Declercq, Eugene PhD1; Barger, Mary CNM, MPH1; Cabral, Howard J. MPH, PhD2; Evans, Stephen R. MPH3; Kotelchuck, Milton MPH, PhD1; Simon, Carol MS, PhD5; Weiss, Judith ScD1; Heffner, Linda J. MD, PhD4
The overall (29.1%) and primary (20.6%) cesarean rates in the United States reached an all-time high in 2004, each rate representing a 40% increase since 1996.1,2 It has been suggested that “maternal request cesareans” may play a significant role in this increase,3,4 but there is very limited systematic evidence on maternal perspectives toward cesareans. Maternal request cesareans are presumably a subset of medically elective cesareans, a phenomenon that has been systematically documented,5–7 although without analysis of outcomes of these births.
A U.S. National Institutes of Health (NIH) State-of-the-Science Conference on “Cesarean Delivery by Maternal Request,” in March 2006 commissioned an evidence report that restructured the question to focus on a comparison of what they defined as planned cesarean delivery and planned vaginal delivery.8 The report found little research structured in a format to compare outcomes of these methods of delivery, and the expert panel formed by NIH to review the evidence concluded there was a need for research that explicitly compared planned cesarean to planned vaginal births.4
This study used a data system that links almost one half million birth certificates to hospital discharge records for the state of Massachusetts over a 6-year period. Our objective was to examine the trend in planned cesareans and compare maternal morbidities and costs associated with planned cesarean and planned vaginal births using the model recommended by the NIH State-of-the-Science Conference report.
MATERIALS AND METHODS
Data for these analyses were derived from the Pregnancy to Early Life Longitudinal Data System. Pregnancy to Early Life Longitudinal is a public–private partnership among the Boston University School of Public Health, Massachusetts Department of Public Health, and the federal Centers for Disease Control and Prevention, the funding agency. The core Pregnancy to Early Life Longitudinal data sets consist of Massachusetts birth certificates and fetal death records linked to birth-related hospital discharge data for both mothers and infants. Using LinkPro 2.0 (InfoSoft, Inc, Winnipeg, Manitoba, Canada) we linked the birth certificate to the infant's hospital discharge records for the birth event, and birth and fetal death certificate records were linked to the mother's delivery record. Because there are no unique identifiers common to both data sources, matches were made on several combinations of selected variables for the mother (hospital, ZIP code, mother's date of birth, and delivery date) and infant (hospital, medical record number, delivery date, ZIP code, and sex). The linkage rate for the 470,857 deliveries of live births to Massachusetts residents in Massachusetts hospitals from 1998–2003 to the mothers' corresponding hospital discharge delivery records was 99.2%. Subsequent maternal hospitalizations were identified by linking birth event data longitudinally to maternal hospitalizations in Massachusetts hospitals in the year after the birth, using encrypted social security numbers when available (94% of the time), or a unique nonmissing combination of hospital number and medical record number (remaining 6% of births). The small number of cases (1,186) where a woman gave birth a second time within a year of the index birth were excluded. Once linked, records stripped of identifying information were analyzed in accordance with Health Insurance Portability and Accountability Act privacy rules (1996) and Massachusetts confidentiality statutes. Institutional review board approval was obtained from the Massachusetts Department of Public Health and the Boston University School of Medicine.
The subjects of this analysis are mothers who have not had a prior cesarean (thereby excluding mothers who had a vaginal birth after cesarean or repeat cesarean) and a singleton, full term (37 – 41 weeks) birth with a vertex presentation (Fig. 1).9 This reduced the study group to 321,827 births. A “no documented prior risk” group was then defined based on two measures that had been previously developed using either birth certificate7 or hospital discharge files.10 The birth certificate measure excluded any mother who had one or more of the 18 items available under “medical risk factors” (Fig. 1) and one item listed under labor and delivery complications on the birth certificate (placenta previa) that we included as a medical risk because it would likely be diagnosed before labor. A total of 267,179 mothers met those criteria. The hospital discharge measure was adapted from earlier work by Gregory et al.10 Using a previously validated measure based on selected International Classification of Diseases, 9th Revision, Clinical Modification diagnosis codes,11 they identified mothers with medical risks likely diagnosed during the prenatal period (Fig. 1). A total of 284,052 mothers met these criteria. We then took advantage of our data linkage to limit our analysis to those who met both the birth certificate and hospital discharge criteria, resulting in 244,088 mothers who would have approached labor as candidates for a vaginal birth.
