Abstract
Objective
The optimal timing of TNF antagonists before elective surgery is unknown. This study evaluated the association between infliximab timing and serious infection after elective hip or knee arthroplasty.
Methods
A retrospective cohort study evaluated U.S. Medicare patients with rheumatoid arthritis, inflammatory bowel disease, psoriasis, psoriatic arthritis, or ankylosing spondylitis who received infliximab within 6 months of elective knee or hip arthroplasty from 2007–2013. Propensity adjusted analyses examined whether infliximab stop timing (time between most recent infusion and surgery) was associated with hospitalized infection within 30 days or prosthetic joint infection (PJI) within 1 year.
Results
Hospitalized infection within 30 days occurred after 270/4288 (6.3%) surgeries. Infliximab stop timing < 4 weeks vs 8–12 weeks was not associated with an increase in infection within 30 days: propensity adjusted OR 0.90 (95% CI 0.60, 1.34). The rate of PJI was 2.9 per 100 person-years and was not increased in patients with stop timing < 4 weeks vs 8–12 weeks: HR 0.98 (95% CI 0.52, 1.87). Glucocorticoid dose > 10mg/day was associated with increased risk of 30-day infection (OR 2.11, 95% CI 1.30, 3.40) and PJI (HR 2.70, 95% CI 1.30, 5.60). Other risk factors for infection included elderly age, comorbidities, revision surgery, and previous hospitalized infection.
Conclusion
Administering infliximab within 4 weeks of elective knee or hip arthroplasty was not associated with a higher risk of short or long-term serious infection compared to withholding infliximab for longer time periods. Glucocorticoid use, especially > 10mg/day, was associated with an increased infection risk.
INTRODUCTION
Biologic disease modifying anti-rheumatic drugs (DMARDs), of which TNF antagonists such as infliximab are the most commonly used, are increasingly used in the treatment of patients with rheumatoid arthritis (RA) and other autoimmune diseases. Clinicians are frequently faced with the question of how to manage these medications in patients undergoing elective surgery. The association of TNF antagonists with serious infections has led to concerns that these medications could lead to increased post-operative infections (1–5). Few studies, however, have compared continuation versus interruption of TNF antagonist therapy. Existing studies are limited by the small number of patients, rare outcomes, and, in some cases, inability to accurately determine when TNF antagonists were given in the perioperative period (6–12). Furthermore, many studies compare TNF antagonist users to non-users (with greater concern for confounding) instead of comparing the timing of treatment among TNF antagonist users.
Despite or perhaps because of limited data, guidelines from many professional societies recommend holding therapy prior to surgery. Some suggest timing be individualized based on medication half-life, surgery type, or patient characteristics (13, 14), while others recommend holding therapy for 3–5 half-lives (4–7 weeks for infliximab) (15). Yet stopping therapy is not without risk, as disease flares may impair rehabilitation, adversely affect functional outcomes, or lead to excess glucocorticoid exposure that could potentially increase infection risk (16).
This study evaluated a large cohort of older patients undergoing elective knee or hip arthroplasty treated preoperatively with infliximab. Infliximab specifically was evaluated to allow accurate assessment of the timing of therapy before surgery; as an infusion, the precise administration date of infliximab can be identified accurately in claims data. The objective was to evaluate whether withholding infliximab perioperatively (consistent with current recommendations) was associated with a lower risk of post-operative infection.
METHODS
This retrospective cohort study evaluated patients with autoimmune diseases undergoing hip or knee arthroplasty using U.S. Medicare claims data from 2006–2013. Medicare is a public health plan covering more than 90% of US adults age 65 or older (17). Younger individuals with certain disabilities (e.g. RA) may also be covered. Coverage includes in-hospital care (Part A), physician and outpatient services (Part B), and, since 2006, outpatient pharmaceuticals (Part D). Claims for infusions of biologics are covered as medical procedures under Part B and are therefore identifiable with a specific date of administration.
Cohort identification
Included patients were ≥ 18 years old with rheumatoid arthritis (RA), inflammatory bowel disease (IBD), psoriasis (PsO), psoriatic arthritis (PsA), or ankylosing spondylitis (AS) based on 2 physician office or inpatient ICD-9 (International Classification of Diseases, 9th edition) codes at least 7 days apart who underwent elective primary or revision hip or knee arthroplasty from 2007–2013 and had received at least one infliximab infusion within 6 months before surgery. These diagnosis codes have been shown in previous population-based studies to have high validity, particularly when used in combination with DMARD or biologic use, with positive predictive values (PPVs) that exceed 80% (18–22). Included patients must have received at least 3 infusions in the year prior to surgery (to identify long term, stable therapy) and been enrolled in Medicare Part A, B, and D without enrollment in a Medicare Advantage Plan for at least 1 year prior to surgery (“baseline”) to allow uniform assessment of covariates including prescription medications.
Total hip and knee arthroplasties were identified by validated algorithms based on ICD-9 procedure codes or CPT (Current Procedural Terminology) codes during an inpatient admission for primary surgeries and using CPT codes during an inpatient admission for revision surgeries (Supplemental Table 6) (23–26). Patients with CPT or ICD-9 codes for multiple different surgery types on the same day were excluded.
Exclusion criteria were designed to exclude pre-existing infections and non-elective surgery and included joint infection or procedure to treat infection in the year prior to surgery, fracture, metastatic or bone cancer, admission through the emergency department, transfer from another acute hospital, surgery after hospital day 3, or previous hip or knee arthroplasty within 6 months of surgery (additional details in Supplemental Table 6). Patients could contribute multiple surgeries if more than 6 months apart, and standard errors were adjusted to account for this clustering (see below).
Timing of infliximab infusions
The main exposure of interest was the pre-operative “infliximab stop timing” – defined as the time between the most recent infliximab infusion and surgery. Although previous studies have treated TNF stopping vs. not stopping as a dichotomous exposure based on an arbitrary (and variable) stopping definition, in this study the primary analysis evaluated stop timing as a more general categorical exposure using 4-week intervals (half the standard RA dosing interval) to allow better assessment of the optimal stop timing. Stop timing of 8–12 weeks (at least one full dosing interval) was chosen as the reference category, as there were relatively few patients in longer stop timing categories (e.g. >12 weeks).
