Low-Dose Perioperative Corticosteroids Can Be Administered Without Additional Morbidity in Patients Undergoing Bilateral Total Knee Replacement: A Retrospective Follow-up Study of a Randomized Controlled Trial

Abstract

Background: Short-term benefits of perioperative corticosteroid injections (CSIs) for bilateral total knee replacement (BTKR) include suppressed inflammation, improved knee motion, and reduced pain. Very little is known about the long-term benefits, complications, and safety of corticosteroids administered in the perioperative period. Purpose: We sought to compare 3-year follow-up outcomes of BTKR patients who received perioperative CSI with those who received placebo. We hypothesized that there would be no statistically significant differences in functional outcomes or adverse events based on whether or not CSIs were administered in the perioperative period. Methods: We conducted a retrospective review of chart and registry data of BTKR patients from a prior randomized controlled trial to compare outcomes in patients who received hydrocortisone vs placebo injections after BTKR (ClinicalTrials.gov: NCT01399268 and NCT01815918). Outcomes were compared at 6 and 12 weeks and at 1, 2, and 3 years. The Knee Injury and Osteoarthritis Outcome Scores (KOOS) and Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) were used to evaluate clinical outcomes. Cochran-Mantel-Haenszel tests were used to compare the risk of complications between treatments after adjustment for trial. When possible, summary relative risk estimates were calculated using the Mantel-Haenszel method. Results: No BTKR patients in the treatment group developed an infection. The risk of complications did not increase in patients who received CSI compared with those who received placebo. Patients in the CSI group experienced greater reductions in pain and stiffness, though these results were not statistically significant. There were no statistically significant differences in the KOOS-Symptoms, KOOS-Activities of Daily Living, KOOS-Sports, KOOS-Quality of Life, or WOMAC Function scores. Conclusions: Low-dose corticosteroids can be administered in selected patients who undergo BTKR without increasing the risk of adverse events. At 3-year follow-up, administration of low-dose corticosteroids did not result in superior clinical outcomes scores when compared with placebo.

Introduction

Bilateral total knee replacement (BTKR) is associated with a significant systemic inflammatory response, which has been quantified in various studies as a rise in serum inflammatory cytokine levels, such as interleukin-6 (IL-6), tumor necrosis factor-α, C-reactive protein, and leukocyte count [6,7,12]. The inflammatory cascade is triggered by several mechanisms related to the BTKR procedure, including surgical trauma; fat, bone, and cement emboli; and limb ischemia when a tourniquet is used [12]. Given this extensive inflammatory response, it is important to understand how best to mitigate these effects, as significant inflammation can lead to increased pain severity, malaise, and delays in early rehabilitation [7].

Serial administration of low-dose corticosteroids over the first 24 perioperative hours can suppress postoperative cytokine release after BTKR [7], and it is associated with improved knee range of motion (ROM), reduced incidence of postoperative fever, lower pain scores, and reduced analgesic requirement [6,7]. Perioperative corticosteroids also reduce postoperative nausea and vomiting [10]. In unilateral total knee replacement (TKR), this dose regimen has been shown to reduce the rise in serum prothrombin fragment (PF1.2), a marker of thrombin generation, suggesting that perioperative corticosteroids may also blunt the procoagulable state postoperatively and thus diminish the risk for thromboembolism [11].

However, while corticosteroid injections (CSIs) have demonstrated benefits, they have been associated with adverse outcomes. The likelihood of adverse events after corticosteroid administration is related to the dose and duration of therapy [9]. When administered chronically, corticosteroids are associated with several adverse effects, such as osteoporosis, opportunistic infection, avascular necrosis, diabetes, and glaucoma. Short courses have been generally well-tolerated and rarely associated with osteonecrosis. However, other recent data have suggested that intra-articular injections administered 1 to 6 months before TKR have been linked to an elevated risk of periprosthetic infection [1].

There is continued debate surrounding the risks and benefits of perioperative administration of corticosteroids in lower extremity arthroplasty. Specifically, concerns have been raised about the aforementioned associations as well as the nonspecific therapeutic effects of corticosteroids [15]. Very little is known about the long-term benefits, complications, and safety of corticosteroids administered in the perioperative period. The purpose of the current study was to compare the outcomes over 3 years of follow-up of BTKR patients who received perioperative CSI or placebo. The authors hypothesized that there would be no statistically significant differences in functional outcomes or adverse events based on whether or not CSIs were administered in the perioperative period.

