1. Background

Total knee arthroplasty (TKA) is a mainstay treatment option for patients suffering from degenerative osteoarthritis (OA) of the knee, and is expected to grow to a total of 3.48 million annual procedures by 2030 [1]. Prior to surgical intervention, intra-articular injection of corticosteroids or hyaluronic acid (HA) into the knee to improve pain and function is commonly performed in up to 30% of patients undergoing TKA [2]. Despite the expanded use of intra-articular injections, debate regarding their efficacy and durability of effect in the treatment of OA remains [3-6]. This ambiguity is reflected by the American Academy of Orthopedic Surgeons (AAOS) 2013 treatment guidelines, which are unable to recommend for or against the use of intra-articular corticosteroids and recommend against the use of hyaluronic acid for patients with symptomatic knee OA [7]. Further, there is controversy surrounding whether the timing of preoperative intra-articular injections increases the risk of periprosthetic joint infection (PJI) following both total hip arthroplasty (THA) [8-10] and TKA [11-17]. Proposed reasons for this increased risk include the immunosuppressive effects of corticosteroids and the direct infiltration of the joint from the injection itself [13].

The concepts of real world data and real world evidence refer to the use of health information from multiple sources outside of typical clinical research settings, such as electronic health records, claims data, registries and personal devices [18-20]. Evidence from these non-traditional sources can be complimentary to clinical trials and may efficiently allow for the investigation of broad populations, although limitations of data quality and the potential for confounding and bias must be considered when evaluating these studies [20]. To investigate the relationship between preoperative injection timing and risk of postoperative PJI, we analyzed a large, multicenter longitudinal database.

2. Methods

After receiving institutional review board exemption, the TriNetX Research database was retrospectively queried as of October 1, 2019 to evaluate all patients with a diagnosis of osteoarthritis undergoing primary TKA between January 1, 2010 and September 30, 2018. Patients were then classified based on whether they received a preoperative intra-articular injection of hyaluronic acid or corticosteroid within the three months prior to surgery. Analysis was performed using unmatched and propensity score matched cohorts. The primary endpoint was periprosthetic joint infection within 12 months of surgery, identified by relevant diagnosis or procedure codes. Statistical analysis was performed within the TriNetX Analytics platform and Microsoft Excel, with odds ratios calculated as described by Altman [21].

2.1 Cohort definitions

Surgery was defined by the presence of the primary total knee arthroplasty current procedural terminology (CPT) code along with the appropriate international classification of disease 10^th edition (ICD-10) diagnosis code for unilateral post-traumatic or primary osteoarthritis of the left or right knee, respectively, between January 1, 2010 and September 30, 2018. Injection within 3 months of surgery was defined as any instance of the CPT code for arthrocentesis, aspiration and/or injection of a major joint or bursa (with or without ultrasound guidance), along with the matching ICD-10 laterality code, and a healthcare common procedure coding system (HCPCS) code for one of the included injection types. All patients with a concurrent ICD-10 code of post-traumatic or primary OA of the contralateral knee at the time of injection were excluded to control for laterality. Patients not receiving injections were defined by the presence of a surgery with a laterality code, excluding patients with any instance of injection on the ipsilateral knee within 3 months of the surgery date. The 3 month preoperative period was selected to allow for comparison with previous studies, and because our institutional protocol is to not perform TKA within 3 months of intra-articular injection on the operative knee.

2.2 Risk adjustment

Propensity scores were developed based on age and the presence or absence of the following comorbidities, as defined by ICD-10 codes, within one-year prior to surgery: essential (primary) hypertension, overweight and obesity, diabetes mellitus, personal history of nicotine dependence, other cardiac arrhythmias, chronic ischemic heart disease, nicotine dependence, and atrial fibrillation and flutter.

2.3 Endpoint

In alignment with prior studies, PJI was identified by the ICD-10 code for infection due to internal knee prosthesis or the CPT codes for treatment of infection. Surgical treatments of infection that were included were incision and drainage, deep incision with opening of bone cortex, arthrotomy with exploration and drainage or removal of foreign body, or removal of prosthesis with or without insertion of a spacer within one year of primary surgery. A full list of codes used to define the treatments, comorbidities, and endpoint is included as Appendix A.

