Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2020 Feb 1.
Published in final edited form as: Drug Alcohol Depend. 2018 Dec 8;195:33–39. doi: 10.1016/j.drugalcdep.2018.11.019

An ecological momentary assessment study examining posttraumatic stress disorder symptoms, prenatal bonding, and substance use among pregnant women

Pilar M Sanjuan a, Matthew R Pearson a, Carly Poremba a, Hortensia Amaro b, Lawrence Leeman c
PMCID: PMC6359965  NIHMSID: NIHMS1516302  PMID: 30572290

Abstract

Background:

Substance use disorder (SUD) during pregnancy requires efficacious interventions based on understanding the ebb and flow of risk and protective factors for substance use across time. To assess how these fluctuations are associated temporally with substance use, we used ecological momentary assessment (EMA) to evaluate substance use risk (posttraumatic stress disorder [PTSD] symptoms) and protective (prenatal fetal bonding) factors and their associations with prenatal substance use recorded in real time.

Methods:

Pregnant women in SUD treatment (N=33) with prior trauma exposure received smartphones with an EMA application that queried them thrice daily for 28 days about PTSD symptoms, prenatal bonding, and substance use.

Results:

Nearly all (N=32) provided EMA data resulting in 2,049 EMA reports (74% compliance). Most participants reported tobacco (72%), alcohol (22%), heroin (41%), and/or other illicit drug (6% - 31%) use at least once via EMA. There were moderate associations (average β=.23) between greater daily peak PTSD symptoms and substance use with significant effects on illicit drug (β=.37), cannabis (β=.35) and cigarette use (β=.24). Prenatal bonding subscales were modestly associated with substance use with daily intensity of attachment low point associated with lower heroin (β=−.34), but higher alcohol use (β=.24). Quality of attachment low point was associated with higher cigarette use (β=.06).

Conclusions:

Despite the SUD severity and social instability of this sample, we observed high rates of compliance. We found preliminary support suggesting daily PTSD symptoms as a risk factor and less consistent support for prenatal bonding as a protective factor for prenatal substance use.

Keywords: Posttraumatic stress disorder, prenatal, substance use, ecological momentary assessment, prenatal fetal bonding

1. Introduction

Approximately 5% of women entering treatment for a substance use disorder (SUD) are pregnant (SAMHSA, 2013). Between 1999 and 2014, the rate of opioid use disorder documented at in-hospital births quadrupled (Haight et al., 2018). Socially vulnerable and stigmatized, many pregnant women with SUD lack access to resources and face high risks for negative prenatal and postpartum consequences, including homelessness, loss of child custody, preterm delivery, stillbirth, infant death, and child abuse and neglect (Kahila et al., 2010; King et al., 2015). Furthermore, women with deliveries complicated by alcohol/drugs are 38 times more likely than other women to die within nine years of giving birth (Kahila et al., 2010). High rates of posttraumatic stress disorder (PTSD) symptoms disproportionately affect these women (Erickson and Tonigan, 2008; Moylan et al., 2001). Finally, the recent U.S. increase of opioid use corresponds with a marked increase in neonatal abstinence syndrome (NAS) (Patrick et al., 2015), the severity of which can be decreased by medication assisted treatment (MAT) (Reddy et al., 2017). However, pregnancy is associated with poor treatment outcomes (Davie-Gray et al., 2013; Delano et al., 2013; Wilder et al., 2015).

Factors associated with substance use during prenatal SUD treatment must be determined so the specific needs of pregnant women can be better addressed. Many people are exposed to the types of traumatic events that can cause PTSD, but only about 10% of these individuals will ultimately develop PTSD (Kessler et al., 2005, 1995). Evidence suggests that symptoms of PTSD (Linden et al., 2013; Seng et al., 2014) are an important risk factor for substance use in this population, whereas positive prenatal bonding (i.e., the relationship/feelings pregnant women develop toward their fetuses) may be protective against substance use (Lindgren, 2001; Massey et al., 2015). However, the more precise temporal interactions of these factors are not known. It is therefore necessary to more closely examine the association between PTSD symptoms, prenatal bonding, and substance use in this population.

Because symptoms of PTSD are expected to fluctuate (Possemato et al., 2015), and individuals may use substances on some days but not others, we used an ecological momentary assessment (EMA) design to pinpoint how these constructs are associated with each other across time. EMA uses mobile devices (e.g., smartphones) to collect information (e.g., symptoms, emotions, substance use) multiple times per day in real-time. Thus, EMA is used to assess how symptoms and emotions temporally predict substance use (Morgenstern et al., 2014). EMA is less prone to retrospective bias error than traditional surveys and is ideal for identifying targets for precision health interventions. Because emotions and well-being can fluctuate greatly during pregnancy, it has been suggested that EMA may be a better method for measuring prenatal mood than traditional single time-point surveys (Belisario et al., 2017; Newham and Martin, 2013). EMA has been used successfully during pregnancy to examine mood and tobacco use (Allen et al., 2018), cortisol (Entringer et al., 2011), stress (Walsh et al., 2015), and movement (Faherty et al., 2017). Daily fluctuations in prenatal bonding and PTSD symptoms in pregnant women with SUD have not been determined, but if substance use risk factors such as PTSD interfere with prenatal bonding (a potential protective factor against substance use), then prenatal bonding may also be unstable across time in this population. Such instability in PTSD and/or prenatal bonding may be associated with substance use. EMA captures these complex relationships.

