Supplementary Material

Appendix

The algorithms for use of the claims data to identify cases and outcomes in this paper were very similar to those used in the previous paper based on this data set¹. The original algorithms are contained in the appendix of Reference 1. Identification of primary and revision total hip replacements in this paper was identical to that in the previous paper¹, but we made minor modifications to the algorithms for identifying some outcomes. In the prior study, dislocations that occurred after discharge were counted, but those that occurred during the initial hospital admission were not (because of the possibility that dislocations might have been the indication for the revision total hip arthroplasty). In the present paper, we included dislocations occurring during the index hospitalization if they met our case definition (a dislocation that required an open or closed reduction). There were also minor changes to the algorithm for identifying deep infection. In our earlier work, patients were followed until three months postoperatively. In this paper, we followed the patients for six months. With longer follow-up, there is a higher likelihood of infection in a joint other than the index hip. Thus, in this paper, we required a more specific set of codes to ensure that the infection had occurred in the index hip. Specifically, we required that either the diagnosis or procedure codes mentioned the hip or pelvis. The more specific algorithm detected most infections detected by the less specific algorithm. In the primary total hip replacement cohort, it detected 95% of the infections within ninety days after surgery and 93% of those within 180 days. In the revision cohort, the algorithm detected 86% of infections within ninety days after surgery and 89% of infections within 180 days.