Abstract
Postpartum Depression (PPD) occurs at higher rates among impoverished mothers than the general population. Depression during pregnancy is one of the strongest predictors of developing PPD. Research indicates that non-pharmacological interventions are effective in reducing depressive symptoms but engaging and retaining low-income mothers remains a challenge. A brief, educational intervention delivered through home visits may be an ideal setting for low-income pregnant mothers at risk for PPD. Problem Solving Therapy (PST) is an effective treatment for depression. However, its effectiveness has not been tested among low-income mothers at risk of postpartum depression. This research tests the feasibility and pretest to posttest outcome of an adaptation of PST among a sample of low-income, expectant mothers. This research used a pre-post pilot study. Fourteen pregnant women at risk for postpartum depression were provided 5 home visits of motivational interviewing introduction for engagement and PST for symptom reduction. The intervention had a 93% retention rate. Analysis revealed statistically significant improvements on measures of depression symptoms after intervention This pilot study demonstrates the feasibility of implementation of a home visiting PPD intervention in a community based agency and provides the grounds for optimism about the effectiveness of a PST intervention for low-income women at risk for postpartum depression.
Postpartum depression (PPD) is the most frequent cause of maternal morbidity in the United States (Robertson, Grace, Wallington, & Stewart, 2004). It affects roughly one in eight women (Gavin et al., 2005; Wisner, Parry, & Piontek, 2002), including almost one in four low-income women (Hobfoll, Ritter, Lavin, Hulsizer, & Cameron, 1995). Community-based family service agencies are one of the largest providers of services to low-income pregnant and postpartum mothers, yet empirically supported interventions for PPD in these settings are lacking (Abrams & Curran, 2007; Sampson, Zayas, & Seifert, 2012).
Family service programs, especially those with home visiting services, may provide ideal settings for reaching low-income women at risk of PPD. Screening for perinatal (pregnancy or postpartum) depression among low-income women receiving family services has increased substantially in the past decade as agencies like Health Resources and Service Administration and Maternal Child Health Bureau require PPD screening in Healthy Start programs (U.S. Department of Health and Human Services, 2013). The current pilot study took place in a Healthy Start agency, where it sought to assess the feasibility and potential effectiveness of an intervention with previous strong research support for depression (Bell & D’Zurilla, 2009; Cuijpers, van Straten, & Warmerdam, 2007; Malouf, Thorsteinsson, & Schutte, 2007) when it is adapted to fit the circumstances of low-income pregnant women at risk for PPD. We used a federally funded problem solving intervention used by University of Washington (National Institutes of Mental Health; R34 MH083085) to design an intervention for our population. The prime ways in which the problem-solving (PST) intervention was adapted involved referring to it as problem solving “tools”, not therapy, shortening the number of sessions from 8 to 4, using social workers rather than nurses for visits and preceding the PST sessions with one motivational interviewing (MI) session aimed at fostering engagement and retention in the intervention. shortening the number of sessions from eight to four, using social workers (rather than nurses) for home visits and preceding the sessions with one motivational interviewing (MI) session aimed at fostering engagement and retention in the PST.
Depression during pregnancy, high levels of stress, and poverty are all risk factors for PPD (O’Hara, 2009). PPD is sometimes referred to as a “silent thief” (Beck, 2002) since it casts a negative shadow over the early stage of mothering, often inhibiting mother–child positive interaction and attachment. PPD is treatable and possibly preventable in some cases (Beck, 2002; Zlotnick, Johnson, Miller, Pearlstein, & Howard, 2001).
Yet, engaging and retaining low-income expectant mothers in mental health interventions is challenging (Battle & Zlotnick, 2005; Miranda et al., 2003). Most cases of PPD among low-income mothers go undetected and undertreated (Yonkers et al., 2001). Women living in poverty may avoid treatment due to stigma, cultural beliefs that dissuade mental health intervention, and attributing depressive symptoms to everyday stressors (Abrams, Dornig, & Curran, 2009). These barriers to treatment also pose a threat to self-identification of symptoms. Untreated PPD has significant public health implications because of its negative effects on mother–child attachment (Grace, Evindar, & Stewart, 2003; Martins & Gaffan, 2000; Miller, Barr & Eaton, 1993) and an increase in child behavior problems (Hay, Pawlby, Angold, Harold, & Sharp, 2003). Some interventions are not sustainable because they require mental health expertise by interventionists or are hampered by logistical issues such as transportation, child care, and cost (Miranda et al., 2003). Interventions that blend cognitive and behavioral techniques are common and effective for lowering perinatal depressive symptoms (O’Hara, 2009). Specifically, research shows that an increase in problem-solving skills, skills that are learned with the intervention of PST, has a positive benefit on lowering depression (Berry, Elliott, Grant, Edwards, & Fine, 2012).