1.1. Purpose

The present study had four primary objectives: 1) characterize EMA compliance rates among treatment-engaged pregnant women with SUD, 2) characterize their substance use, 3) examine the convergent validity of EMA and retrospective measures of PTSD symptoms and prenatal bonding, and 4) examine whether PTSD symptoms and prenatal bonding are proximally associated with prenatal substance use. We hypothesized that higher peak daily PTSD symptoms would be associated with higher substance use at the daily level, while lower prenatal bonding daily low points (quality of attachment, intensity of attachment) would be associated with higher substance use at the daily level.

2. Method

2.1. Participants

Pregnant women (N=33) with SUD and prior trauma exposure were recruited and assessed. All women were recruited from a comprehensive program for prenatal, maternity, and neonatal care integrating medical and SUD treatment in the clinical setting of a public-sector hospital associated with an academic institution in the southwest United States (see Figure 1 for participant flow).

Figure 1.

Figure 1.

Open in a new tab

Participant flow from recruitment to study end. EMA=Ecological momentary assessment; N=number.

Inclusion criteria were (1) clinic enrollment, (2) prior exposure to DSM-5 Criterion A traumatic events - actual or threatened death, serious injury, or sexual violation (American Psychiatric Association, 2013), (3) pregnancy between 20–35 weeks gestation at baseline, (4) 18 years or older, (5) able to read, write, and speak English. Fetal gestational age was restricted at the lower end to 20-weeks at baseline to maximize fetal bonding, which is typically measureable by 10 weeks gestation (Caccia et al., 1991) and increases throughout gestation (Grace, 1989). The upper-end cutoff of 35 weeks was set to increase likelihood of study completion (4 weeks) before childbirth. Exclusion criteria were (1) acute psychosis, (2) acute and uncontrolled severe mental illness or suicidality, (3) inability to provide informed consent, (4) pregnancy complications likely to interfere with study involvement (e.g., planned hospitalization), (5) immediate plans to relocate from the area, or (6) current incarceration. We also did not approach individuals who were (1) enrolled in other studies that potentially conflicted with our protocol, (2) at the clinic for postpartum visits, or (3) engaged in medical treatment that rendered them unable to participate.

2.2. Recruitment and consent

Prospective participants were clinic patients either approached in private rooms at the prenatal or opioid treatment program clinics while waiting to see medical providers, or who contacted our offices after learning about study via fliers at the clinics or word-of-mouth. At the prenatal clinic, study staff attempted to screen all eligible patients on recruitment days. Study staff were trained to maintain a compassionate and non-judgmental approach. After obtaining verbal consent, study staff administered a brief screening form while patients waited in private exam rooms for medical providers. Other interested clinic patients contacted study staff for screening via telephone or in person at study offices. Eligible participants were scheduled for baseline appointments and given consent forms to review at screening. Informed consent included statements that the study was protected by a Certificate of Confidentiality from NIH, that staff was required by state law to report suicidal and homicidal intent and child or elder abuse, but that drug or alcohol use during pregnancy was not mandated to be reported in this region, and thus this information would remain confidential. All participants gave written informed consent prior to baseline assessments, and the IRB at the university approved the protocol.

2.3. Assessment process

2.3.1. Baseline assessment.

Including informed consent, baseline assessments lasted 4-5 hours and were conducted in one continuous visit with intermittent breaks. Baseline PTSD diagnostic assessments were conducted by the licensed psychologist or comprehensively-trained master’s-level clinicians under supervision (as recommended by the measure). All other parts of the baseline assessment were conducted by either the aforementioned licensed psychologist or a rigorously-trained and supervised staff research assistant (RA) with a psychology bachelor’s degree (as recommended for those measures). Statistical comparisons found no systematic differences in diagnoses associated with educational level of test administrators (data not shown).

2.3.2. Ecological Momentary Assessment (EMA).

At the end of baseline assessments, participants were given smartphones with the EMA application already installed and instructed how to operate the application. They completed their first assessment in the office with opportunities for questions. Participants were asked to complete three surveys per day for 28 days, and each survey took 5-10 minutes to complete. During the EMA survey, participants saw one question at a time on the screen, and all responses were multiple-choice buttons. The smartphone alarm application delivered one audiovisual prompt to complete each survey labeled by time of day (e.g., “morning assessment”). To provide flexibility for this population, there was no time-limit for completing assessments following prompts. All responses were timestamped by the EMA application, and data were cleaned using these timestamps prior to analysis to eliminate duplicate entries. We asked participants to identify set times that worked best for them for survey completion (e.g., not driving, not feeding children) corresponding to 1) upon waking, 2) afternoon around 2pm, and 3) before bedtime. Participants could neither change answers nor access surveys already completed. Data uploaded whenever the phone connected to Wi-Fi. For most participants this only occurred at study offices, but some had data upload via Wi-Fi at home. The smartphones, which participants were allowed to keep after study completion, were not connected to phone, text, or data plans during the EMA phase.

