Linton, Andrea MS; Peterson, Michael R. DVM, DrPH; Williams, Thomas V. PhD
Over 3 decades of study on the rise in cesarean delivery rates in the United States and worldwide have identified a number of significant associations between cesarean rates and maternal, provider, institutional, and health plan–related factors. Initially, research focused on isolating the root cause of cesarean rate escalation with the intent of reducing or arresting further cesarean rate increases. More recently, studies have focused instead on identifying and addressing variation in cesarean delivery that may not be clinically justified. A number of case-mix adjustment methodologies have been proposed to support more meaningful comparisons of cesarean use between individual patients,1 physicians,2 hospitals,3–5 or health plans.6 These adjustment methodologies are based on the underlying assumption that cesarean rate variations that persist following a case-mix adjustment are attributable to differences in practice management and therefore represent an opportunity for utilization review and standardization.
We undertook a study to assess the extent to which differences in clinical case mix may explain cesarean rate variation observed among the Department of Defense health care beneficiaries who deliver in military hospitals. A study of this population provides several advantages. Despite the wide racial, cultural, and geographic diversity of Department of Defense beneficiaries, maternal socioeconomic status is generally not a significant indicator for cesarean delivery, presumably as a result of the uniform health care entitlement that guarantees equal access to health care services, regardless of the beneficiary's location, ability to pay, or prior medical condition.7 Additionally, Department of Defense beneficiaries receive health care services through a managed care plan or other types of health plan similar to those available in the commercial health care industry, all of which suggests that findings from studies of this population may have relevance for the national population.
Examination of the Department of Defense as a health care provider offers further opportunities for controlled study and comparison with the civilian population. All military hospitals are operated by the U.S. federal government, thus controlling for hospital ownership, a variable that impacts cesarean rate studies of the national population.8–12 Department of Defense hospitals are also geographically distributed across the United States, as well as Europe and Asia, enabling comparison with regional variations in cesarean rates that have persisted nationally for the past decade.13,14 These hospitals are, however, diverse in their environmental setting, size, depth, and breadth of health care services supported by each hospital, allowing comparison with civilian-sector cesarean trends associated with a hospital's delivery volume, teaching status, or presence of an neonatal intensive care unit (NICU).9,10,12,15,16
We made a comparison of observed and case mix–adjusted cesarean rates for live deliveries in U.S. military hospitals worldwide in 2002. We hypothesized that predicted rates based on clinical case mix would agree with the observed cesarean rates within a 95% confidence level for subgroups defined by maternal and hospital characteristics. Our methodology focused on adjusting for objectively diagnosed clinical conditions that have been previously associated with increased likelihood of cesarean delivery.
MATERIALS AND METHODS
The study was found to be exempt by the Human Subjects Research Review Board Office of Regulatory Compliance and Quality. The study sample included all live deliveries that occurred in U.S. military hospitals during the calendar year 2002. Electronic discharge records for 53,215 live deliveries were compiled from 66 hospitals worldwide, including 48 hospitals located within the U.S. and 18 hospitals located in the U.S. territories and throughout Europe and Asia. The Department of Defense maintains enterprise-wide information systems that collect and consolidate electronic discharge records for every inpatient stay. The discharge records are coded from inpatient medical documentation per Joint Commission Accreditation of Healthcare Organizations standards and then assigned to a diagnosis-related group using a Department of Defense standard inpatient grouping program. Records are forwarded to a central data repository every month where they are processed for validity and consistency. A census of delivery records from 2002 was extracted from the central repository using SAS 8.2 software (SAS Institute, Cary, NC).
To identify method of delivery, we used diagnosis-related groups from the discharge data. The occurrence of diagnosis-related groups 370 and 371 (cesarean delivery with and without complication, respectively) was used to identify cesarean births. The occurrence of International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) diagnosis code 654.2x (previous cesarean delivery) in combination with diagnosis-related groups 370 and 371 was used to identify women who had had a previous cesarean delivery. Data associated with stillbirths, abortions, or other non–live births were eliminated from the study.
Clinical conditions used to adjust cesarean rates for clinical case mix were identified using ICD-9-CM diagnostic codes listed in the electronic record (up to 8 diagnostic codes may be entered on the discharge record). The clinical conditions selected for inclusion in our study were modified from an adjustment methodology used by Peaceman, Feinglass, and Manheim17 and included previous cesarean delivery, multiple gestation, intrapartum complications (abruption placentae, placenta previa, and malpresentation), and medical risk factors for cesarean (hypertension, diabetes mellitus, excessive fetal growth, hydramnios, oligohydramnios, genital herpes, and human immunodeficiency virus [HIV]). Diagnoses of dystocia and fetal distress were excluded from our adjustment methodology because it has been suggested by Peaceman et al17 and others4,18 that these diagnoses are imprecise and may reflect post ad hoc justifications of cesarean use, rather than objectively assessed conditions. We added genital herpes and HIV to our list of medical risk factors, because both were positively associated with cesarean delivery in previous studies of the study population and elsewhere.19–21 The ICD-9 code for excessive fetal growth (656.6) was added as a proxy for high gestational weight. Maternal age was included as a risk factor in the adjustment methodology because of the general perception that women 35 years and older face a higher risk during pregnancy than their younger counterparts, even in the absence of clinical indications.22,23 We elected to exclude nonclinical maternal characteristics, including maternal race and health plan type, from our adjustment methodology. Instead, we treated these factors as independent variables in our analysis.1 The ICD-9-CM diagnosis codes used to identify each clinical condition are provided in Table 1. A logistic regression analysis was performed to confirm that each clinical condition was a significant indicator for a cesarean outcome in the study population.
The presence or absence of one or more clinical conditions was used to assign each delivery record to a hierarchy of 18 mutually exclusive strata. The first 9 strata included women under the age of 35 years. The first and second strata included all prior cesarean deliveries with and without one or more comorbid conditions, respectively. The third and fourth strata included multiple gestations with or without one or more comorbid conditions, respectively, and with no indication of a prior cesarean delivery. The fifth and sixth strata included intrapartum complications with or without one or more medical risk factors, respectively, and no multiple gestations or prior cesarean deliveries. The seventh and eighth strata included deliveries with multiple or one medical risk factor, respectively, but no intrapartum complications, multiple gestations, or prior cesarean deliveries. The ninth and final stratum included all remaining deliveries that did not fit into one of the other strata, ie, deliveries that contained no indication for any of the clinical conditions examined in this study. The remaining 9 strata included women 35 years and older and repeated the clinical conditions established for the first 9 strata. The distribution of total deliveries, cesarean deliveries, and cesarean delivery rates across each stratum is presented in Table 2.
Observed and predicted cesarean rates (case mix–adjusted rates) were calculated for subgroups defined by institutional characteristics (hospital location, delivery volume, teaching status, and presence of a NICU) and maternal characteristics (race and health plan). Hospital location was defined as 4 regions in the United States (Northeast, South, Midwest, West) and one for outside of the United States. The U.S. regions included in this study are consistent with those defined for studies of the national population.13 Delivery volume was defined by the number of live deliveries that took place in the hospital in 2002 (< 500, 500–1,199, 1,200–2,000, and > 2,000). Teaching status and NICU were defined dichotomously based on the presence of an active graduate medical education program in obstetrics or active NICU, respectively, in 2002. Four self-reported, maternal race categories were used (white, black, Asian, and other). The “other” race category includes American Indians, multiracial women, and women who chose not to classify themselves as white, black, or Asian. Hispanic ethnicity was not available from the data set; thus, no ethnicity-based comparisons were made. The health plan variable was dichotomized into TRICARE Prime if patients used the Department's managed care health plan and non-Prime if patients used one of the Department's preferred provider or fee-for-service plans.
Statistical Package for Social Sciences (SPSS 10.1.3, SPSS Inc, Chicago, IL) was used for the calculation and comparison of observed and predicted cesarean rates for each subgroup. The methodology for calculating predicted cesarean rates is similar to that used by the National Center for Health Statistics to age-adjust mortality rates.24,25 The predicted or case mix–adjusted rate is the rate that would be expected if all study subjects with the same clinical profile (ie, assigned to the same stratum) faced the same probability of delivering by cesarean delivery. Stratum-specific cesarean rates were calculated, and each woman's probability of cesarean delivery was then set equal to the rate associated with the clinical stratum to which she was assigned. Predicted cesarean rates were determined by calculating the mean probability of cesarean delivery for all women included in each subgroup. To determine if the observed and predicted cesarean rates for a subgroup differed significantly, a 95% confidence interval was calculated around the standardized ratio of the observed cesarean rate to the predicted cesarean rate. Algorithms for calculating the standardized ratio, standard error, and 95% confidence intervals are shown here.25
Standardized ratio (SR) = observed cesarean rate/predicted cesarean rate
Standard error (SE) = SR/square root of the number of observed cesarean deliveries
95% Confidence interval (CI) = SR ± (1.96 × SE)
The number of deliveries, cesarean deliveries, and cesarean delivery rate associated with each clinical condition used in our case mix–adjustment methodology is presented in Table 1. The condition-specific cesarean rates ranged from 29.2% to 73.6% for oligohydramnios and previous cesarean delivery, respectively. All clinical conditions were significant indicators for a cesarean outcome. Previous cesarean delivery was the strongest predictor of cesarean delivery (adjusted odds ration [AOR] 13.14, 95% CI 11.86–14.56), and diabetes mellitus was the weakest predictor of cesarean delivery (AOR 1.34, 95% CI 1.20–1.49).
