If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
The purpose of this study was to investigate whether knowledge of ultrasound-obtained estimated fetal weight (US-EFW) is a risk factor for cesarean delivery (CD).
Retrospective cohort from a single center in 2009-2010 of singleton, term live births. CD rates were compared for women with and without US-EFW within 1 month of delivery and adjusted for potential confounders.
Of the 2329 women in our cohort, 50.2% had US-EFW within 1 month of delivery. CD was significantly more common for women with US-EFW (15.7% vs 10.2%; P < .001); after we controlled for confounders, US-EFW remained an independent risk factor for CD (odds ratio, 1.44; 95% confidence interval, 1.1–1.9). The risk increased when US-EFW was >3500 g (odds ratio, 1.8; 95% confidence interval, 1.3–2.7).
Knowledge of US-EFW, above and beyond the impact of fetal size itself, increases the risk of CD. Acquisition of US-EFW near term appears to be an independent and potentially modifiable risk factor for CD.
Utilization of prenatal ultrasonography has risen in parallel. The average number of ultrasound examinations per pregnancy has nearly doubled in the last decade, and the percent of pregnancies with ≥4 ultrasound examinations has almost tripled.
Increased use of prenatal ultrasound scanning may be contributing to the rising CD rate. One study suggested that sonographic overestimation of fetal macrosomia is associated with a lower threshold for CD for labor arrest.
If obtaining an ultrasound-estimated fetal weight (US-EFW) near term places women at increased risk of CD, it may represent a modifiable risk factor that providers could target. We hypothesized that knowledge of US-EFW may impact provider practice or patient preference even if the fetus is not macrosomic. Our study analyzes whether women with any US-EFW within 1 month of delivery are more likely to be delivered by cesarean section compared with those women without an US-EFW.
Materials and Methods
We created a retrospective cohort of deliveries at a single tertiary care center during 1 fiscal year (Oct. 1, 2009, to Sept. 30, 2010) after obtaining approval from our institutional review board (protocol #2011-P-001261/1). We included women who were delivered by provider groups for whom ultrasound reports were available on the electronic medical record. These groups included 1 private practice, 1 group of certified nurse-midwives, and a faculty group that included patients who were seen by maternal-fetal medicine subspecialists, academic generalists, and obstetric residents.
Our cohort was restricted in several ways. We included only singleton live births and excluded those patients with a scheduled CD (both primary and repeat). Patients with a maternal or fetal contraindication to vaginal delivery (including placenta previa, breech/malpresentation, and previous uterine surgery) and patients with diabetes mellitus (both gestational and pregestational) were also excluded because it is standard practice at our institution to get an US-EFW for all patients with diabetes mellitus. Finally, we excluded women who delivered at <39 weeks' gestation to create a lower risk cohort by removing anyone with a serious maternal or fetal indication that necessitated delivery at <39 weeks' gestation. In addition, we assumed that concern for labor dystocia because of fetal size would be greater at ≥39 weeks' gestation compared with earlier gestational ages; thus, the knowledge of US-EFW would have a greater impact at later gestation.
Ultrasound examinations were performed by maternal-fetal medicine specialists and radiologists. Studies were performed in 1 of 3 centralized ultrasound suites; the ultrasound reports were available immediately on our electronic medical record. EFW was provided in grams, and a weight percentile was given that was based on a curve of hospital-specific gestational age–matched birthweights.
We abstracted only ultrasound reports with an EFW assessment that had been performed within 1 month (31 days) of the delivery date. If a patient had >1 US-EFW within the month before delivery, we considered only the most recent. Of note, the provider who ordered the ultrasound scan may not have been the delivering provider. All patients were delivered in group practices.
Demographic and delivery data were abstracted from our computerized medical record system. Missing demographic data were hand-collected by a review of the relevant charts. Gestational ages were rounded to the nearest completed gestational week. Body mass index (BMI) was categorized as obese if prepregnancy BMI was >30 kg/m2 or, if prepregnancy BMI was not available, if delivery BMI was >35 kg/m2, because this was the average BMI increase in our cohort for those women with both sets of data available. The CDs for the midwife group were performed by the faculty practice. Thus, CDs were reviewed, and the provider group was reassigned to the midwives if appropriate. Women with prelabor rupture of membranes and oxytocin administration before the onset of spontaneous contractions were considered to be induced. Ultrasound information was hand-collected, and the date, indication, and EFW were recorded.
We created 2 groups: (1) women with an US-EFW within 1 month of delivery and (2) women with no US-EFW. We reported baseline characteristics for these 2 groups using medians and quartiles (because they were not normally distributed) and compared demographic differences using Wilcoxon 2-sided tests for continuous variables and χ2 testing for categoric variables. We used logistic regression to analyze whether US-EFW was an independent risk factor for CD. The mode of delivery was the outcome variable; both spontaneous and operative vaginal deliveries were included in the vaginal delivery group. The statistical software package (SPSS version 12.0; SPSS Inc, Chicago, IL) was used for all data analyses.
