This is an efficacy trial of the Motivational Interviewing (MI) and Educational Training (MIET) program. The approach includes LARC education and brief MI to encourage LARC adoption.
A randomized trial design with pre- and post-assessments will be used to meet study aims.
Women in both the control and treatment conditions will complete a baseline survey that includes the Mini-International Neuropsychiatric Interview (MINI), a contraceptive use survey, and the WHO Quality of Life-Brief (WHO-QOL-B). Women in the control group will be provided a color brochure with easy-to-read information about the advantages of LARC. Via this brochure, control group women will be provided no-cost options and corresponding referral information for receiving LARC. Women randomized to the treatment group will complete the baseline survey and receive the same brochure; however, they also will receive structured education regarding unintended pregnancies among women using opioids, potential complications associated with opioid, alcohol, and other drug use, and education on contraceptive use (with an emphasis on LARC). This structured education will be coupled with a brief, initial MI conversation.
Consistent with the peak years for having unplanned pregnancies, we will recruit women 18-44 years of age, who are reproductively capable (intact uterus, not sterilized), who are classified with opioid use disorder (OUD) based on a \>3 MINI score, and who indicate having male sex partner(s). Women will be excluded if they: (1) are knowingly pregnant; (2) are identified as pregnant from an administered pregnancy test; (3) express a current intent to conceive; (4) are currently using Depo-Provera to avert pregnancy; or (5) already have LARC insertion. Recruitment and intervention will occur sequentially and will be delivered in the office-based opioid treatment (OBOT) clinic.
All patients meeting the inclusion criteria will be approached for recruitment during the fifth week of their induction into MAT. The research coordinator will recruit patients, gathering informed consent and baseline data, as well as randomizing patients to the control or treatment conditions. She will electronically forward screens to the treatment group interventionist to expedite the intervention by providing data on opioid use, pregnancy risk, and participant quality of life. The control group interventionist will know only that the patient is eligible for LARC and will not receive the baseline data. The research coordinator will collect all survey data at baseline and up to five days before, or seven days after, the one- and six-month follow-up time points. The consent form will request permission to acquire Electronic Health Record (EHR) data from the clinic for a one- and six-month extraction to determine whether the women experienced unplanned pregnancies, had contraceptive consultations, and received LARC implant/insertion.
Study Design and Flow. Figure 1 provides a flow diagram describing the study. Women in both conditions will be notified that they are eligible for a free LARC implant/insertion if they have interest. Upon consent, women will be provided pregnancy tests, and those who are pregnant will be informed and referred for prenatal services.
We anticipate 90% retention at month one (i.e., N = 36 per group). This figure is consistent with current retention rates for patients in the clinic who are retained for four weeks. As primary one- and six-month endpoints are collected via chart extraction, loss to follow-up is not an issue for the outcomes of interest, receipt of contraception consultation, and LARC implantation (i.e., the available N for this final assessment will be 80), as this data is maintained in the EHR and represents the primary set of dependent variables for the RCT. Self-reported measures serve as study covariates and data from follow-up assessments (one and six months) will serve as dependent variables for a sensitivity analysis.
Randomization will be achieved by using a stratified method described by Suresh.42 The strata will include women who have insurance, opposed to those who do not; the "strata" will be one of two blocks. We chose to stratify, as women without insurance will have the added step of seeking an implant from a local provider that offers free LARC implant/insertion, thus adding an additional step to the process and perhaps affecting the outcomes. Within strata, women will be randomized to one of the two study conditions using a computer-generated allocation procedure and a concealment of allocation technique.43,44 The concealment of allocation will be conducted by the data manager generating a random assignment list. Once the sequence has been assembled, it will be uploaded to the REDCap randomization module and employed after the baseline assessment.
Statistical and Power Analysis Power Analysis. We are planning a study examining LARC uptake with 40 experimental and 40 control subjects. For Aim 1 the anticipated percentage point differences in LARC uptake are based on the following estimates. Data from a large retrospective cohort analysis of LARC adoption comparing women with SUDs to those without found that \~6% of women adopted LARC.49 We predict a slightly higher adoption rate among women in the control group (10%) in this study due to their exposure to educational materials in the clinic, including posters and brochures, and their awareness of free access to this services. Preliminary data from our pilot study support a projection of a conservative (40%) uptake probability. If the true probability of uptake among the intervention group is 40% (or a 30-percentage point increase), we will be able to reject the null hypothesis that uptake rates for the intervention and control groups are equal with a power of \~91%. The type I error probability associated with this one-tailed test of the null hypothesis is 0.05. Figure 2 illustrates the power curve for Aim 1 and the expected power given the sample size. Further, estimates are provided given increases or decreases in projected differences in control and intervention. For instance, if the true difference is only 25%, as opposed to the projected 30%, we would still have sufficient power at \~81% power to reject the null hypothesis.