We then used labor and delivery complication codes from the birth certificate and hospital discharge data to classify these mothers in conformance with the guidelines recommended by the NIH evidence report, subdividing them into two groups: primary cesareans with no labor and no documented complications (planned cesareans) and women who had vaginal or cesarean births with labor (planned vaginal births). Assuming that a mother with a labor and delivery complication code had experienced labor, we created our subgroups for analysis. The planned cesarean group involved those mothers with no prior risk and none of the labor and delivery complication codes on the birth certificate7 or hospital discharge file (Fig. 1).10 The resulting subgroup of 3,334 mothers is the most conservative population-based estimate used to this point to identify planned primary cesareans.
Using an “intention-to-treat” approach, the planned vaginal group consisted of the other three possible options (Fig. 1): vaginal births with no documented labor complications (172,329), vaginal births with documented labor complications (47,509), and primary cesareans with documented labor complications (20,916). These unplanned primary cesareans are included in the planned vaginal category because a vaginal birth was likely anticipated at the onset of labor.8 These three groups totaled 240,754 women in the planned vaginal group.
Thus, we have created a comparison between the two groups characterized by the NIH panel as being at the center of the current debate over medically elective cesareans—mothers who approach labor without a major medical risk factor and anticipate either a planned vaginal birth or a planned primary cesarean.
Our primary outcome was maternal hospital readmission to any Massachusetts hospital during three time periods after discharge for the birth: 1–30 days; 31–180 days; and 181–365 days. We examined overall rates of readmission as well as unadjusted and adjusted odds ratios for readmission, using planned vaginal births as the referent. We ran three separate sets of logistic regression analyses (both unadjusted and adjusted), with one set for each period. For the 31–180 days and 181–365 day periods, we excluded observations for subjects who had been readmitted in a prior period. We interpreted odds ratios as measures of relative risk given that rehospitalization was a comparatively rare event in these data. The three covariates included in the model have been documented in past research to be related to primary cesarean rates: mother's age, parity, and race or ethnicity.1,12
Costs of initial and subsequent hospitalizations were calculated from total charges drawn from the hospital discharge file and then adjusted based on hospital-specific cost/charge ratios (from the Massachusetts Division of Health Care Finance and Policy) for Massachusetts hospitals for the time period studied. Computed costs were then adjusted for inflation, with data presented in 2003 dollars.13 Least squares analyses were also run for average costs by method of delivery controlling for age, parity, race or ethnicity, and payer source. We also used the hospital discharge data to examine length of hospital stays for mothers. We used χ2 and Fisher exact tests to perform hypothesis tests related to key study questions in cross-tabulated data. All statistical analyses were performed using SAS 9.1.3 (SAS Institute, Inc., Cary, NC).
From the original 470,857 births in Massachusetts hospitals to Massachusetts residents from 1998–2003, there were 321,827 mothers with no prior cesareans who, between 1998 and 2003, had a singleton, vertex, full term (37—41 weeks) birth. The number of mothers who gave birth with no documented prior risk decreased slightly through the study period (Table 1). Among these births, the unplanned cesarean rate increased steadily during the study period (from 7.6% to 9.9%), whereas the planned primary cesarean rate quintupled from 0.4% in 1998 to 2.2% in 2003, with an average rate for the entire period of 1.4%.
Rates of planned cesareans varied widely across different subgroups (Table 2), with women aged 40 years or older (3.1%), African-American non-Hispanic mothers (2.3%) and first time mothers (2.1%) each substantially more likely than average to have a planned primary cesarean. By 2003, the planned primary cesarean rate for first-time mothers was 3.4% compared with 1.0% for multiparous mothers. The difference in the rate for mothers with private insurance (1.4%) was minimal compared with mothers with public health insurance (1.3%). When race or ethnicity and payer source were combined, African-American non-Hispanic mothers continued to have the highest planned cesarean rates when using either private insurance (2.6%) or a public payer (2.0%). Hispanic mothers planned cesarean rate also varied according to their use of a public (1.1%) or private (1.6%) payer, while white non-Hispanic mothers planned cesarean rates (1.3%) were the same for public or private payer (data not shown).
Women with these lower-risk deliveries typically did not require rehospitalization, but there were substantial differences by method of delivery (Table 3). The rate of rehospitalization (per 1,000) within 1 month of delivery for planned vaginal births (7.5) was significantly lower than that for planned primary cesareans (19.2). Adjusting for age, race or ethnicity, and parity, a mother who had a planned primary cesarean was 2.3 times as likely as a woman who had a planned vaginal birth to be rehospitalized in the first month after the birth. For subsequent periods, mothers with a planned primary cesarean were 45% (31–180 days) and 77% (181–365 days) more likely to require a hospital admission compared with those with a planned vaginal birth.
The leading reason for readmission associated with planned primary cesareans in the first 30 days after birth was surgical wound complications, including cellulitis or abscess of trunk (Table 4). Hospitalizations for wound complications occurred at a rate 14 times higher than in the planned vaginal group (P≤.001), where it was the fifth leading reason, with the latter cases being drawn from planned vaginal births that resulted in unplanned cesareans. Postpartum infections were a major cause of rehospitalization for both groups, with rehospitalization for infection after planned primary cesarean births at a rate almost twice as high as that for planned vaginal births (P=.04). The reasons for rehospitalizations in the subsequent periods were less directly related to birth events and distributed across a much larger number of diagnoses (eg, appendicitis, calculus of gallbladder, volume depletion, pneumonia, lumbar intervertebral disc without myelopathy).
The average initial maternal (excluding infant) hospital costs in 2003 dollars for a planned primary cesarean was $4,372, 76% higher than the initial costs for a planned vaginal birth ($2,487) (Table 5). For the subset of mothers who were rehospitalized, costs for those who had a planned primary cesarean averaged $6,167, 13% higher than the average of $5,434 for those rehospitalized after a planned vaginal birth. Some of these additional costs are undoubtedly associated with longer stays in each case, not surprising in the case of the initial hospitalization, because state and federal laws mandating longer stays for cesareans were passed in the 1990s.14 Mothers who had a planned primary cesarean averaged 4.3 days in their initial stay and 4.4 days in cases of rehospitalization as compared with 2.4 and 3.9 days, respectively, for those with a planned vaginal birth. The same cost analysis was performed using least squares regression controlling for mothers age, parity, race or ethnicity, and payer source (data not shown) and found comparable relationships, with costs associated with a planned primary cesarean higher for both delivery (65%) and postpartum rehospitalization (11%).
This study used 6 years of data from a population-based linked data system to create a refined measure identifying women with planned cesareans and planned vaginal births and comparing maternal outcomes and costs associated with these two options. Planned primary cesareans constituted a small portion of all births during the study period, but did increase rapidly between 1998 and 2003. Although planned primary cesareans seem to be medically elective, they are not necessarily “maternal request cesareans.” Nonetheless, we can say that mothers who have a planned primary cesarean face outcomes they may not have anticipated, including more than double the chance of being rehospitalized in the first month for complications associated with the surgery. These findings do not support one of the rationales proposed for elective cesareans—greater maternal convenience.15
The 2006 NIH panel report noted that the inability to combine unplanned cesareans with vaginal births has handicapped prior analyses of length of stay and costs,4 stating “Planned cesarean delivery on maternal request will have different cost implications that should be modeled explicitly.”4 (p. 1394.) Applying their model, we found mothers with planned cesareans had a 77% longer delivery hospital stay and 76% higher costs.
The greater likelihood of a rehospitalization associated with cesarean birth has been documented in the past, although not in the context of planned, medically elective cesareans. Past studies typically adjusted for some demographic and health-related variables, but did not have a planned cesarean category, nor did they include unplanned cesareans in their cohort of vaginal births. Lydon-Rochelle et al16 found the likelihood of readmission within 60 days after a cesarean to be 80% higher than that after a spontaneous vaginal delivery. Liu and colleagues17 found mothers with cesareans were 87% more likely to be readmitted within 60 days than those with a spontaneous vaginal delivery.
There are several limitations to our study. We are using data from only one state, and although the pattern of growth in primary cesareans in Massachusetts has paralleled the national trend, the state does not demographically mirror the United States, particularly with its lower than average proportion of births to African-American non-Hispanic mothers. The proportion of planned primary cesareans was highest among that group; therefore, the overall rate of these planned cesareans in other states may be higher.
