Abstract
This study evaluated preliminary reliability and validity evidence for novel scores that extend the Motivational Interviewing Treatment Integrity (MITI) coding system to assess elements unique to Motivational Interviewing (MI) with couples. We recruited 20 cismale couples where at least one partner was aged 18-29; reported substance use; and was HIV-negative. Couples were randomized to a three-session MI or attention-matched education control. Four coders rated audio-recordings of initial sessions in both conditions. Novel global scores had high nearly-exact percent agreement (Decreasing Dyadic Ambivalence=100%; Facilitating Dyadic Functioning=94%). Consistency for novel behavior counts was good (ICCdyadic reflections=.72) and above minimum thresholds (nearly-exact agreement for Relationship Affirmations=78%). MI sessions generated novel and established MITI codes that were significantly better than education sessions, providing preliminary evidence of validity. Findings provided initial support for the use of these novel scores in future research evaluating MI-based substance use and HIV risk reduction studies with couples.
Keywords: HIV prevention, HIV testing, gay couples, men who have sex with men, drug use
INTRODUCTION
Recent years have seen a resurgence of interest in couples-focused HIV prevention interventions, particularly for sexual minority men (SMM). SMM continue to face HIV disparities, having accounted for 69% of new HIV diagnoses in 2018 (1). Intervention efforts tailored for partnered SMM have been energized in large part by the observation that a substantial number (as many as 68%) of new HIV infections in this group are transmitted between main or primary relationship partners, as opposed to casual partners(2).
Several dyadic interventions designed for SMM have demonstrated the potential to achieve reductions in sexual risk behavior or drug use (e.g., 3, 4). While some of these integrate psychoeducational or cognitive behavioral strategies and techniques, many are informed by or draw upon principles of Motivational Interviewing (MI). As such, the continued growth of research in this area requires new attention to fidelity monitoring and specifically the need to evaluate the quality of MI delivery with couples.
The Motivational Interviewing Treatment Integrity (MITI) coding system (5) has long been the gold standard for assessing MI fidelity. MITI requires coders to assess a randomly selected twenty-minute segment of the session and assign global scores and behavior counts to clinician statements. Global scores are used to summarize the coders’ overall impression of the clinician’s ability to establish these MI component processes throughout session and overall demonstration of the MI spirit. Coders assign a score from a five-point scale to indicate whether the clinician was able to successfully elicit change talk from a client, avoid discussions against change, establish a partnership with the client, and display empathy and understanding towards the client.
Behavior counts require the coder to record frequencies of specific clinician behaviors. MI-adherent behaviors include: Affirmations (i.e., clinician statements that highlight positive aspects of the client, their behavior or effort), Seeking Collaboration (i.e., clinician statements that acknowledge the importance of the client’s view or clinician attempts to cooperate with the client) and Emphasizing autonomy (i.e., clinician statements that emphasize the client’s control over their behaviors and decisions regarding change). MI non-adherent behaviors include: Persuade (i.e., clinician statements that aim to explicitly change the client’s attitudes or behaviors using convincing arguments or facts), and Confront (i.e., clinician statements that openly criticize, shame, or disapprove of the client or their behavior). Additionally, coders record frequencies of core active-listening skills such as questions as well as simple and complex reflections.
Previous research on MI with couples has utilized the MI for Significant Others (MISO) coding manual. The MISO is designed to evaluate the language of an identified-patient’s significant other, rather than the language of the clinician (6). The purpose of the MISO is to evaluate interactions between patients and their partners and assess the influence the significant other has on patient change talk. Therefore, there has yet to be a coding manual developed to evaluate the quality of MI delivery by the clinician in couples’ sessions.
Recent work in the theory and practice of MI with couples has seen the emergence of an alternative to the identified patient paradigm. Starks and colleagues (7) proposed a theoretical framework for conducting MI with couples in which the couple in toto is conceptualized as the client. They outlined three theoretical principles unique to conducting MI with a couples. First, there is the potential for dyadic ambivalence – or a divergence in partners’ perspectives – towards change. Second, relationship functioning may play a vital role facilitating accommodation – or adaptive responses to dyadic ambivalence that mitigate discord or conflict between partners. Finally, joint-goal formation is a prerequisite for successful dyadic change planning. Unless partners have arrived at some shared vision of what change should look like, discussions of how to enact change are likely to elicit sustain talk from at least one of the partners and exacerbate dyadic ambivalence.
