PREDICTION OF VAGINAL BIRTH AFTER CAESAREAN DELIVERY AMARNATH BHIDEAB

GRINDING TEMPERATURE FIELD PREDICTION BY MESHLESS FINITE BLOCK METHOD
LINEAR DISCRIMINANT ANALYSIS WITH JACKNIFED PREDICTION
201301 YI JUN，LI TAIFU，HOU JIE ET AL DYNAMIC PREDICTION

29 CHILDREN’S PREDICTIONS AND RECOGNITION OF INCLINE MOTION CHANGING
58 1 RM STRENGTH PREDICTION 0 JEPONLINE JOURNAL OF
A METAANALYTIC INVESTIGATION OF CONSCIENTIOUSNESS IN THE PREDICTION OF

Validation of a Predicted Probability of Vaginal Birth After Caesarean Delivery

Prediction of Vaginal Birth After Caesarean Delivery


Amarnath Bhidea,b, Vedrana Carica and Sabaratnam Arulkumarana,b


[a] Department of Obstetrics and Gynecology

St. George’s University Hospital Foundation Trust and

[b] St. George’s University of London,

London, United Kingdom


Type of the article: Clinical article

Word count of the manuscript: 1902 words

Synopsis: The chance of a successful VBAC is 72%. Previous Caesarean section for failure to progress and Asian or African race are associated with unsuccessful VBAC.

Key words: Delivery, Caesarean section, Vaginal Births after Cesarean, Labour

Address for correspondence

Dr. Amar Bhide

Fetal Medicine Unit

Lanesborough Wing, 4th Floor

St. George’s Hospital

Blackshaw Road, SW17 0QT

United Kingdom

Tel: +44 20 87250080

Fax: +44 20 87250079

e-mail: [email protected]


Abstract

Objective: To examine factors associated with a successful vaginal birth following caesarean section, and to validate a previously published prediction model.

Methods: All women with one prior low transverse caesarean section who underwent a trial of labour with a cephalic singleton pregnancy at term were identified retrospectively over 14 years. Univariate analysis identified maternal demographics significantly associated with a successful vaginal birth following a Caesarean (VBAC). Logistic regression analysis identified factors independently associated with a successful VBAC. A prediction model was built. Predicted probabilities were compared to observed frequencies. Probabilities were also calculated based on previously published prediction model for validation.

Results: We identified 1463 women attempting VBAC after one low transverse Caesarean section. 1051 women out of 1463 (71.8%) had a successful VBAC. Asian (OR=1.6) or African (OR=1.8) race and prior Caesarean for failure to progress (OR= 6.4) were the only factors significantly associated with unsuccessful VBAC. Predicted and observed probability of successful VBAC showed a good correlation (Spearman’s rho = 0.905, p = 0.002). Previously published model worked less well.

Conclusion: Previous Caesarean performed for failure to progress in labour and Asian or African race are associated with unsuccessful VBAC. Performance of a published prediction model was inferior.


Word Count: 200 words


Introduction: Increasing rates of primary caesarean section have led to an increased proportion of women who have a history of prior caesarean delivery. Pregnant women with a previous caesarean section may be offered either planned vaginal birth after Caesarean (VBAC) or elective repeat caesarean section. The proportion of women who decline VBAC can further increase caesarean section rate. Women with a prior history of one uncomplicated lower-segment transverse caesarean section, in an otherwise uncomplicated pregnancy at term, with no contraindication for vaginal birth, should be able to discuss the option of planned VBAC and the alternative of a repeat caesarean section, during the first antenatal visit.

The Royal College of Obstetricians and Gynecologists in the United Kingdom suggests that every woman fulfilling the above criteria should discuss these two options prior to 36 weeks and they should be counselled that likelihood of achieving a planned VBAC is between 72 and 76 %[1]. The likelihood of a successful vaginal delivery is one of the most important factors in a decision making process in the antenatal counselling of these women. A number of factors have been described to be associated with an improved chance of successful VBAC. Indication for the Caesarean section, ethnicity, body mass index, prior vaginal delivery are some of them [2-11]. Assessing an individual woman’s risk of a successful vaginal birth following a Caesarean section is possible using them. There are several published prediction models for estimating the success of VBAC. Grobman et al [7], produced a model, which includes factors available at the first antenatal visit enabling a clinician to give the individual probability outcome to the women, and is perhaps the most widely used one. That outcome is a reasonably accurate assessment of a woman’s chance of achieving a VBAC if she opts for a trial of labour. The aim of the study was to examine factors associated with a successful attempt at vaginal birth following a Caesarean section in the study population, and to compare the accuracy of the Grobman prediction model when applied to the study population.


