Rates of cesarean delivery vary markedly in the United States, despite practice recommendations promulgated by the National Institutes of Health and ACOG.1,2 In 1999, the overall cesarean delivery rate in the United States increased to 22% from 21.2% in 1998, with considerable variation in rates across states, from 14.8% in Alaska to 27.3% in Mississippi.3 In California, the cesarean delivery rate was 22.7% in 1999, up from 21.7% in 1998.3
As the overall cesarean rate increased in the 1970s and 80s, the rate of elective repeat cesareans accounted for an increasing proportion of the total cesarean deliveries. In 1998 in California, repeat cesareans comprised nearly one third of all cesareans,4 and in the United States 37% of all cesareans were repeats.5 Vaginal birth after cesarean (VBAC) has been shown to be a safe alternative to repeat cesarean delivery for many women.6 Although recent studies have shown a higher risk of maternal morbidity among unsuccessful VBACs than for elective repeat cesareans, overall complication rates are low, and successful VBAC has lower maternal morbidity than elective repeat cesarean.7
An important approach to lowering or stabilizing the cesarean delivery rate is to offer trial of labor to women with one or two previous low transverse cesareans.6 The rate of VBAC in the United States increased steadily from 18.9% in 1989 to 28.3% in 1996,5 and then decreased to 23.4% in 1999.3 The overall cesarean rate is likely affected by whether eligible women are consistently counseled about this option and offered the opportunity for informed consent. Some of the substantial variation seen in cesarean delivery rates at the hospital level4,6 might be related to policies regarding trial of labor. We studied prenatal and obstetric medical records from a representative, stratified, random sample of California hospitals to determine whether the rate at which women were offered and consented to trial of labor after cesarean was associated with differences in risk-adjusted hospital cesarean delivery rates.
MATERIALS AND METHODS
Hospitals were selected for this study based on their participation in a previous study of maternal readmission after vaginal or cesarean delivery. That study used data for the years 1992–1993 from the California Patient Discharge Data System, which is maintained by the Office of Statewide Health Planning and Development, to develop a logistic regression model to estimate each woman's risk of readmission for a postpartum complication within 6 weeks of delivery.8 To create the sampling frame, the 267 nonfederal acute-care hospitals in California that performed more than 678 deliveries in 1992–1993 were stratified according to their ownership status and risk-adjusted readmission rate. Forty-six hospitals were sampled within strata by random number generation to participate in this study. Five hospitals declined to participate and were replaced by ten randomly selected alternates. The number of hospitals within each stratum was chosen to ensure an adequate number of cesarean and vaginal deliveries. These 51 hospitals were representative of all California hospitals, except that Kaiser Permanente hospitals (a large group model health maintenance organization) were purposefully oversampled, so that their labor and postpartum management practices could be studied more carefully.
Information for deliveries in all California hospitals in 1992–1993 was used to develop a logistic regression model to estimate each patient's risk of cesarean delivery,8 in accord with other published models.9 This model adjusted for maternal demographic and clinical characteristics and interactions among those characteristics (Table 1). We aggregated these individual probabilities across all included patients at each hospital to estimate the hospital's expected cesarean rate. Risk-adjusted cesarean rates were estimated by indirect standardization (observed rate/expected rate) statewide rate. The sampled hospitals were then classified according to their risk-adjusted cesarean rates as high (75th percentile or greater), intermediate (16th–74th percentile), or low (15th percentile or lower) based on their statewide ranking. Because many sampled hospitals had relatively low cesarean rates, the “low” group was defined as those at the lowest end of the spectrum (with a mean cesarean rate of 15%), whereas the “high” group had a mean cesarean rate of 30%.
The protocol for this study was approved by the California Health and Welfare Agency's Committee for the Protection of Human Subjects, and exempted from review by the University of California Davis Human Subjects Research Committee. A file of all 1992–1993 obstetric admissions to California hospitals resulting in a delivery had been created previously, based on the appropriate International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) codes.7 For each participating hospital, we generated a random sample of births (including stillbirths) from 1992–1993 proportional to the total number of deliveries at that hospital. To provide an adequate number of cesarean deliveries, cesarean deliveries were oversampled so that about 50% of requested records were cesarean deliveries. A total of 1662 births were sampled, of which the 369 women with previous cesareans were included in the current study.
