A Combined Motivational Interviewing and Behavioral Couples Therapy Intervention to Reduce Intimate Partner Violence and Alcohol Use in South India
A Combined Motivational Interviewing and Behavioral Couples Therapy Intervention to Reduce Intimate Partner Violence and Alcohol Use in South India
The goal of this randomized clinical trial is to test an intervention consisting of a combination of behavioral couples therapy and motivational interviewing to improve communication and reduce conflicts between couples and decrease harmful drinking among spouses in urban primary health centers, South India. The intervention will be delivered by nurses in primary health centers who will be supervised by a clinical psychologist.
The main question[s] it aims to answer are:
These intervention participants will be compared to a control group who will receive only referral information for intimate partner violence and an educational session and referral for alcohol use disorder.
All participants will participate in quantitative interviews at baseline, and every three months thereafter, for a period of one year. In-depth qualitative interviews will be done with a subgroup of couples to try to understand how the intervention led to the observed outcomes.
Aim 1: PILOTING. Prior to launching the trial, 10 male and 10 female participants will pilot test any measures that need to be adapted for this study and complete cognitive interviews. The study team and stakeholders then review the pilot data and finalize the measures. The investigators finalize and refine the intervention protocol based on direct observation of nurses delivering it to research staff. Mock sessions will be observed to ensure protocol fidelity when MI and BCT are delivered sequentially. Training manuals, guides, fidelity checklists and trainers are already available from previous studies. For MI, Motivational Interviewing Treatment Integrity (MITI) coding is used; for BCT, a standard checklist from a pilot study, to document proportion of relevant topics covered in each session and the quality of counseling, interaction with participants to assess engagement with the material, and any factors that may have affected implementation.
Aim 2: RANDOMIZED CONTROLLED TRIAL: 400 married couples will be recruited from government-run urban primary health centers (PHCs). Community health workers, called ASHAs (Accredited Social Health Activists) in India, refer couples they know are experiencing intimate partner violence (IPV) while the husband uses alcohol. ASHAs bring the woman to the clinic for a routine "women's health" appointment in which research staff screen the woman for eligibility, after obtaining informed consent. If the woman is eligible (criteria see elsewhere), she is asked if her husband drinks alcohol. If yes, the study staff member determines the appropriate method for engaging her husband (wife brings him, study calls him directly, or ASHA speaks with him).
In the next visit for the couple, the research team obtains screening consent of the husband and assess his eligibility (criteria see elsewhere). For all patients who are ineligible for the study but screen positive for IPV or AUD (alcohol use disorder), study staff will provide a brief educational session on AUD (to applicable participants) and immediately refer all of them to NIMHANS in Bangalore, a premier tertiary care system that can address IPV issues, psychology and psychiatry services, as well as de-addiction for severe alcohol dependence and alcohol withdrawal. After confirming that the couple is eligible and interested, the research staff obtains consent to participate in the trial, enroll them in the study and, if randomized to the intervention arm, schedule the first session with the study. Participants are randomized at the couple level, prior to the start of the intervention.
Participants in the intervention arm receive 4 weekly 1-hour Motivational Interviewing (MI) sessions (#1-3 for husband only, #4 for couple), followed by six weekly 1-hour Behavioral Couples Therapy (BCT) sessions; all are delivered by a trained study nurse and occur in person - details elsewhere.
Participants in the control arm receive enhanced usual care, since usual care for IPV and AUD in PHCs is limited and even when protocols exist, they are not usually followed. Enhanced usual care includes: a) trained staff conduct initial safety assessments for all patients; b) for IPV, wives are referred to a tertiary care center (NIMHANS), and informed of their options and local resources; c) for AUD, participants receive a brief educational session and referral to NIMHANS.
Assessment procedures include (1) tablet-administered questionnaires by experienced, trained interviewers fluent in the local language, and (2) breathalyzer tests over a 1-week period to measure alcohol use. The tablet-administered questionnaires collect information on demographics, self-efficacy, motivation to change, coping skills, alcohol consumption and dependence, alcohol expectancies, context of drinking, context of violent situations, attitudes towards violence, IPV, marital relationship, gender norms, communication skills, anger management, impulsivity, stressors, decision making power, stigma, anxiety, and depression. Participants are reassessed quarterly for 12 months. A week prior to the interview, ASHAs bring the breathalyzer to the patient's home. Participants get a daily reminder to use the breathalyzer every day for the next week. On day 8, ASHAs return to the participant's home to obtain the breathalyzer, and escort them to the PHC to complete follow-up questionnaires.
Primary intention-to-treat (ITT) data analyses will compare the intervention arms on the primary outcomes: 1) mean AUDIT (Alcohol Use Disorders Identification Test) scores for alcohol use; and 2) mean overall IFVCS score (Indian Family Violence and Control Scale) for IPV. It is hypothesized that at 12 months, the treatment group will show lower mean AUDIT and IFVCS scores than the control group. These 2 outcomes will be examined repeatedly, to compare the trajectories of both treatment arms over time, using linear regression. For both, models will control for baseline outcome score and any relevant covariates for which the treatment groups might differ at baseline despite randomization. A multilevel version of the models will be used for the longitudinal analyses, with repeated measures nested within participants and a random intercept for participants. The treatment arm x wave interaction serves as the test of the intervention effect.
