Background
Women who seek vaginal birth after cesarean delivery may find limited in-hospital options. Increasing numbers of women in the United States are delivering by vaginal birth after cesarean delivery out-of-hospital. Little is known about neonatal outcomes among those who deliver by vaginal birth after cesarean delivery in- vs out-of-hospital.
Objective
The purpose of this study was to compare neonatal outcomes between women who deliver via vaginal birth after cesarean delivery in-hospital vs out-of-hospital (home and freestanding birth center).
Study Design
We conducted a retrospective cohort study using 2007–2010 linked United States birth and death records to compare singleton, term, vertex, nonanomolous, and liveborn neonates who delivered by vaginal birth after cesarean delivery in- or out-of-hospital. Descriptive statistics and multivariate regression analyses were conducted to estimate unadjusted, absolute, and relative birth-setting risk differences. Analyses were stratified by parity and history of vaginal birth. Sensitivity analyses that involved 3 transfer status scenarios were conducted.
Results
Of women in the United States with a history of cesarean delivery (n=1,138,813), only a small proportion delivered by vaginal birth after cesarean delivery with the subsequent pregnancy (n=109,970; 9.65%). The proportion of home vaginal birth after cesarean delivery births increased from 1.78–2.45%. A pattern of increased neonatal morbidity was noted in unadjusted analysis (neonatal seizures, Apgar score <7 or <4, neonatal seizures), with higher morbidity noted in the out-of-hospital setting (neonatal seizures, 23 [0.02%] vs 6 [0.19%; P <.001]; Apgar score <7, 2859 [2.68%] vs 139 [4.42%; P <.001; Apgar score <4, 431 [0.4%] vs 23 [0.73; P =.01]). A similar, but nonsignificant, pattern of increased risk was observed for neonatal death and ventilator support among those neonates who were born in the out-of-hospital setting. Multivariate regression estimated that neonates who were born in an out-of-hospital setting had higher odds of poor outcomes (neonatal seizures [adjusted odds ratio, 8.53; 95% confidence interval, 2.87–25.4); Apgar score <7 [adjusted odds ratio, 1.62; 95% confidence interval, 1.35–1.96]; Apgar score <4 [adjusted odds ratio, 1.77; 95% confidence interval, 1.12–2.79]). Although the odds of neonatal death (adjusted odds ratio, 2.1; 95% confidence interval, 0.73–6.05; P =.18) and ventilator support (adjusted odds ratio, 1.36; 95% confidence interval, 0.75–2.46) appeared to be increased in out-of-hospital settings, findings did not reach statistical significance. Women birthing their second child by vaginal birth after cesarean delivery in out-of-hospital settings had higher odds of neonatal morbidity and death compared with women of higher parity. Women who had not birthed vaginally prior to out-of-hospital vaginal birth after cesarean delivery had higher odds of neonatal morbidity and mortality compared with women who had birthed vaginally prior to out-of-hospital vaginal birth after cesarean delivery. Sensitivity analyses generated distributions of plausible alternative estimates by outcome.
Conclusion
Fewer than 1 in 10 women in the United States with a previous cesarean delivery delivered by vaginal birth after cesarean delivery in any setting, and increasing proportions of these women delivered in an out-of-hospital setting. Adverse outcomes were more frequent for neonates who were born in an out-of-hospital setting, with risk concentrated among women birthing their second child and women without a history of vaginal birth. This information urgently signals the need to increase availability of in-hospital vaginal birth after cesarean delivery and suggests that there may be benefit associated with increasing options that support physiologic birth and may prevent primary cesarean delivery safely. Results may inform evidence-based recommendations for birthplace among women who seek vaginal birth after cesarean delivery.
Reducing the overall cesarean delivery (CD) rate is a national priority, and vaginal birth after cesarean delivery (VBAC) is one proven strategy for reaching this goal. National guidelines recommend that women with one previous CD have the option for trial of labor after cesarean (TOLAC) ; however, not all hospitals allow TOLAC. Increasing numbers of women in the United States seek out-of-hospital birth, including those who seek VBAC. Although it is well-established that in-hospital VBAC carries a small but significantly increased risk for neonatal death and evidence suggests a small but significant relative increase in intrapartum fetal and neonatal death that is associated with out-of-hospital birth, less is known about neonatal outcomes for out-of-hospital VBAC.