Outcomes
The two pre-specified primary outcomes were serious infection within 30 days and prosthetic joint infection (PJI) within 1 year after surgery. Outcome ascertainment began post-operatively for all patients, with no patients contributing exposed time prior to surgery. Serious infections were identified using a previously validated set of ICD-9 diagnosis codes from any position of the discharge diagnoses (PPV > 80%) from any hospitalization with an admission date within 30 days of surgery, including the index hospitalization (Supplemental Table 3) (1, 27). In the primary analysis, urinary tract infections were not included to avoid identification of minor infections diagnosed during the index hospitalization. PJI within 1 year after surgery was defined as diagnosis code 996.66 from any part of the record after surgery, excluding discharge diagnoses from the index hospitalization because this situation might indicate that revision surgery was performed because of a pre-existing infection (28, 29).
Secondary outcomes included prolonged length of stay as a proxy for post-operative complications (defined as greater than 5 days, the 90th percentile of length of stay) (30), and surgical site infection within 90 days of surgery based on the presence of any of four previously used diagnosis codes from inpatient discharge diagnoses (Supplemental Table 6) (31).
Covariates
Covariates measured during the baseline period included baseline demographics, comorbid conditions, an adaption of the Charlson comorbidity index (32), skilled nursing facility stay, number of outpatient visits, prior hospitalization, prior hospitalized infection, antibiotic use, dual Medicare/Medicaid eligibility (as a proxy for low income), quintiles of median household income based on zip code from the American Community Survey 2009–2013 (33), non-biologic DMARD use, number of prior biologic therapies (using all available data from the 12 month baseline and earlier, if available), and infliximab dose and dosing interval (details in Supplemental Table 6). Average glucocorticoid dose in the month prior to surgery was calculated based on oral prescriptions for prednisone, prednisolone, and methylprednisolone, using prescribed dose in prednisone equivalents and days supply to determine each daily dose and truncating prescriptions if a new prescription was filled before the prescription end date. Concomitant non-biologic DMARD use was based on filled prescriptions in the 3 months prior to surgery. Surgeon and hospital volume were estimated among a larger cohort of 13,631 THA or TKA procedures among patients with any biologic infusion within 6 months.
Statistical analysis
The association between infliximab stop timing and hospitalized infection within 30 days was evaluated using logistic regression. Because inclusion of all measured covariates might result in overfitting of traditional multivariable models, in the primary analysis propensity scores based on the likelihood of being in each exposure category were used to balance measured covariates across exposure categories using inverse probability weights (details of propensity scores below). In a secondary analysis, multivariable logistic regression was used to identify other risk factors for 30-day infection using generalized estimating equations with an unstructured correlation structure to account for multiple surgeries within subjects. These models were built using step-wise backward selection of covariates with p > 0.2 by likelihood ratio testing that did not change the coefficient for the exposure of interest by > 15%, forcing in age, sex, glucocorticoid dose, inflammatory disease type, and non-biologic DMARD use.
Differences in time to prosthetic joint infection within 1 year after surgery by infliximab stop timing were evaluated with Kaplan-Meier curves and log-rank tests. Cox models, again using inverse probability weights to balance covariates across exposure categories, evaluated the association between infliximab stop timing and time to PJI with clustering by subject and right censoring for death, end of follow-up, or subsequent hip or knee arthroplasty. In a secondary analysis, individual contributions of other risk factors for PJI were assessed in a multivariable Cox model using stepwise backward selection of covariates with p > 0.1 to avoid over-fitting the model, forcing in age, sex, and glucocorticoid dose. Timing for restarting infliximab post-operatively was explored by including infliximab restart timing as a time-varying covariate. Proportional hazards were assessed using log-log plots.
Propensity Scores
Propensity scores were calculated by predicting the probability of being in each infliximab stop timing category using a multinomial logistic regression model including all covariates of interest measured during baseline (Table 1 and Supplemental Table 1) with a squared term for age (34, 35). Inverse probability treatment weights (the reciprocal of the subject’s probability of being in the observed treatment category) were stabilized by multiplying by the proportion of patients in that treatment category (35–37). Comparability was assessed by evaluating the overlap of the predicted probabilities of being in a given treatment category across all treatment category groups, with excellent overlap observed (Supplemental Figure 1). The balance of covariates across treatment categories before and after applying inverse probability weights was assessed using F-tests or Chi2 tests from linear or logistic regression (Table 1 and Supplemental Table 1) (34, 36). No extreme weights > 10 were observed.
Table 1.
