factors associated with suicidal ideation among South African youth: A 28-year longitudinal study of the Birth to Twenty Plus cohort.

BACKGROUND: Although early life factors are associated with increased suicide risk in youth, there is a dearth of research on these associations for individuals growing up in disadvantaged socioeconomic contexts, particularly in low- and middle-income countries (LMICs). We documented the association between individual, familial, and environmental factors in childhood with suicidal ideation among South African youth. METHODS AND
FINDINGS: We used data from 2,020 participants in the Birth to Twenty Plus (Bt20+) study, a South African cohort following children born in Soweto, Johannesburg from birth (1990) to age 28 years (2018). Suicidal ideation was self-reported at ages 14, 17, 22, and 28 years, and the primary outcome of interest was suicidal ideation reported at any age. We assessed individual, familial, and socioeconomic characteristics at childbirth and during infancy, adverse childhood experiences (ACEs) between ages 5 and 13 years, and externalizing and internalizing problems between 5 and 10 years. We estimated odds ratios (ORs) of suicidal ideation for individuals exposed to selected childhood factors using logistic regression. Lifetime suicidal ideation was reported by 469 (23.2%) participants, with a 1.7:1 female/male ratio. Suicidal ideation rates peaked at age 17 and decreased thereafter. Socioeconomic adversity, low birth weight, higher birth order (i.e., increase in the order of birth in the family: first, second, third, fourth, or later born child), ACEs, and childhood externalizing problems were associated with suicidal ideation, differently patterned among males and females. Socioeconomic adversity (OR 1.13, CI 1.01 to 1.27, P = 0.031) was significantly associated with suicidal ideation among males only, while birth weight (OR 1.20, CI 1.02 to 1.41, P = 0.03), ACEs (OR 1.11, CI 1.01 to 1.21, P = 0.030), and higher birth order (OR 1.15, CI 1.07 to 1.243, P < 0.001) were significantly associated with suicidal ideation among females only. Externalizing problems in childhood were significantly associated with suicidal ideation among both males (OR 1.23, 1.08 to 1.40, P = 0.002) and females (OR 1.16, CI 1.03 to 1.30, P = 0.011). Main limitations of the study are the high attrition rate (62% of the original sample was included in this analysis) and the heterogeneity in the measurements of suicidal ideation.
CONCLUSIONS: In this study from South Africa, we observed that early life social and environmental adversities as well as childhood externalizing problems are associated with increased risk of suicidal ideation during adolescence and early adulthood.

Introduction

Suicide is an important cause of mortality worldwide, accounting for 800,000 deaths each year [1]. Among youth aged 15 to 25 years, suicide ranks as the second or third most common cause of death in most countries [2]. Suicidal ideation—the consideration of or desire to end one’s own life [2]—is a strong predictor of subsequent suicidal acts [3]. In the United States, 1 in 5 youth experiencing suicidal ideation transition to elaborate concrete suicidal plans within a year, and 60% of those having such plans attempt suicide in this time frame [3]. These data suggest that preventing suicidal ideation is critical to reduce suicide risk. The burden of youth suicide is disproportionally high in low- and middle-income countries (LMICs), where nearly 90% of the world’s youth live [4] and 78% of all suicides occur [5]. In particular, young people in LMICs in the African region have the highest prevalence of suicidal ideation, with 1 in 5 adolescents reporting seriously considering suicide in the past 12 months [6]. Evidence suggests that these high rates are likely to be underestimated [7]. Despite this elevated prevalence, research on suicide prevention in LMICs accounts for only a small fraction of the available evidence [8]. This strongly limits our understanding of risk factors for suicide-related outcomes (i.e., suicidal ideation, suicide attempt, and death by suicide) and precludes the development of tailored public health suicide prevention strategies in LMIC contexts.

Beyond the contribution of childhood externalizing problems (i.e., behaviors such as conduct disorder and hyperactivity/impulsivity that are overt and can result in conflict with others) [9] and internalizing problems (i.e., emotional symptoms such as anxiety or depression which reflect internal distress) [9-15], longitudinal studies conducted in high-income countries (HICs) have shown that early life factors—including perinatal and childhood factors—play an important role in increasing vulnerability to suicidal ideation and suicide attempts during the life course [15,16]. For example, a recent meta-analysis found that exposure to socioeconomic adversity (indexed as low socioeconomic status, low maternal age, or low parental education at childbirth) is associated with increased suicide risk later in life [16]. Similarly, early growth deficits as indexed by low birth weight were associated with increased suicide risk in prior HIC studies [16,17]. Adverse childhood experiences (ACEs), such as exposure to violence and abuse/neglect [18-20], family difficulties (e.g., early separation and single-parent families) [18], and parental substance use problems [21], have all consistently been associated with mental health problems and suicide risk in HIC contexts [22].

However, it is unknown if and how such childhood factors are associated with suicide risk in LMIC contexts, which are markedly different from HIC contexts in a number of ways. First, the population prevalence of most childhood risk factors—including socioeconomic adversity and low birth weight—in LMICs is much higher than in HICs [23]. Second, several studies have shown that exposure to ACEs is more common in LMICs than in HICs. For example, in South Africa and Brazil, more than 85% of children reported at least 1 ACE [24,25], compared to 46% of children in the US and the United Kingdom [26]. Third, externalizing and internalizing problems may be differently interpreted within each socioeconomic context. For example, in social environments characterized by violence, displaying aggressive behaviors may be considered normative or a demonstration of invulnerability to potential threats [27]. Finally, cultural norms may differentially impact how environmental factors influence mental health and suicide risk specifically. For example, although the consistent findings of sex differences in the prevalence of suicidal ideation and attempt (higher in females) and death by suicide (higher in males) in HICs have been replicated in LMICs, including South Africa, country-specific sociocultural gender norms—particularly those related to masculinity—may explain these differences in LMIC contexts [6,28-30].

The objective of this study was to investigate childhood risk factors for suicidal ideation in adolescence and young adulthood using data from the largest and longest-running birth cohort in Africa, the Birth to Twenty Plus (Bt20+) study in South Africa. Following Turecki and Brent’s developmental model of suicide risk [1], a wide range of potential childhood risk factors were investigated, including early life socioeconomic adversity, ACEs, and children’s externalizing and internalizing problems. These factors—considered in the model as predisposing and developmental, as opposed to precipitating, factors—were hypothesized to increase vulnerability to suicide during the life course and may be targeted by population-based suicide prevention strategies. Furthermore, we aimed to systematically document sex differences in these associations, given the known differences in the prevalence of suicide-related outcomes [30-32].

Methods

Study participants

This study used data from the Bt20+ cohort, a population-based longitudinal study that followed-up children from birth to adulthood [33]. The initial sample included 3,273 mothers and their singleton children born during a 7-week period in 1990 in Soweto (a historically informal settlement in Johannesburg), South Africa. Mothers were recruited from public antenatal clinics in the area in late 1989, when researchers began interviewing women who were predicted to deliver their babies during the study’s enrolment period. The aim of the Bt20+ cohort was to describe the effects of rapid urbanization on the physical and psychosocial development of children during a period of dramatic political and social change in South Africa, when violence and social disorder peaked. The study is still ongoing, and the last data collection was performed in 2018 when participants were 28 years of age. In over 22 waves of data collection, the investigators assessed social and economic circumstances, family relationships, children’s growth and health, schooling and employment, and mental health (). To overcome potential language problems, data collection was performed in the participants’ language of choice (isiZulu, Sesotho, or English); consensual agreement on the phrasing of questions in the different languages was reached for each item when instrument validation was performed [34]. For this study, we analyzed data from a sample of 2,020 participants with at least 1 measure of suicidal ideation at ages 14, 17, 22, or 28 years (of the initial 3,273 participants data on suicidal ideation were not available for 1,253 participants). This analytical sample differed from the original cohort on a number of variables, including maternal age and schooling, household crowding, and assets (). Inverse probability weighting was therefore used in all analyses to address biases due to differential attrition. Weights were derived from the independent baseline characteristics (maternal age, maternal schooling, household crowding, and assets) predicting inclusion in the sample using a logistic regression model’s individual predicted probabilities [35]. Ethical approval for the Bt20+ study was obtained from the Committee for Research on Human Subjects at the University of Witwatersrand, South Africa. Ethical clearance for the use of secondary data is only applicable where use of the data is not covered by primary data collection (the Bt20+ study) ethics approval. As this study is based on secondary data analysis within the parameters of the primary data collection’s ethics clearance, no further ethical approval is necessary. All participants gave written informed consent at each data collection wave. Written consent was obtained by parents or guardians when children were minor, with verbal assent from children. From ages 16 to 28, Bt20+ participants provided written consent for each individual data collection wave.

