Sex And Female Empowerment (SAFE): A Randomized Trial Comparing Sexual Health Interventions for Women in Treatment for Opioid Use Disorder

Abstract

Background

Unintended pregnancies are prevalent among women with opioid use disorder (OUD). The Sex and Female Empowerment (SAFE) project developed a social-cognitive, theory-driven intervention to increase acceptance of and adherence to contraceptive practices among women receiving medication for OUD (MOUD). This study evaluated the feasibility and acceptability of two SAFE interventions (Face-to-face and Computer-adapted) compared to usual care as well as their efficacy to improve contraception utilization.

Methods

This pilot randomized trial enrolled 90 heterosexual, non-pregnant, reproductive-age women receiving MOUD. Participants were randomized into either a: SAFE Face-to-face intervention, SAFE Computer-adapted intervention, or usual care (UC) condition (n=30 each) and followed for 6 months. Outcome measures included intervention completion, intervention satisfaction, attendance at a contraception consultation appointment, and long-acting reversible contraceptive (LARC) method receipt. A generalized linear model was used for inferential testing and to estimate least squares means (predicted probabilities for binary outcomes) and their standard errors.

Results

Compared to the UC condition, both the SAFE Face-to-face and the SAFE Computer-adapted intervention had higher intervention completion [Means (Standard Errors) = 0.97 (.03) and 0.97 (.03), respectively, vs. 0.53 (.09); ps<.001], higher intervention satisfaction [Ms (SEs) = 3.7 (.11) and 3.8 (.11), respectively, vs. 3.1 (.11); ps<0.001), higher contraception consultation visit attendance [Ms(SEs) = 0.80 (.07) and 0.73 (.08) vs. 0.33 (.09); p<.001], and greater LARC receipt [Ms(SEs) = 0.77 (.08) and 0.73 (.08) vs. 0.23 (.08); p<.001).

Conclusions

SAFE appears feasible and efficacious for supporting women in contraception decision-making. Integrating SAFE into women’s comprehensive OUD treatment services holds promise to increase contraceptive decision-making and initiation of a chosen method.

1. Introduction

Women with opioid use disorder (OUD) often need tailored, effective interventions to improve contraceptive and safer-sex practices. While the prevalence of unintended pregnancies in the general population in the United States is roughly 50%, it is much higher among women with OUD, ranging in the literature from about 60% (MacAfee et al., 2019a; MacAfee et al., 2020) to over 80% (Black et al., 2012; Heil et al., 2011). Unintended pregnancy is associated with a wide range of negative outcomes for women, children, and their families (Cheng et al., 2009). For women with OUD, these associated risks become more varied to also include outcomes such as prenatal substance exposure (Larson et al., 2019) and possibly compromised OUD treatment retention (Martin et al., 2018).

Major drivers of this high burden of unintended pregnancies among women with OUD are both the need for reproductive autonomy and the low utilization of effective contraception, in part due to concerns about side effects, insertion and removal of methods (Black et al., 2012; Bornstein et al., 2019; Matusiewicz et al., 2017; Rey et al., 2020). The most commonly used contraceptive method by women with OUD is condoms while long-acting reversible contraceptive (LARC) methods are less used (Terplan et al., 2015). Despite increasing awareness of this reproductive health disparity concurrent with attention to the opioid crisis (Wright, 2019), recent studies continue to highlight low contraception uptake among women with OUD (Krans et al., 2018; MacAfee et al., 2020). A paradigm shift in how reproductive/sexual health is conceptualized for women with OUD is needed to effectively address this disparity. This paradigm shift will need to encompass not only a narrow focus on the prevention of and treatment for sexually transmitted infections but also a broader focus on other areas such as sexual health education and reproductive life planning underpinned by choice and autonomy.

Other reasons for low contraception use in this population are complex, involving challenges shared with the general population (e.g., cost, provider knowledge, transportation) in addition to barriers unique to women with OUD. Such unique barriers include stigma/discrimination related to substance use and fear of legal involvement on the part of the women (MacAfee et al., 2019b) as well as low perceived risk of pregnancy while on medications treating OUD (MOUD) and fear of infertility (Bornstein et al., 2019). Integration of reproductive and sexual health services into addiction treatment services is a proposed avenue to overcome these challenges (Wright, 2019). However, significant barriers to successful provision of reproductive/sexual health services within OUD treatment programs exist, including lack of trained staff, supplies, childcare and organizational support (Klaman et al., 2019).

Overall, studies assessing the efficacy of co-localization of reproductive/sexual health services and OUD treatment have shown inconsistent results and are largely focused on pregnant to immediate postpartum patients (Black and Day, 2016; Collier et al., 2019). Such variability in results may reflect inconsistencies across sites in adequately addressing patient barriers to care and coordinating clinical services through transitions in OUD treatment, such as the postpartum period. Additionally, prior studies have largely evaluated only face-to-face single contraceptive educational sessions as interventions and have done so using a variety of methods of outcome measurement.