Select patient characteristics by infliximab stop timing prior to surgery
| Total n = 4288 |
Stop timing < 4 weeks n = 1162 |
Stop timing 4–8 weeks n = 1969 |
Stop timing 8 12 weeks n = 747 |
Stop timing 12 16 weeks n = 218 |
Stop timing ≥ 16 weeks n = 192 |
p value prior to weighting |
p value after inverse probability weighting |
|
|---|---|---|---|---|---|---|---|---|
| Age, yearsa | 71 [66,75] | 70 [66,75] | 70 [66,75] | 71 [67, 76] | 71 [66, 76] | 70 [66, 75] | 0.10 | 0.96 |
| Age ≥ 65b | 3382 (79) | 909 (78) | 1548 (79) | 611 (82) | 166 (76) | 148 (77) | 0.24 | 0.67 |
| Female | 3385 (79) | 924 (80) | 1547 (79) | 591 (79) | 173 (79) | 150 (78) | 0.97 | 0.75 |
| Black | 298 (7) | 80 (7) | 127 (6) | 56 (8) | 16 (7) | 19 (10) | 0.45 | 0.83 |
| RA | 3251 (76) | 872 (75) | 1475 (75) | 587 (79) | 173 (79) | 144 (75) | 0.21 | 0.98 |
| IBD | 475 (11) | 159 (14) | 222 (11) | 57 (8) | 19 (9) | 18 (9) | <0.001 | 0.99 |
| PsA/AS/PsO | 562 (13) | 129 (11) | 272 (14) | 103 (14) | 26 (12) | 30 (16) | 0.21 | 0.85 |
| Extra-articular RA | 306 (7) | 86 (7) | 144 (7) | 56 (8) | 12 (6) | 8 (4) | 0.45 | 0.98 |
| Diabetes | 890 (21) | 245 (21) | 408 (21) | 155 (21) | 46 (21) | 36 (19) | 0.97 | 0.96 |
| COPD/asthma | 920 (21) | 268 (23) | 381 (19) | 160 (21) | 53 (24) | 58 (30) | <0.01 | 0.90 |
| Kidney disease | 295 (7) | 79 (7) | 129 (7) | 60 (8) | 15 (7) | 12 (6) | 0.74 | 0.93 |
| Obesity | 333 (8) | 80 (7) | 146 (7) | 66 (9) | 23 (11) | 18 (9) | 0.21 | 0.97 |
| Charlson Score | 3 [1,5] | 3 [1,5] | 2 [1,5] | 3 [1,5] | 3 [1,5] | 3 [2,5] | 0.04 | 0.99 |
| Non-biologic DMARD past 3 months |
2632 (61) | 731 (63) | 1220 (62) | 456 (61) | 133 (61) | 91 (47) | <0.01 | 0.99 |
| Methotrexate past 3 monthsb |
2063 (48) | 555 (48) | 974 (49) | 357 (48) | 104 (48) | 73 (38) | 0.05 | 0.97 |
| Previous biologic DMARDc |
565 (13) | 144 (12) | 244 (12) | 122 (16) | 32 (15) | 23 (12) | 0.07 | 1.00 |
| Glucocorticoid dose past month |
||||||||
| none | 3033 (71) | 819 (70) | 1395 (71) | 527 (71) | 158 (72) | 134 (70) | 0.97 | 1.00 |
| ≤ 5mg/day | 561 (13) | 155 (13) | 252 (13) | 100 (13) | 29 (13) | 25 (13) | 0.99 | 0.99 |
| 5 to ≤10mg/day | 500 (12) | 144 (12) | 232 (12) | 88 (12) | 19 (9) | 21 (11) | 0.71 | 0.94 |
| > 10mg/day | 194 (5) | 48 (4) | 90 (5) | 32 (4) | 12 (6) | 12 (6) | 0.68 | 0.79 |
| Surgery type | ||||||||
| Primary knee | 2848 (66) | 785 (68) | 1308 (66) | 491 (66) | 132 (61) | 132 (69) | 0.33 | 0.83 |
| Primary hip | 1032 (24) | 273 (23) | 472 (24) | 179 (24) | 61 (28) | 47 (24) | 0.72 | 0.89 |
| Revision knee | 232 (5) | 56 (5) | 107 (5) | 43 (6) | 16 (7) | 10 (5) | 0.64 | 0.98 |
| Revision hip | 176 (4) | 48 (4) | 82 (4) | 34 (5) | 9 (4) | 3 (2) | 0.51 | 0.94 |
n (%) and median [IQR] calculated before weighting; p-values based on F-test or Chi2 test for differences across infliximab stop timing groups before and after applying inverse probability weights.
Included in propensity score as continuous variable and with squared term,
not included in the propensity score,
biologic use at any time prior to surgery (all available data)
RA: rheumatoid arthritis, IBD: inflammatory bowel disease, PsA: psoriatic arthritis, PsO: psoriasis, AS: ankylosing spondylitis, COPD: chronic obstructive pulmonary disease
Sensitivity analyses
Three alternative definitions of serious infection were evaluated. These included urinary infections and/or excluded infections from the index hospitalization or acute rehabilitation facilities (excluding diagnoses from already hospitalized patients). Three alternative definitions of PJI required inpatient diagnosis codes and/or required accompanying procedure codes for arthrotomy, prosthesis removal, central catheter insertion, spacer insertion, or revision surgery similar to previously used definitions (Supplemental Table 3) (24). Additional sensitivity analyses included truncating inverse probability weights at the 99th percentile (38), analyses limited to primary arthroplasties, limited to patients with RA, stratified by calendar year (2007–2009 and 2010–2013), or stratified by glucocorticoid use (Supplemental Tables 4 and Supplemental Tables 5). Infliximab stop timing was also evaluated with a < 2-week stop timing category and as a continuous variable.
Pre-analysis power calculations
Based on the observed sample size of 4288 patients in the infliximab group with 1162 receiving infliximab within 4 weeks of surgery and a baseline risk of 30 day infection of 6%, there was 86% power to detect an OR of 1.5 for infection in the < 4 week exposure group based on Chi2 testing, corresponding to an absolute increase in the 30-day risk of serious infection of 2.7%. In time to event analysis, based on the observed sample size and number of events and assuming a baseline prosthetic joint infection-free survival of 98% at 2 years, there was 76% power to detect a HR of 1.8, corresponding to an increase in the one-year risk of prosthetic joint infection from 2% to 3.6% in those with infliximab within 4 weeks of surgery.
The dataset was created with SAS 9.4 and analysis was performed using STATA 13.1 (StataCorp, LP, College Station, TX). The protocol was approved by the University of Alabama institutional review board. Use of the data was governed by a Data Use Agreement from CMS.
RESULTS
Among 8795 arthroplasties occurring from 2007–2013 in patients with RA, IBD, PsA, AS, or PsO treated with infliximab, 4288 (48.8%) among 3867 patients met all inclusion and exclusion criteria (Figure 1). Patient characteristics were largely similar by infliximab stop timing even before propensity adjustment (Table 1 and Supplemental Table 1). After application of inverse probability weights, there were no significant differences in covariates across infliximab stop timing groups (right-most column p-values), and the groups were well balanced in their characteristics (not shown).
Figure 1.