Secondary outcomes are CPQ (Communication Patterns Questionnaire) score and number of days with a negative breathalyzer test over a 1-week period. It is hypothesized that post intervention (a) the treatment group will show more positive and less negative communication patterns than the control group; (b) the treatment group will show a higher mean number of negative breathalyzer tests than the control group. For the continuous CPQ scores, analyses will consist of similar ITT linear regressions as described for the primary outcomes. The breathalyzer variable is a count variable, so Poisson regression will be used, with an offset of number of days of measurement to account for missing daily test results. CPQ data will be collected from both spouses. CPQ subscale scores will be constructed based on each spouse's report separately, and agreement assessed via intraclass correlations. The treatment effect will be explored using husband and wife reports as the outcome variable in separate models to allow a differential effect of the treatment on mean CPQ subscale scores as reported by each partner in the couple. Other secondary analyses will rerun the ITT models above, but with intervention dose (number of sessions attended) added to the models.
Power calculation: Enrolling 400 couples, with an expected 20% attrition, results in n=320. For the IFVCS outcome, assuming a standard deviation of 14 based on pilot results, n=320 results in 80% power for a minimum detectable effect size of a difference in mean IFVCS score between treatment and control group of 4.5, corresponding to a standardized effect size of d=0.31, a small effect size. Furthermore, an effective sample size of 320 provides sufficient power for the Poisson regression of number of negative breathalyzer tests in a week. Based on pilot data we expect that the control group will have a mean of 4-6 negative daily breathalyzer tests (base rate). With alpha=0.05, n=320 results in 80% power to detect a minimum effect size of around 1.2 times as many negative breathalyzer tests in the treatment than control group, i.e. an increase to a mean of at least 4.68, 5.75 and 6.82 negative tests for a base rate of 4, 5, or 6 negative daily tests, respectively.
Aim 3: MIXED METHOD ASSESSMENT: 40 intervention arm couples will take part in serial in-depth qualitative interviews (IDI). Using a purposive sampling strategy, couples are selected for a variation in engagement with intervention activities, as well as differences in factors related to engagement in particular clinics to account for a diversity of experiences in the intervention. Because all Aim 3 participants will be individuals recruited from the Aim 2 cohort, study investigators will already have contact information and will invite them to take part in the interview once they are eligible. To ensure the experiences in the intervention can be recalled accurately, enrollment of intervention participants for interviews will be ongoing, with each selected participant taking part in an in-depth interview at the end of intervention activities (3 months) and at study completion (12 months). After obtaining consent, study investigators will set up a 30- to 60-minute-long individual interview. Male and female partners are interviewed separately. All qualitative interviews follow a semi-structured guide and are conducted by an experienced interviewer. In the first interview participants are asked about relationship quality and experiences with violence, experiences with alcohol use and impact on relationship, engagement in intervention activities, barriers and facilitators to participation, perceived quality of each activity, and perceptions of how participation has affected hypothesized mechanisms of change and study outcomes. In interview 2, participants are asked about how behaviors and experiences related to alcohol use or IPV have changed or been sustained since the end of the intervention, perceptions of the role of the intervention in these changes, and any unexpected positive or negative consequences of participation. Interviewers ask follow-up questions to encourage a narrative response to understand the intervention's role in mechanisms to change. Interviews are digitally audio recorded, translated, and transcribed verbatim with all identifying information deleted and uploaded on a shared, encrypted, password-protected server.
Quantitative data on mechanisms of change (all study participants in aim 2) and perceived intervention quality and contribution of each component to change mechanisms is collected as part of the surveys administered at each study visit described in aim 2.
For qualitative data, analyses of IDI transcripts use a thematic approach to coding and summarization using Dedoose qualitative analysis software. A preliminary codebook is developed from interview guides and rapid analysis content, then refined after all analysts code a set of transcripts independently and resolve discrepancies through consensus. The remaining transcripts are single-coded, maintaining intercoder reliability of at least 80% throughout the coding process. Coding memos are used to summarize and explore the relationship between the constructs of the theory of change framework. Analysts also review coded serial IDI content within each participant's dataset to look specifically for patterns over time to further understand sustained results.
Mediation analyses of quantitative data assess if the effect of the intervention on the primary outcomes is mediated by change in the theorized drivers and intermediate variables, by assessing the effect of the intervention on the mediators and of both the mediators and the intervention on the outcome. This will be tested via structural equation models (SEM) with bootstrapped confidence intervals for the indirect effects. SEM can also accommodate dyadic analyses in which variables reported by both spouses are included in the same model and estimate both 'actor' and 'partner' effects. In addition, it will be explored if personality factors such as impulsivity moderate the intervention effect. Moderation is assessed by including an interaction between the potential moderator and the intervention variable in the regression model. Findings from the qualitative and quantitative analyses are triangulated to build a more comprehensive understanding of how the hypothesized changes in mechanisms of action resulted in the intended outcomes and their relative weight in creating change.
Inclusion Criteria:
Exclusion Criteria: