Kabir, Azad A. MB, MSPH*; Pridjian, Gabriella MD†; Steinmann, William C. MD, MSc‡; Herrera, Eduardo A. MD†; Khan, M Mahmud PhD§
In recent years, there has been an increase in surgical procedures, particularly cesareans.1–5 To decrease the number of cesareans, Healthy People 2010 proposed national targets of a 15% primary cesarean rate and a 63% repeat cesarean rate. These targets were chosen to reflect the importance of patient safety or case mix, and to focus the national reduction goal to low-risk nulliparous women.6
In certain clinical situations, vaginal delivery is clearly contraindicated because of maternal, fetal, or delivery characteristics, and cesarean deliveries are necessary. However, the increase in cesarean rates in recent years has been attributed to a variety of causes including changing pregnancy characteristics, medicolegal concerns, obstetrician's practice patterns, work schedules, financial incentives, and patient's, race, age, education, and insurance status.6–12 It is uncertain whether some of these nonclinical factors are associated with improved clinical outcomes. Several experts13–15 have suggested that many of the cesareans performed in the United States may be unnecessary. However, given the multifaceted nature of the decision-making process to perform a cesarean, evaluating what we describe in this study as “potentially unnecessary” will allow further understanding of the influences to the process of choosing cesarean delivery.
We have shown in a previous study of unnecessary cesarean deliveries in Louisiana that, whereas white women were more likely to experience potentially unnecessary repeat cesareans, they were less likely to experience potentially unnecessary primary cesareans when compared with black women.16 However, the data source did not allow further analysis to control for other important differences that could confound the association between race and cesarean. The current study data set allows for the evaluation of other patient, provider, and system influences on primary and repeat cesarean, as well as the expansion of our hypothesis of the existence of racial differences in cesareans beyond the state of Louisiana.
MATERIALS AND METHODS
This is a cross-sectional study of the 2001 Healthcare Cost and Utilization Project National Inpatient Sample database, sponsored by the Agency for Healthcare and Quality. The National Inpatient Sample is a public database available in the United States containing 7.4 million hospital stays for the year 2001. The database was created from about 1,000 hospitals sampled to approximate a 20% stratified sample of U.S. hospitals. This database includes 15 different diagnoses and 15 different procedures per patient as identified by the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) codes. This database also includes patient demographics, socioeconomics, and provider characteristics. Of the 33 Healthcare Cost and Utilization Project partner states, 24 states report race (AZ, CA, CO, CT, FL, HI, IA, KS, MD, MA, MI, MO, NC, NJ, NY, PA, RI, SC, TN, TX, UT, VA, VT, WI), and 9 states (GA, IL, KY, ME, MN, NE, OR, WA, WV) did not report any race in the National Inpatient Sample data set, but they reported the other demographics studied.
The criteria based on ICD-9-CM codes, proposed by Henry et al17 as indications for cesarean delivery were used (Table 1). To facilitate comparison with the results from other studies, the analyses of cesareans was conducted separately for primary and repeat procedures.8,18 It should be noted that this methodology does not include simply having a prior cesarean delivery as a reason for performing the repeat cesarean delivery.
The study population consisted of all singleton live-birth delivery discharges in the National Inpatient Sample database for the calendar year 2001. Patients with multiple pregnancies were excluded from the analysis so comparison to data of other investigators would be facilitated.19 Cesarean delivery was defined as “potentially unnecessary” if none of the indications noted in Table 1 were present in the patient's discharge records. Women were divided into those who never had a prior cesarean, the primary cesarean group, and those with at least 1 previous cesarean, the repeat cesarean group. The proportions of apparently or potentially unnecessary primary or repeat cesareans were calculated by dividing the number of cesareans without any documented clinical indication by the number of women who received the respective cesarean.
