OBJECTIVE: The purpose of this study was to quantify patient populations and practice patterns at perinatal centers with the highest and lowest cesarean delivery rates.
METHODS: The 2 perinatal centers in our state with the lowest (Hospital A-16.6%) and highest (Hospital B-20.3%) overall cesarean rates for Robson group 1 (term primigravidas, vertex, spontaneous labor) and group 2 (term primigravidas, vertex, induced labor) were identified. A total of 174 medical records at Hospital A and 150 records at Hospital B were reviewed. Statistical analysis was performed using independent-sample t tests, χ2, and multiple logistic regression.
RESULTS: Indications for cesarean delivery were not different between the 2 groups, with the majority being for failure to progress in labor and nonreassuring fetal status. There were no differences between groups in rates of postpartum hemorrhage, chorioamnionitis, or endometritis. There were no differences in neonatal outcomes.
Although women delivering in hospital A were not more likely to receive oxytocin augmentation (P = .291), their mean maximal oxytocin dosage was higher (14.5 units compared with 11.6 units, P < .001), and they were more likely to receive both fetal scalp electrodes (60.9% compared with 37.3%, P < .001) and intrauterine pressure catheters (63.8% compared with 26.0%, P < .001).
CONCLUSION: Because safe reduction in cesarean delivery rates for primigravidas will proportionately reduce the number of repeat cesarean delivery required, benchmarking practices as described in this study can be considered in obstetric practices interested in long-term reductions of their cesarean delivery rates.
LEVEL OF EVIDENCE: III
Comparison between hospitals can identify benchmarking practices that could safely lower cesarean rates.
From the *Department of Obstetrics and Gynecology, University of Utah School of Medicine, and †Utah Department of Health, Salt Lake City, Utah.
Presented in part at the annual meeting of the Society for Gynecologic Investigation, Houston, Texas, March 24-27, 2004.
Address reprint requests to: D. Yvette LaCoursiere MD, MPH, University of Utah, Department of Obstetrics and Gynecology, 30N 1900 E Suite 2B200, Salt Lake City, UT 84108; e-mail: firstname.lastname@example.org.
Received August 16, 2004. Received in revised form December 13, 2004. Accepted December 16, 2004.
Decreasing the overall cesarean delivery rate continues to be a major goal for obstetricians and health policy makers since the national cesarean delivery rate peaked in 1985 (23% of all deliveries).1,2 With the wide-scale endorsement of vaginal birth after cesarean delivery (VBAC) in the 1980s as a means to reduce the number of repeat cesarean deliveries3 plus the adoption of active labor management protocols to decrease the primary cesarean delivery rate,4,5 many centers across the country have found that cesarean delivery rates can be reduced without untoward harm to the mother or the infant.1,6,7
Nonetheless, cesarean rates vary substantially between geographic areas in the United States and often vary substantially between hospitals in the same community.8 Numerous factors contribute to these differences, including the availability of ancillary staff (anesthesia, pediatrics, etc), training and experience of the surgeon(s), and characteristics of particular patient populations.9 The recent VBAC policy changes by the American College of Obstetricians and Gynecologists have also increased the number of repeat cesarean deliveries both nationally and in our state.10,11
The Robson cesarean classification system was developed as a way to categorize the population of pregnant women requiring cesarean delivery and thereby to compare cesarean rates among equivalent subpopulations.12 In this system any cesarean delivery can be placed in 1, but only 1, of 10 mutually exclusive patient population categories (Table 1). Although not yet widely applied to U.S. populations, this classification system could be of value for defining and comparing optimal cesarean delivery rates for different patient populations.