Materials and methods: We searched the maternity database of St George’s Hospital in London, from January 2000 to August 2013 for women with previous a Caesarean section attempting a vaginal birth. We excluded women with multiple pregnancies, more then one previous caesarean birth, preterm deliveries and those who underwent a pre-planned repeat caesarean section. We collected maternal demographic variables: maternal age, body mass index and ethnicity, smoking status, medical disease complicating pregnancy, recurrent indication of prior caesarean delivery (Caesarean section for failure to progress) and variables related to the obstetric history (any prior vaginal delivery). All these variables can be ascertained at the first antenatal visit. Univariate analysis was conducted to identify variables that were significantly different between those with successful versus failed attempt of vaginal birth following Caesarean section. Logistic regression was conducted using variables that were significantly different. Receiver operating characteristic (ROC) curve was constructed with the probabilities of successful vaginal birth following Caesarean section with the equation derived from the logistic regression analysis. Forced entry model was used. We also calculated probabilities predicted by a previously published prediction model[7] in an attempt to validate it. SPSS V10 was used for analysis of the data. A probability of 0.05 was considered significant.

We categorized probability of having a VBAC into deciles: <0.1, 0.1-0.2, 0.2-0.3, 0.3-0.4, 0.4-0.5, 0.5-0.6, 0.6-0.7, 0.7-0.8, 0.8-0.9 and >0.9 and calculated the predicted and observed number of successful VBAC in each group. The observed and predicted probabilities were plotted to produce a scatter plot.

Results: We identified 7253 women with a previous Caesarean section giving birth from January 2000 to August 2013. Out of these 7253 women, 4221 attempted labour (58.2%). Indication for the previous Caesarean section was known for 1463 women, and forms the cohort for this study. 1050 out of 1463 (71.9%) women who attempted VBAC had a successful vaginal birth. Descriptive statistics of the study population are shown in Table 1. A significantly higher proportion of women carrying a male fetus were unsuccessful in the attempt at VBAC. Prior vaginal birth did not alter the chance of successful VBAC. Birth weight was significantly higher with a failed VBAC attempt. In the univariate analysis, ethnicity, maternal BMI, induction of labour and prior Caesarean section for failure to progress in labour were factors associated with the probability of successful VBAC. For the logistic regression analysis we only included factors known at the time a delivery plan is made. Therefore, birth weight, gestational age at birth, induction of labour and fetal gender were not included. Table 2 shows that the only independent significant predictors were ethnicity and prior Caesarean section for failure to progress in labour. We also performed bootstrapping, and the odds ratios did not change. Table 3 shows observed and predicted probability data. Table 4 shows the predicted probabilities (minimum, maximum and three quantiles) of five women from the database with a previous caesarean section attempting VBAC.


The ROC curve had an area under the curve of 0.72 (0.69 - 0.76). The scatter diagram of the correlation between two different prediction models (Figure 1) shows a fair degree of variation, although the correlation was statistically significant (r = 0.53, p < 0.001). Figure 2 shows the scatter plot of predicted median probability versus observed probability for each of the deciles for the data. The correlation is highly significant (Spearman’s Rho = 0.905, p = 0.002).

Discussion:

The current data show that the overall success of a VBAC attempt is 71.9%. The result of the logistic regression analysis shows that prediction of a successful VBAC attempt is possible with moderate success (Area under ROC curve 0.72). Women of African or Asian ethnicity were more likely to fail in an attempt at VBAC. We also confirmed findings that race and history of a previous Caesarean for failure to progress in labour are two most important factors in determining a success of a successful VBAC. We were unable to confirm the effect of maternal booking BMI, prior vaginal delivery or maternal smoking status on the likelihood of a successful attempt at vaginal birth following Caesarean section. This may be due to a smaller sample size. Cahill et al[12] have previously reported that previous vaginal birth improves the chance of a successful VBAC, and that composite maternal morbidity is less common in women undergoing a repeat elective Caesarean as compared to those who attempt vaginal birth.

