Reply
2018; Elsevier BV; Volume: 219; Issue: 4 Linguagem: Inglês
10.1016/j.ajog.2018.04.057
ISSN1097-6868
AutoresLisa D. Levine, Katheryne Downes,
ResumoWe thank Drs D’Souza, Abraham, and San Roman for their interest in our manuscript. We would like to address the individual concerns and inquiries within this combined response. An incredibly important point to reiterate to the readership is that our calculator was specifically derived and validated in a population of women undergoing a term induction of labor with intact membranes and an unfavorable cervix. This is an inherently different population from many other studies that have evaluated predictors of cesarean and failed induction, which often include women with a favorable cervix and/or women with ruptured membranes.1Tolcher M.C. Holbert M.R. Weaver A.L. et al.Predicting cesarean delivery after induction of labor among nulliparous women at term.Obstet Gynecol. 2015; 126: 1059-1068Crossref PubMed Scopus (56) Google Scholar, 2Luthy D.A. Malmgren J.A. Zingheim R.W. Leininger C.J. Physician contribution to a cesarean delivery risk model.Am J Obstet Gynecol. 2003; 188: 1579-1585Abstract Full Text Full Text PDF PubMed Scopus (37) Google Scholar, 3Crane J.M. Factors predicting labor induction success: a critical analysis.Clin Obstet Gynecol. 2006; 49: 573-584Crossref PubMed Scopus (148) Google Scholar, 4Vrouenraets F.P. Roumen F.J. Dehing C.J. van den Akker E.S. Aarts M.J. Scheve E.J. Bishop score and risk of cesarean delivery after induction of labor in nulliparous women.Obstet Gynecol. 2005; 105: 690-697Crossref PubMed Scopus (298) Google Scholar Therefore, we continue to stress the importance of limiting the applicability of this calculator to that specific population. Furthermore, the authors again urge clinicians not to use this calculator in isolation without taking into account other external clinical factors that may be at play. We thank Dr D’Souza and colleagues for the attempts to externally validate our calculator in a population outside the United States. He did not mention the demographics of the patient population in which the calculator was being evaluated, and it would be interesting to compare and contrast our populations. Assuming that Dr D’Souza’s hospital population is at least somewhat similar to the overall Toronto population, the women have a very different race/ethnic composition compared with our cohort: 77% of our derivation cohort was African American, whereas they make up 4000 g) as a predictor variable. For the derivation group, there were 38 women (7.8%) with birthweight >4000 g with no difference in the risk of cesarean between those with and without macrosomia (P = .14). While macrosomia has been shown to be a predictor of cesarean delivery, it is possible that we did not have that same finding because we used a strict labor protocol of when to call an induction failed and labor arrested. This labor protocol was used to remove practice variation and potential for provider bias that may occur during labor and delivery.2Luthy D.A. Malmgren J.A. Zingheim R.W. Leininger C.J. Physician contribution to a cesarean delivery risk model.Am J Obstet Gynecol. 2003; 188: 1579-1585Abstract Full Text Full Text PDF PubMed Scopus (37) Google Scholar One place for this bias to occur is if a provider thinks the fetus is large, they may have a lower threshold to perform a cesarean delivery. Furthermore, many of the studies that have evaluated macrosomia as a predictor of cesarean were not solely limited to women undergoing an induction with an unfavorable cervix,1Tolcher M.C. Holbert M.R. Weaver A.L. et al.Predicting cesarean delivery after induction of labor among nulliparous women at term.Obstet Gynecol. 2015; 126: 1059-1068Crossref PubMed Scopus (56) Google Scholar, 2Luthy D.A. Malmgren J.A. Zingheim R.W. Leininger C.J. Physician contribution to a cesarean delivery risk model.Am J Obstet Gynecol. 