For this project, we formed an academic/community partnership with a local Healthy Start program. The primary goal of that program is to decrease infant mortality, preterm birth, and low birth weight among their clients. In addition, caseworkers are required to screen for PPD at enrollment and refer to external counseling. Our research team has been in collaboration with the program staff since 2011 when we conducted focus groups to explore cultural beliefs of PPD among their clients. Results from the focus groups revealed that high levels of stigma of PPD pervaded the communities and likely prevented mothers from admitting symptoms or seeking treatment (Sampson, November 2014). Most of the agency’s clients are African American and Latina with household incomes less than $20,000 per year. According to agency staff, 30% of mothers in the program screen positive for PPD, yet less than one third have followed through on their treatment referrals for community-based counseling. A home-visiting approach therefore was developed to enhance treatment engagement and retention.
Method
The primary aim of this study was to assess whether a PST intervention adapted for low-income women at risk of PPD (1) appears to be a feasible way to engage and retain this target population in the treatment and research protocol and (2) shows promise regarding reducing depressive symptoms. Mothers were considered “at risk” for developing PPD because of their scores on a depression measure during pregnancy. Prenatal depression is one of the strongest predictors of PPD (Robertson et al., 2004). A secondary aim was to explore the utility of preceding the PST intervention with a session of MI.
We used Rounsaville, Carroll, and Onken’s (2001) stage model for behavioral health intervention research. The researchers present a method for progressive development of interventions. Stage 1 involves intervention development and initial evaluation. We followed suggestions for doing a Stage 1b study. Referred to as a “pilot trial,” this stage involves testing a known treatment on a new population for the purpose of refining the intervention, developing of an appropriate training of the intervention, and tailor making a treatment manual. Because this is a home-visiting intervention delivered by community agency workers, we were concerned with the feasibility of acceptance of the treatment by workers and participants and feasibility of implementation (Rounsaville, Carroll, & Onken, 2001).
The study was funded by University of Houston, Graduate College of Social Work small grant mechanism and Postpartum Support International. We also wish to acknowledge Neighborhood Centers Inc. Healthy Start for its collaboration.
Sample
Because the aim of the Healthy Start agency is physiological (to decrease infant mortality, preterm birth, and low birth weight) and not focused on mental health, our first step was to identify those agency clients who had at least some identifiable risk of PPD. Caseworkers verbally administer the Edinburgh Postnatal Depression Scale (EPDS), which assesses risk of depression, during enrollment to the Healthy Start program (Cox, Holden, & Sagovsky, 1987). Agency guidelines stipulated a cutoff score of 9 to indicate at least some risk of depression so our inclusion criteria required a score of 9 or higher. Higher scores indicate a higher risk of depression and can reach a possible maximum of 30. Additional study inclusion criteria were 18 years or older, not currently receiving mental health treatment, and at least 12 weeks pregnant. English-speaking knowledge was not required. Nonpregnant clients were excluded as were—per agency policy—pregnant women who had answered affirmatively when asked about the risk of homicide or suicide. Participants were given US$25 after intake and US$25 at completion of all sessions, so a total of US$50 as compensation. Because caseworkers were responsible for delivering the intervention, they were compensated with a Kindle Fire for completing at least four participants each. Fourteen women initially participated in the intervention, but one dropped out of the study after three sessions due to loss of pregnancy.
