A Pilot Trial of a Health Promotion and Illness Prevention Paradigm in the Perinatal Period

Abstract

Objectives

This is a pilot study of the Vermont Family Based Approach, an innovative health promotion program designed to address behavioral health prevention in primary care, adapted for perinatal women. We aimed to determine the acceptability of the intervention across socioeconomic strata, and to identify if participation improves perinatal mental health.

Methods

Recruitment occurred at a general obstetrics practice. Women 12–25 weeks gestation were paired with a wellness coach who administered a wellness assessment and used motivational interviewing to facilitate individualized plans based on evidence-based domains of health promotion. Participants were offered access to free, co-located wellness activities through the peripartum, and referred to behavioral health services if appropriate.

Results

93 women consented; 16 Medicaid Insured women (MI) and 30 Privately Insured (PI) were randomized to the intervention. Of all activities, yoga and parenting activities were most appealing, with 58% of women attending. PI (M = 12.30, SD = 11.71) attended significantly more activities than MI (M = 3.81, SD = 12.30; p = .001). Trauma exposure was inversely associated with attendance (p = .004). Randomization to the intervention was not associated with improvements in internalizing symptoms or perceived stress at 12 months postpartum, however, attending three or more wellness activities was associated with a decrease in perceived stress between baseline and 12 months postpartum.

Conclusion

This program appeared acceptable and engaging to women with private insurance, but less so with Medicaid. The trial failed to demonstrate improvement in internalizing scores, but of those randomized to the intervention, higher engagement was associated with decreased stress one year after giving birth.

Significance

Perinatal mental illness is recognized as a serious issue for women with long-term repercussions on offspring. Screening and detection are improving but treatment options remain limited. As with any illness, prevention is ideal, yet there has been little research in this area. This is the first RCT to utilize health coaching with individualized goalsetting and access to free and co-located wellness activities throughout the peripartum. This study describes the feasibility, acceptability and impact of the approach offered in a university obstetrics clinic to women with private insurance and women insured by Medicaid.

Objectives

Perinatal mood and anxiety disorders (PMADs) are prevalent, under-recognized, and under-treated. In addition to the suffering this causes child-bearing women and their families, a 2017 estimate of the cost of PMADS in the US was 14.2 billion dollars. About 60% of the costs were maternally attributed, with loss of productivity and higher obstetric-related healthcare costs, while the remainder were associated with child outcomes, such as preterm birth, behavioral, and developmental disorders. (Luca et al., 2019). Children of women with PMADS are at increased risk for internalizing disorders (Santos et al., 2014), behavioral problems (Stein et al., 2014), impaired socioemotional development (Junge et al., 2017), lower academic achievement (van der Waerden et al., 2017), and even compromised growth (Surkan et al., 2008).

There have been many initiatives to improve screening and early intervention, but nothing is more cost effective than illness prevention. A recent systematic review of prevention studies found a small to medium effect associated with programs designed to decrease the risk of developing PMADS (O’Connor et al., 2019). The programs reviewed varied in scope, including psychotherapeutic interventions and medications. A few studies focused on lifestyle-based health promotion, including yoga and exercise interventions, which demonstrated trend-level benefit, but were limited in scope and generalizability.

Up to this point, most wellness-based interventions are prescriptive in nature, based on a single type of intervention and a set number of classes. These programs may have limited ecologic validity and sustainability. Research suggests that motivational interviewing (Rubak et al., 2005) and collaborative goal setting (Epton et al., 2017) are highly effective in health behavior change. We are not aware of any perinatal prevention studies that utilize motivational interviewing or involve individually tailored prevention strategies which may lead to habit change sustained beyond the duration of the intervention. Another limitation associated with many existing prevention programs relates to generalizability, many efficacious interventions report samples with disproportionately high SES (Davis et al., 2015; Muzik et al., 2012; Woolhouse et al., 2014). We aim to include women on Medicaid insurance to specifically understand effectiveness in a more representative, universal sample who engage in obstetrics care.