Methods

Institutional review board approval was received prior to initiating this trial. Between February 2009 and March 2012, patients undergoing TKR were assessed for eligibility in 2 randomized controlled trials (RCTs) at an urban, specialty orthopedic surgery hospital (ClinicalTrials.gov Identifiers: NCT01399268 and NCT01815918). Protocols and results of these studies have been published previously [4,5,7,11]. In brief, inclusion criteria for these trials were adult TKR patients aged 50 to 90 years. Patients were excluded if they were on preoperative corticosteroids or required stress-dose corticosteroids, had a history of intolerance or adverse reaction attributable to corticosteroids, had diabetes, or were active smokers. In addition, exclusion criteria for BTKR included age greater than 80 years, American Society of Anesthesiologists class of 3 or higher, active ischemic heart disease, left ventricular ejection fraction of less than 50%, pulmonary disease, body mass index (BMI) greater than 50 kg/m², serum creatinine over 1.6 mg/dL, chronic hepatic disease, stroke history, or prior lower extremity vascular surgery.

Enrolled patients were randomized to receive either 3 doses of 100 mg intravenous (IV) hydrocortisone, each 8 hours apart, or placebo (IV normal saline) at the same time points. The first dose was given 2 hours prior to surgery. All study personnel except the pharmacy were blinded. Patients received identical surgical and postsurgical care by a single surgeon (T.P.S.). Combined spinal/epidural anesthesia and bilateral femoral nerve blocks were given. Tourniquets were inflated prior to incision and deflated prior to medial parapatellar arthrotomy closure. An extramedullary tibial cutting guide and an intramedullary femoral cutting guide were used. A deep drain was placed at closure and removed routinely on the first postoperative day. Cemented posterior stabilized TKRs were implanted in all cases. Warfarin was used for venous thromboembolic prophylaxis.

Patient charts and institutional registry data were reviewed by a trained research coordinator from the health care research institute (K.G.F.) at 6 weeks, 12 weeks, 1 year, 2 years, and 3 years postoperatively. Patient age at surgery, BMI at surgery, sex, and knee ROM were recorded. Baseline and 2-year Knee Injury and Osteoarthritis Outcome Score (KOOS) and Western Ontario and McMaster Universities Arthritis Index (WOMAC) scores were recorded. All complications and reoperations were recorded throughout the study period.

Statistical Analysis

Based on consultation with a biostatistician, statistical analyses were stratified by trial, with NCT01399268 referred to as “Trial A” and NCT01815918 as “Trial B.” No a priori power analysis was performed because this study was undertaken as a retrospective review of all eligible patients. Continuous variables were summarized as means and standard deviations. Age, BMI, and patient-reported outcome scores were compared between treatment groups, stratified by trial, using 2-sample t tests. Postoperative longitudinal ROM measurements were compared between treatment groups, stratified by trial, using regression based on a generalized estimating equations approach to allow for adjustment for baseline ROM and to account for the correlation between repeated measurements on a given patient. Binary variables were presented as frequencies with percentages and compared between treatment groups, stratified by trial, using χ² or Fisher exact test, as appropriate.

The risk of complications and reoperations was compared between treatment groups after adjustment for trial using Cochran-Mantel-Haenszel tests. When possible, summary relative risk (RR) estimates were calculated using the Mantel-Haenszel method. The cumulative proportion of patients with complications and reoperations across trials were summarized as point estimates with 95% Wilson score confidence intervals (CIs).

All statistical hypothesis tests were 2-sided with P < .05 considered statistically significant. Statistical analyses were performed with SAS Version 9.3 (SAS Institute, Cary, North Carolina) and the metafor package implemented in R software version 3.2.4 (R Foundation for Statistical Computing, Vienna, Austria).

Results

Sixty-three BTKR patients (31 placebo, 32 treatment) were included from the index trials (Table 1). Baseline patient characteristics were not different between groups. Fifty-one (81.0%) patients provided 2-year patient reported outcome measures (Trial A: 11 treatment, 12 placebo; Trial B: 17 treatment, 11 placebo).

Table 1.

Patient demographics.