Co-located services are likely to be successful if services are provided using low-threshold access – that is, provide reproductive/sexual health services in OUD treatment settings with few or no barriers to access (Black and Day, 2016). One successful example of co-located reproductive/sexual health services and OUD treatment involves incentives (Heil et al., 2016). Specifically, in a preliminary controlled trial, women were assigned to the usual care (offered condoms, emergency contraception, and external service referrals) or to the intervention group who received the World Health Organization’s contraception initiation protocol, free prescription contraceptives, and financial incentives for follow-up visit attendance. At 6-month follow-up, uptake of prescription contraception was significantly higher in the intervention compared to usual care group (94% vs. 13%) (Heil et al., 2016). With the increasing burden of the opioid crisis affecting women (Woolf and Schoomaker, 2019), innovative methods to meet reproductive/sexual health needs that are feasible within a busy OUD clinical environment are needed. One such intervention for women with OUD is Sex and Female Empowerment (SAFE), a comprehensive approach that focuses on meeting the sexual health, family planning, and contraceptive needs of this often overlooked patient population of non-pregnant women in OUD treatment.

SAFE is a social-cognitive, theory-driven intervention designed to empower women receiving MOUD and increase acceptance of and adherence to contraceptive practices specific for women on MOUD. Two different delivery formats for the SAFE intervention were created to maximize clinically applicability across settings: Face-to-face and Computer-adapted. Given time and budget demands on MOUD treatment programs, as well as the need for virtual interventions in the COVID-19 pandemic, a computer-adapted intervention may be more acceptable to providers. Many patients may also prefer it over a face-to-face intervention, given its greater anonymity, self-paced completion of materials and information delivery flexibility. Alternately, this sensitive topic may be best addressed in a personal, face-to-face counselor format with established patient rapport. Therefore, the main objective of the current study was to evaluate the feasibility, acceptability, and efficacy of these two SAFE interventions relative to each other and compared to usual care (i.e., the patient is provided a brochure about family planning and contraceptive choices and then given a referral to a contraceptive service provider) in terms of intervention completion, intervention satisfaction, attendance at a contraception consultation appointment, and long-acting reversible contraceptive (LARC) method receipt.

2. Materials and Methods

This study was approved by the University of North Carolina Institutional Review Board and registered at ClinicalTrials.gov (NCT02197715) before enrollment of any participants. This study was designed and is being reported using CONSORT guidelines.

2.1. Participants

All women receiving treatment at two MOUD treatment programs in urban areas of central North Carolina from March 1, 2015 to March 30, 2017 were provided with study flyers and could complete study screening on-site or via phone. Eligibility screening included a 10-minute structured interview to collect basic demographic and other characteristics as well as eligibility information. Participants were offered a hygiene gift basket of items worth approximately $10 (e.g., tampons/pads, condoms, body lotion, soap). Inclusion criteria were: (1) ages 18–40 years old; (2) able to provide informed consent; (3) currently enrolled (and stabilized for more than 90 days) in one of the two MOUD program sites of the study; (4) currently not pregnant (confirmed with urine testing); (5) reported heterosexual orientation; (6) no tubal ligation or other sterilization; (7) no plans to become pregnant in the next 6 months; and (8) provided verifiable locator information. Those screened eligible were offered study participation via an informed consent meeting.

2.2. Treatment Setting

Two clinics that accept self-pay or Medicaid for providing methadone and buprenorphine to treat OUD were sites for the study. The clinics are representative of typical MOUD programs providing men, women, and non-binary patients an array of outpatient treatment services (e.g., screening, assessment and treatment for OUD, development and follow-up of individualized person-centered plans, access to treatment of comorbid mental health diagnoses, on-site individual and group therapy). The intervention occurred in individual office space provided by the respective MOUD clinic..

2.3. Design and Description of Study Conditions

Intervention allocation.

Once informed consent was obtained, participants were block randomized on a 1:1:1 basis into one of the three study conditions: (1) SAFE Face-to-face, (2) SAFE Computer-adapted, (3) Usual Care (UC), using a series of numbered envelopes that contained the random assignment of each participant to one of the three conditions. Neither the research staff nor the participant knew assignment of a participant to an intervention condition until the envelope was opened, at which time the assignment to treatment condition for that participant was revealed.

2.4. Intervention Conditions

Both the SAFE Face-to-face and the SAFE Computer-adapted interventions, described below, were designed to be completed in three separate sessions, each of 45–60 minutes duration and provided as part of the participant’s MOUD treatment visit. However, in order to help support participants in completing the intervention, participants who missed a session were allowed to reschedule for the next week on a day and time of their choosing, so participants had up to 4 weeks to successfully complete the 3-session intervention.

At the end of each intervention session participants were asked to complete a two-minute session evaluation.