The Vermont Family Based Approach (VFBA) is an illness prevention program that originated in pediatric care (Hudziak & Ivanova, 2016), and is directly applicable to pregnant women. The VFBA is a population-based, universal intervention, which focuses on emotional and behavioral health as the key to overall health with an emphasis on wellness promotion. The approach involves a multi-disciplinary team and is tailored to an individual’s risk level. All participants are offered a “wellness coach” and co-located wellness activities, while individuals found to be at higher risk are referred for psychotherapy and/or medication management with a psychiatrist as needed. The paradigm includes an initial assessment component, triage to higher level of care when appropriate, and then incorporates motivational interviewing to facilitate all women to set goals across various wellness domains, each linked to evidence for improving mental health, decreasing stress, and in some cases targeting improved bonding with offspring, which is a potential mitigator of the effects of PMADs on offspring. (Reck et al., 2016). The evidence-based domains include prenatal nutrition, (Navarro et al., 2017), exercise (Bø et al., 2016), yoga (Gong et al., 2015), mindfulness (Perez-Blasco et al., 2013), parenting competency (Rosenblum et al., 2018) and social support (Ohara et al., 2017).

Results of the first VFBA randomized control trial in pediatrics have established improved emotional and behavioral outcomes for children and improved health-related quality of life for parents (Ivanova et al., 2019). Given that the perinatal period is a time of inherently increased motivation, (Cooper et al., 2017) but also of vulnerability, we proposed a trial to test the VFBA in women through pregnancy and the postpartum period. The purpose of the paper is twofold. First, we describe how women interacted with the intervention, which includes an examination of recruitment into the study, and then engagement in the study activities, with consideration of insurance status and other participant characteristics associated with participation. The second focus is to characterize the effect of participation in the VFBA on internalizing symptoms and perceived stress at one year postpartum, which marks the end of the intervention.

Methods

Recruitment

This research was conducted in accord with prevailing ethical principles and reviewed by an Institutional Review Board at the University of Vermont. Women between 12 and 25 weeks gestation were eligible and recruited between December, 2016 and May, 2018 from two University of (Blinded) obstetrics clinics. They were excluded if they were under 18 or unable to fill out questionnaires or provide informed consent due to language barrier or intellectual disability. Women were approached by their provider or clinic staff with a flyer explaining the research and asking for contact information if interested in participating. Research staff called all interested women to set up an initial visit. Posters were also distributed around the clinics encouraging women to call for more information, at which time the initial visit could also be scheduled. Women were offered a total of $90 in compensation for completing questionnaires at baseline and postpartum.

Study Design

Participants were randomized to receive the VFBA-based wellness promotion and illness prevention paradigm, or to the control group, who filled out questionnaires and otherwise received prenatal care as usual. To test a range of socioeconomic backgrounds, our recruitment targeted equal numbers of women with private insurance and Medicaid. Due to uneven recruitment rates in the first six months, women on Medicaid were randomized to the intervention at a higher rate (See Fig. 1). This protocol was approved by the University of Vermont Institutional Review Board.

Fig. 1.

Flow of women through recruitment and study engagement

Assessments

Study data were collected and managed using REDCap (Research Electronic Data Capture), a secure, web-based application. To measure mental health at baseline and follow up, we chose the Adult Self Report (ASR), a comprehensive mental health questionnaire, because it includes a series of problem-based subscales empirically derived through factor analysis, such as “attention problems”, or “somatic complaints”. Each subscale offered valuable information to provide women feedback in the initial session. The “internalizing” subscale, which encompasses both anxiety and depressive problems, was chosen as a targeted outcome, since both symptoms are elevated in the perinatal period. This scale has high internal consistency (α = 0.92) and test-retest reliability (r = 0.89), (Achenbach & Rescorla, 2003). The other outcome measure, administered at baseline and 12 months postpartum, was the Perceived Stress Scale (PSS) (Cohen et al., 1983) is a valid and reliable 10-item scale chosen as the most widely used instrument to measure perception of stress. Additional scales used at baseline only were chosen to measure various domains of wellness, for the purpose of providing women with feedback on their current wellness. These scales included the Substance Use Risk Profile-Pregnancy, (Yonkers, et al., 2010), Evaluation of Physical Exercise During Pregnancy (Nascimento et al., 2015), a list of Adverse Childhood Events, the Dyadic Adjustment Scale (a measure of partner satisfaction), and a health behavior and nutrition questionnaire developed specifically for this study (Can be made available upon request).

Engagement in the study was measured via number of activities attended. Women who attended three or more activities were considered “participators”, as opposed to “non-participators” who attended three or fewer.

