Materials and Methods
To compare the health-related and economic consequences of a trial of labor with those of an elective cesarean delivery, we examined a hypothetical cohort of 100,000 women whose only prior pregnancy was delivered through a low transverse cesarean incision. A decision-tree model was constructed using Tree Plan, version 1.61 (Decision Support Services, San Francisco, CA). In this model, a woman makes an initial decision whether to proceed with an elective repeat cesarean delivery. The decision tree is then extended with the addition of probabilities and costs of the clinical scenarios that develop subsequent to this decision. A Markov technique (Figure 1) was used to extend the analysis beyond the second pregnancy and follow the 100,000 women through the completion of their reproductive lives. After each delivery, a woman chooses whether to have another child; the women who choose further childbearing have a subsequent delivery and related complications that are dependent on their previous obstetric experience. After the cohort has completed childbearing, the model calculates the cumulative probabilities for the outcomes of interest. Given that the choice to proceed with trial of labor after one cesarean delivery is the decision under consideration in this analysis, women who had an elective repeat cesarean for their second pregnancy and women who had an unsuccessful trial of labor will continue to have cesarean deliveries in subsequent pregnancies. Women who had a vaginal birth after a trial of labor will again undergo a trial of labor in this model.
Probability and cost variables that were entered into the model were derived from data in the published literature, or when these were not available, from expert opinion, as indicated. The point of view of this analysis is that of the medical system and third-party payers. All monetary values used in the analysis are direct medical costs; in circumstances in which only charge data were available, these were converted to cost data using a cost-to-charge ratio of 0.6. Costs derived before 1999 were adjusted for inflation to reflect 1999 dollars according to the medical care component of the consumer price index. Future costs were discounted at a rate of 3%.
The most reasonable estimate for each variable was used to create a base-case result for the model. The sensitivity of this result to changes in the probability and cost variables was tested using univariate sensitivity analysis in which each variable was altered between the upper and lower limits of its plausible value. The variables to which the model was most sensitive were then further tested using multivariate sensitivity analysis.
Probability estimates for the different variables are summarized in Table 1. Given that the rate of vaginal birth for women who undergo a trial of labor after one prior low transverse cesarean delivery consistently has been shown to be at least 75%, this value was used as the base-case estimate, with a range of 65–85% examined in the sensitivity analysis.2–5 However, once a woman has a vaginal birth after cesarean, her chance of a vaginal delivery in a subsequent pregnancy increases.19 Consequently, in our model, trials of labor after the first vaginal delivery were accorded a higher vaginal delivery rate of 92.8%, with a range from 85% to 95% tested in the sensitivity analysis.
The definition of uterine scar rupture has varied in the literature, and reports have often grouped together uterine ruptures with asymptomatic scar dehiscences. Additionally, many reports include ruptures that occur after cesarean delivery by classical incision, as well as those that occur in the absence of a trial of labor. This analysis considers only prior low transverse incisions and the excess number of uterine ruptures that occur during a trial of labor. Uterine scar ruptures are assumed to be symptomatic occurrences that require laparotomy to repair and have potential adverse consequences for mother and fetus. If these criteria are used, most large series have found the incidence of rupture to be in the range of 0.3–0.8%.2–6,8,14,16 For the base case, a rate of rupture of 0.5% was used, with a range from 0.3% to 1.0% examined in the sensitivity analysis.
This model incorporates six maternal morbidities whose association with a trial of labor and an elective cesarean delivery has been well studied. These morbidities include postpartum endometritis, blood transfusion, deep venous thrombosis, hysterectomy, and in the case of cesarean delivery, abdominal wound infection and operative bladder injury. In addition, we incorporated the risk of maternal mortality into the model. Because the rates of morbidity and mortality are related to the delivery method, frequencies were estimated for elective cesarean delivery, vaginal delivery after a trial of labor, cesarean delivery after a trial of labor, and uterine rupture during a trial of labor.