SAFE Face-to-face Intervention

Like other behavioral interventions that improve condom use, SAFE is grounded in social-cognitive theory, which posits that an individual must have both the knowledge and the skills necessary to perform a desired behavior and perceived self-efficacy in the ability to perform it successfully (Bandura, 2004; van Empelen et al., 2003). SAFE face-to-face was developed using qualitative data gathered separately from women in treatment for OUD, men in OUD treatment who were not the women’s partners and health care providers who care for women with OUD. The findings from the qualitative data yielded the development of a 4-session intervention that, with pilot-testing feedback, was reduced to a 3-session intervention with each session lasting 45–60 minutes. The intervention utilized Motivational Interviewing (Miller and Rollnick, 2002) for all three sessions. Each session was guided by flip-charts that the interventionist used to guide the discussion. Session one focused on basic body anatomy and reproductive biology as well as exploring the types of contraceptive methods available and the pros and cons of each method. Contraceptive method examples were used to demonstrate how different methods look, feel, and work. Session two focused on a discussion about the participant’s thoughts on which contraceptive method would work best for her, and how to talk about it with her medical care provider and her partner. Tips for how to talk to a doctor about contraceptive options were provided as well as tips for how, where, and when to talk to her partner about condoms and other contraceptive practices. In order to talk to providers and partners, skills of active listening, positive language, “I feel” statements, and safer-sex negotiation were provided and role-played. Physical and emotional safety techniques were provided. Session three focused on developing an action plan and looking at what resources are needed to meet the participant’s immediate and long-term personal reproductive life goals. In doing so, a shared decision-making approach was used to aid women in resolving their ambivalence and changing behavior through various intervention practices (e.g., personalized feedback on contraceptive use, discussing the pros and cons of such use, decisional balance, goal setting). Trained BA to MA level staff of the MOUD program provided the intervention. Staff training consisted of four one-hour training sessions that reviewed the study and the content of the 3-session intervention and provided role-play practices of all intervention components. Staff received feedback on role-plays, and supervision was provided every other week to troubleshoot issues. At the end of each session, a participant was asked if she wanted to complete a plan to decide a method of contraception. If she wanted to decide, she was asked what method(s) she wanted to discuss with a health care provider, and an appointment was made for her to visit a local contraceptive practice provider.

SAFE Computer-adapted Intervention

The SAFE Computer-adapted intervention provided the same information in the same order as the SAFE Face-to-face intervention. Trained staff led the participant to a computer in a private office in the MOUD clinic and set her up with the computerized information on the screen. The participant could then click through the information, use www.bedsider.org to find out more information about the contraceptive options, listen to testimonials about contraceptive options, and listen and read about how to talk to a doctor about contraceptive practice options and how, where and when to talk to her partner about condoms and other contraceptive practices. Participants observed active listening skills, positive language, “I feel” statements, and safer-sex negotiation being performed. Physical and emotional safety techniques were also provided via video testimonials. At the end of every computerized session, each lasting 45–60 minutes, a set of questions were completed by the participant asking if she wanted to make a contraceptive decision and, if so, what method(s) did she want to discuss with her health care provider. A summary of her questions and answers was printed for her by the study-trained BA and/or MA level staff already employed in the MOUD treatment facility (interventionist). The interventionist then briefly discussed her responses with her and the participant was asked if she wanted to complete a plan to make a decision regarding a method of contraception and, if so, what method(s) did she want to discuss with her health care provider. The interventionist then made an appointment for her to visit a local contraceptive practice provider.

Usual Care

The usual care (UC) condition was defined as what normally occurs in the OUD treatment clinic. It comprised having the treating medical provider or nurse on site provide each participant a brochure with written information about sexual health and contraceptive practices and an offer for referral to the local health department for more information and discussion. This meeting typically lasted 20 minutes. For study purposes, the interventionist met with the participant immediately after the meeting and asked her if she wanted to discuss with her health care provider one or more methods of contraception. If yes, then an appointment was made for her to visit a local contraceptive practice provider. If no, then after this initial session the study interventionist contacted the participant weekly for two more weeks to asked her if she wanted to discuss with her health care provider one or more methods of contraception. If the participant then decided yes, then an appointment was made for her to visit a local contraceptive practice provider.

2.5. Contraception Consultation Appointment

As soon as a participant indicated she wanted a contraception consultation appointment, she selected the appointment day and time best for her. The participant was given a picture and information about the provider, free transportation was arranged, and planning was completed around any issues regarding appointment attendance. These participants also received follow-up calls and text reminders regarding their appointments.

2.6. Assessments and Retention Efforts

Participants were followed for 6 months, with assessments at baseline (study entry), post-treatment, 3- and 6-month follow-up. All participants completed an updated extensive locator form and received bi-weekly phone calls to maintain rapport and assist with basic case management resources (if needed). Participants could choose the location of follow-up study visits. Every effort was made to have all study interactions be warm, non-judgmental and empathetic. Snacks, drinks, and gift baskets of health, beauty and home items worth approximately $25 each were given at the completion of each study visit. Women completed surveys as to what items they most liked, and these items were obtained for their baskets.

