Burrows, Lara J. MD; Meyn, Leslie A. MS; Weber, Anne M. MD, MS
Recently interest has grown with idea that injury sustained during childbirth might contribute to the subsequent development of pelvic floor disorders. This has prompted some clinicians to question the potential benefits of a “prophylactic” cesarean delivery. Concurrently, the debate over whether “elective” cesarean deliveries should be performed on maternal request is escalating.
Whether cesarean delivery is performed for medical indications or at the patient’s request, data regarding the respective morbidities for vaginal and cesarean deliveries are poorly defined. There are few reports on morbidities associated with vaginal birth. Until recently, events such as anal sphincter lacerations were not even considered significant sources of morbidity.
The earliest reports on cesarean deliveries focused mainly on mortality associated with the procedure. However, maternal mortality has become increasingly rare in developed countries. Many studies evaluating maternal morbidity associated with cesarean delivery were performed in the 1970s and 1980s; since then, obstetric practice has changed considerably. Therefore, the goal of this study was to provide a current description of maternal morbidity for cesarean and vaginal deliveries.
MATERIALS AND METHODS
This study was a review, which used the Magee Obstetric Medical and Infant database, of 32,834 women who delivered at Magee-Womens Hospital from 1995 to 2000. This number included patients who were term (37 weeks of gestation or greater) with a singleton pregnancy and who had only 1 delivery or their first delivery at this institution. The data were captured as they passed through medical records and electronically interfaced with the database. This interface contained all International Classification of Diseases, 9th Revision, diagnosis and procedure codes. In addition, trained chart abstractors reviewed each medical record after every delivery and coded the information into an electronic format for demographic variables, such as maternal history of smoking. Variables were validated later by manual review of 100 arbitrarily selected medical records. This study was approved by the Magee-Womens Hospital Institutional Review Board.
Patients were categorized as follows: spontaneous vaginal delivery, operative vaginal delivery, primary cesarean delivery without trial of labor, primary cesarean delivery with trial of labor, repeat cesarean delivery without trial of labor, and repeat cesarean delivery with trial of labor. A trial of labor was defined as an attempt to achieve a vaginal delivery in patients who had no maternal or fetal contraindications to such an attempt. This included both spontaneous and induced labor. In the database, the existing “labor” variable categorized patients as those who had “spontaneous labor,” “induced labor,” or “no labor.” Patients who were categorized as “no labor” had undergone a scheduled cesarean delivery, either primary or repeat for a given indication. For the purposes of our analysis, we combined the spontaneous and induced categories to represent patients who labored in any capacity. This is how we were able to group our cesarean delivery patients into those who did and did not have a trial of labor. International Classification of Diseases, 9th Revision, codes were used to identify patients with the following complications: postpartum hemorrhage, blood transfusion, endometritis, deep venous thrombosis (DVT), pulmonary embolism, urinary tract infection, intraoperative complications, and pneumonia.
Morbidities were tabulated for each mode of delivery. Patients were analyzed based on gravidity, parity, age, race, level of education, marital status, history of smoking, maternal weight on admission, and the following maternal conditions: preeclampsia, pregestational diabetes, systemic lupus erythematosus, sickle cell disease, hypertensive disease, premature rupture of membranes, prolonged rupture of membranes, preterm spontaneous uterine contractions, prolonged second stage of labor, and chorioamnionitis in labor. Because maternal height and prepregnancy weight were not recorded in the database, maternal weight on admission was dichotomized as less than or equal to 91 kg (200 lb) and greater than 91 kg (200 lb) as a proxy for obesity.
By using a χ2 test at the .05 significance level, this study had greater than 99% power to detect a significant difference in the rate of morbidities across all 6 delivery groups based on the occurrence of these morbidities in the database. Except for DVT in the operative vaginal delivery group, which had a rate of 0.04%, all morbidities occurred at a rate of 0.1% or more. All statistical analyses were performed by using SPSS statistical software 10.1.4 (SPSS Inc, Chicago, IL). Normality was defined by visually assessing the shape of the frequency distributions. Normally distributed variables were presented as mean ± standard deviation. Data that were not normally distributed were presented as median values with overall ranges. The associations of delivery mode, demographic characteristics, and maternal conditions with maternal morbidity were evaluated by using Pearson’s χ2 or Fisher exact tests where appropriate. The associations of demographic characteristics and maternal conditions with delivery mode were evaluated by using Pearson’s χ2, one-way analysis of variance, or Kruskal–Wallis tests where appropriate. All statistical tests were evaluated at the .05 significance level. Separate multivariable logistic regression models were developed to identify factors independently associated with each morbidity. Variables with P values less than .1 were considered for inclusion in the models. Models were developed by using forward stepwise regression based on the likelihood ratio test statistic. Variables were retained in the model if the Wald χ2 test statistic had a P value of .05 or less.1
Of 32,834 subjects, 27,178 had vaginal delivery (operative = 4,908; spontaneous = 22,270) and 5,656 had cesarean delivery. Third- or fourth-degree lacerations occurred in 1,733 (7.8%) women who had spontaneous vaginal delivery compared with 1,098 (22.3%) who had operative vaginal delivery. There was a total of 12,602 episiotomies performed.
The demographics for patients in the database stratified by delivery mode are provided in Table 1. The method of delivery differed significantly by age and racial group (P ≤ .001). White women were more likely than black women to undergo cesarean delivery without a trial of labor. Women who underwent repeat cesarean delivery were older. The frequencies of comorbidities stratified by delivery mode are shown in Table 2. Table 3 shows the adjusted odds ratios (OR) stratified by delivery type for endometritis, pneumonia, postpartum hemorrhage, transfusion, and DVT.
Endometritis, the most frequent morbidity, was most strongly associated with cesarean delivery. Compared with spontaneous vaginal delivery, cesarean delivery after trial of labor was associated with a 14.6-fold increased risk of endometritis for repeat procedures and a 21.2-fold increased risk after primary procedures. Women with chorioamnionitis were 3.2 times more likely to have endometritis when adjusted for delivery mode, age, race, smoking, admission weight, prolonged rupture of membranes, and vaginal laceration. Those who sustained an anal sphincter laceration were 2.9 to 4.5 times more likely to develop endometritis compared with those who did not have a laceration when adjusted for delivery mode, age, race, smoking, admission weight, prolonged rupture of membranes, and chorioamnionitis. Prolonged rupture of membranes conferred a 2-fold increased risk of endometritis. Patients whose weight on admission was greater than 200 pounds were 30% more likely to develop endometritis.
Of the 41 patients (0.1%) who developed pneumonia, those who underwent cesarean delivery incurred the greatest risk. Women who had repeat cesarean delivery after trial of labor were 9.3 times more likely to have pneumonia than women after spontaneous vaginal delivery. Smokers had an adjusted 2-fold increased risk of developing pneumonia.
Regarding postpartum hemorrhage, the spontaneous vaginal delivery and operative vaginal delivery groups had the highest rates, which were 5.0% and 4.7%, respectively. Independent risk factors for postpartum hemorrhage were episiotomy (OR 1.3; 95% confidence interval [CI] 1.1, 1.4), pregnancy-induced hypertension (OR 1.7; 95% CI 1.4, 2.1), preeclampsia (OR 2.3; 95% CI 1.8, 2.9), chorioamnionitis (OR 1.9; 95% CI 1.6, 2.3), and prolonged second stage of labor (OR 2.1; 95% CI 1.5, 2.8).
Of the 5,656 patients who underwent cesarean delivery, 49 (0.86%) received blood transfusions compared with 61 of 27,178 patients after vaginal delivery (0.22%). Patients who had cesarean delivery with trial of labor sustained the highest risk, more than a 4-fold increase. Compared with spontaneous vaginal delivery, patients had an adjusted 5-fold increased risk of being transfused if they had preeclampsia or sickle cell disease. The presence of chorioamnionitis increased the risk (adjusted) of transfusion 3-fold.
Of the 29 patients who had DVT, 15 (0.1%) had spontaneous vaginal delivery, 2 (0.04%) had operative vaginal delivery, and 12 had cesarean delivery (0.2%). A prolonged second stage of labor conferred a 6.1-fold increased risk of DVT. When stratified by delivery mode, primary cesarean delivery with trial of labor imposed the highest risk of DVT, which was 3.9 times higher than spontaneous vaginal delivery. Four patients had pulmonary emboli, 2 after spontaneous vaginal delivery (0.007%) and 2 after primary cesarean delivery (0.04%), 1 with trial of labor and 1 without.
Nine patients underwent hysterectomy at the time of delivery: 2 after spontaneous vaginal delivery, 2 after primary cesarean without trial of labor, 3 after primary cesarean with trial of labor, and 2 after repeat cesarean with trial of labor. These patients were too few in number to undergo statistical modeling. The median age of these 9 patients was 28 years, 3 were primiparous, 5 had chorioamnionitis, and 1 had preeclampsia.
Of the 5,656 cesarean deliveries, 22 intraoperative injuries were reported (0.38%) in the database, 11 of which were cystotomies. Six occurred during primary and 5 during repeat cesarean delivery. Other injuries included broad ligament and bladder flap hematomas, lacerations of the uterine artery or urethra, and injury to the mesosalpinx.