Cohort Creation
Infection within 30 days
Serious infection within 30 days occurred after 270 (6.3%) surgeries with bacteremia/septicemia, skin/soft tissue infection, and pneumonia the most common infections. Rate of infection was similar across categories of infliximab stop timing (Table 2). In propensity-adjusted analysis, subjects receiving infliximab within 4 weeks or 4–8 weeks before surgery were not more likely to have infection than subjects with stop timing of 8–12 weeks (Figure 2) [OR 0.90 (95% CI 0.60 to 1.34) and OR 0.95 (95% CI 0.62 to 1.36) respectively]. In similar propensity-adjusted analyses, infliximab stop timing < 2 weeks (n = 254) also was not associated with an increased risk of infection compared to stop timing of 8–12 weeks [OR 0.52 (95% CI 0.24 to 1.15)]. Results were similar in sensitivity analyses, including using the three alternative outcome definitions (Supplemental Table 2).
Table 2.
Frequency of hospitalized infection within 30 days and incidence rate of prosthetic joint infection within 1 year by infliximab stop timing
| Infliximab stop timing |
N | Person years |
Hospitalized Infection within 30 days, N (%, 95% CI) |
Prosthetic Joint Infection within 1 year, N (Incidence Rate/100py, 95% CI) |
|---|---|---|---|---|
| < 4 weeks | 1162 | 979 | 70 (6.0, 4.7–7.6) | 30 (3.1, 2.1–4.4) |
| 4–8 weeks | 1969 | 1665 | 117 (5.9, 4.9–7.1) | 47 (2.8, 2.1–3.8) |
| 8–12 weeks | 747 | 617 | 53 (7.1, 5.4–9.2) | 18 (2.9, 1.8–4.6) |
| 12–16 weeks | 218 | 179 | 15 (6.9, 3.9–11.1) | 7 (3.9, 1.9–8.2) |
| ≥ 16 weeks | 192 | 156 | 15 (7.8, 4.4–12.6) | 3 (1.9, 0.6–6.0) |
| Total | 4288 | 3596 | 270 (6.3, 5.6–7.1) | 105 (2.9, 2.4–3.5) |
Figure 2.
Propensity adjusted analysis of serious infection within 30 days and prosthetic joint infection within 1 year by infliximab stop timing
Prosthetic joint infection within 1 year
Prosthetic joint infection occurred after 105 surgeries (2.9/100 person years). Incidence rates were similar across infliximab stop timing categories (Table 2, Figure 3). In propensity-adjusted analysis, subjects receiving infliximab within 4 weeks vs 8–12 weeks before surgery did not have higher rates of PJI [HR 0.98 (95% CI 0.52 to 1.87)] (Figure 2). Results were similar but less precise for infliximab stop timing < 2 weeks vs 8–12 weeks [HR 1.14 (95% CI 0.48 to 2.71)]. Results were also similar in sensitivity analyses, including using the three alternative outcome definitions of PJI (Supplemental Table 3).
Figure 3.
Kaplan-Meier curves for prosthetic joint infection within 1 year of surgery by infliximab stop timing
Additional risk factors from multivariable models
In multivariable models, infliximab stop timing was not associated with infection within 30 days or time to PJI (Table 3). In contrast, glucocorticoid use > 10mg was associated with increased risk of infection within 30 days [OR 2.11 (95% CI 1.30 to 3.40)] and rate of PJI [HR 2.70 (95% CI 1.30 to 5.60)]. Concomitant non-biologic DMARD use was not associated with the risk of either outcome. Additional risk factors for infection within 30 days included age > 80, higher Charlson score, hospitalization for infection in the past year, greater number of outpatient visits in the past year, and lower surgeon volume (Table 3). Risk factors for PJI included revision knee surgery, hospitalization for infection in the past year, and extra-articular RA diagnosis codes (Table 3). Rates of both outcomes were lower in surgeries occurring 2010–2013 vs 2007–2009. Disposition at index hospitalization discharge (home, home health care, skilled nursing facility, inpatient rehabilitation) was not associated with PJI and was not a confounder (not shown).
Table 3.
Multivariable logistic regression analysis for serious infection within 30 days (4283 observations among 3863 patients), and multivariable Cox analysis for prosthetic joint infection within 1 year (4288 observations among 3867 patients, 105 events)
| Infection within 30 days OR (95% CI) |
Prosthetic joint infection within 1 year HR (95% CI) |
|
|---|---|---|
| Infliximab pre-operative stop timing | ||
| < 4 weeks | 0.81 (0.55, 1.18) | 1.02 (0.56, 1.83) |
| 4–8 weeks | 0.88 (0.62, 1.24) | 0.97 (0.56, 1.66) |
| 8–12 weeks | Reference | Reference |
| 12–16 weeks | 0.84 (0.44, 1.58) | 1.31 (0.55, 3.11) |
| ≥ 16 weeks | 0.95 (0.52, 1.70) | 0.60 (0.17, 2.06) |
| Glucocorticoid dose | ||
| None | Reference | Reference |
| ≤ 5mg/day | 0.98 (0.67, 1.44) | 1.90 (1.14, 3.18) |
| > 5 and ≤ 10mg/day | 1.13 (0.76, 1.67) | 1.93 (1.13, 3.31) |
| > 10mg/day | 2.11 (1.30, 3.40) | 2.70 (1.30, 5.60) |
| Non-biologic DMARD use | 1.05 (0.80, 1.38) | - |
| Previous non-infliximab biologic DMARD | - | 0.44 (0.20, 0.98) |
| Age ≥ 80 years old | 1.75 (1.23, 2.50) | 1.27 (0.69, 2.34) |
| Disease type (vs RA) | ||
| Inflammatory bowel disease | 1.02 (0.66, 1.56) | - |
| PsA/PsO/AS | 0.97 (0.64, 1.47) | - |
| Extra-articular RA | - | 2.11 (1.11, 4.03) |
| Surgery type (vs primary knee) | ||
| Primary Hip | - | 1.15 (0.72, 1.87) |
| Revision Knee | - | 3.77 (2.15, 6.61) |
| Revision Hip | - | 1.84 (0.84, 4.05) |
| Charlson score, per 1 point increase | 1.08 (1.03, 1.12) | - |
| Past year hospitalizations (vs none) | ||
| Hospitalization without infection | 0.71 (0.49, 1.02) | 1.57 (0.97, 2.54) |
| Hospitalization with infection | 1.91 (1.28, 2.84) | 2.81 (1.63, 4.86) |
| Outpatient visits past year, per visit | 1.02 (1.01, 1.03) | - |
| Calendar year (vs 2007–2009) | ||
| 2010–2013 | 0.67 (0.52, 0.87) | 0.65 (0.43, 0.96) |
| Surgeon volume (vs lowest tertile) | ||
| Middle tertile | 0.77 (0.57, 1.04) | - |
| Highest tertile | 0.62 (0.46, 0.85) | - |
Sex included in both models, region and inflammatory disease type in 30-day infection model, and skilled nursing facility stay in past year in prosthetic joint infection model but not shown (p > 0.05). Tested but not included in either final model with p > 0.2 for 30-day infection and p > 0.1 for prosthetic joint infection: race, urban, zip-code-based median household income quintiles, dual eligibility status, osteonecrosis, diabetes, chronic obstructive pulmonary disease/asthma, kidney disease, obesity, infliximab infusion interval, infliximab dose, antibiotic prescription in the past year, hospital volume