2.7. Measures

Baseline measures

Demographic and background characteristics.

Participants completed a set of questions asking about demographics, life context characteristics, and sexual and reproductive history.

Measures repeated at baseline, post-treatment, 3- and 6-month follow-up

Contraceptive Method Receipt.

Participants were asked which, if any, contraceptive method they were using. At post-treatment and follow-up appointments, they were asked if they attended their contraceptive method appointment (because they could attend the visit at any time during the study).

Intervention Satisfaction.

Acceptability was assessed by a single survey item, “How satisfied were you with the sexual health education or support you received today,” to which participants responded using a 5-point Likert-type scale (0 = “not at all helpful”; 1 = “a little helpful”; 2 = somewhat helpful”; 3 = “very helpful”; 4 = “extremely helpful”).

Primary Outcome Measures

Feasibility was measured by intervention completion and intervention satisfaction, while efficacy was measured by attendance at a contraception consultation visit and long-acting reversible contraceptive (LARC) method receipt.

There was a secondary set of outcome measures that assess knowledge of reproduction and contraceptive methods and high-risk sexual activity, and results in regard to these measures will be reported separately.

Intervention Completion.

SAFE participants who attended all 3 sessions (regardless of intervention type or if one session was missed and subsequently rescheduled and attended) were coded as completing the intervention whereas others were coded as non-completers. For participants in UC, they only needed to attend one visit for completion.

Intervention Satisfaction.

The mean of the satisfaction score across the four time points was considered to represent an overall measure of intervention satisfaction for each participant.

Attendance at Contraception Consultation Visit.

If a participant reported she made an appointment and attended it, this was deemed attendance at a consultation visit; otherwise, if the participant either never made an appointment, or made an appointment and failed to keep it, then this was deemed non-attendance.

Long-Acting Reversible Contraceptive (LARC) Method Receipt.

Participants self-reported if they received a LARC within the 6-month follow-up. If a participant chose a LARC method at the consultation visit and received one, this was coded as LARC receipt. If the participant received a different method or did not attend the contraception consultation appointment, this was coded as non-receipt.

2.8. Statistical Analysis

Baseline and demographic measures were analyzed with either a logistic regression or an analysis of variance that included a single factor with four levels: SAFE Face-to-face condition, SAFE Computer-adapted condition, UC condition, or the Ineligible group (participants found to be ineligible as detailed in 3.1 Participants). The difference between the three treatment conditions and the Ineligible group was determined by a 1-df contrast between the three treatment conditions, pooled, versus the Ineligible group. Differences among the treatment conditions were determined by a 2-df contrast among the three conditions.

A generalized linear model (GLiM), with a single fixed between-subjects factor, Intervention Condition (SAFE Face-to-face vs. SAFE Computer-adapted vs. UC) was used to conduct all inferential analyses. GLiM is a flexible generalization of ordinary least squares regression and analysis of variance that allows for outcomes in which the errors may follow a distribution other than the normal distribution (McCullagh and Nelder, 1989). A major strength of the GLiM model is that it allows estimation of least squares means and their standard errors from model parameters regardless of the distributional assumption of the outcome. In the case of a binary outcome (intervention completion, attendance at contraception consultation appointment, LARC method receipt), these least squares means represent the predicted probability of the event for a given level of the factor. Post hoc testing utilized Bonferroni-corrected pairwise mean comparisons. α for all significance tests effects was determined to be .13 at the time the study was designed and funded. This value was chosen in order to be able to detect a medium effect with 90 participants. Thus, pairwise mean comparisons were each conducted at .043 (.13/3).

In addition, ancillary analyses were conducted to further appreciate the data following examination of results. The statistical methods that were utilized are described in 3.1 Participants and 3.2 Outcomes, as findings are presented. Because these analyses occurred following examination of the data, caution should be utilized when interpreting these results.

3. Results

3.1. Participants

Of the 132 women who were screened, 20 were ineligible: 7 were older than 40 years of age; 3 were pregnant; 9 reported non-heterosexual orientation; and 1 had tubal ligation or other sterilization. Of the 112 remaining eligible women, 22 women refused participation: 4 stated they were “too busy,” 5 stated they “do not trust/bad experience with medical providers,” while 13 stated they were “not interested.” Thus, 90 women gave consent and were randomized to an intervention condition (Figure 1). There were 90 participants assessed at baseline, 90 (100%) assessed at intervention completion (with 1 participant failing to answer a question regarding anal sex), 90 (100%) assessed at 3-month follow-up, and 90 (100%) assessed at 6-month follow-up (with 4 participants failing to answer questions regarding vaginal, oral, and anal sex, and choice of birth control practices at this assessment). Thus, no participant was lost to follow-up, and no missing data occurred for the four primary outcomes.

Figure 1.