2.4. Measures

2.4.1. PTSD severity, diagnosis, and symptomatology.

The Clinician Administered PTSD Scale-5 (CAPS-5: Weathers et al., 2013a) - the gold-standard for PTSD diagnosis - was used to determine baseline PTSD severity and diagnosis. We used the well-validated National Women’s Study Event History-PTSD Module (Resnick, 1996) to collect a detailed list of prior traumatic experiences. For EMA data collection, the PTSD Checklist (PCL-5: Weathers et al., 2013b) was administered. The PCL-5 is a well-validated measure consisting of a list of PTSD symptoms to assess current PTSD severity. We calculated the highest reported PTSD symptom severity daily (peak PTSD symptoms), because we expected high PTSD symptom severity to be associated with substance use.

2.4.2. Prenatal bonding.

The Maternal Antenatal Attachment Scale (MAAS) is a validated measure of attachment (Condon, 1993), and we used the full scale to measure prenatal fetal bonding at baseline. This scale consists of two factors, (a) “quality of attachment,” which represents quality of maternal affective experiences (e.g., closeness/distance, positive/negative, tenderness/irritation) and (b) “time spent in attachment mode,” which represents intensity of preoccupation with the fetus (based on amount of time a woman spends thinking about, talking to, or palpating her fetus). We then used the three highest loading items as reported in Condon (1993) of each of the subscales for the EMA survey, and these were queried at each EMA report. We calculated the lowest reported prenatal bonding scores for each scale daily (Quality of Attachment Low and Intensity of Attachment Low), because we expected low prenatal bonding to be associated with substance use. Thus, a lower score for “quality of attachment low” on a particular day indicates that when the participant was feeling the least attached that day she felt more distant, negative, and irritated towards the fetus than she did on days with higher scores when she felt closer, more positive, and more tender during her least attached periods those days (i.e., lower lows represent less bonding, while higher lows represent more bonding).

2.4.3. Alcohol, tobacco, and other drug use.

Substance use disorders were determined at baseline using the Structured Clinical Interview for DSM5 Disorders, Section E (First et al., 2015). The Fagerstrom Test for Nicotine Dependence (FTND; Heatherton et al., 1991), a well-validated measure, was used to determine severity of nicotine dependence at baseline. Substance use was assessed at each EMA survey with an item that queried any use of drugs (heroin, benzodiazepines, amphetamines, cocaine, cannabis, or other drugs) since the previous EMA session. Alcohol use was assessed during each EMA survey with an item asking number of standard drinks since the previous EMA session, and tobacco use was assessed at the same times via an item asking number of cigarettes smoked.

2.4.4. Participant profile.

A study-specific demographics-history questionnaire administered at baseline assessed age, race, ethnicity, education, criminal justice, and socioeconomic variables.

2.5. Participant compensation

Participants were compensated with merchant gift cards based on appointment type and EMA completion rate: $50 for the 4-5-hour baseline appointment, $10-40 for each EMA download (dependent upon EMA completion rate), and $25 extra for the final EMA download. Some participants also received an additional $25 for an extra interview. Including the 4-5-hour baseline assessment, the weekly download and survey sessions, and the 28-day EMA reporting task, participants could earn a maximum of $260.00 for participation.

2.6. Analytic approach

We compared three groups of participants 1) screened out, 2) screened in/did not enter the study, and 3) screened in/consented at baseline. These analyses used ANOVA and chi square analyses in SPSS (IBM Corp., 2017). We conducted multilevel models in Mplus 8 (Muthén and Muthén, 1998-2017) using a Bayesian estimator with non-informative priors to examine the effects of peak daily PTSD symptoms and daily prenatal bonding subscale lows (Quality of attachment low and Intensity of attachment low) on six same-day substance use outcomes aggregated across time for each day: any drug use (0=no, l=any drug use except alcohol/tobacco), cannabis use (0=no, l=yes), heroin use (0=no, l=yes), benzodiazepine use (0=no, l=yes), alcohol use (continuous number of drinks), and cigarette use (continuous number of cigarettes) (see Table 3). “Other” drug use, cocaine use, and amphetamine use were not examined due to low endorsement rates. Statistical significance was reported based on a 95% credibility interval not containing zero. At level 1 (within-subjects), predictors were person-mean centered (n=813 days); at level 2 (between-subjects), predictors were grand-mean centered (n=32 participants). This centering results in the ability to examine independent within-subject and between-subject effects (Raudenbush and Bryk, 2002).

Table 3.

Models Predicting Drug Use, Alcohol Use, and Cigarette Use from Peak PTSD Symptoms and Prenatal Bonding Low (Quality of Attachment and Intensity of Attachment)

Drug Use Cannabis Use Heroin Use Alcohol Use Cigarette Use
WITHIN-SUBJECTS β B 95% CI β B 95% CI β B 95% CI β B 95% CI β B 95% CI
Peak PTSD Symptoms .37 .39 .05, .71 .35 .37 .10, .57 .09 .09 −.22, .28 .11 .02 −.01, .06 .24 .08 .02, .16
Quality of Attachment Low .00 .01 −.31, .31 −.02 −.04 −.22, .02 .02 .07 −.09, .14 .04 .02 −.01, .04 .06 .06 .01, .10
Intensity of Attachment Low −.09 −.08 −.34, .22 −.22 −.21 −.36, .11 −.34 −.32 −.50, −.04 .24 .04 .02, .07 .05 .02 −.03, .05
BETWEEN-SUBJECTS B B 95% CI β B 95% CI β B 95% CI β B 95% CI B B 95% CI
Peak PTSD Symptoms .15 .25 −.45, .97 .27 .84 −.04, 1.0 −.07 −.08 −.41, .42 .17 .06 −.16, .18 −.20 −.19 −.50, .12
Quality of Attachment Low −.07 −.38 −1.4, .59 −.09 −1.2 −1.5, .30 −.16 −.54 −1.5, .30 .07 .09 −.40, .57 −.31 −.98 −2.3, .26
Intensity of Attachment Low −.19 −.31 −1.1, .37 −.01 −.37 −.44, .56 .04 .05 −.35, .56 −.30 −.09 −.22, .04 −.04 −.03 −.35, .31
Open in a new tab