Assessment of suicidal ideation

Suicidal ideation was assessed at ages 14, 17, 22, and 28 years using single self-reported questions. At age 14 the item “I think about killing myself” was included in the Youth Self Report questionnaire [36]. Adolescents answered in reference to the previous 6 months using a 4-point scale (not true, sometimes true, true, and very true). These responses were dichotomized into yes (sometimes to very true) and no (not true). At ages 17 and 22, the question was asked “Have you recently found that the ideas of taking your own life kept coming into your mind?” from the General Health Questionnaire [37]. Participants answered using a 4-point scale (definitely not, I don’t think so, has crossed my mind, and definitely yes), also dichotomized into yes (has crossed my mind and definitely yes) and no (otherwise). At age 28, the question “Has the thought of ending your life been on your mind?” from WHO’s Self Reporting Questionnaire was asked [38]. Responses were yes or no in reference to the past 30 days. Our outcome was lifetime suicidal ideation, defined as having reported suicidal ideation at any time versus never. Follow-up intervention of care for participants was conducted at 2 points. Firstly, fieldworkers indicated if a participant appeared to be under psychological distress in study notes immediately after administration of the questionnaire, which were reviewed at the end of each day of data collection. Secondly, during data entry, a built-in calculation on items probing for “suicide ideation in the past month” gave a positive for a “mental health referral needed” variable. Participants expressing distress were accordingly put in contact with social and health services linked to the study.

Assessment of childhood risk factors

The following childhood risk factors were investigated (see also for the items used to assess these variables).

Child characteristics

Child sex (male/female), birth weight (considered as a continuous variable measured in kg following a peer-reviewer’s remark; previously dichotomized into low, <2.5 kg versus nonlow, ≥2.5 kg), and birth order (first, second, third, fourth, or later born) were reported by the mother at enrollment into the study.

Sociodemographic and maternal characteristics

Parity (i.e., total number of pregnancies); any previous abortion (spontaneous or induced); and maternal postnatal depression were assessed when the child was 6 months of age using the Pitt Inventory [39], a validated measure consisting of 24 items (Cronbach’s alpha, 0.85) assessing current feelings and changes in mood answered on a 3-point scale (yes, no, and I don’t know). The inventory has been previously used in South Africa and shows good correlation with the Edinburgh Postnatal Depression Scale [40,41]. An index of socioeconomic adversity in the early life environment was created from the following variables (all reported at the time of childbirth): maternal age at childbirth; number of assets, a proxy of wealth, derived from a site-specific list (TV, fridge, car, washing machine, and phone) according to the methodology of Filmer and Pritchett [42]; maternal education, measured as years of completed schooling; and household crowding, measured as the number of people per room living in the same household. The index was derived as the sum of poverty (3 people per room) as previously described [43] and was standardized with a mean of 0 and standard deviation (SD) of 1.

Childhood externalizing and internalizing problems

The South African Child Assessment Schedule (SACAS) was used to ascertain child externalizing problems at ages 5, 7, and 10 years from maternal reports. The SACAS is a questionnaire based on the Child Behaviour Checklist [44] and was adapted idiomatically to the South African culture and translated into isiZulu, Sotho, and Afrikaans [40]. Externalizing problems were assessed with 35 items from the SACAS describing aggressive and rule-breaking behaviors (e.g., “Is he/she disobedient at school?” and “Does he/she physically attack people?”). Internalizing problems were assessed with 32 items from the SACAS describing anxiety and depressive symptoms (e.g., “Is he/she sad or depressed?” and “Is he/she too fearful or anxious?”). We averaged the 5-, 7-, and 10-year assessments to create the final scores that were standardized. Reliability of the SACAS has been demonstrated and validity was established in a clinical group that was shown to have significantly higher mean scores compared to a nonclinical group of children on all scales [45]. In our sample, both externalizing and internalizing problems measures showed good reliability (Cronbach’s alpha were 0.80 to 0.85 and 0.76 to 0.80, respectively).

Adverse childhood experiences

Mothers (at child ages 5, 7, and 11) and children (at ages 11 and 13) were asked about several ACEs including experiences of material deprivation (i.e., chronic unemployment, legal problems, and chronic poverty), loss (i.e., chronic illness, disability, or death of a family member), negative family dynamics (i.e., parental substance abuse, divorce, intimate partner violence, lack of cohesion, and child separation), and child abuse and violence (i.e., child sexual and physical abuse and exposure to violence) [24]. The overall level of exposure to ACEs was computed by summing the number of reported ACEs (by either mothers or participants). This follows the cumulative risk model of ACEs, which states that it is the accumulation of adverse events, rather than the exposure to specific events, which is detrimental to health [46]. The final score was then standardized (z-score transformed).

Data analysis

The analysis protocol was decided upon during study group meetings that took place between February and September 2020. However, there was no formal prospective analysis plan for the study. Changes to the initial planned analyses implemented following peer review were documented in the Methods section. We described continuous and categorical variables using means and SDs and counts and percentages, respectively, and used binary logistic regression to estimate the univariable associations between each childhood factor and suicidal ideation. We systematically tested the interaction between each factor and child sex and report analyses separately for males and females. Then, variables were jointly entered in a multivariable logistic regression model to estimate their independent associations with suicidal ideation, avoiding the use of collinear variables. This approach to the multivariable modeling was implemented following the comment of a reviewer. Initially, only variables that showed evidence of an association (P < 0.05) in the univariable analysis were used in the multivariable model. The 2 methods led to consistent results. To account for missing data in childhood factors, we estimated our model using multiple imputations: 50 imputed datasets were generated using the Amelia II package in R, which relies on a bootstrap expectation–maximization algorithm to impute missing multivariate data [47]. Models were then estimated across all imputed datasets and results pooled. The amount of missing data is shown in . Analyses were performed in R version 3.6, and the statistical significance level used was P < 0.05, 2 sided. This study is reported as per the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline (S1 Checklist).

Results

Among the 2,020 participants in this study, 1,042 (51.6%) were female, and 978 (48.4%) were male (). Approximately 11% were born of low birth weight, and 38.1% were firstborn children. Concerning mothers, 7.4% were 17 years old or younger at childbirth, and 44.6% reported living in overcrowded households.

In our sample, 469 (23.2%) participants reported suicidal ideation between the ages of 14 and 28 years. Females (lifetime estimate, 29.6%) were more likely to report suicidal ideation than males (lifetime estimate, 16.5%), with a female/male ratio of 1.7:1. As shown in and , the prevalence of suicidal ideation increased sharply from 14 to 17 years (7.8% to 15.1%) and subsequently decreased (10.4% and 6.9% at 22 and 28 years, respectively). Although suicidal ideation was always higher in females than males, the sex gap decreased over time, and male–female rates were similar at age 28 ().

Prevalence of suicidal ideation in the Bt20+ cohort.

The figure illustrates the prevalence of suicidal ideation at each assessment occasion and their change over time. Percentages are relative to the individuals in the cohort for which suicidal ideation data were available at the given age, namely 1,209 (632 females and 577 males), 1,532 (814 females and 718 males), 1,582 (821 females and 761 males), and 1,388 (728 females and 660 males) participants at ages 14, 17, 22, and 28 years, respectively. Vertical bars represent 95% CIs around the point prevalence. Bt20+, Birth to Twenty Plus.

Female-to-male RR for suicidal ideation at each age.