Consort Diagram

All participants were assessed at all 4 time points. However, there were rare instances where either a participant failed to respond to a specific self-report question, or declined to respond to a particular self-report measure. These instances of missing data were rare, and were almost exclusively related to questions regarding sexual activity or choice of birth control methods. No data are missing for the 4 primary outcome measures reported in the present paper.

Table 1 indicates that there were no significant differences between the three Intervention Conditions [Eligible (n=90)] and the Ineligible group (n=20; data were not available for the 22 eligible women who declined participation) nor among the three Intervention Conditions. Study participants were relatively young [M=27.4 years (SD=4.8)], with 65/90 (72%) less than 30 years of age], majority White [54/90 (60%)], predominantly single [65/90 (72%)] with 55/90 (61%) having less than a high school education [years of education M=11.2 (SD=1.0)], largely unemployed [57/90 (63%)], and currently smoking cigarettes [77/90 (86%)]. In terms of sexual and reproductive history, age of first sexual intercourse was 15.1 years (SD=1.0), with 35/90 (39%) having engaged in sexual intercourse before the age of 15 and only 8/90 (9%) at age 18 or older. Mean lifetime number of pregnancies was 2.1 (SD=1.4) with a range of 0 to 6 past pregnancies. The mean lifetime number of live deliveries was somewhat more than 50% of the mean lifetime number of pregnancies [1.2 (SD=1.0)], and the mean lifetime number of abortions [.4 (SD=.7)] was approximately 20% of the mean lifetime number of pregnancies. Most participants (73/90; 81%) reported they had used a prescribed contraception at some point in the past (Table 1). However, more than 40% reported no past-30-day contraception use at baseline with the most prevalent used method being condoms among those participants using a contraceptive method (Table 2).

Table 1:

Baseline Demographic and Background Characteristics (N=110)

	Eligible (n=90)				Ineligible (n=20)
	SAFE Face-to-Face (n=30)	SAFE Computer-adapted (n=30)	Usual Care (n=30)	p		p
	n (%)	n (%)	n (%)		n (%)
Race				.99		.89
White	18 (60%)	18 (60%)	18 (60%)		11 (55%)
Black	4 (13%)	4 (13%)	6 (20%)		6 (30%)
American Indian	2 (7%)	3 (10%)	3 (10%)		1 (5%)
Multiracial	6 (20%)	5 (17%)	3 (10%)		2 (10%)
Marital Status: Single	23 (77%)	21 (70%)	21 (70%)	.81	14 (70%)	.81
Currently Employed: yes	12 (40%)	10 (33%)	11 (37%)	.84	8 (40%)	.92
Current Smoker: yes	26 (87%)	25 (83%)	26 (87%)	.80	18 (90%)	.97
Lifetime Use of Prescription Contraceptives: yes	24 (80%)	25 (83%)	24 (80%)	.89	17 (85%)	.73
	M (SD)	M (SD)	M (SD)		M (SD)
Age	27.2 (4.7)	27.4 (5.1)	27.7 (4.8)	.93	28.0 (4.9)	.63
Education	11.2 (1.0)	11.1 (1.1)	11.2 (1.1)	.93	11.0 (.9)	.37
Age of First Sexual Intercourse	15.1 (2.1)	14.9 (2.4)	15.2 (1.6)	.88	15.3 (2.1)	.65
Lifetime Number of Pregnancies	2.1 (1.6)	2.0 (1.4)	2.1 (1.3)	.93	2.2 (1.1)	.83
Lifetime Number of Live Deliveries	1.4 (1.3)	1.2 (.8)	1.1 (.7)	.44	1.2 (.8)	.81
Lifetime Number of Miscarriages/Stillbirths	.2 (.4)	.4 (.6)	.5 (.6)	.10	.5 (.5)	.48
Lifetime Number of Abortions	.4 (.7)	.4 (.8)	.4 (.6)	.97	.5 (.8)

Notes. The p values under the Eligible column heading are the results of 2-df contrasts among the three treatment conditions, while the p values under the Ineligible column heading are the results of a 1-df contrast between the three treatment conditions, pooled, versus the Ineligible condition. Percentages are for the respective condition.

Table 2.

Baseline, Intervention Completion, and 6-month Follow-up for Utilization of Selected Contraceptive Methods for the three Intervention Conditions (N=90)

	Baseline			Intervention Completion			6-month Follow-up
	SAFE Faceto-face Condition (n=30)	SAFE Computer-adapted Condition (n=30)	SAFE Face-toface Condition (n=30)	SAFE Face-toface Condition (n=30)	SAFE Computer-adapted Condition (n=30)	Usual Care Condition (n=30)	SAFE Face-toface Condition (n=29)	SAFE Computer-adapted Condition (n=29)	Usual Care Condition (n=28)
None	13 (43%)	12 (40%)	14 (47%)	1 (3%)	1 (3%)	6 (20%)	0	0	2 (7%)
Male Condom	12 (40%)	12 (40%)	9 (30%)	5 (17%)	5 (17%)	9 (30%)	6 (21%)	4 (14%)	7 (25%)
Combined pill or progestin-only pill	4 (13%)	5 (17%)	5 (17%)	0	0	5 (17%)	0	0	3 (11%)
Intrauterine Device (IUD)	0	0	0	6 (20%)	5 (17%)	2 (7%)	7 (24%)	5 (17%)	2 (7%)