We performed logistic regressions on several different population subsets to analyze the potential relationship and explanation for our findings in the entire cohort. The different subsets included (1) a low-risk group (we excluded any patient with an ultrasound scan or induction of labor for maternal disease or fetal concerns), (2) a group with smaller EFWs (<3000 g), (3) a group with larger EFWs (>3500 g), and (4) a group with actual birthweights of >4000 g.
Power calculations were performed with the assumption that the average CD rate would be 15% for the entire population and that 50% of the population would have an US-EFW performed within 1 month of delivery. We assumed that a 5% absolute difference in CD rates would be meaningful clinically and used a power of 80%, which required a sample size of 1600 women. We chose to use all the deliveries within 1 fiscal year (approximately 2300 women after exclusions) as a convenience sample, which gave us 92% power to detect a 5% absolute difference in CD rates.
During the study period, 7880 women delivered. Figure 1 displays all our exclusions to arrive at a final cohort of 2329 women. Our cohort was composed of 557 women who were delivered by the private practice, 701 women who were delivered by the faculty practice, and 1071 women who were cared for antepartum by the midwife practice.
In total, 1168 women (50.2%) had an US-EFW within 1 month of delivery. Of those with an US-EFW, 540 women (46%) had the procedure within 1 week of delivery; 276 women (24%) had the procedure 1-2 weeks before delivery; 179 women (15%) had the procedure 2-3 weeks before delivery, and 173 women (15%) had the procedure 3 weeks to 1 month before delivery. The most common indications for an ultrasound scan, as listed on the ultrasound report, were postdates (19%), uterine size less than dates (17%), uterine size greater than dates (14%), and advanced maternal age (16%). The ultrasound findings did not significantly impact delivery timing because there were no new fetal anomalies detected and because only 13 women were induced for oligohydramnios as seen on ultrasound scans within a week of delivery.
Demographic characteristics by ultrasound status are displayed in Table 1. As shown, women with an US-EFW within 1 month of delivery were significantly older and more likely to be white, to be cared for by an physician as compared with a midwife, to be obese, and to undergo an induction of labor. Women with an US-EFW also delivered at slightly later gestational ages. The median for both groups was 40 weeks' gestation, but the range was higher in the US-EFW group, which was statistically significant (P = .03). There were no differences in parity, previous CD, or birthweight.
TABLE 1Demographic characteristics of women with and without ultrasound-estimated fetal weight
Ultrasound-estimated fetal weight within 1 month of delivery
Table 2 shows the unadjusted CD rates between those women with and without an US-EFW within 1 month of delivery. Women with an US-EFW were 55% more likely to have a CD (15.7% vs 10.2%; P < .01). The increased risk of CD with US-EFW was observed in all provider groups. Women with an US-EFW were more likely to have a CD compared with women without an US-EFW in the private physician, faculty physician, and midwife practices with odds ratios (ORs) of 2.1 (95% confidence interval [CI], 1.1–3.7), 1.3 (95% CI, 0.8–1.9), and 1.6 (95% CI, 1.1–2.3), respectively. For every 18 additional US-EFWs that were obtained, 1 additional CD was performed. The timing of ultrasound ascertainment within the 1-month period before delivery had little impact. The CD rate was 16% with an ultrasound within 1 month of delivery, 16% with an ultrasound within 3 weeks of delivery, 15.5% within 2 weeks of delivery, and 12% within 1 week of delivery.
TABLE 2Delivery outcome by ultrasound-estimated fetal weight status (unadjusted)
Ultrasound-estimated fetal weight within 1 month of delivery
CD indications were similar between the groups (Table 2). The indication for CD was assigned by the delivering provider, and there were no standardized protocols in place on the labor and delivery unit as a whole for assigning the indication. The 2 main indications for CD were failure to progress and nonreassuring fetal heart tracings; there was no statistical difference in the frequency of CD for these indications between the 2 groups. There were 17 CDs for presumed macrosomia, all within the US-EFW group. Of these, only 3 deliveries had an EFW of >5000 g.
Our logistic regression model controlled for the potential confounders of birthweight, maternal age, race, BMI, previous CD, parity, gestational age, induction vs spontaneous labor, and provider group (Table 3). US-EFW was a statistically significant independent risk factor for CD. As shown, after the data were controlled for these potential confounders, women with an US-EFW within 1 month of delivery were 44% more likely to have a CD. The other independent predictors of CD were birthweight, advanced maternal age, black or Hispanic race, nulliparity, obesity, previous CD, and faculty provider group (as compared with either private or midwife practices).