The figure represents the comparisons of suicidal ideation rates in females versus males as risk ratios (RR) at each age. RR were calculated by dividing the prevalence rate in females by the prevalence rate in males, e.g., for age 14 years: (71/632)/(23/577) = 2.81. Vertical bars represent 95% CIs of the RR. The horizontal dashed line represents the null (RR = 1), and CI bars indicates that the RR is not statistically significant at P < 0.05, except at age 28 years. RR, risk ratio.

reports the univariable associations between childhood factors and suicidal ideation. In the overall sample, female children (OR 2.09, CI 1.84 to 2.38, P < 0.001), those born of lower birth weight (OR for each kg lower birth weight 1.25, CI 1.11 to 1.41, P < 0.001), and those of higher birth order (i.e., second, third, and fourth+ born; OR for trend 1.08, CI 1.02 to 1.14, P = 0.007) were more likely to report suicidal ideation. Indicators of socioeconomic adversity at birth were not statistically significantly associated with suicidal ideation in the whole sample but showed evidence of a statistically significant association uniquely among males (OR 1.17, CI 1.05 to 1.29 for males; OR 0.98, CI 0.90 to 1.06 for females for each SD increase in socioeconomic adversity index; P = 0.009). This was especially the case for poverty (OR for males: 1.25, CI 1.02 to 1.52, P = 0.031; OR for females: 0.89, CI 0.76 to 1.04, P = 0.152; P = 0.010). Both externalizing (OR 1.15, CI 1.06 to 1.25, P < 0.001) and internalizing (OR 1.12, CI 1.04 to 1.21, P = 0.003) problems in childhood were significantly associated with increasing odds of suicidal ideation. In subgroup analyses, childhood externalizing problems were significantly associated with an increased likelihood of reporting suicidal ideation among both males (OR 1.28, CI 1.13 to 1.45, P < 0.001 for 1 SD increase in externalizing problems score) and females (OR 1.19, CI 1.06 to 1.34, P = 0.001; P = 0.380), while internalizing problems were significantly associated with suicidal ideation among males only (OR for males: 1.19, CI 1.05 to 1.34; P = 0.003; OR for females: 1.05, CI 0.95 to 1.16, P = 0.328), although the interaction did not reach statistical significance (P = 0.118). Finally, increased exposure to ACEs in childhood was significantly associated with higher odds of subsequently reporting suicidal ideation (OR 1.10, CI 1.03 to 1.18, P = 0.003 for 1 SD increase in ACEs score) and stratified analyses by sex suggested that the association was stronger among females (OR 1.14, CI 1.05 to 1.24, P = 0.003) than males (OR 1.06, CI 0.95 to 1.18, P = 0.299), although the interaction was not statistically significant (P = 0.308).

Multivariable analyses () showed that socioeconomic adversity (OR 1.13, CI 1.02 to 1.25, P = 0.031) and externalizing problems (OR 1.22, CI 1.08 to 1.39, P = 0.002) were independently associated with an increased likelihood of reporting suicidal ideation among males, while internalizing problems only approached statistical significance (OR 1.12, CI 0.99 to 1.27, P = 0.083). Among females, independent factors significantly associated with suicidal ideation were lower birth weight (OR 1.20, CI 1.02 to 1.41, P = 0.030), ACEs (OR 1.11, CI 1.01 to 1.21, P = 0.030), higher birth order (for each birth order increase, OR 1.14, CI 1.06 to 1.23, P < 0.001), and externalizing problems (OR 1.16, CI 1.03 to 1.31, P = 0.011). Multivariable analysis including only factors associated with suicidal ideation at P < 0.05 in the univariable analyses yielded consistent results ().

Discussion

This study described the prevalence of suicidal ideation from ages 14 to 28 and identified key socioeconomic and individual-level childhood factors associated with suicidal ideation among South African youth. We found that, overall, 23.2% of participants reported suicidal ideation, with a peak in prevalence at age 17 years and an overall 1.7:1 female/male ratio. Females reported significantly higher rates than males at all time points except for age 28 when rates for males and females were similar. We found that males exposed to socioeconomic adversity and those who experienced externalizing problems in childhood were more likely to consider ending their lives during their teenage years and early adulthood. Among females, childhood factors associated with suicidal ideation included lower birth weight, ACEs, higher birth order, and externalizing problems.

Added knowledge to existing research

To the best of our knowledge, this is the first study to prospectively examine the prevalence and childhood risk factors of suicidal ideation across adolescence and young adulthood among youth in sub-Saharan Africa. As most previous longitudinal studies on the topic are based on samples from HICs and countries outside the African continent, our findings add to the current literature about the prevalence and etiology of suicidal ideation in LMICs in Africa that can be used by policymakers to elaborate local suicide prevention strategies.

Comparison of the findings with previous research

Our findings of the prevalence of suicidal ideation, the sex distribution, and the peak at age 17 years are in line with previous LMIC and South African prevalence studies that have reported increased rates of suicidal ideation in later adolescence (17.8%; 15 to 17 years) compared to earlier adolescence (15.9%; 13 to 14 years) [6,48] and higher rates in females (8.5%; 10 to 18 years) compared to males (5.6%; 10 to 18 years) [29,48,49]. Consistently, previous meta-analyses of LMIC studies reported a higher prevalence of suicidal ideation, planning, and attempt among female youth, with those in the African region reporting the highest rates [6,31]. Our findings on the risk factors for suicidal ideation are also consistent with prior cross-sectional studies in Africa. For example, a previous investigation among young South African men also found an association between poverty and past month suicidal ideation [50]. In another South African study, ACEs measured at baseline (10 to 18 years) were significantly associated with suicidal ideation among adolescents 1 year later [29]. This study also reported a higher prevalence of suicidal ideation among females compared to males but did not examine whether the association between ACEs and suicidal ideation differed by sex. These findings are also consistent with those from studies conducted in HICs that reported associations between poverty and ACEs in childhood and suicide-related outcomes over the life span [51,52], suggesting that such factors are deleterious for human development and mental health irrespective of cultural and socioeconomic variations. However, studies examining sex differences in the association between childhood factors and suicidal outcomes have found a different pattern of results from ours. For example, a recent meta-analysis of longitudinal studies (mostly from HICs) reported no sex differences in the association of ACEs and internalizing problems with suicide attempt and death by suicide, but found that externalizing problems were only associated with suicide attempt in males [32]. Our finding of an association between externalizing problems and suicidal ideation among both males and females could be explained by previous findings of a higher prevalence of childhood externalizing problems in LMICs compared to HICs among both males and females [27,53]. Importantly, while internalizing problems are often associated with suicidal ideation in HICs [3,54], they were not independent predictors in our study. This is consistent with previous studies in HICs that have shown that externalizing behavior and comorbid internalizing–externalizing behaviors are more strongly associated with suicide-related outcomes than internalizing behavior alone [12,14,15]. While other studies found independent associations of internalizing behavior with suicide-related outcomes, these studies often measured internalizing behavior in adolescence rather than in childhood [54,55]. However, it is also important to consider that the lack of association for internalizing problems may be due to the developmental age in which internalizing problems have been measured, since some previous studies in HICs suggested associations between internalizing problems measured in adolescence and suicide-related outcomes. Finally, different from studies in HICs [56,57], we did not find an association between maternal postnatal depression and offspring suicidal ideation. However, previous Bt20+ studies have found that maternal postnatal depression is associated with offspring internalizing problems in both childhood [40] and early adulthood [58]. Therefore, the absence of associations with suicidal ideation in the present study may indicate that, although maternal depression increases risk for offspring internalizing problems, it is not sufficient to differentiate participants who consider suicide from those who do not in the specific socioeconomic context of South Africa. However, maternal depression had a high rate of missing data in our sample (). Therefore, although maternal depression was not associated with attrition, caution should be used to interpret this lack of association, as the mothers with the highest depressive symptoms may have dropped out from the study.

Strengths and limitations

A key strength of this paper is the use of longitudinal data from the longest birth cohort in Africa to prospectively examine the association of the childhood environment with suicidal ideation in youth. Assessments relied on validated measures and care was taken to adapt them to the local setting. Additionally, the majority of childhood factors were reported by mothers while suicidal ideation was self-reported by offspring, thus reducing bias due to shared method variance. The longitudinal nature of this study also enabled us to assess risk factors multiple times throughout participants’ childhood. In addition to these strengths, certain limitations of the study should also be acknowledged. Attrition was substantial, although comparable with other longitudinal cohorts [59]. We used inverse probability weighting to address differential attrition, which may partially account for related biases. Furthermore, due to differential attrition [33,60], the majority of our sample was Black South African, the overwhelmingly largest group in the country. There is some evidence suggesting that prevalence rates of death by suicide and suicidal ideation are higher in population groups not represented in our sample [49]; thus, our results may not be generalizable to South African youth from minority population groups. Another limitation was the use of 3 different questionnaires to assess suicidal ideation. Although the items used assessed the same construct, broadly defined as having thought about ending one’s life and was appropriate to the life stage of the sample, measurement differences may have introduced bias. This limitation is due to the fact that the Bt20+ cohort was not initially designed to study specific suicide-related outcomes. However, it worth noting that (1) this study considered “any suicidal ideation” as an outcome rather than a measure of its intensity, thus reducing bias arising from using different instruments; and (2) bias would lead to underestimation of our associations, leading to conservative results. Moreover, although suicidal ideation is a key aspect of suicide risk, this study did not measure other important aspects such as suicide attempt and transition from ideation to attempt. Another limitation is that some of the childhood factors were only measured once, and our analysis did not take into account their time-varying nature (e.g., socioeconomic adversity). Furthermore, only parent reports of child behavior were available in early and middle childhood. This may have introduced measurement errors, especially for internalizing behavior measures that can be affected by maternal state of mind and attitudes. Finally, the childhood factors investigated in our study reflect a conceptual model that mostly emphasizes risk factors rather than protective and mitigating factors. To enhance suicide prevention, further research is needed to understand the factors decreasing and buffering suicide risk, especially for children exposed to socioeconomic adversity and those exhibiting behavioral problems.