In these data analyses, the dependent variables of interest were unnecessary primary and repeat cesareans. Demographic information obtained from the discharge data for this study included maternal age, race, and ZIP income. Age was divided into 2 categories: younger than 35 years and 35 years or older. Race was categorized as white, black, Hispanic, other, or unknown. The “other” category of race included Asian or Pacific Islander and Native American. ZIP income (median household income for patient's ZIP code) was categorized as $1–$24,999, $25,000–$34,999, $35,000–$44,999, or $45,000 and above. Provider characteristics were also categorized according to payment source (Medicaid, Medicare, private, and others), hospital region (Northeast, Midwest, South, or West), hospital location and teaching status (rural, urban nonteaching, or urban teaching), hospital ownership (government, private, government or public), and hospital size (small, medium, or large). To evaluate the effect of physician work schedule (leisure incentive), the admission day (weekday or weekend) was also used as a categorical variable. Leisure time was defined as the “weekend,” which included Saturday or Sunday day and night. For the purpose of this study, Friday night or other weekday nights were not considered leisure time.
Because the database represented a cluster sampling, the subjects were not truly independent but nested within hospitals. This created correlation among the observations within hospitals and violated the assumption of independence required by the traditional logistic regression model. To correct for the dependence of clustered data and the potential that the standard errors of the estimates would be biased and underestimated, a generalized estimating equation was used.20 SUDAAN 8.00 software (Research Triangle Institute, Research Triangle Park, NC) adapted the generalized estimating equation technique to estimate the variances (exchangeable) when estimating the logistic regression equation. Because the sample size was large, a relatively low P value (< .01, 2-tailed) was used to test the significance of various parameters.
Using our U.S. study population made up of 33 reporting states, we calculated that 540,174 primary cesareans and 371,863 repeat cesareans were performed during calendar year 2001. Based on discharge ICD-9 coding methodology, the overall proportions of potentially unnecessary primary and repeat cesareans were 11.03% and 65.35%, respectively (Table 2).
The potentially unnecessary primary cesarean proportion in black women was 14.4%, higher than the proportions in white (10.4%) or Hispanic women (10.7%) (Table 2). Women aged 35 years or older had a somewhat greater rate of potentially unnecessary primary cesareans when compared with women younger than 35 years of age (11.9% versus 10.9%, respectively). Differences in potentially unnecessary primary cesareans were also noted by payment source. Women with Medicare had the highest proportion of potentially unnecessary primary cesareans (15.8%) compared with those with other payment sources. Women admitted to the hospital on a weekend had an 11.7% rate of potentially unnecessary cesareans; those admitted on a weekday had a 10.9% rate. Women who lived in a ZIP code with the lowest median income (< $25,000) had the highest potentially unnecessary primary cesareans (12.4%) compared with women of higher-income families (10.9%). The Northeast region had the highest (12.4%) and the West had the lowest (9.2%) potentially unnecessary primary cesarean proportions. Urban teaching hospitals had a slightly higher proportion of unnecessary primary cesareans compared with nonteaching hospitals and rural hospitals. Neither hospital ownership nor hospital size had an effect on the proportion of potentially unnecessary primary cesarean deliveries.
Explanatory variables for potentially unnecessary repeat cesareans were in the opposite direction to the associations found with unnecessary primary cesareans (Table 2). White women had the highest (66.3%) and black women had the lowest (61.9%) rate of potentially unnecessary repeat cesareans. In contrast to primary cesareans, those with the highest rate of potentially unnecessary repeat cesareans were more often younger than 35 years, admitted during the week, located in small to medium rural hospitals, and located in the South or West. However, similar to the unnecessary primary cesarean pattern, women living in ZIP codes with the lowest median income had the highest potentially unnecessary repeat cesarean rate (68.4%) when compared with those in greater ZIP-income categories (64.0% for income of $45,000 or greater). Of various payment sources, women with Medicare had the lowest rate of potentially unnecessary repeat cesareans when compared with other payment sources.
