Outcomes of Implementing in the Real World the Women's Health CoOp Intervention in Cape Town, South Africa

This study used a modified cluster-randomized stepped-wedge design. In a cluster-randomized stepped-wedge design, each cluster receives the intervention; however the order in which clusters receive the intervention is randomized. This approach overcomes the ethical issues that would arise from a traditional cluster-randomized trial in which participants in clusters assigned to the control arm would not receive an intervention that had already been proven to be efficacious. It also overcomes logistical issues associated with trying to implement an intervention in a large number of clusters simultaneously. These features make the stepped-wedge design particularly well suited for studies assessing the implementation of interventions with established efficacy. A more simplified study design is shown in Fig. 1.

In this study, clusters comprised eight clinics—4 HIV/antenatal clinics and 4 substance use treatment rehab clinics—in Cape Town, selected and run by the City of Cape Town Health Department. The clinics were each embedded within clinic compounds that served populations of African descent in historically underserved communities where there is a higher prevalence of HIV. All clinics received the intervention. Randomization was conducted using a SAS computer program to assign clinic pairs matched by geography—one HIV clinic and one substance use treatment clinic—to 1 of 4 cycles. The implementation strategies and the WHC intervention were conducted one cycle at a time—with Cycle 1 occurring for the first 6 months, and each subsequent cycle following until Cycle 4 was complete. Prior to each cycle, the study’s Principal Investigator trained clinic staff at both clinics to administer the WHC intervention.

This modified stepped-wedge design recruited 480 women living with HIV. Project staff conducted outreach activities in the communities around the study clinics and in-reach activities within the clinics. Project staff asked interested individuals a set of eligibility questions using a screener. To be eligible, potential participants had to meet the following criteria: be female; be 18 to 45 years of age; report the use of at least one substance—which could include alcohol—at least weekly during the previous 3 months; report unprotected sex (sex without a condom) with a male partner in the past 6 months; had a positive HIV test result from either the participating health clinic or treatment clinic, or a clinic-issued ART appointment card or ART prescription as verification of HIV status; and intend to remain in the area for at least 6 months. Individuals were excluded if they were not HIV-positive or if they were unwilling to participate in alcohol and other drug screening tests. Individuals who met eligibility criteria and were interested in participating were scheduled for an intake appointment (if not the same day) at one of the clinics, where they were rescreened for eligibility and then enrolled if they provided informed consent to participate.

If the participant could not attend the baseline appointment on the same day of her screening, she was given a date of her choice for the appointment where she was rescreened for eligibility using the same screening form used previously to verify responses. The intake appointment was conducted in the study clinic. Research staff obtained informed written and signed consent and a signed release for referral for health services and administered a face-to-face baseline interview using a secure tablet with skip patterns and quality-control checks programmed into the interview. The baseline interview included measures of socioeconomic status; sex risk behaviors; alcohol and other drug use; substance use treatment readiness; ART use and barriers to use; and other self-reported clinical outcomes, such as tuberculosis (TB) and symptoms of sexually transmitted infections (STIs). Biological data were obtained from a urine specimen tested for pregnancy and metabolites for recent drug use. Participants also underwent a breathalyzer test for recent alcohol use. Active linkage to ART initiation, antenatal care, and substance use treatment was offered at the end of the appointment. Participants were given R100 for their time and transportation reimbursement. During this appointment, they also were scheduled for their first WHC intervention workshop.

A master trainer trained clinic staff after the initial training with the PI at each study clinic on the intervention two weeks prior to implementation launch to practice and role-play competence [23]. These trained clinic staff led the WHC intervention—a 2-session group workshop—at the participating study clinics. Clinic staff included HIV counseling and testing staff, nurses, clerks, and paraprofessionals. Workshop groups averaged in size with 5 attendees—and participants could attend the workshop in either order for flexibility of the workshop and for real-world implementation. Workshop 1 content consisted of facts about HIV, STIs, and TB; male and female condoms; sex risk reductions; skills to negotiate safer sex; and how women living with HIV can stay healthy. Workshop 2 content consisted of facts about alcohol and other drugs, injection drugs, reducing alcohol and other drug risk, violence, and prevention strategies. An individual personalized action plan was developed, when possible, at the end of the second workshop. Each workshop lasted an average of 46 min (depending on the discussion) and the two workshops usually took a total of less than two hours together and both workshops were held approximately within a month, although this time varied depending on when enough participants were available to convene a workshop group. An overview of the intervention content and topics for each workshop is presented in Table 1.

Core elements of the original intervention on which the WHC is based include (1) role-playing condom use negotiation; (2) developing an individualized action plan to establish concrete risk-reduction steps, such as substance use and sexual risk behavior; (3) accessing HIV testing or making referrals to local service agencies; and (4) distributing risk-reduction materials, such as male and female condoms and lubricants. A formative phase that included focus group discussions with stakeholders and meetings with a Community Collaborative Board (CCB) was conducted to adapt and refine the intervention specifically for women living with HIV [37].