The most important finding of this study is that, although overall rates of morbidity for cesarean deliveries are higher than for vaginal births, they are not as high as reported previously. The most common postoperative morbidity attributable to cesarean delivery across studies are infectious; endometritis has been reported to occur in 6–18% of procedures.2–5 The overall rate of 6.9% after cesarean delivery in our study is on the low end of that range. Primary cesarean deliveries after labor carry a higher risk of infection than repeat procedures without labor and our findings are consistent with this.6
Postpartum hemorrhage occurred most commonly in the vaginal delivery groups. This was an unexpected finding and may reflect underestimation of the blood loss at cesarean delivery. Alternatively, this may reflect increased blood loss related to episiotomy use or other perineal or vaginal trauma. Although patients after spontaneous vaginal delivery had the highest rate of postpartum hemorrhage in this study, patients who had cesarean delivery had a higher transfusion rate. Perhaps this discrepancy is the result of inaccuracies in estimating blood loss, different definitions of postpartum hemorrhage for vaginal and cesarean deliveries, or the possibility that clinicians are more likely to transfuse if a patient has had surgery. As with the other morbidities, parity was a not a significant predictor of postpartum hemorrhage.
Regarding postpartum hemorrhage, the OR for primary cesarean deliveries without trial of labor was decreased. The ORs for other cesarean delivery types were also decreased but not to a statistically significant level. Interestingly, a recent study examining maternal morbidity associated with cesarean delivery without labor found that women who underwent primary cesarean delivery without labor were less likely to have postpartum hemorrhage compared with those who had spontaneous onset of labor.7
Previous reports of blood transfusions with cesarean delivery range from 1.2% to 6.3%,5,8 which is higher than the rates found in this study (0.3–1.1%). This may reflect improvements in surgical technique or the increasing reluctance of practitioners to transfuse patients. Patients who had cesarean delivery after trial of labor had a 4-fold higher transfusion rate compared with spontaneous vaginal delivery. This may be the result of the fact that performing cesarean delivery after substantial descent of the fetus increases the chance of uterine, cervical, and vaginal lacerations. Furthermore, this group inclues patients who had urgent and emergent deliveries. The need for blood transfusions is more common with emergency surgery.5
The occurrence of DVTs in the cesarean groups (0.21%) was lower than the 0.6–1.8% reported to date and highlights the rarity of this condition.6,9 Pulmonary emboli have been reported in 0.1–0.2% of cesarean deliveries.5,10 In our series, the rate was even lower at 0.04%. Given that this was such a rare occurrence regardless of delivery type, one cannot draw any meaningful conclusions regarding risk by mode of delivery.
It is interesting that white women were more likely to undergo cesarean delivery without trial of labor. This may reflect different practice patterns among private practitioners and resident staff, especially regarding vaginal birth after cesarean delivery. Irwin et al11 found that black women were more likely to undergo a cesarean delivery than their white counterparts. Alternatively, Gould et al12 found that white women had the highest primary cesarean delivery rate, and this association appeared to be related to socioeconomic status.
This study has the inherent limitations of a retrospective review. As with any database, there is the potential for errors in data entry. Unfortunately, variables such as the urgency with which cesarean deliveries were performed, intraoperative injury, paralytic ileus, and wound cellulitis were not reliably recorded in this database or in the medical records system. This was a limitation, especially regarding intraoperative injuries. Although there are specific codes for the different procedures performed for various intraoperative injuries, there is enough variation in the procedures that we did not feel the data were robust enough to allow comparisons between the different delivery types. A small number of complications were documented, prohibiting us from drawing valid conclusions from these data. Additionally, because of limitations in the information contained in the database, some patients were artificially categorized. For example, we were unable to consider situations such as those who did not progress past the first stage of labor but who ultimately had a cesarean delivery. Ideally we would have been able to consider such details as they may have an impact on maternal morbidity. Additionally, our data are limited to the index hospitalization (the delivery); if rehospitalization occurred, this information would not have been captured. Finally, these data were obtained from a tertiary medical center; therefore, these results may not be generalized to many community settings.
Although several studies have examined morbidity associated with cesarean delivery, few studies have compared these with morbidity of vaginal delivery, such as anal sphincter lacerations and perineal wound breakdown or postoperative complaints common to both types of delivery, such as persistent pain, urinary incontinence, and fecal incontinence. Hannah et al13 found that patients who had planned cesarean delivery for breech presentation were 38% less likely to report postpartum urinary incontinence, and they did not have an increased risk of various other problems at three months postpartum. We were unable to assess the aforementioned morbidities because they were not recorded in the database or they primarily occurred in the outpatient setting. Prospective studies taking these measures into account are needed to fully characterize maternal morbidity by delivery type.