RA: rheumatoid arthritis, PsA: psoriatic arthritis, PsO: psoriasis, AS: ankylosing spondylitis
Secondary outcomes
Prolonged length of stay > 5 days occurred in 197 (4.7%) surgeries; in propensity-adjusted analyses, infliximab stop timing < 4 weeks vs 8–12 weeks was associated with lower risk of prolonged length of stay [OR 0.61 (95% CI 0.37 to 0.99) vs 8–12 weeks]. Hospitalized surgical site infection occurred within 90 days after surgery in 80/3996 (2.0%) surgeries with 90 days of follow up; infliximab stop timing < 4 weeks was not associated with increased risk [OR 0.87 (95% CI 0.41 to 1.90)]. Mortality was infrequent occurring in 4 (0.1%) within 30 days, 14 (0.3%) within 90 days, and 55 (1.3%) within 1 year.
Restart timing of infliximab
Infliximab was restarted within 4 weeks after 847 (20%), 4–8 weeks after 1913 (45%), 8–12 weeks after 514 (12%), 12–16 weeks after 174 (4%) and ≥ 16 weeks or never restarted after 840 (19%) surgeries. Patients who had restarted infliximab at any given time had lower rates of PJI than those who had not yet restarted [adjusted HR 0.50 (95% CI 0.28 to 0.89)]. Adding infliximab restart timing as a time-varying covariate to multivariable models did not change stop timing associations.
DISCUSSION
This study evaluated the association between the timing of infliximab in the perioperative period and the risk of infection among a large cohort. The risks of short-term hospitalized infection within 30 days and longer-term prosthetic joint infection within 12 months were not increased in patients who received infliximab within 4 weeks of elective hip or knee arthroplasty. In contrast, the risk of infection was substantially higher in patients who were receiving >10 mg per day of glucocorticoids.
Existing data to inform management of TNF antagonists in the perioperative period is limited. Some observational studies, including a recent meta-analysis, suggest that patients treated with TNF antagonists have higher rates of post-operative infection than patients treated with traditional DMARDs. These results, however, may be due to differences in disease severity or other patient characteristics that reflect residual confounding problematic in analyses that compare anti-TNF users to non-biologic DMARD users (12). These prior studies therefore do not directly address whether stopping anti-TNF therapy before surgery lowers the risk of infection. Results from small studies that have evaluated pre-operative timing of anti-TNFs generally have been underpowered, unable to fully address potential confounding, and have used varied definitions for stopping therapy based on dosing interval or 4–5 medication half-lives. Results of these studies are mixed, with some showing a lower risk of infection among patients stopping anti-TNFs before surgery (10, 39) and several showing no significant difference in the risk of infection (6–9, 11).
This study was designed specifically to evaluate perioperative biologic timing and has several advantages over previous studies. First, analysis was restricted to infliximab, because the precise timing of infusion therapy administration is available in claims data based on procedure codes from infusion centers, while precise timing of actual medication use cannot be achieved for non-infusion medications using administrative claims (or electronic health record data). At the time of this analysis, the data were too sparse for other IV biologics (e.g. abatacept) to include these other therapies. Secondly, including only patients treated with infliximab and comparing shorter stop timing to longer stop timing, rather than comparing users to non-users, can mitigate the substantial confounding present in studies comparing patients treated with biologics to those treated only with traditional DMARDs. In this study, there were few differences in patient characteristics between those with shorter and longer stop timing, which might be expected given the paucity of evidence on this topic. In addition, results were robust before and after adjustment for potential confounders, suggesting that there was not substantial confounding by indication based on stop timing.
The large size of this study is critical for evaluation of the relatively rare serious infections after arthroplasty and allows a degree of precision that is lacking from previous studies. Patients who received infliximab within 4 weeks of surgery were, if anything, slightly less likely to have a hospitalized infection. Although fewer patients received infliximab within 2 weeks of surgery these patients also showed no increase in the risk of infection. The large sample size makes this an informative negative study – as evidenced by the confidence intervals presented, if there is in fact a difference in the risk of infection by infliximab stop timing this difference is likely to be quite small. Analyses of prosthetic joint infection are less precise because of the smaller number of infections, but the similarity of results to those of hospitalized infection is reassuring.
Although infliximab timing was not found to be associated with the risk of post-operative infection, this study was large enough to identify other risk factors for infection including older age, certain comorbidities, previous infections, revision surgery, and lower surgeon volume. Of particular interest is the increase in the risk of 30-day infection with glucocorticoid doses above 10mg per day in the month prior to surgery and PJI with any pre-operative dose of glucocorticoids. This increased risk with glucocorticoids has been suggested by previous studies (7, 10, 40–44). Although this risk may be related in part to increased disease severity among glucocorticoid treated patients, a direct medication effect is likely. This data suggests that prolonged interruptions in infliximab therapy prior to surgery may be counterproductive if higher dose glucocorticoid therapy is used in substitution. Indeed, one possible explanation for this study’s results is that any small benefit to prolonged holding of infliximab is counteracted by an increased risk of infection from disease flares or glucocorticoid use (45).
The optimal time to restart therapy post-operatively is also of interest but is more difficult to address retrospectively because post-operative complications lead to delays in restarting therapy. As seen here, restarting therapy quickly may appear protective because physicians restart therapy sooner in patients who are doing well after surgery. Similar results were observed in a previous study of post-operative infection (11). The results of the current study do not contradict current recommendations to wait for evidence of satisfactory wound healing before restarting therapy (14).
Unmeasured confounders such as disease severity and surgeon/institution characteristics could conceivably impact stop timing and infection risk, although proxies for disease severity such as extra-articular RA and previous biologic use were included, as were surgeon and hospital volume. Reassuringly, minimal confounding was observed in the covariates assessed. Although it was not possible to identify whether surgery was required because of inflammatory arthritis or co-existing primary osteoarthritis, the underlying mechanism of joint destruction is not likely to be strongly associated with infliximab timing and thus would not be expected to be a confounder. Outcome misclassification, even if non-differential, could bias results towards the null. While the hospitalized infection outcome has been validated, performance of this outcome in the post-operative setting may differ. Strict exclusion criteria makes it unlikely that infections preceded surgery. Minor infections that might not lead to hospitalization but occur during the index hospitalization could be misclassified as serious infections. For this reason, urinary infections were excluded in the primary analysis. Sensitivity analyses excluding infections during the index hospitalization showed similar results. The PJI outcome definition has not been validated, although it has been used in a number of administrative database studies (28, 29) and three alternative outcome definitions provided similar results. Finally, results may not be generalizable to a younger, non-Medicare population, although this elderly cohort is arguably of greatest interest with regard to the question at hand.
Results of this study may not apply to other biologic DMARDs, particularly those with different mechanisms of action, but sample size was not sufficient to perform analysis for other infusion biologics. For TNF antagonists, however, the results of this study suggest that the timing of therapy is not strongly associated with the risk of post-operative infection and that administration of medications within a few weeks of surgery is not likely to adversely impact infectious outcomes. While results appeared similar in patients with and without RA, numbers were not sufficient to separately evaluate non-RA patients.
In conclusion, rates of both short and long-term infection were similar in patients who received infliximab within 4 weeks of hip or knee arthroplasty compared to those patients who held the medication for longer periods of time prior to surgery. In contrast, oral glucocorticoid use, especially > 10 mg per day, was associated with an increased risk of post-operative infection.
Supplementary Material
Significance and Innovation.
Patients receiving infliximab within 4 weeks before elective hip or knee arthroplasty were not at a higher risk of post-operative infection within 30 days or prosthetic joint infection within 1 year compared to patients holding therapy prior to surgery
Use of glucocorticoids, especially > 10mg per day, was associated with an increased risk of post-operative infection
Acknowledgments
The authors would like to thank Dr. Kevin Winthrop for his advice and critical review of the manuscript.
Funding/Grant sources: Dr. George was supported by the National Institutes of Health (NIH 5T32AR007442-28) and the Rheumatology Research Foundation Scientist Development Award. Dr. Baker is supported by a VA Clinical Science Research & Development Career Development Award (IK2 CX000955).
Financial interests/commercial sources: Dr. Yun has received research grants from Amgen for unrelated work. Dr. Curtis has received research grants for unrelated work and consulting fees > $10,000 from Pfizer, Amgen, UCB, and Myriad genetics and consulting fees < $10,000 from Bristol-Myers Squibb and Janssen.
References
- 1.Curtis JR, Patkar N, Xie A, Martin C, Allison JJ, Saag M, et al. Risk of serious bacterial infections among rheumatoid arthritis patients exposed to tumor necrosis factor alpha antagonists. Arthritis Rheum. 2007;56:1125–1133. doi: 10.1002/art.22504. [DOI] [PubMed] [Google Scholar]
- 2.Bongartz T, Sutton AJ, Sweeting MJ, Buchan I, Matteson EL, Montori V. Anti-TNF antibody therapy in rheumatoid arthritis and the risk of serious infections and malignancies: systematic review and meta-analysis of rare harmful effects in randomized controlled trials. JAMA. 2006;295:2275–2285. doi: 10.1001/jama.295.19.2275. [DOI] [PubMed] [Google Scholar]
- 3.Galloway JB, Hyrich KL, Mercer LK, Dixon WG, Ustianowski AP, Helbert M, et al. Risk of septic arthritis in patients with rheumatoid arthritis and the effect of anti-TNF therapy: results from the British Society for Rheumatology Biologics Register. Ann Rheum Dis. 2011;70:1810–1814. doi: 10.1136/ard.2011.152769. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Galloway JB, Mercer LK, Moseley A, Dixon WG, Ustianowski AP, Helbert M, et al. Risk of skin and soft tissue infections (including shingles) in patients exposed to anti-tumour necrosis factor therapy: results from the British Society for Rheumatology Biologics Register. Ann Rheum Dis. 2013;72:229–234. doi: 10.1136/annrheumdis-2011-201108. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Dixon WG, Symmons DPM, Lunt M, Watson KD, Hyrich KL, Silman AJ. Serious infection following anti-tumor necrosis factor α therapy in patients with rheumatoid arthritis: Lessons from interpreting data from observational studies. Arthritis Rheum. 2007;56:2896–2904. doi: 10.1002/art.22808. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.den Broeder AA, Creemers MCW, Fransen J, Jong Ede, Rooij D-JR de, Wymenga A, et al. Risk factors for surgical site infections and other complications in elective surgery in patients with rheumatoid arthritis with special attention for anti-tumor necrosis factor: a large retrospective study. J Rheumatol. 2007;34:689–695. [PubMed] [Google Scholar]