Building upon this work, Starks and colleagues (8) outlined five techniques commonly used by clinicians to respond to dyadic discord or conflict in response to dyadic ambivalence. Two of these techniques were diffuse, meaning they involved a sequence of utterances and are best captured by global scores related to a provider’s skill in Decreasing Dyadic Ambivalence and Facilitating Relationship Functioning. The remaining two techniques were discrete utterances similar to other active listening skills captured by behavior counts. Clinicians can foster empathy between partners and ease tension by reflecting on areas of agreement – utilizing content expressed by both partners, or highlighting “common ground” between partners (Dyadic Reflections). Finally, clinicians can reflect the strengths in relationship functioning and identify previous moments when the couple successfully worked together (Dyadic Affirmations).
While the existing literature has identified novel processes and techniques unique to couples’ MI, the field would benefit from the development of an objective coding system that encompasses these components. While the MISO provides one option, it is not well-suited to assess MI fidelity in paradigms where there is not an “identified client”, nor does it assess the quality of the clinician’s MI delivery to a couple. In this paper, our goal was to derive such codes. The purpose of the current study was to evaluate the reliability and preliminary validity of new MITI codes assessing therapist behaviors unique to MI with couples. Furthermore, we evaluated the validity of existing MITI codes when adapted for application to couples’ sessions.
METHODS
Participants
This study utilized demographic data from baseline assessments and session 1 recordings from participants enrolled in the pilot phase of the Couples Health Project. Couples in this initial pilot phase were enrolled between January, 2014 and October, 2015. All participants identified as cismale, were 18 years of age or older; and were able to speak and read in English. In order to attend sessions in person, both partners also reported a residence in the New York City metropolitan area. In each couple, at least one partner was 18-29 years old, at least one was HIV-negative, and at least one reported substance use in the past 30 days (either a day of heavy drinking or illicit drug use). Eligible couples reported a relationship duration of at least three months and recent sexual activity (together or with outside partners).
Procedures
The study utilized both online and in-person recruitment strategies. Online recruitment efforts included advertisements on Facebook as well as websites and popular geosocial networking apps for gay, bi, trans, and queer people seeking sexual partners. Interested participants who clicked on an ad were directed to a screening survey that assessed preliminary eligibility and gathered contact information. In addition, study staff went in person to bars, nightclubs, and other social events for SMM in the metropolitan area. Staff provided study information and also directly screened participants on an iPad device that accessed the same screener used in online recruitment. Study staff then contacted potentially eligible participants and completed a telephone screener with one member of the couple (the index partner).
Where telephone screening indicated couples were eligible, index participants were asked to schedule a baseline assessment appointment in our research center at a time when their partner could also attend. Partners were consented and completed baseline survey instruments separately. Regardless of condition, Session 1 was conducted immediately after baseline. Sessions 2 and 3 occurred at approximately one-week intervals thereafter. Each participant received $10 for completion of the baseline survey; $20 for completion of each intervention session (regardless of condition). All procedures were approved by the Institutional Review Board of Hunter College, CUNY.
The Couples Health Project (CHP) intervention condition comprised 3 sessions of couples MI, lasting 60-75 minutes each. All sessions were conducted with both partners together following principles for MI delivery described elsewhere (7,8). MITI coding was completed for Session 1 audio recordings. This session began with an emphasis on engagement and facilitating dyadic functioning. The clinician then introduced the target behavior of substance use and the couple completed a calendar describing their use (independently or together) in the past 30 days. Following procedures developed by Starks et al. (3) this calendar serves as a form of feedback and provides as a starting point to evoke change talk. The session ended with the creation of goals for the week ahead based upon the couples’ readiness for change. These plans incorporated attention to communication skills practice, pleasant event scheduling, and substance use and were tailored to couples’ goals.
