Institutional review board approval was obtained prior to initiating the study procedures. Eligibility criteria included: adults ( > 18 years old) with chronic ( > 1 year post injury), traumatic SCI with non-complete recovery. Individuals were recruited from a pool of participants from longitudinal studies of health outcomes after SCI. Originally, the participant cohorts were identified from a regional specialty hospital in the southeastern United States; the first cohort enrolled in 1997–1998 and the second enrolled in 2007–2010. An additional cohort was identified from a state-based surveillance system in the southeastern United States and added in 2012–2016. All participants were previously enrolled in follow-up studies of persons with SCI. For the current study, 1407 participants were invited to participate. There were a total of 918 respondents (65.2% response rate), 25.5% of whom were from the state-based surveillance system.

In 2021–2022, potential participants were sent cover letters explaining the study and informing them of forthcoming study materials. The cover letter contained a link to the online REDCap self-report assessment (SRA), which allowed participants to complete the materials virtually. A paper form of the SRA was mailed 4–6 weeks later for all individuals who did not complete the materials in REDCap. A second mailing was used to promote response, followed by a phone call from the study coordinator. For individuals who agreed to participate but had discarded or misplaced the SRA, another set of materials were mailed. Return of completed materials implied consent, and a waiver of signed consent was obtained from the IRB. Participants were offered $50 in remuneration.

The SRA included items on demographic characteristics (age, race-ethnicity, sex), SCI characteristics (injury severity, years since onset), socioeconomic indicators (education and income), and prescription medication items from the National Survey on Drug Use and Health (NSDUH) [19]. We identified self-report of benzodiazepines use based on the tranquilizer and sedative screeners, as well as self-reported prescription pain relievers (opioid) use.

Age was broken down into three categories: <50, 50–64, ≥65. Race-ethnicity was grouped as follows: non-Hispanic White, non-Hispanic Black, and Hispanic and non-Hispanic other. Neurologic level was broken down into four categories, the first of which was all ambulatory, and the other three categories broken down according to neurologic level as follows: high cervical (C1–C4), low cervical (C5–C8), and noncervical. Years since injury was broken down into the following categories: 1–9 years, 10–19 years, and ≥20 years. Years of education was broken down into three levels: high school diploma/GED or less, trade school/associate’s degree, and bachelor’s degree or higher. Family income was broken down into three levels: <$25,000, $25,000–74,999, and > $75,000.

To assess prescription benzodiazepine use, participants were asked, “In the past 12 months, which, if any, of these sedatives or tranquilizers have you used?” and asked to indicate the frequency of use as Never, Once or twice, Monthly, Weekly, or Daily/almost daily. The SRA informed them that “tranquilizers are usually prescribed to relax people, to calm people down, to relieve anxiety, or to relax muscle spasms; and that some people call them nerve pills.” Prescription sedatives were defined as “drugs that are also called sleeping pills or downers. People take these drugs to help them relax or help them sleep.” The specific benzodiazepines included: (a) Alprazolam (Xanax®, Xanax® XR); (b) Lorazepam (Ativan®); (c) Clonazepam (Klonopin®); (d) Diazepam (Valium®); (e) Flurazepam (Dalmane®); (f) Temazepam (Restoril®); and (g) Triazolam (Halcion®).

For the opioid items, participants were asked, “In the past 12 months, which, if any, of these prescription pain relievers have you used?” and reported the frequency of use (Never, Once or twice, Monthly, Weekly, or Daily/almost daily). Participants were asked about their use of prescription pain relievers, because they were more likely to understand the term “pain relievers” rather than “opioids,” since “pain relievers” indicates the purpose for which the drugs are likely to be taken. Eleven common opioids comprise the item set. The methods of assessment have been documented in a previous manuscript [20].

Based on participant responses and regularity of use, we combined “once or twice” with “monthly” use, and “daily/almost daily” use with “weekly” use; we refer to once/monthly or weekly/daily use of the drugs in the cross-tabulation and the logistic regression model. Our operational definition of concurrent use included those individuals who indicated “weekly” or “daily/almost daily” using both any opioid and any benzodiazepine. Although individuals who indicated using these medications “monthly,” or “once or twice,” may also experience risky concurrent use, due the greater uncertainty of overlap in the frequency of use, they were not included in our definition.

Missing data were examined and handled based on two criteria: First, if an individual skipped all of the specific drugs in the benzodiazepine or opioid modules, they were classified as missing (n = 13 (1.4%) were missing all benzodiazepine items, and n = 17 (1.9%) were missing all opioid items. Among those 30 individuals, 4 skipped both sections; in total, 26 individuals were missing benzodiazepines or opioids and excluded from the prescription specific descriptive analyses, therefore we report n = 905 for the presentation of benzodiazepine use groups, and n = 892 for the cross-tabulation of use). Second, if an individual answered one or more specific drugs in either module, while leaving other drugs blank, those missing drugs were assumed to be “no/never use,” and classified as such.

STATA Version 16.1 was used for all statistical analyses. First, we examined the frequencies of use of each prescription medication, then categorized use of benzodiazepines and opioids into any use (yes/no), once/monthly use (yes/no), and weekly/daily use (yes/no). Because participants may report taking more than one prescription benzodiazepine or opioid, they may fall into both categories. We also assessed concurrent use by defining those who responded “yes” to weekly/daily opioid and benzodiazepine usage. Descriptive statistics (n, %) and a contingency table of frequency of use are presented.

The relationships between each variable and the dependent variable (weekly/daily benzodiazepine use) were analyzed. We ran a simple logistic regression and a full multiple logistic regression model of binary indicators of weekly/daily benzodiazepine use (yes/no) using logit model estimation methods. The full model included age, years since injury, SCI injury severity, race/ethnicity, Military Veteran status, education, household income, and weekly/daily opioid use as covariates. Odds ratios (OR) and 95% confidence intervals are reported.