The Online OurRelationship Program for Relationally Distressed Individuals: A Pilot Randomized Controlled Trial

Abstract

Relationship distress has deleterious effects on mental health, physical health, and quality of life. Although many couples report relationship distress, one barrier to seeking services is that one member of a couple may be too busy or refuse to participate. Relationship interventions offered to individuals have shown promising efficacy, but, as most are offered in-person, barriers to their reach remain. To increase the reach of such interventions, the present pilot study examined the efficacy of a fully web-based relationship intervention for individuals (OR-I). The program was adapted from the couple version of the OurRelationship program (OR-C), which is effective in improving relationship and individual functioning (Doss et al., 2016). Results indicated that couples randomized to the OR-I program, compared to couples in a waitlist control group, saw significant improvement in quality of life (d = 0.69), work functioning (d = 0.44), and perceived health (d = 0.49) during treatment. Furthermore, gains in quality of life and perceived health for the intervention group were maintained over short-term follow-up. However, there were no significant improvements in relationship functioning or symptoms of depression or anxiety. When comparing the efficacy of OR-I and a demographically-matched subsample of OR-C, results showed that change in outcomes did not significantly differ by program. Overall, with some adaptations, OR-I may be a viable option for individuals seeking relationship help. Limitations, potential adaptations, and future directions are discussed in-depth.

Within the United States, approximately one third of couples are relationally distressed (Whisman, Beach, & Snyder, 2008) and one half of first marriages end in divorce within 20 years (Copen, Daniels, Vespa, & Mosher, 2012). Moreover, high levels of relationship distress and divorce longitudinally predict poorer global life satisfaction (Be, Whisman, & Uebelacker, 2013) and are associated with greater levels of depressive symptoms, anxiety, and substance use (e.g., Proulx, Helms, & Buehler, 2007; Whisman, 2007), increased risk for metabolic syndrome (Whisman & Uebelacker, 2012), and greater impairment in social and work functioning (Whisman & Uebelacker, 2006).

To address these concerns, two types of in-person couple interventions have been developed: relationship education and couple therapy. In the United States, about 50% of couples seek relationship education, which is most commonly offered as premarital education (Doss, Rhoades, Stanley, & Markman, 2009). Although relationship education shows positive improvements in relationship quality through short-term follow-up, the effects are small (Cohen’s d = 0.36 at post-treatment and 0.31 at follow-up; Blanchard, Hawkins, Baldwin, & Fawcett, 2009; Hawkins, Blanchard, Baldwin, & Fawcett, 2008). Additionally, programs are rarely attended by couples at high risk for future relationship distress such as cohabitating couples, African American couples, and non-religious couples (Doss et al., 2009; Halford, O’Donnell, Lizzio, & Wilson, 2006). While couple therapy has shown greater efficacy in significantly improving relationship satisfaction by end of treatment than relationship education, with effects in the medium to large range (Cohen’s d = 0.56–0.82; Lebow, Chambers, Christensen, & Johnson, 2012), it too has limitations. Indeed, couples show an overall 30–60% deterioration rate in relationship functioning 2–5 years after ending treatment (e.g., Christensen, Atkins, Baucom, & Yi., 2010; Cookerly, 1980; Jacobson, Schmaling, & Holtzworth-Munroe, 1987; Snyder, Wills, & Grady-Fletcher, 1991). Moreover, couple therapy is underutilized by distressed couples, with only 19% of current dyads having previously sought couple therapy and only 37% of divorced couples having attended couple therapy prior to separation (Johnson et al., 2002).

The limited reach of both relationship education and couple therapy likely reflects barriers common to both couple interventions and individual mental health services, such as financial burden for low-income individuals, perceived stigma, and logistical concerns such as childcare, regular transportation to appointments, and difficulty in scheduling (Hoge et al., 2004). Consequently, brief, low-cost interventions with more “flexible delivery” formats are needed to circumvent treatment barriers and increase reach. These types of interventions may be attractive to a larger percentage of couples and more easily accessible.

Numerous interventions with varying delivery formats have been developed with these goals in mind. For example, the Marriage Checkup is a brief two-session assessment and feedback intervention which demonstrates small intervention effects with mild to moderately distressed couples (d = 0.11 to 0.39; Cordova et al., 2015); however, as the intervention requires face-to-face interaction, it is less accessible than other interventions with more flexible, remote administration. More easily accessible, DVD-based relationship education programs have shown small to medium effects (d = 0.21–0.50; Bodenmann, Hilpert, Nussbeck, & Bradbury, 2014; Halford, Moore, Wilson, Farrugia, & Dyer, 2004; Halford et al., 2015) with larger effects demonstrated in trials with more distressed couples (e.g., Halford et al., 2015). Given their effectiveness and potential for high reach (e.g., as indicated by low risk of administration, low cost, and high couple acceptability), these flexibly administered programs may increase the population-level impact of couple interventions if implemented on a large scale.

Of all “flexible delivery” options, those delivered over the computer/internet are arguably the most accessible as they may reach couples who often do not seek or have access to in-person interventions. Indeed, 70% of U.S. households have access to broadband/cable internet (File, 2013) including a majority of low-income families with smartphones or computers (79%; Zickuhr & Smith, 2013). Furthermore, two recent reports indicate that 83–98% of African-Americans (Smith, 2014) and 80–89% of Hispanics (Lopez, Gonzalez-Barrera, & Patten, 2013) under the age of 50 have access to broadband/cable internet. Moreover, many individuals interested in couple-based treatments report they would prefer seeking help via a web-based intervention over other available methods (Georgia & Doss, 2013).