Education Control was composed of 3 sessions, each lasting 45-70 minutes. Sessions were delivered in person by trained educators and consisted of a mixture of didactic presentations; interactive question-and-answer components; and discussion. Delivery of the condition was structured through the use of PowerPoint presentations, and supplemented with relevant publically-sourced video content. Session 1, examined in these analyses, provided an overview of psychological research on sex, relationships, and sexual health among men who have sex with men. Additionally, information on a variety of sexual safety strategies was delivered to the couple.
Measures
Demographics.
Participants reported their age, HIV status, sexual identity (gay or bisexual), race/ethnicity (Black, Latino, White, Other/Mixed; we collapsed several groups, e.g., Native American, Asian, into an Other/Mixed category given the small number of participants), income (below $20,000 and $20,000 or above annually), education (less than college or college and above), and relationship length.
MITI fidelity assessment
MITI coding was completed by a group of 4 experienced coders. All of them were trained on the MITI system by the first author and each had more than 3 years of coding experience. For the first 2 months of coding, MITI coders met weekly with the first author to develop definitions for novel codes and their application to session coding. Across these meetings, the coders and first author discussed codes and resolved disputes by consensus. Once the novel code definitions were established, all coders listened to 3 full-session recordings and completed MITI coding for reliability analyses. After establishing reliability, coders subsequently rated all initial sessions completed in the MI and Education control conditions.
Reliability for the 4 coders – across the 3 sessions that were coded for reliability analyses – was assessed using metrics of percent agreement for MITI codes with ranges of 5 or less and intra-class correlations (ICC) for MITI codes with ranges exceeding 5. All ICC’s were specified in SPSS version 25 as two-way mixed effects with absolute agreement. We utilized standard guidelines for the interpretation of the magnitude of ICCs used in medical and clinical research (9). ICCs less than 0.40 were considered poor; 0.40 to 0.59 were considered fair; 0.60 to 0.74 were considered good, and ICCs of 0.75 to 1.00 were considered excellent. Percent agreement was first calculated as exact agreement – a conservative threshold capturing the proportion of ratings that were identical. Subsequently, we also calculated the percentage of exact and nearly-exact (or adjacent) ratings (10). This latter metric is more liberal. Two raters are considered to “agree” on a rating if their scores differ by no more than 1 point (i.e., they give ratings that are either exact or at least “adjacent” to one another). Graham et al. (10) suggested that exact and nearly-exact agreement of 90% or more could be considered high, while 75% represented a minimum threshold for adequate agreement. Values closer to 90% should be expected when the more liberal calculation of exact or nearly-exact agreement is used.
Coders applied standard MITI global and behavior count codes. Exact agreement for global scores ranged from 56% for Partnership and Empathy to 77% for Cultivating Change Talk and 83% for Softening Sustain Talk. The percentage of exact and nearly exact agreement was very high, ranging from 0.94 to 1.0. For a number of behavior counts (Giving information; Persuade with Permission; Affirm; Seek; Emphasize Autonomy; and Confront) the scores across all sessions coded for reliability calculation ranged from 0 to 2. The ICC is a poor metric of reliability when range is restricted (9). As a result, percent agreement metrics were utilized instead. Affirm, Seek, Emphasize, and Confront codes all had 100% exact agreement. Persuade with Permission had exact agreement of 83% and exact or nearly-exact agreement of 100%. The Giving Information code had exact agreement of 28%; however, exact or nearly-exact agreement was substantially higher (83%) suggesting the magnitude of most disagreements was 1 point. The ICC for Persuade codes was excellent (ICC = 0.97) as was the ICC for Complex Reflections (ICC = 0.79). The ICC for Questions and Simple Reflections were fair in magnitude (0.41 and 0.43 respectively). Debriefing of scores revealed that these modest reliabilities were each adversely impacted by a single rating on one session. When these single ratings were corrected following consensus discussion, reliability for these scores was good to excellent (ICC = 0.74 for Questions and 0.91 for Simple Reflections).