In the largest series to examine the fetal consequences of uterine rupture, Leung et al13 studied 99 cases of symptomatic uterine scar separation. There were six fetal deaths, only two of which occurred during a trial of labor. An additional five neonates had perinatal asphyxia. Because these cases include ruptures of classical incisions, which is a risk factor for poor fetal outcome because of the greater risk of fetal extrusion, the true neonatal morbidity after rupture of a low-transverse incision is likely lower. Indeed, Sachs et al7 noted that the risk of serious complications after uterine rupture might be as low as 1% or less. Although rates of complications might be this low at some institutions, we conservatively estimated that 10% of neonates would have major complications (death or cerebral palsy) after uterine rupture. We further estimated that an additional 15% of neonates without long-term morbidity would require care in the special care nursery because of low Apgar scores and/or acidemia. Because other reports of smaller case series have noted both lower and higher rates of neonatal complications,2,4,14,16,17 we tested a range of possible complication rates in the sensitivity analysis.
Neonatal morbidity during a trial of labor might not be related solely to uterine rupture. The events during an otherwise uncomplicated vaginal delivery could also contribute to the development of cerebral palsy. This possibility was tested in the sensitivity analysis by estimating that up to 20% of the cases of cerebral palsy, with an incidence of one per 1000, occur intrapartum.18
To extend the model beyond the second pregnancy, we estimated the proportion of women who continue with childbearing after their second delivery. Forty-six percent of women who have two children will have a third, and 37% of those who have a third child will have a fourth.20 For cost discounting, we chose an average interval between childbearing of 3 years. We limited this analysis to childbearing through the fourth child, because the proportion of women in the United States who have more than four children is small.
Costs of the different health scenarios are summarized in Table 2. In terms of delivery type, the vaginal route is least costly, followed by elective cesarean, and then cesarean delivery after labor.15,21,22 Because this analysis considers uterine ruptures that occur during a trial of labor and require laparotomy, a uterine rupture was accorded the cost of an unsuccessful trial of labor in addition to the extra costs incurred by this complication.
Costs of various maternal complications also were calculated. The costs of deep venous thrombosis and transfusion of packed red blood cells were derived from estimates in the literature.23,24 It was assumed that patients requiring transfusion would receive on average two units of packed red blood cells, with the exception of patients transfused in the setting of uterine rupture, who would receive four units.8,28 The costs of other maternal complications were based on the costs of additional operating room time, hospital days, home health care visits, and antibiotics that these complications typically require at our institution. The cost of an additional hospital day was calculated to be $400 and a home health visit was $67.22,29 Thus, the cost of endometritis was estimated as the additional cost incurred by 1 extra hospital day and 48 hours of gentamycin and clindamycin.30 The cost of a wound infection comprised 1 extra hospital day, three home health visits, and 10 total days of antibiotics (72 hours of ampicillin or sulbactam and 7 days of amoxicillin or clavulanate).30 The incremental cost of a hysterectomy was approximated by the additional cost of operating room time (three times that required for a cesarean delivery),22 whereas a bladder injury was considered to require 1 extra hospital day and two home health visits. Because the exact composition of the costs of these complications could vary according to regional practice differences, costs above and below the base costs were examined in the sensitivity analysis. The most difficult cost to estimate was that of a maternal death because the exact circumstances of this event vary greatly, and correspondingly, additional medical costs vary greatly as well. For the base case, we have assumed that there are no marginal costs in excess of those of the complications (eg, deep venous thrombosis) that resulted in the death. Because of the great variation in possible scenarios, however, we have tested this assumption in the sensitivity analysis by including excess costs of up to $100,000.