Note. CI=Bayesian Credibility Interval. Drug use, cannabis use, and heroin use were modeled as categorical variables (0=no use, 1=use) with a probit link function, alcohol use and cigarette use were modeled as continuous variables (number of drinks, number of cigarettes, respectively). Statistically significant effects (95% CI not including zero) are in bold typeface for emphasis.

3. Results

3.1. Recruitment of sample

Study flow is illustrated in Figure 1. We contacted 98 patients, 12 were not interested in participating, 3 lived too far away, 2 were in a conflicting study, and we lost contact with 4 interested participants before screening. Of the 77 participants screened, we found 60 eligible patients, of whom 33 participated in the study. As shown in Figure 1, about half the participants who screened in but did not participate failed to show up for baseline appointments while the other half repeatedly rescheduled until they became ineligible due to gestational age past 35-weeks (timed-out). We compared all tracking variables collected at screening between ineligible individuals, eligible individuals who did not participate, and participants. We found no significant differences between groups on age, race, ethnicity, gestational age, MAT type, number of reported mental health diagnoses, or reported depression. Ineligible patients, as a group, had less reported prior PTSD diagnoses than patients who screened in (see Table 1). This was as expected, because trauma exposure was an inclusion criterion.

Table 1.

Sample versus Non-Participants

Measure Screened-out (N=16) Screened-in but did not participate (N=28) Final Participants (N=33)
Mean (SD) Mean (SD) Mean (SD) F (p) Partial Squared eta
Age 27.4 (5.54) 26.8 (4.52) 27.8 (4.8) 0.36 (0.68) 0.01
Mental Health Diagnoses* 0.8 (0.98) 1.6(1.04) 1.58 (1.30) 2.97 (0.06) 0.08
Gestation (weeks at baseline) 22.6 (7.73) 21.7 (8.59) 22.5 (6.44) 0.10 (0.91) 0.00
χ2 Cramer’s V
Hispanic 68.8% 67.9% 66.7% 0.02 (0.99) 0.02
Caucasian 87.5% 85.7% 78.8% 0.79 (0.67) 0.10
African American 6.3% 3.6% 3.0% 0.31 (0.86) 0.06
Native American 6.3% 10.7% 21.2% 2.43 (0.30) 0.18
MAT (reported at screening) None 20.0% None 3.6%% None 6.3% 7.49 (0.11) 0.22
Methadone 20.0% Methadone 57.1% Methadone 43.8%
Suboxone 60.0% Suboxone 39.3% Suboxone 50.0%
Prior reported PTSD* 18.8% 59.3% 48.7% 7.40 (0.03)** 0.31
Prior reported Depression* 25.0% 59.3% 50.0% 5.20 (0.07) 0.07
Open in a new tab

Note. N= Number, SD=Standard Deviation, MAT=Medication Assisted Treatment, PTSD=Posttraumatic Stress Disorder

*

Mental health diagnoses were self-reported at screening.

**

Differences were expected between the screened out and screened in groups on PTSD, as prior trauma exposure was inclusion criteria.

3.2. Participant profile

On average, participants were in the late 2nd trimester (M=26.8 weeks, SD=5.29 weeks, range 18-35 weeks), 27.8 years of age (SD=4.48, range 21-38), and predominantly Hispanic (67%, see Table 2). The majority were unemployed (70%), never married (55%), and had annual household incomes below $20,000 (55%). Nearly half (49%) had been homeless at some point in the last 12-months. Most participants had some college education (61%) and had prior pregnancies (88%). Almost half the participants had other children living with them (42%), and many had children not in their custody (39%). Nearly all (91%) had been arrested, with a mean of 6.6 arrests (SD=7.13, range 1-30) and 36% reported incarceration during a pregnancy.

Table 2.

Sample Profile

Measure (range of scale or norms where available) Range Mean SD
Age 21-38 27.8 4.48
Gestational age of fetus (weeks) 18-35* 26.8 5.29
Percentage (%)**
Non-Hispanic Caucasian/White 27
Hispanic/Latina 67
American Indian/Alaska Native 15
African American/Black 4.5
Range Mean SD
PTSD Severity (CAPS5) (range 0-80) 0-61 25.1 19.22
Maternal Antenatal Attachment Scale (MAAS, Norm 75.7, SD 8.1) 50-94 77.9 10.73
 MAAS Quality of Attachment (Norm 49.2, SD 4.9) 33-50 44.6 4.79
 MAAS Time Spent in Attachment Mode (Norm 26.5, SD 4.8) 13-39 28.7 6.38
Anxiety Sensitivity Index (>17=moderate to high anxiety sensitivity) 0-55 18.7 16.43
Difficulty in Emotion Regulation Scale (Women Norm 78, SD 20.7) 10-126 87.7 26.47
Percentage (%)
Alcohol Use Disorder (Current) 27
Opioid Use Disorder (Current or in remission on MAT) 91
Substance Use Disorder - Non-opioid (Current) 64
Subutex (at baseline - Buprenorphine) 45.5
Methadone (at baseline) 45.5
Open in a new tab
*

Gestational age was recalculated by medical staff in one case between screening and baseline, resulting in one baseline at 18 weeks.

**

Multiracial identification was allowed.

PTSD=Posttraumatic stress disorder; CAPS5=Clinician-Administered PTSD Scale for DSM5; MAAS=Maternal Antenatal Attachment Scale; MAT=Medication Assisted Treatment; SD=Standard deviation

Nearly half (46%) the sample met criteria for PTSD, with a wide range of PSTD severity (Table 2) associated with PTSD diagnosis, t(31)=10.1, p<0.0001, d=3.47. The majority reported an opioid use disorder (91%), and many had current non-opioid SUD (64%). Of those being treated with MAT (91% of participants), half were prescribed buprenorphine and half methadone. Sixty-one percent used tobacco, with nicotine dependence ranging from low to high severity (M=3.6, SD=2.42) as measured by the FTND. At baseline, prenatal bonding and difficulty in emotion regulation were in the normal range, while anxiety sensitivity was in the moderate-to-high range (see Table 2).

3.3. Compliance rates

Of 33 participants who consented and completed baseline assessments, 32 provided EMA data. Each subject was asked to complete three EMA reports per day for 28 days (84 expected reports per person). Thus, 100% compliance would have resulted in a total 2,772 EMA reports from 33 participants. Over the 4-week periods, a final total 2,049 EMA reports were collected (74% compliance).

3.4. Alcohol, tobacco, and other drug (ATOD) use EMA reports

About one-third (34.4%) of the sample reported no illicit drug use during the study period, leaving 65.6% of the sample reporting at least one day of drug use (a total of 149 use days, 18.3% of days). Of these, 37.5% of participants reported heroin (32 use days, 3.9% of days), 28.1% reported cannabis (62 use days, 7.6% of days), 18.7% reported amphetamine (45 use days, 5.5% of days), 12.5% reported benzodiazepine (29 use days, 3.6% of days), 6.2% reported cocaine (2 use days, 0.2% of days), and 3.1% reported “other” drug use (3 use days, 0.4% of days). Some participants used multiple drugs on the same day. Most of this sample (71.9%) reported cigarette (431 use days, 53% of days), and 18.7% reported alcohol use (45 use days, 5.5% of days).

3.5. Convergent validity

We examined the degree to which baseline retrospective assessment scores converged with EMA assessment scores. Baseline PTSD severity scores were strongly correlated with averaged EMA PTSD severity scores, r=0.741. Similarly, baseline MAAS subscale scores correlated strongly with averaged EMA-adapted MAAS subscale scores (quality of attachment, r=0.733; intensity of attachment r=0.681).

3.6. Alcohol, tobacco, and other drug use outcomes

We observed moderate associations (average β=.23) between daily peak PTSD symptoms and substance use at a daily level (see Table 3) with significant effects on drug use (β=.37), cannabis use (β=.35) and cigarette use (β=.24), and non-significant (though directionally consistent) effects on heroin use (β=.09) and alcohol use (β=.11). The effects of prenatal bonding subscales on substance use were more modest and less consistent across models with intensity of attachment low predicting higher alcohol (β=.24), but lower heroin use at the daily level (β=−.34). Quality of attachment low was significantly associated with higher cigarette use, although this effect was quite small (β=.06).

4. Discussion

We collected EMA data with a sample of pregnant women with prior trauma exposure and SUD enrolled in a comprehensive prenatal, maternity, neonatal, and SUD treatment program to examine associations between PTSD symptoms, prenatal bonding, and substance use.

Although our participants on average had relatively severe SUD, economic and social instability (i.e., low income, single, unemployed mothers with justice system involvement; see Section 3.2) and childcare responsibilities (i.e., almost half had other children in their care; see Section 3.2), compliance and retention rates were high. Our compliance rates of 74% were close to rates in less disadvantaged samples (e.g., binge eaters 76%-92% [Haedt-Matt and Keel, 2011], children/adolescents 78%, [Wen et al., 2017]) and in other samples of pregnant women, (e.g., pregnant ex-smokers 84% [Allen et al., 2018] and pregnant women with depression 72% [Faherty et al., 2017]).

More than two-thirds of our sample reported illicit drug use via EMA, nearly one-fifth reported drinking alcohol, and nearly three-fourths reported use of tobacco. Despite legitimate concerns pregnant women may otherwise have about disclosing prenatal substance use, our participants appeared willing to disclose substance use during EMA. Substance use disclosure rates might differ in regions with mandatory reporting laws for prenatal substance use or where cannabis laws are different. State law in this region did not mandate this reporting, and medical use of cannabis was legal (but recreational cannabis use was not). Additionally, our research was covered by a Certificate of Confidentiality from the CT.S. Federal Government, which protected participants’ identities from subpoena, and all participants were notified of this during informed consent. Substance use disclosure rates might also be lower when EMA is collected for clinical purposes instead of research, since demand characteristics or consequences might be greater in that context.

At the within-subject level, we observed associations between PTSD symptoms and substance use with effects ranging from small to moderate. This supports the plausibility of PTSD symptoms as a temporally proximal risk factor for same-day substance use in this population. Given that these substances differ in prevalence, treatment approaches, and teratogenicity, additional research is needed to determine how PTSD symptoms may place pregnant women at immediate risk for using specific substances.

Our results were less clear regarding the potential role of prenatal bonding as a temporally proximal protective factor against prenatal substance use. Lower intensity of attachment low points was significantly associated with more alcohol use, but less heroin use. It is possible that this discrepancy related to treatment approach. Women with opioid use disorders received MAT, but there was no equivalent MAT for prenatal alcohol use disorders. Overall, associations between prenatal bonding and substance use were weaker and less consistent across substance use outcome, highlighting the need to conduct additional research with greater statistical power.

As shown in Table 1, our participant sample did not significantly differ on tracking variables from patients who did not participate nor from those who screened out, except on variables that reflected eligibility criteria, i.e., PTSD requires prior trauma exposure (American Psychiatric Association, 2013). Although participants who opted not to be screened for the study were likely different from those who were interested in participation, our finding reduces concerns that interested pregnant women who were able to participate were substantially and systematically different from individuals who were initially interested but did not participate. Our results showed that pregnant women with SUD were willing and able to adhere to an EMA protocol. Baseline PTSD severity and prenatal bonding were strongly correlated with the average of corresponding EMA reports, supporting the convergent validity of these measures. However, EMA provided the additional benefit of documenting the instability of these processes over time.

4.1. Limitations

Because this was a pilot study, our statistical power was limited for subgroup analyses due to the small sample size (N=32) of the study and was also limited at the within-subject level by the sparseness of our substance use outcomes. Because of this our analyses were conducted at the daily level to maximize statistical power and additional work is needed to examine the lagged and bidirectional associations among these study constructs (e.g., morning PTSD symptoms predicting later prenatal bonding or substance use). Demographics of participants were consistent with the lower socioeconomic population in this region and, thus, represented the general patient population of the clinic. However, African Americans, Asians, and Pacific Islanders are underrepresented in this region, and therefore our results may not generalize to these populations. We had a high attrition rate between screening and baseline visits, and, although those who participated did not differ significantly as a group on tracking variables from patients who did not participate or from those who screened out, they may have differed from participants on unknown variables. We also could not conduct baseline assessments at the clinic or offer taxi service to participants for appointments (which could have affected the number of participants who screened in but did not make it to a baseline appointment before falling outside the gestation eligibility window). Additionally, we do not know how the 12 women who declined to be screened, or the nine women who were unable to be screened differed from the 77 screened women. Consequently, our sample is susceptible to some selection bias. We did not have biological confirmation of self-reported substance use and therefore may have missed some episodes of use.

4.2. Conclusions

In a sample of treatment-engaged pregnant women with SUD and trauma, we found evidence that suggests PTSD symptom severity may be a temporally proximal risk factor for prenatal substance use and that facets of prenatal bonding may be temporally proximal protective factors for alcohol use at a day-to-day level. Given the known emotional fluctuations associated with pregnancy (Newham and Martin, 2013) and consequences of prenatal substance use to mothers and children (Quesada et al., 2012), it is critical to determine proximal antecedents to substance use in this at-risk population. Additional work is needed with EMA to disentangle within-subject and between-subject effects of these factors on prenatal substance use and to determine the best interventions targeting such fluctuating potential proximal risk/protective factors. We demonstrated that EMA research can be conducted with high-risk, disadvantaged, pregnant women with SUD. We had excellent retention, were able to collect EMA data at rates similar to studies with lower-risk populations, and participants reported substantial substance use via EMA.

Highlights.

  • Peak post-traumatic stress disorder (PTSD) symptoms may put women at risk for same day prenatal substance use.

  • Facets of prenatal bonding may protect against same day alcohol use.

  • Women reported prenatal substance use/trauma via ecological momentary assessment.

  • Ecological momentary assessment can be used with pregnant women using substances.

  • Retention rates were similar to studies with lower-risk populations.

Acknowledgements

We kindly thank the medical staff at the clinic, especially Mandy Hatley and Mija Serrano, who accommodated our staff and welcomed us into their space.

Role of the Funding Source

Funding for this study was provided by The National Institutes of Health (grant numbers R25DA035163, K23AA025094, and K01AA023233). Funding sources have had no involvement in study design; in the collection, analysis and interpretation of data; in the writing of the report; or in the decision to submit the paper for publication.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Conflict of Interest

All authors declare they have no conflict of interest.