A limited body of research addresses questions of VBAC outcomes by birth setting. One German study compared women who planned their second birth at home (n=24,545) and found that women with previous CD had worse neonatal outcomes. A prospective British study (n=1436) found a nonsignificant pattern that suggested increased neonatal morbidity and mortality rates in out-of-hospital VBAC, as compared with hospital VBAC. Three studies assessed US neonatal outcomes for out-of-hospital TOLAC. A retrospective study found 1 stillbirth among 57 women who planned TOLAC at home. A second prospective study of parous women who delivered in a birth center found increased neonatal mortality rates among women with a previous CD (n=1453). The third study found significantly elevated rates of poor neonatal outcomes among women who sought TOLAC in a large national registry of planned home births to parous women (n=13,144). Despite the increasing frequency of out-of-hospital VBAC, to our knowledge there are no US studies that have compared VBAC outcomes between hospital and out-of-hospital settings.
With the use of vital statistics data, the purpose of this study was to (1) characterize US women who completed VBACs in home, birth center, and hospital settings, (2) compare neonatal outcomes by birth setting (in vs out-of-hospital), and (3) apply advanced statistical modeling techniques to account for intended vs. actual delivery setting.
Materials and Methods
This was a retrospective cohort study that used 2007–2010 US birth and death records. An infant death file was linked to the corresponding birth certificate in year of birth, regardless of year of death. Cohort-linked birth/infant death files are available through 2010.
Records were excluded if the woman (1) did not have a CD history, (2) delivered by repeat CD, (3) had a multiple gestation, (4) delivered preterm (<37 weeks) or extremely postterm (>43 weeks), (5) had a breech fetus, (6) had a fetus with congenital anomalies, (7) experienced stillbirth (fetal death), or (8) if there was obvious data miscoding (eg, both nulliparity and CD history were marked; this removed <2% of the sample). Thus, our study compares outcomes for women with singleton, term, vertex, nonanomolous, liveborn neonates who were delivered by VBAC in or out-of-hospital (n=109,970). Our key exposure variable was birth setting , which was defined as completed home or freestanding birth center birth vs completed hospital birth.
We analyzed several neonatal outcomes that included neonatal death (death within first 28 days), depressed 5-minute Apgar score (both <7 and <4), neonatal seizures, infant ventilator support >6 hours, neonatal intensive care unit (NICU) admission, and birth injury (skeletal fracture, peripheral nerve injury, and/or soft tissue/solid organ hemorrhage that required intervention).
We calculated descriptive statistics for demographic and outcome variables, comparing completed VBACs in and out-of-hospital. All statistical tests were performed with the use of the chi-square or Fisher’s exact test; statistical significance was P < .05. Logistic regressions were performed with the use of multivariable models that controlled for demographic and prenatal care characteristics. Models compared outcomes in the out-of-hospital to in-hospital (the referent) categories. For regression analyses and descriptive statistics of outcomes, we combined completed birth center and completed home VBACs into a single out-of-hospital category, a decision scientifically and practically motivated. We posit that, during obstetric emergencies, proximity to medical intervention is often similar for those at home and at free-standing birth centers. Practically, given the rare exposure (out-of-hospital birth), exposure subgroup (VBAC), and very rare outcomes (neonatal death), small cell sizes and limited statistical power precluded us from analyzing out-of-hospital settings separately.
Covariates included in the models were maternal race/ethnicity (white [referent], black, Hispanic, Asian/Pacific Islander, and American Indian/Alaskan Native), parity (1 [referent], 2, and ≥3), tobacco use (no/yes), maternal age (≤20, 21–34 [referent], and ≥35 years old), maternal education (less than high school education, high school only [referent], undergraduate college of any duration, and graduate school of any type), number of previous CDs (1 [referent], 2, ≥3), prenatal care after first trimester (no/yes), prenatal care adequacy as identified by the Kotelchuck Index (adequate plus [referent], adequate, intermediate, and inadequate), and weight gain >40 pounds (no/yes). It is valuable to consider measures of both relative and absolute risk when comparing outcomes by birth setting ; thus, absolute birth setting risk differences were calculated for each regression. All analyses were conducted with Stata software (version 12; StataCorp, College Station, TX).