Our use of the term “no documented prior risk,” reflects the reality that there may be medical indications for cesareans not recorded or captured by the variables reported on the birth certificate and hospital discharge data studied here. Two recent studies18,19 have examined the reliability of birth certificate and hospital discharge data and find underreporting of some medical risk factors. Both cite limitations in the use of either hospital discharge or birth certificate data sets alone and recommend the use of the type of linked data set employed in this analysis.
It is still possible that in our study, the planned cesareans that resulted in readmissions are those for which an unrecorded medical reason was the basis for the cesarean. It is not clear, however, why there would be a systematic bias toward not recording a medical risk factor or labor and delivery complication in the case of primary cesareans, when it would seem more likely that a reason for a clinically indicated cesarean would be noted. Also, the pattern of growth in planned primary cesareans found here is similar to that reported in other studies relying on different data sets.6 In addition, the records used to create the “no documented prior risk” group that is at the core of the planned cesarean measure were drawn from two entirely different data sources, and we only analyzed those cases where both birth certificate and hospital discharge data were in agreement. The fact that the leading causes of readmission were complications specifically associated with surgery suggests it is doubtful that recording bias alone could account for such a substantial difference in rehospitalization, but it may influence the overall result.
This study also examined outcomes only up to 1 year after birth, and that limits our analysis in two directions. It is the contention of those advocating for elective cesareans that the problems with vaginal birth are manifested over a longer period of time in terms of urinary and fecal incontinence.20 These problems would seldom manifest themselves so seriously that it would require hospitalization, the outcome measure used here. Our 1-year period also prevents us from examining the effect of planned primary cesareans on risks associated with subsequent birth experiences, a concern raised in a recent multicenter study.21
Applying the term “maternal request” to these planned primary cesareans is problematic, because none of the population-based studies,22 including ours, measure maternal request. Studies from other countries have cast doubt on a simple characterization of medically elective cesareans as involving maternal request.23,24 The cost analysis presents a challenge for funders to consider. Despite some suggestions that cesareans on demand will have minimal effect on costs,25 we found substantial differences in maternal hospital costs associated with both the initial hospitalization and rehospitalizations. This raises the issue of who should be responsible for paying these costs. If these are truly cases of patient choice elective surgery, should individual patients bear the additional costs? Who should be responsible for the costs of additional hospital space that must be allocated to maternity care, given the longer postpartum stays required for cesarean deliveries?
There are clinical implications of these findings. The higher rates of readmission for surgical complications and infection in the cesarean cases suggest that despite the improvements in surgical techniques, continued efforts to reduce complications and infections are necessary. Universal use of intraoperative prophylactic antibiotics, minimization of the length of time that indwelling urinary catheters remain in place, and routine use of early (2-week) postoperative office visits are examples of ways in which the higher rates of rehospitalization after cesarean might be reduced.
Improvements in surgical training and techniques and the institution of quality review systems have made cesareans safer than they have been in the past. It is less clear that planned primary cesareans are safer than planned vaginal births, even when the latter category includes unplanned cesareans. Our study used the most extensive data set and rigorous exclusion criteria yet applied to an analytic model recommended by a U.S. expert panel. It found a small, but consistent growth in planned primary cesareans, but higher costs, longer hospital stays, and substantially greater risks of maternal rehospitalization associated with these deliveries. Our findings suggest that planned primary cesareans are not without immediate health consequences for mothers and financial implications for society. Clinicians should be aware of the increased risk for maternal rehospitalization after cesarean deliveries to low-risk mothers when counseling woman about their choices.
1. Martin JA, Hamilton BE, Sutton PD, Ventura S, Menacker F, Kirmeyer S. Births: final data for 2004. Natl Vital Stat Rep 2006;55:1–101.
2. Menacker F. Trends in cesarean rates for first births and repeat cesarean rates for low-risk women: United States, 1990–2003. Natl Vital Stat Rep 2005;54:1–8.
3. Harer WB. Patient choice cesarean. ACOG Clin Rev 2000;5:13–6.
4. National Institutes of Health state-of-the-science conference statement: Cesarean delivery on maternal request March 27-29, 2006. Obstet Gynecol 2006;107:1386–97.
