Incidence, Microbiological Studies, and Factors Associated With Prosthetic Joint Infection After Total Knee Arthroplasty

This cohort study examines incidents rates, microbiological studies, and patient and surgical factors associated with prosthetic joint infection at 3 time points after total knee arthroplasty among US veterans.


Introduction
Total knee arthroplasty (TKA) is one of the most common elective surgeries in the US due to increasing rates of obesity and an aging population. 1,2Prosthetic joint infection (PJI), one of the most serious complications of TKA, can result in substantial morbidity, mortality, and health care costs. 3spite the clinical implications of PJIs, there are major knowledge gaps regarding the epidemiological features, microbiological studies, and factors of these infections in the US.
9][20][21] To our knowledge, no US study has compared the incidence, microbiological studies, and factors of PJI occurring in key periods after TKA: early PJI (Յ3 months), delayed PJI (between >3 and Յ12 months), and late PJI (>12 months).Prior literature has suggested that the pathogenesis of PJI differs by time points after TKA and that these specified periods may classify the different mechanisms and organisms associated with this infection. 22,23These data could identify subgroups at highest risk for PJI after primary TKA and could inform empirical antibiotic therapy for suspected PJI after this surgery.
To address these knowledge gaps, we evaluated a national cohort of veterans who underwent primary TKA within the US Department of Veterans Affairs (VA) Healthcare System.We aimed to identify the incidence rates, organisms isolated from microbiological cultures, and patient and surgical factors of PJI occurring early, delayed, and late after primary TKA.

Study Design and Data Source
We conducted a retrospective cohort study of patients who underwent primary TKA in the national VA system.The Human Investigations Committee at the VA Connecticut Health System, Yale University, and University of Pennsylvania approved the study and waived the informed consent requirement because this research could not practicably be conducted without this waiver.We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline. 24 collected electronic health record (EHR) data from the VA Corporate Data Warehouse.The data set included demographic characteristics, hospital and ambulatory diagnoses (recorded using To obtain operative variables, we linked patients' records to the Veterans Affairs Surgical Quality Improvement Program (VASQIP), which included surgical data entered by nurses who reviewed records using standardized definitions. 25In general, surgical data at each VA center are identified on

Study Patients
Patients were eligible for inclusion in the study if they (1) underwent elective primary TKA (eTable 1 in Supplement 1) between October 1, 1999, and September 30, 2019, and (2) had at least 1 year of care in the VA prior to TKA.Primary TKAs can be accurately identified from the VA data set with a positive predictive value (PPV) of 95.2% (95% CI, 92%-99%). 27Patients were excluded if they underwent partial or revision TKA, underwent TKA for nonelective reasons (malignant neoplasm or fracture), or had a history of prior PJI or native septic joint arthritis of the knee (eTable 1 in Supplement 1).
We defined the index date as the date of hospital admission for primary TKA.The 12 months prior to admission represented the baseline period.Follow-up continued until the occurrence of one of the following: hospitalization with PJI, death, aseptic revision arthroplasty of the knee or contralateral TKA (since these procedures could affect PJI risk), or last VA visit before October 1, 2020 (to ensure patients had the opportunity for at least 1 year of follow-up after TKA).

Main Study Outcome
The primary end point was incident hospitalization with PJI.We classified patients as being hospitalized with PJI after primary TKA if they had (1) principal or contributory PJI ICD-9 or ICD-10 diagnosis codes at hospital discharge; (2) knee radiography within 90 days of PJI diagnosis; and (3) microbiological culture, arthrocentesis, or arthrotomy of the knee within 90 days of PJI diagnosis (eFigure 1 in Supplement 1).This algorithm accurately identified PJI with a PPV of 75.0%(95% CI, 64.1%-84.0%) in the ICD-9 era and 85.0% (95% CI, 75.3%-92.0%)in the ICD-10 era. 28e PJI date was defined as the date of hospital discharge with PJI diagnosis.We defined early PJI as occurring 3 months or less after primary TKA, delayed PJI as occurring between more than 3 months and 12 months or less after TKA, and late PJI as occurring more than 12 months after TKA.
Patients stopped contributing to follow-up at their initial PJI. 29