Implications and next steps for research, clinical practice, and public policy

Given that the African region bears the heaviest burden of suicide-related outcomes among youth, further studies across the continent are needed to understand how country-specific factors impact the association between the childhood environment and suicide-related outcomes and to explore sex differences in these associations to inform prevention. Further qualitative studies are also needed to complement epidemiological studies by unpacking how childhood risk factors and sociodemographic characteristics are associated with suicidal ideation in the sociocultural contexts of countries in the African region [61]. For example, the handful of qualitative studies conducted in South Africa [30,61-65] highlight societal expectations of masculinity (e.g., escaping from situations in which they were unable to live up to traditional understandings of masculinity) and protest against masculine dominance (e.g., for females, experiencing intimate partner violence) as important factors for suicidal behavior. These factors play a role in the sociocultural context within which suicidal behaviors occur and how context can inform interventions [66]. Our findings also emphasize the need to consider how individual and social factors interact in increasing risk of suicide-related outcomes. Prior studies point to the high rates of poverty and unemployment in South Africa and their relation to a breakdown in family life, which can result in a number of social problems including child abuse and neglect [65]. Inability to cope with such adverse socioeconomic experiences could result in psychological distress, which is an important risk factor for suicide.

Conclusions

In this longitudinal study on the prevalence and etiology of suicidal ideation in South African youth, we found that prevalence rates peaked at age 17 and decreased continuously until age 28. Prevalence rates were higher among females than males, and we found sex differences in the association of childhood individual, familial, and environmental factors with youth suicidal ideation. As these factors (e.g., externalizing problems, socioeconomic adversity, and ACEs) are highly prevalent in South Africa, our findings support the need for a population-based approach to suicide prevention aiming at reducing the pervasiveness of childhood adversity and increasing societal well-being. Given the high burden of suicide-related outcomes in LMICs, especially in the African region, further quantitative and qualitative research is needed to understand how country-specific factors influence suicide-related outcomes. Such research can inform the development of interventions to reduce the burden of youth suicide.

Disclaimers

Opinions expressed and conclusions arrived at are those of the authors and are not to be attributed to the CoE in Human Development.

STROBE Checklist.

STROBE, Strengthening the Reporting of Observational Studies in Epidemiology.

(DOCX)

Click here for additional data file.

Questionnaires.

(PDF)

Click here for additional data file.

Comparison of the characteristics of participants included and not included in the analysis sample.

(DOCX)

Click here for additional data file.

Count and proportion of missing data in the analysis variable.

(DOCX)

Click here for additional data file.

Multivariable associations between childhood factors and suicidal ideation using factors associated with suicidal ideation in the univariable analysis at P < 0.05.

(DOCX)

Click here for additional data file.

Representation of the Bt20+ cohort assessments used in this investigation.

Bt20+, Birth to Twenty Plus.

(DOCX)

Click here for additional data file.

12 Aug 2021

Dear Dr Orri,

Thank you for submitting your manuscript entitled "Childhood factors associated with suicidal ideation among South African youth A 28-year longitudinal study using the Birth to Twenty Plus cohort" for consideration by PLOS Medicine.

Your manuscript has now been evaluated by the PLOS Medicine editorial staff as well and I am writing to let you know that we would like to send your submission out for external peer review.

However, before we can send your manuscript to reviewers, we need you to complete your submission by providing the metadata that is required for full assessment. To this end, please login to Editorial Manager where you will find the paper in the 'Submissions Needing Revisions' folder on your homepage. Please click 'Revise Submission' from the Action Links and complete all additional questions in the submission questionnaire.

Please re-submit your manuscript within two working days, i.e. by Aug 16 2021 11:59PM.

Login to Editorial Manager here: https://www.editorialmanager.com/pmedicine

Once your full submission is complete, your paper will undergo a series of checks in preparation for peer review. Once your manuscript has passed all checks it will be sent out for review.

Feel free to email us at plosmedicine@plos.org if you have any queries relating to your submission.

Kind regards,

Caitlin Moyer, Ph.D.

Associate Editor

PLOS Medicine

30 Nov 2021

Dear Dr. Orri,

Thank you very much for submitting your manuscript "Childhood factors associated with suicidal ideation among South African youth A 28-year longitudinal study using the Birth to Twenty Plus cohort" (PMEDICINE-D-21-03460R1) for consideration at PLOS Medicine.

Your paper was evaluated by a senior editor and discussed among all the editors here. It was also discussed with an academic editor with relevant expertise, and sent to four independent reviewers, including a statistical reviewer. The reviews are appended at the bottom of this email and any accompanying reviewer attachments can be seen via the link below:

[LINK]

In light of these reviews, I am afraid that we will not be able to accept the manuscript for publication in the journal in its current form, but we would like to consider a revised version that addresses the reviewers' and editors' comments. Obviously we cannot make any decision about publication until we have seen the revised manuscript and your response, and we plan to seek re-review by one or more of the reviewers.

In revising the manuscript for further consideration, your revisions should address the specific points made by each reviewer and the editors. Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments, the changes you have made in the manuscript, and include either an excerpt of the revised text or the location (eg: page and line number) where each change can be found. Please submit a clean version of the paper as the main article file; a version with changes marked should be uploaded as a marked up manuscript.

In addition, we request that you upload any figures associated with your paper as individual TIF or EPS files with 300dpi resolution at resubmission; please read our figure guidelines for more information on our requirements: http://journals.plos.org/plosmedicine/s/figures. While revising your submission, please upload your figure files to the PACE digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at PLOSMedicine@plos.org.

We expect to receive your revised manuscript by Dec 21 2021 11:59PM. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

We ask every co-author listed on the manuscript to fill in a contributing author statement, making sure to declare all competing interests. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. If new competing interests are declared later in the revision process, this may also hold up the submission. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT. You can see our competing interests policy here: http://journals.plos.org/plosmedicine/s/competing-interests.

Please use the following link to submit the revised manuscript:

https://www.editorialmanager.com/pmedicine/

Your article can be found in the "Submissions Needing Revision" folder.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

We look forward to receiving your revised manuscript.

Sincerely,

Caitlin Moyer, Ph.D.

Associate Editor

PLOS Medicine

plosmedicine.org

-----------------------------------------------------------

Requests from the editors:

Comments from the Academic Editor:

1. The authors have selectively focused on adversity and not on potentially protective or mitigating factors e.g. relating to schooling/educational success, social capital/extended family indicators (if measured). This appears to be a limitation.

2. Was there any conceptual model that drove the selection the individual, familial and environmental exposures? for example, was substance use considered as a factor?

3. is it reasonable to expect that predictors of suicidal ideation at those different time-points would be the same?

4. It is surprising that postnatal depression is not a predictor. What could be the reason for that?

5. I agree with the statistical reviewer's concerns about the way that variables were selected for inclusion into the multivariable analysis, and also the interpretation of interactions.

Other editorial points:

6. Data availability statement: Thank you for providing the link to request data access. Please clarify if there is a more direct link available (e.g. https://www.wits.ac.za/coe-human/open-access-datasets/). If possible please provide a (non-author) contact email address to which data inquiries can be directed.

According to this website, the following text should be included in the acknowledgements: “The support of the DSI-NRF Centre of Excellence in Human Development at the University of the Witwatersrand, Johannesburg in the Republic of South Africa towards this research is hereby acknowledged. Opinions expressed and conclusions arrived at, are those of the author and are not to be attributed to the CoE in Human Development.” It seems only part of this statement was included.