Illness Prevention Paradigm

Goal-Setting Session

Women randomized to the intervention were scheduled for an initial meeting with the Wellness Coach (WC). WCs are master’s-level healthcare professionals who had received training in the theoretical foundations of behavior change, Motivational Interviewing, and evidence-based interventions. They directly coach women vis-à-vis the pillars of wellness. Prior to this meeting, the research team analyzed the results of each participant’s questionnaires underscoring areas of strengths and challenges across wellness domains. On each wellness questionnaire, a challenge threshold was determined and if women scored beyond that threshold this domain was identified as an area with room for improvement. For example, women would be considered to have a challenge in the exercise domain if they endorsed less than 150 min of any type of exercise. Table 1 lists the challenge thresholds and the number of women with challenges in each domain.

Table 1.

Provider Identified Challenges at Baseline by Wellness Domain (n = 52)

Wellness domain	Challenge threshold	Total (n = 53) % (n)	PI (n = 30) % (n)	MI (n = 22) % (n)
Exercise	0–150 min per week	65.4% (34)	63.3% (19)	68.2% (15)
Sleep	Fairly or Very Bad Quality	30.8% (16)	16.7% (5)	50% (11)
Stress	> 18 On PSS	31.4% (16)	16.7% (5)	52.4% (11)
Partner Support	“None” or “Unhappy” with help raising child	19.2% (10)	10% (3)	31.8% (7)
Trauma	4 or more ACES	32.7% (17)	16.7% (5)	54.5% (12)
Drug Use	Any drug use in last month	5.8% (3)	3.3% (1)	9.1% (2)
Mental Health	Clinically Elevated (T > 65) Internalizing Problems	3.8% (2)	0% (0)	9.1% (2)
Nutrition	Excessive fats, excessive sweet/caffeinated beverages or low water intake	19.2% (10)	6.7% (2)	36.4% (8)

During this meeting, the WC presented these areas of strengths and challenges. In cases where women demonstrated clinical elevations on the ASR scales, the team psychologist joined the meeting to introduce herself as a resource for psychotherapy. The WC used a motivational interviewing approach to help women set goals for their wellness based on the feedback they received, and the activities offered by the program.

Wellness Activities

All activities were free and co-located at the hospital. Prenatal offerings included individual meetings with the WC, private and group yoga sessions, nutrition groups, parenting seminars, mindfulness classes, community dinners and prenatal fitness. The WC was certified in prenatal yoga and Mindfulness Based Stress Reduction and taught all yoga and mindfulness classes. Mother-baby yoga, postnatal fitness, and a “Mom Power” parenting group (Rosenblum et al., 2018) were offered postnatally. All women were made aware that psychology and psychiatry services were available, and the WC would coordinate referrals.

Contacts

WC contacted women weekly by their method of choice, i.e., phone, text, email to advertise new activities, and to check in regarding women’s self-identified goals. In-person meetings were also offered, though women largely did not schedule recurring individual meetings.

Statistical Plan

We conducted t-tests and chi squares to compare women across insurance groups, and correlations, regressions and chi-squares to examine baseline characteristics associated with engagement. To examine outcomes we used General Linear Models (GLM) repeated measures to determine the effects of randomization on 4 and 12 month ASR internalizing and PSS scores, with respect to baseline. Then, follow up GLM analyses were repeated among women randomized to the intervention only, by engager status, to test whether for a “dose effect” of engagement on internalizing or stress scores at follow up points. A threshold for engagement status was set at three activities, based on the fact that almost half of the sample (42%) attended 2 or less. Medicaid status was controlled for in all of the GLMs, to account for the disproportionate number of MI women in the intervention group due to oversampling.

Results

Recruitment

Over the span of 18 months, the clinic throughput was estimated as 825 women who were Medicaid Insured (MI) and 1140 who were Privately Insured (PI). The enrollment target was 60 women for both MI and PI. We received interest (as indicated by a returned flyer) from 54 MI (6.5%) and 104 PI (9%). Of the women interested, 33 (4%) MI, and 60 (5.3%) PI women went on to provide consent (see Fig. 1). The most common reason interested women did not consent was a failure to schedule before 25 weeks gestation.