Excess neonatal costs include those incurred as a result of cerebral palsy or neonatal death. An individual with cerebral palsy incurs additional medical costs of approximately $179,000 throughout his or her lifetime.25 Given the impact that differing discount rates can have on this cost of care, a range of costs, derived using different discount rates, was used in the sensitivity analysis. The marginal cost of care for neonates who are transiently admitted to the special care nursery was calculated as the cost of the additional days in the special care nursery that such infants require. Socol et al26 reported that infants with severely depressed Apgar scores and acidemia who do not have long-term complications usually remain in the special care nursery less than 48 hours. Therefore, the cost of transient asphyxia was calculated as the additional cost of 2 days in the special care nursery.27 Estimations of the cost of a neonatal death are hampered by the same variations in cost noted for maternal death. Neonates may never be resuscitated, in which case marginal costs are low, or they may have a long and complicated course before their demise, in which case marginal costs are high. To maintain consistency, the base-case assumption places the marginal cost of care for a neonatal death at $0, although the impact of this assumption is tested in the sensitivity analysis by varying this cost of care to $100,000.
The results for the base-case assumptions are summarized in Table 3. Under these assumptions, elective repeat cesarean delivery in 100,000 women whose first birth was a cesarean through a low transverse incision will prevent 37 cases of cerebral palsy and 37 neonatal deaths. To achieve this health benefit requires an excess of 117,748 cesarean deliveries, seven maternal deaths, and 5500 maternal morbid events. Further, this reduction in neonatal complications comes at a total cost of $179 million to the health care system. Viewed from a cost-effectiveness perspective, the prevention of one major adverse neonatal outcome (death or cerebral palsy) requires 1591 cesarean deliveries and a cost of more than $2.4 million to the health care system. Additionally, one maternal death occurs for every five neonatal lives saved.
Under one-way sensitivity analysis, the cost-effectiveness results of this model are insensitive to changes in the frequency and costs of maternal complications listed in Table 1. Despite varying each of these estimates through its entire range, the cost per averted adverse neonatal outcome remains $2.4 million. The model is also relatively insensitive to changes in the other probability and cost variables under consideration (Table 4). Only at the upper bound of the frequency of uterine rupture does the cost per averted adverse neonatal outcome decrease to approach $1 million. Under most circumstances, more than 1000 excess cesarean deliveries are necessary to prevent one major neonatal complication.
With multivariate sensitivity analysis, the model remained insensitive to changes in most variables (Table 5). When all the risks and costs of maternal complications associated with a trial of labor were simultaneously increased to their upper limits, the elective cesarean policy continued to incur more than $2 million per adverse neonatal outcome averted. Similarly, despite changes in multiple variables related to neonatal complications, a minimum of 806 cesarean deliveries and a cost of $1.1 million per poor neonatal outcome resulted from a policy of elective cesarean birth.
Last, we assessed the circumstances under which the probabilities of combined vaginal delivery during the initial trial of labor and uterine rupture would achieve more favorable cost-effectiveness ratios than those obtained so far in the analysis (Figure 2). For example, a cost-effectiveness of $250,000 per major adverse neonatal outcome prevented would not be achieved unless a woman with a vaginal birth rate of 70% has a risk of uterine rupture that exceeds 3.3%. The policy of routine elective cesarean delivery would not be a cost-saving policy at any vaginal birth rate greater than 40% unless the risk of uterine rupture was extremely high.
Rupture of a prior uterine incision during a trial of labor has the potential to be a physically and emotionally devastating event. In their recent editorial assessing the risks of a trial of labor, Sachs et al7 enumerate the number of neonatal and maternal complications that can occur after a uterine rupture and note the medical and legal costs that these complications can generate. Consequently, they suggest a reevaluation of any policy that mandates trials of labor and a weighing of the risks and benefits that result from this policy.
In view of these concerns as well as those of other authors regarding the safety of a trial of labor,6,17 this analysis was undertaken to formally assess the health and economic consequences of a policy that encourages vaginal birth after one prior low transverse cesarean delivery. We examined the reproductive life of a hypothetical cohort of 100,000 women whose first birth was a cesarean through a low transverse incision. The results of our model suggest that the consequences of proceeding with routine elective cesarean delivery for their second birth are significant, with an additional 117,748 cesarean deliveries, 5500 maternal morbid events, and $179 million cost incurred. The prevention of one major adverse neonatal outcome would require 1591 cesarean deliveries and $2.4 million.
