Objective
The purpose of this study was to determine the influence of planned mode and planned timing of delivery on neonatal outcomes in infants with gastroschisis.
Study Design
Data from the Canadian Pediatric Surgery Network cohort were used to identify 519 fetuses with isolated gastroschisis who were delivered at all tertiary-level perinatal centers in Canada from 2005-2013 (n = 16). Neonatal outcomes (including length of stay, duration of total parenteral nutrition, and a composite of perinatal death or prolonged exclusive total parenteral nutrition) were compared according to the 32-week gestation planned mode and timing of delivery with the use of the multivariable quantile and logistic regression.
Results
Planned induction of labor was not associated with decreased length of stay (adjusted median difference, –2.6 days; 95% confidence interval [CI], –9.9 to 4.8), total parenteral nutrition duration (adjusted median difference, −0.2 days; 95% CI, –6.4 to 6.0), or risk of the composite adverse outcome (relative risk, 1.7; 95% CI, 0.1–3.2) compared with planned vaginal delivery after spontaneous onset of labor. Planned delivery at 36-37 weeks’ gestation was not associated with decreased length of stay (adjusted median difference, 5.9 days; 95% CI, –5.7 to 17.5), total parenteral nutrition duration (adjusted median difference, 3.2 days; 95% CI, –7.9 to 14.3), or risk of composite outcome (relative risk, 2.3; 95% CI, 0.8–5.4) compared with planned delivery at ≥38 weeks’ gestation.
Conclusion
Infants with gastroschisis who were delivered after planned induction or planned delivery at 36-37 weeks’ gestation did not have significantly better neonatal outcomes than planned vaginal delivery after spontaneous onset of labor and planned delivery at ≥38 weeks’ gestation.
Despite numerous studies, the optimal mode and timing of delivery for fetuses with gastroschisis remains uncertain. A major limitation of existing studies is that they have examined differences in newborn infant outcomes according to infants’ actual mode and timing of delivery rather than what was planned . Although data on outcomes according to actual mode and timing of delivery are valuable for informing the newborn infant’s postnatal prognosis, they are less relevant for antenatal delivery planning. For antenatal delivery planning, information is needed on newborn infant outcomes according to planned mode of delivery because this best reflects the information available at the time that obstetrics decision-making occurs (often ≤32-34 weeks’ gestation). A planned vaginal delivery may result in a successful vaginal delivery, but it may also result in an emergency intrapartum cesarean delivery. Because it is unknown at the time of delivery planning which delivery will occur for a given pregnancy, information on the risks and benefits that are associated with a planned vaginal delivery must reflect newborn infant outcomes after either of these potential events. Examining outcomes according to planned rather than actual mode and timing of delivery thus best mimics the “intention-to-treat” analysis of a randomized trial.
In this study, our goal was to compare neonatal outcomes according to planned mode and timing of delivery in pregnancies with gastroschisis. Our primary objective was to determine whether planned vaginal delivery after induction of labor improves neonatal outcome at birth compared with planned vaginal delivery after onset of spontaneous labor. Our secondary objective was to determine whether planned timing of delivery at earlier gestational ages (36-37 weeks’ gestation) improves neonatal outcome compared with planned delivery ≥38 weeks’ gestation. We tested the hypotheses that outcomes would be improved after planned induction of labor and planned birth at 36-37 weeks’ gestation.
Materials and Methods
Data source
We used data from the Canadian Pediatric Surgery Network (CAPSNet) Database, which is an ongoing prospective national database that contains data on all fetuses who are diagnosed with gastroschisis or congenital diaphragmatic hernia at all tertiary-level perinatal centers in Canada (n = 16). The CAPSNet data registry has been described in detail elsewhere. Briefly, data are abstracted from maternal and infant charts by trained research assistants, deidentified, and sent to a centralized repository that is managed by a multidisciplinary and geographically representative steering committee. CAPSNet data collection was approved by the institutional review boards of each perinatal center; this study was approved by the Research Ethics Board of the University of British Columbia/British Columbia Children’s & Women’s Hospital (H13-03017).
Study population
All pregnancies with confirmed gastroschisis that were admitted to a CAPSNet center between May 1, 2005, and May 30, 2013, were eligible for inclusion. Infants were excluded if they had additional malformations or if they were born at <32 weeks’ gestation (because information on planned mode of delivery at 32 weeks’ gestation would not be available for these infants). Multiple births and pregnancies with elective terminations were also excluded.
