Prevalence of suicide thoughts, plans and attempts
The estimated prevalence of suicidal thoughts, plans and attempts varies widely across different studies and countries.Reference Ndetei, Mutiso, Weisz, Okoth, Musyimi and Muia1–Reference Abdu, Hajure and Desalegn10 Among young people, the prevalence of suicidal thoughts has been reported to range from 10.7% in a cross-sectional epidemiological study in GermanyReference Voss, Ollmann, Miché, Venz, Hoyer and Pieper7 to 17% in a prospective cohort study in NorwayReference Strandheim, Bjerkeset, Gunnell, Bjørnelv, Holmen and Bentzen4 and 17.9% in a study in Turkey,Reference Canbaz and Terzi5 whereas the prevalence of suicide attempts ranges from 3.6 to 10.3%.Reference Pandey, Bista, Dhungana, Aryal, Chalise and Dhimal6,Reference Voss, Ollmann, Miché, Venz, Hoyer and Pieper7 In the USA, the prevalence of suicidal thoughts ranged from 4.0% in the north-east and south to 4.8% in the west, and from 3.3% in New Jersey to 6.9% in Utah in the National Survey on Drug Use and Health covering the period between 2015 and 2019.Reference Ivey-Stephenson, Crosby, Hoenig, Gyawali, Park-Lee and Hedden11
Differences in prevalence have been observed in African studies, with lifetime prevalence rates of suicidal thoughts, plans and attempts of 58.3, 37.3 and 4.4%, respectively, reported in an Ethiopian cross-sectional study.Reference Abdu, Hajure and Desalegn10 A cross-sectional study in GhanaReference Quarshie, Cheataa-Plange, Annor, Asare-Doku and Lartey9 found prevalence rates of 15.4, 6.6 and 2.3%, respectively. The study in Ghana focused on nursing students specialising in midwifery; thus, it provided a broader perspective on the prevalence of suicidality among various populations. In addition, a study on suicidal ideation among youth in Kampala, Uganda, reported a prevalence of 23.53%.Reference Culbreth, Swahn, Ndetei, Ametewee and Kasirye8
A study of suicidal ideation among Kenyan students aged 15+ years found a prevalence of 22.6%.Reference Ndetei, Mutiso, Weisz, Okoth, Musyimi and Muia1 In an out-patient psychiatric clinic in a youth centre of a referral hospital in Kenya, 82% of youth showed suicidal behaviours (plans and attempts); those aged 16–18 years were more likely to show suicidal behaviour compared with those aged 19–22 years.Reference Khasakhala, Ndetei and Mathai2
Methods in suicidality
According to the World Health Organization (WHO), hanging, ingestion of pesticides and use of firearms are the most common methods of suicide globally.12 Among adolescents, a meta-analytic review of worldwide suicide rates reported that the most common suicide methods were hanging and jumping from a height or in front of a moving object.Reference Glenn, Kleiman, Kellerman, Pollak, Cha and Esposito13 A study in Western KenyaReference Ongeri, Larsen, Jenkins, Shaw, Connolly and Lyon14 found the most commonly reported method of suicide to be self-poisoning. Research has shown that suicidality methods tend to differ, with less lethal methods reported in attempters and highly lethal methods for completers.Reference Lim, Lee and Park15 A study in Korea found drug poisoning (which they classified as a less lethal method) to be the most frequent method used by suicide attempters, whereas hanging (classified as highly lethal) was the most common method among suicide completers.Reference Lim, Lee and Park15
Suicidality associations
There is evidence of an association of suicide attempts with thoughts and plans. In the German study mentioned previously,Reference Voss, Ollmann, Miché, Venz, Hoyer and Pieper7 47.0% of participants who had lifetime suicidal ideation reported a suicide plan, and 23.9% reported a suicide attempt. Suicidal plans or attempts occurred mainly in the year of suicidal ideation, with 74.9% of plans and 71.2% of attempts associated with ideation and 85% of attempts associated with plans.Reference Voss, Ollmann, Miché, Venz, Hoyer and Pieper7
Prevention of suicide requires good evidence on both the prevalence and probable hot spots, as well as suicidal behaviours, to inform the development of appropriate interventions for communities and for health and social systems.Reference Venturo-Conerly, Wasil, Puffer, Weisz and Wasanga16 There is a gap in such knowledge in the Kenyan context. Therefore, in this study, we sought to determine the levels of suicidal thoughts, plans, and attempts; suicidality patterns; and sociodemographic associations of suicidality in high school students in three different regional settings in Kenya.