Two new global scores were coded. Consistent with the general format of the MITI 4.2.3 coders assigned scores of 1 through 5 that captured the clinicians’ overall performance in these domains. These scores, like all MITI global scores, are best thought of as ordinal categories in which higher scores are indicative of better MI performance. Rather than a Likert-type scale, prototype descriptions of performance at each level (1 through 5) were developed for both global scores.
The Decreasing Dyadic Ambivalence global measured the extent to which the clinician reduced the couple’s ambivalence about behavior change. This scale was meant to capture efforts made by the clinician to help couples achieve consensus around behavior change and come to a shared understanding around behavioral goals. To achieve high scores, clinicians must work with the couple to improve dyadic agreement around behavioral goals and potential steps to achieve them. Similar to Softening Sustain talk, high scores can also be achieved in the absence of indications of dyadic ambivalence if the clinician does not engage in behaviors that evoke it. Therefore, like Softening Sustain Talk, the coder defaults to a “4” in the absence of dyadic ambivalence. Low scores on Decreasing Dyadic Ambivalence are appropriate when the clinician focuses on areas of disagreement that amplify divisions between the partners around the change behavior. This is true even if the clinician uses MI-consistent utterances to evoke and reflect couple disagreement. Low scores should be given when the clinician undermines the couple’s ability to achieve consensus on the target goal throughout the session. Exact agreement for decreasing dyadic ambivalence was 44%; however, exact or nearly exact agreement was 100% indicating that no disagreements were larger than 1 point in magnitude.
The Facilitating Dyadic Functioning global score measured the extent to which the clinician made efforts to identify and foster the couple’s strengths, investment, past successes, and overall relationship functioning. To achieve high scores, clinicians must recognize and elicit discussion of the couples’ strengths in order to enhance each partner’s emotional investment, commitment, and satisfaction in the relationship. Clinicians who score high on this scale may use various techniques such as Relationship Affirmations or Dyadic Reflections that highlight the partner’s similarities and their shared understanding of the target behavior. Additionally, clinicians may achieve high scores by eliciting the couple’s past successes in a manner that highlights their capacity to work together towards shared goals. Low scores are warranted when clinicians amplify or emphasize factors that would diminish satisfaction and investment, such as relationship problems or partners’ dissatisfaction with one another. Clinicians who score low on this scale may focus on relationship problems, confront one or both partners, fail to mitigate conflict that arises between partners, or focus primarily on each partner’s individual concerns rather than addressing the couple as a unit. Exact agreement was 56%; however, exact or nearly exact agreement was 94% indicating that only 1 difference between ratings exceeded 1.
Two new behavior count codes were applied. Similar to existing MITI behavior counts, these codes were applied to individual utterances (5). Traditional MITI rules for parsing utterances were utilized. Utterances were awarded a maximum of one code each.
Dyadic Reflections must be directed at the couple as a unit. These reflections either integrate content expressed by both partners or they demonstrate an understanding of the relationship as a whole. They serve the purpose of emphasizing similarities or differences between partners as well as the values and strengths of the relationship. At times, Dyadic Reflections may appear as two separate complex reflections in one volley, but if the clinician is integrating separate content stated by each individual into one reflection it is coded as a dyadic reflection. Reliability for Dyadic Reflections was good (ICC = 0.72).
A Relationship Affirm (RAF) is a clinician utterance that genuinely accentuates something positive about the couple and their relationship. To be considered, the utterance addresses the couples’ strengths, or emphasize the partners’ care, concern, and capacity for empathy with one another. Relationship Affirmations are also coded when the clinician emphasizes how the couple adaptively handled conflict or successfully addressed a challenge together. Exact agreement for Relationship Affirm was 50%, exact or nearly exact agreement was 61%. Debriefing of coding revealed that agreement was substantially diminished by the score of a single coder on one session. Correction of this code resulted in 78% exact or nearly exact agreement.
RESULTS
A total of 92 index participants completed the study screener. Of these, 59 (64.1%) were preliminarily eligible. Of these, 30 scheduled a baseline appointment. In total, 20 couples completed baseline appointments and were subsequently randomized.
