Inter-hospital comparison of Cesarean delivery rates should not be considered to reflect quality of care without consideration of patient heterogeneity: An observational study

Benchmarking Cesarean Delivery Rates using Machine Learning-Derived Optimal Classification Trees

OBJECTIVE

To establish a case-adjusted hospital-specific performance evaluation tool using machine learning methodology for cesarean delivery.

DATA SOURCES

Secondary data were collected from patients between 1/1/2015-2/28/2018 using a hospital's "Electronic Data Warehouse" database from Illinois, USA.

STUDY DESIGN

The machine learning methodology of Optimal Classification Trees (OCT's) was used to predict cesarean delivery rate by physician group, thereby establishing the case-adjusted benchmarking standards in comparison to the overall hospital cesarean delivery rate. Outcomes of specific patient populations of each participating practice were predicted, as if each were treated in the overall hospital environment. The resulting OCTs estimate physician group expected cesarean delivery outcomes, both aggregate and in specific clinical situations.

DATA COLLECTION/EXTRACTION METHODS

12,841 singleton, vertex, term deliveries, cared for by practices with ≥50 births.

PRINCIPAL FINDINGS

The overall rate of cesarean delivery was 18.6% (n = 2384), with a range of 13.3%-33.7% amongst 22 physician practices. An optimal decision tree was used to create a prediction model for the overall hospital which defined 23 patient cohorts, divided by 46 nodes. The model's performance for prediction of cesarean delivery is as follows: area under the curve- 0.73, sensitivity- 98.4%, specificity- 16.1%, positive predictive value 83.7%, negative predictive value 70.6%. Comparisons with the overall hospital's specific-case adjusted benchmark groups revealed that several groups outperformed the overall hospital and some practice groups underperformed in comparison to the overall hospital.

CONCLUSIONS

OCT benchmarking can assess physician practice specific case-adjusted performance, both overall and clinical situation specific, and can serve as a valuable tool for hospital self-assessment and quality improvement. This article is protected by copyright. All rights reserved.

Caesarean section rates in primigravid women categorised by age and BMI

The rising caesarean section (CS) rate is a complex issue, particularly in an increasingly heterogenous nulliparous population. The study aim was to stratify the CS rate in nulliparous women by age and BMI to determine if any difference existed. This was a retrospective review of CS procedures of nulliparous women in two centres in Ireland (2014 through 2017). Data were obtained for 17,177 women from the hospital databases and CS procedures determined for each age and BMI category. Significant differences were observed when CS rates were stratified in this manner. The CS rates for women <20 years/BMI < 18.5 was 8.8 versus 57.6% for women 35 - 39 years/BMI 30 - 34 and 76 - 100% for all women >45 years (p<.005). The development of customised charts subdivided by age and BMI may be a useful counselling tool and assist in the comparison of rates between units.Impact statementWhat is already known on this subject? It is well known that along with rising CS rates globally, there have also been significant changes in maternal demographics-with increasing maternal age at first birth and increasing maternal BMI. It is well established that both of these factors affect the rate of CS in a population.What do the results of this study add? This study sought to stratify the CS rate in nulliparous women by age and BMI to determine if any difference existed. The results of the study showed an increasing CS rate for increasing age and BMI categories that was statistically significant.What are the implications of these findings for clinical practice and/or further research? Additional research using larger population data sets could allow the development of customised charts for nulliparous women subdivided by age and BMI which could act as a useful counselling tool in clinical practice, as well as assist in the comparison of CS rates between units.

The prediction of morbidity related to vaginal delivery in nulliparous women - A secondary analysis from the genesis multicenter trial

OBJECTIVE

In the prospective multicenter Genesis study, we developed a prediction model for Cesarean delivery (CD) in term nulliparous women. The objective of this secondary analysis was to determine whether the Genesis model has the potential to predict maternal and neonatal morbidity associated with vaginal delivery.

STUDY DESIGN

The national prospective Genesis trial recruited 2,336 nulliparous women with a vertex presentation between 39 + 0- and 40 + 6-weeks' gestation from seven tertiary centers. The prediction model used five parameters to assess the risk of CD: maternal age, maternal height, body mass index, fetal head circumference and fetal abdominal circumference. Simple and multiple logistic regression analyses were used to develop the Genesis model. The risk score calculated using this model were correlated with maternal and neonatal morbidity in women who delivered vaginally: postpartum hemorrhage (PPH), obstetric anal sphincter injury (OASI), shoulder dystocia, one- and five-minute Apgar score ≤ 7, neonatal intensive care (NICU) admission, cephalohematoma, fetal laceration, nerve palsy and fractures. The morbidities associated with spontaneous vaginal delivery were compared with those associated with operative vaginal delivery (OVD). The likelihood ratios for composite morbidity and the morbidity associated with OVD based on the Genesis risk scores were also calculated.

RESULTS

A total of 1,845 (79%) nulliparous women had a vaginal delivery. A trend of increasing intervention and morbidity was observed with increasing Genesis risk score, including OVD (p < 0.001), PPH (p < 0.008), NICU admission (p < 0.001), low Apgar score at one-minute (p < 0.001) and OASI (p = 0.009). The morbidity associated with OVD was significantly higher compared to spontaneous vaginal delivery, including NICU admission (p < 0.001), PPH (p = 0.022), birth injury (p < 0.001), shoulder dystocia (p = 0.002) and Apgar score of<7 at one-minute (p < 0.001). The positive likelihood ratios for composite outcomes (where the OVD was excluded) increases with increasing risk score from 1.005 at risk score of 5% to 2.507 for risk score of>50%.

CONCLUSION

In women who ultimately achieved a vaginal birth, we have shown more maternal and neonatal morbidity in the setting of a Genesis nomogram-determined high-risk score for intrapartum CD. Therefore, the Genesis prediction tool also has the potential to predict a more morbid vaginal delivery.