Data Collection
At the time of TKA, we collected data on age, sex, self-reported race and ethnicity, year of surgery, urban vs rural VA center, 30 body mass index (BMI; calculated as weight in kilograms divided by height in meters squared), and tobacco use (current, former, or never smokers). 31Race and ethnicity data were collected as variables to contribute data on the representativeness of the study sample.
Validated algorithms based on 1 or more inpatient or 2 or more outpatient ICD-9 or ICD-10 diagnosis codes were used to identify baseline comorbidities, including alcohol use disorder (AUD), 32 diabetes, 33 heart failure, 34,35 hypertension, 36 hepatitis B virus infection, hepatitis C virus infection, 37 HIV infection, 38 and peripheral artery disease (PAD) 39,40 (eTable 2 in Supplement 1).Autoimmune inflammatory arthritis was defined by 1 inpatient or 2 outpatient ICD-9 or ICD-10 diagnosis codes and required at least 1 rheumatologic clinic visit. 41

Statistical Analysis
Among the overall cohort, we estimated incidence rates (events per 10 000 person-months) with 95% CIs of early, delayed, and late PJI.We used unadjusted Poisson regression to estimate the incidence rate ratio (IRR) with 95% CI of early and delayed PJI compared with late PJI.The standard Kaplan-Meier method with noninformative censoring was used to demonstrate the cumulative incidence of PJI at 3, 12, and 24 months.We measured the frequency of gram-positive, gramnegative, fungal, polymicrobial, and culture-negative PJI within each postoperative period and compared the frequencies of organisms using χ 2 tests.
For Variables were retained in multivariable models if they were associated with PJI (statistical significance: P < .10) in univariable analysis and had statistical significance of P < .05 in multivariable analysis.We retained BMI in all models given its clinical importance 44 and conflicting association with PJI in prior studies, 12 regardless of statistical significance.We used the svy module in Stata to account for clustering by VA center and also adjusted for year of TKA (categorized in 5-year periods from 1999 to 2019). 45Data were analyzed between December 9, 2021, and September 18, 2023, using Stata, version 16.1 (StataCorp LLC).

Results
A total of 83 973 patients underwent elective primary TKA during the study period, of whom 4606 were excluded, leaving 79 367 patients in the overall cohort (Figure 1).

Risk Factors for Early, Delayed, and Late PJI
Among the 65 188 patients in the VASQIP cohort, 61 701 (94.6%) had complete data on risk factors and so were included in the complete case analysis.Hepatitis C virus infection, PAD, and autoimmune inflammatory arthritis were associated with PJI across all postoperative periods (Table 3).Early PJI was also associated with current smoking, heart failure, hypertension, urban location, and prolonged operative time (Ն2 hours).Autoimmune inflammatory arthritis was a robust factor of early PJI (adjusted IRR, 2.3; 95% CI, 1.5-3.5).Delayed PJI was also associated with BMI of 40 or higher, AUD, general anesthesia, anemia, and prolonged operative time (Ն2 hours).Body mass index of 40 or higher was a robust factor of delayed PJI (adjusted IRR, 2.7; 95% CI, 1.5-4.9).Late PJI was also associated with AUD, heart failure, anemia, and younger age (<70 years).Anemia was a factor of late PJI (adjusted IRR, 2.2; 95% CI, 1.5-3.1).Factors asociated with PJI at any time after TKA are reported in eTable 4 in Supplement 1. Diabetes and chronic kidney disease were not associated factors.

