Short and long interpregnancy intervals have been shown to be associated with increased risk for adverse maternal and perinatal outcomes.1–5 However, the effects of birth interval on the safety and efficacy of vaginal birth after cesarean delivery (VBAC) are less well characterized because of study design constraints and the small number of publications.6–11
We hypothesized that short interpregnancy intervals may lead to altered wound healing and an increased risk of uterine rupture in patients who attempt a vaginal birth after cesarean. Our hypothesis is based on previous observational studies that suggest an association between short birth interval and increased adverse perinatal outcomes1,2,6–10,12 and wound-healing research that indicates that uterine smooth muscle tissue repair evolves over several months.13–16 Because myometrial tissue regenerates slowly, cesarean incisions heal predominantly from a proliferation of fibroblasts and resultant replacement of myometrium with connective tissue.14,16,17 Importantly, there is radiographic and hysteroscopic evidence that cesarean scar development is incomplete as long as 6 or 12 months postoperatively.13,15
Using the same rationale, we hypothesized that a short interpregnancy interval is independently associated with higher rates of two secondary outcomes: a composite of delivery-related major maternal morbidity and blood transfusion requirement. Although less biologically plausible, observational research also suggests an adverse effect of long interpregnancy interval on maternal outcome.1–3,12 Thus, we also assessed long interpregnancy interval as a risk factor for uterine rupture and other VBAC complications. Our specific aim for this study was to investigate whether a short or long interpregnancy interval is associated with uterine rupture and other major maternal morbidity in patients who attempt a VBAC.
MATERIALS AND METHODS
This study is a secondary analysis of a multi-center, record-based, retrospective cohort study of pregnant women with at least one prior cesarean. The detailed methods of the initial cohort study have been published previously, but a brief description follows.18,19 Seventeen hospitals in the Northeastern United States participated, including six university hospitals, five medical teaching community hospitals, and six nonteaching community hospitals. The study was approved by all local institutional review boards before its initiation. Patients were identified by a previously validated International Classification of Disease, 9th Revision, code search using the code for “previous cesarean delivery, delivered.” More than 25,000 patients who delivered between 1995 and 2000 at participating centers made up the cohort. Any patient was excluded from the initial cohort if she had a prior classic or unknown type of uterine incision or if she had a fetus with a major anomaly. Trained research nurses abstracted information from the medical records using structured, closed-ended data forms.
This secondary analysis includes only patients who elected to attempt a VBAC. The exposures for this analysis are short and long interpregnancy intervals, defined by the number of months between the immediate prior delivery and the subsequent conception. We calculated the conception date from the estimated due date and validated the conception date estimate with the delivery date, gestational age, and birth weight variables. Of note, the interpregnancy interval was calculated by using the delivery immediately before the study pregnancy, regardless of the delivery mode of that prior delivery. Thus, a proportion of patients (49.9%) had a vaginal delivery before the study pregnancy because they achieved VBAC status from a cesarean delivery performed in an earlier pregnancy. We assessed the effect of immediate prior delivery mode on the outcomes by using stratified χ2 and multivariable analyses because it is plausible that risk may be higher with cesarean delivery as the prior delivery mode than with vaginal delivery. Because the cohort database included only the year of the prior delivery, we conservatively set the prior delivery date to be January 1 of the delivery year for each patient. This date approximation forces misclassification into a single direction, assuring that misclassification of short interpregnancy interval will incorrectly assign a proportion of truly short interpregnancy interval patients to the normal interval group. Consequently, in this case, misclassification makes the “normal interval” group more similar to the short interval group and should slightly underestimate the short interval effect, biasing results toward the null hypothesis. To validate this assumption, we performed a sensitivity analysis, setting the delivery date at July 1, the mid-year point. This mid-year estimate should minimize the amount of error in delivery date on average and classify more truly short interval patients correctly but will allow misclassification in both directions and, thus, misclassify many truly normal interval patients as short interval patients.
