# Development and Validation of Case‐Finding Algorithms to Identify Periprosthetic Joint Infections After Total Hip Arthroplasty in Veterans Health Administration Data

## TL;DR

This study developed and validated algorithms to identify hip joint infections after surgery in veterans' health data, finding them accurate for future research.

## Contribution

The paper introduces and validates new case-finding algorithms using ICD-9 and ICD-10 codes combined with CPT codes for hip PJI in VHA data.

## Key findings

- ICD-9-based algorithm had a PPV of 87.8% for confirmed hip PJI.
- ICD-10-based algorithm had a PPV of 80.0% for confirmed hip PJI.
- Algorithms could be used for future pharmacoepidemiology studies in VHA data.

## Abstract

To determine the positive predictive values (PPVs) of ICD‐9‐ and ICD‐10‐based diagnostic coding algorithms to identify periprosthetic joint infection (PJI) following total hip arthroplasty (THA) within the United States (US) Veterans Health Administration (VHA).

We selected patients with: (1) any position hospital discharge ICD‐9 or ICD‐10 diagnosis of PJI, (2) ICD‐9, ICD‐10, or current procedural terminology (CPT) procedure codes for THA any time prior to PJI diagnosis, (3) CPT code for hip X‐ray within ±90 days of the PJI diagnosis, and (4) 1 or more CPT codes for arthrocentesis, arthrotomy, or revision arthroplasty all occurring within ±90 days of the PJI diagnosis date. We obtained separate samples of patients for ICD‐9 and ICD‐10‐based PJI diagnoses. These samples were stratified by THA medical center volume. Infectious disease physicians adjudicated each identified PJI event. The PPV (95% confidence interval [CI]) for the ICD‐9 and ICD‐10 PJI algorithms were calculated.

Among the 90 sampled hip PJI events for the ICD‐9 era, 79 were confirmed PJIs (PPV 87.8%, 95% CI 79.2%–93.7%). For the 90 sampled hip PJI events for the ICD‐10 era, 72 were confirmed PJIs (PPV 80.0%, 95% CI 70.3%–87.7%).

These algorithms yielded a PPV of 87.8% (ICD‐9) and 80.0% (ICD‐10), for confirmed PJI events and could be considered for use in future pharmacoepidemiologic studies.

Electronic healthcare databases may serve as valuable resources to study the pharmacoepidemiology of periprosthetic joint infection (PJI) following total hip arthroplasty (THA), but case finding algorithms to identify hip PJI have not been developed or validated within US healthcare databases.We developed and validated two case finding algorithms based on International Classification of Diseases, Ninth Revision (ICD‐9) and International Classification of Diseases, Tenth Revision (ICD‐10) diagnoses for hip PJI, in combination with Current Procedural Terminology (CPT) codes, to identify hip PJI within US Veterans Health Administration data.We identified possible hip PJI events based on the following criteria: (i) hospital discharge ICD‐9 or ICD‐10 diagnosis for PJI, (ii) ICD‐9, ICD‐10 or CPT procedure code for THA prior to PJI diagnosis, (iii) CPT code for hip X‐ray within ±90 days of the hip PJI diagnosis, and (iv) at least 1 CPT code for arthrocentesis, arthrotomy, or revision arthroplasty all occurring within ±90 days of the hip PJI diagnosis date.An algorithm consisting of an ICD‐9 or ICD‐10 code for PJI following a procedure code for THA, combined with CPT codes for a hip X‐ray and any one of arthrocentesis or arthrotomy within ±90 days of the PJI diagnosis date, had a positive predictive value (PPV) of 87.8% (ICD‐9) and 80.0% (ICD‐10), respectively, for confirmed hip PJI.These algorithms could be considered as fit for purpose for future pharmacoepidemiology studies of hip PJI within US Veterans Health Administration data.

Electronic healthcare databases may serve as valuable resources to study the pharmacoepidemiology of periprosthetic joint infection (PJI) following total hip arthroplasty (THA), but case finding algorithms to identify hip PJI have not been developed or validated within US healthcare databases.

We developed and validated two case finding algorithms based on International Classification of Diseases, Ninth Revision (ICD‐9) and International Classification of Diseases, Tenth Revision (ICD‐10) diagnoses for hip PJI, in combination with Current Procedural Terminology (CPT) codes, to identify hip PJI within US Veterans Health Administration data.

We identified possible hip PJI events based on the following criteria: (i) hospital discharge ICD‐9 or ICD‐10 diagnosis for PJI, (ii) ICD‐9, ICD‐10 or CPT procedure code for THA prior to PJI diagnosis, (iii) CPT code for hip X‐ray within ±90 days of the hip PJI diagnosis, and (iv) at least 1 CPT code for arthrocentesis, arthrotomy, or revision arthroplasty all occurring within ±90 days of the hip PJI diagnosis date.

An algorithm consisting of an ICD‐9 or ICD‐10 code for PJI following a procedure code for THA, combined with CPT codes for a hip X‐ray and any one of arthrocentesis or arthrotomy within ±90 days of the PJI diagnosis date, had a positive predictive value (PPV) of 87.8% (ICD‐9) and 80.0% (ICD‐10), respectively, for confirmed hip PJI.

These algorithms could be considered as fit for purpose for future pharmacoepidemiology studies of hip PJI within US Veterans Health Administration data.

Joint replacement surgery is very common and growing in the United States (US). PJI is a serious complication of joint replacement surgery. There are few large pharmacoepidemiological studies of hip PJI in the US. The Veterans Health Administration (VHA) is the largest integrated health system in the US and may be a valuable setting to study hip PJI. In order to study hip PJI within VHA data we developed four separate algorithms to identify patients who developed a hip PJI after undergoing hip replacement surgery. The algorithms used a combination of diagnosis codes and procedural codes to identify patients with a hip PJI. We then assessed the accuracy of each algorithm by having two independent physicians review 90 cases of hip PJI identified by the algorithm to confirm that a hip PJI truly occurred. We found that Algorithms 2A and 2B accurately identified hip PJI in greater than 80% of the cases.