2.4 About TriNetX

TriNetX is a “global health research network that optimizes clinical research and enables discoveries through the generation of real-world evidence” [22]. The research platform includes a federated health research network providing access to statistics on electronic medical records (diagnoses, procedures, medications, laboratory values, genomic information) including longitudinal data from 26 health care organizations and includes over 37 million patients. TriNetX received a waiver from Western IRB, as no protected health information is included in the database. On average, participants submit data retrospectively for seven years, with some providing historical data 13 years or older [23]. Diagnoses and procedures coded using ICD-9 (prior to October 1, 2015) are converted to ICD-10 using General Equivalence Mapping (GEMS) [24]. Using this methodology, all ICD-9 OA codes lacking laterality are mapped to non-laterality specific ICD-10 codes and are therefore excluded from this analysis. Statistical analysis is performed within the analytics platform [23].

3. Results

A total of 19,510 patients undergoing primary unilateral TKA between January 1, 2010 and September 30, 2018 were retrospectively reviewed. Of the 903 patients receiving injections within 3 months of surgery, 125 (14%) received hyaluronic acid only, 715 (79%) received corticosteroid only, and the remaining 63 (7%) received a combination of hyaluronic acid and corticosteroid. In unmatched analysis of the total sample of 19,510 subjects, no significant difference in one-year PJI rate was observed (No injection: 416 PJI (2.24%) vs. Injection: 28 PJI (3.10%), No Injection OR=0.715, 95% CI: 0.485-1.054, p=.090) (Table 1). Patients were then propensity score matched to control for potentially confounding comorbidities. No significant differences in age or comorbidities remained after propensity score matching and controlling for laterality (Table 2). After propensity score matching, no significant difference in infection rate between groups at one year was observed (No injection: 20 PJI (2.21%) vs. Injection: 28 PJI (3.10%), No Injection OR=0.708, 95% CI: 0.396-1.266, p=.244) (Table 1).

* Number of subjects in each propensity score matched cohort = 903.

Table 1: One year PJI rates after TKA: unmatched and propensity score matched analysis.

Table 2: TKA propensity score matching, controlling for laterality.

4. Discussion

The rates of infection following TKA observed in our study are similar to those previously reported [13, 16, 17]. Our findings suggest that even after controlling for age and comorbidities, patients undergoing ipsilateral intra-articular injections within three months of TKA may not be at significantly increased risk for PJI.

Our findings align with those of Amin et al, whose 2016 retrospective review of 1,628 TKA patients is the largest study concluding that there does not appear to be a correlation between timing of injection before surgery and increased risk of infection [25]. The authors found a deep postoperative infection rate of 0.77% in patients receiving preoperative injections within 12 months prior to surgery and 1.18% in the control group of those who did not. No differences in infection rate between the no injection group (1.18%), patients receiving steroid injections (1.11%) or patients receiving viscosupplementation (0.47%) were observed. No difference in infection rate was found regardless of the timing of the preoperative injection, with patients undergoing injections within 3 months of surgery demonstrating a PJI rate of 1.4%. Average time to deep infection was 5 months in the control group and 8.67 months in the injection group. While our study and Amin’s reached similar conclusions, at 2.24% for no injection and 3.10% for injections, the rates of PJI observed in our study are higher than those reported by Amin. A strength of our study is that it further supports these previous conclusions, while incorporating risk adjustment through propensity score matching to control for the impact of age and comorbidities on risk for PJI.

Other smaller studies have also reached the conclusion that preoperative injections do not increase the risk of PJI after TKA. In a review of 442 patients undergoing primary TKA, Kokubun et al found that after controlling for confounding variables, intra-articular corticosteroid, viscosupplementation, and any injection within 90 days were not associated with an increase in complications, infection, or poor functional outcomes after TKA (all p> .05). On multivariate logistic regression analysis, injection within 90 days resulted in no significant increase in risk for infection (OR=0.534, 95% CI 0.116-2.446, p= .419) [16].  Desai et al examined the risk of superficial and deep infections in patients receiving intra-articular steroid injections within 12 months of TKA. They found a superficial infection rate of 4.4% in the injection group (n=45) and 2.8% in the control group (n=180)—a difference that was not statistically significant—and no deep infections in either group [17]. In addition to these primary evaluations, two systematic reviews [2, 26] concluded that preoperative intra-articular injection does not increase the risk of PJI following TKA, but also highlighted that current studies are often underpowered and may suffer from selection bias.