7. Throughout: Please include line numbers running throughout the text with the revised version.

8. Abstract: Please combine the Methods and Findings sections into one section, “Methods and findings”.

9. Abstract: Methods: We suggest emphasizing early on that the primary outcome of interest was suicidal ideation reported at any age.

10. Abstract: Methods and Findings: Please quantify the main results (with 95% CIs and p values). Please mention the important dependent variables that are adjusted for in the analyses.

11. Abstract: Methods and Findings: Please clarify “high birth order” briefly.

12. Abstract: Methods and Findings: In the last sentence of the Abstract Methods and Findings section, please describe the main limitation(s) of the study's methodology.

13. Abstract: Conclusions: Please address the study implications without overreaching what can be concluded from the data; the phrase "In this study, we observed ..." may be useful.

14. Author summary: At this stage, we ask that you include a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists. The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract. Please see our author guidelines for more information: https://journals.plos.org/plosmedicine/s/revising-your-manuscript#loc-author-summary

15. Throughout: Please use square brackets for in-text citations, rather than superscript numbers. Please place the reference before the sentence punctuation, and please do not include spaces within brackets where multiple references are indicated, for example [1,2].

16. Methods: Please ensure that the study is reported according to the STROBE guideline, and include the completed STROBE checklist as Supporting Information. Please add the following statement, or similar, to the Methods: "This study is reported as per the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline (S1 Checklist)."

The STROBE guideline can be found here: http://www.equator-network.org/reporting-guidelines/strobe/

When completing the checklist, please use section and paragraph numbers, rather than page numbers.

17. Methods: Did your study have a prospective protocol or analysis plan? Please state this (either way) early in the Methods section.

a) If a prospective analysis plan (from your funding proposal, IRB or other ethics committee submission, study protocol, or other planning document written before analyzing the data) was used in designing the study, please include the relevant prospectively written document with your revised manuscript as a Supporting Information file to be published alongside your study, and cite it in the Methods section. A legend for this file should be included at the end of your manuscript.

b) If no such document exists, please make sure that the Methods section transparently describes when analyses were planned, and when/why any data-driven changes to analyses took place.

c) In either case, changes in the analysis-- including those made in response to peer review comments-- should be identified as such in the Methods section of the paper, with rationale.

18. Methods: Please provide some brief details on how mothers were recruited for the Bt20+ cohort.

19. Methods: Please clarify the information on phrasing of questions across languages: “Data

collection was performed mainly in isiZulu, Sesotho, or English, with consensual agreement on the

phrasing of questions asked across the different languages.”

20. Methods: “For this study, we analysed data from a sample of 2,020 participants with at least one measure of suicidal ideation at ages 14, 17, 22, or 28 years. This analytical sample differed from the original cohort on a number of variables, including maternal age and schooling, household crowding, and assets. Inverse probability weighting was therefore used in all analyses to address biases due to differential attrition.”

Please provide a table comparing the sample of 2,020 participants included here and the original cohort. A participant flowchart would be helpful.

21. Ethical approval: Please clarify if this report is a secondary analysis of data collected from the Bt20+ study and if ethical approval for this analysis was waived.

Please clarify the nature of written consent from all participants at all waves- was written consent obtained from parents/guardians with written assent from the children, with written consent provide at waves conducted during adulthood?

Please also note whether and how individuals indicating having experienced suicidal ideation were followed up for additional intervention and care.

22. Methods: Study waves at age 17 and 22: Please clarify if the response of “No more than usual” to “Have you recently found that the ideas of taking your own life kept coming into your mind?” was dichotomized as yes or no, and if categorized as “no” please explain as this seems as if it could indicate suicidal ideation.

23. Methods: It would be helpful to include copies of the relevant questions used to obtain the primary outcome and childhood risk factor data, as supporting information.

24. Methods: Please describe the inverse probability weighting that was used to account for study attrition. Please include more description of the multiple imputation methods. It would be helpful to indicate the amount of missing responses for each relevant outcome/factor within a Table.

25. Methods: Please specify the significance level used (e.g., P<0.05, two-sided).

26. Results: Please provide an analysis comparing characteristics of those for which data were missing compared with those included in the study.

27. Results: For all results presented in the text (e.g from univariable and multivariable analyses) please present both the 95% CIs and p values. Please present the overall associations for each factor. If results stratified by sex are presented, please present the findings for both sexes, and please also report the interaction.

28. Results: “This was especially true for poverty (1.25, CI 1.02- 1.52), low maternal education (OR 1.31, CI 1.04-1.64), and household crowding (OR 1.24, CI 0.98-1.57).” Please clarify that this describes the associations for males.

29. Results: Childhood internalizing problems: For differences reported between males and females, it seems as if the interaction by sex did not reach statistical significance. Please mention in the text that while the association with internalizing problems was significant for males (and please report the result for females) there was no evidence to support an interaction effect by sex. Please present the overall associations for externalizing and internalizing problems in addition to the sex-stratified analyses.

30. Discussion: Please present and organize the Discussion as follows: a short, clear summary of the article's findings; what the study adds to existing research and where and why the results may differ from previous research; strengths and limitations of the study; implications and next steps for research, clinical practice, and/or public policy; one-paragraph conclusion.

31. Discussion: Please temper statements related to primacy with “To the best of our knowledge” or similar: “This is the first study to prospectively examine associations…”

32. Discussion: “A key strength of this paper is the use of longitudinal data from the longest birth cohort in Africa to prospectively examine the impact of the early childhood environment on suicidal ideation in youth.” Here and throughout, please revise to avoid language that implies causality. Instead, please refer to associations.

33. Figure 1: Please present this information in a table format. Please note the numbers as well as percentages reported for both males and females. Please indicate in the legend what is represented by the “error bars” and please note that the percentages are relative to the individuals in the cohort for which suicidal ideation data were available at the given age.

34. Figure 2: Please provide a descriptive legend describing the figure, including how risk ratio was determined and the meaning of the points and bars.

35. Table 1: Please specify the significance level used (eg, P<0.05, two-sided) and the statistical test used to derive the p value.

36. Table 2: Please provide the p values for the associations for the whole sample, males, and females. In the legend, please note the statistical tests used.

37. Table 3: Please provide the p values for these associations. In the legend, please note the statistical tests used.

38. References: Please use the "Vancouver" style for reference formatting, and see our website for other reference guidelines https://journals.plos.org/plosmedicine/s/submission-guidelines#loc-references

Comments from the reviewers:

Reviewer #1: I confine my remarks to statistical aspects of this paper. The general approach is fine, but I have a few issues to resolve before I can recommend publication.

Abstract (and similar in main text) - when comparing males to females be careful to not accept the null. You can either insert a bunch of "significantly" or else, rather than say "not associated" say things like "higher in males" or "higher in females".

p. 3 I'm not sure you can say that preventing suicidal ideation would enhance mental well being. It could be the other way around.

"seriously considering and attempting suicide" - which one? Either?

p. 6 Don't categorize birth weight and (maybe) don't categorize depression. Categorizing a continuous variable is nearly always a bad idea. See my blog post https://medium.com/@peterflom/what-happens-when-we-categorize-an-independent-variable-in-regression-77d4c5862b6c

p. 7 This method of model building is known as bivariate screening and it is not recommended. All the results will be wrong. P values are too low, standard errors are too small, parameter estimates are biased away from 0. See Harrell, Regression Modeling Strategies.

Peter Flom

Reviewer #2: Title: Childhood factors associated with suicidal ideation among South African youth A 28-year longitudinal study using the Birth to Twenty Plus cohort

* What are the main claims of the paper and how significant are they for the discipline?

This is a well-written manuscript focusing on an important subject in child mental health. The research aims to show the possible impact of individual, familial and environmental factors on the suicidal ideation of young people in South Africa taken from a cohort sample.

* Are the claims properly placed in the context of the previous literature? Have the authors treated the literature fairly?

Discussion was good and comprehensive.

* Do the data and analyses fully support the claims? If not, what other evidence is required?

Overall, the sample and methods of the study are robust. However, could the authors explain why they have not used ACE-IQ to find out about adverse childhood experiences? Furthermore, why did they not ask for ACEs after age 13? I believe this might have changed the cumulative ACE score and therefore its possible association with suicidal ideation. What is the mean ACE score for the sample? Why did the authors use cumulative ACE score rather than the mean score which is more commonly used in the field? Have the authors looked into the impacts of individual ACE components on suicidal ideation?

