Recruitment occurred at six prenatal care (PNC) clinics between July 2001 and October 2003. Primary and secondary hypotheses were specified in advance. To ensure statistical power for testing the hypotheses that the integrated intervention would achieve statistically significant reductions in the targeted risk factors, sample size determination was essential. Assuming a 5% level of significance, 80% power, 20% drop-out rate, and 20% loss to follow-up, a sample size of 1,750 pregnant women was required in order to retain 1,050 at the end of the study (525 in each care group). This number allowed detection of 10-20% reductions in risk-specific factors in the intervention group (IG) as compared to the usual care (UCG). Details are published in El-Khorazaty et al.19
During 28 months of recruitment, 4,213 women were approached for Audio-Computer Assisted Survey Interview screening, and 2,913 women consented. Eligibility criteria included minority status, age ≥ 18 years, ≤ 28 weeks of pregnancy, DC resident, English speaking, and reporting any of the four designated risks (active smoking, ETSE, depression, and IPV). Smoking was defined as smoking in the six months prior to pregnancy or since learning they were pregnant. ETSE was defined as exposure to smokers at home, in the same room, or in a car. Depression was evaluated using the Beck Depression Inventory II.20,21
For IPV, women were asked if a current or previous partner, boyfriend, husband, or baby's father had pushed, shoved, slapped, kicked, or physically hurt them or forced them to have sexual intercourse in the last year, or if they were afraid of their current partner. Eligible women were invited to participate and those interested consented to study participation.
A baseline interview gathered sociodemographic data and information on the designated risks. Survey data items assessing smoking and ETSE were taken or adapted from the Smoke-Free Families screening and baseline questionnaires,22
the 1990 and 1998 National Health Interview Survey supplements,23-26
and ETSE intervention studies.27
Additional items assessed stage of change for smoking cessation28,29
and processes of change.30
Depression was identified using the Hopkins Symptom Checklist, a validated, widely used depression score based on a set of twenty items.31,32
Respondents rated each item on a 5-point scale from 0=Not at all to 4=Extremely distressed by various symptoms within the past month (e.g., poor appetite, feeling lonely or blue, restless sleep, thoughts of death and dying). Responses were summed, divided by 20, and resulting scores classified into one of two groups; ≤.75 = remission or no depression and > .75 depression. Survey items for IPV included the Conflict Tactics Scale33-35
, a validated scale to measure risk of intimate partner violence and safety behaviors items.36-38
The conflict tactics scale measured annual frequency of physical assault and sexual coercion (partner to self). A single item assessed whether any IPV on the Conflict Tactics Scale occurred during pregnancy (either partner to self or self to partner). Active smoking was defined as smoking a puff of a cigarette in the previous week, ETSE as having been in the same room or area where someone was smoking in the past week; depression as having a score >0.75 on Hopkins Symptom Checklist; and IPV was confirmed if a woman reported being subjected to any of the actions on the revised Conflict Tactics Scale at least once by her partner in the past year. After telephone interview, women were randomized to IG or UCG.
Interventions for active and passive smoking were based on the Social Cognitive Theory39,40
and the Transtheoretical Model.41
The smoking interventions used a Cognitive-Behavioral Therapy orientation42-44
tailored to individuals' “stage” of readiness for change. Active smoking intervention included the “Smoking Cessation or Reduction in Pregnancy Treatment” materials,45-47
and incorporated some content from the “Pathways to Change” manual. 48
ETSE intervention paralleled the active smoking intervention. The primary objectives were to promote smoking cessation, ETSE avoidance and/or reduction depending on the participants' readiness.