One of the most studied web-based relationship interventions, ePREP (Braithwaite & Fincham, 2007), is a relationship education program translated from the in-person PREP program (Markman, Stanley, & Blumber, 2001). While ePREP has been shown to improve facets of relationship functioning such as constructive communication (e.g., Braithwaite & Fincham, 2007, 2011), the intervention has only been tested with non-distressed couples and does not consistently improve relationship satisfaction (e.g., Braithwaite & Fincham, 2011, 2014). To date, the only “flexible delivery” intervention which has targeted distressed couples is the OurRelationship.com program (OR-C; Doss et al., 2016), a web-based translation of Integrative Behavioral Couple Therapy (IBCT; Christensen & Jacobson, 2000).

During the OR-C program, couples first identify a “core issue” or their biggest relationship problem. Couples then work to develop a more accurate and objective understanding of the core issue, learn about acceptance and self-change, and develop behavioral solutions to ameliorate the core issue. The program is divided into three phases and takes 5–6 hours to complete over a 4- to 6-week period. Couples work through material individually and then come together for conversations at the end of each of the three phases to share what they learned. They then meet with a “coach”, a paraprofessional who monitors progress and provides technical assistance during three brief, 15-minute calls via Skype/phone. In a randomized controlled trial of 300 relationally distressed heterosexual couples, compared to a waitlist control group, OR-C resulted in significant pre-post improvements across several domains, such as relationship satisfaction (d = 0.69), relationship confidence (d = 0.47), work functioning (d = 0.19), quality of life (d = 0.18), perceived health (d = 0.23), depressive symptoms (d = 0.50), and anxiety symptoms (d = 0.21). Further, as almost one third of couples were low-income, a population that often faces financial barriers to seeking treatment, results support the possibility that web-based interventions can circumvent obstacles to seeking in-person treatment (Doss et al., 2016).

Overall, despite strengths of flexible couple-based interventions, one important limitation they share with traditional, in-person couple interventions is they require attendance by both members of a couple. However, individuals may not want to include their partner in treatment or their partners may be too busy or refuse to participate. For example, in a study that surveyed recently divorced individuals, about one third of individuals who did not seek help before divorce cited their partner’s unwillingness as a reason for not seeking couple therapy (Wolcott, 1986). Unfortunately, research suggests that in-person interventions offered to only one member of a couple, such as individual therapy, are not very effective in improving relationship distress for relationally distressed couples (e.g., Atkins, Dimidjian, Bedics, & Christensen, 2009; Emanuels-Zuurveen & Emmelkamp, 1996; Foley, Rounsaville, Weissman, Sholomskas, & Chevron, 1989). Additionally, while individual-oriented, in-person relationship education programs such as the Within My Reach program (WMR; Pearson, Stanley, & Rhoades, 2008), Fatherhood, Relationship, and Marriage Education intervention (FRAME; Wadsworth et al., 2011), and PREP Inside and Out Program (Einhorn et al., 2008) have shown small improvements across multiple relationship functioning domains by end of treatment (e.g., Antle et al., 2013; Einhorn et al., 2008; Wadsworth et al., 2011), these interventions did not include a majority of relationally distressed individuals. Thus, it is unclear how results would generalize to individuals with more significant relationship distress.

Moreover, as is true for all in-person interventions, barriers such as access to treatment, stigma, and logistic concerns are still present. As such, there has been a recent push for individual-oriented programs offered in more flexible formats (Rhoades & Stanley, 2011), such as through the internet. A recent study found that individuals who sought a web-based relationship intervention alone were more relationally distressed than couples who sought the program together (Nowlan, Cicila, & Doss, 2014), further demonstrating a need for an efficacious web-based option. In the only study to examine a web-based program for individuals, the Power of Two Online program showed small but significant improvements in relationship satisfaction and conflict management for new and expectant parents by the end of treatment (Hedges g of 0.2–0.49; Kalinka, Fincham, & Hirsch, 2012). However, due to an almost 50% rate of program non-completion and greater than 50% attrition at post-treatment, the intent-to-treat effects of the program are unclear. Moreover, as the sample included relationally satisfied individuals, no study has yet explored a web-based program targeting distressed individuals.

To address this gap in the literature, the present pilot study endeavors to test the efficacy of a web-based relationship intervention for relationally distressed individuals and explore future program development. The intervention, OurRelationship for Individuals (OR-I), was adapted from the OR-C program (Doss et al., 2016). The study’s first aim was to test OR-I’s efficacy on both relationship and individual functioning by the end of treatment. Given past research showing that individual-oriented programs produce small improvements in certain relationship and individual functioning domains (e.g., FRAME; Wadsworth et al., 2010; WMR; Pearson et al., 2008; PREP Inside and Out; Einhorn et al., 2008), we hypothesized that the program would also result in small, but significant improvements during the intervention period. The second aim was to test the hypothesis that pre-post gains would be maintained through short-term follow-up for the intervention group. We expected that domains that did not improve during the active intervention would improve over the follow-up period because it would give individuals more opportunities to implement changes in their relationship. The third aim of the study was to test for differential rates of pre-post change in relationship and individual functioning between individuals in OR-I and a demographically-matched subsample of individuals in OR-C (comparing each intervention group to their own waitlist control group). As it is harder to change relationship dynamics when only one person is attempting to change and because OR-I participants may have been less committed to their relationships (e.g., program may have been used as a last ditch effort before ending the relationship), we hypothesized that OR-I participants would show significantly less change in relationship functioning during treatment than those in the demographically-matched subsample of OR-C. Lastly, informed by study results, the final aim was to explore potential future program adaptations.