Medical records were abstracted in two phases. First, one of four accredited records technicians or certified coding specialists abstracted maternal demographic data, ICD-9-CM diagnostic and procedural codes, indications for cesarean delivery, prenatal history, admission history, delivery outcomes, and postpartum complications. They were supervised by an experienced medical records consultant, who verified at least 10% of each abstractor's records to ensure greater than 95% accuracy on each data element. In the second phase, one of two experienced obstetric nurses abstracted more complex clinical data, including documentation in the prenatal record of counseling for a trial of labor, as well as other prenatal and intrapartum clinical variables. Periodic double abstraction was conducted for these nurses, initially with the Principal Investigator and subsequently between nurse abstractors at specified intervals. A separate database was used to generate detailed abstraction reports. Discrepancies were discussed at weekly project meetings, and resolution was achieved by guideline and obstetric practice review.
All analyses were weighted to account for oversampling of cesarean deliveries; weights were constructed as the inverse of the probability of sampling each record at that hospital. Consequently, the weighed results provided unbiased estimates of patient characteristics and processes of care for all deliveries, not just sampled deliveries, at the 51 hospitals. Weighted proportions for variables of interest were examined for hospitals grouped according to high, intermediate, and low risk-adjusted cesarean delivery rates. Further multivariable adjustment was not done because all available patient characteristics had already been considered in the first-order logistic regression model that was used to classify hospitals as having high, intermediate, or low risk-adjusted cesarean rates. Differences in proportions were compared using the first-order Taylor series linear approximation for estimation of variance. This approximation was required because subjects were clustered within hospitals and therefore could not be considered independent observations.10 To account for oversampling of cesarean deliveries and clustering of nonindependent observations within hospitals, we applied the second-order Rao and Scott11 correction of the Pearson χ2 statistic. This corrected χ2 statistic was then converted to an F statistic by dividing it by its degrees of freedom. These procedures were implemented using the svytab options in Stata Release 6 (Stata Corporation, College Station, TX).
Of 1662 records requested, 1573 were abstracted and included in the total sample; 73 records were missing and 16 records were excluded (three were not a delivery, eight were fetal deaths, two were incomplete, and three were emergency cesareans without labor).
Characteristics of the 51 sampled hospitals are shown in Table 2. Risk-adjusted cesarean rates among hospitals ranged from a low of 13% to a high of 37%. Hospitals were distributed geographically throughout California. In the sample, all hospitals with low risk-adjusted cesarean rates were hospitals of a single staff-model health maintenance organization.
Of records abstracted, 798 were vaginal deliveries and 775 were cesareans. Of the cesareans, 477 were primary and 298 were repeat operations. Indications for the cesareans, according to the hierarchy of Henry and colleagues,12 included elective repeat 40%, dystocia 16%, breech 7%, and other 28%. Three hundred sixty-nine women had prior cesareans and were included in the current study. These women were categorized according to their eligibility for trial of labor. Fifty-seven women with clear indications for repeat cesarean, including vertical uterine incision (n = 12), three or more previous cesareans (n = 22), breech or other malpresentation (n = 20), placenta previa (n = 3), or active herpes lesions at the time of labor (n = 0) were considered ineligible. After those exclusions, 312 women were considered potentially eligible for trial of labor. Characteristics of those women are shown in Table 3. These characteristics did not differ from those of the overall group of women who had a prior cesarean delivery (data not shown). In addition, the percentage of women eligible for a trial of labor was similar at hospitals with high, intermediate, and low risk-adjusted cesarean rates (87%, 84%, and 93%, respectively).