Based on prior research, we analyzed short interpregnancy interval by using cutoffs of less than 6, less than 12, and less than 18 months and defined long interpregnancy interval as 60 months or more between the prior delivery and the subsequent conception. We used two different approaches to define exposed and unexposed study groups. First, we completed the analysis by using a series of dichotomous exposure variables, dividing the cohort into normal or abnormal interval groups based on the time cutoffs described above. This method was used to maximize the study’s power to detect group differences. Second, we completed the analysis with a five-category interpregnancy interval variable with the same four cutoff points, defining the 18–59 month group as reference. Although this method decreases study power by dividing the cohort into smaller groups, it allows us to compare among various severity levels of abnormal birth interval and to determine whether risk is confined to the most extreme intervals or is related to interval in a dose-dependent fashion. This second method also enables us to unmask whether risk in less stringently defined dichotomous “short interval” variables (ie, less than 12 or less than 18 months) is attributable only to patients with less-than-6-month intervals, falsely elevating risk between 6 and 18 months.
The outcomes for this analysis were strictly defined a priori. Uterine rupture, the primary outcome for this study, was defined as a uterine scar separation determined at laparotomy that was preceded by a nonreassuring fetal heart rate pattern, maternal signs or symptoms of acute blood loss, or hemoperitoneum. This definition excludes asymptomatic uterine dehiscence.18,19 The two secondary outcomes included a composite of maternal major morbidity and maternal blood transfusion requirement. The dichotomous composite maternal morbidity variable included uterine rupture; bladder, ureter, or bowel injury; and uterine artery laceration. Patients were categorized as having morbidity if they had one or more of the events. Blood transfusion requirement was also a dichotomous variable and was extracted directly from the medical record.
Our statistical analysis consisted of descriptive statistics, univariable and stratified analyses, and finally, multivariable logistic regression. In the unadjusted analysis, we used χ2 or Fisher exact tests for categorical data and unpaired t tests or analysis of variance for continuous variables. We used the results of our unadjusted and stratified analyses (P<.2) to guide selection of candidate variables for inclusion in the logistic regression models but also included covariates that previously were identified to be biologically important. Most data were complete; but four variables, namely maternal hemoglobin, birth weight, maternal race, and labor type, had missing data rates of 14% (n=1,916), 14.5% (n=1,930), 2.8% (n=370), and 1.9% (n=249), respectively. Missing data points for these variables were imputed from covariates for use in regression modeling. However, birth weight was not retained as a significant confounder in any model, and because of the higher missing data rate, maternal hemoglobin was modeled both with and without imputation, which produced no variation in the interpregnancy interval adjusted odds ratios for morbidity. The final logistic regression model for each primary outcome includes the most accurate and parsimonious group of variables that describe the association between interpregnancy interval and outcome. The final regression equation is a result of sequentially adding or removing candidate variables, testing differences in the accuracy of hierarchical models with the likelihood ratio test or Wald test. We assessed each final logistic regression model for its effectiveness in describing the primary outcome by using the Hosmer-Lemeshow goodness-of-fit summary statistic.20,21
In this cohort of 25,005 pregnant women with prior cesareans, 55% elected to attempt a VBAC. Among the 13,706 patients who elected a VBAC trial, 2.7% did not have interpregnancy interval data, leaving 13,331 for this analysis. There were 128 cases of uterine rupture, yielding a rupture rate of 0.9% in patients who attempted VBAC. Patients with a short interpregnancy interval (18 months or less) attempted a VBAC trial approximately 10% more often than patients with a longer interpregnancy interval of more than 18 months (relative risk 1.09, 95% confidence interval [CI] 1.06–1.12). The rate of short interpregnancy interval, defined as less than 6, 6–11, and 12–17 completed months, was 2.2%, 8.3%, and 13.1%, respectively. The rate of long interpregnancy interval (60 months or more) was 19.7%. These rates of extreme interpregnancy interval are similar to prior published research performed in the United States and Latin America,4,6,8,10,22 but some estimates of short interval rates are slightly higher (eg, rate of less-than-6-month interval=5–10%).2,23–25
In the univariable analysis, patients with extreme interpregnancy intervals—either short or long—delivered smaller neonates on average and were more likely to self-report African-American race, smoke cigarettes, have a preterm delivery, have more than one prior cesarean delivery, undergo a prior cesarean delivery for a maternal indication or malpresentation, have a prior vaginal delivery, and develop preeclampsia compared with patients with a normal interpregnancy interval of 18–59 months (Table 1). The rates of undergoing a prior cesarean delivery for cephalopelvic disproportion and for nonreassuring fetal status were similar across all interpregnancy interval groups. Patients with a short interpregnancy interval had a lower hemoglobin level on average, were younger, less likely to develop gestational diabetes, and more likely to use Medicaid and deliver at a university hospital (Table 1). Patients with a long interpregnancy interval were older, more frequently developed gestational diabetes, and more commonly had chronic hypertension than normal interval patients (Table 1).