Delivery planning
Planned mode of delivery at 32 weeks’ gestation is classified in the CAPSNet data registry as either (1) planned vaginal delivery after spontaneous onset of labor, (2) planned induction of labor, (3) planned cesarean delivery, or (4) other or unspecified mode of delivery. Planned timing of delivery at 32 weeks’ gestation is classified in the CAPSNet data registry as <36 + 0, 36 + 0 to 37 + 6, or ≥38 + 0 weeks’ gestation. These variables indicate the intended delivery plan as of 32 weeks’ gestation, not what actually happened during delivery. As such, the “planned vaginal delivery after spontaneous onset of labor” group would include women who ultimately delivered by induction or cesarean should an indication develop. The characteristics of women with a planned cesarean delivery or with other, unknown, or no plans are presented for comparison but were excluded from formal analyses. We also did not analyze deliveries that were planned for <36 + 0 weeks’ gestation, because these likely represented cases in which there was already a medical indication for delivery by 32 weeks’ gestation.
Outcomes
Our study outcomes were neonatal length of stay in days (primary outcome), duration of total parenteral nutrition (TPN) in days, days until the infant’s first enteral feed, and number of postoperative ventilation days. We also created a composite of adverse events that included the occurrence of any of the following events: stillbirth, newborn infant death, or prolonged exclusive TPN (≥6 weeks until first enteral feed).
Confounders
Potential confounders included maternal comorbidities, year of admission, and hospital center. We created a composite maternal comorbidity variable that included women who reported the use of antihypertensives or were diagnosed with gestational or preexisting diabetes mellitus. We grouped all centers with observations of ≤20 into 1 small center category to ensure that there were sufficient numbers to adjust by center.
Statistical analysis
Our primary analysis compared neonatal outcomes according to planned induction vs planned vaginal delivery after the onset of spontaneous labor. Our secondary analysis compared planned early delivery at 36-37 completed weeks vs ≥38 weeks’ gestation. We did not examine women with a planned cesarean delivery or a planned delivery at <36 weeks’ gestation because the numbers in these groups were small and likely reflected higher risk cases. Because of highly skewed continuous outcomes (eg, number of days on TPN), we compared medians (50th percentiles) rather than means using multivariable quantile regression. Risk ratios (rather than odds ratios) with 95% confidence intervals (CIs) were calculated directly from logistic regression using the margins command for the composite variable because the outcome was common. Analyses were performed with the use of Stata/SE software (version 12.1; StataCorp, College Station, TX).
Multivariable models
Variation in patient characteristics or neonatal care between centers could produce variation in newborn infant outcomes, which made it appropriate to adjust for center as a confounder. However, adjustment for center would be an over-adjustment if it were a strong predictor of delivery planning. We therefore presented 2 models: the first adjusted for maternal comorbidities and calendar year (model 1), and the second additionally adjusted for hospital center (model 2).
Sensitivity analysis
Infants who died in hospital may have had short lengths of stay or days on TPN. To avoid misinterpretation of short stays after infant death as a positive outcome, we conducted a sensitivity analysis in which we assigned the highest reported value of an outcome to babies who died. For example, the longest recorded length of time on TPN in the cohort was 603 days. Thus, every infant who died during hospital stay was assigned a TPN value of 603 days in our sensitivity analysis. This allowed us to examine the effect of delivery plan without excluding infants with the worst-case outcomes.
Results
We identified a total of 829 confirmed infants with gastroschisis who were delivered between May 1, 2005, and May 30, 2013, in the CAPSNet database. Excluding 16 deliveries at <32 weeks’ gestation, 104 infants with other reported malformations and 22 elective terminations/unknown outcomes left 687 infants. In this final cohort, 519 pregnancies (76%) had a documented plan for mode of delivery. Of the 280 deliveries with a planned induction or planned cesarean delivery, 270 pregnancies (96%) had documentation on the gestational age for which the procedure was planned.
Baseline characteristics of the cohort according to delivery plan are given in Table 1 and described here. There were 190 planned vaginal deliveries (28%) after spontaneous onset of labor, 49 planned cesarean deliveries (7%), 280 planned inductions (41%), and 168 pregnancies (24%) with other, unknown, or no plans. Among the 270 planned inductions with a documented gestational age for the planned procedure, 8 women (3%) planned to deliver at <36 weeks’ gestation; 193 women (71%) planned to deliver at 36-37 completed weeks’ gestation, and 69 women (26%) planned to deliver at ≥38 weeks’ gestation.