Using birth certificate data and the Robson cesarean classification system12 we previously surveyed the cesarean rates in the 6 perinatal centers in our state.8 The average cesarean delivery rates for Robson group 1 and 2 were 12.7% ± 3.0 and 13.7% ± 3.3. We found that rates varied substantially between centers both for Robson group 1 (term primigravidas with vertex presentation in spontaneous labor), 7.8% (Hospital A) to 16.3% (Hospital B), and Robson group 2 (term primigravidas with vertex presentation whose labors were induced), 8.3% (Hospital A) to 18.3% (Hospital B). The overall cesarean rates, regardless of Robson criteria varied from 16.6% at Hospital A to 20.3% at Hospital B. The objective of this study was to quantify patient populations and practice patterns at the institutions with the highest and lowest cesarean delivery rates in the State of Utah.
The State of Utah has 6 designated perinatal centers. When birth certificate data from 1998 to 2000 (n = 127,462) from these 6 hospitals were compared using Robson categories 1 and 2, one hospital (Hospital A) had the lowest cesarean rates in both categories and another hospital (Hospital B) had the highest cesarean rates in both categories. The protocol was reviewed and approved by the Utah Department of Health and the University of Utah Institutional Review Boards.
One hundred temporally consecutive maternal and fetal medical records identified from State of Utah birth certificate data were requested from each of the 2 Robson criteria from both Hospital A and Hospital B. The sampling technique oversampled for cesarean delivery, such that one half of each sample was delivered by cesarean.
Inclusion criteria were the following: primigravidas with singleton vertex presentations at more than 37 weeks gestation who were admitted in spontaneous labor (Robson group 1) or for induction of labor (Robson group 2). Medical records were reviewed for baseline demographic characteristics, antepartum and medical complications, labor management variables, and neonatal outcomes. Demographic characteristics included maternal age, gestational age at delivery, maternal body mass index (BMI) at delivery, past and current tobacco use, number of prenatal care visits, and the woman's choice of perinatal health care provider.
Method of delivery was determined (spontaneous vaginal delivery, forceps- or vacuum-assisted vaginal delivery, cesarean delivery) as were the specifics of intrapartum management, including oxytocin use, maximal oxytocin dose and duration, fetal scalp electrode or external monitoring to assess fetal heart rate, intrauterine pressure catheter placement, and corresponding maximal Montevideo units. Indications for cesarean delivery were recorded. Records were also reviewed for the occurrence of postpartum hemorrhage requiring transfusion, chorioamnionitis, and endometritis. Neonatal complications reviewed included birth weight, 1- and 5-minute Apgar scores, and need for newborn intensive care unit admission.
The calculation of the sample size to detect a 50% difference in intrapartum management characteristics was 162 total women per group at power of 80% and a type I error of 0.05. The specific characteristics compared for sample size calculation were intrauterine pressure catheter and fetal scalp electrode use; both the fetal scalp electrode and intrauterine pressure catheter groups were each estimated to have a 30% usage rate at one hospital. We estimated that an increase to 45% would be a clinically meaningful increase.
Data analysis was performed using SPSS 12.0 (SPSS Inc., Chicago, IL). Continuous normal data were described using means and standard deviations, then analyzed using independent t tests. Categorical variables were described using frequency and percentages. Statistical analysis on these variables was performed using χ2 analyses (uncorrected or Fisher exact test when expected cells were less than 5), For analyses using t tests and χ2, statistical significance was set at P < .05. An a priori list of independent and dependent variable were selected for the multiple logistic regression models. Scale variables were coded using dummy variables (Tobacco, Robson, and Hospital). Maternal age was included as a continuous variable. The variables were entered in block. Statistical hypotheses were tested using 2-tailed 95% confidence intervals.
A total of 174 records were reviewed at Hospital A and 150 records from Hospital B. At Hospital A, 78 records fell into Robson category 1, with 96 in Robson category 2. At Hospital B, 50 records fell into Robson category 1, with 100 records in Robson category 2. All patients who were classified into Robson cesarean classifications 3 through 10 (Table 1) were excluded from the analysis. For Hospital A (low cesarean delivery rate), 12 patient records belonged to Robson categories 3–10 (6% misclassification rate), and a total of 14 were unavailable for analysis. At Hospital B (high cesarean delivery rate), a total of 21 records belonged to Robson categories 3–10 (10.5% misclassification rate), and 29 were unavailable for analysis.