Method

Participants

Ninety individuals in heterosexual relationships from 35 states participated in the present study. The sample included eligible individuals who enrolled directly in the OR-I program and others who applied to join the OR-C program and were eligible, but because their partners refused participation, were offered participation in the individual program. Not all participants who applied directly to OR-I informed us of why they chose the individual program; however, some indicated they did not want to involve their partner (perhaps because they were considering whether to end their relationship) or that their partner was unwilling/unavailable to participate.

Seventy-four percent of participants were married, 14.4% were engaged, and 11.1% had lived with their partner for more than 6 months. The average length of relationship was 10.55 years (Mdn = 8.0; SD = 8.86) and 77.8% of the sample had children living with them. Participants were primarily female (80%), in their late thirties (M = 36.84; SD = 10.47), and employed part (13.3%) or full time (64.4%). Individuals reported their highest level of education as high school or less (31.1%), some college or technical training (25.6%), or a Bachelor’s degree or higher (43.3%), Moreover, participants reported a median annual household income of $55,000 (M = $69,868; SD = $78,617). Most of the participants were White, non-Hispanic (66.7%) or African American (14.8%) with smaller numbers of Asian/Pacific Islander (5.6%), and American Indian/Alaska Native (2.3%) participants. Ten percent of participants identified as Hispanic. As measured by the four-item Couples Satisfaction Index, 98% of participants presented to the study with relationship satisfaction in the distressed range (<13.5; Funk & Rogge, 2007; M = 6.32; SD = 3.48).

As described in detail in the Data Analyses section, propensity scores were used to create the demographically-matched subsample of individuals in OR-C, which resulted in a sample of 210 individuals. There were no significant demographic differences between the two samples. Like participants in OR-I, individuals in the demographically-matched subsample of OR-C were primarily female (85%), in their late thirties (M = 36.70; SD = 10.20), and mostly Caucasian (84.2% White, non-Hispanic; 15.8% African American). About 6% of individuals identified as Hispanic. Additionally, the majority of the demographically-matched subsample were employed full (61.7%) or part (10.8%) time. Participants in this subsample reported their highest level of education as high school or less (22.5%), some college (24.2%), or a Bachelor’s degree or higher (53.3%). The subsample’s mean household income was $88,704 (SD = $82,264). On the four-item Couples Satisfaction Index (Funk & Rogge, 2007), the subsample had a similar level of relationship satisfaction as individuals in OR-I (M = 6.66, SD = 3.66).

Procedures

All procedures were approved by the university IRB. Advertisements on Facebook and Google AdWords, unpaid “organic” search results, and media coverage directed individuals to the program’s web page and an online eligibility screener. To meet eligibility, individuals had to be at least 21 years old, in a heterosexual relationship, be married, engaged, or living with their partner for at least 6 months, have access to high-speed broadband/cable internet, and report moderate to severe levels of relationship distress (scoring at least 0.5 standard deviations below the population mean). Individuals were excluded from the study if they lived outside the U.S., were attending or seeking couple therapy, reported an affair during the last month, had decided to separate, reported intimate partner violence with actual or feared injury during the past 3 months, or experienced moderate to severe suicidal thoughts during the past 3 months.

All eligible individuals (including those whose partners refused participation) were sent an email encouraging them to complete OR-C instead of OR-I as it was hypothesized that the couple version would be more effective. Those who remained interested in OR-I completed a second online assessment followed by a brief call with project staff to give verbal consent and ask questions about participation. They were then randomized to either the intervention (n = 46) or to a 1-month waitlist (n = 44) condition. There were no significant pre-treatment between-group differences on any dependent or demographic variables.

Treatment Description

OR-I’s goal was to help individuals work through a specific relationship problem by promoting emotional acceptance and positive behavior change to ameliorate negative relationship patterns. Eighty percent of individuals randomized to the intervention condition completed the program over a 1-month period (see Figure 1); mean completion time was 25.46 days (SD = 10.87). Intervention material—which took 4–6 hours to complete—was generally consistent with OR-C including being divided into three phases (e.g., Observe, Understand, and Response). However, individuals did not have conversations with their partners throughout the program as they did in OR-C. Instead, following the intervention, participants had the opportunity to share their work with their partner via a letter. We did not encourage participants to share before that point because we felt that, with only one partner working on the relationship, sharing with a partner before gaining sufficient acceptance of relationship problems and committing to self-change could lead the couple to fight and fall into their maladaptive patterns, create more resentment, and decrease likelihood for positive change.

Figure 1.

CONSORT Figure

a Numbers do not sum to 349 because some individuals were ineligible for multiple reasons.

b Declined indicates that participant did not answer any questions at this assessment time point.