2003; 188: 1579-1585Abstract Full Text Full Text PDF PubMed Scopus (37) Google Scholar a group of women already at an increased risk of cesarean. Second is pelvic adequacy. This is likely correlated with patient height, which is already a variable included as a predictor of cesarean delivery. However, we do want to reiterate the importance of using other pieces of clinical information to predict success. We, the authors, continue to strongly remind the readership that this calculator is neither 100% accurate nor should it be used in isolation from other known clinical factors. Third, we chose to use modified Bishop score because the complete 5-point Bishop score was not available in the external CSL database. The variables that were predictors of cesarean and the effect they had on the determined calculation of the risk of cesarean were no different when we compared the full Bishop score with the modified Bishop score in the derivation data set. Fourth, information from the validation group (attached is a table with demographic information from the validation group). It is important to note there are many demographic differences between the derivation group and validation group. The fact the calculator remained valid in an external population that had different maternal characteristics increases the generalizability of the calculator.TableOverall maternal characteristics from the validation group (CSL database)Maternal characteristicn = 8466Age28 (6)Race White3789 (44.8) African American1969 (23.3) Hispanic1826 (21.6) Other882 (10.4)Insurance Private4131 (48.8) Public2027 (23.9) Self-pay99 (1.2) Other50 (0.6) Unknown2159 (25.5)BMI category, kg/m2 <25.01071 (12.6) 25.0–29.93230 (38.1) 30.0–34.92306 (27.2) 35.0–39.91089 (12.9) ≥40.0770 (9.1)Maternal height category <62.01867 (22.0) 62.0–63.91457 (17.2) 64.0–65.92285 (27.0) 66.0+2857 (33.7)Pregestational diabetes159 (1.9)Gestational diabetes540 (6.4)Chronic hypertension186 (2.2)Gestational hypertension425 (5.0)Preeclampsia459 (5.4)Postdate3257 (38.5)Nulliparous5639 (66.6)Cervical dilation at first examination1 (0.5–2)Modified Bishop score, median (IQR)3 (2–5)GA at delivery, median (IQR)39 (38–40)Indication for induction Postdate965 (11.4) Maternal909 (10.7) Fetal939 (11.1) Elective1431 (16.9) Unknown4213 (49.8)BMI, body mass index; GA, gestational age; IQR, interquartile range.Levine. Reply. Am J Obstet Gynecol 2018. Open table in a new tab BMI, body mass index; GA, gestational age; IQR, interquartile range. Levine. Reply. Am J Obstet Gynecol 2018. Fifth is criteria for failed induction and arrest disorders. Within the derivation group, cesarean delivery indications were ultimately up to the discretion of the provider who was managing the labor.6Levine L.D. Downes K.L. Elovitz M.A. Parry S. Sammel M.D. Srinivas S.K. Mechanical and pharmacologic methods of labor induction: a randomized controlled trial.Obstet Gynecol. 2016; 128: 1357-1364Crossref PubMed Scopus (88) Google Scholar Within the labor protocol that was utilized for the derivation group (trial), the authors encouraged the use of the current guidelines prior to diagnosing an arrest disorder.7Spong C.Y. Berghella V. Wenstrom K.D. Mercer B.M. Saade G.R. Preventing the first cesarean delivery: summary of a joint Eunice Kennedy Shriver National Institute of Child Health and Human Development, Society for Maternal-Fetal Medicine, and American College of Obstetricians and Gynecologists Workshop.Obstet Gynecol. 2012; 120: 1181-1193Crossref PubMed Scopus (0) Google Scholar The authors are unable to speak to the criteria utilized in the validation group because this was a publicly available database with minimal details regarding how indications for cesarean were determined. Dr San Roman had several concerns and comments regarding the manuscript that are addressed here. Maternal age was not included as a predictor variable in the calculator because it was not an independent risk factor for cesarean within our statistical testing. One large difference with our study vs many of the prior studies performed evaluating the risk of cesarean is that the derivation portion of this calculator was performed within a randomized trial that had a standard labor protocol utilized. Therefore, external factors, such as provider bias, would not come into play. Additionally, previously recognized risk factors that were determined by retrospective data may be inherently biased and affected by confounding by indication. The demographic makeup of the derivation group and validation group were different (table included), and therefore, our predictor variables are robust because they were derived in one population and validated in an external population with a different composition from the derivation group. In the example that Dr San Roman provided with the difference in cesarean for an 18 year old and 35 year old, our results would argue that it is not the age that is the biggest driver. Perhaps, instead, an obstetrician is more concerned about performing a cesarean in an 18 year old and therefore allows them to have a longer labor but has a lower threshold in a 35 year old woman (provider bias, confounding by indication). Deriving the predictor variables within a population that had a standardized labor