Notes. Percentages are for the respective Intervention Condition and do not sum to 100% because only a subset of the questions regarding possible birth control methods are provided in this table (see Supplementary Table S1 for more information). Contraceptive methods that were selected represent methods that were (1) prevalent in the sample and (2) appeared to show substantive change (see Supplementary Table S1 for more information). Data are missing at the 6-month follow-up assessment due to failure to respond to questions regarding birth control practices by 4 participants.

Supplementary Table S1 details the wide variability in the choice of contraceptive practices at enrollment, treatment completion, and 6-month follow-up assessment. In an attempt to economize the presentation of the information in Supplementary Table S1, contraceptive methods that were (1) prevalent in the sample and (2) appeared to show substantive change in uptake over time are shown in Table 2. Although the general interpretation of this table is that over the intervention period, participants began use of some contraceptive method, given the pronounced decline in the percentages for the report of non-use of any birth control method (“None”), and that the use of male condom shows some decline over the period, the large number of zero cells precludes the analysis of even this selected table. However, it was possible to conduct two separate ancillary analyses: (1) on non-use of any birth control method and (2) on use of a male condom, using a weighted least squares approach in which the Intervention Condition is a fixed between-subjects factor and Assessment Time Point is a repeated or within-subjects factor with 3 levels (baseline, post-treatment, and 6-month follow-up). For non-use of any birth control method, the only significant effect found was for the main effect for Assessment Time Point, χ²(2) = 52.5, p<.001, in support of the contention that non-use of any birth control method declined, seemingly sharply, over the course of the trial. Similarly, for male condom use, the only significant effect found was for the main effect for Assessment Time Point, χ²(2) = 7.5, p<.03.

3.2. Outcomes

Table 3 provides the F test statistics and their associated p values, and the least squares means and their standard errors for the Intervention Condition effect. The Intervention Condition effect was significant for all four primary outcomes (ps<.001): intervention completion, intervention satisfaction, contraception consultation visit attendance, and LARC receipt. The least squares means for all four outcomes (in the case of the binary outcomes of intervention completion, attended contraception consultation appointment, and LARC method receipt, these means are the model-derived predicted probabilities, while in the case of intervention satisfaction, means are the ordinary least squares model-derived means) were non-significantly different from each other in comparisons of the two SAFE conditions. However, across the four outcomes, the least squares means for the two SAFE intervention conditions were both significantly higher than the least squares means for the UC Condition (all ps<.0039 for the eight tests comparing SAFE Face-to-face to UC and SAFE Computer-adapted to UC, respectively).

Table 3.

Feasibility, Acceptability, and Efficacy Outcomes for the three Intervention Conditions (N=90)

	SAFE Face-to-face Condition (n=30)	SAFE Computer-adapted Condition (n=30)	Usual Care Condition (n=30)
Outcome:	Least Squares Means (Standard Errors)			F	p
Intervention Completion: yes	.97 (.03)	.97 (.03)	.53 (.09)	7.8	<.001
Intervention Satisfaction (range 0–4)	3.7 (.11)	3.8 (.11)	3.1 (.11)	12.0	<.001
Attended Contraception Consultation Appointment: yes	.80 (.07)	.73 (.08)	.33 (.09)	7.1	<.001
LARC Method Receipt: yes	.77 (.08)	.73 (.08)	.23 (.08)	9.2	<.001

Notes. α=.13 for all tests. Degrees of freedom (df) for inferential F tests were 2, 87. For intervention completion, attendance at contraception consultation appointment, and LARC method receipt, least squares means represent the predicted probability of the event for a given level of the factor, while Intervention Satisfaction is a continuous score [0 = “not at all helpful”; 1 = “a little helpful”; 2 = somewhat helpful”; 3 = “very helpful”; 4 = “extremely helpful”]. Post hoc testing indicated that for all four outcomes, the two SAFE Conditions were not significantly different from each other (all ps>.043) but were significantly different from the Usual Care Condition (all ps<.0039).

To present a concise summary for the binary outcomes, turning the proportions into counts revealed: 29/30 participants in each of the SAFE Face-to-face and SAFE Computer-adapted Conditions completed the intervention in contrast to 7/30 participants in the UC Condition; 24/30 participants in the SAFE Face-to-Face Condition and 22/30 participants in the SAFE Computer-adapted Condition attended a contraception consultation in contrast to 10/30 participants in the UC Condition; and, 23/30 participants in the SAFE Face-to-Face Condition and 22/30 participants in the SAFE Computer-adapted Condition received LARC in contrast to 7/30 participants in the UC Condition. In regard to intervention satisfaction, the means of the two SAFE conditions approached the Likert scale anchor point of “extremely helpful”, while the mean for the UC condition, although in the positive direction, was only slightly above the Likert scale anchor point of “very helpful”.