Discussion
Using EHR data from the national VA system, we found that PJI was uncommon after TKA (2.0%), and the incidence rate was highest in the early and delayed periods (20.7 and 4.2 times higher, respectively) compared with the late period.Staphylococcus and Enterococcus species were more commonly isolated from early PJIs, whereas Streptococcus species were more commonly isolated from delayed or late PJIs.Gram-negative infections were more common in the early postoperative period.Hepatitis C infection, PAD, and autoimmune inflammatory arthritis were associated with PJI across all postoperative periods.
This study demonstrated that the rate of PJI was highest within the initial 3 months after primary TKA, and the cumulative incidence at 24 months was 1.60%.These results are consistent with the findings of a cohort study of 69 663 US Medicare beneficiaries in which the cumulative incidence of PJI diagnoses was 1.55% within the first 24 months after elective TKA. 7Since the incidence of PJI was highest soon after primary TKA, development and testing of interventions to modify risk factors for early PJI should be prioritized.
Few studies have examined the microbiological characteristics of PJI after TKA.In a crosssectional study of 1651 patients who had a primary or revision TKA or total hip arthroplasty at Mayo Clinic (2010-2019) and who were observed for PJI, coagulase-negative staphylococci were the most   46 In a separate cohort study of 231 patients with hip or knee PJI from a single hospital in China (2006-2015), coagulase-negative staphylococci were also the most common organisms isolated. 47The finding of the present study that Staphylococcus aureus was the most frequently identified causative organism may be due to inclusion of only primary TKAs, whereas previous studies included PJIs of the knee and hip as well as primary and revision arthroplasties.
Gram-negative organisms were isolated from 11.1% of all PJIs and 15.4% of early PJIs.The gramnegative organisms identified from the PJIs in this study, including the early infections, are typically resistant to first-generation cephalosporins.Given these findings, empirical gram-negative antibiotic therapy, in addition to the usual gram-positive coverage, should be considered during suspected early PJI.Cefazolin, the recommended antibiotic for perioperative TKA prophylaxis, would not provide adequate protection against most of the gram-negative organisms identified in this study. 48 date, the only intervention associated with a decreased rate of PJI is perioperative antibiotic prophylaxis. 49,50Future studies should examine the antimicrobial susceptibility profiles of gramnegative infections that were isolated during early PJI to determine the extent to which broader perioperative gram-negative antibiotic prophylaxis is warranted.
We explored factors of early, delayed, and late PJI.Factors associated with inflammation and tissue hypoxia were associated with PJI in each postoperative period.2][53] Peripheral artery disease was also a significant factor in tissue hypoxia and compromised skin or soft tissue integrity 54 across the 3 postoperative periods.0][61][62] Prognostic models should be developed to classify patients according to risk of PJI, which could help guide informed consent, surgical planning, and postoperative monitoring.
We observed a reduced rate of late infections among patients 70 years or older compared with those younger than 60 years, which is consistent with findings of large European and Canadian studies and may reflect a tendency to avoid invasive diagnostic testing and workup for painful joints in patients 70 years or older. 17,64Chronic kidney disease and diabetes were not associated with PJI during any period after TKA.The lack of association between CKD and PJI has been reported in a meta-analysis. 6517]66 The lack of a similar association in the present study might be due to the definition of diabetes used.No pattern was observed between BMI and PJI risk, making the clinical significance of the association between BMI of 40 or higher and risk of PJI in the delayed postoperative period unclear.The role that obesity plays in postoperative infection risk has been inconsistent, with some studies reporting no association 2,14,21,67 and other studies observing an increased risk. 12,17,68Results of the present study suggest that patients with obesity or CKD should not be restricted from undergoing TKA because of concerns about increased PJI risk.

Strengths and Limitations
Strengths of this study included the large sample of veterans undergoing primary TKAs, access to national EHR data that included microbiological culture results, and linkage to surgical-specific data.
This study also had several potential limitations.First, misclassification of outcomes was possible, but use of a validated PJI end point minimized this risk.It is possible that some PJI outcomes were missed in patients who presented to a non-VA hospital.Second, some culture results could have included organisms that were contaminants.Third, residual confounding by unmeasured factors, such as perioperative antibiotic prophylaxis and chronic skin conditions, could have affected risk factor analyses.Moreover, we evaluated comorbidities as binary (present or absent), without accounting for illness severity, and accounted only for baseline comorbidities or surgical characteristics without evaluating time-varying factors.Fourth, the sample consisted of mostly males with a high prevalence of comorbid conditions.Results may have less generalizability to females and the overall population receiving TKAs.

JAMA Network Open | Infectious Diseases
Incidence of Prosthetic Joint Infection After Total Knee Arthroplasty [ICD-9] and International Statistical Classification of Diseases and Related Health Problems, Tenth Revision [ICD-10] diagnosis codes), procedures (recorded using Current Procedural Terminology codes), laboratory results, microbiological culture results, and dispensed medications.Death date was ascertained from the VA Vital Status File.