TABLE 3Risk factors for cesarean delivery (adjusted odds ratios by logistic regression)
The relationship between US-EFW and CD changed when the data were analyzed in population subsets (Figure 2). In the entire cohort, US-EFW was associated with increased odds of CD (OR, 1.44; 95% CI, 1.1–1.9). For those with smaller fetuses, which are categorized here as EFW of <3000 g, US-EFW was no longer associated with an increased risk of CD (OR, 0.97; 95% CI, 0.45–2.1). In contrast, the magnitude of the association increased for those with larger fetuses. With EFW of >3500 g, the OR was greater than that of the cohort (OR, 1.8; 95% CI, 1.3–2.7) and was stronger still when considering only the population with an actual birthweight >4000 g (OR, 3.1; 95% CI, 1.2–8.3). In a low-risk subset (excluding those with an induction or ultrasound scan for maternal disease or fetal indication), there was a nonsignificant trend toward US-EFW as a risk factor for CD (OR, 1.3; 95% CI, 0.94–1.9).
Women with an US-EFW within 1 month of delivery were 44% more likely to be delivered by cesarean section, even after the data were controlled for birthweight, age, race, parity, gestational age, provider group, obesity, previous CD, and induction of labor. This association appears to be due to provider concern over larger US-EFWs, because there was no increased risk for CD if the US-EFW was <3000 g; whereas for women with an US-EFW of >3500 g, there was an 85% increased risk of CD with US-EFW ascertainment. It is important to note that we controlled for birthweight; thus, it is the patient and provider knowledge of the US-EFW, above and beyond the impact of fetal size itself, which is associated with the increased risk of CD.
This study is not without limitations. First, the clinical indication for ultrasound examination was not well coded in our dataset. Indications were entered by the sonographer, and the accuracy of these is unknown. Similarly, the reliability of coding for CD indications is also unknown. However, our main finding was that US-EFW for any indication was associated with CD and thus would not be affected by inaccurate coding. Second, there is the possibility of residual confounding; the relationship between ultrasound scanning and mode of delivery may be complicated by the fact that the same underlying high-risk complication may increase the likelihood of both an ultrasound and CD. We attempted to limit this possibility by controlling for any known confounders in our regression model and by restricting our cohort to those who delivered at ≥39 weeks' gestation to reduce the proportion of women in our sample with pregnancies at higher risk. In addition, we also created a lower risk cohort by eliminating anyone with an ultrasound scan or induction for maternal disease or fetal concerns. The relationship between US-EFW and CD held true in this subset as well, although was no longer statistically significant. Although certainly residual confounding may still be present and explain some of our findings, what lends further support to a true association between US-EFW and CD, is the fact that women with an EFW of <3000 g do not have an increased risk of CD. If the association between US-EFW and CD were due to residual confounding, we would expect to see increased odds of cesarean delivery even in women with a smaller EFW. Thus, the disappearance of the association between US-EFW and CD at lower weights suggests that it is truly knowledge of a larger EFW that is driving this association.
Our study has several strengths. We had a large sample size and were able to obtain complete demographic information on all the women, which allowed us to control for many possible confounders. In addition, our cohort was drawn from a variety of provider groups that included a private practice, midwife group, and faculty practice with both resident and maternal-fetal medicine clinics. Our findings were consistent across provider groups, which allowed for generalizability to a wide variety of patient populations.
Previous work has suggested that overestimation of fetal weight by ultrasound scanning may be associated with an elevated risk of CD and that women with a false diagnosis of macrosomia were at increased risk of CD compared with birthweight-matched control subjects.
Similar to these findings, our findings were that women with an US-EFW of >3500 g are at an increased risk of CD. Unlike previous work, however, we chose not to limit our outcome to CD for labor dystocia only but rather all cesarean deliveries. CD is a hard outcome, whereas CD for labor dystocia relies on provider coding, which is subject to provider interpretation. Additionally, US-EFW potentially could increase the risk of CD for any indication, by affecting the patient or provider beliefs with regards to the probability of vaginal delivery and the assessment of labor length. Our study also extends previous findings to a broader population. We find that larger EFWs, even if not formally macrosomic, still increase the risk of CD. Thus, knowledge of sonographic EFW may mediate subtly the relationship between provider behavior and mode of delivery. Perhaps obstetricians believe an US-EFW is more accurate than a clinical estimation of fetal size alone, and the availability of US-EFW leads providers to give more consideration to fetal size when managing labor than they otherwise would.
The rising CD rate, especially primary CD, is of great concern to policy organizations, providers, and patients alike. Unfortunately, many of the reasons behind the rise, such as advancing maternal age and increasing maternal obesity, are not easily modifiable. Sonographic ascertainment of fetal weight near term in the absence of clear and evidence-based indications; however, could potentially be avoided. Although there are surely many benefits to the increased use of prenatal ultrasound scanning over recent years, providers must be careful also to consider potential harms. Our findings suggest that provider knowledge of larger US-EFW may be associated with a higher rate of CD; thus, limiting the ultrasound ascertainment of fetal weight near term may help to reduce the CD rate in the United States.
We thank Louise Wilkins-Haug, MD, PhD, for her assistance in interpreting the data and Vineeta Vaidya for her assistance in data collection (both are from Brigham and Women's Hospital, Boston MA), and the Excellence in Clinical Research Course, where the project was conceived.
Births: final data for 2008: national vital statistics reports; vol 59, no 1.
National Center for Health Statistics,