The number of deliveries, cesarean deliveries, and cesarean delivery rate associated with each of the 18 clinical strata are presented in Table 2. The strata-specific cesarean rates ranged from 10.5% (stratum 9, under age 35 with no medical risk factors or complications) to 84.3% (stratum 11, age 35 or older with previous cesarean delivery and 1 or more comorbid conditions). Stratum 9 contained the largest number of deliveries, comprising roughly 61% of the total records examined in this study.
A comparison of the observed and predicted cesarean rates across institutional and maternal subgroups is presented in Table 3. The majority of deliveries, 46,656 of 53,215 (88%), took place in 48 hospitals located in the United States. The South had the largest number of hospitals and deliveries, followed by the West, Midwest, and Northeast. The Northeast and the Midwest were associated with the lowest and highest observed cesarean rates, respectively (19.7% and 23.4%). The lowest and highest predicted cesarean rates were observed for the Northeast and South (20.1% and 22.0%, respectively). No significant differences between observed and predicted cesarean rates were found for any location-based subgroup. The observed and predicted cesarean rates inside and outside the United States were similar.
Roughly half of all deliveries in 2002 occurred in one of the 55 hospitals with a delivery volume of less than 1,200. Observed cesarean rates ranged from 20.5% to 23.1% in a bimodal pattern, reflecting the highest observed cesarean rates for the hospitals with fewer than 500 deliveries and more than 2,000 deliveries (23.1% and 23.0%, respectively). In contrast, the predicted cesarean rate was positively associated with increasing delivery volume (20.4% and 22.5% for the smallest and largest hospital subgroups, respectively). The observed cesarean rate was significantly higher than the predicted rate in hospitals with fewer than 500 deliveries (SR 1.13, 95% CI 1.08–1.19) and significantly lower than the predicted rate in hospitals with workload between 1,200 and 2,000 deliveries (SR 0.92, 95% CI 0.88–0.96).
Approximately one third of deliveries (17,342 deliveries) in this study occurred in the 9 teaching hospitals. Teaching hospitals were associated with higher observed and predicted cesarean rates than nonteaching hospitals. The observed cesarean rate in teaching hospitals was significantly higher than the predicted rate (SR 1.05, 95% CI 1.02–1.08). Conversely, the observed cesarean rate in nonteaching hospitals was significantly lower than the predicted rate (SR 0.97, 95% CI 0.95–0.99).
A total of 12 hospitals in the study sample had an active NICU during 2002. Both observed and predicted cesarean rates were higher in hospitals with active NICUs. The hospitals with NICUs had a higher-than-predicted cesarean rate, and those without NICUs had a lower-than-predicted cesarean rate, but the differences between the observed and predicted cesarean rates in each subgroup were not statistically significant. We performed an additional calculation to compare observed and predicted cesarean rates in the 9 teaching hospitals with NICUs and the 3 nonteaching hospitals with NICUs. The observed cesarean delivery rates were significantly higher than expected for teaching hospitals with NICUs (SR 1.06, 95% CI 1.02–1.09) and significantly lower for the nonteaching hospitals with NICUs (SR 0.92, 95% CI 0.87–0.97).
Observed cesarean delivery rates across race subgroups ranged from 20.5% for white woman to 25.1% for black women. Predicted cesarean delivery rates ranged from 21.6% for white and other women to 22.8% for black women. The observed cesarean rate for white mothers was significantly lower than expected (SR 0.95, 95% CI 0.93–0.97), and observed cesarean rates for black and “other” mothers were significantly higher than expected (SR 1.10, 95% CI 1.05–1.14 and SR 1.05, 95% CI 1.01–1.09 for blacks and others, respectively). Observed and predicted cesarean rates for Asian women were similar. Additional calculations performed to assess the impact of hospital teaching status on cesarean rates for race subgroups yielded significantly higher-than-predicted rates for black women in teaching hospitals (SR 1.20, 95% CI 1.12–1.28). Although not statistically significant, higher values for observed cesarean rates relative to predicted rates were found for white (SR 1.01, 95% CI 0.97–1.06) and other mothers (SR 1.07, 95% CI 0.99–1.14) as well (tabular data not shown).
