Internet-Delivered Therapy

Internet delivery systems could overcome many of these face-to-face treatment barriers. Receiving treatment from the comfort and convenience of a woman’s home would reduce mobility and childcare concerns. Relatively anonymous Internet-delivered therapy may circumvent women’s concerns of stigma [7]. Further, evidence is accumulating to suggest that Internet-delivered treatments are as efficacious and more cost effective in terms of the provider’s time and resources as compared to face-to-face treatments [8].

Ample empirical evidence supports the efficacy of Internet-delivered Cognitive Behavior Therapy (ICBT) in the treatment of major depression and sub-threshold depression. Particularly with therapist-assistance, ICBT is an effective method for reducing depressive symptoms when compared to treatment as usual (TAU), control groups, and treatment from a general practitioner [9–11]. Meta-analytic data from 21 studies indicate that effects obtained by supported Internet-delivered therapy for major and sub-threshold depression is analogous to those obtained in face-to-face therapy [9]. Moreover, supported Internet-delivered interventions evidence both greater program adherence [12] and stronger effects than self-directed interventions [13], with more frequent support resulting in greater symptom change [14].

Internet-Delivered Therapy for Postpartum Depression

Preliminary research suggests that Internet-delivered treatment is efficacious for women in the postnatal period. In one study, researchers tested the efficacy of an Internet-delivered Behavioral Activation (iBA) program compared with TAU for PPD [15]. The 11-session program consisted of psychoeducation and behavioral exercises. Although the iBA program was self-directed, it included access to an Internet chat room that was moderated by parent supporters and supervised by specialist health visitors. Findings from 910 women with PPD indicated that participants in the iBA group reported greater reductions in depressive symptoms compared to the TAU group; attrition rates were high, however, with less than one-third of the participants completing the program. In an effort to improve treatment understanding and adherence rates, these same researchers recently created an 11-module iBA program and added a guided support component (i.e., telephone calls; [16]). Again, attrition rates were high. Only 5% of the women completed more than eight modules, with less than 2% completing all iBA modules. Women who reported lower perceived support, who were working or studying, and who reported lower socio-economic status completed fewer modules.

Another Internet-delivered intervention, Mom-Net, provided Therapist-Assisted ICBT (TA-ICBT) to economically disadvantaged mothers of young children [17]. Consisting of eight modules, Mom-Net included cognitive behavioral skills with adaptations relevant to mothers. Weekly phone calls were arranged between coaches and participants to assist with understanding and applying the materials. Women (n = 70) with elevated levels of depression, a child under five years of age, and attending a low-income program were randomly assigned to TA-ICBT or TAU. When compared to the TAU participants, Mom-Net participants reported improvements on measures of depression, parental satisfaction and efficacy, as well as a reduction on observational indices of harsh parenting behavior [17]. For the Mom-Net intervention, there was less attrition, with nearly two-thirds of women completing all Internet-delivered modules [17].

While preliminary research supports the efficacy of TA-ICBT for women of young children [17] and Internet-delivered behavioral intervention for PPD [15,16], researchers have not yet determined whether ICBT is an efficacious modality for treating sub-threshold and clinical PPD within the first post-natal year. Prior to implementation of TA-ICBT within community settings, it is important to ensure that the approach is more efficacious than symptom reduction over time as PPD is often a spontaneously remitting condition [18]. The purpose of the present investigation was to explore the efficacy of a specialized TA-ICBT program for women with symptoms of sub-threshold and clinical PPD. It was hypothesized that TA-ICBT participants would demonstrate greater symptom improvement than participants who received the delayed intervention on primary and secondary outcome measures (i.e., PPD, postnatal anxiety, general stress, parental stress, quality of life). These changes were expected to be clinically significant and maintained at follow-up.

Participants

Using the GPower program [19], to achieve power at 80% (alpha = 0.05), n = 23 participants per group were required to detect a medium effect size for a mixed model ANOVA. A medium effect size is consistent with previous meta-analyses of therapist-assisted ICBT [11,13].

A total of 50 participants were recruited to account for participant attrition. Participant recruitment and follow-up occurred between March, 2012 and February, 2013. Recruitment efforts included notifying physicians, perinatal nurses, and the public through the use of posters, information cards, Internet advertisements, and media interviews. Participant recruitment concluded when data collection was complete. Eligibility criteria were: (a) 18 years of age or older; (b) gave birth to an infant within the past year; (c) residing in Saskatchewan; (d) self-reported access to and comfort using a computer and the Internet; (e) score of ≥ 10 on the Edinburgh Postnatal Depression Scale (EPDS [20]); (f) consent to notify a physician of their participation; (g) not receiving other psychotherapy; (h) if taking medication, stable dose for more than a month; and (i) no past or present psychotic mental illness (schizophrenia), bipolar disorder, or current suicide plan or intent. S1 Table provides demographic details for the participants in the study. As displayed in S1 File, a total of 56 adult women responded to recruitment efforts. Of these, six participants were considered ineligible for reasons including: (1) they did not meet PPD symptom inclusion criteria, n = 2; (2) they declined to participate, n = 1; and (3) they recently had a change in their medication, n = 3.