As part of the program, individuals had three brief Skype/phone calls with an assigned coach; the coaches included three graduate students and a Bachelor’s-level project coordinator. Unlike during the OR-C program, individuals did not have a scheduled call with their coach at the end of the Understand phase (because there was no conversation with the partner to debrief). Across the three calls, the average total call time was 22.08 minutes (SD = 10.72). Each of the calls was tightly scripted, with coaches answering questions related to program content, addressing technical problems, and monitoring progress. Coaches did not provide therapy or provide information that was not covered in the online program. Following each call, coaches reported adherence to the call script. In total, two deviations were reported (2.6% of calls). For both deviations, individuals were not given all script information during the call; however, they were subsequently provided this information via e-mail.

Measures

Measures were administered to all participants at pre-, mid-, and post-treatment. Participants in the intervention condition completed two additional brief assessments during the program and a more extensive 3-month follow-up assessment. Descriptive statistics for outcomes by condition and assessment time point are presented in Table 1.

Table 1.

Relationship and Individual Functioning Outcomes by Condition and Assessment Point

	Intervention Group						Waitlist Control Group
	Pre	Post O	Mid	Post U	Post	Three- Month	Pre	Post O	Mid	Post U	Post	Three- Month
Relationship Measures
Relationship Satisfaction	6.35	8.21	9.95	9.97	9.95	10.20	6.30	—	8.02	—	7.75	—
	(3.39)	(4.25)	(4.40)	(4.45)	(5.52)	(6.14)	(3.61)		(5.21)		(4.97)
Relationship Confidence	7.46	7.39	9.11	9.08	8.53	8.27	6.36	—	7.31	—	7.18	—
	(2.66)	(3.10)	(2.60)	(2.51)	(3.38)	(3.83)	(3.46)		(3.35)		(3.51)
Positive Relationship Quality	16.96	14.53	16.27	16.32	16.08	16.55	17.23	—	14.71	—	15.68	—
	(4.83)	(6.02)	(5.44)	(5.68)	(5.58)	(5.34)	(5.29)		(6.38)		(6.42)
Negative Relationship Quality	14.37	13.61	11.30	11.92	11.47	10.15	14.45	—	13.05	—	13.65	—
	(6.42)	(6.01)	(6.24)	(6.10)	(7.42)	(7.68)	(5.65)		(6.32)		(6.43)
Individual Measures
Depressive Symptoms	11.93	—	—	—	8.50	7.11	12.68	—	—	—	9.83	—
	(6.34)				(5.74)	(6.72)	(6.73)				(6.44)
Anxiety Symptoms	10.54	—	—	—	6.21	5.50	9.52	—	—	—	6.55	—
	(6.00)				(4.74)	(6.26)	(6.05)				(4.58)
Quality of Life	3.02	—	—	—	3.89	3.83	3.39	—	—	—	3.55	—
	(1.06)				(0.86)	(0.85)	(0.99)				(0.85)
Work Functioning	3.76	—	—	—	4.18	3.86	3.64	—	—	—	3.63	—
	(0.95)				(0.73)	(0.96)	(0.99)				(0.95)
Perceived Health	3.37	—	—	—	3.89	3.81	3.39	—	—	—	3.48	—
	(0.93)				(0.95)	(0.92)	(0.99)				(1.15)

Note: SDs are in parentheses

Demographic information

At pre-treatment, participants reported their age, gender, ethnicity, education, employment status, household income, and number of children. Participants also reported relationship status at pre-treatment, post-treatment, and 3-month follow-up.

Relationship satisfaction

Global relationship satisfaction was measured using the Couples Satisfaction Index-4 Item (CSI-4; Funk & Rogge, 2007). Lower scores indicate greater relationship distress with a clinical cut-off score of 13.5 (Funk & Rogge, 2007). Responses in this sample showed good internal consistency (α = .89)

Positive and negative relationship quality

Positive and negative relationship quality were assessed using the Positive and Negative Relationship Quality Scale (PNRQ; Fincham & Rogge, 2010). The PNRQ is an 8-item self-report scale that asks participants to separately rate four positive (e.g., enjoyable, alive) and four negative (e.g., bad, lifeless) dimensions of relationship quality. Higher scores on the positive relationship quality scale and lower scores on the negative relationship quality scale indicate greater relationship quality. In this study, α =.91 for the positive scale and α =.90 for the negative scale.

Relationship confidence

Following Rhoades, Stanley, and Markman (2009), relationship confidence was measured by combining two items (“I believe we can handle whatever conflicts will arise in the future” and “I feel good about our prospects to make this relationship work”) from the 10-item Confidence Scale (Stanley, Hoyer, & Trathen, 1994). Lower scores indicate less relationship confidence. Cronbach’s alpha was .80 in the present study, indicating good internal consistency.

Anxiety

Anxiety symptoms were measured using the Generalized Anxiety Disorder-7-item (GAD-7; Spitzer, Kroenke, Williams, & Lowe, 2006). Higher scores indicate a higher level of anxiety. Scores of 5, 10, and 15 indicate cut-off points for mild, moderate and severe anxiety, respectively. Cronbach’s alpha was .87 in this sample.

Depression

Depressive symptoms were assessed using the Center for Epidemiologic Studies Short Depression Scale (CES-D 10; Cole, Rabin, Smith, & Kaufman, 2004). A cut-off score of greater than or equal to 10 indicates risk for clinical depression. The subscale showed good internal reliability (α =.85).

Quality of life

Overall life satisfaction was assessed using one item from the abbreviated version of the World Health Organization Quality of Life-100 scale (WHOQOL-BREF; WHOQOL Group, 1998). Participants rated their quality of life during the past 4 weeks on a scale from 1 (very poor) to 5 (very good).