Illness Prevention Paradigm

Fifty-two women were randomized to the intervention group (22 MI, 30 PI). If a woman randomized to the intervention did not respond to attempts to schedule the feedback session prior to giving birth, this was considered a failure to engage. 22% of MI (5), and no PI met this criterion. Additionally, 1 MI withdrew prior to participation. Final intervention group composition was 16 MI and 30 PI. After enrolling, three women (1MI, 2PI) moved out of state, and one (PI) miscarried during the study. Their engagement is included in the following results, limited to their time in the study.

Sample Characteristics

A total of 92 women (40 controls) completed baseline surveys after consenting. The sample was representative of the overall age, race and SES served. Average gestational age at recruitment was 20.6 weeks (SD = 3.6) and average age of participant was 30.6, (18–42; SD: 5.14). At baseline, providers identified participant challenges based on standardized survey cutoffs and/or mean participant data (see Table 1). MI reported more baseline challenges (M = 3.50, SD = 1.89) than PI (M = 1.46, SD = 1.33, p < 0.001). MI were more likely to experience sleep (x(1) = 6.62, p = 0.015), stress (x(1) = 7.318, p = 0.013), trauma (x(1) = 8.276, p = 0.007), and nutrition (x(1) = 7.21, p = 0.012) challenges than PI. Exercise, partner support, drug use, and mental health domains did not significantly differ by insurance. Fifty-three percent (49 women) followed up at 12 months. They were similar in terms of baseline characteristics though more likely to have PI (x(1) = 5.90, p = 0.015).

Engagement

Of women randomized to the intervention, MI attended fewer total wellness offerings (M = 3.81, SD = 12.30) than PI (M = 12.30, SD = 11.71) (t(41.64) = 3.49, p = 0.001). Women were most likely to attend Yoga and Parenting Classes, followed by Nutrition Classes, Community Dinners, Mindfulness, and Exercise Classes. Insurance status did not determine type of program women attended. 27 women attended three or more activities and were categorized as “engagers”. Eight of those were MI. Six women participated in individual therapy and no women were referred to psychiatry as a result of this study (the study psychiatrist also worked in the clinic so had contact with study participants via obstetrician’s referrals).

PI (B = − 8.49, p = 0.008) predicted higher program attendance (R² = 0.15, F(1,45) = 7.69, p = 0.008). Additionally, after controlling for insurance in separate models, more childhood trauma (ACES) (B = − 1.33, p = 0.038) predicted lower attendance (R² = 0.23, F(2,45) = 6.45, p = 0.004). No other baseline characteristics influenced engagement over and above insurance. Age was associated with program attendance in opposite directions by insurance group: In MI, older women attended more programming (r(22) = 0.637, p = 0.008), however, in PI, younger women attended more programming (r(30) = − 0.613, p < 0.001).

Outcomes (Table 2)

Table 2.

Mean scores for perceived stress and internalizing at baseline and 12 month postpartum (n = 49) controlled for Medicaid

		Perceived stress scores (PSS)		Internalizing raw scores (ASR)
	Group	Baseline M (SD)	Follow up M (SD)	Baseline M (SD)	Follow up M (SD)
	All women	12.53 (.95)	12.69 (1.04)	12.78 (1.41)	8.70 (1.08)
Randomization	Control (n = 20)	11.16 (1.30)	11.42 (1.43)	12.81 (2.42)	6.15 (2.00)
	VFBA (n = 29)	13.91 (1.38)	13.96 (1.52)	12.62 (2.11)	10.45 (1.74)
Intervention engagement	VFBA engaged (n = 12)	16.38 (2.26)*	12.65 (2.35)*	11.95 (1.93)	10.65 (1.96)
	VFBA disengaged (n = 11)	11.73 (2.15)*	15.23 (2.24)*	14.99 (2.85)	8.52 (2.89)

^*p < .05;

n included 12 MI and 37 PI

49 women completed 12 month questionnaires, (12 MI and 37 PI, 29 VFBA and 20 Controls).

There were no differences in internalizing scores across time by intervention (vs controls) or by engagers (3 + classes vs not) from baseline to 12 months postpartum, controlling for insurance status. There were no differences in perceived stress scores across time by intervention (vs controls) or by engagers from baseline to 12 months. However, women who engaged reported a significant decrease in perceived stress relative to controls from baseline to 12 months, controlling for insurance status (F(1,20) = 6.494, p = 0.019) (Fig. 2).

Fig. 2.