Because both parity and history of vaginal birth are important predictors of perinatal outcomes for women who choose TOLAC, we conducted stratified regression analyses examining the association between birth setting and each outcome. Models were fit as described earlier, stratified by parity (1 previous birth, 2, and ≥3) and history of vaginal birth (no/yes).
Because it is not possible to disaggregate women who intended out-of-hospital birth but ultimately delivered in-hospital from women who planned and birthed in-hospital, we conducted sensitivity analyses to estimate effects of this misclassification bias. We used random resampling to estimate how odds of neonatal death and Apgar score <4 might be affected if transfers were identifiable. Because Oregon birth certificates enable disaggregation by intended birth setting, Oregon vital statistics were used to estimate that 0.44% of observed hospital births were likely out-of-hospital–to-hospital transfers. This rate informed calculation of the number of transfers from the out- to in-hospital setting; records equaling this number were chosen randomly under different scenarios and placed in the out-of-hospital birth category to assess outcome odds ratio sensitivity reclassification.
Three scenarios were explored: (1) hospital births with the outcome of interest were given the lowest probability of being reclassified, (2) hospital births with the outcome of interest were given a high probability of being reclassified (modeling presumed transfers had >5 times greater neonatal death risk and 4 times greater risk of an Apgar score of <4), (3) transfer status was assigned randomly to 470 hospital records (0.44% of hospital births sample). After reassignments, the changed outcome odds ratio was computed with the use of the same logistic regression approach. For both extreme scenarios (1 and 2), a single random reclassification was made for each outcome. To explore odds ratio variations that resulted from different random reassignments, bootstrap analysis of the permutation distribution of random transfer status was performed for 470 new births 50,000 times for the neonatal death outcome and 100,000 times for the Apgar score <4 outcome. This repeatedly and randomly scrambled which 470 births were reclassified.
Results
Demographics
Women who delivered by VBAC out-of-hospital were significantly more likely to be white and older (≥35 years) compared with women who delivered by VBAC in-hospital ( P <.001; Table 1 ). Those who completed VBAC out-of-hospital were less likely to smoke, to initiate prenatal care in the first trimester, to meet criteria for adequate or adequate plus prenatal care according to the Kotelchuck index, and also achieved different levels of education than those who completed VBAC in-hospital (P<.001).
| Variable | Hospital, n (%) | Home birth | Birth center | Out-of-hospital location (home+birth center) | |||
|---|---|---|---|---|---|---|---|
| n (%) | P value | n (%) | P value | n (%) | P value | ||
| N | 106,823 (97.14) | 2352 (2.14) | <.001 | 795 (0.72) | <.001 | 3147 (2.86) | <.001 |
| Race/ethnicity | <.001 a | <.001 a | <.001 a | ||||
| White | 53,226 (49.83) | 2094 (89.03) | 634 (79.75) | 2728 (86.69) | |||