We searched for documentation of counseling regarding VBAC throughout the prenatal and hospital records. Most often, counseling was documented in the prenatal records or hospital history and physical. Refusal of VBAC was often documented in the history and physical. There were no cases where VBAC occurred without documentation of counseling.
Weighted proportions of women scheduled for a repeat cesarean without documentation of counseling for trial of labor, those who refused trial of labor, failed a trial of labor, and had successful VBAC are shown in Table 4. Hospitals with high risk-adjusted cesarean rates were significantly more likely than hospitals with low rates to have scheduled a repeat cesarean without documented counseling of women for trial of labor (21% and 0.3%, respectively) and had a significantly larger proportion of women who refused a trial of labor among those who received counseling (36% and 10%, respectively). The completed VBAC rate in hospitals with high risk-adjusted cesarean rates was less than half that in hospitals with low rates (31% and 71%, respectively). The proportions of women who failed a trial of labor were similar in all three groups of hospitals, indicating that the higher success rate in the hospitals with low risk-adjusted cesarean rates was attributable to the higher proportion of women who were offered and subsequently agreed to undergo a trial of labor.
In this study of a representative, stratified random sample of California hospitals, facilities with low risk-adjusted cesarean rates offered a trial of labor to women with previous cesarean significantly more often than hospitals with high cesarean rates. This offer was accepted more frequently, and the completed VBAC rate was significantly higher in the hospitals with low cesarean rates. It has been shown that a trial of labor for women with a previous cesarean and without contraindications is a safe alternative to repeat cesarean delivery in hospitals with the capacity for rapid emergency cesarean delivery.6 Although large studies have found a higher risk of uterine rupture with trial of labor, the rate of that adverse outcome is 1.5% or less.7,13,14 The manner in which counseling on trial of labor is provided might influence patient choice. It is not known how much of the decline in the VBAC rate during the past several years relates to provider concern about malpractice issues or to patient preferences. If obstetric care providers are not interested in performing VBACs, they might counsel against a trial of labor, or they might schedule a repeat cesarean without documenting counseling regarding the alternative.
In our study, the proportion of failed trials of labor did not differ among hospitals with high, intermediate, and low risk-adjusted cesarean rates. This finding suggests that once a patient had been counseled and consented to a trial of labor, she had an equal likelihood of delivering vaginally regardless of the hospital cesarean rate. Our study is limited by its observational nature, which makes it more susceptible to confounding than a randomized trial. Because of the importance of patient preference in choosing a trial of labor versus cesarean delivery, no randomized study of trial of labor and scheduled repeat cesarean delivery has been conducted. Hospitals run by health maintenance organizations were overrepresented in our sample and tended to be in the low cesarean group. In addition, we could only examine clinical practice as documented in the clinical record, which is unlikely to capture the complete picture of patient care. In particular, hospitals can differ systematically in the extent to which they encourage complete and accurate documentation of counseling on trial of labor. What physicians or midwives actually say when counseling patients might not be reflected by the documentation. A biased presentation might influence patient choice. Finally, our study was not designed to examine outcomes of labor and delivery.
Our findings are strengthened by the statewide representation and the large number of hospitals from which records were examined. Further, because we directly examined the medical records, we could examine the clinical indications for repeat cesareans and actual documentation of counseling regarding trial of labor, which has not been possible in studies based on state databases or hospital discharge abstracts.
The safety of trial of labor has been the subject of controversy for many years.6,14–16 Increasing rates of trial of labor in the late 1980s and early 1990s were followed by large cohort studies documenting the small but increased risk of uterine rupture and other maternal morbidity in failed trials of labor.6,7,14 The decrease in VBAC rates in the United States since 1996 and the subsequent increase in the cesarean rate might reflect patients' and clinicians' concerns regarding adverse outcomes, as well as clinicians' concerns about medicolegal consequences.14 It remains clear, however, that trial of labor is a safe alternative to repeat cesarean in most cases and is an important component of achieving a cesarean rate that is comparable to those in European countries with low rates of maternal and infant morbidity and mortality.6
All perspectives in the controversy surrounding trial of labor acknowledge the importance of informed patient choice. Counseling about trial of labor and documentation of counseling in the medical record are critical elements of this process. Documentation of informed consent is essential for medicolegal purposes. The markedly higher rates of women who have scheduled repeat cesarean without documentation of counseling regarding trial of labor at hospitals with high cesarean rates is cause for concern. Future studies and interventions should be designed to improve communication between physicians and their pregnant patients with a history of prior cesarean delivery. The elements of informed consent for VBAC might be best provided to the patient through specific hospital documentation policies for all obstetric patients with a previous cesarean. Our results and previous studies of physician opinion leaders17 suggest that such communication could represent one strategy for safely reducing overall cesarean rates.
