Secondary analysis of retrospective and prospective reports of adverse childhood experiences and mental health in young adulthood: Filtered through recent stressors

Background: Evidence has identified the detrimental effects that adverse childhood experiences (ACEs) have on outcomes across the life course. We assess associations between prospective and retrospective ACEs and mental health in young adulthood and the influence of recent stressors. Methods: Secondary analysis of a sample of 1592 young adults from the Birth to Twenty Plus cohort, from 1990 to 2013, were assessed throughout their first 18 years for prospective ACEs. Retrospective ACEs and an assessment of mental health were collected at the 22–23-year data point. Findings: Prospective physical and sexual abuse are associated with an increased risk of depression (OR 1·7 [95% CI 1·37–1·93, p = 0·034], and OR 1·8 [95% CI 1·27–2·07, p = 0·018], respectively). Retrospective emotional abuse/neglect is associated with increased anxiety (OR 1·8 [95% CI 1·32–2·36, p = 0·000]), depression (OR 1·6 [95% CI 1·08–2·25, p = 0·018]) and overall psychological distress (OR 1·6 [95% CI 1·18–2·17, p = 0·002]). Prospectively reporting four or more ACEs is associated with a twofold increase in risk for overall psychological distress (OR 2·2 [95% CI 1·58–3.12, p = 0·008]). Retrospectively reporting four or more ACEs is associated with increased likelihood of somatization (p = 0·004), anxiety (p = 0·002), depression (p = 0·021), and overall psychological distress (p = 0·005). Interpretation: Both individual and combined retrospective and prospective ACEs are related to mental health in young adulthood. Recent stressors reinforce this relationship; the likelihood of those who report more ACEs experiencing psychological distress increases when adjusting for recent stressors. Funding: Wellcome Trust (UK), South African Medical Research Council, Human Sciences Research Council, University of the Witwatersrand and supported by the DSI-NRF Centre of Excellence in Human Development. The Birth to Twenty Plus study (BT20+) is a South African birth cohort of 3273 singleton children born to mothers who were residents of Soweto-Johannesburg in a 7-week period of enrolment in 1990. The study is unique in that it is the largest and longest running study of child and adolescent health and development in Africa. Current participants are 30 years old, have been assessed up to 22 times and, since 2005 when the first participant birth occurred, includes the 3rd generation of the cohort. A detailed description of the study, its birth cohort and participants is published elsewhere [29]. This study uses data from birth to age 22–23-years old for prospective and retrospective reports of ACEs and covers the period between 1990 and 2013. A total of 1636 participants were surveyed at the 22–23-year wave and a sample of 1592 participants from this group that had both retrospective and prospective reports of ACEs was included in this analysis. Ethical clearance was obtained from the Witwatersrand University Committee for Research on Human Subjects (protocol number: M140726). All participants and/or their caregivers gave informed written consent for the data reported. Adverse childhood experiences have been defined as physical abuse, sexual abuse, emotional abuse and/or neglect, child separation, divorce or parent separation, parent death, exposure to violence, exposure to intimate partner violence (IPV), chronic unemployment, household substance abuse, household legal trouble, household serious illness or disability, and household death. The ACEs survey questions are included in Supplementary Table A. For prospective reports, caregivers were asked to report on their children at participant ages 5, 7 and 11, and participants provided self-reports at ages 11, 15 and 18. A participant was recorded as having experienced a particular ACE if there was a positive response at any one of these time points. For the retrospective report, participants were asked at the 22–23-year wave to indicate if they had experienced each of the ACEs during the first 18 years of their life. A full detailed account of individual ACEs reported at each of the 7 time points, as well as an analysis of the level of agreement between sources and timing, has been published [14]. In summary, that analysis found the reports of prospective and retrospective ACEs, used in this study, had little overall agreement; 80% of the kappa values were below the moderate agreement cut-off of k = 0·41. The highest levels of agreement were between reports on parental death (k = 0·52) and household death (k = 0·51). Reporting on early life ACEs by caregivers (at ages 5, 7 and 11) showed the greatest concordance with retrospective reports of ACEs on sexual abuse (91·0% agreement), physical abuse (87·7% agreement), and exposure to intimate partner violence (80·2% agreement) [14]. For the purposes of this paper, we conceptualise the ACEs directly impacting an individual – physical, sexual, and emotional abuse – as proximal ACEs, and those occurring in their environment – exposure to IPV, household illness, chronic unemployment – as distal ACEs. For ease of reading, retrospectively reported ACEs may be referred to as ‘retrospective ACEs’ and vice versa; similarly, for individual ACEs we may use the shorthand ‘prospective physical abuse’ rather than ‘prospectively reported physical abuse’ but the method of data collection for all ACEs is either self-reported or parent-reported (in the case of children under the age of 7 years). An assessment of recent stressors, adapted from the Township Life Event Scale [30], was added to the analysis. Participants were asked at the 22–23-year data collection wave to indicate if they had experienced any of 9 negative life events ‒ considered stressors ‒ in the past 6 months. The 9 events included violence in the household (1), workplace (2) or community (3), household illness (4), disability (5) or death (6) in the family, household substance abuse (7), alienation from family (8), and legal trouble (9). Full questions are available in Supplementary Table A. Young adult mental health was assessed using the self-reported GHQ-28 which comprises 4 sub-scales of 7 items each probing for somatic symptoms, anxiety and insomnia, social dysfunction, and major depression. The 28 items are scored in a binary 0011 method. Higher scores on the GHQ-28 represent higher levels of psychological distress. The GHQ-28 is used in epidemiological studies as a screening for minor psychiatric morbidity caseness (clinically significant anxiety and/or depression). Any score above 4 on a subscale and above 23 on the total scale indicates the presence of distress or a positive diagnostic [31]. Data was analyzed using STATA statistical software version 13·0. The ACEs data was transformed into a retrospective and a prospective categorical score for each participant as follows: 0= ‘no reported ACEs’, 1= ‘one reported ACE’, 2= ‘two reported ACEs’, 3= ‘three reported ACEs’ and 4= ‘four or more reported ACEs’. In parts of the analyses outcomes are compared between ‘less than four’ and ‘four or more’ reported ACEs. There are currently no guidelines on the ACEs scoring in the available literature but some studies do point to the ‘four or more’ cut-off functioning as a threshold level, with noticeable deviations in a range of outcomes at that mark [6,32]. The four mental health outcomes, somatization, anxiety, social dysfunction, and depression were transformed into categorical data and the co-occurrence of psychological distress with reports of ACEs was evaluated using the chi-square statistic. Unadjusted effects of each individual ACE, followed by each composite measure of ACEs, separately for prospective and retrospective ACEs, were tested for effects on somatization, anxiety, social dysfunction, depression and GHQ total. Five adjusted logistic regression models were fitted including significant predictors from the unadjusted models, controlling for sex, socio-economic status, maternal education and recent stressors, to estimate the association between the ACE scores and each outcome. Odds ratios and 95% confidence intervals were calculated separately for each outcome. In the fully adjusted models, retrospective and prospective ACEs are entered in the same model together with the selected covariates, therefore the ORs for prospective ACEs indicate the contribution of prospective ACEs independently from retrospective ACEs and vice versa. Factorial analysis of variance was used to test for the unique contribution of prospective and retrospective reports of ACEs, as well as any interactions between them, to the variance in each mental health outcome. Regression analysis was used to test for mediation and moderation effects of recent stressors on mental health outcomes. Factorial analysis of variance tested for interaction effects between retrospectively and prospectively reported ACEs and recent stressors on psychological distress. These analyses yielded no significant results and the recent stressors were subsequently added to the regression models as a covariate. The funder of the study had no role in the study design, data collection, data analysis, and data interpretation, or writing of the report. All authors had full access to the data and accept final responsibility to submit for publication. Access to the data is available to all authors for as long as they are part of the study team. Authors are permitted to keep their own copy of a dataset specific to a publication ad infinitum.