The participants were all African American with a mean age of 24 (Table 1). Most were unmarried (85%), had other children besides this pregnancy (61%), and reported a high school degree or general equivalency diploma (46%). All participants were unemployed at time of study and the mean monthly income for the sample was US$1,153. Caseworkers conduct written and verbal assessments during the intake session. The agency uses a “risk assessment form” that has a list of “low-”(such as environmental surroundings), “medium-,” and “high-risk” (domestic violence, substance abuse, and stress) items. According to the data taken from this risk assessment form (see Table 2), 69% of the sample were considered high risk for infant mortality. High-risk category is a result of the woman reporting one or more of the high-risk factors. According to the risk assessment data, the sample of women report little to no use of alcohol, drugs, or tobacco, but 23% experienced domestic violence in the year prior to intake and 69% report experiences of chronic stress or anxiety. Although not all of these factors are known predictors of PPD, the agency risk assessment data provide broader context of the sample. We did not conduct inferential statistics on the risk assessment data, since there was no scholarly or psychometric data available for the form and the inherent measures.
Sample Demographic Characteristics.
Note. GED = general equivalency diploma. De-identified data to match case numbers was provided by the agency.
Data From Agency-Based Risk Assessment.
Recruitment and Screening
Agency caseworkers recruited participants during their usual intake process. A typical intake session requires assessment of need for resources and risk of infant mortality. Included in their intake materials is the aforementioned risk assessment form that asks about domestic violence, substance abuse, smoking, and so on. Intake also requires a routine screen of perinatal-related depression using the EPDS. If a pregnant woman scored 9 or higher on the EPDS and met other eligibility criteria, the caseworker asked whether she would like to take part in an educational program to learn about depression. If the woman agreed to be in the study, she completed consent. If she declined to be in the study, she received treatment as usual, which is referral to an outside free counseling source. No advertising was made per request of agency staff. Agency staff advised the research team to let caseworkers do all promoting and recruitment for study to establish client trust. The clients at the agency are skeptical of outsiders, especially researchers from the university. The researchers deferred to the agency staff as experts on their own clients and did not interfere with recruitment among potential participants.
All caseworkers received 1.5-day training on using MI for assessing readiness for treatment and PST protocol. The lead investigator of this study conducted MI training and an external consultant (a research assistant on funded PST intervention) trained staff on PST. In addition, our research team trained them on how to invite eligible mothers to participate in the study. A consent “cheat sheet” was distributed to each caseworker so she could cover the risks and benefits of the study in a conversational tone while the participant read the document.
Procedures
Depression outcomes measured with the EPDS (Cox et al., 1987) were assessed at baseline (T1), immediate post (T2), and follow-up which was approximately 3 months after intervention (T3). Depression severity was measured more often with the PHQ-9 (Kroenke, Spitzer, & Williams, 2001) at the beginning of each problem-solving session. The feasibility of study implementation by staff was tracked throughout the study by measuring how many sessions each caseworker completed with each participant and how many homework exercises were completed by the participant. Satisfaction of participants with the intervention was tracked through exit surveys completed in the last session. A debriefing session among caseworkers was conducted after all intervention sessions with participants were complete. All participants completed a written exit interview and those who agreed to follow up were called and asked for their feedback on the intervention. All procedures were approved by the Committee for Protection of Human Subjects at the University of Houston.
Intervention
Three caseworkers from the Healthy Start program administered the home-visiting intervention, including both the first session of MI and the four sessions of PST. Two of the three caseworkers held an associate’s degree and one certified as licensed professional counselor. We adapted a federally funded PST intervention with a Nurses for Newborn program (NIMH, 2008). Main outcome results from that study are forthcoming. The original study utilized eight PST sessions. For our study, we shortened the number of sessions and added an MI component.
PST is a cognitive behavioral intervention that helps increase adaptive social problem-solving skills. It relies on the belief that effective problem solving will help to buffer the effects of stress on mental well-being and thereby can lower depression (Bell & D’Zurilla, 2009). The intervention focuses on problem orientation and problem-solving style. Developers of the intervention advise 8–16 sessions lasting 1–2 hr in duration. The main components of the PST intervention are identification and definition of a problem, understanding the problem, setting goals, generating a list of solutions, choosing a feasible solution, and trying the solution at home. Intervention components are delivered through guided sessions to work through the steps, homework, and worksheets. The interventionist and participant discuss efficacy of the attempted problem solving each week (D’Zurilla & Nezu, 1999). All PST protocol materials (scripts, homework sheets, and handouts) were borrowed with permission from the NIMH (2008). Based on the feedback from the Johnson-Reid study staff, the intervention for the Healthy Start project was shortened to four sessions from eight sessions because decrease in depressive symptoms was seen by second session through fourth. Because it is possible to complete all homework and in-session activities in four sessions, we were able to shorten the number of sessions without missing any intervention components or the original design.