References

  1. Allen A, Tosun N, Carlson S, Allen S, 2018. Postpartum changes in mood and smoking-related symptomatology: An ecological momentary assessment investigation. Nicotine Tob. Res. 20, 681–689. 10.1093/ntr/ntx118. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. American Psychiatric Association, 2013. Diagnostic and statistical manual of mental disorders: DSM-5™ (5th ed.). American Psychiatric Publishing, Inc., Arlington, VA US. [Google Scholar]
  3. Belisario JSM, Doherty K, O’Donoghue J, Ramchandani P, Majeed A, Doherty G, Morrison C, Car J, 2017. A bespoke mobile application for the longitudinal assessment of depression and mood during pregnancy: Protocol of a feasibility study. BMJ Open 7, e014469 10.1136/bmjopen-2016-014469. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Caccia N, Johnson JM, Robinson GE, Barna T, 1991. Impact of prenatal testing on maternal-fetal bonding: Chorionic villus sampling versus amniocentesis. Am. J. Obstet. Gynecol 165, 1122–1125. 10.1016/0002-9378(91)90484-9. [DOI] [PubMed] [Google Scholar]
  5. Condon JT, 1993. The assessment of antenatal emotional attachment: Development of a questionnaire instrument. Br. J. Med. Psychol 66, 167–183. 10.1111/j.2044-8341.1993.tb01739.x. [DOI] [PubMed] [Google Scholar]
  6. Davie-Gray A, Moor S, Spencer C, Woodward LJ, 2013. Psychosocial characteristics and poly-drug use of pregnant women enrolled in methadone maintenance treatment. Neurotoxicol. Teratol 38, 46–52. 10.1016/j.ntt.2013.04.006. [DOI] [PubMed] [Google Scholar]
  7. Delano K, Gareri J, Koren G, 2013. Rates of fetal polydrug exposures in methadone-maintained pregnancies from a high-risk population. PLoS ONE 8, e82647 10.1371/journal.pone.0082647. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Entringer S, Buss C, Andersen J, Chicz-DeMet A, Wadhwa PD, 2011. Ecological momentary assessment of maternal cortisol profiles over a multiple-day period predicts the length of human gestation. Psychosom. Med 73, 469 10.1097/PSY.0b013e31821fbf9a. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Erickson SJ, Tonigan JS, 2008. Trauma and intravenous drug use among pregnant alcohol/other drug abusing women: Factors in predicting child abuse potential. Alcohol. Treat. Q 26, 313–332. 10.1080/07347320802072040. [DOI] [Google Scholar]
  10. Faherty LJ, Hantsoo L, Appleby D, Sammel MD, Bennett IM, Wiebe DJ, 2017. Movement patterns in women at risk for perinatal depression: Use of a mood-monitoring mobile application in pregnancy. J. Am. Med. Inform. Assoc 24, 746–753. 10.1093/jamia/ocx005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. First MB, Williams JBW, Karg RS, Spitzer RL, 2015. Structured Clinical Interview for DSM-5 Disorders (SCID-5-RV), Research Version, 1st ed. American Psychiatric Association, Arlington, VA. [Google Scholar]
  12. Grace J, 1989. Development of maternal-fetal attachment during pregnancy. Nurs. Res 38, 228–232. [PubMed] [Google Scholar]
  13. Haedt-Matt AA, Keel PK, 2011. Revisiting the affect regulation model of binge eating: A meta-analysis of studies using ecological momentary assessment. Psychol. Bull 137, 660 10.1037/a0023660. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Haight SC, Ko JY, Tong VT, Bohm MK, Callaghan WM, 2018. Opioid use disorder documented at delivery hospitalization — United States, 1999–2014. MMWR Morb. Mortal. Wkly. Rep. 67, 845–849. 10.15585/mmwr.mm6731a1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Heatherton TF, Kozlowski LT, Frecker RC, Fagerstrom K-O, 1991. The Fagerstrom Test for Nicotine Dependence: A revision of the Fagerstrom Tolerance Questionnaire. Br. J. Addict 86, 1119–1127. [DOI] [PubMed] [Google Scholar]
  16. IBM Corp., 2017. IBM Statistics for Windows. IBM Corp., Armonk, NY. [Google Scholar]
  17. Kahila H, Gissler M, Sarkola T, Autti-Rämö I, Halmesmäki E, 2010. Maternal welfare, morbidity and mortality 6–15 years after a pregnancy complicated by alcohol and substance abuse: A register-based case-control follow-up study of 524 women. Drug Alcohol Depend. 111, 215–221. 10.1016/j.drugalcdep.2010.04.014. [DOI] [PubMed] [Google Scholar]
  18. Kessler RC, Chiu WT, Demler O, Walters EE, 2005. Prevalence, severity, and comorbidity of 12-month DSM-IV disorders in the National Comorbidity Survey Replication. Arch. Gen. Psychiatry 62, 617–627. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Kessler RC, Sonnega A, Bromet E, Hughes M, 1995. Posttraumatic stress disorder in the National Comorbidity Survey. Arch. Gen. Psychiatry 52, 1048–1060. [DOI] [PubMed] [Google Scholar]
  20. King PAL, Duan L, Amaro H, 2015. Clinical needs of in-treatment pregnant women with co-occurring disorders: Implications for primary care. Matern. Child Health J. 19, 180–187. 10.1007/s10995-014-1508-x. [DOI] [PubMed] [Google Scholar]