In contrast to our findings, a retrospective review of 144 patients by Papavasiliou et al raised concern that intra-articular steroid injections may increase risk for postoperative deep infection following TKA, as evidenced by a 22.2% rate of wound complications in the 54 patients who underwent injection prior to surgery compared to a rate of 11.1% in 90 controls. Three of 54 (5.6%) patients undergoing preoperative injections within 12 months of TKA had deep infections, compared to zero in the control group [14]. Due to the small sample size [27], ambiguous definition of infection [28], and lack of correlation between injection timing and infection rate, the validity of the study has been debated [17].

The strongest evidence supporting an increased risk of postoperative PJI for patients undergoing viscosupplementation or steroid injections prior to TKA is presented by Richardson et al’s 2019 review of 58,337 patients from a national database [13]. The overall 6-month postoperative infection rate, including the control group of patients not receiving injections within 12 months of TKA, was 2.83% (No injection PJI rate=2.74%, HA within 3 months of TKA PJI rate=4.18% and corticosteroid within 3 months of TKA PJI rate=3.25%). These results closely align with our finding of a 2.24% PJI rate in the unmatched and 2.21% in the propensity score matched no-injection cohorts, compared to a PJI rate of 3.10% in patients receiving an injection within 3 months of TKA. [13]. Despite demonstrating similar trends in infection rates, our studies reached opposing conclusions based on the statistical significance of the differences observed. Both studies are comparable in their use of large, administrative databases to assess risk of PJI, and utilization of statistical controls for potentially confounding factors. Significant differences in approach include Richardson’s exclusion of patients receiving both corticosteroids and hyaluronic acid injection, which were included in our analysis, Richardson’s use of a 6-month postoperative infection rate in comparison to our evaluation of infections up to 12 months postoperatively, and Richardson’s stratification of PJI rate and risk by injection type, whether multiple injections were received, and comparison of risk between multiple preoperative injection time points.

Based on the conflicting conclusions reached by the studies presented, we suggest the current state of the literature is not sufficient to reach a consensus regarding the risk of PJI following TKA after injections within three months of surgery. Until consensus is reached, our institution will continue to utilize a conservative approach and not perform TKA within three months of intra-articular injection. Given the inherent limitations of administrative datasets relying on coded data—which has been demonstrated to have inaccuracies, [29] we suggest a large-scale, multisite retrospective review of outcomes using clinician validated measures is needed if a consensus treatment guideline is to be reached.

5. Limitations

The primary limitation of our study is its reliance on coded data submitted to an administrative database in a blinded fashion. This data structure inherently limits our granularity of analysis, and limits the ability to evaluate potentially clinically significant factors such as whether dosing vials were single or multi-use, exact number of doses received, and methods of controlling for sterility at the time of injection. Despite these potential limitations, which are inherently present in aggregated databases, we suggest the use of large-scale real world data is a valuable supplement to smaller, but potentially more robust data sets from institutional reviews. Finally, our study was limited by the relatively small number of patients receiving injections within the three-month preoperative period (903). This left the study underpowered to assess differences in PJI rate by injection type, as sample sizes of 2,000 patients per group are recommended to rule out a 50% increase in infection rate across cohorts [17].

6. Conclusion

Based on analysis of 19,510 patients undergoing TKA from a longitudinal multicenter database, after controlling for age and comorbidities, preoperative injection within three months of surgery was not associated with increased risk for postoperative periprosthetic joint infection. A large, multicenter, retrospective review of outcomes is warranted to establish a true consensus regarding appropriate preoperative timing of injections. We continue to recommend caution in administering injections in the three months preceding surgery until a consensus can be reached.

Conflict of Interest

None of the authors have any relevant conflict of interest to disclose.

Funding

No corporate entity or foundation provided any funding for this investigation.

Institutional Review Board

Study was deemed exempt by the institutional Clinical Research Committee.