PST sessions were preceded with the caseworker using an “MI primer script” to help increase awareness of depressive symptoms and the prevalence and treatability of PPD. MI has been used successfully with African Americans to increase treatment engagement and reduce stigma (Breland, Bell, & Nicolas, 2006). It can promote a sense of collaboration and respect for the participant’s perspective on their depression. The MI session may also help with the “problem orientation” that is vital to effective PST (Malouf et al., 2007). The script was created by the lead investigator (first author) of this study who has experience in the MI approach and was able to consult with expert trainers who are members of Motivational Interviewing Network of Therapists. Exercises to assess motivation to accept treatment were also conducted in this primer. Motivation to accept treatment was not measured for this pilot study.
PST is a cognitive behavioral intervention that does not require a mental health degree to deliver; lay people can learn and deliver this intervention in a standardized and measureable manner. The research team chose PST for the agency because it had various levels of education ranging from an associate’s degree to a graduate degree in psychology. In addition, both PST and MI are effective among low-income, low-education populations (Berry et al., 2012; Ruger, Weinstein, Hammond, Kearney, & Emmons, 2008).
Each PST session included a progress update, topic discussion, and goal setting. Participants were asked to identify problems, brainstorm possible solutions, set goals, and come up with a specific action plan, to be carried out as “homework,” on how to solve the initially identified problems. Examples of identified problems were “stop arguing so much with boyfriend,” “clean out cluttered closet,” “ask boyfriend to help with baby more often so I can sleep.” Caseworkers reviewed accomplishment of these tasks at the beginning of the next session. Additionally, participants were given a copy of different strategies that they could implement when feeling depressed between sessions. Caseworkers were given a copy of the “protocol flow” to keep (Supplementary Appendix).
Outcome Measures
Two depression-related instruments were used to assess outcome. One was the EPDS, which as mentioned earlier was also used to screen for depression risk. It has been widely used and has very good internal consistency and validity for assessing risk of neonatal and postnatal depression (Cox et al., 1987). It was first administered at the beginning of the preliminary, MI session and then later administered at the end of the final PST session. The other outcome measure was the PHQ-9, which measures the severity of depressive symptoms (Kroenke et al., 2001). It was administered at the start of each PST session.
Analysis
Fourteen participants enrolled in the study, and 13 completed all sessions. Paired sample t-tests were conducted to assess any changes in the pre- and posttest scores on the two outcome measures among intervention completers (N = 13). Although there was no comparison group, we sought to measure practical effect size. Within group effect sizes were also calculated by dividing the difference between pre- and postscores using the Cohen’s d method (Cohen, 1988).
Results
For the EDPS measure, there was a significant reduction in scores, t(12) = 3.49, p < .05, from a mean of 13.36 at pretest to a mean of 7.69 at posttest. Dividing that mean difference of 5.67 by the pooled standard deviation of 5.51, the within-group effect size was 1.03. For the PHQ-9, there was a significant reduction in depressive symptoms, t(12) = 4.54, p < .05, from a mean of 10.85 at pretest to a mean of 5.23 at posttest. Dividing that mean difference of 5.62 by the pooled standard deviation of 4.55, the within-group effect size was 1.24.
Conclusions
As indicated by the 93% program retention rate in the intervention, this study demonstrated the feasibility of adapting PST for low-income, pregnant women at risk for PPD via a home visiting–based approach. The feasibility was also supported by a 100% completion of homework by the participants. A decrease in depression scores from pretest to posttest indicates promise for effectiveness of the intervention. Given that all measures of depression were self-report and provided to a caseworker that the participant is already working with, there may have been some social desirability bias. This may be a common dilemma in community-based research where skepticism to outsiders is a barrier to data collection. By using caseworkers who were familiar (and typically culturally matched) to the participant, skeptical participants may have been more willing than usual to participate in a research study. In this case and possibly other similar sites, limited funding prevents the hire of additional program staff. Another limitation in our findings is that we were not able to measure treatment fidelity by caseworker with our planned approach of listening to recorded sessions. During the staff training, the staff declined to tape-record saying that oftentimes participants are too “paranoid” to allow recording. This may be a recurring concern in other community-based research with highly stigmatized mental disorders. Thus, our research team adapted to the needs of the program by “hiring from within” and allowing the research to proceed without tape-recorded sessions.