Several models have been published previously in order to predict the success of vaginal birth following Caesarean section. Grobman et al [7] published a prediction model based on a large number of women attempting VBAC. 73% women who attempted a vaginal birth, and the area under the ROC curve was 0.75 (95% confidence interval; 0.74–0.77) in that study. Our results compare favourably with this study. However, the area under the ROC curve for prediction model derived from the Grobman model was 0.61 (95% CI; 0.58 – 0.65) when applied to the current data. Thus, the predictive nomogram developed in the USA, which includes six variables identifiable at the first antenatal visit, predicts success of a trial of labour less accurately in our population. Recently, Fagerberg et al [11] studied a large cohort (n = 49472) of Swedish women undergoing an attempt at vaginal birth following Caesarean section, and developed a prediction model. They reported good correlation between predicted and observed VBAC success rates both using the Grobman model and one derived from their own data. Fagerberg et al [11] used the delivery unit Caesarean section rate as an explanatory variable, and were able to improve the accuracy of prediction further. It is interesting to note that the prediction with both models worked as well, despite the dissimilarities in the Caesarean section rate in Scandinavia versus USA. The reported success rate of VBAC can also be variable. Yokoi et al [13] reported that 664/725 (91.6%) women were successful in achieving a vaginal birth following caesarean.

It would be interesting to see how the prediction model will perform in real life. A failed attempt at VBAC after a long trial is not only disheartening for the mother, but also increases the morbidity as compared to a planned repeat Caesarean section [12]. The best use of health resources is if the majority of women with a good predicted success attempt to labour, and the majority of women with poor predicted success chose to undergo a planned Caesarean section. The model is valuable if it is successful in increasing the uptake of VBAC attempt by women with a high predicted success rate. On the other hand, use of such a model may put off women with a low predicted success from attempting VBAC, and may reduce the overall uptake of VBAC. Frost et al [14] in a qualitative study assessed women’s views on decision aids for decision making for the mode of delivery following a previous caesarean section. They reported that women value some form of structured information to help decisional conflict. Therefore, an individualised prediction model is likely to help women.


The uptake of VBAC is variable. It was reported as 8.5% in USA in 2006 [15]. It was 58.2% in the current study. The overall rate of Caesarean section is also variable in reported literature, and various factors are more or less likely to be associated with Caesarean section. It may be better to generate likelihood ratios of predicted success to take the background Caesarean section rate into account, since observed rate of success is so widely different.

It may be argued that the number of cases was smaller than some of the previously published reports [7,11]. However, the analysis is based on adequate number of cases. It is recommended that at least 10 events per predictor should be present in a logistic regression model [16]. In the prediction model described here, the ratio of number of predictors to the number of events is 200, and the confidence intervals around the predictors are narrow. When we compared the predicted probabilities to observed probabilities the correlation is good, showing that the prediction model works on the study population.

Retrospective nature of the data is a drawback. The prediction model was derived from the study population, and it is likely that the performance will be inferior in another population. Part of the reason why the performance of the Grobman model is inferior is because it was developed on a different population. However, the Grobman model has been validated on another population in the US [17], and the area under the curve was 0.70 (95% CI; 0.65 – 0.74), which was higher than that seen in the current population (0.61, 95% CI; 0.58 – 0.65). Table 3 shows that the number of women in the probability range are unevenly distributed. There were no predicted probabilities in the first two deciles. The number of women in the third and the seventh decile (one and seven respectively), is small, and predicted probability frequency has two peaks at the ends. We did not have a separate derivation and validation dataset. However, we used bootstrapping, and the log odds ratios did not change from those obtained from multivariate analysis.

We conclude that the chance of a successful VBAC in the current dataset was 72%. Previous Caesarean section for failure to progress and Asian or African race significantly reduced the chance of VBAC success. Published prediction models need to be validated before use.


