Measuring Variability In Acute Myocardial Infarction Coding Using A Statistical Process Control and Probabilistic Temporal Data Quality Control Approaches

Acute Myocardial Infarction (AMI) is frequently reported when coding hos-pital encounters, being commonly monitored through acute care outcomes. Variability in clinical coding within hospital administrative databases, how-ever, may indicate data quality issues and thereby negatively affect quality assessment of care delivered to AMI patients, apart from impacting health care management, decision making and research as a whole. In this study, we applied statistical process control and probabilistic temporal data quality control approaches to assess inter-hospital and temporal variability in coding AMI episodes within a nationwide Portuguese hospitalization database. The application of the present methodologies identified affected data distribu-tions that can be potentially linked to data quality issues. A total of 12 out of 36 institutions substantially differed in coding AMI when compared to their peers, mostly presenting lower than expected incidences of AMI. Results al-so indicate the existence of abnormal temporal patterns demanding addition-al investigation, as well as dissimilarities of temporal data batches in the pe-riods comprising the recent transition to the International Classification of Diseases, 10th revision, Clinical Modification (ICD-10-CM) and changes in the Diagnosis-Related Group (DRG) software. Hence, the main contribution of this paper is the use of reproducible, feasible and easy-to-interpret meth-ods that can be employed to monitor the variability in clinical coding and that could be integrated into data quality assessment frameworks.