One-way sensitivity analyses were performed for all variables. These sensitivity analyses revealed that changes in three variables—the probability of antepartum hemorrhage with expectant management, the probability of perinatal mortality associated with antepartum hemorrhage requiring delivery, and the probability of ICU admission after scheduled cesarean delivery—affected the order of the strategy rankings. Figure 1 demonstrates that there are two thresholds at which the most effective strategy is no longer delivery at 34 weeks of gestation. If the risk of antepartum hemorrhage necessitating delivery is between 1% and 7% at 34 weeks of gestation, then the preferred strategy is to deliver at 37 weeks of gestation without performing amniocentesis. Only when the risk of serious hemorrhage necessitating delivery is less than 1%, a probability outside the range of published values, does expectant management until 39 weeks of gestation become the most preferred strategy. With respect to perinatal mortality associated with antepartum hemorrhage, only when the risk of perinatal mortality related to hemorrhage is 0.1% or lower is the preferred strategy delivery at 37 weeks of gestation without performing amniocentesis. Varying the probability of ICU admission after scheduled cesarean delivery resulted in the preferred strategy being either delivery at 34 weeks of gestation or delivery at 37 weeks of gestation without amniocentesis. When the risk of ICU admission after emergent cesarean delivery is 36% or higher, the preferred strategy is a scheduled delivery at 34 weeks of gestation after administration of steroids. In cases in which risk of ICU admission after scheduled cesarean delivery is higher than 36%, only in situations in which the scheduled delivery is associated with a higher maternal risk of ICU admission than emergent delivery does the preferred strategy switch in certain instances to delivery at 37 weeks of gestation with no testing for fetal lung maturity. Such a situation seems implausible.
The results were also determined after incorporation of a more liberal betamethasone administration policy. Under this policy, mothers delivering in a nonemergent fashion between 34 and 38 weeks of gestation received betamethasone 2 days to 1 week before delivery. The only mothers who did not receive betamethasone were those for whom amniocentesis demonstrated lung maturity, those delivering emergently, or those delivered at 39 weeks of gestation. Individuals for whom amniocentesis demonstrated fetal lung immaturity were administered betamethasone and delivered 1 week later. When administered beyond 34 weeks, steroids were still presumed to decrease the risks of RDS by 50%. Even under this alternate policy, no material differences were noted in the quality-adjusted life years or ranking of the delivery strategies from those displayed in Table 5.
Although two-way sensitivity analyses were performed for all combinations of variables, none of these analyses led to a change in the preferred strategy except those analyses that included at least one of the three variables of interest identified by the one-way sensitivity analyses. Figure 2 is a two-way sensitivity analysis examining the relationship between the probability of ICU admission after emergent delivery and the risk of hemorrhage requiring emergent delivery. The graph depicts that if an individual's risk of hemorrhage necessitating emergent delivery is approximately 11% or higher, delivery at 34 weeks of gestation after steroids is always the preferred strategy. When the risk of hemorrhage is between 2% and 8%, the preferred strategy is delivery at 37 weeks of gestation without testing for fetal lung maturity. However, when the risk of hemorrhage is less than 2%, the preferred strategy shifts to planned delivery at 39 weeks of gestation only if the risk of ICU admission after emergent cesarean delivery is higher than 1.5%.
A two-way sensitivity analysis examining the relationship between probability of ICU admission after emergent delivery and probability of perinatal mortality during an acute hemorrhagic episode revealed that when the probability of perinatal mortality is 0.6% or higher, the preferred strategy is always delivery at 34 weeks of gestation after administration of steroids, irrespective of the maternal risk of ICU admission after an emergent delivery. When the risk of perinatal mortality reaches 0.1% or less, the preferred strategy switches to delivery at 37 weeks of gestation, except when the risk of maternal ICU admission after emergent cesarean delivery is between 1.5% and 8.4%.
An additional two-way sensitivity analysis examining the relationship between probability of perinatal mortality during an acute hemorrhagic event and the maternal utility assigned to ICU admission revealed that for situations in which the probability of perinatal mortality during acute hemorrhage is 0.5% or higher, the preferred strategy remains delivery at 34 weeks of gestation after administration of steroids, regardless of the maternal utility assigned to ICU admission. At lower probabilities of perinatal mortality, the preferred strategy becomes dependent on the maternal utility for ICU admission.