Approximately 90% (47,965) of deliveries were to beneficiaries using TRICARE Prime, the Department's managed care health plan. TRICARE Prime beneficiaries were associated with a lower observed cesarean rate, but a higher predicted cesarean rate relative to their non-Prime counterparts. The observed cesarean rate for non-Prime beneficiaries was significantly higher than their predicted rate (SR 1.10, 95% CI 1.04–1.16). An additional calculation of the observed and predicted cesarean rates for non-Prime beneficiaries found that more than half of all deliveries to non-Prime beneficiaries occurred in teaching hospitals. Significantly higher-than-predicted cesarean rates were found for deliveries in teaching hospitals (SR 1.23, 95% CI 1.13–1.33), but observed and predicted cesarean rates were statistically similar for non-Prime births in nonteaching hospitals (SR 1.00, 95% CI 0.91–1.08).
The findings of this study provide mixed support for our initial hypothesis. Case mix–predicted cesarean rates differed across subgroups, indicating that subgroups should be expected to have different cesarean delivery rates because of their differing case mix of patients. For subgroups defined by hospital location and NICU status, the observed cesarean rate was statistically similar to the rate predicted on the basis of the case mix of their patients. This was not the case for delivery volume, teaching status, maternal race, or health plan subgroups, however, where statistically significant differences between observed and predicted cesarean rates were found. This finding suggests that differences in practice management, as well as clinical case mix, contributed to variations in observed cesarean use for specific hospital settings and maternal subgroups.
Our findings were not consistent with studies of the national population that found persistently higher cesarean rates in the South, followed by the Northeast, and then the West and Midwest.13,14 The absence of significant differences in cesarean rates across locations is not surprising. The high degree of mobility for military physicians suggests that regional patterns of practice management are less likely to influence their clinical decision making over time. The high mobility of the patients (active duty service members and their families) suggests that even nonmilitary physicians who render care in military hospitals are exposed to a greater demographic diversity of patients than they would be if treating an exclusively civilian population. This expanded realm of clinical experience may provide a more uniform basis for clinical decision making that may also contribute to typically lower cesarean rates observed for Department of Defense beneficiaries when compared with the national population.7
Adjustment for case mix failed to explain the observed bimodal pattern of cesarean delivery across delivery volume subgroups. The smallest hospitals (< 500 deliveries) had the least complex case mix of patients but the highest observed cesarean rates. The relatively low predicted cesarean rate for this subgroup may be explained by anecdotal evidence that more complicated obstetrical cases were referred out of smaller hospitals that may not have been adequately staffed or resourced to support high-risk deliveries. Pregnancies initially considered low risk, however, may still have encountered problems during the onset of labor or delivery. In these situations, physicians in a resource-limited environment may have elected to deliver by cesarean rather than risk a complicated delivery for which the hospital was not properly equipped. This explanation is consistent with the higher observed cesarean rates in small hospitals, as well as with studies of cesarean use in the national population, which have suggested that changes in pregnancy and labor management practices may promote cesarean use as an alternative before entering into a potentially complicated labor.26,27
The literature contains mixed findings regarding the influence of teaching status on a hospital's cesarean rate.9,10,12,15,16 In our study, military teaching hospitals had a significantly higher cesarean rate than would be predicted on the basis of case mix. The rationale offered to explain the higher cesarean rates observed for smaller hospitals is not consistent with this observation. Although the 9 military teaching hospitals vary in size (at least 1 teaching hospital falls into each delivery volume subgroup), these facilities are generally staffed and equipped sufficiently to ensure that their graduate medical education programs can maintain their accreditation. Each of these teaching facilities has a NICU, which has generally been associated with lower cesarean rates in the literature.10,16 Furthermore, the disparity in observed cesarean rates between Prime and non-Prime beneficiaries appears to be attributable to the teaching status of the facility in which they delivered. Together, these findings suggests practice management in a teaching environment that may have promoted greater use of cesarean delivery than that in a nonteaching environment.