Table 2 shows that women delivering at Hospital A were significantly younger (23.7 compared with 25.0 years, P = .008) and had higher mean gestational ages (39.8 compared with 39.3 weeks, P < .001). Women at Hospital A were also significantly less likely to have current or past use of tobacco. The 2 groups were not different in any characteristics of percentage of total prenatal care visits (P = .131), level of care provider (P = .065), or BMI (P = .540). There was no difference between hospitals in the indication for induction.
There were significant differences in labor management characteristics between the 2 hospitals (Table 2). Although women delivering in Hospital A (low cesarean rates) were not more likely to receive oxytocin augmentation (75.9% vs 70.7%, P = .291), their mean maximal oxytocin dosage was higher (14.5 units compared with 11.6 units, P < .001). There was no difference in the duration of oxytocin administration. Hospital A used an active labor management algorithm similar to that described by the National Maternity Hospital in Dublin.4 Women at Hospital A were more likely to receive both fetal scalp electrodes (60.9% compared with 37.3%, P < .001) and intrauterine pressure catheters (63.8% compared with 26.0%, P < .001). The unadjusted odds of having a fetal scalp electrode or intrauterine pressure catheter at the low cesarean delivery rate hospital were 2.78-fold (95% confidence interval 1.72, 4.49) and 5.01-fold (95% confidence interval 3.03, 8.34) higher, respectively, than at the high cesarean delivery hospital. The maximal Montevideo units at Hospital A were not statistically different from Hospital B (252.2 compared with 230.0, respectively). Fetal scalp electrode and intrauterine pressure catheter remained significantly higher at Hospital A after controlling for maternal age, tobacco use, gestational age at delivery, and Robson criteria (Table 3).
The oversampling technique yielded 70 women from the random sample of patients at Hospital A and 73 at Hospital B who underwent cesarean delivery. Because cesarean deliveries were oversampled, cesarean delivery rates cannot be calculated from this sample. Cesarean delivery rates at both hospitals were previously identified as follows: Hospital A 7.8% (Robson 1) and 8.3% (Robson 2) and Hospital B 16.3% (Robson 1) and 18.3% (Robson 2).8 Table 2 shows that indications for cesarean delivery were not different between the 2 groups, with the majority being for failure to progress in labor (A = 41.1% compared with B = 32.9%, P = .436) and nonreassuring fetal monitor tracings (A = 53.4% compared with B = 45.7%, P = .543).
When comparing Hospital A to Hospital B, there were no differences in postpartum hemorrhage requiring transfusion (P = .999), chorioamnionitis (P = .690), or endometritis (P = .627) (Table 4). No statistically significant differences develop for these complications when controlling for maternal age, tobacco use, gestational age at delivery, and Robson criteria. As shown in table 4, there were no differences in neonatal outcomes, including birth weight (P = .260), 1- or 5-minute Apgar scores less than 7 (P = .664 and P = .127, respectively), and neonatal intensive care unit admissions (P = .065).
Although not widely used in the United States, the Robson cesarean delivery classification system12 allows comparison of cesarean rates within specific subsets of an obstetric population and thereby obviates many of the historic arguments that have arisen when comparing overall cesarean rates between different populations. By reviewing patient demographics and practice patterns for 2 similar patient populations (Robson groups 1 and 2, Table 1) we sought to identify issues that could explain the greater than 2-fold differences in cesarean rates between Hospital A (low cesarean rates) and Hospital B (high cesarean rates).
As described by others, we have found only a modest correlation between birth certificate data and actual medical record reviews. Of the 100 patients in each of the 2 groups at each hospital that were identified from birth certificate data, we were only able to find records on 89.3% of women. Of these, 90.7% were correctly classified by Robson criteria.
