Young women (aged 14−25 years, n=1,047) entering prenatal care at two publicly funded clinics were randomly assigned to standard individual care or group care (). The differences in the quantity and quality of prenatal care are substantial between individual care and group care as described in . Individual prenatal care across the pregnancy occurs over the course of approximately 2 hours. Group prenatal care across the pregnancy occurs over the course of approximately 20 hours.
Fig. 1 CONSORT study description. All outcomes were measured using medical records or at the trimester 3 interview, with the exception of breastfeeding initiation, which was measured at the interview conducted 6 months postpartum (n=783). There was no differential (more ...)
Participants were randomly assigned by using a blocked randomized controlled design, stratified based on site and expected month of delivery. Allocation was concealed from participant and research staff until eligibility screening was completed and study condition was assigned. These tasks were completed by trained research team members who were independent of prenatal care. A computer-generated randomization sequence, password protected to recruitment staff and participants, was used to assign participants. Although it was not possible to have treatment blinded (common practice in clinical interventions), all measurement and data collection were conducted in blinded fashion independently of the care setting. Moreover, medical record abstracters were independent of clinical care.
Participants were recruited from large obstetrics clinics in two university-affiliated hospitals. Procedures were approved by Human Investigation Committees at both sites (No. 11972, Yale University, New Haven, CT, and No. 197−2001, Emory University, Atlanta, GA). African-American women with limited financial resources are overrepresented, reflecting clinic use patterns. There were no deviations from the study procedures as originally planned, with the exception of expanded access by age at study entry from 14−19 years to 14−25 years; this expanded access was implemented before randomization.
Between September 2001 and December 2004, women attending their first or second prenatal care visit were referred by a provider or approached directly by research staff. Inclusion criteria were as follows: less than 24 weeks of gestation, age 25 years or less, no medical problems requiring individualized care as “high-risk pregnancy” (eg, diabetes, HIV), English or Spanish language, and willingness to be randomized. Potential participants were screened; if eligible, research staff explained the study in detail and obtained informed consent. Baseline interviews occurred at an average gestational age of 18 weeks (standard deviation [SD] 3.3). Each patient underwent second-trimester ultrasonography for confirmation of dating and anatomy. Estimated date of confinement was established by an attending obstetrician who was independent of the study, and this was confirmed by ultrasonography. Participants were followed prospectively through 1 year postpartum. All participants were paid $20 for each interview.
Of the 1,538 eligible women, 1,047 (68%) enrolled. Compared with those who declined enrollment, participants were more likely to be African American, older, and at a later gestational age at initial screening (all P<.01). Recruitment was nearly equivalent between the two study sites: Atlanta (n=546, 52%) and New Haven (n=503, 48%). Intervention effects were not statistically different on primary outcomes by study site; therefore, analyses were combined across sites.
Even with randomization, baseline differences can emerge by chance. To evaluate this, we conducted χ2 and t tests comparing the study conditions on demographic, medical history, and major study variables assessed at the baseline interview (). Despite randomization, three differences by study condition were documented. By chance, individuals assigned to group prenatal care were more likely to be African American, less likely to have a history of preterm birth, and more likely to have high levels of prenatal distress. Therefore, all subsequent analyses controlled for these variables.
Baseline Differences by Study Condition
Groups of eight women (on average) are formed based on estimated delivery month and led by a trained practitioner (eg, midwife, obstetrician). The model provides integrative prenatal care by combining three primary components: assessment, education and skills building, and support. All prenatal care occurs within the group setting, except for the initial assessment at entry to care, health concerns involving need for privacy, and cervical assessments late in pregnancy. After the first visit, participants in the study were randomly assigned to continue care individually or in the group setting. When group participants arrive, they engage in self-care activities of weight and blood pressure assessment and update their medical records. Individual prenatal assessments (eg, fundal height, fetal heart rate) are completed by the practitioner during the first 25−30 minutes within the group space. The majority of time is spent with women and clinicians engaging in discussion, education, and skills building to address explicit learning objectives in prenatal care, child birth preparation, and postpartum and parenting roles. Handouts and self-assessment sheets facilitate group discussions and stimulate self-care and evaluation. The full curriculum consists of 10 structured sessions (120 minutes each) conducted from 16 through 40 weeks of gestation. provides a comparative assessment of traditional care and group care.