In an attempt to further understand findings regarding contraception consultation appointment, we examined at what point in the intervention follow-up the participants in each of the three intervention conditions chose to make such an appointment (see Table 4). A post hoc ordered logistic regression analysis with Intervention Condition as a fixed between-subjects factor and session in which a contraception consultation appointment was received (session 1, 2, 3; and 4 for no receipt of LARC) as the outcome indicated that the three Intervention Conditions differed in their respective patterns of attendance, Wald χ²(2) = 9.7, p<.008; however, subsequent pairwise comparisons indicated that each of the two SAFE Conditions differed from the UC Condition (Face-to-Face vs. UC: Wald χ²(1) = 8.9, p<.003; SAFE Computer-adapted vs. UC: Wald χ²(1) = 5.9, p<.02), while the comparison between the two SAFE Conditions failed to reach significance (Wald χ²(1) = 0.4, p>.5). A subsequent one-way test of the six cell proportions of timing of the contraception consultation appointment (that is, at sessions 1, 2, and 3, respectively) for the two SAFE conditions restricted to those participants who did attend a contraception consultation appointment (n=52) failed to reject the null hypothesis that the probability of attendance was equal (that is, in each SAFE Condition, the null hypothesis was that there was a 33.3% chance of making a contraception decision at any one of the three sessions).

Table 4.

Session during which Contraception Decision was made for the three Intervention Conditions (N=90)

	SAFE Face-to-face Condition (n=30)	SAFE Computer-adapted Condition (n=30)	Usual Care Condition (n=30)
	n (%)
At Session 1	6 (20.0)	5 (16.7)	5 (16.7)
At Session 2	11 (36.7)	9 (20.0)	4 (13.3)
At Session 3	7 (23.3)	8 (26.7)	1 (3.3)
No decision made	6 (20.0)	8 (26.7)	20 (66.7)

Notes. Percentages are for the respective Intervention Condition and may not sum to 100% due to rounding.

Finally, given the finding that use of a male condom declined over the course of the trial in the sample as a whole, an important question would be whether those participants who received LARC also used condoms. Of the 52 participants who received LARC at study entry, 19/52 (36.5%) reported use of a male condom and 0/52 reported use of a female condom in the 30 days prior to study entry. The corresponding figures for male and female condom use, respectively, at 6-month follow-up were 0/50 (0%) and 1/50 (2%), respectively (two participants who received LARC were 2 of the 4 participants who did not report their birth control practices at 6-month follow-up).

4. Discussion

Women with OUD receiving MOUD are at high risk of unintended pregnancy, yet contraceptive use is often low, and less effective method use is common (Terplan et al., 2015). Women with OUD have unique or heightened concerns about effective contraceptive options (i.e., LARC), such as compromised reproductive autonomy and pain (Rey et al., 2020). Additionally, it has been suggested that women receiving MOUD may benefit from education about LARCs, including their efficacy and information about their insertion and removal (Matusiewicz et al., 2017). Thus, contraceptive interventions tailored for this specific population are needed. This small-scale randomized controlled trial provides an initial evaluation of a novel cognitive-based approach (SAFE) for women enrolled in MOUD treatment. Both SAFE Face-to-face and Computer-adapted interventions were feasible, with increased adoption of LARCs at the post-treatment interval compared to the UC condition.

The number of sessions it took for participants in the two SAFE conditions who attended a contraception consultation appointment to make a contraception method decision were fairly evenly distributed over the three sessions. Such results suggest neither a priming nor latency effect. Our finding of how more than 20% of participants in both SAFE Face-to-face and Computer-adapted interventions did not make a contraception method decision after 3 sessions suggests that future studies should examine ways to resolve continued ambivalence about making any such decision (i.e., use or intended avoidance of contraceptive methods).

Paradoxically, the sample as a whole decreased its use of male condoms during the study follow-up. Subsequent examination of data for participants who opted for LARC showed that by the 6-month follow-up assessment, only 1 of 52 participants who chose LARC reported they were also using female condoms. Such low condom use before and after contraceptive method interventions is consistent with findings from previous studies and may be partly due to high rates of monogamous relationships (e.g., Heil et al., 2016). Nonetheless, future studies need to address more effective ways to increase condom use by women with OUD (and their partners) who also use LARC in order to best promote dual method use for prevention of both sexually transmitted infections and unintended pregnancies.