The specific objectives were to determine the following:
(a) the prevalence of suicidal thoughts, plans and attempts;
(b) methods contemplated in suicidal plans and employed in suicidal attempts;
(c) patterns of suicidal thoughts and plans to suicidal attempts;
(d) sociodemographic associations of suicidal thoughts, plans and attempts.
Method
Study setting
We undertook a cross-sectional study in ten secondary schools. The schools were sampled from three regions of 47 (otherwise referred to as counties) in Kenya, which are geographical units created by the 2010 Constitution of Kenya. Our decision to focus on these three regions was primarily influenced by an ongoing project we had in those counties. It made logistical sense to extend our efforts into these areas where we already had established connections, networks and perhaps even existing data that could inform our work. The sampling sought to ensure we recruited a demographically representative sample that reflected the diversity of adolescents within Kenyan secondary schools. The institutions also represented all four levels of government-funded schools (national, extra-county, county and sub-county – the main differences between these types of school in Kenya are their ownership, management and the level of academic performance required for admission) as well as urban and rural school locations. Ten schools were selected as follows: one national, one extra-county, five county and three subcounty. Three schools were day schools, five were boarding, and two had both day and boarding facilities. All the schools had a guidance and counselling department supported by trained teachers. This is in line with a policy directive, where the government has committed resources towards school counselling programmes to assist students to deal with behavioural and psychological challenges that affect their academic and social relations.Reference Gachenia and Mwenje17 Our deliberate choice to prioritise public schools over private ones was informed by a range of factors, including resource limitations and the potential psychological impact on students. Public schools often face more significant resource constraints compared with their private counterparts, and this can have profound effects on students’ well-being. These limitations can manifest in various ways, including inadequate mental health support systems, fewer extracurricular opportunities and larger class sizes. These in turn can contribute to increased levels of stress, anxiety and even suicidal ideation among students in public schools.
Sampling procedure
Sampling was multi-stage, with a broad stratification rationale applied to first select the regions, then the secondary schools that represented each of the four levels of government-funded schools within the three regions selected. In each of the schools, students were randomly divided into small groups assembled for the study. Each group comprised 10–15 students, with one research assistant in charge of each group. The questionnaire was thereafter self-administered to the students using paper and pencil.
Data collection to achieve objectives 1, 2 and 3
We asked structured questions to determine the prevalence and methods of suicidal behaviour. Five questions were asked, as follows. (a) ‘Have you thought seriously about ending your life?’ (‘No, I have not,’ ‘Yes, once,’ ‘Yes, more than once’). For this analysis, the response options were dichotomised into ‘No’ and ‘Yes’. (b) ‘Have you tried ending your life?’ (‘No, I have not,’ ‘Yes, once,’ ‘Yes, more than once’). For this analysis, the response options were dichotomised into ‘No’ ‘and ‘Yes’. (c) ‘If yes to question 1 above, did you think of a possible way to end your life?’ (Yes/No). (d) ‘If yes to question 3 above, how?’ (list the methods). (e) ‘If yes to question 2 above, what methods did you use?’ (list them).
To achieve objective 4
Demographic data were assessed using three self-reported questions: (a) gender (male/female/other); (b) age (in years)’; and (c) ‘In what form (high-school grade) are you?’ (1, 2, 3 or 4).
Statistical analysis
Data analysis was performed with SPSS version 25 (Armonk, NY: IBM Corp) for Microsoft Windows. Descriptive summary statistics in the form of frequency, percentage, mean of the age and standard deviation were generated to examine the variables. χ2 and Fisher's exact tests were used where appropriate to determine significance. Univariate and multivariate logistic regression was performed to determine: (a) which socio-demographic variables were associated with suicidality; (b) which suicide plan methods were associated with suicide attempts; and (c) associations between sociodemographic variables and the various suicidality patterns.
Ethics
The authors assert that all procedures contributing to this work complied with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008. All procedures involving human subjects were approved by Kenyatta University Ethics Review Committee (protocol number: PKU/2456/E1587). A research license was granted from the National Commission for Science, Technology, and Innovation (licence number: NACOSTI/P/22/17173). Permission to conduct this research was sought from institutional heads. Informed written consent was obtained from those students over 18 years and assent from those under 18 years. In addition, written consent was obtained from parents/guardians of participants under 18 years via school officials. We assumed there would be students who had suicidal thoughts and/or plans or had attempted suicide. As this was an anonymous survey, we had no way of identifying the students with any of these. In mitigation, we advised the students that in the event they had any of these, they could discuss them with trusted friends, teachers, school counsellors and family members if they felt comfortable with them and gave them the option of a helpline at the Africa Mental Health Research and Training Foundation (AMHRTF).