As depicted in Table 2, patients with a short interpregnancy interval of less than 6 months who attempted a VBAC had a significant increase in uterine rupture rate, with an absolute risk of 2.7%, which is nearly a threefold increase in rupture rate relative to patients with birth intervals longer than 6 months. This short interval group also had a significant twofold increase in the composite morbidity rate, with an absolute risk of 4.2%. They had a significant threefold increase in the risk of receiving a blood transfusion compared with patients with longer birth intervals. These risk increases persisted when we adjusted for relevant confounders such as prior cesarean number, prior vaginal delivery, delivery gestational age, hospital type, anemia, cigarette use, maternal age, race, parity, and whether labor was induced (Table 2). When we used the less-than-12-month cutoff point to discriminate between normal or abnormal interpregnancy interval, relative risks of uterine rupture, composite morbidity, and blood transfusion remained increased, but the magnitude of the effect size was attenuated, and the associations no longer reached statistical significance. Using the less-than-18-month cutoff further attenuated the magnitude and significance of the relative risks (data not shown). Long interpregnancy interval was not an independent risk factor for uterine rupture, composite morbidity, or blood transfusion (Table 3).
Performing a similar analysis using the multi-category interpregnancy interval variable to assign exposure status, we confirmed that risk increases for uterine rupture, composite morbidity, and blood transfusion due to short interpregnancy interval appear to be confined to the less-than-6-month interval (Table 4). The risks for uterine rupture and transfusion remained threefold higher, and the risk for composite maternal morbidity remained twofold higher than those of patients with a normal interpregnancy interval (18–59 months) when birth interval was stratified by five categories. In addition, there was no evidence that maternal risk is related to short interpregnancy interval in a dose-dependent fashion. All final logistic regression models that we report effectively explain the outcome variable in terms of the covariates as determined by the Hosmer-Lemeshow goodness-of-fit summary statistic (P=0.2–0.9, in which P<.05 indicates a poorly fit model).
Although the delivery mode (cesarean compared with vaginal delivery) of the pregnancy immediately before the study pregnancy was not a significant covariate or effect modifier in the multivariate analyses of interpregnancy interval and morbidity, it is biologically plausible that risk of uterine rupture and other morbid events may be highest in or confined to VBAC patients who most recently had a cesarean delivery. Thus, we analyzed the relationship between short interpregnancy interval (less than 6 months) and maternal morbidity stratified on the immediate prior delivery mode to assess for effect modification not detected, potentially due to type 2 error, in more complex multivariable models. The prior delivery mode-specific relative risks for uterine rupture seem different for cesarean delivery (relative risk 5.1, 95% CI 2.4–10.8) and vaginal delivery (relative risk 0.8, 95% CI 0.1–6.0), but the Mantel-Haenszel test for homogeneity used to assess effect modification is not quite statistically significant (P=.07). There was no evidence for effect modification by immediate prior delivery mode for the association between short interpregnancy interval and composite morbidity (P=.5) or blood transfusion (P=.2).