An encouraging finding from the present study was the almost universal participant completion of both the SAFE Face-to-face and Computer-adapted interventions. This reflects the high feasibility of this intervention within a clinical setting. There is a call for integration of innovative approaches like ours aimed to increase contraceptive uptake into clinical settings, including OUD treatment (Heil et al., 2019). Integration overcomes many practical barriers to service receipt, vital aspects to consider as they do impede contraceptive uptake among vulnerable populations (Higgins et al., 2018). However, both patients and substance use providers perceive multiple barriers to this integration, such as competing priorities and lack of knowledge or training as well as cost (Klaman et al., 2019; Klaman et al., 2020; MacAfee et al., 2019b). Allowing treatment centers to adopt either the SAFE Face-to-face or Computer-adapted option of the intervention could potentially overcome some of these barriers by offering flexibility in intervention delivery without compromising gains in patient benefits.

Further, women reported high satisfaction with both SAFE interventions. Participant satisfaction in the two SAFE conditions was very high, with means of 3.7 and 3.8 (out of 4) for SAFE Computer-adapted and SAFE Face-to-face conditions, respectively, indicating average evaluations in the two conditions bordering on “extremely helpful”. SAFE is grounded in social-cognitive theory, but its content and structure were devised from robust qualitative data gathered for the purposes of intervention development. Our finding of high intervention satisfaction, considerably higher than that of usual care, reflects how the foundation of our intervention was created from key stakeholder input, namely women and men in treatment for substance use disorder. Stigma and discrimination are strong barriers for women to avoid seeking not only OUD treatment (Frazer et al., 2019), but also reproductive and sexual health services (MacAfee et al., 2020; MacAfee et al., 2019b). This stigma and discrimination extends into the exam room where women commonly don’t trust their healthcare providers to provide them with reliable contraceptive education (Heil et al., 2016). One potential benefit of integrating reproductive and sexual health services into OUD treatment settings is that women may place more trust in their relationships with OUD treatment than outside providers, given their non-judgmental nature and perceived safety at the treatment center (MacAfee et al., 2019b). Interventions targeted at mitigating stigma and discrimination represent an avenue to facilitate women receiving appropriate contraceptive education and a reliable method of birth control. Thus, we designed SAFE not only to be intended for delivery within an OUD treatment setting but also tailored to the needs, desires, and concerns elicited by women and men currently in treatment for substance use disorder. As a result, SAFE’s high participant satisfaction indicates its potential to effectively equip OUD treatment providers to meet their patients’ unmet reproductive and sexual health needs (MacAfee et al., 2020) as well as promote their reproductive autonomy in a non-stigmatizing manner.

Overall, both SAFE interventions increased the probability of women attending a contraceptive counseling provider appointment and receiving a LARC relative to women receiving usual care. Even though these were the primary outcomes of our study, it is notable how SAFE emphasized reproductive autonomy in its provision of contraceptive education, not the ‘need’ for women with OUD to use contraception. SAFE provided women with tools to promote self-efficacy and choice, which could include not using a reliable contraceptive method. Women with substance use disorder often have significant trauma histories with high burdens of childhood abuse and intimate partner violence (Engstrom et al., 2012; Schneider et al., 2009). Thus, trauma-informed approaches are critical in the care of this population (SAMHSA, 2014), including non-coercive approaches to reproductive and sexual health service provision. With the high prevalence of unintended pregnancies among women with OUD (Heil et al., 2011) and prenatally opioid-exposed neonates (Leech et al., 2020), emerging contraceptive clinical practices may be inclined to respond with top-down, tiered approaches emphasizing the urgent need for women with OUD to use LARCs. While women in OUD treatment do have significant unmet family planning needs (Terplan et al., 2015), the importance of prioritizing compassionate, person-centered, trauma-informed approaches for this population cannot be understated. Our findings highlight the effectiveness of SAFE at meeting reproductive and sexual health needs of women with OUD while simultaneously promoting reproductive autonomy.

Our study is not without its limitations. First, the relatively small sample size and requirement of 90 or more days in MOUD treatment in each treatment condition is important to note. Moreover, the intervention was conducted at two treatment centers in a single state which limits generalizability of our findings to women not engaged or newly enrolled in MOUD treatment as well as in different geographic locations. Thus, our findings should be replicated in a larger sample elsewhere, such as treatment centers with a higher proportion of women of color and/or ethnic diversity as well as in sexual minority samples.

Future intervention trials should also incorporate additional person-centered outcomes beyond participant satisfaction; comprehensively assessing the patient perspective combined with other clinically meaningful outcomes, such as those measured in our study, would allow for further translation of research findings into clinical care. Lastly, all our measures were based on self-reported outcomes which could have introduced a possible response bias into our results and future studies should include medical record and billing data verification. Limitations notwithstanding, these initial trial results yield encouraging findings regarding the efficacy of the SAFE intervention for helping women treated with MOUD make and enact informed decisions about their sexual health goals.

HIGHLIGHTS.

• Many women in opioid medication treatment clinics want sexual health education

• The SAFE intervention was feasible, efficacious and acceptable to women

• SAFE in opioid use disorder treatment for women helps meet sexual health goals