Artificial Intelligence for Automated Implant Identification in Total Hip Arthroplasty: A Multicenter External Validation Study Exceeding Two Million Plain Radiographs.

BACKGROUND: The surgical management of complications after total hip arthroplasty (THA) necessitates accurate identification of the femoral implant manufacturer and model. Automated image processing using deep learning has been previously developed and internally validated; however, external validation is necessary prior to responsible application of artificial intelligence (AI)-based technologies. METHODS: We trained, validated, and externally tested a deep learning system to classify femoral-sided THA implants as one of the 8 models from 2 manufacturers derived from 2,954 original, deidentified, retrospectively collected anteroposterior plain radiographs across 3 academic referral centers and 13 surgeons. From these radiographs, 2,117 were used for training, 249 for validation, and 588 for external testing. Augmentation was applied to the training set (n = 2,117,000) to increase model robustness. Performance was evaluated by area under the receiver operating characteristic curve, sensitivity, specificity, and accuracy. Implant identification processing speed was calculated. RESULTS: The training and testing sets were drawn from statistically different populations of implants (P < .001). After 1,000 training epochs by the deep learning system, the system discriminated 8 implant models with a mean area under the receiver operating characteristic curve of 0.991, accuracy of 97.9%, sensitivity of 88.6%, and specificity of 98.9% in the external testing dataset of 588 anteroposterior radiographs. The software classified implants at a mean speed of 0.02 seconds per image. CONCLUSION: An AI-based software demonstrated excellent internal and external validation. Although continued surveillance is necessary with implant library expansion, this software represents responsible and meaningful clinical application of AI with immediate potential to globally scale and assist in preoperative planning prior to revision THA.

Worsening Arthroplasty Utilization With Widening Racial Variance During the COVID-19 Pandemic.

BACKGROUND: Elective arthroplasty surgery in the United States came to a near-complete halt in the spring of 2019 as a response to the COVID-19 pandemic. Racial disparity has been a long-term concern in healthcare with increased focus during the pandemic. The purpose of this study is to evaluate the effects of COVID-19 and race on arthroplasty utilization trends during the pandemic. METHODS: We used 2019 and 2020 Center for Medicare and Medicaid Service fee-for-service claims data to compare arthroplasty volumes prior to and during the COVID-19 pandemic. We compared overall arthroplasty utilization rates between 2019 and 2020 and then sought to determine the effect of race and COVID-19, both independently and combined. RESULTS: There was a decrease in primary total knee arthroplasty (-28%), primary total hip arthroplasty (-14%), primary total hip arthroplasty for fracture (-2%), and revision arthroplasty (-14%) utilization between 2019 and 2020. The highest decrease in overall arthroplasty utilization was in the Hispanic population (34% decrease vs 19% decrease in the White population). We found that a non-White patient was 39.9% (P < .001) less likely to receive a total joint arthroplasty prior to COVID-19. The COVID-19 pandemic further exacerbated the pre-existing racial differences in arthroplasty utilization by decreasing the probability of receiving a total joint arthroplasty for non-White patient by another 12.9% (P < .001). CONCLUSION: We found an overall decreased utilization rate of arthroplasty during the COVID-19 pandemic with further decrease noted in all non-White populations. This raises significant concern for worsening racial disparity in arthroplasty caused by the ongoing pandemic.

Hospital Costs for Unsuccessful Two-Stage Revisions for Periprosthetic Joint Infection.

BACKGROUND: Although 2-stage exchange arthroplasty is the preferred surgical treatment for periprosthetic joint infection (PJI) in the United States, little is known about the risk of complications between stages, mortality, and the economic burden of unsuccessful 2-stage procedures. METHODS: The 2015-2019 Medicare 100% inpatient sample was used to identify 2-stage PJI revisions in total hip and knee arthroplasty patients using procedural codes. We used the Fine and Gray sub-distribution adaptation of the conventional Kaplan-Meier method to estimate the probability of completing the second stage of the 2-stage PJI infection treatment, accounting for death as a competing risk. Hospital costs were estimated from the hospital charges using "cost-to-charge" ratios from Centers for Medicare and Medicaid Services. RESULTS: A total of 5094 total hip arthroplasty and 13,062 total knee arthroplasty patients had an index revision for PJI during the study period. In the first 12 months following the first-stage explantation, the likelihood of completing a second-stage PJI revision was 43.1% (95% confidence interval [CI] 41.7-44.5) for hips and 47.9% (95% CI 47.0-48.8) for knees. Following explantation, 1-year patient survival rates for hip and knee patients were 87.4% (95% CI 85.8-88.9) and 91.4% (95% CI 90.6-92.2), respectively. The median additional cost for hospitalizations between stages was $23,582 and $20,965 per patient for hips and knees, respectively. Hospital volume, Northeast or Midwest region, and younger age were associated with reduced PJI costs (P < .05). CONCLUSION: Although viewed as the most preferred, the 2-stage revision strategy for PJI had less than a 50% chance of successful completion within the first year, and was associated with high mortality rates and substantial costs for treatment failure.

An Examination of the Adoption of Outpatient Total Knee Arthroplasty Since 2018.

BACKGROUND: The Centers for Medicare and Medicaid Services removed total knee arthroplasties (TKAs) from the inpatient-only (IPO) list on January 1, 2018, which meant that TKAs could be performed on a hospital outpatient basis. We examined the following: (1) the national rate of adoption of outpatient TKAs over time, (2) how adoption varied across hospitals, and (3) whether adoption of outpatient TKAs has positively or negatively impacted 90-day TKA readmission rates. METHODS: We used national patient-level Medicare Fee-for-Service Part A claims data (100% sample) from January 2017 through June 2019 to look at the quarterly trend in percent of TKAs performed as outpatient, and the distribution in this percentage across hospitals in the country. We ran a case-level regression to understand whether inpatient vs outpatient coding status relates to 90-day readmission rates. RESULTS: In 2017 prior to the removal of TKAs from the IPO list, 0.2% were performed as outpatient. In the first quarter (Q1 2018) after the rule change, 24.9% were performed as outpatient, and by the second quarter of 2019, 36.4% were performed as outpatient. These rates varied widely across hospitals from 0% (10th and 25th percentiles) to 78% (90th percentile) from January 2018 through March 2019. There was no difference in readmission rates for same-day discharges, but outpatient cases discharged after one or more nights in the hospital had statistically lower readmissions than inpatient cases. CONCLUSION: There was a rapid increase in the adoption of hospital outpatient TKAs following their removal from the Medicare IPO, which has resulted in lower readmission rates, and so adoption is likely to continue.