protocol would eliminate the provider bias that Dr San Roman suggested. The concern regarding fetal weight has been addressed in the previous text in response to Dr Abraham. We appreciate and agree with the remarks regarding issues that can arise when calculated probabilities are grouped for the area under the curve (AUC) analysis. However, that is not what was done in our study. The AUC was calculated based on the continuous probability value (no groupings) that was generated from the calculator. In the case of the of the derivation group, the actual calculated probability of cesarean ranged from 1.52657% to 77.90622%. The graphics that divided calculated probability into 10 percentage point groupings were for general illustrative purposes only and were not used in the calculation of the AUC itself. We apologize for any confusion that this may have created. We appreciate Dr San Roman pointing out the 2 additional risk models for cesarean that are available on-line. As noted in the previous text, it is critical to remember that our calculator was specifically derived and validated in a population of women undergoing a term induction of labor with intact membranes and an unfavorable cervix. This is an inherently different population from other studies that have previously identified risk factors for cesarean and include women who may start labor with a favorable cervix and/or women who present in labor.1Tolcher M.C. Holbert M.R. Weaver A.L. et al.Predicting cesarean delivery after induction of labor among nulliparous women at term.Obstet Gynecol. 2015; 126: 1059-1068Crossref PubMed Scopus (56) Google Scholar, 2Luthy D.A. Malmgren J.A. Zingheim R.W. Leininger C.J. Physician contribution to a cesarean delivery risk model.Am J Obstet Gynecol. 2003; 188: 1579-1585Abstract Full Text Full Text PDF PubMed Scopus (37) Google Scholar, 3Crane J.M. Factors predicting labor induction success: a critical analysis.Clin Obstet Gynecol. 2006; 49: 573-584Crossref PubMed Scopus (148) Google Scholar, 4Vrouenraets F.P. Roumen F.J. Dehing C.J. van den Akker E.S. Aarts M.J. Scheve E.J. Bishop score and risk of cesarean delivery after induction of labor in nulliparous women.Obstet Gynecol. 2005; 105: 690-697Crossref PubMed Scopus (298) Google Scholar The examples that Dr San Roman provided are looking at overall risk of cesarean in a pregnancy, not specifically focusing on women undergoing an induction with an unfavorable cervix. While they are interesting tools, they do not address the same question nor the same population studied in our article. We thank you for the interest in our manuscript and appreciate the opportunity to address the authors’ concerns and comments. We hope this provides additional clarification. A validated calculator to estimate risk of cesarean after an induction of labor with an unfavorable cervixAmerican Journal of Obstetrics & GynecologyVol. 219Issue 4PreviewI am commenting on the article by Levine et al1 in which a predictive model was created to estimate the risk of cesarean delivery after labor induction with an unfavorable cervix. The following 5 variables were used in model construction: nulliparity, gestational age greater than or equal to 40 weeks, body mass index before delivery, height, and modified Bishop score at induction start. I commend the authors on derivation of this predictive model and subsequent validation. However, there are several flaws in this study that affect the interpretation and generalization of their conclusions. Full-Text PDF Concerns regarding a validated calculator to estimate risk of cesarean delivery after an induction of labor with an unfavorable cervixAmerican Journal of Obstetrics & GynecologyVol. 219Issue 4PreviewI believe that there are several issues that adversely affect the results of a recent study on estimating risk of cesarean delivery.1 The derivation data set included only 491 patients and failed to confirm the previously proven risk factor of maternal age. This contradicts many other studies, including references 5 and 9 from the study itself.2 Full-Text PDF Prediction calculator for induction of labor: no Holy Grail yet!American Journal of Obstetrics & GynecologyVol. 219Issue 4PreviewWe congratulate Levine et al1 on deriving and validating a model and nomogram for cesarean risk after term labor induction with an unfavorable cervix. With increasing numbers of pregnant women requiring labor induction for various indications, the importance of a tool such as this to predict the success of labor induction accurately is more pertinent than ever. Full-Text PDF
Referência(s)