Reoperations may be a better way of tracking adverse outcomes than complications. Repeat surgery causes cost to the system, and often indicate failure of the primary procedure resulting in the patient not achieving the expected improvement in pain and function. Understanding the cause of repeat surgery at the primary site may result in design improvements to implants or improvements to fusion techniques resulting in better outcomes in the future. The COFAS group have designed a reoperation classification system. The purpose of this study was to outline the inter and intra observer reliability of this classification scheme.

To verify the inter- and intra-observer reliability of this new coding system, six fellow ship trained practicing foot and ankle Orthopaedic surgeons were asked to classify 62 repeat surgeries from a single surgeons practice. The six surgeons read the operation reports in random order, and reread the reports 2 weeks later in a different order. Reliability was determined using intraclass correlation coefficients (ICC) and proportions of agreement. The agreement between pairs of readings (915 for inter observer for the first and second read – 61 readings with 15 comparisons, observer 1 with observer 2, observer 1 with observer 3, etc) was determined by seeing how often each observer agreed. This was repeated for the 366 ratings for intra observer readings (61 times 6).

The inter-observer reliability on the first read had a mean intra-class correlation coefficient (ICC) of 0.89. The range for the 15 comparisons was 0.81 to 1.0. Amongst all 1830 paired codings between two observers, 1605 (88%) were in agreement. Across the 61 cases, 45 (74%) were given the same code by all six observers. However, the difference when present was larger with more observers not agreeing. The inter-observer reliability test on the second read had a mean ICC of 0.94, with a range of 0.90. There were 43 (72%) observations that were the same across all six observers. Of all pairs (915 in total) there was agreement in 804 pairs for the first reading (88%) and disagreement in 111 (12%). For the second reading there was agreement in 801 pairs (86%) and disagreement in 114 (14%). The intra-observer reliability averaged an ICC value of 0.92, with a range of 0.86 to 0.98. The observers agreed with their own previous observations 324 times out of 366 paired readings (89% agreement of pairs).

The COFAS classification of reoperations for end stage ankle arthritis was reliable. This scheme potentially could be applied to other areas of Orthopaedic surgery and should replace the Claiden Dindo modifications that do not accurately reflect Orthopaedic outcomes. As complications are hard to define and lack consistent terminology reoperations and resource utilisation (extra clinic visits, extra days in hospital and extra hours of surgery) may be more reliable measures of the negative effects of surgery.