Multiple logistic regression with primary and repeat unnecessary cesareans as dependent variables demonstrated maternal sociodemographic factors and provider characteristics that influenced the likelihood of potentially unnecessary cesarean deliveries (Table 3). White women, Hispanic women, and women of other ethnicities had a lower likelihood of receiving a potentially unnecessary primary cesarean when compared with black women (odds ratio [OR], 95% confidence interval [CI]: 0.75, 0.70–0.81 for white women; 0.73, 0.66–0.80 for Hispanic women; 0.78, 0.70–0.86 for other ethnicities, all significantly less than black women, P < .001). However, the likelihood of a potentially unnecessary repeat cesarean was higher among white women (OR, 95% CI: 1.10, 1.03–1.18) when compared with black women (P < .001). Women aged 35 years or older were 1.13 times more likely to receive a potentially unnecessary primary cesarean (95% CI 1.07–1.19, P < .001) than those younger than 35 years. However, older women received unnecessary repeat cesareans at a lower rate than their younger counterparts (OR, 95% CI: 0.78, 0.75–0.81; P < .001). Women with Medicare had a higher likelihood (OR, 95% CI: 1.44, 1.13–1.83; P < .001) of having potentially unnecessary primary cesareans but not repeat cesareans (OR, 95% CI: 0.77, 0.61–0.98; P < .05, considered not significant) compared with the privately insured. Women admitted during the weekend had the higher likelihood (OR, 95% CI: 1.08, 1.03–1.14; P < .001) of having a potentially unnecessary primary cesarean, but a lower likelihood (OR, 95% CI: 0.66, 0.62–0.70; P < .001) of having a potentially unnecessary repeat cesarean compared with those admitted during the week.
The region of the United States in which the hospital was located was significantly associated with potentially unnecessary primary cesareans but not associated with potentially unnecessary repeat cesareans (Table 3). Compared with the Western region of the United States, the likelihood of having an apparently unnecessary primary cesarean was 1.26 times higher in the Northeast (95% CI 1.06–1.49, P < .01), and 1.24 times higher in the South (95% CI 1.12–1.36, P < .001). Although location/teaching status of hospitals was not associated with potentially unnecessary primary cesareans, it was associated with potentially unnecessary repeat cesareans. Urban teaching hospitals were found less likely to provide potentially unnecessary repeat cesareans compared with rural hospitals. Hospital ownership, patient income as judged by ZIP-code income (used as a surrogate measure of income), and hospital bed size were not associated with potentially unnecessary cesareans when adjusted for all the factors in the models.
This study describes certain characteristics associated with potentially unnecessary cesarean deliveries. To our knowledge, this is the first report of factors related to apparently unnecessary cesarean deliveries derived from a large U.S. database. We have searched the medical literature in the English language from 1966 to 2004 using Ovid MEDLINE (Ovid Technologies Incorporated) with the following keywords: “cesarean,” “US,” “unnecessary procedures,” “race,” and “racial disparity,” and were unable to find a study similar to ours. Studies that address factors that influence performance and appropriateness of common procedures such as cesarean delivery are important, particularly if strategies to influence their rates are to be developed.
In our study, potentially unnecessary cesarean deliveries were defined as those not having a corresponding discharge diagnosis related to the indication for cesarean (Table 1). This method was used before by Henry et al,17 who noted that the discharge ICD-9 codes correlated well with specific chart review for the cesarean indication. The case number in our clustered samples precluded detailed chart review, rendering use of the discharge diagnosis method more feasible. Use of the SUDAAN software and a generalized estimating equation for data analysis allowed correction for the lack of independence of cluster data, minimizing bias and overestimation.
Notable in our study method is that a diagnosis of a prior cesarean alone was not sufficient to consider the index repeat cesarean as necessary. We chose to exclude this diagnosis as a reason for necessity of repeat cesarean to better compare to work of prior investigators.16,18,21,22
Using our methodology, we found that in the United States in the year 2001, the estimated proportion of potentially unnecessary primary and repeat cesareans was 11% and 65.3%, respectively. Thus, 59,581 primary and 243,012 repeat cesarean deliveries in that year were apparently unnecessary using our definition. Limited state-specific information is available. Proportions of unnecessary primary cesareans in Louisiana16 from 1993 to 2000 and in Ohio18 from 1991 to 1993 were 17% and 7%, respectively. The proportion of unnecessary repeat cesareans in Louisiana (1993–2000),16 Ohio (1991–1992),18 California (1992),21 and Colorado (1986–1988)22 ranged from 40% to 45%. These wide differences in rates of potentially unnecessary repeat cesareans between prior studies and the present study are likely due to vaginal births after cesarean (VBACs). Although VBACs were encouraged more in the past,23 there has been a greater trend toward elective repeat cesareans in recent years. Differences in documentation technique between the hospital discharge certificate and the birth certificate, used in some studies, may also explain these differences.16,18
Variances in cesarean delivery rates have been attributed to maternal factors (race, age, weight, and insurance status), physician factors (physician gender, training, work experience, practice characteristics, work schedules, economic incentives, and malpractice experience), hospital characteristics (hospital ownership, size, location, and teaching status), and patient preference.9–12,18,19 In our 2001 U.S. database, we were able to evaluated patient characteristics such as ethnicity, age, payment source, ZIP income, and day of admission, as well as hospital characteristics including location of the hospital, ownership, size, and teaching status.