The authors also mention about attrition being substantial despite being comparable with other cohorts. Have the authors compared baseline characteristics of participants who were lost to follow-up with clinical characteristics of those remaining? If so, what are the results, and could they have an effect on the results of the paper?

* PLOS Medicine encourages authors to publish detailed methods as supporting information online. Do any particular methods used in the manuscript warrant such publication? If a protocol is already provided, for example for a randomized controlled trial, are there any important deviations from it? If so, have the authors explained adequately why the deviations occurred?

I have not seen any information about this.

* Is this paper outstanding in its discipline? If yes, what makes it outstanding? If not, why not?

It is an important paper taken from a cohort sample, looking into an important subject in mental health, and aiming to find out the factors associated with suicidal ideation in a population from a low/middle income country.

* Does the study conform to any relevant guidelines such as CONSORT, MIAME, QUORUM, STROBE, and the Fort Lauderdale agreement?

N/A

* Are details of the methodology sufficient to allow the experiments?

Yes.

* Is any software created by the authors freely available?

The dataset is not available online. Authors state that the data underlying the results presented in the study are available to authorized researchers and provide a link to apply for authorization to access data.

* Is the manuscript well organized and written clearly enough to be accessible to non-specialists?

The manuscript is well-written and easy to understand. The English and Scientific language is of adequate quality throughout the manuscript.

Reviewer #3: Introduction

Please elaborate or define or give examples of internalizing and externalizing problems for the benefit of the reader. Especially since externalizing factor is one of the significant findings in this study.

Methods

It is quite a complicated methodology with various timelines, from participant and mothers as well as various questionnaires used. Perhaps a diagrammatic flowchart will benefit the readers in understanding the process.

A few of the childhood risk factors were elicited at time of child's birth such as birth order and socioeconomic adversity, which may have changed during their lifetime. This step cannot be undone but can be mentioned as a limitation of this study.

Childhood internalizing and externalizing problems:

1. It is not very clear regarding SACAS. You mentioned in the manuscript:" The South African Child Assessment Schedule (SACAS) was used to ascertain child externalising problems……The SACAS is an 85-item questionnaire based on the Child Behaviour Checklist" and "Externalising problems were assessed with 35 items describing…"

Is the 35 items part of SACAS?

2. Perhaps can put in the internal reliability (Cronbach alpha) of the SACAS to further strengthen the strength of this tool

Throughout the study, there seems to be different ages at which the assessments took place which does not tally with each other. For example:

- Child externalizing problems were assessed at 5,7, 10 from their mothers.

- Adverse Childhood Experiences (ACEs) were assessed at 5,7, 11

Why is this so?

Adverse Childhood Experience (ACEs)

1. Are the questions to assess ACEs from a reliable and validated questionnaire?

2. "Overall exposure to ACEs was computed by summing the number of reported ACEs (by either mothers or participants) and the final score was standardized" …is not a very clear description of the scoring process.

Reviewer #4: This is a well-written paper that investigates an important public health problem about which little is known, re: childhood risk factors for youth suicide (14-28 years) in a LMIC setting (South Africa). The study data is from a population-based longitudinal prospective study that followed-up children from birth to adulthood. Study methods, measures and data analyses are adequately described. Suicidal ideation (outcome measure) was assessed as self-reported suicidal ideation at any time (at ages 14, 17, 22, and 28 years). Independent measures included early-life adversity (data collected at baseline), children's externalizing and internalising problems (assessed at ages 5, 7, and 10 years, from maternal reports using the SACAS), and ACEs (obtained from mothers at child ages 5, 7, and 11, and from children at ages 11 and 13). The main findings were:

1. 22.3% participants reported SI between ages of 14 and 28; peak SI was at age 17 years

2. Females reported higher SI at all time points except for age 28 when males = females

3. Externalising problems in childhood independently predicted SI

4. Internalising problems in childhood was not an independent predictor of S

5. Childhood socioeconomic adversity was associated with SI in males alone

6. ACEs, low birth weight, and high birth order, were associated with SI in females alone

The lack of an association between internalizing behaviors and risk of suicide is a major albeit surprising finding that is inconsistent with existing literature (1). This discrepancy raises the question of the extent to which the study (measures, procedures) establishes a trustworthy cause and effect relationship between internalizing behaviors and suicidal ideation. That is, the study's internal validity that makes it possible to eliminate alternative explanations for the finding. Threats to internal validity and thus potential sources of bias include:

1. Use of a parent report measure to assess internalizing behaviors. Parent ratings of children's behaviors may reflect parental attitudes and stress just as much as they reflect the child's behavior

2. Repeated testing of participants using the same measures. This influences study outcomes as participants will often do better as they become used to the testing process

3. Maturation effects: this refers to the impact of time as a variable in a study (for example, participants growing older). This can make it impossible to rule out whether effects seen in the study were simply due to the effect of time.

4. Attrition: in the study, the analytical sample was a biased one as it differed from the original cohort on many variables, including maternal age and schooling, household crowding, and assets.

5. Inverse probability weighting: used to address biases due to differential attrition in data analyses. However, one problem issue with IP-weighting is that participants who are extremely unlikely to be treated (that is, those with negative association for SI) will end up with a large weight, potentially making the weighted estimator highly unstable. A common alternative to the conventional weights that at least "kind of" addresses this problem are the stabilized weights, which use the marginal probability of treatment instead of 1 in the weight numerator (2).

I would recommend that the authors address these issues pertaining to internal validity.

References:

1. Auerbach, R. P., Mortier, P., Bruffaerts, R., Alonso, J., Benjet, C., Cuijpers, P., Demyttenaere, K., Ebert, D. D., Green, J. G., Hasking, P., Lee, S., Lochner, C., McLafferty, M., Nock, M. K., Petukhova, M. V., Pinder-Amaker, S., Rosellini, A. J., Sampson, N. A., Vilagut, G., Zaslavsky, A. M., … WHO WMH-ICS Collaborators (2019). Mental disorder comorbidity and suicidal thoughts and behaviors in the World Health Organization World Mental Health Surveys International College Student initiative. International journal of methods in psychiatric research, 28(2), e1752. https://doi.org/10.1002/mpr.1752

2. The intuition behind inverse probability weighting in causal inference https://www.rebeccabarter.com/blog/2017-07-05-ip-weighting/

Any attachments provided with reviews can be seen via the following link:

[LINK]

7 Jan 2022

Submitted filename: 2022-01-06_R1_Response Letter.docx

Click here for additional data file.

8 Feb 2022

Dear Dr. Orri,

Thank you very much for re-submitting your manuscript "Childhood factors associated with suicidal ideation among South African youth A 28-year longitudinal study using the Birth to Twenty Plus cohort" (PMEDICINE-D-21-03460R2) for review by PLOS Medicine.

I have discussed the paper with my colleagues and the academic editor and it was also seen again by 3 reviewers. I am pleased to say that provided the remaining editorial and production issues are dealt with we are planning to accept the paper for publication in the journal.

The remaining issues that need to be addressed are listed at the end of this email. Any accompanying reviewer attachments can be seen via the link below. Please take these into account before resubmitting your manuscript:

[LINK]

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

In revising the manuscript for further consideration here, please ensure you address the specific points made by each reviewer and the editors. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments and the changes you have made in the manuscript. Please submit a clean version of the paper as the main article file. A version with changes marked must also be uploaded as a marked up manuscript file.

Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. If you haven't already, we ask that you provide a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists. The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract.

We expect to receive your revised manuscript within 1 week. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

We ask every co-author listed on the manuscript to fill in a contributing author statement. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT.

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript.

Please note, when your manuscript is accepted, an uncorrected proof of your manuscript will be published online ahead of the final version, unless you've already opted out via the online submission form. If, for any reason, you do not want an earlier version of your manuscript published online or are unsure if you have already indicated as such, please let the journal staff know immediately at plosmedicine@plos.org.

If you have any questions in the meantime, please contact me or the journal staff on plosmedicine@plos.org.

We look forward to receiving the revised manuscript by Feb 15 2022 11:59PM.

Sincerely,

Caitlin Moyer, Ph.D.