The rate of failure to achieve a VBAC was approximately 23% and did not vary significantly by interpregnancy interval category, even when controlling for confounders in a multivariable analysis. This finding suggests that the excess morbidity associated with a short or long interpregnancy interval is not mediated by VBAC failure risk.
Lastly, the sensitivity analysis, in which we varied the assumed prior delivery date from January 1 to July 1 of the year of delivery, validated our assumptions. As predicted, the number of patients with a short interpregnancy interval increased, suggesting that using the less stringent assumption results in misclassification of many truly normal birth interval patients as having a short interval. For example, there were 1,402 patients (10.5%) with an interpregnancy interval less than 6 months using the mid-year date assumption compared with 286 (2.2%) in the primary analysis. This sensitivity analysis produced relative risks in the same direction of association but with less magnitude and less statistical significance. For example, the adjusted odds ratio for uterine rupture with exposure to interpregnancy interval less than 6 months was 1.60 (95% CI 0.97–2.63).
In this large cohort study we demonstrated that a short interpregnancy interval of less than 6 months is an independent risk factor for uterine rupture in patients who attempt a vaginal birth after cesarean delivery. Patients with a short interpregnancy interval who attempt VBAC also have an increased risk for other delivery-related major maternal morbid events, such as operative injury and the need for a blood transfusion. These morbidity risks were not mediated by a difference in the risk of having an unsuccessful VBAC attempt. Although even this sizable cohort study is underpowered to definitively assess the modifying effect of the immediate prior delivery mode on the risk of maternal morbidity imparted by short interpregnancy interval, our stratified analyses suggest that uterine rupture risk may be highest in short interpregnancy interval VBAC patients who last had a cesarean delivery compared with those who had last delivered vaginally. In this study, intermediate interpregnancy intervals of 6–18 months do not seem to significantly increase the risk of maternal morbidity associated with a VBAC attempt. Our results also indicate that a long interpregnancy interval of 60 months or greater is not a risk factor for major maternal morbidity in VBAC patients.
Our results confirm the finding of some prior studies that short birth interval is associated with a twofold to fourfold increase in uterine rupture risk.6,7,10 However, our large multi-center cohort represents a significant addition to the literature because our study design enabled us to better characterize this association for clinical application. The large sample size, which includes the largest number of symptomatic uterine rupture cases in publication, enabled us to explore effects of variable short and long interpregnancy intervals while maintaining adequate precision for assessing exposure-outcome associations. The cohort database is robust with information on multiple risk factors, allowing us to control for important confounders. Primary outcomes were strictly defined a priori, limiting information bias related to maternal outcome.
The existing body of literature regarding VBAC and short birth interval is difficult to interpret because of conflicting results, inconsistent exposure and outcome definitions, and study design limitations. Of the five studies that we identified in our systematic search of English publications on the topic of VBAC and birth interval, three reported an increase in risk of uterine rupture associated with short birth interval,6,7,10 one detected no difference in uterine rupture rate,8 and one reported increased risk with short interdelivery interval only among a subgroup of patients with prior preterm cesarean deliveries.9 All of these studies were restricted by sample size, each containing between three and 29 cases of symptomatic uterine rupture.6–10 Two research groups used a case–control design, which did not allow them to estimate incidence rates of uterine rupture or compare relative risks across various interpregnancy interval lengths.7,9 One of the case–control studies included symptomatic uterine rupture (n=23) and dehiscence (n=43) in the primary outcome definition,7 making it difficult to gauge the clinical significance of the outcome, and the other study did not control for confounders in multivariable analysis.9 In the three cohort studies, the investigators restricted their inclusion criteria to allow only patients with one prior cesarean delivery,6,8,10 term delivery,8,10 and no prior vaginal delivery,6,10 limiting the generalizability of the results. These cohort studies defined short birth interval using interdelivery interval (time between delivery dates) instead of interpregnancy interval.6,8,10 This definition is subject to confounding by variable gestational lengths, and although the investigators controlled for the occurrence of postterm delivery10 or birth weight6 in a multivariable analysis, there is potential for residual confounding by other gestational age differences. In a subanalysis of 223 labor induction patients with prior cesarean deliveries, Huang and colleagues8 reported a reduced VBAC success rate in patients with an interdelivery interval less than 19 months compared with women with a longer interval (14% compared with 86%, P<.01). This finding is contrary to ours and, perhaps, could be attributed to the small number of short interval patients (n=7) in their subanalysis. We found no effect of short interpregnancy interval on VBAC failure rate regardless of interpregnancy interval definition or labor type (spontaneous compared with induced).
