Page 6 - Layout 1

and major flaw in the study that would invalidate Analyses) diagram below depicts the flow of records
results. from identification to inclusion (Figure B). Most
records were excluded at full-text screening (n=197)
Appendix D of the full report provides the data
based on the reasons listed in the diagram. Appendix E
extraction, risk of bias, and applicability forms.
provides a list of excluded studies, and Appendix F
Data Synthesis and Analysis provides individual-level study data.
We meta-analyzed the RCTs with a random effects Study Characteristics
model, following a DerSimonian and Laird approach,
Table A presents general summary characteristics of the
when they were clinically and methodologically similar.
included studies. Most studies were observational and
To assess statistical heterogeneity and the magnitude of
included cohorts and case series. Two studies were
heterogeneity, we used Cochran’s Q (a=0.10) and the I 2
RCTs, and one was a nonrandomized trial. Sample
statistic respectively. Odds ratios (ORs) were calculated
sizes ranged from 9 to 1,366, but greater than 70
for dichotomous outcomes and mean differences for
percent of studies included at least 200 participants
continuous outcomes. All analyses were performed
(average 291 ± 395). All studies were from the United
using Comprehensive Meta Analysis version 2.2.046 or
States, and participants were recruited either from
version 2.2.055 (New Jersey, USA). We did not meta-
single-center study sites or from a national proprietary
analyze observational studies because of potential
database run by Matria Healthcare. The Matria database
differences in confounders, nor did we combine studies
provides an outpatient perinatal program consisting of
of singleton and multiple pregnancies. Synthesis of
24-hour nursing and pharmacy support, home uterine
evidence from observational studies was, therefore,
activity monitoring, individualized education, and
undertaken qualitatively. Due to the small number of
provision of tocolytic therapy to women with preterm
studies, we could not perform any meta-regression to
labor. Because five studies originated in the Matria
explore statistical heterogeneity in effect estimates.
database, and not all reported geographic region and/or
Strength of Evidence and Applicability years over which participants were recruited, the
question of overlap in participants across these studies
Based on published guidance for the Effective Health was an important concern of reviewers. Through the
9
Care Program, two reviewers graded the strength of Scientific Resource Center (SRC), we requested this
evidence using the four primary domains (i.e., risk of missing information from Matria (now called Alere)
bias, consistency, directness, and precision) for the Healthcare but did not receive a response. Therefore,
following outcomes: incidence of delivery at various where appropriate, we report this risk of double-
gestational ages (<28 weeks, <32 weeks, <34 weeks, counting of participants.
<37 weeks), mean prolongation of pregnancy,
Several studies included women with RPTL and
bronchopulmonary dysplasia, significant
singleton gestation. Comparator groups included
intraventricular hemorrhage (grade III/IV), neonatal
placebo, no treatment, oral terbutaline, oral nifedipine,
death, death within initial hospitalization, and maternal
and mixed oral tocolytics. The definition of labor was
withdrawal due to adverse effects (Withdrawal-AE). We
unclear in 36 percent of the included studies. The
described population, intervention, comparison,
remaining studies included women with persistent
outcome, timing, and setting characteristics to
contractions and cervical change.
summarize the applicability of the body of evidence.
Results

Study Selection

We screened 427 citations and included 14 unique
records in the review. The PRISMA (Preferred
Reporting Items for Systematic Reviews and Meta-

6