Perceived health

Perceived health was assessed using one item from the abbreviated version of the World Health Organization Quality of Life-100 scale (WHOQOL-BREF; WHOQOL Group, 1998). Participants rated how satisfied they were with their health during the past 4 weeks on a scale from 1 (very dissatisfied) to 5 (very satisfied).

Work functioning

Work functioning was measured with the following question: “Please rate your ability to function well at work on the following scale. If you do not work outside the home, please rate your ability to complete household tasks.” Responses were scored on a Likert scale from 1 (poor) to 5 (excellent).

Missing Data

Preliminary analyses were conducted to assess amount and patterns of missing data. Following the initial assessment, we collected at least one measure of relationship functioning in 99% of cases and at least one measure of individual functioning in 88% of cases. During the intervention period, relationship functioning data were missing at 9.9% of time points (13.0% in the intervention group and 4.5% in the waitlist group). Individual functioning data were missing at 6.7% of time points (8.7% in the intervention group and 4.5% in the waitlist group). At the 3-month follow-up (intervention group only), 23.9% of relationship functioning data and 21.7% of individual functioning data were missing. The high percentage of missing data at follow-up is in part because 10.9% of intervention couples had separated or divorced by 3-month follow-up; therefore, they were not asked to complete relationship functioning questions, which likely also reduced their desire to complete individual functioning measures. Missing data were not related to treatment condition or pre-treatment levels of any dependent variable. However, within the intervention condition, occurrence of missing data was related to program non-completion at 3-month follow-up and was thus controlled for in Aim 2 analyses.

Data Analyses

Aim 1

Due to the nested nature of the data, analyses were conducted within the Hierarchical Linear Modeling program (HLM; Raudenbush & Bryk, 2002) using two-level models where repeated measures at Level 1 were nested within individuals at Level 2. Time was entered at Level 1 [elapsed in weeks for relationship functioning and uncentered (0 = pre-treatment; 1 = post-treatment) for individual functioning]. Treatment condition (uncentered) and gender (grand mean centered) were added as Level 2 predictors of both the Level 1 intercept and slope. Power analyses were run in the Optimal Design Program; analyses showed we could detect an effect size of 0.66 with 80% power for relationship outcomes and an effect size of 0.59 with 80% power for individual outcomes.

Aim 2

To examine the short-term maintenance of effects through follow-up (intervention condition only), analyses were again conducted in HLM using two-level models where repeated measures were nested within individuals. An uncentered time term (0 = post-treatment, 1 = 3-month follow-up) was entered at Level 1. Gender and program completion status (0 = non-completer; 1 = completer) were grand mean centered and added as Level 2 predictors of both the Level 1 intercept and slope. To determine if there was sufficient power to detect effect sizes, power analyses were run using the G-Power Program. Results showed we could detect an effect size of 0.49 with 80% power for all Aim 2 analyses.

Aim 3

To compare the intervention effects for OR-I and OR-C, a demographically-matched subsample from OR-C was created using propensity score matching (Rosenbaum and Rubin, 1983). Propensity scores indicate the probability of being in a treatment condition based on measured covariates. In this study, propensity scores for participants in OR-I and OR-C were created using the following variables: age, gender, ethnicity, education, household income, and initial level of relationship satisfaction. Each OR-I participant was then matched with the participants from the full OR-C sample whose estimated propensity scores were the closest in number using the nearest neighbor matching approach. Given the nested nature of couple data, only one member of any couple from OR-C was included in analyses. Analyses were again conducted in HLM using two-level models. As was true for Aim 1, time was entered at Level 1 (elapsed in weeks for relationship functioning and uncentered for individual functioning). Additionally, treatment group (uncentered; 0 = OR-C; 1 = OR-I), treatment condition (uncentered; 0 = Waitlist; 1 =Intervention), gender (grand mean centered), and a group by condition interaction were added as Level 2 predictors of both the Level 1 intercept and slope.

Results

Aim 1: Treatment Gains During the Intervention Period

Full results for treatment gains during the intervention period and the between-group effect sizes can be found in Table 2. While changes in relationship outcomes were in expected directions (e.g., improvement), there were no significant differences between the intervention and waitlist conditions in rates of change in relationship satisfaction (b = 0.221, p = 0.107, d = 0.23), relationship confidence (b = 0.104, p = 0.432, d = 0.12), relationship negatives (b = −0.424, p = 0.095, d = −0.26), or positive relationship quality (b = 0.127, p = 0.523, d = 0.09). In contrast, compared to the waitlist group, the intervention group showed significantly more improvement in quality of life (b = 0.715, p = 0.002, d = 0.69), work functioning (b = 0.420, p = 0.022, d = 0.44), and perceived health (b = 0.466, p = 0.030, d = 0.49). While change was in the hypothesized direction for other individual functioning outcomes, the intervention and waitlist groups did not significantly differ in their amount of change in depressive (b = −0.349, p = 0.809, d = −0.05) or anxious (b = −1.378, p = 0.245, d = −0.23) symptoms during the intervention.

Table 2.