Predicted cesarean rates differed significantly from the observed rate for 3 of the 4 race subgroups in this study. This finding suggests either a weakness in the adjustment methodology or variations in practice management based on the race of the mother. The persistently higher cesarean rates for black women relative to whites and Asians in the study population and the national population warrants continued study.21
Advantages of this study include the large diverse population, the quality and completeness of the data source, and an ability to control certain variables that may confound studies of the national population, as previously discussed. Although not immune from the effects of selection bias, hospital and health plan selection bias is also likely to be diminished in the study population. TRICARE policy during 2002 mandated that all obstetrical patients receive care from a military hospital if one was located within 40 miles of their residence. Active duty women are de facto enrolled in TRICARE Prime, whereas family members and retirees are offered enrollment only if it is available in their area.
Several limitations must be considered when interpreting these findings, one of which involves potential risk factors for cesarean that were not included in our adjustment methodology. These include gestational age and congenital anomalies of the newborn, neither of which was available from the study data set. Neither was maternal parity readily available. Past studies have shown that primiparity is a significant risk factor for primary cesarean across all maternal age groups.22 In the absence of adjustment for parity, cesarean rate variation may reflect differing ratios of primiparous to multiparous women across subgroups.
Although our methodology made a broad attempt to account for the presence of comorbid conditions, we did not address cesarean rate differences for specific combinations of comorbidities. To do so would likely have yielded a more refined estimate of the predicted cesarean rate, but it is not clear that the overall findings would have been altered by the creation of additional clinical strata in the adjustment methodology.
The potential for inconsistencies in the coding of birth records represents a challenge to the accurate assessment of cesarean delivery rates in both the study population and other populations. Variability in the interpretation and coding of obstetric complications for a patient population of the size, diversity, and distribution of our study population is inherent. Because this study used the presence of specific ICD-9 codes in the discharge records to identify clinical complications, omission or misassignment of a code may have resulted in the patient's assignment to the wrong clinical strata. At a minimum, inconsistent coding would introduce random variation in cesarean rates across subgroups. There is no evidence to assume, however, that the coding inconsistencies would occur more frequently for one subgroup than for another or change the patterns of cesarean delivery observed across study subgroups.
The clinical significance of the statistically significant findings from this study are difficult to assess. Because the predicted cesarean rates for each subgroup were calculated using the study population as the reference rather than an external population or objective benchmark, our findings only speak to the relative difference in rates between subgroups. For example, the differences in observed and predicted cesarean rates for teaching hospitals translates into approximately 49 women in teaching hospitals who had cesarean deliveries that may not have been clinically justified according to mean obstetrical practice across all U.S. military hospitals in 2002. In contrast, 45 women in nonteaching hospitals delivered vaginally despite a clinical profile that predicted a cesarean delivery using the same mean obstetrical practice as the standard. When the size of the study population is considered, this finding indicates that cesarean decision making was not consistent with mean obstetrical practice in less than 0.2% of obstetrical cases. The percentage of patients affected would inevitably change if a different standard were used for case mix adjustment. Whether the Department of Defense mean obstetrical practice represents the best standard for assessing whether cesarean rates are appropriate is not clear. Regardless of the standard used, the contribution of this study lay, not so much with the absolute rates or rate differences, but with the patterns of unexplained cesarean rate variation that guide Department of Defense policy makers in their efforts to promote uniform delivery of obstetrical care throughout the beneficiary population.
Although the case mix–adjusted cesarean rates calculated in this study have little direct applicability for rate comparisons with other populations, our overall findings support previous conclusions that advocate the importance of case mix when comparing cesarean rates across a population. Case mix adjustment exclusively, however, is not adequate for predicting the frequency of cesarean use in all situations. Despite the promulgation of The American College of Obstetricians and Gynecologists and other guidelines designed to standardize clinical practice, clinical decision making appears to be influenced by factors independent of a woman's medical condition. It is reasonable for a physician to consider environmental factors, such as facility staffing and resources available to support a complicated delivery, when making decisions about a patient's course of treatment. An attempt to consider both the case mix of patients and limitations posed by delivery environment is recommended for future studies that focus on assessing and comparing performance across institutions, health plans, or individual providers.
1. Aron DC, Gordon HS, DiGiuseppe DL, Harper DL, Rosenthal GE. Variations in risk-adjusted cesarean delivery rates according to race and health insurance. Med Care 2000;38:35–44.
2. Eldenburg L, Waller WS. Decision-case mix model for analyzing variation in cesarean rates. Med Decis Making 2001;21:170–9.
3. Bailit JL, Dooley SL, Peaceman AN. Risk adjustment for interhospital comparison of primary cesarean rates. Obstet Gynecol 1999;93:1025–30.