	Trial A			Trial B
	Placebo	Treatment	P value	Treatment	Placebo	P value
	(n = 17)	(n = 17)		(n = 15)	(n = 14)
Age, y
Mean	68.8	68.0	.733	68.4	64.7	.201
SD	5.0	8.2		7.5	7.3
Minimum	59	48		54	52
Maximum	76	80		80	77
Body mass index, kg/m2
Mean	31.8	29.2	.266	28.5	28.9	.809
SD	5.0	8.2		4.8	5.5
Minimum	23.8	21.0		20.5	21.6
Maximum	39.0	55.8		38.3	39.5
Female
Count	11	10	.724	8	8	.837
%	64.7	58.8		53.3	57.1

There were no differences in clinical and functional outcomes as assessed by KOOS or WOMAC subscale scores at baseline or at 2 years (Table 2). Mean differences in subscale scores between baseline and 2 years were also not different. For Trial A, the adjusted differences (treatment minus placebo) in knee extension and flexion were 0° (95% CI = −2° to 2°; P = .941) and 3° (95% CI = −2° to 8°; P = .230), respectively. For Trial B, the adjusted differences in knee extension and flexion were 0° (95% CI = −3° to 2°; P = .787) and −4° (95% CI = −15° to 7°; P = .474).

Table 2.

Comparison of patient-reported outcomes in Trial A and Trial B among patients who were randomized to receive corticosteroid injections or placebo.

	Trial A						Trial B
	n	Placebo	n	Treatment	Difference in means (95% CI)	P value	n	Placebo	n	Treatment	Difference in means (95% CI)	P value
KOOS Pain Score
Baseline	16	45 ± 15	12	53 ± 22	8 (−6 to 23)	.246	3	49 ± 11	9	47 ± 13	−2 (−21 to 17)	.804
3-year	11	96 ± 6	11	93 ± 11	−3 (−11 to 5)	.495	2	86 ± 20	9	85 ± 17	−1 (−31 to 30)	.961
Delta	11	49 ± 15	11	37 ± 21	−12 (−28 to 4)	.124	2	32 ± 10	9	39 ± 23	7 (−33 to 46)	.715
KOOS-Symptoms Score
Baseline	17	48 ± 16	12	55 ± 21	7 (−7 to 20)	.337	3	57 ± 29	9	53 ± 11	−5 (−28 to 19)	.674
3-year	9	91 ± 10	9	79 ± 19	−12 (−27 to 4)	.125	1	54 ± 0	5	60 ± 7	N/A	N/A
Delta	9	51 ± 11	9	28 ± 36	−24 (−52 to 5)	.090	1	−4 ± 0	5	5 ± 16	N/A	N/A
KOOS-ADL Score
Baseline	16	46 ± 16	12	59 ± 19	13 (−1 to 27)	.064	3	53 ± 11	9	56 ± 12	3 (−14 to 21)	.687
3-year	12	92 ± 11	11	92 ± 10	0 (−9 to 9)	.992	3	97 ± 3	8	86 ± 17	−11 (−25 to 4)	.131
Delta	12	42 ± 16	11	31 ± 17	−12 (−26 to 3)	.103	3	45 ± 13	8	32 ± 17	−13 (−38 to 19)	.267
KOOS-Sports Score
Baseline	15	20 ± 24	9	20 ± 13	0 (−18 to 18)	.970	3	17 ± 8	9	28 ± 29	11 (−28 to 50)	.539
3-year	8	77 ± 20	7	60 ± 29	−17 (−44 to 10)	.201	1	50 ± 0	3	65 ± 26	N/A	N/A
Delta	7	59 ± 24	6	33 ± 21	−26 (−53 to 2)	.065	1	40 ± 0	3	53 ± 15	N/A	N/A
KOOS-Quality of Life Score
Baseline	17	17 ± 17	12	26 ± 15	9 (−4 to 21)	.171	3	33 ± 16	9	26 ± 16	−8 (31 to 16)	.490
3-year	9	85 ± 13	9	77 ± 15	−8 (−22 to 6)	.264	1	44 ± 0	4	78 ± 27	N/A	N/A
Delta	9	69 ± 15	9	54 ± 19	−16 (−33 to 1)	.062	1	25 ± 0	4	56 ± 14	N/A	N/A
WOMAC Pain Score
Baseline	16	51 ± 16	11	58 ± 23	7 (−9 to 22)	.383	3	53 ± 13	9	51 ± 11	−3 (−20 to 14)	.721
3-year	12	96 ± 6	11	94 ± 10	−2 (−9 to 5)	.593	3	97 ± 6	9	86 ± 17	−10 (−34 to 13)	.341
Delta	12	43 ± 19	10	32 ± 20	−11 (−29 to 6)	.188	3	43 ± 14	9	36 ± 23	−8 (−39 to 24)	.603
WOMAC Stiffness Score
Baseline	17	40 ± 21	12	47 ± 21	7 (−9 to 23)	.369	3	46 ± 26	9	47 ± 10	1 (−59 to 62)	.936
3-year	12	87 ± 20	11	86 ± 22	0 (−18 to 18)	.991	3	92 ± 14	9	85 ± 14	−7 (−27 to 14)	.468
Delta	12	49 ± 25	11	40 ± 26	−9 (−31 to 13)	.396	3	46 ± 26	9	38 ± 21	−8 (−41 to 24)	.581
WOMAC Function Score
Baseline	16	46 ± 16	12	59 ± 19	13 (−1 to 27)	.064	3	53 ± 11	9	56 ± 12	3 (−14 to 21)	.687
3-year	12	92 ± 11	11	92 ± 10	0 (−9 to 9)	.992	3	97 ± 3	8	86 ± 17	−11 (−25 to 4)	.131
Delta	12	42 ± 16	11	31 ± 17	−12 (−26 to 3)	.103	3	45 ± 13	8	32 ± 17	−13 (−38 to 12)	.267