| Black | 15,396 (14.41) | 76 (3.23) | 18 (2.26) | 94 (2.99) | |||
| Hispanic | 31,030 (29.05) | 141 (5.99) | 132 (16.60) | 273 (8.67) | |||
| Asian | 6,106 (5.72) | 34 (1.45) | 8 (1.01) | 42 (1.33) | |||
| American Indian/Alaska Native | 1,065 (1.00) | 7 (0.30) | 3 (0.38) | 10 (0.32) | |||
| Parity | <.001 | <.001 | <.001 | ||||
| 1 | 33,813 (31.65) | 483 (20.54) | 165 (20.75) | 648 (20.59) | |||
| 2 | 28,423 (26.61) | 495 (21.05) | 165 (20.75) | 660 (20.97) | |||
| ≥3 | 43,237 (40.48) | 1331 (56.59) | 454 (57.11) | 1785 (56.72) | |||
| Tobacco use | 7,842 (7.34) | 12 (0.51) | <.001 | 13 (1.64) | <.001 | 25 (0.79) | <.001 |
| Maternal age, y | <.001 | <.001 | <.001 | ||||
| ≤20 | 2,302 (2.15) | 2 (0.09) | 3 (0.38) | 5 (0.16) | |||
| 21–34 | 81,386 (76.19) | 1623 (69.01) | 569 (71.57) | 2192 (69.65) | |||
| ≥35 | 23,135 (21.66) | 727 (30.91) | 223 (28.05) | 950 (30.19) | |||
| Education | <.001 a | <.001 a | <.001 a | ||||
| Not completed high school | 9,044 (8.47) | 421 (17.90) | 252 (31.70) | 673 (21.39) | |||
| High school | 41,668 (39.01) | 404 (17.18) | 181 (22.77) | 585 (18.59) | |||
| College | 45,093 (42.21) | 1289 (54.80) | 298 (37.48) | 1587 (50.43) | |||
| Graduate school | 9,744 (9.12) | 222 (9.44) | 61 (7.67) | 283 (8.99) | |||
| Previous cesarean deliveries, n | .105 a | .033 a | .137 a | ||||
| 1 | 94,478 (88.44) | 2061 (87.63) | 730 (91.82) | 2791 (88.69) | |||
| 2 | 7,916 (7.41) | 187 (7.95) | 45 (5.66) | 232 (7.37) | |||
| ≥3 | 1,676 (1.57) | 36 (1.53) | 6 (0.75) | 42 (1.33) | |||
| History of vaginal birth | 67,217 (62.92) | 1780 (75.68) | <.001 | 612 (76.98) | <.001 | 2392 (76.01) | <.001 |
| Prenatal care initiation after 1st trimester | 31,415 (29.41) | 1143 (48.60) | <.001 | 441 (55.47) | <.001 | 1584 (50.33) | <.001 |
| Kotelchuck Index | <.001 a | <.001 a | <.001 a | ||||
| Adequate Plus | 21,011 (19.67) | 159 (6.76) | 49 (6.16) | 208 (6.61) | |||
| Adequate | 35,629 (33.35) | 744 (31.63) | 254 (31.95) | 998 (31.71) | |||
| Intermediate | 20,748 (19.42) | 594 (25.26) | 128 (16.10) | 722 (22.94) | |||
| Inadequate | 22,486 (21.05) | 800 (34.01) | 331 (41.64) | 1131 (35.94) | |||
| Weight gain > 40 lb | 20,477 (19.17) | 431 (18.32) | .252 | 138 (17.36) | .115 | 569 (18.08) | .072 |
More than one-half of VBAC births out-of-hospital were among women with parity ≥3 (out-of-hospital 56.72% vs hospital 40.48%; P <.001). In contrast, women with in-hospital VBAC were more frequently of lower parity (eg, parity 1: hospital, 31.65%; out-of-hospital, 20.59; P <.001). Most women in all settings had a history of 1 previous CD (out-of-hospital 88.69% vs in-hospital 88.44%; P =.14). A higher proportion of women who delivered via VBAC out-of-hospital had a history of previous vaginal birth than women who delivered via VBAC in-hospital (out-of-hospital 76.01% vs in-hospital 62.92%; P <.001).
A small proportion of the total number of US women with a history of CD (n=1,138,813) delivered by VBAC in any setting (n=109,970; 9.65%); a large majority of these delivered in-hospital (n=106,823; 97.14%; Table 1 ). Between 2007 and 2010, the proportion of completed VBAC births in the home setting increased from 1.78–2.45% ( Table 2 ).
