Abstract
Background
There is little literature examining the association of corticosteroid injections into shoulders with a pre-existing arthroplasty. The aim of the current study was to determine the risk of early infection following intra-articular corticosteroid injection into a pre-existing shoulder arthroplasty.
Methods
The PearlDiver database was retrospectively reviewed to identify patients with a pre-existing shoulder arthroplasty from 2007 to 2017. Patients with an ipsilateral shoulder corticosteroid injection in the postoperative period were identified. A control group of patients without an injection was matched 4:1 by age, gender, and postoperative timepoint. Periprosthetic infection within six months after the injection was then assessed and compared using a logistic regression analysis.
Results
Nine hundred and fifty-eight patients were identified who underwent a postoperative corticosteroid injection into a pre-existing shoulder arthroplasty and compared to 3832 control patients. After controlling for demographics, comorbidities, and procedure type, the rate of infection in patients who received a postoperative corticosteroid injection (1.77%) was significantly higher than control patients who did not receive an injection (0.91%) (OR 1.98 (95% CI 1.31–2.98), p = 0.0253).
Conclusions
There is a significant association between intra-articular shoulder corticosteroid injections in patients with pre-existing shoulder arthroplasties and prosthetic joint infection compared to matched controls without postoperative injections.
Study Design
Level III, retrospective cohort study.
Introduction
The rate of total shoulder arthroplasty (TSA) surgery continues to increase and is substantially higher than total knee and hip arthroplasties.1,2 Shoulder arthroplasty in the correctly indicated patient has been shown to be a successful operation with durable and sustainable mid and long term outcomes.3,4 Despite such promising clinical outcomes, similar to lower extremity arthroplasty reports, there remain a subset of patients who experience persistent pain following surgery. 5 In fact, a recent Danish Shoulder Arthroplasty Register study reported a 22% prevalence of persistent pain one to two years following shoulder arthroplasty. 4 The management of such pain is predicated on its etiology, which often include the clinical workup and evaluation of infection, aseptic loosening, or periprosthetic fracture. When the source of pain is identifiable and correctable, treatment strategies are often successful. However, in some cases the source of pain is more complex and multifactorial, with a cause that is more uncertain. This subset of patients represents a treatment dilemma for the treating surgeon, and many patients are often dissatisfied with conservative treatments such as physical therapy, anti-inflammatory medications, and activity modification.
Although corticosteroid injections are an accepted and well adopted part of the preoperative treatment algorithm for patients who ultimately undergo shoulder arthroplasty, its use postoperatively remains a controversial treatment strategy. Like many other philosophies adopted from the lower extremity arthroplasty literature, physicians are typically hesitant to administer corticosteroid injections into a pre-existing shoulder arthroplasty due to concern for increasing the risk of infection. While such concern is certainly justified, there is little evidence to support this claim and the practice has not been completely abandoned across the country. Therefore, the purpose of the present study was to utilize a national database to determine the risk of prosthetic joint infection following administration of a corticosteroid injection into a pre-existing shoulder arthroplasty.
Materials and methods
The PearlDiver database, an insurance-based, for-fee database of patient records, was used for the present study (PearlDiver Patient Records Database, www.pearldiverinc.com, Fort Wayne, IN). The Humana, private-payer, dataset was utilized for the current study, because it contains laterality for arthroplasty procedures, which is not the case for the Medicare dataset within PearlDiver. This dataset contains procedural volumes, patient demographics, and laterality modifiers for patients with International Classification of Diseases, 9th Revision (ICD-9), diagnoses and procedures or Current Procedural Terminology (CPT) codes. The Humana dataset currently includes patients insured from 2007 to 2017, and, in total, contains approximately 20 million unique patients with orthopedic diagnoses. Data within the system are deidentified and this study is thus exempt from institutional review board approval.
Study groups
The database was first queried for patients who underwent primary TSA or hemiarthroplasty using CPT codes 23470 and 23472. Any patients without a coded laterality, those with a history of revision shoulder arthroplasty, and any patients with infectious-related procedural or diagnostic shoulder codes were excluded. Patients who underwent bilateral shoulder arthroplasty procedures were accounted for as two separate shoulders using laterality modifiers.
Patients who underwent ipsilateral postoperative intra-articular shoulder injections were then identified by searching for patients with a large joint injection (CPT 20610) for an associated shoulder-specific diagnosis (ICD-9 coding) with a steroid “J” code for corticosteroid preparations administered as a local injection. These included J3300 (injection, triamcinolone acetonide, 1 mg), J3301 (injection, triamcinolone acetonide, 10 mg), J1020 (injection, methylprednisolone acetate, 20 mg), J1030 (injection, methylprednisolone acetate, 40 mg), J1040 (injection, methylprednisolone acetate, 80 mg), JO702, JO704, J1100, J3302, J3303, and J1094. These injections were subsequently divided into “left” and “right” using CPT laterality modifier codes. Any patient without an associated laterality modifier for the injection was excluded. Those patients with a history of a shoulder arthroplasty with a documented ipsilateral postoperative intra-articular shoulder steroid injection were identified using Boolean coding (e.g. and, or, not) within the database. For comparison purposes, a control group of patients who did not undergo a postoperative injection during the study period was created and matched in a 4:1 fashion by age, gender, procedure (hemi or TSA), and postoperative time period to the injection study group.