Demographic data for the total sample are presented in Table I. Tests of between-condition differences on baseline demographic characteristics suggested that conditions were equivalent with respect to demographic covariates except race and ethnicity. Table 1 also contains baseline data for target behaviors of focus in the intervention. Rates of drug use and heavy drinking did not differ between conditions. Groups did not differ with respect to the odds of condomless anal sex with casual partners.
Table 1.
Demographic Characteristics
| Total | Motivational Interviewing | Education Control | |||||
|---|---|---|---|---|---|---|---|
|
|
|
|
|||||
| nindividuals = 40 | nindividuals = 20 | nindividuals = 20 | Test Statistic | ||||
|
|
|
|
|||||
| n | (%) | n | (%) | n | (%) | Wald χ2(1) | |
|
|
|
|
|
||||
| Race and Ethnicity | 4.67* | ||||||
| White | 19 | (47.5) | 13 | (65.0) | 6 | (30.0) | |
| Non-White | 21 | (52.5) | 7 | (35.0) | 14 | (70.0) | |
| Education | 1.88 | ||||||
| Less than a 4-year degree | 17 | (42.5) | 6 | (30.0) | 11 | (55.0) | |
| 4-year degree or more | 23 | (57.5) | 14 | (70.0) | 9 | (45.0) | |
| Income | 0.08 | ||||||
| Less than $20,000 per year | 19 | (47.5) | 9 | (45.0) | 10 | (50.0) | |
| $20,000 or more per year | 21 | (52.5) | 11 | (55.0) | 10 | (50.0) | |
| Sexual Identity | 0.22 | ||||||
| Gay | 33 | (82.5) | 17 | (85.0) | 16 | (80.0) | |
| Bisexual/Queer | 7 | (17.5) | 3 | (15.0) | 4 | (20.0) | |
| HIV status | † | ||||||
| Negative/Unknown | 38 | (95.0) | 20 | (100.0) | 18 | (90.0) | |
| Positive | 2 | (5.0) | 0 | (0.0) | 2 | (10.0) | |
| CAS with Casual Partners | 14 | (35.0) | 10 | (50.0) | 4 | (20.0) | 3.46 |
|
|
|
|
|||||
| M | SD | M | SE | M | SE | Wald χ2(1) | |
|
|
|
|
|||||
| Age | 25.50 | 4.33 | 25.45 | 1.25 | 25.55 | 1.04 | 0.004 |
| Relationship Length | 27.36 | 25.46 | 29.40 | 7.65 | 25.33 | 8.18 | 0.13 |
| Heavy Drinking Days | 5.93 | 5.07 | 5.15 | 1.34 | 6.70 | 1.30 | 0.69 |
| Drug Use Instances | 13.53 | 12.86 | 11.90 | 3.80 | 15.15 | 3.52 | 0.39 |
| Drug use Instances (excluding Marijuana) | 2.25 | 4.07 | 2.65 | 1.43 | 1.85 | 0.85 | 0.23 |
Test unavailable due to complete-separation of the data.
p < .05
CAS = Condomless Anal Sex
Between group differences in MITI Global scores
Consistent with expectations, coders rated clinicians in the MI condition significantly higher on Cultivating Change Talk, Partnership, and Empathy. The between-group difference on Softening Sustain talk was non-significant. This result is likely an artifact of the instructions to default to a score of 4.0 where no sustain talk is present to be softened. This score was commonly awarded to education condition sessions.
Coders rated MI clinicians significantly higher on both Decreasing Dyadic Ambivalence and Facilitating Dyadic functioning scores. Mean scores of 2.1 (SD = 0.74) and 1.6 (SD = 0.97) respectively in the control condition indicate that these factors were largely or entirely absent from most education sessions. Meanwhile, mean scores of 4.2 (SD = 0.63) and 4.10 (SD = 0.88) indicate that these factors were highly characteristic of MI sessions.