Additional slides were developed around ART, understanding viral load and the importance of “good” adherence. This resulted in a new published intervention curriculum.

Participants were followed up 6 months post enrollment to take part in a follow-up appointment involving activities similar to the intake appointment for repeated measures. At this appointment, participants re-consented, participated in a risk behavior survey via computer-assisted personal interviewing, and provided a urine sample for drug screening and a breath scan via a breathalyzer for recent alcohol use. Follow-up appointments were conducted either at the study clinic or at the project field site. Participants were given R150, a certificate of study completion, and transportation at this final appointment.

The following patient-level measures were assessed at baseline and 6-month follow-up.

We began by calculating descriptive statistics at baseline and follow-up for each cycle. We identified nine variables that could plausibly be related to the outcomes and that differed significantly across cycles. We included these nine variables in a multiple logistic regression analysis and saved the predicted probabilities. These predicted probabilities were entered into the outcome models as propensity scores [39]. Cycle 1 was used as the reference group to compare subsequent cycles to the first cycle, as the study design allowed for modifications and refinements after each cycle through the mixed methods design [36]. We used a generalized estimating equations (GEE) approach with an exchangeable correlation matrix to logistic regression to adjust for correlations among participants within each cycle. Analyses were conducted in SAS Enterprise Guide Version 7.15 and Stata MP/IC Version 16.

The study’s CONSORT Diagram is presented in Fig. 2. A total of 2165 people were screened for eligibility, and 594 (27%) were eligible. Reasons for ineligibility are noted in the CONSORT Diagram. A total of 480 participants were enrolled and consented. An average of 83% of participants attended Workshop 1, ranging from 79 to 88% across the 4 cycles. An average of 84% of participants attended Workshop 2, ranging from 82 to 85% across the 4 cycles. Approximately 91% of participants received at least one workshop (not shown). Approximately 95% of participants returned for their 6-month follow-up—ranging from 93 to 99% across the 4 cycles. There were 4 non-study-related deaths over the course of the study, which were attributed to comorbidity. There was no differential attrition by cycle.

Table 2 presents the demographics of the study participants enrolled overall and across the 4 cycles (N = 480). The mean age of the study participants was 33 years (SD = 6). Most participants (93%) were Black African (not shown). A third of participants (36%) had some form of food insecurity and 46% of participants did not have running water in their home. Participants drank an average of 11 days (SD = 7) during the past month, had 14 drinks (SD = 6) per drinking day during the past month, and had an average of 11 days (SD = 7) in the past month when they drank 4 or more drinks. Among participants, 92% met the thresholds for NIAAA’s definition of risk for an alcohol use disorder. These alcohol measures differed significantly by cycle. An estimated 70% of participants reported taking ART in the past month, 72% reported taking ART in the past 6 months, and 85% reported ever taking ART—the latter was statistically significant by cycle.

In addition, Table 2 presents follow-up information to correspond to the baseline values. Overall, report of homelessness increased (from 23 to 30%) whereas reports of participants currently having a main partner decreased (from 86 to 74%). Overall alcohol use also reduced in average days drinking, days drinking 4 or more drinks (Mean = 7, SD = 7 for both), and average drinks per day (Mean = 10, SD = 6). Increases in ART use also were observed overall, with 88% reporting ART use in the past 6 months, 83% reporting past month ART use, and ART adherence at an average of 89% (from 75%) of ART taken in the past month.

GEE models examined associations between cycle and intervention outcomes of interest—alcohol use disorder risk and 85% ART adherence at follow-up. Cycle 1 was used as the reference group in the models, which also included the baseline value and a propensity score variable to control for relevant baseline differences in participant characteristics across the cycles. Compared with Cycle 1, women in Cycle 4 were significantly less likely (OR = 0.10 [95% CI 0.04, 0.24], p < 0.001) to report alcohol use disorder risk at 6-month follow-up. The odds were not statistically significant for alcohol use disorder risk for Cycle 2 or Cycle 3 as compared with Cycle 1 (Table 3).

At follow-up, women in Cycle 4 were significantly more likely (OR = 4.16; 95% CI 1.05, 16.51) to report being at least 85% adherent to ART in the past 6 months compared with women in Cycle 1 (Table 4). As predicted, the likelihood of taking ART increases as women are enrolled in the later cycles and the risk of alcohol use disorder decreases.

Intervention satisfaction was high among the 435 participants who participated in at least one of the intervention’s workshops, with a mean score of 30.9 (SD = 1.9) out of the possible range of 8 to 32.