Sample

A total of 244 adults (18 years or older) with OUD who were initiating office-based buprenorphine treatment at one of five Federally Qualified Health Centers (FQHCs) in Eastern Pennsylvania and New Jersey were enrolled in one of two randomized, controlled trials (Study 1 n = 159; Study 2 n = 85). Participating FQHCs were located in urban and suburban areas and provided comprehensive primary care services including mental health treatment. Both studies were designed to evaluate the effectiveness of providing adjunctive psychosocial treatment (i.e., cognitive-behavioral therapy and peer support services) to patients receiving buprenorphine treatment for their OUD. Because the studies targeted the same population, utilized the same inclusion criteria and research assessments, and employed similar methodologies, baseline data from the two studies were combined to form the analytic sample for the current paper. Study 1 began enrollment in July 2020 and Study 2 began in October 2020; both studies were actively recruiting at the time of this analysis. As seen in Table 1, 68% of participants identified as male, 38.5% as White, 38% as Black, and 25% as Hispanic or Latinx. At baseline, participants reported a mean age of 43.23 years (SD = 11.16). The most common medical conditions endorsed were chronic pain (37%), hypertension (36%), and respiratory problems (29%). On average, participants reported having 2.16 conditions (SD = 1.86; range = 0–9).

Table 1 Characteristics of Participants with Opioid Use Disorder (N = 244)

Across both studies, individuals were eligible to participate if they had moderate to severe OUD based on the number of symptoms endorsed, had initiated office-based buprenorphine treatment in the past 28 days, and displayed no evidence of psychiatric instability or cognitive impairment. Potentially eligible participants were identified by FQHC providers or staff and then introduced to trained, onsite research staff. Interested individuals were screened for eligibility, informed about the study, and then guided through the informed consent process. Following randomization, participants completed the baseline assessment battery that included a number of standardized assessments including the Addiction Severity Index-Lite (ASI-Lite; 14; see description below) and provided a urine specimen to screen for recent substance use. Participants were compensated $50.00 for their time. Studies were approved by the lead institutions’ review boards.

Transparency and openness

These trials are currently ongoing and upon completion, we will report how we determined our sample size, any data exclusions or manipulations, and all measures in the study. Data from all participants enrolled in the studies to date were examined; cases with missing data for the predictor and/or outcome variable were excluded from the corresponding logistic regression analysis. For the present analyses, no exclusions or manipulations were conducted. Data and code will be available upon study completion at https://doi.org/10.5061/dryad.stqjq2c5n.

Measures

All data used in the present analyses were obtained from the ASI-Lite [14], a reliable and valid multidimensional assessment of current and lifetime psychosocial functioning across seven domains commonly impacted by substance use (i.e., medical, employment, psychiatric, alcohol use, drug use, legal, and family/social). Although the instrument can be used to produce summary scores for each domain, individual binary items were used in the analyses including those measuring:

Lifetime [1] thoughts of suicide (seriously considered a plan for taking their life and [2] lifetime suicide attempts;

Years of heroin and other illicit opioid use and years of polysubstance use (including alcohol);

Presence of a chronic pain-related condition and other medical conditions;

History of physical abuse (caused person physical harm) and history of sexual abuse (forced sexual advances/acts);

History of serious depression and anxiety; and

Demographic information including gender identity, race, ethnicity, employment, education, legal history, and age.

Data capture and statistical approach

All study data were collected and managed using REDCap (Research Electronic Data Capture) hosted at the University of Pennsylvania [15]. Descriptive statistics were calculated to characterize the sample at study entry and to generate prevalence rates for lifetime suicidal ideation and attempts. Next, a series of bivariate correlations were calculated between each of the binary suicide-related response variables (i.e., lifetime ideation, lifetime attempts) and clinical and demographic factors that may be associated with suicidality (age, education completed [years], gender identity [male/female with additional responses added for non-binary, etc.], race [White, Black, Other], ethnicity, years of heroin use and other opioid use, lifetime history of depression or anxiety [yes/no], physical and/or sexual abuse [yes/no], presence of a chronic pain-related condition [yes/no], parole/probation [yes/no], days worked [past 30 days], and lifetime years of daily use of more than one substance [including alcohol]). Finally, a series of logistic regression analyses were performed to predict each of the suicide-related binary response variables. Predictors that reached statistical significance (i.e., p < . 05) for at least one of the suicide-related response variables were included in regression models.