| Year | Hospital, n (%) | Home birth, n (%) | Birth center, n (%) | Out-of-hospital setting, n (%) |
|---|---|---|---|---|
| 2007 | 21,505 (96.79) | 395 (1.78) | 150 (0.68) | 545 (2.47) |
| 2008 | 25,273 (96.65) | 514 (1.97) | 197 (0.75) | 711 (2.74) |
| 2009 | 27,062 (96.51) | 604 (2.15) | 209 (0.75) | 813 (2.92) |
| 2010 | 32,983 (96.21) | 839 (2.45) | 239 (0.70) | 1078 (3.16) |
Outcomes
A pattern of increased neonatal morbidity was noted in unadjusted analysis of outcomes (Apgar score <7 or <4, neonatal seizures) by birth setting, with higher morbidity in the out-of-hospital setting ( Table 3 ). For example, Apgar score <7 was observed in 4.4% of out-of-hospital VBACs and 2.7% of in-hospital VBACs ( P <.001). A similar, but nonsignificant, pattern of increased risk was observed for neonatal death and ventilator support in the out-of-hospital setting. NICU admission was significantly lower among neonates who were delivered via VBAC out-of-hospital as compared with in-hospital (1.1% vs 3.1%; P <.001); birth injury was lower, but nonsignificant, out-of-hospital compared with in-hospital.
| Variable | Hospital, n (%) | Out-of-hospital setting, n (%) | P value |
|---|---|---|---|
| Neonatal death | 84 (0.08) | 4 (0.13) | .326 |
| Apgar score <7 | 2859 (2.68) | 139 (4.42) | <.001 |
| Apgar score <4 | 431 (0.40) | 23 (0.73) | .010 |
| Neonatal seizures | 23 (0.02) | 6 (0.19) | <.001 |
| Ventilator support | 309 (0.29) | 12 (0.38) | .315 |
| Neonatal intensive care unit admission | 3292 (3.10) | 35 (1.11) | <.001 |
| Birth injury | 109 (0.10) | 1 (0.03) | .382 |
In multivariable regression analyses that were adjusted for confounders, associations between birth setting and rare adverse outcomes were estimated with low statistical precision because of small cell sizes. Neonates who were born in the out-of-hospital setting were >8 times as likely to have neonatal seizures and almost twice as likely to have lower Apgar scores compared with babies delivered in-hospital (neonatal seizures: adjusted odds ratio [aOR] 8.53; 95% confidence interval [CI], 2.87–25.4); Apgar score <7 (aOR, 1.62; 95% CI, 1.35–1.96); Apgar score <4 (aOR 1.77; 95% CI, 1.12–2.79; Table 4 ). Although odds of neonatal death (aOR, 2.1; 95% CI, 0.73–6.05; P =.18) and ventilator support (aOR, 1.36; 95% CI, 0.75–2.46; P =.31) were increased in out-of-hospital settings, findings did not reach statistical significance. Decreased NICU admission and birth injury among out-of-hospital VBACs persisted after regression analysis: NICU admission (aOR, 0.4; 95% CI, 0.29–0.57); birth injury (aOR, 0.78; 95% CI, 0.58–1.04).
| Variable | Adjusted for covariates | ||
|---|---|---|---|
| Adjusted odds ratio (95% confidence interval) | Adjusted risk difference b (95% confidence interval) | P value | |
| Neonatal death | 2.10 (0.73–6.05) | 0.0005 (–0.0002–0.0013) | .176 |
| Apgar score <7 | 1.62 (1.35–1.96) | 0.0129 (0.0079–0.0179) | <.001 |
| Apgar score <4 | 1.77 (1.12–2.79) | 0.0022 (0.0004–0.0041) | .016 |
| Neonatal seizures | 8.53 (2.87–25.4) | 0.0005 (0.0002–0.0009) | .003 |
| Ventilator support | 1.36 (0.75–2.46) | 0.001 (–0.0009–0.0028) | .311 |
| Neonatal intensive care unit admission | 0.40 (0.29–0.57) | –0.0262 (–0.0363––0.0161) | <.001 |
| Birth injury | 0.78 (0.58–1.04) | –0.0002 (–0.0004–0.0000) | .089 |
a Models controlled for maternal race, maternal age, maternal education (high school education as referent), Kotelchuck Index (adequate plus as referent), parity, number of previous cesarean deliveries, and maternal weight gain ≥40 lbs
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