Structured interviews by audio computer-assisted self-interviewing (audio-CASI) were conducted upon study entry: before session 1 among group participants, and before 24 weeks of gestation in individual care. Audio-CASI allows respondents to simultaneously listen with headphones and see questions on a computer laptop, facilitating completion for participants with lower reading skills. Audio-CASI has been previously validated among pregnant women.15
Trained study staff was present to facilitate the self-interview process by answering any questions and assisting with any technical issues. Medical records were reviewed for 993 participants (95%) by trained medical abstractors who were independent of care and blinded to study assignment.
All encounters to one of the facilities (Yale New Haven Hospital) are recorded electronically in a computerized database used for billing purposes, which identifies sites of care, inpatient compared with outpatient status, International Classification of Diseases, 9th Revision (ICD-9) codes, and cost of care. Therefore, cost data were available at this site (n=503). Cost data included charges, revenue, and actual costs, but only actual costs were used because they are not dependent on reimbursement rates.
Primary outcomes included gestational age at delivery, dichotomized as term or preterm (less than 37 weeks), and infant birth weight, dichotomized as normal or low birth weight (less than 2,500 g).16
All patients underwent second-trimester ultrasound examination for confirmation of dating and anatomy. Estimated date of confinement was established by a consulting obstetrician who was independent of the study, and the date was confirmed by ultrasonography. Decisions on inpatient management and delivery were made by attending physicians and midwives, who were independent of the site of outpatient care, on a pre-established rotating call schedule.
Adequacy of prenatal care was measured by using standard scoring on the Kotelchuck Index.17
Apgar scores at 5 minutes were taken from hospital labor logs. Breastfeeding initiation was based on participant self-report at the first postpartum interview.
All psychosocial outcomes were measured during the third trimester of pregnancy (average gestational age 35 weeks, SD 3.1). Psychosocial outcomes included five domains. Pregnancy knowledge was measured by using a tool developed by the research team to assess prenatal and infant care knowledge; this was not validated. Prenatal distress was measured with the established Pregnancy Distress Questionnaire.18
Readiness for labor and delivery and readiness for infant care scales queried preparedness for delivery and infant care. Satisfaction with prenatal care was measured by using an adaptation of the Patient Participation and Satisfaction Questionnaire.19
Initial analyses were based on intention-to-treat models, with randomized study condition as the primary independent variable: individual compared with group prenatal care. General linear model and logistic regression analyses for basic group differences on birth outcomes, psychosocial factors, and patient satisfaction were conducted. Given differences despite randomization, race, preterm distress, and history of preterm birth were statistically controlled. Additionally, analyses controlled for relevant clinical risks for adverse perinatal outcomes (ie, smoking, history of preterm birth, history of miscarriage or stillborn birth).
Beyond the primary intention-to-treat analyses, several additional analytic approaches were used. Cox proportional hazards analysis was conducted to provide more detailed assessment of time to preterm delivery. Post hoc analysis was conducted to determine if group care had differential outcome for African Americans, who represented 80% of participants. Finally, we evaluated whether there was a potential “dose-response” intervention effect for the primary outcome variables of gestational age and birth weight.
Because the study was originally powered statistically to detect differences in incident sexually transmitted infection, secondary power analyses were conducted for the purposes of these analyses, based on preterm birth as the outcome. With a targeted sample size of 1,040 (n=416 in control and n=624 in the intervention group), we calculated 80% power to detect a 33% reduction in preterm birth (P<.05). This calculation was based on national U.S. base rates for preterm birth, weighted by racial and ethnic distribution in this sample, equivalent to a weighted preterm birth rate of 16.4% (U.S. National Center for Health Statistics, 2006).