Between group differences in behavior counts
Analyses focused primarily on behavior counts and summary proportions that are typically used to evaluate proficient MI delivery. Clinicians in the MI condition used significantly more reflections overall, and also significantly more dyadic reflections specifically. The average number of reflections in the MI condition was 35.80 (SD = 19.52), Approximately 2.7 times the average number in the education condition (M = 13.20; SD = 9.47; t(18) = 3.29, p = .004, d = 1.47). Meanwhile, the use of dyadic reflections was rare in the education condition (M = 1.60; SD = 1.78), while occurring with substantial frequency in the MI condition (M = 14.40; SD = 7.04; t(18) = 5.57, p < .001, d = 2.49). Clinicians in the MI condition also made significantly more MI consistent statements (M = 2.60, SD = 1.51) compared to those in the education condition (M = 0.20, SD = 0.63; t(18) = 4.65, p = .001, d = 2.07). In contrast, clinicians in the education condition had significantly higher scores on Giving Information (M = 18.90, SD = 5.72) compared to those in the MI condition (M = 7.20, SD = 13.15; t(18) = −2.58, p = .019, d = 1.15). The groups did not differ with respect to the average number of Questions (MMI = 31.20, SD = 16.77; Meducation = 31.90, SD = 12.78) and Persuade codes (MMI = 1.10, SD = 1.52; Meducation = 0.30, SD = 0.67).
The reflection to question ratio was significantly higher in the MI condition, as was the dyadic reflection to question ratio. MI clinicians utilized approximately 1.28 (SD = 0.63) reflections per question whereas clinicians in the education condition utilized only 0.38 (SD = 0.16) reflections per question, t(18) = 4.37, p = .001, d = 1.96. Differences in Dyadic Reflection to Question ratio were even more pronounced. Clinicians in the education condition employed an average of only 0.17 (SD = 0.18) dyadic reflections per question versus those in the MI condition who employed an average of 0.43 (SD = 0.22) dyadic reflections per question, t(18) = 2.98, p = .008, d = 1.29. Interestingly, the percentage of reflections coded as complex did not differ between conditions (MMI = 0.35, SD = 0.13; Meducation = 0.36, SD = 0.18).
DISCUSSION
These findings provide initial evidence that clinician skills and techniques used in MI with couples can be reliability coded using an expanded application of the MITI 4.2.3 system. Novel global scores and behavior counts specific to couples that demonstrated adequate to excellent reliability and differed as anticipated between the intervention and control conditions. In addition, clinicians in the MI condition obtained MITI scores that indicated significantly higher proficiency in MI than those in the education condition for 3 out of four established MITI global scales and the majority of behavior counts. This suggests that the implementation of skills unique to MI with couples does not substantially disrupt fidelity to other core aspects of MI.
We found that Decreasing Dyadic Ambivalence and Cultivating Change Talk scores were significantly higher in the MI condition. This is consistent with Starks et al’s (7) assertion that addressing dyadic ambivalence, and achieving consensus around change is a prerequisite to evoking change talk. Many recently developed dyadic interventions addressing HIV risk and drug use rely on agreement formation between partners as a mechanism of individual behavior change (e.g., 3, 4). These findings suggest that objective ratings of provider skills directly relevant to agreement formation and subsequent discussions of change might be applied to inform fidelity monitoring and provider training.
Starks and colleagues (7, 8) suggested that facilitating dyadic functioning is an essential process unique to MI with couples. The findings of this study provide some initial indication of the construct validity of their suggestion. Coders were able to reliably evaluate the clinicians’ use of this process. In addition, clinicians in the MI condition obtained significantly higher scores than educators in the control condition. Notably, the education condition contained information about safer sex practices in couples specifically. The fact that educators obtained consistently lower scores on this global item suggests that merely providing information tailored for couples is functionally different from the skills and strategies captured by the Facilitating Dyadic Functioning process.
Dyadic reflections and relationship affirmations have been suggested as specific techniques that serves to enhance dyadic functioning, decrease dyadic ambivalence, and capture the couples’ motivation and potential for change (8). The findings of the current study suggest that coders can reliably recognize these strategies in session based upon audio recordings alone and without the benefit of visual information indicating whether the clinician is addressing one participant or both members of the couple together. Interestingly, it is at least plausible that the use of dyadic reflections may have been partly responsible for the absence of between condition differences in the percentage of complex reflections. Coders followed an exclusion rule wherein if a reflection was dyadic as well as complex (or simple), only the dyadic code would be applied.