This decision analysis indicates that in women with placenta previa and ultrasonographic evidence of placenta accreta, the preferred strategy for timing of delivery under a variety of (but not all) circumstances is a scheduled delivery at 34 weeks of gestation. Additionally, the model suggests that in circumstances when delivery at a given gestational age later than 34 weeks is the preferred option, the advantages of amniocentesis for confirmation of fetal lung maturity never outweigh the disadvantages.
The infrequency of individuals with placenta previa and accreta likely precludes the undertaking of a prospective trial to address the proper timing of delivery, making decision analysis potentially the only means by which to investigate the question. The values placed into the model were ascertained through a thorough search of the literature and included utilities that were contained in a recently published dataset dealing with the explicit pediatric disease states in question.23 Moreover, the elucidation of an optimal strategy for these individuals is of particular relevance today. The increasing incidence of cesarean delivery likely portends a future increase in this population of individuals, meaning that this clinical scenario may become more frequent in the coming years. Although obstetrics as a field has made significant advances with respect to individual safety, one complication that still not infrequently jeopardizes both maternal and perinatal health is peripartum hemorrhage.
This decision analysis demonstrates two particularly salient findings with respect to the timing of delivery in women with placenta previa and suspected accreta. The first point is that under most reasonable clinical circumstances, there is no benefit to be gained by expectantly managing these individuals beyond 37 weeks of gestation and, in many cases, delivery at 34 weeks of gestation is preferred. The second point is that in those individuals not delivered by 36 weeks of gestation, the practice of using amniocentesis for determination of fetal lung maturity does not improve outcome. Although the current literature detailing the probability of the events related to emergent delivery of women with placenta previa and suspected accreta is not definitive enough to allow the unequivocal establishment of a single ideal gestational age at which to deliver these individuals at high risk, the preferred gestational age is almost certainly at or beyond 34 weeks but not later than 37 weeks.
Examination of the two-way sensitivity analyses is notable for the stepwise appearance of the depiction in Figure 2. The preferred strategy progresses from a scheduled 34-week delivery under most circumstances to scheduled 37-week delivery under limited circumstances, to a 39-week delivery under unlikely circumstances. As the probabilities of adverse outcomes related to expectant management are decreased, the marginal benefits gained by delivery only 1 or 2 weeks later are outweighed by the additional benefit associated with delivery of a term neonate.
The differences in quality-adjusted life years among the various strategies certainly bear mentioning. Although a fair amount of controversy exists regarding the number of quality-adjusted life years that represent a clinically significant gain, Richardson et al36 have argued that a gain in quality-adjusted life years of 2 or more months should be considered an important gain. This is in contrast to gaining just a few days, which would more often be deemed as representing a “toss up.” Applying the standard of 2 months (2 months out of a 12-month year, or 0.166) to the differences in the ranking strategies would suggest that the marginal benefit of delivering at 34 weeks of gestation compared with 35 weeks of gestation achieves a gain in quality-adjusted life years of borderline clinical significance (just less than 2 months), whereas delivering at 34 weeks of gestation compared with 36 weeks of gestation or any strategy resulting in a delivery beyond 36 weeks of gestation achieves a gain of more than 3 months in quality-adjusted life years.
This analysis has several limitations. First, this is not a clinical trial. Because of the relatively infrequent occurrence of placenta accreta, it is unlikely that a prospective, randomized clinical trial will ever be performed to address the decision of delivery timing in this population of individuals. Utilization of a decision analysis is a sound alternative to help inform our decision-making, particularly when competing risks of preterm birth and the consequences of emergent delivery are weighed against one another. Another limitation, inherent to all decision analyses, is the choice and precision of the estimates included in the model. Although some probabilities were derived from large studies containing thousands of participants, other probability estimates relied on studies containing substantially smaller population sizes. Also, because of the relative paucity of data in the literature detailing all possible outcomes in women with placenta previa and ultrasonographic evidence of accreta, as well as the dearth of studies addressing emergent compared with scheduled delivery in such individuals, this analysis utilized numerous publications that reported on outcomes for women with similar morbidities, such as those having a peripartum hysterectomy. Similarly, the incidence and neonatal consequences of emergent delivery were derived from the abundance of data detailing clinical outcomes in women with placenta previa, although not always with accreta. Because of the uncertainty of the base case estimates, extensive sensitivity analyses that included varying the variable inputs over wide ranges were performed. Thus, although some may disagree with the baseline values chosen, few would disagree that the range of values over which the sensitivity analyses were conducted were likely to contain all reasonable estimates.
As demonstrated by the decision tree model, the preferred strategy for timing of delivery in individuals with placenta previa and placenta accreta discovered by ultrasonography under a variety of circumstances is delivery at 34 weeks of gestation. This result takes into account both long-term and short-term outcomes of the mother and child. This result does not imply that all women with this condition should be delivered at this gestational age. As is also clear from the model, there are some circumstances (ie, when the probability of antepartum hemorrhage or the adverse consequences from such bleeding are lower) that would lead one to advocate for awaiting planned delivery until approximately 37 weeks of gestation. Unfortunately, there are no validated predictive models at this point in time that would allow physicians to know with confidence which individuals are most likely to bleed or have adverse outcomes related to such hemorrhagic episodes. Thus, clinical judgment to the best of one's ability will be crucial in discussions and decision-making with the individual. The results of the present decision analysis should support the concept that planned delivery at 34 weeks of gestation may be warranted under many circumstances and that there is little reason to believe that awaiting labor beyond 37 weeks of gestation is a beneficial strategy. Furthermore, this analysis should support the concept that for any given gestational age at which delivery is planned for women with a previa and suspected accreta, there seems little reason to use amniocentesis for confirmation of fetal lung maturity.
1. Wu S, Kocherginsky M, Hibbard JU. Abnormal placentation: twenty-year analysis. Am J Obstet Gynecol 2005;192:1458–61.
2. Clark SL, Koonings PP, Phelan JP. Placenta previa/accreta and prior cesarean section. Obstet Gynecol 1985;66:89–92.
3. Warshak CR, Eskander R, Hull AD, Scioscia AL, Mattrey RF, Benirschke K, et al. Accuracy of ultrasonography and magnetic resonance imaging in the diagnosis of placenta accreta. Obstet Gynecol 2006;108:573–81.
4. Warshak CR, Ramos GA, Eskander R, Benirschke K, Saenz CC, Kelly TF, et al. Effect of predelivery diagnosis in 99 consecutive cases of placenta accreta. Obstet Gynecol 2010;115:65–9.
5. Briery CM, Rose CH, Hudson WT, et al. Planned vs emergent cesarean hysterectomy. Am J Obstet Gynecol 2007;197:154.e1–5.
6. Gonsoulin W, Kennedy RT, Guidry KH. Elective versus emergency cesarean hysterectomy cases in a residency program setting: a review of 129 cases from 1984 to 1988. Am J Obstet Gynecol 1991;165:91–4.
7. Grosvenor A, Silver R, Porter TF, Zempolich K. Optimal management of placenta accreta. Am J Obstet Gynecol 2006;10:261.
8. Lu PY, Pastorek JG II, Letellier RL, Bey MA. Elective vs emergency cesarean hysterectomy on a teaching service—1981 to 1991. South Med J 1997;90:50–4.
9. Chestnut DH, Dewan DM, Redick LF, Caton D, Spielman FJ. Anesthetic management for obstetric hysterectomy: a multi-institutional study. Anesthesiol 1989;70:607–10.
10. Plauche WC, Wycheck JG, Ianessa MJ, Rousset KM, Mickal A. Cesarean hysterectomy at Louisiana State University, 1975 through 1981. South Med J 1983;76:1261–3.
11. Plauche WC, Gruich FG, Bourgeois MO. Hysterectomy at the time of cesarean section: analysis of 108 cases. Obstet Gynecol 1981;58:459–64.
12. Cotton DB, Read JA, Paul RH, Quilligan EJ. The conservative aggressive management of placenta previa. Am J Obstet Gynecol 1980;137:687–95.
13. Silver R, Depp R, Sabbagha RE, Dooley SL, Socol ML, Tamura RK. Placenta previa: aggressive expectant management. Am J Obstet Gynecol 1984;150:15–22.
14. Brenner WE, Edelman DA, Hendricks CH. Characteristics of individuals with placenta previa and results of expectant management. Am J Obstet Gynecol 1978;132:180–91.