Funding: This study was carried out without funding.


Ethics approval: Written confirmation was obtained from the local ethics committee that formal approval is not necessary for this retrospective study.


Conflicts of interest: None of the authors have any conflicts of interests to declare.

References

  1. Birth after previous Caesarean birth. RCOG Green top guideline No. 45, February 2007.

  2. Weinstein D, Benshushan A, Tanos V, Zilberstein R, Rojansky N. Predictive score for vaginal birth after cesarean section. Am J Obstet Gynecol. 1996;174(1 Pt 1):192-8.

  3. Flamm BL, Geiger AM. Vaginal birth after cesarean delivery: an admission scoring system. Obstet Gynecol. 1997;90(6):907-10..

  4. Smith GC, White IR, Pell JP, Dobbie R. Predicting cesarean section and uterine rupture among women attempting vaginal birth after prior cesarean section. PLoS Med. 2005;2(9):e252..

  5. Landon MB, Leindecker S, Spong CY, et al. The MFMU Cesarean Registry: factors affecting the success of trial of labor after previous cesarean delivery. Am J Obstet Gynecol. 2005;193(3 Pt 2):1016-23

  6. Srinivas SK, Stamilio DM, Stevens EJ, Odibo AO, Peipert JF, Macones GA. Predicting failure of a vaginal birth attempt after cesarean delivery. Obstet Gynecol. 2007;109(4):800-5.

  7. Grobman WA, Lai Y, Landon MB, et al. Development of a nomogram for prediction of vaginal birth after cesarean delivery. Obstet Gynecol. 2007;109(4):806-12.

  8. Hashima JN, Guise JM. Vaginal birth after cesarean: a prenatal scoring tool. Am J Obstet Gynecol. 2007;196(5):e22-3.

  9. Madaan M, Agrawal S, Nigam A, Aggarwal R, Trivedi SS. Trial of labour after previous caesarean section: the predictive factors affecting outcome. J Obstet Gynaecol. 2011;31(3):224-8.

  10. Metz TD, Stoddard GJ, Henry E, Jackson M, Holmgren C, Esplin S. Simple, validated vaginal birth after cesarean delivery prediction model for use at the time of admission. Obstet Gynecol. 2013;122(3):571-8.

  11. Fagerberg MC, Maršál K, Källén K. Predicting the chance of vaginal delivery after one cesarean section: validation and elaboration of a published prediction model. Eur J Obstet Gynecol Reprod Biol. 2015;188:88-94

  12. Cahill AG, Stamilio DM, Odibo AO, Peipert JF, Ratcliffe SJ, Stevens EJ, Sammel MD, Macones GA. Is vaginal birth after cesarean (VBAC) or elective repeat cesarean safer in women with a prior vaginal delivery? Am J Obstet Gynecol. 2006;195(4):1143-7.

  13. Yokoi A, Ishikawa K, Miyazaki K, Yoshida K, Furuhashi M, Tamakoshi K. Validation of the prediction model for success of vaginal birth after cesarean delivery in Japanese women. Int J Med Sci. 2012;9(6):488-91

  14. Frost J, Shaw A, Montgomery A, Murphy DJ. Women's views on the use of decision aids for decision making about the method of delivery following a previous caesarean section: qualitative interview study. BJOG. 2009;116(7):896-905.

  15. Vaginal birth after previous cesarean delivery. Practice Bulletin No. 115. American College of Obstetricians and Gynecologists. Obstet Gynecol. 2010;116(2 Pt 1):450-63.

  16. Peduzzi P, Concato J, Kemper E, Holford TR, Feinstein AR. A simulation study of the number of events per variable in logistic regression analysis. J Clin Epidemiol. 1996;49(12):1373-9

  17. Costantine MM, Fox KA, Pacheco LD, Mateus J, Hankins GD, Grobman WA, Saade GR. Does information available at delivery improve the accuracy of predicting vaginal birth after cesarean? Validation of the published models in an independent patient cohort. Am J Perinatol. 2011;28(4):293-8.




