Adverse childhood experiences and comorbidity in a cohort of people who have injected drugs

Study participants were a subset of individuals from the AIDS Linked to the Intravenous Experience (ALIVE) study. Complete details of the ALIVE study, which began recruitment in 1988, are described elsewhere [15]. Briefly, the ALIVE study is an ongoing, community-based, prospective cohort study of people who have a lifetime history of injection drug use, living in and around Baltimore, Maryland. Approximately 2,600 adults were originally recruited in 1988, and subsequent waves of recruitment occurred in 1994–1995, 2005–2008, and 2015–2018. Participants attend twice-annual study visits where they complete a physical exam; complete a standardized interviewer-administered and audio-computer assisted surveys about drug use, related health-risk behaviors, and physical and mental health outcomes; and provide a blood sample for HIV, Hepatitis C (HCV) antibody testing, and other biomarker measurements.

Detailed methods for the adverse childhood experiences sub-study, data collection, and survey instruments can be found in Additional file 1: Appendix A. All ALIVE participants were eligible to participate in the adverse childhood experiences sub-study if they attended a study visit during the period of August 1^st 2018 – December 31^st 2019 and had attended at least one prior ALIVE study visit. Of the 1,127 eligible participants, 735 were recruited for this non-selective convenience sample. A total of 653 participants from three separate recruitment cohorts provided complete data on adverse childhood experiences and were included in the present analysis (see Additional file 1: Appendix B). We compared the analytic sample (n = 653) to those individuals missing any adverse childhood experience data (n = 69) on demographic characteristics (i.e., age, sex, race [Black vs. white], past six-month homelessness and income) and other study characteristics (past six-month injection drug use, HCV and HIV status, cohort group). Compared to those with complete data on adverse childhood experiences, there was a higher percentage of participants who were HIV positive among participants with missing data on childhood adversity, χ² (1) = 5.1, p = 0.02. There were no other differences between participants with and without complete adversity data (see Additional file 1: Appendix C). The Johns Hopkins University Institutional Review Board has continuously approved the ALIVE study and approved the protocol for this sub-study. All study protocol was conducted in accordance with Johns Hopkins University Institutional Review Board’s standards and the Declaration of Helsinki. All participants provide written informed consent upon study entry and provided additional consent to participate in this sub-study.

ALIVE cohort participants are asked at each study visit “Has a doctor or other health care provider ever said you have [condition name].” All conditions assessed – which included diabetes, hypertension, or cerebrovascular, cardiovascular, renal, chronic lung, cancer, or liver disease – were included in this analysis. One additional comorbidity – obesity (defined as a body mass index ≥ 30) – was also calculated and included in this analysis based on measurements taken during the study visit physical exam. These conditions were selected based on known associations with adverse childhood experiences (e.g., obesity) [16], prior comorbidity indices created in the ALIVE cohort [17], and medical conditions included in comorbidity burden indices (e.g., Charlson Comorbidity Index) [18]. Based on the distribution of comorbidities in the study sample, scores were categorized into the following groups: 0 comorbidities, 1–2 comorbidities, and ≥ 3 comorbidities (see Table 1). Within ALIVE, this measure of comorbidity burden has been independently associated with geriatric phenotypes, frailty transitions, hospitalization, and mortality [17, 19, 20].

Participants’ childhood exposure to adversity was assessed using a modified version of the adverse childhood experiences questionnaire (Additional file 1: Appendix D) [21]. Briefly, the questionnaire included 21 questions assessing 14 adverse childhood experiences occurring before the age of 18 years: physical and emotional neglect, physical, emotional, and sexual abuse, loss of a parent, domestic violence, parent substance abuse, parent mental illness, incarceration of a household member, bullying, social ostracization, neighborhood violence, and poverty. For adversities assessed with multiple items, endorsing any one of those items was sufficient to indicate the presence of that adversity. Scores were summed and then categorized into the following groups: 0–1 adversities, 2–4 adversities, 5–9 adversities, ≥ 10 adversities. These cut-offs were based on the distribution of ACEs in the current sample. Specifically, “0–1 adversities” was chosen as the reference category given the large percentage of the sample that reported exposure to at least one adverse childhood experience (~ 50%); other categories were selected based on what we deemed to be adequate group sizes that also reflected conceptually different adversity exposure groups.