| Characteristics | Planned mode (n = 687) | Planned timing a (n = 270) | |||||
|---|---|---|---|---|---|---|---|
| Spontaneous labor | Cesarean delivery | Induction | Other, no plan, unknown | ≤35 wk b | 36-37 wk | ≥38 wk | |
| n | 190 | 49 | 280 | 168 | 8 | 193 | 69 |
| Actual mode of delivery (cesarean), n (%) | 49 (26) | 42 (88) | 79 (29) | 44 (27) | 2 (25) | 53 (28) | 17 (25) |
| Infant sex (male), n (%) | 110 (58) | 21 (44) | 148 (54) | 79 (49) | 4 (50) | 102 (54) | 39 (57) |
| Maternal age, y c,d | 23 ± 4.3 | 26 ± 5.5 | 23 ± 4.9 | 23 ± 4.7 | 23 ± 3.8 | 23 ± 5.1 | 24 ± 4.8 |
| Gravidity, n (%) | |||||||
| 1 | 107 (56) | 18 (37) | 153 (55) | 91 (55) | 3 (38) | 107 (55) | 38 (55) |
| 2 | 41 (21) | 10 (20) | 69 (24) | 41 (24) | 3 (38) | 48 (25) | 13 (19) |
| ≥3 | 42 (22) | 21 (43) | 58 (21) | 36 (21) | 2 (25) | 38 (20) | 18 (26) |
| Maternal comorbidities, n (%) e | 5 (3) | 2 (4) | 7 (3) | 2 (1) | 0 (0) | 5 (3) | 2 (3) |
| Gestational age at birth, wk f | 36 [35–37] | 37 [35–38] | 36 [35–37] | 36 [35-37] | 35 [35–35] | 36 [35–37] | 37 [36–38] |
| Birthweight, g d | 2556 ± 437 | 2568 ± 572 | 2595 ± 494 | 2525 ± 474 | 2116 ± 333 | 2564 ± 456 | 2732 ± 566 |
| Bowel dilation ≥18 mm, n (%) g | 6 (6) | 2 (9) | 17 (11) | 5 (6) | 2 (50) | 12 (12) | 1 (2) |
a Among pregnancies with planned induction
b In completed weeks (eg, 36-37 weeks means 36 + 0 to 37 + 6 weeks)
e Use of antihypertensive medications or diagnosis of diabetes in pregnancy
f Data are given as median [interquartile range]
Women with a planned induction were of comparable age and gravidity with women with a planned vaginal delivery after onset of spontaneous labor (23 years old; 56% with a gravidity of 1). Women in the planned induction group also had a similar proportion of comorbidities compared with women with planned vaginal delivery after onset of spontaneous labor (3%). The actual mode of delivery was comparable between the planned vaginal delivery after onset of spontaneous labor and the planned induction groups (actual cesarean delivery proportions of 26% vs 29%, respectively); in both groups, the median gestational age at delivery was 36 weeks.
We graphed planned mode and timing of delivery by hospital center to determine the extent to which between-site practice variation accounted for the variation in delivery plans ( Figures 1 and 2 ). There was considerable variation in plans by center, which suggests that the treatment plan was not based solely on patient characteristics or presumed severity of gastroschisis. Hospital names and error bars were withheld to protect anonymity.
The results from the multivariable quantile and logistic regressions comparing various planned modes of delivery are presented in Table 2 . Compared with planned vaginal delivery after onset of spontaneous labor, planned induction of labor was not associated with significant differences in length of stay, duration of TPN, or time until first enteral feed in either the crude or adjusted models. In the crude analysis, planned induction was associated with a median reduction of 1 day in postoperation ventilation days (95% CI, –1.4 to –0.6). However, this association was no longer significant after adjustment for hospital center and other covariates (model 2).
| Neonatal outcomes | Planned spontaneous labor | Planned induction of labor | Median difference, planned induction vs planned spontaneous labor (95% CI) | ||
|---|---|---|---|---|---|
| Crude | Model 1 a | Model 2 b | |||
| Length of stay, d c | 37.5 [26.5–59] | 36 [26–63] | –2 (–7.6 to 3.6) | –1.6 (–7.5 to 4.2) | –2.6 (–9.9 to 4.8) |
| Time on total parenteral nutrition, d c | 27 [19–42] | 27 [19–48] | 0 (–4.3 to 4.3) | –0.16 (–4.6 to 4.3) | –0.2 (–6.4 to 6.0) |
| Time until first enteral feed, d c | 13 [10–18] | 13 [10–19] | 0 (–1.6 to 1.6) | 0.13 (–1.4 to 1.6) | 1 (–1.1 to 3.1) |
| Postoperative ventilation days c | 4 [2–5] | 3 [2–5] | –1 (–1.4 to –0.6) | –1 (–1.4 to –0.6) | 0.0 (–0.5 to 0.5) |
| Prolonged total parenteral nutrition or death, n (%) | 11 (5.8) | 26 (9.3) | 1.6 (0.5–2.7) d | 1.5 (0.4–2.7) d | 1.7 (0.1–3.2) d |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