Change in Relationship and Individual Functioning During Intervention and Follow-up

	Between-Group Change Pre-Treatment to Post-Treatment					Within-Group Change (Intervention Only) Post-Treatment to Three-Month Follow-up
	b	SE	t	p	d	b	SE	t	p	d
Relationship Measures
Relationship Satisfaction	0.221	0.136	1.625	0.107	0.23	0.270	0.655	0.412	0.682	0.05
Relationship Confidence	0.104	0.131	0.790	0.432	0.12	−0.125	0.508	−0.245	0.808	−0.04
Positive Relationship Quality	0.127	0.198	0.641	0.523	0.09	0.598	0.662	0.904	0.372	0.11
Negative Relationship Quality	−0.424	0.252	−1.684	0.095	−0.26	−1.240	1.098	−1.129	0.267	−0.17
Individual Measures
Depressive Symptoms	−0.349	1.444	−0.242	0.809	−0.05	−0.935	0.961	−0.973	0.337	−0.16
Anxiety Symptoms	−1.378	1.179	−1.169	0.245	−0.23	−0.489	0.938	−0.522	0.604	−0.10
Quality of Life	0.715	0.213	3.3348	0.002	0.69	−0.105	0.138	−0.760	0.452	−0.12
Work Functioning	0.420	0.181	2.327	0.022	0.44	−0.349	0.164	−2.129	0.040	−0.48
Perceived Health	0.466	0.212	2.201	0.030	0.49	−0.116	0.154	−0.752	0.457	−0.12

Note: Bolded text indicates significance at p < .05.

Aim 2: Short-Term Maintenance of Treatment Effects

Full results for the amount of change in outcomes over follow-up for the intervention group and the within-group effect sizes are presented in Table 2. When exploring gains from post-intervention to 3-month follow-up for the intervention group, results demonstrated maintenance of treatment effects over follow-up in quality of life (b = −0.105, p = 0.452, d = −0.12), and perceived health (b = −0.116, p = 0.457, d = −0.12). In contrast, work functioning significantly deteriorated during the follow-up period (b = −0.349, p = 0.040, d = −0.48).

Contrary to our hypotheses, neither relationship functioning outcomes nor depressive or anxious symptoms showed significant within-group improvements between post-treatment and 3-month follow-up: relationship satisfaction (b = 0.270, p = 0.682, d = 0.05), relationship confidence (b = −0.125, p = 0.808, d = −0.04), positive relationship quality (b = 0.598, p = 0.372, d = 0.11), depressive (b = −0.935, p = 0.337, d = −0.16) and anxiety (b = −0.489, p = 0.604, d = −0.10) symptoms, and negative relationship quality (b = −1.240, p = 0.267, d = −0.17). Post-hoc analyses (available from the authors) also indicated that the intervention group did not show significant within-group gains in these variables from pre-treatment to 3-month follow-up.

Aim 3: Comparison of Treatment Gains Between OR-I and OR-C

For Aim 3 analyses, the three-way interaction of group by condition by time examined differential treatment effects. The interaction term indicated whether differences between the intervention and control groups were significantly different by treatment program over time.

Relationship functioning outcomes

When comparing relationship functioning outcomes between the intervention and waitlist conditions for individuals in OR-I and those in the matched subsample of OR-C, results revealed no significant interactions with treatment type for any of the four relationship functioning measures: relationship satisfaction (b = −0.181, SE = 0.164, t = −1.103, p = 0.272, d = −0.38), relationship confidence (b = 0.004, SE = 0.151, t = 0.026, p = 0.980, d = 0.01), positive relationship quality (b = −0.102, SE = 0.230, t = −0.445, p = 0.657, d = −0.14), and negative relationship quality (b = 0.026, SE = 0.290, t = 0.091, p = 0.928, d = −0.03). Thus, the difference in the linear rate of change in relationship functioning during the intervention (compared to a waitlist group) did not significantly differ between OR-I and OR-C.

Individual functioning outcomes

Results showed no significant differences between the treatment and waitlist conditions for the OR-I sample and for the matched subsample of OR-C in the pre-post amount of change for depressive (b = 1.681, SE = 1.821, t = 0.923, p = 0.357, d = 0.26) or anxiety symptoms (b = −1.142, SE = 1.524, t = −0.933, p = 0.352, d = −0.24), quality of life (b = 0.375, SE = 0.283, t = 1.323, p = 0.187, d = 0.37), perceived health (b = 0.254, SE = 0.300, t = 0.847, p = 0.398, d = 0.24), or work functioning (b = 0.064, SE = 0.281, t = 0.226, p = 0.821, d = 0.06). Taken together, results suggest that neither program outperformed the other on change in individual functioning during the intervention (compared to their own waitlist groups).

Discussion

Findings from the present study indicate the program was acceptable to clients, had good reach, and improved some individual functioning domains. Indeed, a nationally-representative sample from across all four contiguous U.S. time zones sought and enrolled in the OR-I program. Moreover, 97% of participants from OR-I reported they were mostly or very satisfied with the services received (analyses available from the first author). Furthermore, the program resulted in small-sized, significant between-group pre-post improvement in work functioning (d = 0.44) and perceived health (d = 0.49), and medium-sized, significant between-group improvement in quality of life (d = 0.69). Lastly, participants in the intervention group did not experience significant deterioration in quality of life (d = −0.12) or perceived health (d = −0.12) through short-term follow-up, suggesting general maintenance of treatment effects.

