Defining major surgical complications using administrative data in Ontario: a validation study.

Area of Science:

• Health Informatics
• Surgical Outcomes Research
• Data Validation

Background:

• Surgical complications are crucial outcomes in research but often lack validation when using administrative data.
• Limited validation exists for algorithms designed to capture surgical complications from health administrative data.

Purpose of the Study:

• To evaluate the diagnostic performance of an algorithm for capturing major surgical complications using health administrative data.
• To assess the accuracy of administrative data in identifying postoperative complications.

Main Methods:

• Retrospective study of 270 patients undergoing high-risk elective general surgery.
• Comparison of an administrative data algorithm against clinician-abstracted data from medical records.
• Evaluation using sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy.

Main Results:

• The administrative data algorithm achieved 72% sensitivity, 80% specificity, 82% PPV, 70% NPV, and 76% accuracy for a composite outcome of major surgical complications.
• 55% of patients experienced at least one major complication based on chart audit.
• Diagnostic performance varied, with poor accuracy for several individual complications.

Conclusions:

• Health administrative data can effectively capture a composite indicator of major surgical complications with adequate sensitivity and specificity.
• Further research is needed to develop precise algorithms for identifying specific surgical complications.