As with any decision analysis, the results depend on the variables included and the values that are assigned to those variables. To demonstrate the validity of our model, we tested a wide range of probability and cost estimates. The robustness of the model's results are supported by a sensitivity analysis that, despite changes in single or multiple variables, continued to demonstrate that the avoidance of one major adverse neonatal outcome required more than $1 million and at least 806 additional cesarean deliveries. In some scenarios, such as low frequency of neonatal neurologic injury or death after uterine rupture, economic costs and adverse maternal health events are substantially higher.
It is useful to compare these results with cost-effectiveness analyses that examined other potential indications for cesarean delivery. Randolph et al15 assessed the policy of performing cesarean delivery for women presenting with an outbreak of genital herpes at the time of labor. That analysis, in which the neonatal complications were similar to those of the present analysis (death or permanent neurologic injury), found that for women with a history of genital herpes, more than 1500 cesarean deliveries and $2.5 million were needed to avoid one poor neonatal outcome. Rouse et al21 found that nearly $5 million and more than 2000 cesarean deliveries were needed to prevent one permanent brachial plexus injury in children of women without diabetes mellitus who had an ultrasonographic diagnosis of macrosomia. Both studies concluded that the medical consequences and economic costs of undergoing cesarean delivery for those indications were unacceptably high. These conclusions support our judgment that the risks and costs associated with routine elective repeat cesarean delivery are excessive.
Inevitably, not all possible events that are related to the decision to proceed with a trial of labor can be modeled. For example, infants delivered vaginally could incur other infection or trauma, whereas those delivered by cesarean might be at greater risk for respiratory complications. From a maternal perspective, a vaginal delivery could increase the risks of morbidities such as perineal fistula and anal incontinence, whereas a cesarean delivery might increase the chance of future complications such as placenta previa or accreta.28 However, we attempted to include the most significant, frequent, and well-defined events that occur throughout a woman's reproductive life, and we do not believe that the omission of other scenarios would significantly or systematically bias the model toward either policy.
An understanding of the relative costs and risks incurred by undergoing an elective repeat cesarean delivery does not translate into a policy of mandated trial of labor. There are some patients, for example, with relatively high risk of uterine rupture and low possibility of vaginal delivery, for whom an elective cesarean delivery is substantially more cost effective. Each patient's clinical presentation must be assessed, with treatment decisions based on this evaluation. We do believe, however, that a decision analysis, by quantifying the consequences of a treatment plan, can provide information that helps health care providers and their patients make more informed choices. Research into the impact that the consequences of different delivery policies have on a patient's quality of life could provide further assistance in the decision-making process.
1. Taffel SM, Planek PJ, Moien M, Kosary CL. 1989 U.S. cesarean section rate steadies—VBAC rate rises to nearly one in five. Birth 1991;18:73–7.
2. Flamm BL, Goings JR, Liu Y, Wolde-Tsadik G. Elective repeat cesarean delivery versus trial of labor: A prospective multicenter study. Obstet Gynecol 1994;83:927–32.
3. Phelan JP, Clark SL, Diaz F, Paul RH. Vaginal birth after cesarean. Am J Obstet Gynecol 1987;157:1510–5.
4. Miller DA, Diaz FG, Paul RH. Vaginal birth after cesarean: A 10-year experience. Obstet Gynecol 1994;84:255–8.
5. Cowan RK, Kinch RA, Ellis B, Anderson R. Trial of labor following cesarean delivery. Obstet Gynecol 1994:83:933–6.
6. McMahon MJ, Luther ER, Bowes WA, Olshan AF. Comparison of trial of labor with an elective second cesarean section. N Engl J Med 1996;335:689–95.
7. Sachs BP, Kobelin C, Castro MA, Frigoletto F. The risks of lowering the cesarean delivery rate. N Engl J Med 1999;340:54–7.
8. Farmer RM, Kirshbaum T, Potter D, Strong TH, Medearis AL. Uterine rupture during trial of labor after previous cesarean section. Am J Obstet Gynecol 1991:165:996–1001.
