Retrospective Derivation and Validation of an Automated Electronic Search Algorithm to Identify Post Operative...

Area of Science:

• Medical Informatics
• Clinical Research
• Surgical Outcomes

Background:

• Electronic medical records (EMRs) are increasingly adopted in hospitals.
• Electronic search algorithms can aid in identifying postoperative complications.
• This can expedite data abstraction and clinical research for improved patient outcomes.

Purpose of the Study:

• To develop and validate an electronic search algorithm.
• The algorithm aims to identify specific postoperative complications: deep venous thrombosis, pulmonary embolism, and myocardial infarction.
• The focus is on complications occurring within 30 days of total hip or knee arthroplasty.

Main Methods:

• A large cohort of 34,517 patients undergoing hip or knee arthroplasty was identified.
• An electronic search algorithm was iteratively developed and refined using a derivation cohort (418 patients).
• The algorithm was validated on an independent cohort (2,857 patients), comparing its performance to manual chart review.

Main Results:

• The algorithm demonstrated high sensitivity and specificity for all targeted complications in both derivation and validation cohorts.
• For deep vein thrombosis, sensitivity ranged from 91%-97% and specificity from 85%-99%.
• For pulmonary embolism, sensitivity was 96%-97% and specificity 100%; for myocardial infarction, sensitivity was 100% and specificity 95%-99%.

Conclusions:

• An electronic search strategy was successfully derived and validated.
• This strategy significantly accelerates data abstraction for research and quality improvement.
• The electronic approach maintains superb reliability compared to manual chart review, enhancing patient care.