The impact of care environment on the mental health of orphaned, separated and street-connected children and adolescents in western Kenya: A prospective cohort analysis

Introduction The effect of care environment on orphaned and separated children and adolescents’ (OSCA) mental health is not well characterised in sub-Saharan Africa. We compared the risk of incident post-traumatic stress disorder (PTSD), depression, anxiety and suicidality among OSCA living in Charitable Children’s Institutions (CCIs), family-based care (FBC) and street-connected children and youth (SCY). Methods This prospective cohort followed up OSCA from 300 randomly selected households (FBC), 19 CCIs and 100 SCY in western Kenya from 2009 to 2019. Annual data were collected through standardised assessments. We fit survival regression models to investigate the association between care environment and mental health diagnoses. Results The analysis included 1931 participants: 1069 in FBC, 783 in CCIs and 79 SCY. At baseline, 1004 participants (52%) were male with a mean age (SD) of 13 years (2.37); 54% were double orphans. In adjusted analysis (adjusted HR, AHR), OSCA in CCIs were significantly less likely to be diagnosed with PTSD (AHR 0.69, 95% CI 0.49 to 0.97), depression (AHR 0.48 95% CI 0.24 to 0.97), anxiety (AHR 0.56, 95% CI 0.45 to 0.68) and suicidality (AHR 0.73, 95% CI 0.56 to 0.95) compared with those in FBC. SCY were significantly more likely to be diagnosed with PTSD (AHR 4.52, 95% CI 4.10 to 4.97), depression (AHR 4.72, 95% CI 3.12 to 7.15), anxiety (AHR 4.71, 95% CI 1.56 to 14.26) and suicidality (AHR 3.10, 95% CI 2.14 to 4.48) compared with those in FBC. Conclusion OSCA living in CCIs in this setting were significantly less likely to have incident mental illness, while SCY were significantly more, compared with OSCA in FBC. The Orphaned and Separated Children’s Assessments Related to their (OSCAR’s) Health and Well-Being Project is a two-phase longitudinal cohort investigating the effects of care environment on the physical and psychosocial well-being of OSCA in Uasin Gishu (UG) County, Kenya. Phase 1 ran from 2010 to 2015 and Phase 2 ran from 2016 to 2019. The study enrolled participants aged 18 years or less from May 2010 to April 2013. The OSCAR cohort comprises participants from communities within eight administrative locations in UG County and includes 300 randomly selected households caring for OSCA (FBC), 19 Charitable Children’s Institutions (CCIs) (institutional care) (of 21 in the county at the time of study start-up) and a convenience sample of 100 street-connected children and youth (SCY) in ‘self-care’ on the streets. Children were eligible to participate in the study if they were resident of the care environment at the time, irrespective of orphan status (non-orphaned children were included in order to reduce the risk of stigma against orphans and to provide a small, nested comparator group of non-orphaned children in these same environments), and irrespective of the cause of orphanhood (ie, HIV and all other causes). The present analysis included participants with at least one follow-up visit with a psychosocial assessment. In-depth details about the OSCAR cohort’s study design, setting and recruitment have been previously reported.24 This study conforms to the principles embodied in the Declaration of Helsinki. Written informed consent for participation was provided by the head of household, Director of the CCI, or in the case of SCY, by the District (now County) Children’s Officer. Individual written informed assent was provided by each child aged 7 years and above. Fingerprints were used for both children and guardians who were unable to sign or write their name. This study used community-based, participatory processes to inform the research questions, hypotheses and methods, detailed elsewhere.24 To summarise briefly, the Children’s Officers in the region and representatives from CCIs were initially consulted prior to the funding application. They were requested to provide input as to whether such a study would be important from their perspective, and what their priority questions and concerns were. In addition, traditional community assemblies were held in some of the target communities to identify community concerns and priorities with respect to the care of orphaned and vulnerable children. These assemblies were also held following the initiation of the study to maintain regular contact with the community and disseminate findings. We formed an Advisory Board early on, consisting of representatives from communities, CCIs and Children’s Officers, and this board met regularly throughout the life of the study. Our study disseminated findings through the monthly Uasin Gishu Children’s Services Forum, through