4. Glantz JC. Cesarean delivery risk adjustment for regional interhospital comparisons. Am J Obstet Gynecol 1999;181:1425–31.
5. Aron DC, Harper DL, Shepardson LB, Rosenthal GE. Impact of risk-adjusting cesarean delivery rates when reporting hospital performance. JAMA 1998;279:1968–72.
6. Roohan PJ, Josberger RE, Gesten FC. Risk-adjusted primary cesarean delivery rates for managed care plans in New York State, 1998. Matern Child Health J 2001;5:169–177.
7. Linton A, Peterson MR, Williams TV. Effects of maternal characteristics on cesarean delivery rates among U.S. Department of Defense healthcare beneficiaries, 1996–2002. Birth 2004;31:3–11.
8. Placek PJ, Taffel S, Moien M. Cesarean section delivery rates: United States, 1981. Am J Public Health 1983;73:861–2.
9. Williams RL, Chen PM. Controlling the rise in cesarean section rates by the dissemination of information from vital records. Am J Public Health 1983;73:863–7.
10. Tussing AD, Wojtowycz MA. The cesarean decision in New York State, 1986: economic and noneconomic aspects. Med Care 1992;30:529–40.
11. Curtin SC, Kozak LJ, Gregory KD. U.S. cesarean and VBAC rates stalled in the mid-1990s. Birth 2000;27:54–7.
12. Gregory KD, Ramicone E, Chan L, Kahn KL. Cesarean deliveries for medicaid patients: a comparison in public and private hospitals in Los Angeles County. Am J Obstet Gynecol 1999;180:1177–84.
13. Menacker F, Curtin SC. Trends in cesarean birth and vaginal birth after previous cesarean, 1991–99. Natl Vital Stat Rep 2001;49:1–16.
14. Kabir AA, Steinmann WC, Myers L, Khan MM, Herrera EA, Yu S, Jooma N. Unnecessary cesarean delivery in Louisiana: an analysis of birth certificate data. Am J Obstet Gynecol 2004;190:10–9; discussion 3A.
15. Oleske DM, Glandon GL, Giacomelli GJ, Hohmann SF. The cesarean birth rate: influence of hospital teaching status. Health Serv Res 1991;26:325–37.
16. Burns LR, Geller SE, Wholey DR. The effect of physician factors on the cesarean section decision. Med Care 1995;33:365–82.
17. Peaceman AM, Feinglass J, Manheim LM. Risk-adjustment of cesarean delivery rates: a practical method for use in quality improvement. Am J Med Qual 2002;17:113–7.
18. Lieberman E, Lang JM, Heffner LJ, Cohen A. Assessing the role of case mix in cesarean delivery rates. Obstet Gynecol 1998;92:1–7.
19. Preboth M. ACOG practice bulletin on management of herpes in pregnancy. American College of Obstetricians and Gynecologists. Am Fam Physician 2000;61:556, 561.
20. ACOG committee opinion scheduled Cesarean delivery and the prevention of vertical transmission of HIV infection. Number 234. Int J Gynaecol Obstet 2001;73:279–81.
21. Linton A, Peterson MR. The effect of preexisting chronic disease on primary cesarean delivery rates by race for births in U.S. military hospitals, 1999–2002. Birth 2004;31:165–75.
22. Parrish KM, Holt VL, Easterling TR, Connell FA, LoGerfo JP. Effect of changes in maternal age, parity, and birth weight distribution on primary cesarean delivery rates. JAMA 1994;271:443–7.
23. Heffner LJ, Elkin E, Fretts RC. Impact of labor induction, gestational age, and maternal age on cesarean delivery rates. Obstet Gynecol 2003;102:287–93.
24. Anderson RN, Rosenberg HM. Age standardization of death rates: implementation of the year 2000 standard. Natl Vital Stat Rep 1998;47(3):1–16, 20.
25. Arias E. Unites States Life Tables, 2001. Natl Vital Stat Rep 2004;52(14):1–38.
26. Schindl M, Birner P, Reingrabner M, Joura E, Husslein P, Langer M. Elective cesarean section vs. spontaneous delivery: a comparative study of birth experience. Acta Obstet Gynecol Scand 2003;82:834–40.
27. Wilkes PT, Wolf DM, Kronbach DW, Kunze M, Gibbs RS. Risk factors for cesarean delivery at presentation of nulliparous patients in labor. Obstet Gynecol 2003;102:1352–7.