The cognitive behavioral therapy intervention for depression was adapted from a group intervention by Miranda and Munoz.49
The IPV intervention was adapted from the Parker-McFarlane structured intervention,13
to individualized counseling sessions. Behavioral counseling for IPV was integrated from a brochure-based approach using Dutton's Empowerment theory50
found to be effective in increasing safety behaviors and reducing IPV at 6 and 12 months postpartum.13
To be considered adequate, the intervention was intended to be delivered prenatally in a minimum of four sessions, with eight sessions considered complete. Up to two postpartum booster sessions were offered, reinforcing skills and goals and adapted to specific postpartum stressors.
The intervention delivered by Masters' degree trained counselors, lasted 36±15 minutes per session prenatally and 38±13 minutes postpartum. On average, participants attended 3.9 prenatal sessions and 0.8 postpartum sessions. Fifty-four percent of the IG attended at least 4 prenatal sessions, while 51% attended at least one postpartum. (For details on Project DC-HOPE, see El-Khorazaty et al.19
). Eight women (6 IG and 2 UCG) were identified as suicidal (during intervention or data collection) and were referred immediately to mental health care and excluded from further participation. The anticipated additional cost to providing the intervention would be the hiring of a Master's level social or mental health professional and the space needed to deliver the service privately. Additional infrastructure costs are limited.
Women randomized to IG received behavioral counseling addressing the risk factors they reported. Two follow-up interviews were administered during the second trimester (22-26 weeks gestation) and third trimester (34-38 weeks gestation). A final interview was conducted at 10.3 weeks postpartum on average. During these interviews women were evaluated for smoking, depression, and ETSE using same the definitions as at baseline. IPV was defined as victimization during the time interval between two consecutive interviews. Interviewers were blinded to group assignment, while participants and the intervention team were not.
To preserve the randomization, participant data were analyzed according to their randomized group assignment, regardless of receipt of intervention, using an intent-to-treat approach. Of the 913 African American women with one or more risks at baseline, 723 participated in the postpartum telephone interview. In order to make an intent-to-treat analysis possible, multiple imputation (MI) methodology was implemented to estimate missing data for the remaining 190 (49 drop-outs, 111 lost to follow-up, and 30 deemed ineligible) using the sequential regression imputation method described by Raghunathan, et al.51
A linear, logistic, or polytomous regression model was used to impute continuous, binary, or categorical missing values. MI was performed to create five complete datasets, using IVEware imputation and variance estimation software.52
Analyses were conducted for each of the five imputed datasets and the results were combined results using the MIANALYZE procedure in SAS53
to obtain the multiple imputation inference. The final parameter estimates and the associated standard errors account for both within- and between-imputation variance. Basic descriptive statistics provided information regarding characteristics of participating women stratified by randomization assignment. Comparisons were conducted using General Linear Models for continuous variables and Generalized Linear Models for categorical variables. The GLM and GENMOD procedures in SAS were used. Frequencies presented in the tables are rounded to the nearest integer.
To quantify postpartum behavioral changes and to assess effects of the intervention, the distributions of risks in the two randomized groups at baseline and postpartum were compared. Within-person changes in number of risks over time was also evaluated. Each woman was characterized with respect to number of risks (1-4) at baseline. Each of the 913 participants was categorized postpartum in one of three ways: (1) Resolving all risks (RA); (2) Resolving some, but not all, risks (RS); and (3) Resolving no risks or increasing the number of risks (RN).
Intervention effectiveness was measured contrasting the proportion of women RA versus the proportion RN in the two groups. The proportions of participants resolving all or some risk were also compared between groups. To evaluate the impact of the intervention on risk status at postpartum, we used logistic regression analyses. Analyses were conducted using the LOGISTIC procedure in SAS version 9.1.53
Control variables included age; education; marital status; employment status; enrollment in Medicaid and WIC; illicit drug and alcohol use at baseline; previous premature delivery, pregnancy loss, or live birth; time of PNC initiation; gestational age at baseline; smoking, ETSE, depression, and IPV at baseline.
All study related activities were approved by the Howard University IRB, the designated IRB of record for the clinical study sites, the RTI International IRB and the National Institute of Child Health and Human Development IRB.