Results
Sociodemographic characteristics
Of the 2652 participants, 66.6% were male, 33.2% were female and 0.2% were ‘other’. The mean age was 16.3 years, with a standard deviation of ±1.38. The age of the participants ranged from 14.75 to 17.51 years, where average ages were as follows: form 1, 14–15 years; form 2, 15–16 years; form 3, 16–17 years; and form 4, 17–18 years. The fewest respondents (13.6%) were in form 4 (the final class in high school). More of the schools were rural (61.3%) than urban (38.7%). Table 1 provides a summary of the sociodemographic characteristics.
Prevalence of suicidality and its patterns
The response rate for the suicidality questions (suicidal thoughts, suicide plans and suicide attempts) was 95.6% (2534/2652). The responses for various aspects of suicidality could be grouped into the following four categories (percentages/n):
(a) suicidal attempts (regardless of suicidal ideas or plans): 15.7% (397/2534);
(b) suicidal plans but no attempt (regardless of suicidal ideation) 14.9% (377/2534);
(c) suicidal ideations (regardless of plans and attempts) 26.8% (680/2534);
(d) no suicidality (no ideations, no plans, no attempts) 42.6% (1080/2534).
The associations and co-occurrences among suicidal thoughts, plans and attempts can be summarised as follows.
Suicidal thoughts
Of the 680 respondents who had had suicidal thoughts: (a) 55.4% (377/680) had made suicide plans, whereas 44.6% (303/680) had not; (b) 51.3% (349/680) had attempted suicide; (c) 36.0% (245/680) had attempted once and 15.3% (104/680) more than once.
Suicidal plans
Of the 377 respondents who had made suicide plans: (a) 67.1% (253/377) had attempted suicide; (b) 44.8% (169/377) had attempted suicide once and 22.3% (84/377) more than once.
No suicidal plans
Of the respondents who had not made suicide plans but had had suicidal thoughts: (a) 31.7% (96/303) had attempted suicide; and (b) 25.1% (76/303) had attempted suicide once and 6.6% (20/303) more than once. Of the respondents who had no suicide plans or suicidal thoughts: (a) 6.7% (144/2157) had attempted suicide; (b) 5.6% (120/2157) had attempted suicide once and 1.1% (24/2157) more than once.
Methods in suicidal plans and attempts
The methods contemplated in suicidal plans were the same as those used in suicide attempts. The most common method used in suicide attempts was drinking chemicals/poison (18.8%), whereas the least used method was ‘suicide by cop’ (0.3%). On the other hand, hanging and drinking chemicals/poisons were the most common methods contemplated in suicidal plans. There were no significant associations between gender and suicide attempt methods.
Significant gender differences were observed between suicide plan methods for drowning, drug overdose and suffocation by carbon monoxide. Males had a significantly (P < 0.05) higher proportion of drowning suicide plans compared with females. (6.2% v. 0%), and females had a significantly higher proportion of drug overdose suicide plans compared with males. (14.8% v. 7.4%). Males and females had significantly lower proportions of suffocation by carbon monoxide suicide plans compared with ‘other’ gender. There were no significant differences between rural and urban locations in proportions of methods reported in suicide plans and suicide attempts. See Fig. 2 for a summary of the methods contemplated in suicidal plans and used in attempted suicides, and Table 3 for details of sociodemographic differences in methods involved in suicide plans and attempts.
P-values are from Fisher's exact test unless otherwise specified. Superscript letters (a, b) denote a subset of gender categories whose column proportions did not differ significantly from each other at the 0.05 level.
Suicide attempt was associated with having a jumping suicide plan (χ2 = 6.81; P = 0.009) or a drug overdose suicide plan (χ2 = 4.01; P = 0.045). In the univariate analysis, suicide attempt was positively associated with a jumping suicide plan. Participants with a jumping suicide plan had 60% lower odds of attempting suicide than those who did not have a jumping suicide plan. The multivariate analysis revealed that suicide attempts were significantly predicted by participants having a jumping suicide plan. See Table 4 for details of suicide plan methods associated with attempts.
COR, crude odds ratio; AOR, adjusted odds ratio; Ref., reference category; Inf., positive large number.