additional traditional community assemblies and through the study website (https://www.oscarcohort.comhttps://www.oscarcohort.com). Data collection processes were conducted in situ at CCIs and at the OSCAR Project clinic for participants from households and SCY. Annually (semiannually for SCY), participants completed a standardised clinical encounter and those ≥10 years of age also completed a psychosocial encounter. The clinical encounter was an enhanced well-child ‘check-up’ that included a complete physical history and review of health symptoms. The psychosocial encounter measured education and employment, material well-being, behaviours and risks, peer and family relationships, and mental health. The psychosocial assessment was self-administered for those who could read and write, or psychologist-administered for those that could not adequately read or write. In OSCAR Phase 2, two versions of the psychosocial encounter were employed: one for adolescents aged 10–17 years and one for young adults (≥18 years of age) using age-appropriate validated scales and tools to assess PTEs and mental health outcomes. A clinical psychologist was always available during the assessments to assist in case of questions, lack of understanding or distress. Follow-up of cases requiring individual counselling or social work took place on a case-by-case basis as needed, by study staff. The primary exposure of interest was care environment (institutional, family-based or street-based), determined by a participant’s living circumstances at enrolment.21 Separated children were defined as those whose biological mother or father was potentially alive, but functionally not part of the child’s life. Sociodemographic characteristics, ascertained through the clinical encounter, included age, sex, orphan/separated status (maternal, paternal, both or neither, in the case of non-orphaned children living in these environments), HIV status (positive, negative, unknown) and time living with caregiver at baseline (<6 months, 6–2 years, 2–5 years, >5 years, all the child’s life). PTEs (physical abuse, sexual abuse, emotional abuse and bullying) were ascertained through the psychosocial encounters. PTEs for those less than 18 years of age were ascertained using the Child Abuse Screening Tool for Children at Home (ICAST-CH) which measures violence against children.25 The OSCAR study used 11 questions from ICAST-CH based on extensive consultations with Kenyan psychiatrists, psychologists and paediatricians. There were four questions specific to the emotional domain, three to the physical abuse domain and four questions specific to sexual abuse domain. For all questions, responses took on four levels and included ‘Never’, ‘Sometimes’, ‘Many times’ and ‘Not in the past 6 months, but this has happened’. For participants age ≥18, a history of emotional, sexual and physical abuse was assessed using the Childhood Trauma Questionnaire (CTQ) through a five-item subscale: emotional, sexual, and physical abuse, emotional, and physical neglect, as well as three-item subscale to screen for false-negative trauma reports.26–28 For all participants, bullying was measured using the eight items from Strengths and Difficulties Questionnaire,29 30 one item from the Social and Health Assessment Peer Victimisation Scale,31 and one question regarding bullying due to orphan status developed by the research team. If participants answered yes to any of these items, a participant was categorised as having experienced bullying. For this analysis, we dichotomised (yes/no) the PTE variables to capture any history of physical abuse, sexual abuse, emotional abuse or bullying at baseline and at each follow-up. Hypothesised factors that may mediate or confound the relationship between care environment and mental health outcomes included: social support (continuous score measured using the 12-item Multidimensional Scale of Perceived Social Support),32 33 importance of religion in a participant’s life, having basic material possessions (shoes, blanket and at least two pairs of non-school clothes),34 participation in sports and spending time in nature. These factors were measured only on the psychosocial encounter administered to participants 10–17 years of age in OSCAR Phase 2, with the exception of ‘importance of religion’ that was measured in Phase 1 and Phase 2. They assessed these issues currently and are not able to establish temporality with regard to which came first, care environment, outcome or the third factor. We tested whether there were major differences between these factors and outcomes of interest, and differences between these factors and care environment, and concluded that it would be most appropriate to treat them as potential confounding variables. Depression