These findings should be understood in the context of several limitations. First, this study represents only a preliminary investigation of the reliability and validity of these novel MITI codes. Reliability findings warrant replication in larger samples of recordings, with a wider array of clients and more diverse coders. This study provides only one relatively modest indicator of validity – between-condition differences in MITI scores. While this distinction is relatively crude, it is commensurate with this early stage of measurement development. Future studies should examine concurrent validity with other measures of intervention quality and also the predictive validity of novel MI scores and intervention outcomes.
Despite these limitations, the current study is the first to propose and pilot test fidelity codes that captures clinician strategies and techniques unique to conducting MI with couples. Findings indicated that coders can reliably recognize both global MI processes unique to couples (Decreasing Dyadic Ambivalence and Facilitating Dyadic Functioning). They were also able to reliably identify the use of specific MI techniques applicable to couples’ MI (dyadic reflections and relationship affirmations). These findings support continued research utilizing these novel extensions of the MITI 4.2.3 system to evaluate the fidelity of MI delivery with dyads in MI paradigms informed by Interdependence Theory (e.g., 3, 7, 8).
Table 2.
Between condition differences in Session 1 MITI scores
| Total | Motivational Interviewing | Education Control | ||||||
|---|---|---|---|---|---|---|---|---|
|
|
|
|
||||||
| ncouples = 20 | ncouples = 10 | ncouples = 10 | Test Statistic | |||||
|
|
|
|
||||||
| M | SD | M | SD | M | SD | t(18) | p | |
|
|
|
|
||||||
| Global Scores | ||||||||
| Cultivating Change | 2.55 | 1.32 | 3.40 | 0.84 | 1.70 | 1.16 | 3.75 | .001 |
| Softening Sustain | 3.70 | 0.57 | 3.50 | 0.71 | 3.90 | 0.32 | −1.63 | .127 |
| Decreasing Dyadic Ambivalence | 3.15 | 1.27 | 4.20 | 0.63 | 2.10 | 0.74 | 6.83 | <.001 |
| Partnership | 3.40 | 1.50 | 4.20 | 0.63 | 2.60 | 1.71 | 2.77 | .018 |
| Empathy | 3.50 | 1.57 | 4.50 | 0.53 | 2.50 | 1.65 | 3.65 | .004 |
| Enhancing Dyadic Function | 2.85 | 1.57 | 4.10 | 0.88 | 1.60 | 0.97 | 6.06 | <.001 |
| Behavior Count Totals | ||||||||
| All Reflections | 24.50 | 18.91 | 35.80 | 19.52 | 13.20 | 9.47 | 3.29 | .004 |
| Dyadic Reflections | 8.00 | 8.25 | 14.40 | 7.04 | 1.60 | 1.78 | 5.57 | <.001 |
| Questions | 31.55 | 14.51 | 31.20 | 16.77 | 31.90 | 12.78 | −0.11 | .918 |
| Persuade (With and Without Permission) | 0.70 | 1.22 | 1.10 | 1.52 | 0.30 | 0.67 | 1.52 | .154 |
| Giving Information | 13.05 | 11.55 | 7.20 | 13.15 | 18.90 | 5.72 | −2.58 | .019 |
| MI Consistent | 1.40 | 1.67 | 2.60 | 1.51 | 0.20 | 0.63 | 4.65 | .001 |
| Behavior Count Ratio scores | ||||||||
| Dyadic Reflection Percent | 0.30 | 0.24 | 0.43 | 0.22 | 0.17 | 0.18 | 2.98 | .008 |
| Complex Reflection Percent | 0.36 | 0.15 | 0.35 | 0.13 | 0.36 | 0.18 | −0.24 | .816 |
| Reflection to Question Ratio | 0.83 | 0.64 | 1.28 | 0.63 | 0.38 | 0.16 | 4.37 | .001 |
| Dyadic Reflection to Question Ratio | 0.29 | 0.31 | 0.53 | 0.26 | 0.05 | 0.05 | 5.67 | <.001 |
| Relationship Affirm (% of MI Consistent) | 0.25 | 0.37 | 0.49 | 0.39 | 0.00 | 0.00 | 3.95 | .003 |
Acknowledgements
Collection of these data were supported in part by a National Institute on Drug Abuse grant (R34 DA036419; PI Starks) and an award from the Hunter College Presidential Fund for Faculty Advancement. Analyses of these data were funded by a National Institute on Drug Abuse grant (R34 DA043422; PI Starks). The authors acknowledge the contributions of the Couples Health Project Team, particularly Trey Dellucci, Daniel Sauermilch, and Paula Bertone as well as Jeffrey T. Parsons, Mark Pawson, Andrew Cortopassi, Nahuel Smith, Ruben Jimenez, Scott Jones, Alex Brousset, and Jonathan Lassiter. Finally, we thank the staff and interns who contributed to the project and our participants.