P-values are from Fisher's exact test unless otherwise specified.
Bold indicates significant results.
Sociodemographic associations of suicidal thoughts, plans and attempts
Suicidal thoughts were significantly associated with age (P = <0.001), school form (χ2 = 20.46; P = <0.001) and location of school (χ2 = 12.61; P = <0.001). Those with ‘other’ gender had five times greater odds of suicidal thoughts than males. For a 1 year increase in age, there was a 20% increase in the odds of suicidal thoughts. Participants in forms 3 and 4 (the third and final years in high school, respectively) had greater odds of thinking about suicide compared with those in form 1 (the first year in high school). Location of school was also a key determinant of suicidal thoughts, with respondents in urban areas more likely to think about suicide than those in rural areas. Suicidal thoughts were significantly predicted by female gender and older age.
Suicide plans were significantly associated with age (P = <0.001), school form/class (χ2 = 16.72; P = <0.001) and location of school (χ2 = 6.65; P = 0.05). For a 1 year increase in age, there was an 18% increase in the odds of having a suicide plan. Participants in forms 3 and 4 had greater odds of suicide plans compared with those in form 1. Location of school (rural or urban) was also a key determinant of suicide plans, with respondents in urban areas more likely to plan suicide than those in rural areas. Suicide plans were significantly predicted by ‘other’ gender.
Suicide attempts were significantly associated with gender (χ2 = 5.69; P = 0.046), age (P = 0.002), school form (χ2 = 9.17; P = 0.027) and location of school (χ2 = 4.44; P = 0.035). Females had greater odds of attempting suicide than males. For a one-unit increase in age, there was a 15% increase in the odds of a suicide attempt. Participants in form 4 had greater odds of attempting suicide compared with those in form 1. The location of the school was also a key determinant of suicide attempts, with respondents in urban areas more likely to attempt suicide than those in rural areas. In the multivariate analysis, suicide attempts were significantly predicted by female gender and older age. See Table 5 for details of sociodemographic associations of thoughts, plans and attempts.
COR, crude odds ratio; AOR, adjusted odds ratio; Ref., reference category.
*P < 0.05, **P < 0.01, ***P < 0.001.
Combinations of suicidality
Table 6 summarises the different combinations and overlaps of suicidal thoughts, plans and attempts and their associations with sociodemographic variables. Table 1 and Fig. 1 summarise the different patterns found in suicide attempts. These patterns were:
(a) suicidal thoughts and plans associated with attempts;
(b) suicidal thoughts and attempts not associated with plans;
(c) attempts not associated with thoughts or plans.
COR, crude odds ratio; AOR, adjusted odds ratio; Ref., reference category.
*P< 0.05, **P < 0.01, ***P < 0.001.
Gender, age, school form and location of school were associated with suicidal attempts.
Discussion
We present the first study of the prevalence of suicidal thoughts, suicidal plans and suicidal attempts in Kenyan high school students. This was also the first Kenyan study to report on the methods conceived by students as part of their plans to end their life, as well as the different methods used in their suicide attempts. We report that although there is an assumed association of suicide attempts with plans and attempts, this is not always the case; suicide attempts were not always associated with thoughts and/or plans. However, the probability of having plans was doubled for those with thoughts, and those with plans had a much higher probability of making attempts. However, this is a cross-sectional situation. We had no information on the sequential progression over time in a given individual. This would be best addressed by both qualitative and longitudinal studies. However, our findings underline the importance of vigilance regarding any suicidal thoughts and, in particular, suicidal plans. As not all suicidal attempts are associated with thoughts or plans, vigilance should also be applied to the associations we report here. We have demonstrated that there are sociodemographic and environmental factors that are significantly associated with attempts and that also predict attempts. There are also environmental factors related to the availability of certain methods that could be manipulated to mitigate suicide.
Response rate
The high response rate of 95.6% (n = 2534 out of 2652 participants) was not unique to this study. Similar rates have been consistently found in Kenyan studies, particularly those involving school-going children.Reference Ndetei, Khasakhala, Mutiso, Ongecha-Owuor and Kokonya18,Reference Ndetei, Khasakhala, Mutiso, Ongecha-Owuor and Kokonya19 There are several explanations, including our approach of explaining the nature of the study to schools and communities to achieve buy-in, and the willingness to participate in an activity that has the potential to improve mental health – specifically, the mental health of students – and in turn improve academic performance.