We also adjusted for several sociodemographic and clinical characteristics that may confound (e.g., age, sex) or mediate (e.g., HIV status, injection drug use) the association between childhood adversity and adult chronic disease. Sociodemographic characteristics included self-reported age, race (Black versus White), sex, cohort group, past six-month income below $5,000 and past six-month homelessness. Clinical characteristics included self-reported HIV status, anti-HCV status (from antibody test), and past six-month injection drug use.

First, we described the prevalence of adverse childhood experiences in the ALIVE cohort and compared these data with data from Maryland adults (state-wide and in Baltimore City) who participated in the 2018 Maryland Behavioral Risk Factors Surveillance System (BRFSS), which contained an adverse childhood experiences module [22]. We also examined the distribution and co-occurrence (i.e., conditional probabilities) of adverse childhood experiences in our sample. We then evaluated the association between adverse childhood experiences and comorbid conditions. During exploratory analysis, we tested the crude association between adverse childhood experiences and medical comorbidities using an ordinal logistic regression model with robust standard errors. Since the data violated the proportional odds assumption (χ²(3) = 40.23, p < 0.001), multinomial logistic regression with robust standard errors was selected for final models. After a crude model (Model A) was tested without adjustment for covariates, a second model (Model B) was tested, adjusting for demographic characteristics (i.e., age, race [Black vs. White], sex, cohort) and socioeconomic indicators (i.e., income below $5,000/year, past six-month homelessness). The third model (Model C) additionally controlled for current anti-HCV status and HIV status (yes/no), and past six-month injection drug use (yes/no). Analyses were organized in this way to attempt to tease apart whether sociodemographic or clinical characteristics were influencing the association between childhood adversity and adult comorbid conditions. Models will be referred to herein as Model A, B, and C. Covariates in the adjusted model were selected based on known associations with adverse childhood experiences and the crude association with the outcome. Lastly, a sensitivity analysis was conducted with all recruited participants (n = 722), with missing responses for adverse childhood experience items recoded to 0 (i.e., no exposure). This approach was used since minimal adversity data were missing (i.e., < 10% of the sample), and when assuming a response of “yes” to the missing adversity items, few participants (n = 26) moved from on adversity category to another based on our categorical coding strategy. All statistical tests used a p-value threshold of 0.05. All analyses were conducted using Stata version 15 [23].

Results from this study should be considered in the context of several limitations. First, the sample is comprised of a mostly male, mostly Black, urban, east-coast cohort, many of whom came of age during the peak of the HIV epidemic. Thus, this could limit the generalizability of the current findings. ALIVE participants are also unique in that they were recruited because they had a history of injecting drugs. Injection drug use may be a marker for severity of drug-related problems; different effects of adversity might be observed in the broader cohort of adults who use heroin or cocaine but have never injected. Third, adverse childhood experiences were assessed retrospectively. Although there is a temporal sequence from childhood adversity to chronic conditions developed later in life, recall bias might differentially impact different age groups and differentially impact associations between adverse childhood experiences and comorbidity burden. We also did not adjust for all possible confounders in the association between childhood adversity and comorbid conditions. Other social determinants of health, such as food insecurity, diet, housing, and transportation insecurity are associated with exposure to adversity and health and may influence this association. Finally, the Maryland state law requirement that we report items indicating a history of child abuse to the Baltimore Department of Social Services – and the need to notify participants of that requirement – may have led to an undercount of the number of participants who experienced childhood abuse (see Additional file 1: Appendix A). However, the number of participants who declined to respond to the childhood abuse questions was limited (n = 13), and characteristics of participants with and without the missing data on childhood abuse questionnaire were similar.