CI confidence interval, KOOS Knee Injury and Osteoarthritis Outcome Score, Delta 2-year score minus baseline score, ADL Activities of Daily Living, WOMAC Western Ontario and McMaster Universities Arthritis Index.

The frequency of adverse events was not different between the 2 groups. Overall, 24 placebo patients (77% [95% CI = 60–89]) and 23 treated patients (72% [95% CI = 55–84]) had subjective complaints (ie, pain, stiffness, swelling, and mechanical symptoms), complications, or reoperations during the follow-up period. No patient in either group developed venous thromboembolism or a deep infection (0% [0–6]). No patient who received corticosteroids showed signs of superficial infection, wound drainage, or arthrofibrosis (0% [0–11]), whereas 2 patients (6% [2–21]) receiving placebo had superficial infections.

At the 3-year follow-up, pain symptoms were present in 9 patients (28% [16–45]) who received corticosteroids vs 12 patients who received placebo (39% [24–56]). Complaints of knee stiffness occurred in 3 patients who received corticosteroids (9% [3–24]) vs 6 who received placebo (19% [9–36]). Of the patients who received placebo, 2 showed signs of arthrofibrosis (6% [2–21]), and 1 (3% [0–16]) required bilateral manipulation under anesthesia (MUA). Of the placebo patients who complained of knee stiffness, 2 (6% [2–21]) required unilateral arthroscopic debridement with MUA.

Within 6 weeks of surgery, 1 corticosteroid patient (3% [0–16]) had a traumatic quadriceps tendon rupture resulting from a fall and requiring operative repair. Thus, 3 patients who received placebo (10% [3–25]) vs 1 who received corticosteroids (3% [0–16]) required additional procedures. With the numbers available, there was no evidence of a statistically significant difference between treatment and placebo either stratified by or summarized across trials (Figure 1; Table 3).

Fig. 1.

Estimates of treatment effect size (relative risk) are summarized in this forest plot. The point effect estimate for relative risk from each study is presented as a square, and the 95% confidence interval is depicted as a horizontal line. The diamond represents the summary effect estimate for common relative risk, and the diamond’s width represents the CI for the estimate. Estimates to the left of the vertical dashed line favor perioperative steroids and estimates to the right of the dashed vertical line favor placebo. Outcomes without an associated plot had incalculable relative risks because of sample size and/or frequency of the outcome. “A” refers to trial A, and “B” refers to trial B. CI confidence interval, MUA manipulation under anesthesia, DVT deep vein thrombosis, PE pulmonary embolism.

Table 3.

Comparison of adverse events for Trial A and Trial B among patients who were randomized to receive corticosteroid injections or placebo.