Associate Editor

PLOS Medicine

plosmedicine.org

------------------------------------------------------------

Requests from Editors:

1. From the academic editor: The amount of missing data for violence/abuse and for postnatal depression was very high. So, although postnatal depression was not associated with attrition from the cohort, I wonder if it was associated with whether or not data were available on these variables. This may help to explain the lack of association between postnatal depression and later suicidality (although the authors' arguments about why that association may not be seen in this context are also relevant). Please address and comment on the potential relationship between the data that were not available and depression and how this could factor into the absence of a significant association with suicidal ideation.

2. Title: We suggest the following revision to the title. Please make this change in the manuscript submission system as well as the text: “Childhood factors associated with suicidal ideation among South African youth: A 28-year longitudinal study of the Birth to Twenty Plus cohort”

3. Data availability statement: Please update this information in the manuscript submission system (Data availability section) rather than including it at page 20. Please note that there was no non-author contact information provided, as indicated in your response.

4. Response to editor’s comment 20: Please do include Table S1. It is not necessary to include the flowchart, but please do provide the numbers and explanation (e.g. “Of the original 3,273, suicidal ideation data were not available for…” or similar) for those not included in the text of the Methods (approximately at lines 154-155: “For this study, we analyzed data from a sample of 2,020 participants with at least one measure of suicidal ideation at ages 14, 17, 22, or 28 years.”).

5. Abstract: Line 31-32: We suggest revising to: “We documented the associations between individual, familial, and environmental factors in childhood with suicidal ideation…”

6. Abstract: Line 47: Please clarify if p<0.030 should be p=0.03, or otherwise, please report the exact p value unless p<0.001.

7. Author summary: Line 65: Please revise to “To the best of our knowledge, no longitudinal study has been conducted…” or similar.

8. Author summary: Line 72: Please revise to “2,020” in the second bullet point.

9. Author summary: Line 81-83: We suggest revising these two points slightly to avoid suggesting causal implications (e.g. suggesting a direct link to prevention) of the findings.

10. Author summary: Line 84: Please provide slightly more explanation for this point.

11. Introduction: Line 87: Please clarify to “...the second or third most common cause of death…” or similar.

12. Methods: Line 207: Please change the superscript reference to [43] if this is correct.

13. Methods: Line 232: Data analysis plan: Please state that there was no formal prospective analysis plan for the study, but please do mention that the analysis protocol was decided upon during study group meetings (please indicate when this took place, i.e. prior to initiation of the analyses), and please indicate that changes to the planned analyses with rationale (e.g. following peer reviewer comments) are described in the Methods.

14. Results: Line 264: Please change to P=0.007 if this is accurate. Please report the exact p value, unless p<0.001.

15. Results: Line 265 and 266: Please change to “statistically significantly” and “statistically significant” in this sentence. Please also provide the p values for the sex-specific associations reported.

16. Results: Line 279-281: We suggest revising to: “...stratified analyses by sex suggested that the association was stronger among females (OR 1.14, CI 1.05-1.24, P=0.003) than males (OR 1.06, CI 0.95-1.18, P=0.299), although the interaction was not statistically significant…”.

17. Results: Line 288: Please change to P=0.011 if this is accurate. Please report the exact p value, unless P<0.001.

18. Discussion: Line 294: We suggest that the heading “Main Findings” is not necessarily needed.

19. Discussion: Line 299: We suggest revising to: “when rates for males and females were similar…”

20. Discussion: Lines 402-404: We suggest revising slightly to: “Prevalence rates were higher among females than males, and we found sex differences in the associations of childhood individual, familial, and environmental factors with youth suicidal ideation. As these factors (e.g., externalizing problems, socioeconomic adversity, ACEs) are highly prevalent…”

21. Page 20: Please remove the sections titled Contributions, Conflict of Interest disclosures, and Data Access from the main text. Please make sure that all information is entered completely and accurately into the relevant sections of the manuscript submission system.

22. Table 2: Please define RR in the legend.

23. Table 3 and Table 4: Please define OR in the legend.

24. Figure 1: Please use a color scheme that does not involve differentiating red and green colors.

25. Reference 24: Please change the journal title abbreviation to PLoS One.

26. Reference 51: Please change the journal title abbreviation to Lancet.

27. Reference 53: Please update with complete citation information.

28. Table S1 and S2: Please define SD in the legend.

29. Table S3: Please define ACE in the legend.

Comments from Reviewers:

Reviewer #1: The authors have addressed my concerns and I now recommend publication.

Peter Flom

Reviewer #3: All comments were addressed and relevant amendments made. Thank you.

Reviewer #4: The issues I raised in the earlier review have been satisfactorily addressed by the authors.

Any attachments provided with reviews can be seen via the following link:

[LINK]

10 Feb 2022

Submitted filename: 2022-02-09_R2_ResponseLetter.docx

Click here for additional data file.

14 Feb 2022

Dear Dr Orri,

On behalf of my colleagues and the Academic Editor, Charlotte Hanlon, I am pleased to inform you that we have agreed to publish your manuscript "Childhood factors associated with suicidal ideation among South African youth A 28-year longitudinal study of the Birth to Twenty Plus cohort" (PMEDICINE-D-21-03460R3) in PLOS Medicine.

Before your manuscript can be formally accepted you will need to complete some formatting changes, which you will receive in a follow up email. Please be aware that it may take several days for you to receive this email; during this time no action is required by you. Once you have received these formatting requests, please note that your manuscript will not be scheduled for publication until you have made the required changes.

In the meantime, please log into Editorial Manager at http://www.editorialmanager.com/pmedicine/, click the "Update My Information" link at the top of the page, and update your user information to ensure an efficient production process.

Please also address the following editorial points:

-Title: Please update the title in the manuscript submission system: “Childhood factors associated with suicidal ideation among South African youth: A 28-year longitudinal study of the Birth to Twenty Plus cohort”

-Methods: Line 157: Please correct to “12,530” if this is accurate.

-Table 1: Please define the abbreviation for ACE in the legend.

PRESS

We frequently collaborate with press offices. If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximise its impact. If the press office is planning to promote your findings, we would be grateful if they could coordinate with medicinepress@plos.org. If you have not yet opted out of the early version process, we ask that you notify us immediately of any press plans so that we may do so on your behalf.

We also ask that you take this opportunity to read our Embargo Policy regarding the discussion, promotion and media coverage of work that is yet to be published by PLOS. As your manuscript is not yet published, it is bound by the conditions of our Embargo Policy. Please be aware that this policy is in place both to ensure that any press coverage of your article is fully substantiated and to provide a direct link between such coverage and the published work. For full details of our Embargo Policy, please visit http://www.plos.org/about/media-inquiries/embargo-policy/.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Thank you again for submitting to PLOS Medicine. We look forward to publishing your paper.

Sincerely,

Caitlin Moyer, Ph.D.

Associate Editor

PLOS Medicine

Table 1

Sociodemographic characteristics of the sample.

	Whole sample (N = 2,020)	Males (n = 978)	Females (n = 1,042)	P
Female sex	1,042 (51.6)
Low birth weight	219 (10.9)	94 (9.6)	125 (12.0)	0.102
Birth order				0.308
First	769 (38.1)	363 (37.1)	406 (39.0)
Second	584 (28.9)	284 (29.0)	300 (28.8)
Third	352 (17.4)	164 (16.8)	188 (18.0)
Fourth+	315 (15.6)	167 (17.1)	148 (14.2)
Mean*	2.11 (1.08)	2.14 (1.10)	2.07 (1.06)	0.189
Socioeconomic adversity index (z-score)*	0.00 (1.00)	0.05 (1.01)	−0.04 (0.99)	0.064
Low maternal age	149 (7.4)	68 (7.0)	81 (7.8)	0.527
Low maternal education	1,096 (59.1)	546 (61.1)	550 (57.3)	0.102
Household crowding	772 (44.6)	384 (46.0)	388 (43.4)	0.292
Poverty	890 (44.1)	449 (45.9)	441 (42.3)	0.115
Cumulative ACE score (z-score)*a	0.00 (1.00)	−0.01 (1.01)	0.01 (0.99)	0.711
Postnatal maternal depression	199 (17.2)	106 (18.4)	93 (16.0)	0.316
Parity				0.446
First	769 (38.1)	363 (37.1)	406 (39.0)
Second	584 (28.9)	284 (29.0)	300 (28.8)
Third	352 (17.4)	164 (16.8)	188 (18.0)
Fourth	165 (8.2)	86 (8.8)	79 (7.6)
Fifth+	150 (7.4)	81 (8.3)	69 (6.6)
Mean*	2.18 (1.23)	2.22 (1.26)	2.14 (1.20)	0.146
Previous abortions/stillbirths	219 (10.8)	122 (12.5)	97 (9.3)	0.027
Externalizing problems (z-score)*	0.00 (1.00)	−0.03 (1.00)	0.05 (0.97)	<0.001
Internalizing problems (z-score)*	0.00 (1.00)	0.23 (1.02)	−0.05 (0.89)	0.087

Counts and %, except for continuous variables (*), described as mean and SD.

aUnstandardized values for the whole sample, males, and females are 3.00 (2.07), 2.93 (2.03), and 3.05 (2.10), respectively. All tests are 2 sided and considered statistically significant at P < 0.05. P values have been obtained using Student t tests and chi-squared tests for continuous and categorical variables, respectively.