Although our study has several strengths and advantages, it also has limitations. First, there is potential for misclassification of interpregnancy interval type because our database included the year, but not the date, of the prior delivery. Importantly, our delivery date assumption should actually bias the results to the null hypothesis or provide a conservative estimate of the relative risk if significant bias exists, and yet we still detected a significant association between short interpregnancy interval and VBAC-related adverse outcomes. Further, a sensitivity analysis, in which we varied the assigned delivery date, validated our assumptions. Second, there is potential for residual confounding, particularly for the long interpregnancy interval associations, because these patients have many coexisting time- and age-related risk factors. Third, by design, this study has limited neonatal outcome data. Thus, we cannot juxtapose effects of interpregnancy interval on neonatal morbidity with risks of maternal morbidity among VBAC candidates. Lastly, there is potential for selection bias because we were unable to control physician counseling or patient preferences with regard to VBAC and previously known or perceived risk factors of adverse outcome. However, we are confident that we were able to minimize significant bias in our multivariable analyses and suggest that, since many physicians preselect best candidates for VBAC, the effects that we found in our study are most likely underestimated rather than being invalidated by selection bias. Further, the same characteristics of this observational study design that increase likelihood of bias also increase the generalizability of the results to clinical practice across varied populations.
In conclusion, a short interpregnancy interval of less than 6 months is an independent risk factor for uterine rupture and major maternal morbidity in patients who attempt VBAC, increasing the morbidity rate twofold to threefold. A long interpregnancy interval of 60 months or more is not associated with an increase in major maternal morbidity. Our findings suggest that practitioners should consider interpregnancy interval during preconception and VBAC counseling. Further, avoiding a short interpregnancy interval may reduce VBAC-associated major maternal morbidity.
1. Conde-Agudelo A, Rosas-Bermudez A, Kafury-Goeta AC. Birth spacing and risk of adverse perinatal outcomes: a meta-analysis. JAMA 2006;295:1809–23.
2. Zhu BP, Rolfs RT, Nangle BE, Horan JM. Effect of the interval between pregnancies on perinatal outcomes. N Engl J Med 1999;340:589–94.
3. Zhu BP. Effect of interpregnancy interval on birth outcomes: findings from three recent U.S. studies. Int J Gynaecol Obstet 2005;89 suppl 1:S25–33.
4. Conde-Agudelo A, Belizan JM. Maternal morbidity and mortality associated with interpregnancy interval: cross sectional study. BMJ 2000;321:1255–9.
5. Rousso D, Panidis D, Gkoutzioulis F, Kourtis A, Mavromatidis G, Kalahanis I. Effect of the interval between pregnancies on the health of mother and child. Eur J Obstet Gynecol Reprod Biol. 2002;105:4–6.
6. Bujold E, Mehta SH, Bujold C, Gauthier RJ. Interdelivery interval and uterine rupture. Am J Obstet Gynecol 2002;187:1199–202.
7. Esposito MA, Menihan CA, Malee MP. Association of interpregnancy interval with uterine scar failure in labor: a case-control study. Am J Obstet Gynecol 2000;183:1180–3.