5. Kabir AA, Pridjian G, Steinmann WC, Herrera EA, Khan MM. Racial differences in cesareans: an analysis of U.S. 2001 National Inpatient Sample Data [published erratum appears in Obstet Gynecol. 2005;105:1495]. Obstet Gynecol 2005;105:710–8.
6. Meikle SF, Steiner CA, Zhang J, Lawrence WL. A national estimate of the elective primary cesarean delivery rate. Obstet Gynecol 2005;105:751–6.
7. Declercq E, Menacker F, MacDorman M. Rise in “no indicated risk” primary cesareans in the United States, 1991-2001: cross sectional analysis. BMJ 2005;330:71–2.
8. Viswanathan M, Visco AG, Hartmann K, Wechter ME, Gartlehner G, Wu JM, et al. Cesarean delivery on maternal request. Evidence Report/Technology Assessment No. 133. (Prepared by the RTI International-University of North Carolina Evidence-Based Practice Center under Contract No. 290-02-0016.) AHRQ Publication No. 06-E009. Rockville (MD): Agency for Healthcare Research and Quality; 2006.
9. American College of Obstetricians and Gynecologists. Evaluation of Cesarean Delivery. ACOG Task Force on Cesarean Delivery. Washington (DC): ACOG; 2000.
10. Gregory KD, Korst LM, Gornbein JA, Platt LD. Using administrative data to identify indications for elective primary cesarean delivery. Health Serv Res 2002;37:1387–401.
11. 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.
12. Menacker F, Curtin SC. Trends in cesarean birth and vaginal birth after previous cesarean, 1991-99. Natl Vital Stat Rep 2001;49:1–16.
13. Producer Price Index Industry Data [database on internet]. General medical and surgical hospitals 1993-2005; series ID: 622110622110. Washington (DC): U.S. Department of Labor, Bureau of Labor Statistics; 2004. Available from: http://data.bls.gov/cgi-bin/surveymost
14. Newborns' and Mothers' Health Protection Act of 1996. 104th Congress, 2nd Sess (1996) (Senate of the United States, amendment to H.R. 3666).
15. Brink S. Too posh to push? Cesarean sections have spiked dramatically. Progress or convenience? US News World Rep 2002;133:42–3.
16. Lydon-Rochelle M, Holt VL, Martin DP, Easterling TR. Association between method of delivery and maternal rehospitalization. JAMA 2000;283:2411–6.
17. Liu S, Heaman M, Joseph KS, Liston RM, Huang L, Sauve R, et al. Risk of maternal postpartum readmission associated with mode of delivery. Obstet Gynecol 2005;105:836–42.
18. Lydon-Rochelle MT, Holt VL, Nelson JC, Cardenas V, Gardella C, Easterling TR, et al. Accuracy of reporting maternal in-hospital diagnoses and intrapartum procedures in Washington State linked birth records. Paediatr Perinat Epidemiol 2005;19:460–71.
19. Yasmeen S, Romano PS, Schembri ME, Keyzer JM, Gilbert WM. Accuracy of obstetric diagnoses and procedures in hospital discharge data. Am J Obstet Gynecol 2006;194:992–1001.
20. O'Boyle AL, Davis GD, Calhoun BC. Informed consent and birth: protecting the pelvic floor and ourselves. Am J Obstet Gynecol 2002;187:981–3.
21. Silver RM, Landon MB, Rouse DJ, Leveno KJ, Spong CY, Thom EA, et al. Maternal morbidity associated with multiple repeat cesarean deliveries. Obstet Gynecol 2006;107:1226–32.
22. Gossman GL, Joesch JM, Tnafer K. Trends in maternal request cesarean delivery from 1991 to 2004. Obstet Gynecol 2006;108:1506–16.
23. Potter JE, Berquo E, Perpetuo IH, Leal OF, Hopkins K, Souza MR, et al. Unwanted cesarean sections among public and private patients in Brazil: prospective study. BMJ 2001;323:1155–8.
24. Murray SF. Relation between private health insurance and high rates of cesarean section in Chile: qualitative and quantitative study. BMJ 2000;321:1501–5.
25. Bost BW. Cesarean delivery on demand: what will it cost? Am J Obstet Gynecol 2003;188:1418–21.