From our 2001 data set, we found that black women were more likely to have a potentially unnecessary primary cesarean delivery than women who were white, Hispanic, or of other ethnic groups. Potentially unnecessary primary cesareans were also more likely to occur in women who were older, on Medicare, admitted on the weekend, and who lived in the Northeast or the South. Potentially unnecessary repeat cesareans were more likely to occur in white women who were younger than 35 years, admitted on a weekday, and admitted to a rural hospital. This finding is not surprising because of the recent trend to not offer VBAC in rural or community hospitals because of the unavailability of 24-hour anesthesia and other support services.
In the current study, we were unable to evaluate other possible important influences on the decision for cesarean; these include physician factors and patient preference. Given the nearly similar rates for reimbursement for vaginal delivery and cesarean delivery by most payers, a financial incentive to perform a cesarean was likely not a strong factor. However, the finding of more potentially unnecessary primary cesarean deliveries occurring during the weekends suggests that the decision to perform a cesarean might have been more likely during established leisure times. Nonetheless, other factors may well be in play. Other causes of a higher number of potentially unnecessary primary cesareans on weekends could be the availability of anesthesia and adequacy of nursing staff for labor support.
We speculate that in our study population, potentially unnecessary repeat cesareans were more likely to occur during weekdays because of women's resistance to VBAC and request for an elective repeat cesarean, which is consistent with results from other studies.24–26 Scheduled repeat cesareans are generally performed during the workweek and not on weekends. Admittedly, the current standard of practice is to offer a woman with a prior cesarean delivery the option of repeat cesarean even if there are no other indications for repeat cesarean. However, because white women had a higher rate of potentially unnecessary repeat cesareans, we can speculate from our data that white women either may have been more likely to choose repeat cesarean delivery, or may have been given the option more often than black women. Physician resistance to VBAC may have also been a factor, particularly in 2001 when newer literature questioned the safety of VBAC in certain situations.27–30 A limitation of the ICD-9 methodology used in this study was the inability to identify which patients had an absolute contraindication to trial of labor compared with those with failed trial of labor and who had a subsequent repeat cesarean.21
Race was a significant factor in the potentially unnecessary cesarean proportions independent of other factors studied. We speculate that physician practice patterns or patient preference may explain these racial differences. Black women had a significantly greater risk of a potentially unnecessary primary cesarean delivery compared with women of all other races. Black women may have been more agreeable to primary cesarean delivery than others. Although in the year 2001 patient choice alone was not an indication for a primary cesarean, patient preference may have had a subtle effect upon a physician's decision to perform a cesarean. Additionally, black women may also have had a higher rate of unreported or uncommon medical complications (not on our ICD-9 list), which may have swayed the physician's decision for cesarean. If white women are more likely to be higher educated than black women, then level of education may confound the association of race and unnecessary cesarean delivery. A number of published studies have cautioned against VBAC because of reported uterine rupture triggered by induction of labor.27–30 White women may have perceived a higher risk for VBAC than black women, and were more likely to choose repeat cesareans.
In addition to patient preferences, physicians also play an active role in the patient's perception of risk regarding cesarean delivery. Different management practices by physicians independent of clinical indicators may also have an effect on potentially unnecessary cesarean rates. For unclear reasons, physicians may perceive black women as more likely candidates for primary cesarean.
There are limitations to our study. Classifying potentially unnecessary cesareans as the absence of any of the indications described in Table 1 produces a contrived proportion of unnecessary cesarean deliveries. However, others31–34 have suggested more rigid criteria to justify a cesarean and, if used, would likely inflate our proportions.