Study outcomes
Study and control groups were queried for infection within six months after the date of injection using ICD-9 and CPT coding for prosthetic shoulder joint infection.
Statistical analysis
Demographics and comorbidities of the study and control groups were recorded as provided by the database output. The incidence of infectious complications was compared between groups using a multivariable binomial logistic regression analysis which included the following variables: arthroplasty procedure type (hemi versus total), age, sex, race, obesity, morbid obesity, tobacco use, alcohol use, diabetes mellitus, hyperlipidemia, hypertension, peripheral vascular disease, congestive heart failure, coronary artery disease, chronic kidney disease, chronic lung disease, chronic liver disease, thyroid disease, depression, and hemodialysis.
Results
Patient demographics for study and control groups.
Discussion
The present study's findings suggest an association between intra-articular corticosteroid injection into a pre-existing shoulder arthroplasty and infectious complications compared to matched controls who did not receive injections. These findings support a largely previously unfounded concern that corticosteroid injections into pre-existing arthroplasties should be avoided due to a heightened risk of infection. Although causality cannot be established, the association between these two factors remains significant and warrants awareness by surgeons treating patients with continued pain following shoulder arthroplasty.
While corticosteroid injections have long been an accepted component of the conservative treatment for osteoarthritis of the knee, hip, and shoulder, increasing literature has detailed the increased risk of postoperative infection the closer such injections are performed to surgery preoperatively.6–8 For instance, Werner et al. 8 utilized a national database and demonstrated a significantly increased risk of postoperative infection in patients who underwent preoperative corticosteroid injections within three months of shoulder arthroplasty. This is in agreement with the broader hip and knee arthroplasty literature whose authors also suggest waiting at least three months following injections to perform arthroplasty procedures.6,7 While there is general agreement on the safety and efficacy of preoperative injections prior to total joint arthroplasty, there is far less literature and consensus on similar injections into pre-existing total joint arthroplasties for continued pain postoperatively. Despite the inherent risks of introducing a bacterial infection into a pre-existing arthroplasty from an injection, we identified nearly 1000 patients with documented corticosteroid injections into pre-existing shoulder arthroplasties over a 10-year period. Similarly, a recent study by Mills et al. 9 performed an institutional retrospective review and identified nearly 2000 corticosteroid injections into pre-existing total knee arthroplasties over a six-year period. Thus, this practice may be more common than previously thought despite a scarcity of literature on the safety and efficacy of such practices. Limited total knee arthroplasty literature has suggested similar findings of the current study. A recent study by Mills et al. 9 reviewed 736 patients who received an intra-articular corticosteroid injection into a pre-existing total knee arthroplasty, and reported three prosthetic joint infections yielding an infection rate of 0.16% within three months of injection. While the group did not compare this cohort to a matched control group, the authors concluded that injections into pre-existing total knee arthroplasties should be used with extreme caution due to the risk of infection.
In the broader shoulder literature, Kew et al. 10 examined the risk of infection in 3964 patients who underwent corticosteroid injections following arthroscopic rotator cuff repair, debridement, or subacromial decompression compared to controls that did not receive injections. The group reported a significantly higher risk of infection in patients receiving injections within one month of surgery. 10 The present study adds to a growing body of literature linking the use of corticosteroids to an increased risk of infection. The mechanism behind this association is likely multifactorial and beyond the scope of the current study. However, we believe it may be due to one of the following: (1) The etiology of the pain may itself be an indolent infection, and the injection is given in the presence of a low-grade infection that disrupts the body's immune response to fight the infection leading to more systemic symptoms. (2) The injection introduces a bacterial contamination, resulting in prosthetic joint infection. Regardless of the exact mechanism, patients and practitioners should be made aware that intra-articular corticosteroid injections into pre-existing shoulder arthroplasties are associated with nearly double the risk of infectious complications compared to patients who do not receive them. There are several important limitations that must be noted in the current study, many of which are inherent to studies that utilize insurance databases to study infectious complications of shoulder arthroplasty.11–14 First, the results and conclusions of the study rely on the quality and accuracy of data within the database, and miscoding and noncoding are possible sources of error. Second, we are unable to obtain further detail on the clinical workup of patients with shoulder pain who underwent injections and are also unable to determine the concentration of the steroid used. Third, in order to increase the likelihood that the infectious complication was related to the injection, we sought to capture infections within 90 days of the injection. However, this might exclude infections related to the injection that were reported more than 90 days following the injection and could falsely lower the risk of these injections.
Conclusions
There appears to be a significant association between intra-articular shoulder corticosteroid injections in patients with pre-existing shoulder arthroplasties and prosthetic joint infection compared to matched controls who did not receive injections.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Ethical Review and Patient Consent
Deemed exempt by UVA-HSR IRB. No patient consent needed or obtained, existing database of de-identified patient data.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