8. Huang WH, Nakashima DK, Rumney PJ, Keegan KA Jr, Chan K. Interdelivery interval and the success of vaginal birth after cesarean delivery. Obstet Gynecol 2002;99:41–4.
9. Rochelson B, Pagano M, Conetta L, Goldman B, Vohra N, Frey M, et al. Previous preterm cesarean delivery: identification of a new risk factor for uterine rupture in VBAC candidates. J Matern Fetal Neonatal Med 2005;18:339–42.
10. Shipp TD, Zelop CM, Repke JT, Cohen A, Lieberman E. Interdelivery interval and risk of symptomatic uterine rupture. Obstet Gynecol 2001;97:175–7.
11. Conde-Agudelo A, Rosas-Bermudez A, Kafury-Goeta AC. Effects of birth spacing on maternal health: a systematic review. Am J Obstet Gynecol 2007;196:297–308.
12. Adams MM, Delaney KM, Stupp PW, McCarthy BJ, Rawlings JS. The relationship of interpregnancy interval to infant birthweight and length of gestation among low-risk women, Georgia. Paediatr Perinat Epidemiol 1997;11 suppl 1:48–62.
13. Dicle O, Kucukler C, Pirnar T, Erata Y, Posaci C. Magnetic resonance imaging evaluation of incision healing after cesarean sections. Eur Radiol 1997;7:31–4.
14. Herring M, Baughman S, Glover J, Kesler K, Jesseph J, Campbell J, et al. Endothelial seeding of Dacron and polytetrafluoroethylene grafts: the cellular events of healing. Surgery 1984;96:745–55.
15. Mareeva LS, Levashova II, Mil’man DB. Evaluation of the condition of the uterine scar after cesarean section [in Russian]. Akush Ginekol (Mosk) 1989;3:37–8.
16. Schwarz OH, Paddock R, Bortnick AR. The cesarean scar: an experimental study. Am J Obstet Gynecol 1938;36:962–74.
17. Wojdecki J, Grynsztajn A. Scar formation in the uterus after cesarean section. Am J Obstet Gynecol 1970;107:322–4.
18. Macones GA, Cahill A, Pare E, Stamilio DM, Ratcliffe S, Stevens E, et al. Obstetric outcomes in women with two prior cesarean deliveries: is vaginal birth after cesarean delivery a viable option? Am J Obstet Gynecol 2005;192:1223–9.
19. Macones GA, Peipert J, Nelson DB, Odibo A, Stevens EJ, Stamilio DM, et al. Maternal complications with vaginal birth after cesarean delivery: a multicenter study. Am J Obstet Gynecol 2005;193:1656–62.
20. Hosmer D, Lemeshow S. Applied logistic regression. 1st ed. New York (NY): John Wiley & Sons; 1989.
21. Hosmer DW, Hosmer T, Le Cessie S, Lemeshow S. A comparison of goodness-of-fit tests for the logistic regression model. Stat Med 1997;16:965–80.
22. Conde-Agudelo A, Belizan JM, Norton MH, Rosas-Bermudez A. Effect of the interpregnancy interval on perinatal outcomes in Latin America. Obstet Gynecol 2005;106:359–66.
23. Zhu BP, Haines KM, Le T, McGrath-Miller K, Boulton ML. Effect of the interval between pregnancies on perinatal outcomes among white and black women. Am J Obstet Gynecol 2001;185:1403–10.
24. Zhu BP, Le T. Effect of interpregnancy interval on infant low birth weight: a retrospective cohort study using the Michigan Maternally Linked Birth Database. Matern Child Health J 2003;7:169–78.
25. Khoshnood B, Lee KS, Wall S, Hsieh HL, Mittendorf R. Short interpregnancy intervals and the risk of adverse birth outcomes among five racial/ethnic groups in the United States. Am J Epidemiol 1998;148:798–805.
© 2007 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.
This article has been cited