for participants aged 10–17 years was measured with the Child Depression Inventory Short-Form (CDI-SF) with questions specific to the past 2 weeks.35 36 In order to approximate a Diagnostic and Statistical Manual (DSM)-IV Test Revision (TR) diagnosis of depression, scores were summed and a cut-off point of 8 was used to determine probable presence of depression in childhood.37 Depression for participants ≥18 years of age was measured using the Patient Health Questionnaire (PHQ)-9 (Depression) on a four-point Likert scale. Presence of probable depression in those ≥18 years was diagnosed if a participant scored at least 2 (more than half the day) on two or more of the eight items, of which one had to be either item 1 or 2, consistent with the DSM-IV TR criteria.37–39 Adolescent post-traumatic stress was measured using Amaya-Jackson’s ‘Child PTSD Checklist’, a 28-item scale derived from the DSM-IV criteria, which uses a four-point Likert severity scale.40 41 Participants were asked to imagine their worst trauma, and a diagnosis of probable PTSD in adolescents occurred when participants reported currently meeting three diagnostic criteria as indicated by the DSM-IV TR: re-experiencing of the event (a score of 2 or more on any of items 1–5, 10, 11, 14), avoidance symptoms (a score of 2 or more on any three of items 7–9, 12, 13, 22, 25, 28) and arousal symptoms (a score of 2 or more on any two of items 15–21).37 Post-traumatic stress in young adults was measured using the Post-Traumatic Diagnostic Scale (PDS). A diagnosis of probable PTSD in young adults occurred when participants met the same three diagnostic criteria: re-experiencing of the event (a score of 1 or more on items 1–5), avoidance symptoms (a score of at least 3 or more on items 6–12) and arousal symptoms (a score of at least 2 of items 13–17).42 43 Suicidality was measured in adolescents using one question from the CDI-SF scale to ascertain whether they had any suicidal ideation or attempt within the previous 2 weeks. The variable was categorised as no suicidality, suicidal ideation, and suicidal intent, and dichotomised as suicidality (yes/no) in this analysis by combining ideation and intent. Adult suicidality was measured using the PHQ-9 (Suicidality).44–46 A diagnosis of suicidality for adults occurred when a participant scored 1 or above. Adolescent anxiety was measured using the 28-item Revised Children’s Manifest Anxiety Scale (R-CMAS).47 Participants were asked about current thoughts and feelings. A cut-off of 19 was used determine presence of probable anxiety based on the participants score on the R-CMAS. We report mean values and SD or frequencies and percentages for continuous and categorical characteristics, respectively, overall and by care environment. We conducted survival analysis to assess the impact of care environment on four probable diagnostic events: (1) PTSD, (2) depression, (3) anxiety and (4) suicidality. We implemented approaches for recurrent events to account for the fact that individuals may have experienced the events of interest more than once during study follow-up. Time zero was either enrolment into the study or the first visit after the child turned 10 years. For each diagnostic outcome, we present overall survival by care environment using Kaplan-Meier estimates of the cumulative incidence. We use the Prentice-Williams-Peterson (PWP) conditional survival model for recurrent events to evaluate unadjusted and adjusted effects of care environment on each of the four diagnostic outcomes. Adjustment variables were for age, sex, orphan status, length of time in care environment, PTSD/depression/anxiety/suicidality at baseline, emotional/physical/sexual abuse at baseline and during follow-up, bullied at baseline and during follow-up, and clustering by household/institution. The PWP model is a Cox-extended survival model accounting for clustering due to within-subject correlation and stratifying on the order in which the recurrent events occurred. Robust SEs for the model estimates were calculated to also adjust for clustering by care environment. Results from models are presented as HRs with 95% CIs.48 49 We also conducted a subanalysis, restricted to participants aged 10–18 years in CCIs and FBC (data were not collected on SCY) using data from OSCAR Phase 2 to test whether the relationship between care environment and the outcomes were confounded by social support, basic material possessions, importance of religion and time in nature. There was no imputation for missing data. All analyses were conducted using RStudio V.1.3.1056. The funder of the study had no role in study design, data collection, data analysis, data interpretation or writing of the report. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.