1. National Institutes of Health. NIH consensus development statement on cesarean childbirth. Obstet Gynecol 1981;57:537–545.
2. American College of Obstetricians and Gynecologists. Maternal and fetal medicine. Guidelines for vaginal delivery after a previous cesarean birth. ACOG committee opinion no. 64. Washington DC: American College of Obstetricians and Gynecologists, 1988.
3. Curtin SC, Martin JA. Births: Preliminary data for 1999. National Vital Statistics Reports. Vol. 48. No.14. Hyattsville, Maryland: National Center for Health Statistics, 2000.
4. Danielson B, Castles A. Risk-adjusted cesarean section rates for California hospitals 1995 to 1997. San Francisco Report for the California Pacific Business Group on Health, 1999.
5. Ventura SJ, Martin JA, Curtin SC, Mathews TJ, Park MM. Births: Final data for 1998. National Vital Statistics Reports. Vol. 48. No.3. Hyattsville, Maryland: National Center for Health Statistics, 2000.
6. American College of Obstetricians and Gynecologists Task Force on Cesarean Delivery. Evaluation of cesarean delivery. Washington DC: American College of Obstetricians and Gynecologists, 2000.
7. McMahon MJ, Luther ER, Bowes WA Jr, Olshan AF. Comparison of a trial of labor with an elective second cesarean section. N Engl J Med 1996;335:689–95.
8. Romano PS, Luft HS, Remy LL, Rainwater J, Rennie DJ, Zhou H, et al. Report of the California Hospital Outcomes Project. Maternal outcomes following delivery. Risk adjustment methodology and preliminary findings. Sacramento, California: Office of Statewide Health Planning and Development; 1996.™
9. Keeler EB, Park RE, Bell RM, Gifford DS, Keesey J. Adjusting cesarean delivery rates for case mix. Health Services Res 1997;32:511–28.
10. Rao JNK, Scott AJ. On chi-squared tests for multiway contingency tables with cell proportions estimated from survey data. Ann Stat 1984;12:46–60.
11. Rao JNK, Scott AJ. The analysis of categorical data from complex sample surveys: Chi-squared tests for goodness of fit and independence in two-way tables. J Am Stat Assoc 1981;76:221–30.
12. Henry OA, Gregory KD, Hobel CJ, Platt LD. Using ICD-9 codes to identify indications for primary and repeat cesarean sections: Agreement with clinical records. Am J Public Health 1985;75:1143–5.
13. Gregory KD, Korst LM, Cane P, Platt LD, Kahn K. Vaginal birth after cesarean and uterine rupture rates in California. Obstet Gynecol 1999;94:985–99.
14. Flamm BL. Once a cesarean, always a controversy. Obstet Gynecol 1997;90:312–5.
15. Sachs BP, Kobelin C, Castro MA, Frigoletto F. The risks of lowering the cesarean-delivery rate. N Engl J Med 1999;340:54–7.
16. Boyers SP, Gilbert WM. Elective repeat cesarean section versus trial of labour: The neonatologists view. Lancet 1999;351:155.
17. Lomas J, Enkin M, Anderson GM, Hannah WJ, Vayda E, Singer J. Opinion leaders vs audit and feedback to implement practice guidelines. Delivery after previous cesarean section. JAMA 1991;265:2202–7.