Treatment Gains During the Intervention and Follow-up Period

One interpretation for the significant gains in quality of life, work functioning, and perceived health during treatment, but not other variables, is that these variables encompass domains not specific to relationship functioning. Thus, these variables may have improved through an increase in participants’ coping ability, self-efficacy, and hope, even in the absence of change in relationship dynamics. Self-efficacy, or the belief that an individual can accomplish a task or achieve goals (Bandura, 1977), increases as individuals take steps to accomplish goals (Pomaki, Karoly, & Maes, 2009) and is related to overall well-being (David, Okazaki, & Sean, 2009). As study participants were very distressed (e.g., cut off scores indicate high levels of relationship distress, moderate levels of anxiety, and risk for clinical depression), many likely felt little hope in improving their relationship, especially if their partner refused to participate in treatment. As such, actively working on their relationship increased awareness of their own relationship patterns and relationships more broadly. This increased awareness may have provided more clarity on what to do in the relationship (e.g., planning to implement solutions or deciding to end the relationship), improved their coping ability, gave them a sense of resolve, and increased hope and self-efficacy. These changes in turn may have improved well-being (quality of life) and led to downstream effects on work functioning and perceived health during treatment. This interpretation accounts for the fact that some participants may have been more interested in gaining acceptance and improving coping to end the relationship—rather than improving the relationship itself-—which would explain less change in relationship functioning.

As results indicate, OR-I was less successful in improving relationship functioning. Given greater relationship distress, individuals who wanted to improve their relationship likely found it more difficult to make changes without their partners (i.e., in problem-solving activities). One possible conclusion then is that individual-oriented programs may be better suited for less relationally distressed individuals who may have an easier time engaging their partners in activities. Outcomes from other flexible individual-oriented programs with less distressed samples support this interpretation. Indeed, the programs offered to less relationally distressed individuals improve areas such as negative coping, problem solving, relationship satisfaction, and relationship confidence (e.g., FRAME; Wadsworth et al., 2010; WMR; Pearson et al., 2008; PREP Inside and Out; Einhorn et al., 2008; Power of Two; Kalinka et al., 2012). Moreover, past research indicates that when couples with greater relationship distress complete programs together, they often fair better in treatment than less distressed couples (e.g., Halford & Bodenmann, 2008; Halford et al., 2015; Hawkins et al., 2008). The reason interventions may be more effective for more relationally distressed couples is because there is more they can improve upon than less distressed couples. Also, with two partners in treatment (as compared to the present sample), changes are more attainable. Taken together, research suggests that highly relationally distressed couples are amenable to change, but treatment is most effective when both partners participate.

While OR-I had few statistically significant findings, the between-group effect size for improvement in relationship satisfaction during treatment (d = 0.23) was of similar magnitude to significant effects for other individual-oriented programs which had larger sample sizes. Indeed, PREP Inside and Out (d = 0.25 for men and 0.31 for women; Einhorn et al., 2008) and Power of Two (Hedges g = 0.24; Kalinka et al., 2012) both found small-sized between-group effects. The similarity in effect sizes is especially notable given the lower-intensity of contact with OR-I staff and the fact that OR-I was completed entirely online. However, comparing the between-group effect size for relationship confidence (the only other measure assessed in OR-I and another individual-oriented program) shows that PREP Inside and Out produced somewhat larger changes (OR-I: d = 0.12; PREP Inside and Out: d = 0.49 for men and 0.29 for women), possibly due to some OR-I participants leaning towards ending the relationship after treatment.

Regarding change over short-term follow-up, results showed no significant within-group deterioration in quality of life (d = −0.12) or perceived health (d = −0.12), suggesting general maintenance of treatment effects. In contrast, despite significant improvement during treatment, work functioning significantly deteriorated during follow-up (within-group d = −0.48). Finally, there were no significant improvements over follow-up in relationship functioning or symptoms of depression and anxiety. Additional analyses also revealed no significant changes in these variables from pre-treatment to 3-month follow-up. Overall, results provide further evidence that a partner’s involvement is needed for more long-lasting change, especially when there are high levels of relationship distress. Indeed, by 3-month follow-up, many couples were still highly relationally distressed and 10.9% of OR-I intervention couples had divorced.

Comparison of Treatment Gains between OR-I and OR-C

Results comparing changes during the intervention for OR-I participants to the matched subsample of OR-C showed no significant interaction effects across any outcomes, suggesting that one program was not significantly better than the other in improving relationship or individual functioning (compared to their own control groups). However, these non-significant findings may be a result of being underpowered to detect interaction effects. As such, results should be interpreted with caution. Still, a comparison of the between-group effect sizes for each program gives more nuanced information about how the programs faired against each other. For all relationship functioning measures, OR-C showed greater magnitude of change and significant between-group pre-post changes: relationship satisfaction (OR-C, d = 0.68; OR-I, d = 0.23), relationship confidence (OR-C, d = 0.21; OR-I, d = 0.12), positive relationship quality (OR-C, d = 0.28; OR-I, d = 0.09), and negative relationship quality (OR-C, d = −0.49; OR-I, d = −0.26). Moreover, the differences in the effect sizes indicate that in general, OR-C was at least twice as effective in improving relationship functioning. These results strengthen the argument that it is easier to intervene in relationship functioning when both members of the couple are involved. As such, programs should attempt to include both partners in some capacity when intervening with relationally distressed individuals who want to improve the relationship. Regarding individual functioning outcomes, OR-I created notable between-group gains in quality of life (OR-C, d = 0.51; OR-I, d = 0.69), work functioning (OR-C, d = 0.02; OR-I, d = 0.44), and perceived health (OR-C, d = 0.14; OR-I, d = 0.49); however, OR-C was almost 6 times as effective in reducing depressive symptoms (OR-C, d = −0.27; OR-I, d = −0.05) and almost twice as effective in reducing anxiety symptoms (OR-C, d = −0.47; OR-I, d = −0.23).

