gms | German Medical Science

Background: With an increasing rate of caesarean sections (CS) as well as rising numbers of multiple pregnancies, valid classifications are needed. Benchmarking with regard to the CS rate and outcome is only possible, if a commonly accepted classification is established. The Robson classification provides a valid method to group cases with CS.

Research question: In this study we evaluated different methods of classification with regard to case related costs and outcome parameters (the Robson classification, the Swiss classification of procedures Schweizerische Operationsklassifikation (CHOP) and the Swiss Diagnosis Related Groups (SwissDRG)). We hypothesize first that insufficient classifications result in an inhomogeneity of costs and second a correlation of case related costs of mother and newborn.

Method: The study was conducted at the coding department and the department of obstetrics and gynecology Inselspital, University hospital of Berne, Switzerland. The study population contains inpatient cases of the Inselspital from 2014 until 2017. Administrative and health data were extracted from Clinical and Business Data Warehouse and from the discharge documentation (text mining). Cases were classified by a SQL query using Robson criteria and by CHOP categories and SwissDRG algorithm. The log10 transformed data was tested for normality and equal distribution and assessed for kurtosis. Mean, median, variance, linear modelling, analysis of variance, t-testing and computation of p-values and variance reduction were executed. Between-group comparisons and t-test were performed. The correlation of costs were analyzed. Descriptive statistic were used to show differences of clinical indicators.

Results: An automated query to classify the cases with CS according to Robson could be implemented at the Clinical Data Warehouse (CDWH) and successfully validated. The Robson classes could be associated to outcome parameters. A correlation of case related costs of mother and child could be shown for distinct groups, but also negative correlation occurred. The impact on the distribution of case related costs using the three classifications could be demonstrated. The values of the tests for distribution, ANOVA and the paired t-test described clearly the variety of the data and the difficulty to define the predictor for case related costs in the study population, even with different classification methods adopted. The classification according to the CS definitions primary and secondary showed good results in the sense of distribution, amount of high outliers and interquartile range (as a predictor) when used as a single criterion, but not when applied to Robson classes or DRGs. The Robson classification demonstrated a highly usable method to aggregate relevant obstetric information and corresponded to clinical indicators and SwissDRG lier type. Furthermore, benchmarking cases of mother and child with cephalic on term pregnancies for outcome and complications becomes possible. But it did not improve the homogeneity and distribution within the DRGs O01 and will not significantly contribute to the SwissDRG system.

Discussion: As the Robson criteria were already suggested by the Kommission Qualitätssicherung der gynécologie suisse / SGGG in 2015, this study will enhance the discussion to adopt the classification in Switzerland. Analysis of outcome data might influence quality monitoring. With this study it could be demonstrated, that a complex query on routinely collected health data could serve for medical classification, benchmarking and monitoring of quality and outcome. Risk-stratification might be conducted using this data set and should be the next step in order to evaluate the Robson criteria and outcome indicators.

Practical implication: The Robson classification has the capacity to improve international comparability, national benchmarking and improvement of quality indicators and the reimbursement system SwissDRG. With rising numbers in CS (elective and emergency), rising numbers of multiple pregnancies and rising numbers of women with repeat caesarean it is essential to improve the data base by implementing valid classifications to work with. To interpret outcome parameters and to recommend the right mode of delivery, clinical information and validity of data is necessary. The Robson criteria provide this capability.

The developed infrastructure and SQL algorithm could serve as a model for next steps of applying medical classification on routinely collected health data, e.g tumor classifications. Future work should go deeper into the evaluation by stratifying the Robson classes for risk and comorbidities. As refined Robson criteria exist and are already implemented in different health systems they should also be evaluated and might lead to more significant results than we could produce in this project. As to medical homogeneity it would be sensible to adjust the Robson criteria as suggested in different studies.