Figure 2 .
Figure 2. Kaplan-Meier Failure Curves for Prosthetic Joint Infection (PJI) for the Overall Study Period and the First 24 Months After Primary Total Knee Arthroplasty 26cidence of Prosthetic Joint Infection After Total Knee Arthroplasty cycle to ensure that data collection periods begin on different days of the week and are collected for up to 36 surgeries per cycle; thus, not all TKAs are entered into VASQIP.However, exclusion is random and is based on case volume at each center.26 JAMA Network Open.2023;6(10):e2340457. doi:10.1001/jamanetworkopen.2023.40457(Reprinted) October 31, 2023 2/15 Downloaded From: https://jamanetwork.com/ on 11/04/2023 an 8-day
the subgroup of patients with linked VASQIP data (referred to as the VASQIP cohort), we explored demographic, baseline clinical, and perioperative factors of early, delayed, and late PJI.The factors we evaluated included age, sex, urban vs rural center, BMI, AUD, hepatitis B virus infection, hepatitis C virus infection, heart failure, HIV infection, hypertension, PAD, autoimmune inflammatory arthritis, tobacco use, anemia (defined as hemoglobin <12.0 g/dL; to convert to milligrams per 43ciliter, multiply by 10.0), diabetes, chronic kidney disease (CKD; defined by estimated glomerular filtration rate <60 mL/min/1.73m 2 , calculated using the Modified Diet in Renal Disease equation43), anesthesia technique, ASA score, corticosteroid use within 30 days prior to surgery, intraoperative transfusion of packed red blood cells, and prolonged operative time (Ն2 hours).The degree of missing data for variables in the VASQIP cohort was low, with the highest being 2.9% for BMI data.As a result, we implemented a complete case analysis to identify factors for PJI in the VASQIP cohort.We used a piecewise exponential parametric survival model fit through Poisson regression to estimate adjusted IRRs (with 95% CIs) of early, delayed, and late PJI associated with risk factors.

Table 1 .
Demographic, Baseline Clinical, and Perioperative Characteristics of the Overall Cohort (n = 79 367)

Table 1 .
Demographic, Baseline Clinical, and Perioperative Characteristics of the Overall Cohort (n = 79 367) (continued) Abbreviations: ASA, American Society of Anesthesiologists Physical Status Classification System (score range: 1 [indicating a healthy patient] to 4 [indicating a patient with incapacitating severe disease that is a constant threat to life]); AUD, alcohol use disorder; BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; HBV, hepatitis B virus; HCV, hepatitis C virus; ICD-9, International Classification of Diseases, Ninth Revision; ICD-10, International Statistical Classification of Diseases and Related Health Problems, Tenth Revision; NA, not applicable; PAD, peripheral artery disease; RBC, red blood cell; TKA, total knee arthroplasty; VASQIP, Veterans Affairs Surgical Quality Improvement Project.aRace and ethnicity were self-reported.bIncludedAsian, American Indian or Alaska native, Native Hawaiian or Other Pacific Islander, and multiracial.cComorbidities of AUD, heart failure, hypertension, diabetes, PAD, HIV infection, HBV infection, and HCV infection were defined by 1 inpatient or 2 outpatient ICD-9 or ICD-10 diagnosis codes.Autoimmune inflammatory arthritis was defined as 1 inpatient or 2 outpatient ICD-9 or ICD-10 diagnosis codes among those with at least 1 outpatient rheumatologic clinic visit during the baseline period.The ICD-9 or ICD-10 diagnosis codes used to identify comorbidities are provided in eTable 2 in Supplement 1. Anemia was defined as hemoglobin level less than 12 g/dL (to convert to grams per liter, multiply by 10.0), and CKD was defined as an eGFR less than 60 mL/min/1.73m 2 ; eGFR was calculated using the Modified Diet in Renal Disease equation: 175 × (serum creatinine) -1.154 × (age) -0.203 × (0.742, if female) × (1.212, if Black patient).dPerioperative characteristics were available for only a subset of patients in the VASQIP cohort (n = 65 188).Percentages reflect this subset of patients.eIncluded epidural, spinal, or monitored anesthesia.

Table 2 .
Organisms Isolated From Synovial Fluid or Operative Tissue Cultures of Prosthetic Joint Infections (PJIs) After Primary Total Knee Arthroplasty c Included Micrococcus species and diphtheroids.d Included Salmonella, Pasteurella, Citrobacter, e All fungal infections were identified as Downloaded From: https://jamanetwork.com/ on 11/04/2023 common organisms isolated from cultures.

Table 3 .
Adjusted Incidence Rate Ratios (IRRs) of Prosthetic Joint Infections (PJIs) After Primary Total Knee Arthroplasty (TKA) in the Veterans Affairs Surgical Quality Improvement Project Complete Case Cohort (n = 61 701) a The IRR for each variable was adjusted for all other variables as well as by VA center and year of TKA (categorized in 5-year periods from 1999 to 2019).