  21. Linden IA, Torchalla I, Krausz M, 2013. Addiction in maternity: Prevalence of mental illness, substance use, and trauma. J. Aggress. Maltreat. Trauma 22, 1070–1084. 10.1080/10926771.2013.845279. [DOI] [Google Scholar]
  22. Lindgren K, 2001. Relationships among maternal–fetal attachment, prenatal depression, and health practices in pregnancy. Res. Nurs. Health 24, 203–217. 10.1002/nur.1023. [DOI] [PubMed] [Google Scholar]
  23. Massey SH, Bublitz MH, Magee SR, Salisbury A, Niaura RS, Wakschlag LS, Stroud LR, 2015. Maternal–fetal attachment differentiates patterns of prenatal smoking and exposure. Addict. Behav 45, 51–56. 10.1016/j.addbeh.2015.01.028. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Morgenstern J, Kuerbis A, Muench F, 2014. Ecological momentary assessment and alcohol use disorder treatment. Alcohol Res. Curr. Rev 36. [PMC free article] [PubMed] [Google Scholar]
  25. Moylan PL, Jones HE, Haug NA, Kissin WB, Svikis DS, 2001. Clinical and psychosocial characteristics of substance-dependent pregnant women with and without PTSD. Addict. Behav 26, 469–474. 10.1016/S0306-4603(00)00141-6. [DOI] [PubMed] [Google Scholar]
  26. Muthén LK, Muthén BO, 1998. Mplus User’s Guide, Eighth. ed. Muthén and Muthén, Los Angeles, CA. [Google Scholar]
  27. Newham JJ, Martin CR, 2013. Measuring fluctuations in maternal well-being and mood across pregnancy. J. Reprod. Infant Psychol. 31, 531–540. 10.1080/02646838.2013.834040. [DOI] [Google Scholar]
  28. Patrick SW, Davis MM, Lehmann CU, Cooper WO, 2015. Increasing incidence and geographic distribution of neonatal abstinence syndrome: United States 2009 to 2012. J. Perinatol 35, 650–655. 10.1038/jp.2015.36. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Possemato K, Maisto SA, Wade M, Barrie K, McKenzie S, Lantinga LJ, Ouimette P, 2015. Ecological momentary assessment of PTSD symptoms and alcohol use in combat veterans. Psychol. Addict. Behav 29, 894–905. 10.1037/adb0000129. [DOI] [PubMed] [Google Scholar]
  30. Quesada O, Gotman N, Howell HB, Funai EF, Rounsaville BJ, Yonkers KA, 2012. Prenatal hazardous substance use and adverse birth outcomes. J. Matern. Fetal Neonatal Med. 25, 1222–1227. 10.3109/14767058.2011.602143. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Raudenbush SW, Bryk AS, 2002. Hierarchical linear models: Applications and data analysis methods, Second. ed. Sage Publications, Thousand Oaks, CA. [Google Scholar]
  32. Reddy UM, Davis JM, Ren Z, Greene MF, 2017. Opioid use in pregnancy, neonatal abstinence syndrome, and childhood outcomes: Executive summary of a joint workshop by the Eunice Kennedy Shriver National Institute of Child Health and Human Development, American College of Obstetricians and Gynecologists, American Academy of Pediatrics, Society for Maternal-Fetal Medicine, Centers for Disease Control and Prevention, and the March of Dimes Foundation. Obstet. Gynecol 130, 10–28. 10.1097/AOG.0000000000002054. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Resnick H, 1996. Psychometric review of the National Women’s Study (NWS) Event History-PTSD Module, in: Stamm BH (Ed.), Measurement of Stress, Trauma, and Adaptation. Sidran Press. [Google Scholar]
  34. Seng JS, D’Andrea W, Ford JD, 2014. Complex mental health sequelae of psychological trauma among women in prenatal care. Psychol. Trauma Theory Res. Pract. Policy 6, 41–49. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Substance Abuse and Mental Health Services Administration (SAMHSA), 2013. Trends in substances of abuse among pregnant women and women of childbearing age in treatment Substance Abuse and Mental Health Services Administration, Rockville, MD. [Google Scholar]
  36. Walsh K, Basu A, Monk C, 2015. The role of sexual abuse and dysfunctional attitudes in perceived stress and negative mood in pregnant adolescents: An ecological momentary assessment study. J. Pediatr. Adolesc. Gynecol 28, 327–332. 10.1016/j.jpag.2014.09.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Weathers FW, Blake DD, Schnurr PP, Kaloupek DG, Marx BP, Keane TM, 2013a. The Clinician-Administered PTSD Scale for DSM-5 (CAPS-5). National Center for PTSD at www.ptsd.va.gov. [DOI] [PMC free article] [PubMed]
  38. Weathers FW, Litz BT, Keane TM, Palmieri PA, Marx BP, Schnurr PP, 2013b. The PTSD Checklist for DSM-5 (PCL-5). Scale available from the National Center for PTSD at www.ptsd.va.gov.
  39. Wen CKF, Schneider S, Stone AA, Spruijt-Metz D, 2017. Compliance with mobile ecological momentary assessment protocols in children and adolescents: A systematic review and meta-analysis. J. Med. Internet Res. 19 10.2196/jmir.6641. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Wilder C, Lewis D, Winhusen T, 2015. Medication assisted treatment discontinuation in pregnant and postpartum women with opioid use disorder. Drug Alcohol Depend. 149, 225–231. 10.1016/j.drugalcdep.2015.02.012. [DOI] [PubMed] [Google Scholar]

RESOURCES