Funding:
Collection of these data were supported in part by a National Institute on Drug Abuse grant (R34 DA036419; PI Starks) and an award from the Hunter College Presidential Fund for Faculty Advancement. Analyses of these data were funded by a National Institute on Drug Abuse grant (R34 DA043422; PI Starks).
Footnotes
Conflict of Interest: No conflict of interest declared.
Ethical approval: The study was approved by the Institutional Review Board at Hunter College of the City University of New York performed in accordance with the ethical standards as laid down in the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards.
Consent to participate: All participants were provided written informed consent information and this was reviewed verbally prior to the start of the study.
Consent for publication: not applicable
Availability of data and material: Data not publicly available. Please contact the corresponding author.
Code availability: Descriptive statistics and reliability analyses were conducted using SPSS version 25.
REFERENCES
- 1.Centers for Disease Control and Prevention. HIV Surveillance Report, 2018. 2020. [Google Scholar]
- 2.Sullivan PS, Salazar L, Buchbinder S, Sanchez TH. Estimating the proportion of HIV transmissions from main sex partners among men who have sex with men in five US cities. AIDS. 2009;23(9):1153–1162. [DOI] [PubMed] [Google Scholar]
- 3.Starks TJ, Dellucci TV, Gupta S, et al. A Pilot Randomized Trial of Intervention Components Addressing Drug Use in Couples HIV Testing and Counseling (CHTC) with Male Couples. AIDS Behav. 2019;23(9):2407–2420. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Newcomb ME, Sarno EL, Bettin E, et al. Relationship Education and HIV Prevention for Young Male Couples Administered Online via Videoconference: Protocol for a National Randomized Controlled Trial of 2GETHER. JMIR Res Protoc. 2020;9(1):e15883. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Moyers TB, Manuel JK, Ernst D. Motivational Interviewing Treatment Integrity Coding Manual 4.0: Unpublished manual; 2014. [Google Scholar]
- 6.Apodaca T, Manuel JK, Moyers T, Amrhein P. Motivational interviewing with significant others (MISO) coding manual. Unpublished manuscript 2007; https://casaa.unm.edu/download/miso.pdf. [Google Scholar]
- 7.Starks TJ, Millar BM, Doyle KM, Bertone PB, Ohadi J, Parsons JT. Motivational Interviewing with couples: A theoretical framework for clinical practice illustrated in substance use and HIV prevention intervention with gay couples. Psychology of Sexual Orientation and Gender Diversity. 2018;5(4):490–502. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Starks TJ, Robles G, Doyle KM, et al. Motivational Interviewing with male couples to reduce substance use and HIV risk: Manifestations of partner discord and strategies for facilitating dyadic functioning. Psychotherapy. 2020;57(1):58–67. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Li L, Zeng L, Lin Z, Cazzell M, Liu H. Tutorial on the use of intraclass correlation coefficients for assessing intertest reliability and its application in functional near-infrared spectroscopy-based brain imaging. Journal of Biomedical Optics. 2015;20(5):050801. [DOI] [PubMed] [Google Scholar]
- 10.Graham M, Milanowski A, Miller J. Measuring and promoting inter-rater agreement of teacher and principal performance ratings. Online Publication. October 12, 2020. 2012. https://files.eric.ed.gov/fulltext/ED532068.pdf [Google Scholar]