Prevalence of thoughts, plans and attempts
The overall prevalence of 26.8% suicidal thoughts was well above rates reported for Western countries but within the range reported for other African countries, as referenced earlier in this paper. The disparity in prevalence between these regions could be due to socioeconomic factors.Reference Ndetei, Mutiso, Weisz, Okoth, Musyimi and Muia1 African countries often face greater economic instability, poverty and lack of access to basic resources such as healthcare, education and employment opportunities, which can contribute to higher levels of stress and hopelessness, increasing vulnerability to suicide. Furthermore, mental health services and resources are often limited in these countries, and the stigma surrounding mental illness may also prevent individuals from seeking help or accessing appropriate care even where services are available.Reference Ndetei, Mutiso, Maraj, Anderson, Musyimi and McKenzie20,Reference Mutiso, Pike, Musyimi, Rebello, Tele and Gitonga21
However, the prevalence rates of 14.9% for plans and 15.7% for attempts were above those reported in the literature, as indicated earlier. These may reflect the prevailing situation, given the timing of the study with respect to events including the COVID-19 pandemic; media reports of war in Europe; an election year in Kenya, with threats of violence associated with past elections; and the consequences of climate change, presenting in Kenya with unusual extremes of temperature, droughts (failure of crops, death of livestock) often alternating with deadly floods.22 Even those students who did not have personal experience with these factors would have seen many reports in the media, potentially leading to vicarious trauma. In addition, the sample included students in the final year of high school, facing high-stakes exams and thus pressure to excel. Not all suicidal attempts were associated with thoughts and plans. This poses a challenge in predicting suicidal attempts, and there is thus a need for reliance on other associations of suicide attempts, as well as simple screening of mental health in schools.23
Methods in suicidal plans and attempts
Methods were similar for plans and attempts, and they appeared to reflect availability and ease of access; this suggests that an important step toward preventing suicide may be to limit access to those methods. Our study found taking chemicals and poisons to be the leading method reported in plans and attempts in the Kenyan context, a finding similar to that of the WHO report referenced earlier. This finding is not surprising, because in Kenya these chemicals are widely available in homes and at agro-vet outlets for the control of vermin at homesteads and farms, and as herbicides and pesticides for agricultural activities such as the farming of tea, coffee and other crops and of livestock. These are major sources of income for families and the country. There is, therefore, a need for alternatives that are less injurious to human beings, coupled with regulations on availability and access and appropriate secure storage at home. This should be amplified by an awareness campaign targeting communities and families, especially those who use such poisons and chemicals for the economic activities specified above. The same general principles should be applied in relation to drug overdose (the third most common method), especially in relation to regulated access and availability, for prescribed drugs in particular, as well as involving caregivers at home in the oversight and safe custody of prescribed drugs at home. It is noteworthy that participants with a drug overdose suicide plan had two times greater odds of suicide attempt, reflecting easy access to these drugs.
It may be especially challenging to prevent suicides by hanging or cutting because of the wide range of options available at home and outside home. Raising caregiver awareness is paramount. To some extent, jumping from heights (buildings, bridges, etc.) can be mitigated by installing barriers on parapets, windows and other areas where people are likely to jump. It was noteworthy that in our study, having a jumping plan was a significant predictor of suicide attempts but in a negative direction; that is, participants with a jumping suicide plan had lower odds of attempting suicide than those who did not have a suicide plan. Worrisome, though, is the fact that methods traditionally associated with settings outside Kenya – cutting oneself, for example – are beginning to enter the scene in Kenya.
The fact that the students we surveyed had survived their attempts and could thus tell their stories may be a reflection of the mild severity of the attempts or the low lethality of the methods used. Overall, the findings suggest that exerting control over students’ environment at home, school and community may go a long way towards preventing youth suicide, and public and family awareness of methods could contribute to the prevention of suicide.
Mitigating the association of suicidal attempts with thoughts and plans
First, our findings suggest very strongly the need to take suicidal thoughts and plans seriously and on their own merit. This is because those with suicidal thoughts have an increased prevalence of plans, and those with plans have an increased prevalence of attempts (with the caveat that this was a cross-sectional population study, and we had no information on sequential or different staring points for a given individual). Second, our findings suggest that focusing only on thoughts and/or plans as necessary precursors of attempts may be inadequate, as nearly one-third of attempts appeared to be unanticipated on the basis of thoughts and plans.