Aim 1 Analysis Plan. As previously noted, the study has one dependent variable, LARC insertion. In Aim 1, this binary outcome will be evaluated using multivariable generalized estimating equations in conjunction with robust (sandwich) variance procedures. All models will include randomization to MIET versus the control group, and we will adjust for any baseline characteristics that may be distributed differentially between the two study groups (which is unexpected with an effective randomization scheme). To account for the clustering of observations within individuals and the lack of independence of the observations, we will use random effects models. In addition to main effects models, we will explore modifiers of the treatment effect by introducing interaction terms into multivariable models.
Regarding missingness, we will exercise two options. First, missingness from the primary analyses will be partly alleviated by using self-reported values from the one-month follow-up assessment, used to inform imputation. Second, we will determine whether missingness is associated with observed data. If we identify predictors of missingness, multiple imputations will evaluate robustness of the primary outcome analyses and report estimates based on imputations that would be valid under the assumption of random missing data. We will impute using chained equations and produce at least five imputed datasets and derive point and standard error estimates for treatment effects and other parameters from the distribution of estimates with imputed datasets.
A sub-analysis will address the question of whether LARC insertion may lead to subsequent retention in MAT. Although not a specific aim of the study, this question is important, as an affirmative answer would provide added evidence that LARC may be a "gateway" into stabilizing the lives of women with OUD. Conversely, it is possible that uptake and adherence to MAT may stabilize women's lives to the point whereby they begin using LARC. Furthermore, we are interested in examining exploratory aims regarding short-term contraceptive-method retention. If women opt not to have LARC, we will monitor retention on shorter-term methods, including birth control pills and Depo-Provera shots. An appropriate regression-based sub-analysis will detect whether LARC uptake predicts MAT uptake and vice versa (depending on the time sequence).
Aim 2 Analysis Plan. Aim 2 consists of tailoring the MIET intervention to the needs of urban women receiving MAT using the ADAPT ITT model. This aim involves data collection from both review of the interventionist's sessions with patients and qualitative data collection from the patients. The first step, tailoring the interventionist's performance to the patient group, will be achieved by applying a nominal group process, with the group consisting of the research team. The process will occur in two rounds: (1) each team member independently will view a video demonstration of MIET delivered by the two staff members trained to implement the treatment group intervention and provide a written critique of the content and delivery; and (2) the second round will occur through a two-day work meeting, with the goal of sharing these written critiques and proposing adaptations to the intervention until the group comes to full consensus.
We will collect patient data for the second step including: (1) satisfaction (e.g., was session interesting to you?); (2) reaction (e.g., how did it change the way you think about using LARC?); and (3) understandability (e.g., did the wording and pacing of the program meet your needs?). Data will be collected in one-to-one, in-depth interviews with 10 study-eligible women post-intervention. Every fifth woman in the treatment and control groups will be asked to participate in these interviews. If a woman refuses, we will ask each subsequent woman until we identify a woman who is willing to participate. All interviews will be conducted and audio-recorded by the research coordinator. A semi-structured interview guide will be used, and standard coding/analytic procedures will be applied. Interview questions will be designed by the research team to elicit participants' reactions to the core elements of the intervention. Whether reactions are positive, neutral, or negative, each elicitation question will conclude by asking for suggestions as to how the content and/or delivery could be improved for greater relevance. Using standard methods of qualitative data analysis, the transcribed data will be analyzed by members of the research team. First, these data will be analyzed with Atlas T.I. using thematic content analysis to describe and code key themes. Second, open coding will occur through line-by-line reading to identify salient themes through the development of a codebook. Third, using this codebook, each person independently will code five interviews. Finally, the team collectively will verify the codes assigned by the original reviewers to allow for inter-rater reliability and use a consensus-building approach when disagreements occur. Once the themes have been identified, the committee will convene to use the study findings as a basis for refining the brief intervention and tailoring content as needed. Findings from the nominal group process involving staff implementation of the intervention and the qualitative data collected from patients will inform intervention adaptation, leading to an intervention tailored for urban, low-income, opioid-using women, and will be tested in a larger study.
