Page 9 - Layout 1

Risk of Bias Assessment such as compliance, adequacy of sample size, and
selective outcome reporting, were unclear. 22,23
We rated studies as low, medium, or high risk of bias
for the relevant reported outcomes. Although the Neonatal Health Outcomes (Key Question 1)
randomization procedures in the two RCTs were
Strength of evidence is insufficient for
appropriate, we rated one RCT as low risk of bias10
bronchopulmonary dysplasia, death within initial
and the second RCT as high risk of bias because more
hospitalization, and significant intraventricular
than 90 percent of eligible participants declined to
hemorrhage (grade III/IV). Based on one retrospective
participate, the study was underpowered, and blinding
cohort of medium risk of bias, the strength of evidence
was ineffective. 11
favoring the SQ terbutaline pump compared with oral
The single nonrandomized trial was high risk of bias tocolytics for neonatal death in women with twin
for the outcomes of birth weight and gestational age at gestation and RPTL is low (Table B). This study
delivery due to potential prognostic imbalances in investigated women from the Matria database and
groups. However, we did not anticipate that such reported a statistically significant difference in neonatal
imbalances would impact the outcome of maternal death in favor of SQ terbutaline pump (OR = 0.09, 95%
hyperglycemia, which we rated as medium risk of bias, CI: 0.01, 0.70).19 Sparse evidence from underpowered
due to insufficient information to assess several other studies addressed necrotizing enterocolitis, retinopathy
criteria. 12 of prematurity, and sepsis with inconclusive results. 11,13
We rated most of the cohort studies as high risk of bias No data were available for periventricular leukomalacia
because there were important group imbalances in and seizures.
baseline characteristics or prognostic factors. 13-15,18,21 The Three retrospective cohort studies from the Matria
other cohort studies we rated as medium risk of bias; database reported stillbirths in women with RPTL and
although these studies had no identifiable flaws, several single or twin gestation. 17-19 All three studies found
criteria could not be assessed due to incomplete nonsignificant differences between the SQ terbutaline
reporting. 16,17,19,20 pump and oral tocolytics. However, these studies were
Lastly, we rated the two case series as medium risk of likely underpowered to detect a difference in still birth,
bias because neither study provided clear definitions for given the small number of events (<1%).
the pump-related harm outcomes, and several criteria,

9