- 7.Bongartz T, Halligan CS, Osmon DR, Reinalda MS, Bamlet WR, Crowson CS, et al. Incidence and risk factors of prosthetic joint infection after total hip or knee replacement in patients with rheumatoid arthritis. Arthritis Rheum. 2008;59:1713–1720. doi: 10.1002/art.24060. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Ruyssen-Witrand A, Gossec L, Salliot C, Luc M, Duclos M, Guignard S, et al. Complication rates of 127 surgical procedures performed in rheumatic patients receiving tumor necrosis factor alpha blockers. Clin Exp Rheumatol. 2007;25:430–436. [PubMed] [Google Scholar]
- 9.Waterman M, Xu W, Dinani A, Steinhart AH, Croitoru K, Nguyen GC, et al. Preoperative biological therapy and short-term outcomes of abdominal surgery in patients with inflammatory bowel disease. Gut. 2013;62:387–394. doi: 10.1136/gutjnl-2011-301495. [DOI] [PubMed] [Google Scholar]
- 10.Scherrer CB, Mannion AF, Kyburz D, Vogt M, Kramers-de Quervain IA. Infection risk after orthopedic surgery in patients with inflammatory rheumatic diseases treated with immunosuppressive drugs. Arthritis Care Res. 2013;65:2032–2040. doi: 10.1002/acr.22077. [DOI] [PubMed] [Google Scholar]
- 11.Abou Zahr Z, Spiegelman A, Cantu M, Ng B. Perioperative use of anti-rheumatic agents does not increase early postoperative infection risks: a Veteran Affairs’ administrative database study. Rheumatol Int. 2015;35:265–272. doi: 10.1007/s00296-014-3121-0. [DOI] [PubMed] [Google Scholar]
- 12.Goodman SM, Menon I, Christos PJ, Smethurst R, Bykerk VP. Management of perioperative tumour necrosis factor α inhibitors in rheumatoid arthritis patients undergoing arthroplasty: a systematic review and meta-analysis. Rheumatol Oxf Engl. 2016;55:573–582. doi: 10.1093/rheumatology/kev364. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Saag KG, Teng GG, Patkar NM, Anuntiyo J, Finney C, Curtis JR, et al. American College of Rheumatology 2008 recommendations for the use of nonbiologic and biologic disease-modifying antirheumatic drugs in rheumatoid arthritis. Arthritis Rheum. 2008;59:762–784. doi: 10.1002/art.23721. [DOI] [PubMed] [Google Scholar]
- 14.Bombardier C, Hazlewood GS, Akhavan P, Schieir O, Dooley A, Haraoui B, et al. Canadian Rheumatology Association Recommendations for the Pharmacological Management of Rheumatoid Arthritis with Traditional and Biologic Disease-modifying Antirheumatic Drugs: Part II Safety. J Rheumatol. 2012;39:1583–1602. doi: 10.3899/jrheum.120165. [DOI] [PubMed] [Google Scholar]
- 15.Ding T, Ledingham J, Luqmani R, Westlake S, Hyrich K, Lunt M, et al. BSR and BHPR rheumatoid arthritis guidelines on safety of anti-TNF therapies. Rheumatol Oxf Engl. 2010;49:2217–2219. doi: 10.1093/rheumatology/keq249a. [DOI] [PubMed] [Google Scholar]
- 16.Goodman SM. Rheumatoid arthritis: Perioperative management of biologics and DMARDs. Semin Arthritis Rheum. 2015 doi: 10.1016/j.semarthrit.2015.01.008. [DOI] [PubMed] [Google Scholar]
- 17.Medicare C for, Baltimore MS 7500 SB, Usa M. [Accessed May 1, 2015];SummaryMedicareMedicaid. 2015 Available at: http://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/MedicareProgramRatesStats/SummaryMedicareMedicaid.html.
- 18.Herrinton LJ, Curtis JR, Chen L, Liu L, Delzell E, Lewis JD, et al. Study design for a comprehensive assessment of biologic safety using multiple healthcare data systems. Pharmacoepidemiol Drug Saf. 2011;20:1199–1209. doi: 10.1002/pds.2196. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Kim SY, Servi A, Polinski JM, Mogun H, Weinblatt ME, Katz JN, et al. Validation of rheumatoid arthritis diagnoses in health care utilization data. Arthritis Res Ther. 2011;13:R32. doi: 10.1186/ar3260. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Liu L, Allison JE, Herrinton LJ. Validity of computerized diagnoses, procedures, and drugs for inflammatory bowel disease in a northern California managed care organization. Pharmacoepidemiol Drug Saf. 2009;18:1086–1093. doi: 10.1002/pds.1824. [DOI] [PubMed] [Google Scholar]
- 21.Curtis JR, Harrold LR, Asgari MM, Deodhar A, Salman C, Gelfand JM, et al. Diagnostic Prevalence of Ankylosing Spondylitis Using Computerized Health Care Data, 1996 to 2009: Underrecognition in a US Health Care Setting. Perm J. 2016;20:4–10. doi: 10.7812/TPP/15-151. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Asgari MM, Wu JJ, Gelfand JM, Salman C, Curtis JR, Harrold LR, et al. Validity of diagnostic codes and prevalence of psoriasis and psoriatic arthritis in a managed care population, 1996–2009. Pharmacoepidemiol Drug Saf. 2013;22:842–849. doi: 10.1002/pds.3447. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Mahomed NN, Barrett JA, Katz JN, Phillips CB, Losina E, Lew RA, et al. Rates and outcomes of primary and revision total hip replacement in the United States medicare population. J Bone Joint Surg Am. 2003;85–A:27–32. doi: 10.2106/00004623-200301000-00005. [DOI] [PubMed] [Google Scholar]
- 24.Katz JN, Barrett J, Mahomed NN, Baron JA, Wright RJ, Losina E. Association between hospital and surgeon procedure volume and the outcomes of total knee replacement. J Bone Joint Surg Am. 2004;86–A:1909–1916. doi: 10.2106/00004623-200409000-00008. [DOI] [PubMed] [Google Scholar]