References

1. Kurtz S, Ong K, Lau E, et al. Projections of primary and revision hip and knee arthroplasty in the United States from 2005 to 2030. J Bone Joint Surg Am 89 (2007): 780-785.
2. Marsland D, Mumith A, Barlow IW. Systematic review: the safety of intra-articular corticosteroid injection prior to total knee arthroplasty. The Knee 21 (2014): 6-11.
3. Bellamy N, Campbell J, Robinson V, et al. Intraarticular corticosteroid for treatment of osteoarthritis of the knee. Cochrane Database Syst Rev 2 (2006): Cd005328.
4. Campbell KA, Erickson BJ, Saltzman BM, et al. Is Local Viscosupplementation Injection Clinically Superior to Other Therapies in the Treatment of Osteoarthritis of the Knee: A Systematic Review of Overlapping Meta-analyses. Arthroscopy 31 (2015): 2036-2045.e2014.
5. Divine JG, Zazulak BT, Hewett TE. Viscosupplementation for knee osteoarthritis: a systematic review. Clin Orthop Relat Res 455 (2007): 113-122.
6. Jevsevar D, Donnelly P, Brown GA, et al. Viscosupplementation for Osteoarthritis of the Knee: A Systematic Review of the Evidence. J Bone Joint Surg Am 97 (2015): 2047-2060.
7. Brown GA. AAOS clinical practice guideline: treatment of osteoarthritis of the knee: evidence-based guideline, 2nd edition. J Am Acad Orthop Surg 21 (2013): 577-579.
8. McIntosh AL, Hanssen AD, Wenger DE, et al. Recent intraarticular steroid injection may increase infection rates in primary THA. Clin Orthop Relat Res 451 (2006): 50-54.
9. Meermans G, Corten K, Simon JP. Is the infection rate in primary THA increased after steroid injection? Clin Orthop Relat Res 470 (2012): 3213-3219.
10. Schairer WW, Nwachukwu BU, Mayman DJ, et al. Preoperative Hip Injections Increase the Rate of Periprosthetic Infection After Total Hip Arthroplasty. J Arthroplasty 31 (2016): 166-169.e161.
11. Bedard NA, Pugely AJ, Elkins JM, et al. The John N. Insall Award: Do Intraarticular Injections Increase the Risk of Infection After TKA? Clin Orthop Relat Res 475 (2017): 45-52.
12. Cancienne JM, Werner BC, Luetkemeyer LM, et al. Does Timing of Previous Intra-Articular Steroid Injection Affect the Post-Operative Rate of Infection in Total Knee Arthroplasty? J Arthroplasty 30 (2015): 1879-1882.
13. Richardson SS, Schairer WW, Sculco TP, et al. Comparison of Infection Risk with Corticosteroid or Hyaluronic Acid Injection Prior to Total Knee Arthroplasty. J Bone Joint Surg Am 101 (2019): 112-118.
14. Papavasiliou AV, Isaac DL, Marimuthu R, et al. Infection in knee replacements after previous injection of intra-articular steroid. J Bone Joint Surg Br 88 (2006): 321-323.
15. Horne G, Devane P, Davidson A, et al. The influence of steroid injections on the incidence of infection following total knee arthroplasty. N Z Med J 121 (2008): U2896.
16. Kokubun BA, Manista GC, Courtney PM, et al. Intra-Articular Knee Injections Before Total Knee Arthroplasty: Outcomes and Complication Rates. J Arthroplasty 32 (2017): 1798-1802.
17. Desai A, Ramankutty S, Board T, et al. Does intraarticular steroid infiltration increase the rate of infection in subsequent total knee replacements? The Knee 16 (2009): 262-264.
18. Innovation NfEiH. Real World Evidence: A New Era for Health Care Innovation (2015).
19. Administration USFaD. Use of Real-World Evidence to Support Regulatory Decision-Making for Medical Devices (2017).
20. Sherman RE, Anderson SA, Dal Pan GJ, et al. Real-World Evidence-What Is It and What Can It Tell Us? N Engl J Med 375 (2016): 2293-2297.
21. Altman DG. Practical Statistics for medical research. London: Chapman and Hall (1991).
22. Answers at the Speed of Thought (2019).
23. About the Data (2019).
24. Services CfMaM. 2015 ICD-10-CM and GEMS 2014 (2019).
25. Amin NH, Omiyi D, Kuczynski B, et al. The Risk of a Deep Infection Associated With Intraarticular Injections Before a Total Knee Arthroplasty. J Arthroplasty 31 (2016): 240-244.
26. McMahon SE, LeRoux JA, Smith TO, et al. Total joint arthroplasty following intra-articular steroid injection: a literature review. Acta Orthop Belg 79 (2013): 672-679.
27. Dodd LE. Infection in knee replacements after previous injection of intra-articular steroid. J Bone Joint Surg Brn 89 (2007): 422.
28. Little NJ, Chipperfield A, Ricketts DM. Infection in knee replacements after previous injection of intra-articular steroid. J Bone Joint Surg Br 89 (2007): 423.
29. Singh JA, Kundukulam JA, Bhandari M. A systematic review of validated methods for identifying orthopedic implant removal and revision using administrative data. Pharmacoepidemiol Drug Saf 21 (2012): 265-273.