S1 File Flow of process for research participants

Participants who were interested in taking part in the study contacted the researcher. Participants were informed of the need to conduct an initial telephone screening interview to determine eligibility for the study. Verbal informed consent for the telephone screening was obtained at this stage and documented on each participant’s file. Participants then completed the telephone screening. Following this initial verbal consent and telephone screening, eligible participants then completed an electronic informed consent form followed by the online questionnaire battery. Participants were not paid for their participation in the study. The informed consent process was approved by the research ethics boards of the institutions involved. Verbal and electronic consent were deemed sufficient for participation. Ethics approval was granted by the Universities of Regina and Saskatchewan and the Regina Qu’Appelle Health Region in March, 2012. The trial was registered with the International Standard Randomized Controlled Trials (ISRCTN: 85456371) in June 2013. There was a delay in registering the trial as we were made aware of the benefits of registering the trial after data collection had begun. No changes were made to the study protocol between the start of the trial and the time of registration. The authors confirm that all ongoing and related trials for this intervention are registered. The project was partially supported by funding provided by the Canadian Institutes of Health Research (#101526) and the Saskatchewan Health Research Foundation.

Procedure

Measures

Treatment Conditions

Statistical Analysis

All outcome data were analyzed on an intention-to-treat basis such that the analyses included every participant who was randomized according to randomized treatment assignment. Between-group intervention effects on the EPDS from baseline (screening) to T2 (post-intervention/wait-time) were compared using longitudinal mixed effects models [43]. The longitudinal mixed model uses all data on each subject, is unaffected by randomly missing data, and it can flexibly model time affects. The more familiar mixed model ANOVA (condition as the between subject factor; time as the within subject factor) is a special case of a longitudinal mixed effect model. Random intercept models were computed and a maximum likelihood method with covariance type (based on the variance components) was employed to provide the estimates. Chi-square analysis will determine if the difference between the two conditions is statistical significant.

Clinically significant change on the EDPS was calculated using procedures outlined by Matthey [44] in their study of 181 women with PPD. Matthey followed Jacobson and Truax’s [45] calculations to determine the magnitude of change on the EPDS required to be considered clinically significant. Participants whose EPDS score of more than 10 at baseline, decreased by at least four points were classified as recovered based on calculations of Matthey’s Reliable Change Index (RCI) for the EPDS. Those who demonstrated a decrease in their EPDS score of four or more points but still scored 10 or more after treatment were classified as improved but not recovered. Those participants who showed a four point increase in their EPDS score after treatment were classified as deteriorated. Finally, participants whose pre-post difference score was less than four points were classified as no change regardless of whether or not their post-treatment scores fell below the cut-off score of ten. The procedure and recommendations outlined by Matthey [44] have been utilized in other Internet therapy research (e.g., [16])

Multiple regression analyses were used to determine if group differences (TA-ICBT versus WLC) predicted secondary outcome measure scores (i.e., DASS, PSI, WHO-QOL subscales) when controlling for pre-treatment scores on each respective secondary outcome measures. The effects of missing follow-up data were explored by imputing missing data using chained equations [46, 47] that impute data for all relevant variables. All multiple regressions were analyzed twice—once with the original data set and once with the imputed data set. Unstandardized coefficients were used as a measure of effect size to assess the proportion of variance associated with or accounted for by each of the regressions. The results of the two regression analyses were compared to determine if the missing data impacted the results.

A within-subjects analysis for participants receiving TA-ICBT was conducted comparing four-week follow-up (T3) EPDS scores with post-treatment (T2) and baseline EPDS scores (screening). Finally, descriptive statistics are reported for program engagement, satisfaction, and therapeutic alliance reported by the TA-ICBT group (i.e., TSQ, TAQ, CEQ).

Preliminary Analyses

The TA-ICBT and WLC groups did not differ with respect to demographic characteristics (see S1 Table). Descriptive statistics for primary and secondary outcome measures administered at T1 and T2 are presented in S2 Table. There were no differences between completers and non-completers on the baseline EPDS: t(48) = 1.87, p = .067, d = .54. The distribution of the EPDS was normal and similar variance was revealed between the two groups. Attrition rates for TA-ICBT and WLC groups were 16% (n = 4) and 12.5% (n = 3), respectively between T1 and T2. Attrition rates for the TA-ICBT group between T2 and T3 was 28.5% (n = 5).