Limitations

While this study elucidated gaps in the literature on individual-oriented programs, a few limitations were noted that should be addressed in future adaptations. First, the study relied exclusively on self-report measures to assess change in functioning. Further, as self-efficacy and coping ability were not assessed, it was not possible to test whether these variables influenced quality of life as hypothesized. Thus, future adaptations should include measures of these variables to better understand the quality of life finding. Moreover, low statistical power limited our ability to detect small between-group differences and medium-sized interaction terms. For example, while the program yielded a small-sized pre-post effect in relationship satisfaction (d = 0.23) that was comparable in size to significant effects shown by other flexible individual-oriented programs (e.g., PREP Inside and Out; Einhorn et al., 2008; Power of Two; Kalinka et al., 2012), the difference between the intervention and waitlist condition in rate of change was not statistically significant in the current study in large part because of a difference in sample size. Another limitation is that information from the partners of participants was not collected; however, such information could provide a fuller picture of a couple’s functioning, which could potentially improve program efficacy. For example, if more information was known about the partner’s functioning, participants and coaches would gain more insight into the core issue.

A final limitation is that participants who enrolled directly in OR-I were not asked why they participated without their partner. One could hypothesize that the efficacy of an individual program might be different for individuals whose partners were too busy as compared to partners who refused to participate. This conclusion suggests program material should be tailored accordingly as refusing to participate in a program that could ameliorate relationship problems indicates less commitment and suggests larger relationship issues. Indeed, past research has shown that being less committed to the relationship and demonstrating a lack of relationship maintenance efforts may undermine the relationship and increase risk for relationship dissolution (Schoebi, Karney, & Bradbury, 2011). Accordingly, future work in this area should assess reasons for participating individually and level of relationship commitment (from both partners if possible) to better determine for whom the program works.

Potential Program Adaptations and Future Directions

Despite these limitations, given significant pre-post gains in several individual outcomes and change in hypothesized directions in other variables, adapting the program may be warranted. In line with the final aim, potential adaptations have been identified including better assessing participants’ reasons for participation in OR-I over OR-C and using this information to refine the program, such as having some participants invite their partners for limited involvement in the program. For example, for participants who initially did not want partners involved in treatment or for participants whose partners were initially unwilling to participate, the program could highlight the benefit of having partners complete a quick pre-treatment survey of major relationship outcomes (which partners may be more agreeable to complete). Indeed, with this information, coaches could provide more tailored feedback to participants during calls to help them better understand their core issue.

Moreover, without partner participation, identifying and implementing solutions to a core issue may be more difficult—making the program less effective. This hypothesis is consistent with qualitative responses from participants regarding their impressions about the program. Many individuals reported the program helped them better understand their relationship and motivated them to make self-change; however, many participants wanted more concrete plans for implementing their change plan (especially if their partner was too busy or refused to complete the program). Therefore, another adaptation would be giving participants the option to invite their partner to complete a few activities with condensed material from the Understand and Respond phases so partners could be more involved in making positive changes.

Couples in which the partner agrees could then have a conversation following completion of material to share their DEEP understandings and agreed-upon solutions. Then, the final call could be adapted by adding more time and having partners join so that coaches could pull for empathic joining, ensure solutions are specific and manageable, and discuss barriers to making changes. In contrast, for participants whose partners are still not interested in participating, time could be added to the final call focused on helping participants make a concrete plan for implementing change alone. For these individuals, given their partner’s unwillingness to work on improving the relationship, adding more program material on acceptance could help them better accept what is—and is not—within their control. The final call could then follow up on this material and include a discussion of the role of acceptance to increase participants’ commitment to self-change as changing one partner’s negative patterns is still better than neither partner making changes. These revisions would address concerns expressed by some individuals in the current sample. Furthermore, revisions would complement material already covered on the final call including discussing treatment gains, finalizing the letter to the partner, and making a plan for sharing the letter (e.g., when to share, how to express DEEP understanding and change plan without blaming partner, how to effectively ask a partner to make changes, etc.).

On the other hand, some participants may have chosen to complete OR-I in a final effort to decide if they want to save the relationship. Given the high percentage of divorce by follow-up, many individuals likely fell into this category. For these couples, activities focused on helping them assess whether they should end the relationship could be added to the program. Individuals who decide to separate from partners by end of treatment would then be provided resources on initiating and navigating separation (e.g., legal concerns, finances, co-parenting, etc.) and could discuss these resources with their coach during added time to the final call.

Conclusion

This study showed a few promising initial results and elucidated important information on the efficacy of individual-oriented programs. Overall, however, findings highlight how hard it is to intervene with highly relationally distressed individuals when both members of the couple are not involved. Accordingly, such distressed individuals may see more improvement from couple based interventions or individual-oriented programs that are tailored to meet their unique needs. Additional revisions to OR-I are therefore recommended to address this concern and strengthen program content. By getting partners involved in time-limited ways, offering more coach time, and tailoring the program for individuals who are deciding whether to end the relationship or whose partners refuse to participate, OR-I may intervene more effectively with distressed individuals across a variety of relationship and individual functioning domains.