Randomized Controlled Trial of an Internet-Based Educational Intervention for Mothers With Mental Illnesses: An 18-Month Follow-Up

A fuller description of methods and measures can be found in a previous publication (10). The study employed an RCT in which participants were randomly assigned to either an experimental or active control group. The experimental condition consisted of 12 online parenting education sessions that lasted for a period of 3 months and focused on child development, stress reduction, mental illness and parenting, and positive parenting. Parents were instructed to navigate the course (written documents) based on their child’s age and view one session weekly at a time that best accommodated their schedules. From study entrance to 12 months, a peer support–moderated listserv was available. The listserv was moderated by a provider experienced at working with the target population and a mother with lived experience, who answered questions and provided feedback on the educational course. The participants also generated questions to reinforce course content in application to their parenting issues and concerns. In addition, between 3 and 6 months, participants were asked to view weekly prerecorded video conversations with experts as booster sessions to reinforce the educational material from the first 3 months of the intervention. Engagement with the intervention was assessed by how often individuals read messages, wrote messages, and received responses to their messages. Derived from the principles of cognitive-behavioral therapy, the experimental intervention centered on stress inoculation, acquisition of relevant skills, and completion of targeted homework assignments and created an interactive model of education and information sharing. An active control condition was used in which participants were reminded to access healthy-lifestyle educational factsheets on a monthly basis over a period of 12 months. The factsheets offered information and resources about maintaining one’s health.

Participants were recruited between 2011 and 2013 through mental health Web sites, news lists, and listservs. To be eligible, participants had to be mothers with a diagnosis of a schizophrenia spectrum disorder or a mood disorder (e.g., bipolar disorder, depressive disorder). Interested participants were provided with a release-of-information form for a practitioner to confirm their diagnosis. Eligible participants were over the age of 18, had primary or shared custody of a child under the age of 18, had access to the Internet, lived in the United States, and were fluent in spoken and written English. All participants in both conditions were able to engage in any mental health treatment/services throughout the course of the study.

Of the 131 participants enrolled, 66 were assigned to the experimental and 65 to the control group. Participants’ mean±SD age was 37±7.69. The racial and ethnic composition of the sample was 84% (N=106) white, 20% (N=22) black, 7% (N=7) Hispanic, 10% (N=10) Native American, 3% (N=3) Asian, and 2% (N=2) other; participants were able to indicate more than one racial or ethnic category. Of the participants, 74% (N=89) were married or in a domestic partnership, 89% (N=116) had more than a high school education, 53% (N=63) were employed, 21% (N=27) were enrolled in an educational institution, and 45% (N=59) had a disability as indicated by receipt of Supplemental Security Income (N=22, 17%) and Social Security Disability Insurance (N=41, 31%). The mean monthly income was $1,853.29±$1,162.13. Eleven percent (N=14) lived alone in their own home or apartment, 77% (N=101) shared a residence with family, and 36% (N=47) reported being homeless at some point in the past. Fifty-two percent (N=66) had a diagnosis of a bipolar disorder, 40% (N=51) of a depressive disorder, and 8% (N=10) of a schizoaffective disorder, and 35% (N=46) had a substance use history. The average number of children was two, with a range of one to five. Ages of children ranged from infants to early twenties. Eighty-one percent had primary or shared custody of all children. Fifteen percent did not have custody of at least one child, and 4% had had their parental rights terminated for at least one child.

Demographic information and clinical characteristics collected from all participants at baseline and follow-up assessments were evaluated for any changes. Data on outcomes (enhanced parenting skills, improved parental coping skills, and decreased parental stress) were collected at baseline and at 3, 12, and 18 months postbaseline. Outcomes were assessed by using the Healthy Families Parenting Inventory (HFPI; 11) subscales as well as other measures: parenting skills were measured by the HFPI–Parent-Child Interaction and the HFPI–Home Environment, parental coping skills by the Family Coping Inventory (12) and the HFPI–Mobilizing Resources and Personal Care, and parental stress by the HFPI–Role Satisfaction Scale and the Parenting Sense of Competence Scale (PSCS) satisfaction factor (13).

Using an intent-to-treat approach, differences between the experimental and control groups on measures of parenting skills, coping, and stress were assessed over time by using a repeated-measures analysis of variance. This approach allowed for an examination of both conditions based on random assignment without consideration of the level of engagement with the intervention. Analyses were conducted by using PROC MIXED in SAS. The study was approved by the institutional review board from the investigators’ university.