Perceived stress score change from baseline to 12 months postpartum by engagement status

Conclusions for Practice

This study implemented a participant-centered health and illness prevention program for women seeking prenatal care. Recruitment and engagement rates were lower for women with Medicaid. Higher engagement was predicted by insurance type and lower levels of reported trauma. Compared to controls, women who were randomized to the intervention did not experience a decrease in depression or anxiety symptoms or perceived stress. However, women in the intervention who chose to engage in 3 + activities demonstrated a decrease in perceived stress from baseline to 12 months postpartum that was not seen in women who didn’t engage.

Recruitment for women with Medicaid in this study was low, even though signing consent carried virtually no risk and very low commitment beyond completing questionnaires. Our initial goal was to recruit 30 women in each of the MI and PI group but we were unable to meet this goal for the MI group over 18 months. Upon careful review of previous prevention studies, it is not unusual to see samples with predominantly medium to high socioeconomic status. A community study recruiting for a postpartum exercise intervention (Norman et al., 2010) reported a disproportionately high SES in their sample. This bias was also reported by Davis et al. (2015) and Muzik, et al (2012) in prenatal yoga trials, and by Woolhouse et al. in a prenatal mindfulness intervention (2014). There’s a public health argument that universal interventions, such as ours, may only increase the discrepancy between populations who are already thriving and more vulnerable populations who have increased barriers to participate. For this reason, Frohlich and Potvin (2008) suggest that resources may be best utilized by specifically targeting vulnerable populations.

Overall engagement in activities was lower than expected given that activities were free, co-located, and offered at women’s convenience. The most popular events were yoga and parenting groups, with just over half of the women participating. A small number (6) of women engaged in therapy, however only two women were referred based on baseline ASR internalizing score. One of the least popular offerings was individual wellness coaching visits, which we had envisioned as a cornerstone of this program. The FWC either taught or attended many of the activities, so it’s possible women preferred informal check-ins at events or through email or text, rather than separately scheduled meetings. As was the case with recruitment, insurance type clearly predicted participation. Women on Medicaid were less likely to attend the visit where feedback was given and goal setting occurred, and of those who did, they were still engaging in activities at about a third of the rate PI women were. An additional baseline characteristic associated with participation was childhood trauma, unsurprisingly, higher exposure to trauma was a barrier to participation. Prior work has linked childhood trauma to high avoidance and selfreliance, which may limit participation in health regimens (Grote et al., 2007).

We would not have expected women with higher stress at baseline to participate more than those with less stress, yet it was consistent with the aim of this study to find that engagement in these activities was associated with decreased stress during the first year. Notably, the cut off for a “clinical” rating on the perceived stress scale is 20, and our means were all well beneath that. We speculate that having moderate, but not clinically impairing, stress could motivate engagement in wellness offerings which may in turn be efficacious in dampening stress. There may be a specific window of “high, but functional stress” which is best suited for sustainable perinatal wellness promotion programs.

This study had several limitations. First, participants participation was highly variable in terms of number and type of activities attended, making it difficult to see associations between particular health promotion activities and their impact on illness prevention, particularly in such a small sample. We also had limited follow up with 49 of the 92 women filling out questionnaires at 12 months postpartum, only 12 of whom had Medicaid, further constraining our ability to draw conclusions. Finally, our overall sample was unusually well in terms of mental health compared with usual estimates, which may reflect a bias in women who indicated interest in a wellness study, or possibly that the ASR is less sensitive than other tools, such as the Edinburgh Postnatal Depression Scale, which also measures depression and anxiety. No studies have compared the two instruments. A CDC estimate of postpartum depression in the United States in 2018 was around 13% (Bauman, et al., 2020), while at baseline we had only 3.8% of women reaching the clinical cutoff for the internalizing scale. Recruiting and engaging a higher number of MI, whose risk for internalizing symptoms is higher (Ogbo et al., 2019), may have made it more likely to see a decrease. A study of collaborative depression care, which involved screening and treating women with depression early in pregnancy found higher effectiveness in women on public insurance (Katon et al., 2014).

In future iterations of VFBA studies, we aim to increase engagement in the Medicaid group by going further to address barriers to participation for at risk women, such as providing childcare and transportation. We also plan to add financial incentives for participation, as is beneficial in smoking cessation studies (Higgins et al., 2011), which would provide money to help address barriers and to provide a short-term reward to help improve the appeal of a paradigm that aims to provide long-term benefits.

Availability of data and material

Available upon request.