Additionally, the proportions of both primary and repeat potentially unnecessary cesarean deliveries may have been underreported, because certain ICD-9 indications for cesarean delivery were not absolute. For example, a diagnosis of oligohydramnios would place a woman at risk for cesarean delivery, but would not be an absolute indication for surgery. Underreporting of ICD-9 discharge diagnosis codes may have spuriously inflated the potentially unnecessary proportions. However, these limitations affect the entire data sample and may affect the absolute number of apparently unnecessary cesareans but are expected to minimally affect the odds ratios calculated for comparisons within the data set.
The major limitation to our study is that 27% of the Healthcare Cost and Utilization Project partner states did not report race. The current study findings would certainly apply to the states that had reported race. However, we have estimated the study findings considering all the missing races as white and black separately (data not shown). We found similar results in both cases. Because these calculations did not change our findings, we concluded that missing race appeared to have no effect on our odds ratios and that the missing group was not different from the study population.
Technical error and unreliable diagnosis may limit the study's effectiveness. The ICD-9 assignments made by medical record technicians are subject to human error and medical record technician abstraction may vary among institutions. All possible diagnoses may not have been recorded or abstracted. However, Henry et al17 reported that clinical indications and ICD-9 hierarchical codes were concordant for 83% of primary and 86% of repeat cesareans. Given our large sample size, it is unlikely that the results would be affected by random errors in documenting patient-level characteristics.
Definitive conclusions about whether the potentially unnecessary cesareans were clinically unnecessary cannot be made from these data alone. The current method, which relies on the information in the hospital discharge certificate, is not adequate to explain the complex decision-making process that determines the route of delivery. Factors such as patient preferences, physician preferences and concerns regarding malpractice and other legal actions are important in this process. Newer, more recent concerns related to avoidance of vaginal delivery to prevent pelvic floor damage have also emerged. However, none of this information is found in hospital discharge information. To emphasize these limitations, we have used the term potentially unnecessary to categorize cesareans done without the presence of any reasonable indication. Nonetheless, these results clearly demonstrate that a large number of cesareans are not supported by documentation of recognized clinical indicators and vary according to race as well as other patient demographics, and provider and system characteristics. If these potentially unnecessary cesareans are all avoidable, this would lead to a decrease cost of delivery as well as maternal risk, particularly in primary cesareans. Maternal risk related to cesarean delivery occurs not only during the index case, but also during subsequent gestations by increasing the chance of pregnancy complications such as abnormal placentation and uterine rupture.
In summary, unexplained racial differences exist in the potentially unnecessary cesarean deliveries performed in the United States. Black women are more likely to experience potentially unnecessary primary cesareans than are women of other races, and white women are more likely to experience potentially unnecessary repeat cesareans than are black women. This disparity may be explained by differences in patient and physician perceptions, which need further study. Certainly, the influence of patient preference deserves further scrutiny as we enter an era of greater patient autonomy.35
1. Banta HD, Thacker SB. The case for reassessment of health care technology: once is not enough. JAMA 1990;264:235–40.
2. Bertollini R, DiLallo D, Spadea T, Perucci C. Cesarean section rates in Italy by hospital payment made: an analysis based on birth certificates. Am J Public Health 1992;82:257–61.
3. Stafford RS. Alternative strategies for controlling rising cesarean section rates. JAMA 1990;263:683–7.
4. Stafford RS. The impact of nonclinical factors on repeat cesarean section. JAMA 1991;265:59–63.
5. Stafford RS. Cesarean section use and source of payment: an analysis of California hospital discharge abstracts. Am J Public Health 1990;80:313–15.
6. Healthy People 2010. 2nd ed. Understanding and improving health. Objectives for improving health. 2 vols. Washington, DC: U.S. Department of Health and Human Services; 2000.
7. Butcher AH, Fos PJ, Zuniga M, Pane G. Racial variations in cesarean section rates: an analysis of Medicaid data in Louisiana. J Public Health Manag Pract 1997;3:41–8.
8. Koroukian SM, Rimm AA. Declining trends in cesarean deliveries, Ohio 1989–1996: an analysis by indications. Birth 2000;27:12–8.
9. Gould JB, Davey B, Stafford RS. Socioeconomic differences in rates of cesarean section. N Engl J Med 1989;321:233–9.
10. Mitler LK, Rizzo JA, Horwitz SM. Physician gender and cesarean sections. J Clin Epidemiol 2000;53:1030–5.
11. Burns LR, Geller SE, Wholey DR. The effect of physician factors on the cesarean section decision. Med Care 1995;33:365–82.