Modest differences were identified in the primigravid patient populations between Hospital A and Hospital B. Although statistically significant, it seems unlikely that an average maternal age difference of 1.3 years or a difference of a few days gestational age would explain a cesarean rate difference in excess of 2-fold between the 2 hospitals. Likewise, although women at Hospital B were more likely to smoke, there is no reason to think that this would increase the risk of operative delivery to the extent observed. In fact, there is reasonable evidence that smokers are less likely to require operative delivery than are women who do not smoke,13 presumably as a result of the decreased average birth weight associated with smoking.
Another factor that might explain a higher cesarean delivery rate at Hospital B is its more racially diverse patient population as compared with Hospital A. Thom et al14 showed that cesarean delivery rates are significantly higher in black primigravidas compared with whites. Racial characteristics of the overall populations studied at both Hospital A and B were not analyzed, however, and this is a potential limitation of this study that might potentially have served to explain a higher cesarean delivery rate at Hospital B. Caucasian women by far make up the majority of patients at both facilities and lessen this potential effect.
We hoped to identify practice patterns that could safely reduce cesarean rates and serve as a benchmark for our region. The higher average oxytocin dose and the more frequent use of intrauterine pressure catheters and fetal scalp electrodes in Hospital A confirm that that institution's active labor management protocol may be contributing to its lower cesarean delivery rate. The equivalence of other maternal or neonatal outcomes in Hospital A suggests that active labor management practices can lower the cesarean delivery rate within Robson categories 1 and 2 without obvious increased risks.
The differences in practice patterns in this study may have been attenuated given the somewhat increased proportion of Robson 2 (induced) subjects collected at Hospital B. This difference does not affect the cesarean rates documented at these facilities, because these rates were stratified by Robson criteria. The increase in the proportion of women induced may increase the need for intervention, such as oxytocin, intrauterine pressure catheter, or fetal scalp electrode use. This would have a diminutive effect on the overall difference between groups. Despite this, the results remain statistically significant.
The overall results of this study confirm previous reports that cesarean delivery rates can be decreased with the active management of labor.1,2,6,14–23 Active management of labor as described by O'Driscoll et al2,21 consists of strict criteria for the diagnosis of labor, prompt intervention with high-dose oxytocin in the event of inefficient uterine action or ruptured membranes, and a dedicated commitment not to leave a woman unattended during labor.2
Although our results suggest the lower cesarean delivery rate at Hospital A is the result of the institutional protocol of active labor management at Hospital A, it is possible that other factors might explain some proportion of this decrease. There are data that indicate changing physician behavior is also effective at lowering the cesarean rate.22 Myers and Gleicher1 demonstrated that strict adherence by physicians to defined labor management parameters resulted in a significant reduction in the cesarean rate at their institution. Other factors potentially to consider in studies comparing cesarean delivery rates between institutions are the number of obstetricians on staff and how many patients they have, what the nursing care is like, what the litigation history has been at a particular institution, etc. Perhaps there are combinations of these factors occurring at Hospital A that contribute to a lower cesarean delivery rate compared with Hospital B. This study did not address any of these potential factors that might serve to either increase or reduce the cesarean delivery rate, and this represents a potential limitation of this study.
We have not identified the 2 specific hospitals studied in this report and have no intention of doing so. However, we have initiated a statewide education program consisting of presentations at hospital staff meetings by the authors of this study and will circulate reprints of this manuscript to all obstetric care providers and hospital nursing staff in Utah.
In conclusion, reviewing labor management practices by specific patient subgroups can identify issues that could change practice patterns and safely lower cesarean rates. Comparison of cesarean rates by specific Robson criteria can also minimize the contention that differences in patient populations are responsible for differences in cesarean rates. In our study, Hospital A's active labor management practices of oxytocin administration and increased use of internal monitoring (intrauterine pressure catheter or fetal scalp electrode) resulted in higher vaginal delivery rates for primigravidas with no increased maternal or neonatal morbidity. Because safe reduction in cesarean delivery rates for primigravidas will proportionately reduce the number of repeat cesarean delivery required (and therefore contribute to lowering the overall cesarean delivery rate), benchmarking practices as described in this study can be considered in obstetric practices interested in long-term reductions of their cesarean delivery rates.