Sociodemographic and environmental factors associated with suicidal attempts
We emphasise sociodemographic factors and other environmental factors because these are key to potential entry points for interventions; in particular, environmental factors that are amenable to manipulation as part of mitigation, for example, the availability of chemical and poisons, and the need for caregivers to be aware of various methods and hotspots in the community and at home. To this end, our findings confirm literature reports, including our own earlier findings,Reference Ndetei, Mutiso, Weisz, Okoth, Musyimi and Muia1 that older age (here reflected by increasing seniority of high school students) and urban rather than rural areas are associated with suicidality. Female gender and age are independently associated with suicidal attempts. In the Kenyan context, an important finding is that form 4 students, the most senior students in high school, are particularly vulnerable to suicide attempts. This final year is the most stressful as the students prepare for their National Examination, which will affect their future educational and career opportunities, i.e. whether or not they will qualify to continue their education by gaining admission to colleges and universities. Our study provides more information on environmental factors related to available methods; these environmental factors could be manipulated to mitigate suicidal attempts and suicide.
In summary, suicidal thoughts, plans and attempts are prevalent in Kenyan high school students. There are multiple methods with different frequencies of use, with taking chemicals and poisons being the leading method reported in plans and attempts. There are different starting points before suicidal attempts. Although the most common and well-documented starting point is from suicidal thoughts to plans to attempts, there are other starting points to attempts that do not include thoughts and plans. These other starting points may not lend themselves to detection at clinical and community levels except to the extent that various other associations can be identified, in particular, sociodemographic characteristics and environmental factors related to the availability of certain methods. There is, therefore, a need for reliance on the other various associations of suicide attempts. Being a senior-level student and going to a school in an urban area were associated with increased suicide attempts, whereas being in the final year of high school and female gender were independently associated with suicide attempts.
Strengths and limitations of the study
This was a cross-sectional study; thus, no causalities were drawn. Neither could we determine the different starting points for a given individual. As we used a self-report survey, there was a likelihood of social desirability bias, where participants might have responded to questions in a way that they perceived to be socially acceptable or desirable, rather than providing honest or accurate answers. Although this study was conducted in groups meeting in classrooms, with students filling out the questionnaire at their own pace, the students were given the option not to answer any question they did not want to answer. We only worked with the responses that the students were willing to provide, so it is possible that our findings would have been different in some respects if we had had all the information that some students chose not to provide. This study may not represent Kenyan high school students generally, as it was conducted in only three counties, which do not fully account for sociocultural differences across Kenya. In addition, the very sad fact that only those whose suicide attempts were unsuccessful could participate in the study meant that we were not able to assess associations of lethal suicide attempts. The focus of this study was on the prevalence of thoughts, plans and attempts, methods and modifiable environmental factors. It did not explore the mental health aspects. A major strength is the high response rate, as discussed above.
Recommendations
To understand the sequential progression from thoughts to plans to attempts we recommend a prospective study using mixed methods, i.e. qualitative and quantitative.
Data availability
Requests for the data may be sent to the corresponding author.
Acknowledgements
This study was part of an ongoing global multicentre study coordinated by Turku University, Finland, who provided funding for ethics, questionnaire generation, data collection and data entry. The Shamiri Institute supported the logistics of data collection. Africa Mental Health Research and Training Foundation provided staff to coordinate study activities.
Author contributions
D.M.N. – initial development of the protocol, conceptualisation for the Kenyan context, oversight of data collection, drafting of the paper; D.W. – critique of the manuscript; V.M. – oversight of data collection; J.R.S. – critique of the manuscript; C.M. – oversight on ethics; P.N. – literature review, drafting of the paper; T.M. – fieldwork supervision during data collection; M.H.S. – critique of the manuscript; T.L.O. – oversight of data collection and critique of the manuscript; N.E.J. – oversight of data collection and critique of the manuscript; P.M. – critique of the manuscript; K.B. – critique of the manuscript; S.G. – initial development of the protocol and critique of the manuscript; J.R.W. – critique of the manuscript; A.J. – critique of the manuscript; A.S. – initial development of the protocol and critique of the manuscript.
Funding
Turku University provided ethics support and seed funding for data collection. Africa Mental Health Research and Training Foundation provided in-house support for data collection, analysis and write-up. K.B. is part supported by NIHR Thames Valley and Oxford Applied Research Collaboration (ARC) and the Oxford Health Biomedical Research Centre (BRC). The findings do not reflect the views of NIHR-ARC or NIHR-BRC.
Declaration of interest
K.B. is a member of the BJPsych Open editorial board. They did not take part in the review or decision-making process for this paper.
eLetters
No eLetters have been published for this article.