15. Angstmann T, Gard G, Harrington T, Ward E, Thomson A, Giles W. Surgical management of placenta accreta: a cohort series and suggested approach. Am J Obstet Gynecol 2010;202:38.e1–9
16. McShane PM, Heyl PS, Epstein MF. Maternal and perinatal morbidity resulting from placenta previa. Obstet Gynecol 1985;65:176–82.
17. Robertson PA, Sniderman SH, Laros RK Jr, Cowan R, Heilbron D, Goldenberg RL, et al. Neonatal morbidity according to gestational age and birth weight from five tertiary care centers in the United States, 1983 through 1986. Am J Obstet Gynecol 1992;166:1629–45.
18. Roberts D, Dalziel S. Antenatal corticosteroids for accelerating fetal lung maturation for women at risk of preterm birth. The Cochrane Database of Systematic Reviews 2006, Issue 3. Art. No. : CD004454. DOI: 10.1002/14651858.CD004454.pub2.
19. Steinfeld JD, Samuels P, Bulley MA, Cohen AW, Goodman DB, Senior MB. Utility of the TDx test in the assessment of fetal lung maturity. Obstet Gynecol 1992;79:460–4.
20. Parvin CA, Kaplan LA, Chapman JF, McManamon TG, Gronowski AM. Predicting respiratory distress syndrome using gestational age and fetal lung maturity by fluorescent polarization. Am J Obstet Gynecol 2005;192:199–207.
21. Bildirici I, Moga CN, Gronowski AM, Sadovsky Y. Mean weekly increment of amniotic fluid TDX-FLM II ratio is constant during the latter part of pregnancy. Am J Obstet Gynecol 2005;193:1685–90.
22. Kerridge RK, Glasziou PP, Hillman KM. The use of quality adjusted life-years (QALYs) to evaluate treatment in intensive care. Anaesth Int Care 1995;23:322–31.
23. Carroll AE, Downs SM. Improving decision analyses: parent preferences (utility values) for pediatric health outcomes. J Pediatr 2009;155:21–5.
24. Tomashek KM, Shapiro-Mendoza CK, Davidoff MJ, Petrini JR. Differences in mortality between late-preterm and term singleton infants in the United States, 1995–2002. J Pediatr 2007;151:450–6, 456.e1.
25. Tengs TO, Wallace A. One thousand health-related quality of life estimantes. Med Care 2000;38:583–637.
26. Melamed N, Klinger G, Tenenbaum-Gavish K, Herscovici T, Linder N, Hod M, et al. Short-term neonatal outcome in low-risk, spontaneous, singleton, late preterm deliveries. Obstet Gynecol 2009;114:253–60.
27. Bastek JA, Sammel MD, Pare E, et al. Adverse neonatal outcomes: examining the risks between preterm, late preterm, and term infants. Am J Obstet Gynecol 2008;199:367.e1–8.
28. Stotland NE, Lipschitz LS, Caughey AB. Delivery strategies for women with a previous classic cesarean delivery: a decision analysis. Am J Obstet Gynecol 2002;187:1203–8.
29. Herbert WN, Chapman JF. Clinical and economic considerations associated with testing for fetal lung maturity. Am J Obstet Gynecol 1986;155:820–3.
30. Petrini JR, Dias T, McCormick MC, Massolo ML, Green NS, Escobar GJ. Increased risk of adverse neurological development for late preterm infants. J Pediatr 2009;154:169–76.
31. Moster D, Lie RT, Markestad T. Long-term medical and social consequences of preterm birth. N Engl J Med 2008;359:262–73.
32. Young PC, Glasgow TS, Li X, Guest-Warnick G, Stoddard G. Mortality of late-preterm (near-term) newborns in Utah. Pediatrics 2007;119:e659–65.
33. Zlatnik M, Cheng YW, Norton ME, Thiet MP, Caughey AB. Placenta previa and risk of preterm delivery. J Matern Fetal Neonatal Med 2007;20:719–23.
34. Zlatnik M, Little S, Kohli P, et al. When should women with placenta previa be delivered? A decision analysis. Am J Obstet Gynecol 2006;10:69.
35. Madan I, Puri I, Jain NJ, et al. Characteristic of obstetric intensive care unit admissions in New Jersey. J Matern Fetal Neo Med 2009;22:785–90.
36. Richardson WS, Detsky AS. Users' guides to the medical literature. VII. How to use a clinical decision analysis. B. What are the results and will they help me caring for my patients? JAMA 1995;273:1610–3.