1. Myers S, Gleicher N. A successful program to lower cesarean-section rates. N Engl J Med 1988;319:1511–6.
2. Placek PJ, Taffel SM, Moien M. Cesarean rates increases in 1985. Am J Public Health 1987;77:241–2.
3. Flamm BL, Lim OW, Jones C, Fallon D, Newman LA, Mantis JK. Vaginal birth after cesarean section: results of a multicenter study. Am J Obstet Gynecol 1988;158:1079–84.
4. O'Driscoll K, Foley M, MacDonald D. Active management of labor as an alternative to cesarean section for dystocia. Obstet Gynecol 1984;63:485–90.
5. Rouse DJ, Owen J, Savage KG, Hauth JC. Active phase labor arrest: revisiting the 2-hour minimum. Obstet Gynecol 2001;98:550–4.
6. Lopez-Zeno JA, Peaceman AM, Adashek JA, Socol ML. A controlled trial of a program for the active management of labor. N Engl J Med 1992;13:326:450–4.
7. Peaceman AM, Socol ML. Active management of labor. Am J Obstet Gynecol 1996;175:363–8.
8. Varner MW, Shah G, Bloebaum L. Toward optimum cesarean rates in Utah. Am J Obstet Gynecol 2002;187:S106.
9. American College of Obstetricians and Gynecologists. Vaginal birth after previous cesarean delivery. ACOG Practice Bulletin 5. Washington, DC: ACOG; 1999.
12. Robson M. Classification of cesarean sections. Fetal Matern Med Rev 2001;12:23–9.
13. Turcot L, Marcoux S, Fraser WD. Multivariate analysis of risk factors for operative delivery in nulliparous women. Canadian Early Amniotomy Study Group. Am J Obstet Gynecol 1997;176:395–402.
14. Thom MH, Chan KK, Studd JW. Outcome of normal and dysfunctional labor in different racial groups. Am J Obstet Gynecol 1979;135:495–8.
15. Naide J, Deshpande P. Using active management of labor and vaginal birth after previous cesarean delivery to lower cesarean delivery rates: a 10 year experience. Am J Obstet Gynecol 2001;184:1535–41.
16. Glantz J, McNanley T. Active management of labor: a meta-analysis of cesarean delivery rates for dystocia in nulliparas. Obstet Gynecol Surv 1997;52:497–505.
17. Bohra U, Donnelly J, O'Connell M, Geary M, MacQuillan K, Keane DP. Active management of labour revisited: the first 1000 primiparous labours in 2000. J Obstet Gynaecol 2003;23:118–20.
18. Tabowei T, Oboro V. Active management of labour in a district hospital setting. J Obstet Gynaecol 2003;23:9–12.
19. Lagrew D Jr., Adashek J. Lowering the cesarean section rate in a private hospital: comparison of individual physicians’ rates, risk factors, and outcomes. Am J Obstet Gynecol 1998;178:1207–14.
20. Pattinson RC, Howarth GR, Mdluli W, Macdonald AP, Makin D, Funk M. Aggressive or expectant management of labour: a randomised clinical trial. BJOG 2003;110:457–61.
21. O'Driscoll K, Meagher D. Active management of labour: the Dublin experience. 3rd ed. London (UK): Mosby; 1993.
22. Main E. Reducing cesarean birth rates with data-driven quality improvement activities. Pediatrics 1999;103(suppl):374–83.
23. Janni W, Schiessel B, Penchers U, Huber S, Strohl B, Hantschmann P, et al. The prognostic impact of a prolonged second stage of labor on maternal and fetal outcome. Acta Obstet Gynecol Scand 2002;81:214–21.