Identification of prosthetic hip and knee joint infections using administrative databases—A validation study

Abstract

AbstractObjective:To determine whether combinations of diagnosis and procedures codes can improve the detection of prosthetic hip and knee joint infections from administrative databases.Design:We performed a validation study of all readmissions from January 1, 2010, until December 31, 2016, following primary arthroplasty comparing the diagnosis and procedure codes obtained from an administrative database based upon the International Classification of Disease, Tenth Revision (ICD-10) to the reference standard of chart review.Setting:Four tertiary-care hospitals in Toronto, Canada, from 2010 to 2016.Participants:Individuals who had a primary arthroplasty were identified using procedure codes.Intervention:Chart review of readmissions identified the presence of a prosthetic joint infection and, if present, the surgical procedure performed.Results:Overall, 27,802 primary arthroplasties were performed. Among 8,844 readmissions over a median follow-up of 669 days (interquartile range, 256–1,249 days), a PJI was responsible for or present in 586 of 8,844 (6.6%). Diagnosis codes alone exhibited a sensitivity of 0.88 (95% CI, 0.85–0.92) and positive predictive value (PPV) of 0.78 (95% CI, 0.74–0.82) for detecting a PJI. Combining a PJI diagnosis code with procedure codes for an arthroplasty and the insertion of a peripherally inserted central catheter improved detection: sensitivity was 0.92 (95% CI, 0.88–0.94) and PPV was 0.78 (95% CI, 0.74–0.82). However, procedure codes were unable to identify the specific surgical approach to PJI treatment.Conclusions:Compared to PJI diagnosis codes, combinations of diagnosis and procedure codes improve the detection of a PJI in administrative databases.