There were no significant differences between participants in the experimental and control groups on any demographic and clinical characteristics and outcomes at baseline, and degree of engagement (dosage effect) was found to have no effect on outcomes. With regard to the central hypotheses, results revealed that the intervention was effective in decreasing parental stress over time. For parenting stress, the interaction between treatment condition and time on results of the PSCS satisfaction measure was significant (F=3.46, df=257, p=0.017). Specifically, as shown in Figure 1, the conditions differed at 18 months, with the experimental group having greater satisfaction scores (d=4.80, SE=1.67, df=257, t=2.88, p=0.004). The interaction between the treatment condition and time was not significant for the second measure of parental stress, HFPI-role satisfaction. No statistically significant interactions were found between treatment condition and time for the measures of parenting skills (HFPI–Parent-Child Interaction and HFPI–Home Environment) or for measures of parental coping skills (Family Coping Inventory and HFPI–Mobilizing Resources and Personal Care

The purpose of these analyses was to examine the effectiveness of an Internet-based educational intervention for improving longer term parenting outcomes among mothers with serious mental illnesses. The results provided partial support for the hypotheses that engagement in the intervention would yield positive outcomes for parental stress, skills, and coping compared with the control group during an 18-month postbaseline follow-up. Research has shown that mothers with serious mental illnesses experience higher levels of parental stress (7). Furthermore, research has shown that parenting stress exacerbates child behavioral problems (14). Stress-reduction techniques were a component of the intervention and were incorporated into several lessons. These stress-reduction modules focused on managing physical and emotional health and wellness while parenting and described techniques specific to parenting. These modules were designed specifically to address the intersection of parental stress with mental health or illness, which is largely overlooked in general parenting programs. Additionally, participation in the peer support listserv provided participants with the opportunity to receive feedback and support related to their own parenting practices and to reinforce the learning that occurred during the educational sessions. Results of this study indicate that these aspects of the intervention were effective in filling an important gap in currently available services. However, it is unclear what extent peer support helped to reinforce the educational component and contributed to the positive outcome.

This study had several components of innovation and strength. The intervention was delivered via the Internet, which targets several barriers known to be present in face-to-face programs, such as issues surrounding transportation, programmatic availability, scheduling conflicts, and a preference for anonymity. Moreover, because social workers and other service providers are often faced with numerous demands that make it difficult to prioritize the development of parenting skills (e.g., simultaneously managing many cases, creating and implementing treatment plans, etc.), this intervention offers an additional avenue for support that may not otherwise be available to this population.

The results of this study are limited with regards to generalizability, given that the sample had higher levels of education and employment than are generally reported in research on populations of mothers with serious mental illnesses. Additionally, the study was limited by a dearth of measures designed to assess parenting outcomes specifically constructed for mothers with serious mental illnesses. Given the unique needs and barriers of this population, further development and validation of such measures are warranted.

Though only parental stress was found to improve over time as a result of participation in the intervention, this intervention, tailored to the unique needs of mothers with mental illness, provides a valuable and cost-effective resource to mental health providers interested in supporting longer-term stress reduction for this population. Given that these mothers are more likely to have contact with child protective services and are at an increased risk of losing custody of their children, interventions of this nature are important. Further research is needed to identify ways to improve other targeted outcomes related to parenting and to identify ways to modify the intervention to maximize its effectiveness. These results suggest that longer term effects of an educational intervention on stress reduction may improve over time, possibly because of additional time spent practicing the skills associated with the learning that took place during the intervention. Research is needed to further test this effect. It is possible that the slight improvement of the active control group during the 12-month follow-up may be due to the Hawthorne effect, wherein individuals modify their behavior when they know the behavior is under observation, because they were reminded monthly to complete one healthy lifestyle lesson and the effect dropped precipitously after reminders terminated. Alternatively, it is possible that continued access among individuals in the control group to the educational materials over the 12-month period led to modest, positive effects at 12 months that were not sustained at the 18-month follow-up. Finally, additional analyses should focus on potential moderating factors on intervention outcomes, such as children’s’ ages or a mother’s diagnosis.

The results of this analyses provide support for the longer term effectiveness of an Internet-based educational intervention for mothers with serious mental illnesses. Despite study limitations, the results reveal that this is an important area for further investigation. There is a known gap in services available to mothers with serious mental illness—a unique population with specific needs, barriers, and experiences. Although the results of this follow-up study are modest, they offer a foundation for future research and programming designed to support mothers with serious mental illnesses.