Treatment Received Reported by WLC Group

WLC participants reported receiving various support/care during the waiting period. Close to 58% reported regular contact with their family physician, 32% sought psychotherapy for maternal depression, and 25% of participants reported using psychotropic medication.

Primary Outcome Measure

Results from the longitudinal mixed model analysis indicated no effect for condition on EPDS scores, F(1, 48.07) = .35, p = .56, a statistically significant change in EPDS scores over time, F(1,20.99) = 16.23, p = .001, but most importantly and consistent with our primary hypothesis, a condition by time interaction, F(1, 11.82) = 5.15, p = .02. Following the criteria outlined by Matthey [44], approximately 20% of participants receiving TA-ICBT were classified as improved, while over 62% were recovered. On the other hand, while 38% of the participants allocated to WLC condition were classified as recovered, over 50% exhibited no reliable change. Chi-square analysis showed the differences between the two conditions approached statistical significance: χ²(1) = 2.93, p = .08, Cramer’s V = .026.

Secondary Outcome Measures

In order to determine whether the treatment influenced participants’ stress, anxiety, depression, parental stress, and quality of life, scores on these variables at T2 were the outcome variables in a series of multiple regressions. The predictor variables were condition (TA-ICBT or WLC) and scores on the secondary outcome variables at T2. The results for these analyses are presented in S3 Table. When compared to the WLC, the TA-ICBT at T2 (controlling for scores at T1) produced lower stress (β = -.41), lower parental stress (β = -.25) and distress (β = -.41), and better psychological health (β = .34) and environmental quality of life (β = .31). A comparison of the results from the original data set to the pooled results indicated no statistically significant differences.

Follow-Up for the TA-ICBT Condition

We conducted within-subjects analyses comparing follow-up scores (T3) with post-treatment scores (T2) for participants in the TA-ICBT condition. The EPDS collected at follow-up had statistically significant and large gains compared with post-treatment scores, t(14) = 4.13, p < .01, d = 1.10, indicating that intervention effects not only were maintained, but also that the TA-ICBT participants continued to improve.

Program Engagement and Satisfaction

S4 Table presents the descriptive statistics for program engagement. TA-ICBT participants completed on average 5.92 of the seven modules (60% of the participants completed all seven modules). Less than 18% of participants dropped out before the fourth module; the greatest attrition occurred during the 5^th of 7 modules (11.8%). The website was extensively used by participants as indicated by the mean number of program visits (M = 26.88; SD = 11.63), and the number of emails sent (M = 5.4; SD = 4.15) and received by the participants (M = 10.52; SD = 3.95). Over 80% of the participants reported that they liked the overall program and enjoyed communicating with their therapist (TSQ). Participants also reported a high level of therapeutic alliance, giving a rating of 86.42% (TSQ).

Limitations

There were a number of limitations that should be considered. The sample utilized was relatively small and prohibited an examination of other factors. The results, however, are promising as the sample included participants with sub-threshold PPD. An additional limitation was the absence of an active control condition, which may have exaggerated our results. However, the study demonstrates considerable feasibility and acceptability of TA-ICBT for PPD. The limited sample with respect to demographic characteristics was an additional drawback. Moreover, given that more than half of the sample reported some university education, additional research utilizing a more heterogenous sample is required. A group comparison at follow-up was not conducted, which would have been informative given that PPD is often a spontaneously remitting condition [18] and, therefore, may improve over time irrespective of treatment. For ethical reasons, however, we did not unduly extend the waiting period for WLC participants. Additionally, depressive symptoms were assessed using self-report measures, which are known to produce false positives. While we conducted a clinical interview (i.e., MINI) to screen participants, a post-treatment clinical interview would have provided a more objective measure of changes in depressive symptoms. Finally, these findings are specifically related to one TA-ICBT program (i.e., Maternal Depression Online) delivered by Internet therapists who were trained graduate students in clinical psychology. A recent literature review suggests that ICBT outcomes do not differ when delivered by experienced versus inexperienced therapists [51]. It is possible that the structure of TA-ICBT (e.g., modules) results in less variation in the delivery of TA-ICBT than is observed in face-to-face therapy; therefore, similar outcomes may arise regardless of therapist experience [52]. Nevertheless, it is possible that outcomes could have been more impressive had a different TA-ICBT program been used. It would be valuable to conduct further research on which type of allied health provider will obtain the best outcomes with TA-ICBT for PPD and who is best positioned to deliver TA-ICBT if delivered in community settings.