ACE, adverse childhood experience; SD, standard deviation.

Table 2

Prevalence rates of suicidal ideation in the Bt20+ cohort.

	Whole sample		By sex
			Females		Males		Sex comparison
Age	n/N	% (95% CI)	n/N	% (95% CI)	n/N	% (95% CI)	RR (95% CI)	P
14	94/1,209	7.8 (6.3 to 9.3)	71/632	11.2 (8.8 to 13.7)	23/577	4.0 (2.4 to 5.6)	2.81 (1.79 to 4.61)	<0.001
17	231/1,532	15.1 (13.3 to 16.9)	161/814	19.8 (17.0 to 22.5)	70/718	9.8 (7.6 to 11.9)	2.03 (1.54 to 2.70)	<0.001
22	164/1,582	10.4 (8.9 to 11.9)	105/821	12.8 (10.5 to 15.1)	59/761	7.8 (5.9 to 9.6)	1.65 (1.20 to 2.28)	0.001
28	96/1,388	6.9 (5.6 to 8.3)	58/728	8.00 (6.0 to 9.9)	38/660	5.8 (4.0 to 7.5)	1.38 (0.92 to 2.10)	0.130

The table shows the number of participants reporting suicide attempt (n) relative to the number of assessed participants (N) at any assessment age, as well as the rate (%) with 95% CI. Statistics are provided for the whole sample and for females and males separately. The prevalence of suicidal ideation in females versus males is compared using RR with 95% CI, and P values were computed using chi-squared tests. All tests are 2 sided and considered statistically significant at P < 0.05.

Bt20+, Birth to Twenty Plus; CI, confidence interval; RR, risk ratio.

Table 3

Univariable associations between childhood factors and suicidal ideation.

	Whole sample		Males		Females		Interaction
	OR (CI)	P	OR (CI)	P	OR (CI)	P	P
Child characteristics
Female sex	2.09 (1.84 to 2.38)	<0.001	-		-		-
Birth weight (each kg decrease)	1.25 (1.11 to 1.41)	<0.001	1.19 (0.98 to 1.43)	0.075	1.18 (1.01 to 1.39)	0.048	0.931
Birth order
Second	1.11 (0.96 to 1.29)	0.172	1.00 (0.78 to 1.28)	0.992	1.20 (0.99 to 1.46)
Third	0.86 (0.71 to 1.04)	0.119	0.78 (0.57 to 1.07)	0.125	0.90 (0.71 to 1.15)
Fourth+	1.45 (1.21 to 1.73)	<0.001	1.26 (0.95 to 1.66)	0.107	1.79 (1.41 to 2.27)
Trend	1.08 (1.02 to 1.14)	0.007	1.04 (0.95 to 1.14)	0.429	1.14 (1.06 to 1.22)	0.001	0.120
Sociodemographic and maternal characteristics
Socioeconomic adversity index	1.03 (0.97 to 1.10)	0.355	1.17 (1.05 to 1.29)	0.003	0.98 (0.90 to 1.06)	0.576	0.009
Low maternal age	0.95 (0.78 to 1.17)	0.654	0.97 (0.69 to 1.37)	0.868	0.92 (0.70 to 1.19)	0.508	0.788
Low maternal education	1.11 (0.97 to 1.26)	0.132	1.31 (1.04 to 1.64)	0.023	1.06 (0.89 to 1.25)	0.514	0.150
Household crowding	1.06 (0.92 to 1.23)	0.404	1.24 (0.98 to 1.57)	0.068	0.99 (0.82 to 1.19)	0.917	0.137
Poverty	0.99 (0.87 to 1.12)	0.840	1.25 (1.02 to 1.52)	0.031	0.89 (0.76 to 1.04)	0.152	0.010
Postnatal maternal depression	1.17 (0.89 to 1.54)	0.216	1.30 (0.86 to 1.96)	0.235	1.15 (0.81 to 1.62)	0.247	0.628
Parity
Second	1.11 (0.96 to 1.29)	0.172	1.00 (0.78 to 1.28)	0.992	1.20 (0.99 to 1.46)
Third	0.86 (0.71 to 1.04)	0.119	0.78 (0.57 to 1.07)	0.125	0.90 (0.71 to 1.15)
Fourth	1.44 (1.14 to 1.81)	0.002	1.26 (0.88 to 1.80)	0.214	1.73 (1.28 to 2.35)
Fifth+	1.46 (1.15 to 1.85)	0.002	1.26 (0.87 to 1.82)	0.216	1.86 (1.35 to 2.58)
Trend	1.08 (1.03 to 1.13)	0.003	1.04 (0.96 to 1.13)	0.322	1.13 (1.06 to 1.21)	<0.001	0.106
Previous abortions/stillbirths	0.94 (0.77 to 1.16)	0.576	0.92 (0.67 to 1.27)	0.615	1.09 (0.82 to 1.45)	0.549	0.439
Childhood mental health
Externalizing problems	1.15 (1.06 to 1.25)	<0.001	1.28 (1.13 to 1.45)	<0.001	1.19 (1.06 to 1.34)	0.001	0.380
Internalizing problems	1.12 (1.04 to 1.21)	0.003	1.19 (1.05 to 1.34)	0.003	1.05 (0.95 to 1.16)	0.328	0.118
ACEs
Cumulative ACE score	1.10 (1.03 to 1.18)	0.003	1.06 (0.95 to 1.18)	0.299	1.14 (1.05 to 1.24)	0.003	0.308
Material deprivation	1.16 (1.08 to 1.24)	<0.001	1.04 (0.93 to 1.16)	0.535	1.26 (1.16 to 1.38)	<0.001	0.007
Loss	1.04 (0.98 to 1.11)	0.219	1.02 (0.91 to 1.13)	0.789	1.05 (0.97 to 1.14)	0.233	0.609
Family dynamics	1.06 (1.00 to 1.13)	0.057	1.09 (0.98 to 1.21)	0.096	1.04 (0.95 to 1.12)	0.403	0.430
Abuse and violence	1.01 (0.94 to 1.09)	0.796	0.99 (0.87 to 1.12)	0.858	1.05 (0.95 to 1.15)	0.334	0.474

All tests are 2 sided and considered statistically significant at P < 0.05. P values have been obtained using univariable logistic regressions.

ACE, adverse childhood experience; OR, odds ratio.

Table 4

Multivariable associations between childhood factors and suicidal ideation.

	Males		Females
	OR (CI)	P	OR (CI)	P
Birth weight (each kg decrease)	1.15 (0.95 to 1.41)	0.146	1.20 (1.02 to 1.41)	0.030
Birth order (trend)	1.02 (0.93 to 1.12)	0.642	1.15 (1.07 to 1.24)	<0.001
Any previous abortion	0.92 (0.67 to 1.27)	0.616	0.97 (0.73 to 1.30)	0.848
Postnatal maternal depression	1.01 (0.97 to 1.06)	0.576	1.02 (0.98 to 1.06)	0.441
Socioeconomic adversity score	1.13 (1.01 to 1.26)	0.031	0.94 (0.86 to 1.02)	0.132
Externalizing problems	1.23 (1.08 to 1.40)	0.002	1.16 (1.03 to 1.30)	0.011
Internalizing problems	1.12 (0.99 to 1.27)	0.083	1.00 (0.90 to 1.11)	0.939
ACE score	0.97 (0.86 to 1.09)	0.584	1.11 (1.01 to 1.21)	0.030

All tests are 2 sided and considered statistically significant at P < 0.05. P values have been obtained using multivariable logistic regressions.

ACE, adverse childhood experience; OR, odds ratio.