12. Tussing AD, Wojtowycz MA. The cesarean decision in New York State, 1986: economic and noneconomic aspects. Med Care 1992;30:529–40.
13. Norman P, Kostovcik S, Lanning A. Elective repeat cesarean sections: how many could be vaginal births? CMAJ 1993;149:431–5.
14. Flamm BL, Newman LA, Thomas SJ, Fallon D, Yoshida MM. Vaginal birth after cesarean delivery: results of a 5-year multicenter collaborative study. Obstet Gynecol 1990;76:750–4.
15. Paul RH, Miller DA. Cesarean birth: how to reduce the rate. Am J Obstet Gynecol 1995;172:1903–7.
16. Kabir AA, Steinmann WC, Myers L, Khan MM, Herrera EA, Yu S, et al. Unnecessary cesarean delivery in Louisiana: an analysis of birth certificate data. Am J Obstet Gynecol 2004;190:10–9.
17. 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 1995;85:1143–6.
18. Koroukian SM, Trisel B, Rimm AA. Estimating the proportion of unnecessary cesarean sections in Ohio using birth certificate data. J Clin Epidemiol 1998;51:1327–34.
19. Koroukian SM, Bush D, Rimm AA. Comparison of cesarean section rates in fee-for-service versus managed care patients in the Ohio Medicaid population, 1992–1997. Am J Manag Care 2001;7:134–42.
20. Blair RC, Higgins JJ, Topping ME, Mortimer AL. An investigation of the robustness of the t-test to unit of analysis of violations. Educ Psychol Meas 1983;43:69–80.
21. Gregory KD, Henry OA, Gellens AJ, Hobel CJ, Platt LD. Repeat cesareans: how many are elective? Obstet Gynecol 1994;84:574–8.
22. Porreco RP. Meeting the challenge of the rising cesarean birth rate. Obstet Gynecol 1990;75:133–6.
23. Pridjian G, Hibbard JU, Moawad AH. Cesarean: changing the trends. Obstet Gynecol 1991;77:195–200.
24. Joseph GF Jr, Stedman CM, Robichaux AG. Vaginal birth after cesarean section: the impact of patient resistance to a trial of labor. Am J Obstet Gynecol 1991;164:1441–4.
25. Abitbol MM, Castillo I, Taylor UB, Rochelson BL, Shmoys S, Monheit AG. Vaginal birth after cesarean section: the patient's point of view. Am Fam Physician 1993;47:129–34.
26. Wall EM. “But doctor, I'd prefer to have a cesarean section”: when public policy conflicts with patient preference. J Am Board Fam Pract 1995;8:414–6.
27. Approach VBAC deliveries cautiously, experts say. Hosp Peer Rev 1998;23:189.
28. Ravasia DJ, Wood SL, Pollard JK. Uterine rupture during induced trial of labor among women with previous cesarean delivery. Am J Obstet Gynecol 2000;183:1176–9.
29. Mozurkewich EL, Hutton EK. Elective repeat cesarean delivery versus trial of labor: a meta-analysis of the literature from 1989 to 1999. Am J Obstet Gynecol 2000;183:1187–97.
30. Greene MF. Vaginal delivery after cesarean section: is the risk acceptable? N Engl J Med 2001;345:54–5.
31. Sanchez-Ramos L, Kaunitz AM, Peterson HB, Martinez-Schnell B, Thompson RJ. Reducing cesarean sections at a teaching hospital. Am J Obstet Gynecol 1990;163:1081–7.
32. Myers SA, Gleicher N. The Mount Sinai cesarean section reduction program: an update after 6 years. Soc Sci Med 1993;37:1219–22.
33. 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;326:450–4.
34. Chervenak FA, McCullough LB. An ethically justified algorithm for offering, recommending, and performing cesarean delivery and its application in managed care practice. Obstet Gynecol 1996;87:302–5.
35. Gonen R, Tamir A, Degani S. Obstetricians’ opinions regarding patient choice in cesarean delivery. Obstet Gynecol 2002;99:577–80.
Figure. No caption available.