- 25.Mahomed NN, Barrett J, Katz JN, Baron JA, Wright J, Losina E. Epidemiology of total knee replacement in the United States Medicare population. J Bone Joint Surg Am. 2005;87:1222–1228. doi: 10.2106/JBJS.D.02546. [DOI] [PubMed] [Google Scholar]
- 26.Katz JN, Losina E, Barrett J, Phillips CB, Mahomed NN, Lew RA, et al. Association between hospital and surgeon procedure volume and outcomes of total hip replacement in the United States medicare population. J Bone Joint Surg Am. 2001;83–A:1622–1629. doi: 10.2106/00004623-200111000-00002. [DOI] [PubMed] [Google Scholar]
- 27.Schneeweiss S, Setoguchi S, Weinblatt ME, Katz JN, Avorn J, Sax PE, et al. Anti-tumor necrosis factor alpha therapy and the risk of serious bacterial infections in elderly patients with rheumatoid arthritis. Arthritis Rheum. 2007;56:1754–1764. doi: 10.1002/art.22600. [DOI] [PubMed] [Google Scholar]
- 28.Kurtz SM, Ong KL, Lau E, Bozic KJ, Berry D, Parvizi J. Prosthetic joint infection risk after TKA in the Medicare population. Clin Orthop. 2010;468:52–56. doi: 10.1007/s11999-009-1013-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Bozic KJ, Lau E, Kurtz S, Ong K, Rubash H, Vail TP, et al. Patient-Related Risk Factors for Periprosthetic Joint Infection and Postoperative Mortality Following Total Hip Arthroplasty in Medicare Patients. [Accessed March 6, 2015];J Bone Jt Surg Am. 2012 :94. doi: 10.2106/JBJS.K.00072. Available at: http://jbjs.org/cgi/doi/10.2106/JBJS.K.00072. [DOI] [PubMed]
- 30.Fry DE, Pine M, Locke D, Pine G. Prediction models of Medicare 90-day postdischarge deaths, readmissions, and costs in bowel operations. Am J Surg. 2015;209:509–514. doi: 10.1016/j.amjsurg.2014.12.005. [DOI] [PubMed] [Google Scholar]
- 31.Yokoe D, Onufrak F, Olsen M, Stevenson K, Khan Y, Hooper D. Multicenter evaluation of enhanced surgical site infection surveillance following total knee and hip arthroplasty. Abstr Int Conf Healthc-Assoc Infect. 2010 Available at: https://shea.confex.com/shea/2010/webprogram/Paper1547.html.
- 32.Gagne JJ, Glynn RJ, Avorn J, Levin R, Schneeweiss S. A combined comorbidity score predicted mortality in elderly patients better than existing scores. J Clin Epidemiol. 2011;64:749–759. doi: 10.1016/j.jclinepi.2010.10.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Anon. [Accessed October 3, 2015];U.S. Census Bureau, 2009–2013 5-Year American Community Survey. Available at: https://www.census.gov/programs-surveys/acs.
- 34.Spreeuwenberg MD, Bartak A, Croon MA, Hagenaars JA, Busschbach JJV, Andrea H, et al. The multiple propensity score as control for bias in the comparison of more than two treatment arms: an introduction from a case study in mental health. Med Care. 2010;48:166–174. doi: 10.1097/MLR.0b013e3181c1328f. [DOI] [PubMed] [Google Scholar]
- 35.Feng P, Zhou X-H, Zou Q-M, Fan M-Y, Li X-S. Generalized propensity score for estimating the average treatment effect of multiple treatments. Stat Med. 2012;31:681–697. doi: 10.1002/sim.4168. [DOI] [PubMed] [Google Scholar]
- 36.McCaffrey DF, Griffin BA, Almirall D, Slaughter ME, Ramchand R, Burgette LF. A tutorial on propensity score estimation for multiple treatments using generalized boosted models. Stat Med. 2013;32:3388–3414. doi: 10.1002/sim.5753. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Austin PC, Schuster T. The performance of different propensity score methods for estimating absolute effects of treatments on survival outcomes: A simulation study. Stat Methods Med Res. 2014 doi: 10.1177/0962280213519716. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Cole SR, Hernán MA. Constructing inverse probability weights for marginal structural models. Am J Epidemiol. 2008;168:656–664. doi: 10.1093/aje/kwn164. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Dixon W, Lunt M, Watson K, Hyrich K, Symmons D. Anti-TNF therapy and the risk of serious post-operative infection: results from the BSR Biologics register (BSRBR) Ann Rheum Dis. 2007;66(Suppl. II):118. [Google Scholar]
- 40.Michaud K, Fehringer EV, Garvin K, O’Dell JR, Mikuls TR. Rheumatoid arthritis patients are not at increased risk for 30-day cardiovascular events, infections, or mortality after total joint arthroplasty. Arthritis Res Ther. 2013;15:R195. doi: 10.1186/ar4385. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Gilson M, Gossec L, Mariette X, Gherissi D, Guyot M-H, Berthelot J-M, et al. Risk factors for total joint arthroplasty infection in patients receiving tumor necrosis factor α-blockers: a case-control study. Arthritis Res Ther. 2010;12:R145. doi: 10.1186/ar3087. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Somayaji R, Barnabe C, Martin L. Risk factors for infection following total joint arthroplasty in rheumatoid arthritis. Open Rheumatol J. 2013;7:119–124. doi: 10.2174/1874312920131210005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Aberra FN, Lewis JD, Hass D, Rombeau JL, Osborne B, Lichtenstein GR. Corticosteroids and immunomodulators: postoperative infectious complication risk in inflammatory bowel disease patients. Gastroenterology. 2003;125:320–327. doi: 10.1016/s0016-5085(03)00883-7. [DOI] [PubMed] [Google Scholar]
- 44.Ferrante M, D’Hoore A, Vermeire S, Declerck S, Noman M, Van Assche G, et al. Corticosteroids but not infliximab increase short-term postoperative infectious complications in patients with ulcerative colitis. Inflamm Bowel Dis. 2009;15:1062–1070. doi: 10.1002/ibd.20863. [DOI] [PubMed] [Google Scholar]
- 45.Au K, Reed G, Curtis JR, Kremer JM, Greenberg JD, Strand V, et al. High disease activity is associated with an increased risk of infection in patients with rheumatoid arthritis. Ann Rheum Dis. 2011;70:785–791. doi: 10.1136/ard.2010.128637. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.