9. Depp R. Cesarean delivery and other surgical procedures. In: Gabbe SG, Niebyl JR, Simpson JL, eds. Obstetrics: Normal and problem pregnancies. New York: Churchill Livingstone, 1991:635–93.
10. Bashore RA, Phillips WH, Brinkman CR. A comparison of the morbidity of midforceps and cesarean delivery. Am J Obstet Gynecol 1990;162:1428–35.
11. Lasley DS, Eblen A, Yancey MK, Duff P. The effect of placental removal method on the incidence of postcesarean infections. Am J Obstet Gynecol 1997;176:1250–4.
12. Stanco LM, Schrimmer DB, Paul RH, Mishell DR. Emergency peripartum hysterectomy and associated risk factors. Am J Obstet Gynecol 1993;168:879–83.
13. Leung AS, Leung EK, Paul RH. Uterine rupture after previous cesarean delivery: Maternal and fetal consequences. Am J Obstet Gynecol 1993;169:945–50.
14. Jones RO, Nagashima AW, Hartnett-Goodman MM, Goodlin RC. Rupture of low transverse cesarean scars during trial of labor. Obstet Gynecol 1991;77:815–7.
15. Randolph AG, Washington AE, Prober CG. Cesarean delivery for women presenting with genital herpes. JAMA 1993;270:77–82.
16. Nielsen TF, Ljungblad U, Hagberg H. Rupture and dehiscence of cesarean section scar during pregnancy and delivery. Am J Obstet Gynecol 1989;160:569–73.
17. Scott JR. Mandatory trial of labor after cesarean delivery: An alternative viewpoint. Obstet Gynecol 1991;77:811–4.
18. Nelson KB, Ellenberg JH. Antecedents of cerebral palsy: Multivariate analysis of risk. N Engl J Med 1986;315:81–6.
19. Caughey AB, Shipp TD, Repke JT, Zelop C, Cohen A, Lieberman E. Trial of labor after cesarean delivery: The effect of previous vaginal delivery. Am J Obstet Gynecol 1998;179:938–41.
20. Abma JC, Chandra A, Mosher WD, Peterson LS, Piccinino LJ. Fertility, family planning, and women's health: New data from the 1995 National Survey of Family Growth. Vital Health Statistics. 1997;23:1–114.
21. Rouse DJ, Owen J, Goldenberg RL, Cliver SP. The effectiveness and costs of elective cesarean delivery for fetal macrosomia diagnosed by ultrasound. JAMA 1996;276:1480–6.
22. Traynor JD, Peaceman AM. Maternal hospital charges associated with trial of labor versus elective repeat cesarean section. Birth 1998;25:81–4.
23. Menzin J, Colditz GA, Regan MM, Richner RE, Oster G. Cost-effectiveness of enoxaparin versus low-dose warfarin in the prevention of deep-vein thrombosis after total hip replacement surgery. Arch Intern Med 1995;155:757–64.
24. Roberts WA, Kirkley SA, Newby M. A cost comparison of allogeneic and preoperatively or intraoperatively donated autologous blood. Anesth Analg 1996;83:129–33.
25. Waitzman NJ, Romano PS, Scheffler RM. Estimates of the economic costs of birth defects. Inquiry 1994;33:188–205.
26. Socol ML, Garcia PM, Riter S. Depressed Apgar scores, acid-base status, and neurologic outcome. Am J Obstet Gynecol 1994;170:991–9.
27. Kitzmiller JL, Elixhauser A, Carr S, Major CA, De Valencia M, Dang-Kilduff L, et al. Assessement of costs and management of gestational diabetes mellitus. Diabetes Care 1998;21:B123–30.
28. Chazotte C, Cohen WR. Catastrophic complications of previous cesarean section. Am J Obstet Gynecol 1990;163:738–42.
29. Brooten D, Roncoli M, Finkler S, Arnold L, Cohen A, Mennuti M. A randomized trial of early hospital discharge and home follow-up of women having cesarean birth. Obstet Gynecol 1994;84:832–8.
30. Drug topic redbook 1998. Montvale, New Jersey: Medical Economics Company, 1998.