List of figure legends

Figure 1. Probability plot of Grobman et al method versus present method

Predicted probability calculated by the method of Grobman et al is plotted on the Y axis, and that by the current method on the X axis. The correlation is statistically significant (r = 0.53, p < 0.001)

Figure 2. Predicted probability of VBAC plotted against observed probability

Predicted probability (Y axis) of successful vaginal birth following a Caesarean section (VBAC) is plotted against observed probability (X-axis) in the dataset in the ten deciles of the data (Spearman’s rho = 0.905, p = 0.002).

Table 1. Descriptive statistics of the study population

Characteristic

Failed VBAC

n = 413

Successful VBAC

n = 1050

Significance (p)

Mean maternal age in yrs. (SD)

32.6 (4.8)

32.2 (5.2)

0.20

Mean booking BMI (SD)

26.8 (5.5)

26.0 (5.2)

0.021

Ethnicity



<0.001

White European (n = 691)

159 (23.0%)

532 (77%)


Asian (n = 371)

122 (32.9%)

249 (67.1%)


African (n = 278)

94 (33.8%)

184 (66.2%)


Mixed/other (n = 123)

38 (30.9%)

85 (69.1%)


Smoker (n = 71)

10

61

0.008

Prior vaginal birth

60 (14.5%)

143 (13.6%)

0.65

Previous CS for FTP

230 (55.7%)

169 (16.1%)

<0.001

Decimal GA in weeks at birth (SD)

39.7 (2.0)

40.1 (1.2)

<0.001

IOL

84 (20.3%)

117 (11.1%)

<0.001

Mean birth-weight in gm.(SD)

3486 (605)

3400 (485)

0.005

Gender (M/F)

234/179

519/531

0.013

Pre-pregnancy Diabetes

16

19

0.018

Medical condition complicating pregnancy

43

76

0.046

VBAC = Baginal birth after Cesarean, SD = Standard deviation, CS = Cesarea section, FTP = Failure to progress, M = male, F = Female

Table 2. Prediction of failed attempt at VBAC. Results of the logistic regression analysis.


Logistic regression


B

Exp(B)

95% CI

Constant

-1.171

0.31


Booking BMI

0.010

1.01

0.98 – 1.038

Smoking status

-0.81

0.445

0.189 – 1.048

Ethnicity




White

Referent

1.0


Asian

0.466

1.594

1.139 – 2.232

African

0.587

1.798

1.226 – 2.637

Other/mixed

0.333

1.396

0.794 – 2.452

Recurrent indication

1.854

6.388

4.808 – 8.487

Any medical condition complicating pregnancy

0.126

1.134

0.698 – 1.841

VBAC = Vaginal Birth After Cesarean Section, BMI = Body Mass Index. Statistically significant results are shown in bold

Table 3 Predicted and observed probability of successful VBAC

Predicted Probability

Number of women

Observed number of

vaginal births

Observed probability

0.0 - 0.1

0

0

-

- 0.2

0

0

-

- 0.3

1

0

0

- 0.4

157

48

0.29

- 0.5

84

51

0.61

- 0.6

81

50

0.62

- 0.7

7

3

0.43

- 0.8

308

254

0.82

- 0.9

535

441

0.82

- 1.0

38

36

0.95



Table 4. Predicted probabilities of five women from the database with a previous caesarean section attempting VBAC

Cut-point

Maternal BMI

Race

Smoking status

Previous CS for FTP

Any medical condition complicating pregnancy

Predicted probability

Minimum

44.2

African

Non-smoker

Yes

Yes

0.29

25th centile

20.6

White European

Non-smoker

Yes

No

0.51

Median

26.9

Asian

Non-smoker

No

No

0.80

75th centile

25.2

White European

Non-smoker

No

No

0.87

Maximum

17.6

White European

Smoker

No

No

0.94




ADDITIONAL FILE 4 –LIST OF PRIMERS GG2V3 PREDICTIONS WITH
ADDR SPECIAL THEME ISSUE “PREDICTION OF DELIVERY AND THERAPEUTIC
ADVANCES IN STRUCTURE PREDICTION OF INORGANIC COMPOUNDS ARMEL LE


Tags: after caesarean, birth after, delivery, after, bhideab, prediction, caesarean, birth, vaginal, amarnath