Our study also provided outcome data suggesting the possibility that even with only four PST sessions (preceded by one MI session) of the adapted approach, meaningful improvement in symptomatology can be observed and with a potentially meaningful within-group effect size. The similarity in the two effect sizes (1.24 on the PHQ-9 for the reduction in depressive symptoms after four sessions of PST compared to 1.03 on the EDPS for the reduction in depression risk with the MI session added before the 4 PST sessions) suggests that the MI session did not add to the improvement in the level of depression. However, that was not the purpose of the MI session, which was only meant to foster engagement and retention in the PST sessions. Although our data do not address the impact of the MI on engagement and retention, the fact that the only participant who dropped out did so due to the loss of her pregnancy may encourage others seeking to implement our adapted PST approach to include the MI session.
As with most pilot studies, a limitation in ours was its lack of a control group and our consequent inability to derive causal inferences. However, that limitation is somewhat alleviated by the fact that we were not piloting a completely new intervention approach. Instead, we were piloting an adaptation of an approach that already has strong research support for its efficacy in reducing depression, as shown in three meta-analyses (Bell & D’Zurilla, 2009; Cuijpers et al., 2007; Malouf et al., 2007; Figure 1).
Severity of depression symptomatology as indicated by mean PHQ-9 scores at four time points. The four time points refer to the start of the first three sessions of PST and the end of the fourth session.
Lessons Learned
Several challenges were encountered during the recruitment phase. Although we had prepared all intervention materials in Spanish, the agency lost their bilingual caseworker, after we had trained her in protocol, and did not replace her position. This staff changes resulted in one less caseworker to recruit and inability to deliver the intervention in Spanish. We offered to have a research assistant recruit Spanish-speaking participants, but this offer was declined citing the hesitancy of their clients to “outsiders” of the agency. Another challenge was the extra burden caseworkers took on by recruiting participants and providing the PST. The four to five home visits exceeded the typical one visit the caseworker would typically conduct. We addressed these challenges by keeping open lines of communication. The PI came up with a slogan to use internally: Follow the 3Cs for success, namely cheerleading, consistency, and communication. The cheerleading referred to the research team providing positive encouragement to the caseworker supervisor to help her champion the project. We learned that having an internal champion helped the pace of recruitment. The graduate-level research assistant made weekly in person visits to the agency. She would check on the caseworkers and offer any help they needed. This seemed to keep the momentum going. The close partnership that we formed with the agency also seemed to help. Ejiogu et al. (2011) conducted a large population-based study of health disparities among poor African Americans and Whites and concluded that strategies such as developing a hypothesis that is relevant to the studied community and articulation of that hypothesis engenders more participation. Our research team worked closely with the agency staff to develop research questions and design recruitment and follow-up methods that were relevant and accessible.
The PI and research assistant met with the three caseworkers at the end of the study to elicit feedback. Caseworkers reported that it was a challenge to take on increased home visit sessions without being able to drop other cases. The needs of the agency did not allow for the caseworker staff to prioritize study enrollment. Caseworkers reported satisfaction from being able to help the women develop skills that could help them deal with their problems and subsequently their depression. The caseworkers shared that the MI + PST intervention was a different approach than typical in the work setting. Specifically, one caseworker said: They [clients] are used to depending on other resources for help so they forget what they learned…with our clients its generational…they want to break the cycle but don’t know how.
Implications
Mothers who live in poverty face a multitude of stressors, such as limited economic opportunities, difficulty finding suitable housing, and an increased experience of negative life events (Zayas, Jankowski, & McKee, 2003). Living in poverty and experiencing multiple stressors and logistical barriers to treatment contribute to disproportionate levels of PPD among low-income mothers (Hobfoll et al., 1995). The stigma associated with PPD decreases the likelihood of seeking treatment for it, and poor women of color are less likely to seek treatment than are other poor women (D'Angelo et al., 2007). Practitioners urge the use of community interventions since poor mothers often seek local services (Costello, Osrin, & Manandhar, 2004). Our findings offer a way to make a research-supported intervention feasible and potentially effective for this target population.
Footnotes
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The study was funded by University of Houston, Graduate College of Social Work small grant mechanism and Postpartum Support International. We also wish to acknowledge Neighborhood Centers Inc. Healthy Start for its collaboration.
