IDENTIFICATION OF UNDESIRABLE POST-OPERATIVE EVENTS USING ROUTINELY ACQUIRED HOSPITAL DATA: DEVELOPMENT OF AN AUDIT TOOL USING KNEE ARTHROSCOPY AS A MODEL

Purpose of Study: We aimed to reduce the work intensity involved in auditing high volume procedures (eg knee arthroscopy) by developing and validating a tool which uses routinely acquired hospital data, to target those patients most likely to have developed an undesirable post-operative outcome. Methodology: The work was a collaboration effort between the Orthopaedic and Clinical Effectiveness departments. During the period 1997–2003, 2926 elective knee arthroscopies were identified as having been performed in our unit. Linkage of routinely collected data held on the hospital’s computerized Patient Administrative System (PAS), hospital theatre system and A&E system, with data from the Office of National Statistics concerning death, high-lighted 183 cases (Core group) meeting one or more of four indicators: readmission Results: Accuracy of OPCS-4 coding for arthroscopic procedure performed was 77.1% in the core group and 96.4% in the random sample. The new tool yielded a sensitivity of 38% and specificity of 95%. Where major complications were concerned the sensitivity rose to 100%. For major complications the proposed model indicated a 0.6% complication rate vs 0.5% actual rate. For minor complications the proposed model indicated a 1.4% rate vs 3.8% actual rate. Overall complication rate within our unit was comparable to the published literature. Conclusion: The tool has achieved its aim of identifying all major complications and undesirable events, along with many minor complications. As the tool identifies additional information it must be used as an aid to identifying patients for case note review. However, in our study it reduced the number needed to less than 7% of the total.