	Trial A				Trial B				Common relative risk a (95% CI)	P value b
	Placebo (n = 17)	Treatment (n = 17)	Relative risk (95% CI)	P value	Placebo (n = 14)	Treatment (n = 15)	Relative risk (95% CI)	P value
MUA, n (%)	0 (0%)	0 (0%)	Incalculable	N/A	1 (7%)	0 (0%)	Incalculable	.483	Incalculable	.301
Arthroscopy + MUA, n (%)	1 (6%)	0 (0%)	Incalculable	.999	1 (7%)	0 (0%)	Incalculable	.483	Incalculable	.150
Superficial infection, n (%)	1 (6%)	0 (0%)	Incalculable	.999	1 (7%)	0 (0%)	Incalculable	.483	Incalculable	.150
Deep infection, n (%)	0 (0%)	0 (0%)	Incalculable	N/A	0 (0%)	0 (0%)	Incalculable	N/A	Incalculable	N/A
DVT, n (%)	0 (0%)	0 (0%)	Incalculable	N/A	0 (0%)	0 (0%)	Incalculable	N/A	Incalculable	N/A
PE, n (%)	0 (0%)	0 (0%)	Incalculable	N/A	0 (0%)	0 (0%)	Incalculable	N/A	Incalculable	N/A
Stiffness complaints, n (%)	3 (18%)	1 (6%)	0.33 (0.04-2.89)	.601	3 (21%)	2 (13%)	0.62 (0.12-3.19)	.651	0.48 (0.13-1.75)	.260
Quadriceps weakness, n (%)	6 (35%)	5 (29%)	0.83 (0.31-2.22)	.714	1 (7%)	4 (27%)	3.73 (0.47-29.49)	.330	1.26 (0.53-2.97)	.597
Complaints of mechanical symptoms, n (%)	1 (6%)	1 (6%)	1.00 (0.07-14.71)	.999	1 (7%)	0 (0%)	Incalculable	.483	Incalculable	.546
Complaints of pain symptoms, n (%)	5 (29%)	5 (29%)	1.00 (0.35-2.83)	.999	7 (50%)	4 (27%)	0.53 (0.20-2.69)	.196	0.72 (0.36-1.47)	.372
Arthrofibrosis, n (%)	1 (6%)	0 (0%)	Incalculable	.999	1 (7%)	0 (0%)	Incalculable	.483	Incalculable	.150
Complaints of Swelling, n (%)	12 (71%)	12 (71%)	1.00 (0.65-1.54)	.999	7 (50%)	5 (33%)	0.67 (0.28-1.62)	.363	0.87 (0.58-1.32)	.527
Any complication or complaint, n (%)	13 (76%)	14 (82%)	1.08 (0.76-1.52)	.999	11 (79%)	9 (60%)	0.76 (0.47-1.25)	.427	0.93 (0.70-1.24)	.628

CI confidence interval, MUA manipulation under anesthesia, DVT deep vein thrombosis, PE pulmonary embolism.

^aTreatment vs placebo pooled across trials via the Mantel-Henszel method.

^bCochran-Mantel-Haenszel test.

Discussion

The main findings of the current study were (1) patients who underwent BTKR and received perioperative CSIs did not show signs of deep or superficial infection at midterm follow-up of 3 years; (2) patients who received CSIs displayed lower rates of knee stiffness, arthrofibrosis, and pain vs those who received placebo, though this trend was not statistically significant; and (3) there were no statistically significant differences in patient-reported outcome measures at 3-year follow-up.

There are several limitations in the current study. First, the number of participants was small. In general, the frequency of serious postoperative complications (such as thromboembolism, reoperation, and infection) is low. Therefore, the numbers in our study did not have adequate power to demonstrate differences between groups for the outcomes measured. As such, the results are mostly descriptive, but they have value as pilot data for larger future studies. However, RCTs comparing the administration of CSIs vs other injections for patients who underwent unilateral total knee arthroplasty (TKA) have used similar or lower sample sizes. In addition, the surgical technique, anesthetic protocol, and protocol for hydrocortisone administration were identical in both trials. Thus, when interpreting the data for future hypotheses and trial design, there can be greater confidence in supposing that observed differences between placebo and treatment groups are attributable to perioperative hydrocortisone. Second, strict patient selection criteria for the original trials excluded patients on chronic steroids, smokers, patients with diabetes, and those with multiple comorbidities. Thus, we cannot make conjectures about the safety of perioperative corticosteroids in patient populations differing from the trial cohorts. For example, in one of the previous trials, serum glucose was significantly elevated in the corticosteroid group (101 ± 17 mg/dL vs 128 ± 20 mg/dL, P < .01), but less than the recommend limit of 200 mg/dL [2]. Therefore, the safety of these agents in diabetic patients is a concern. Third, this study was a retrospective review of chart and registry data. Compared with a prospective study, it is possible that not all complications or benefits were captured.

The short-term benefits of low-dose corticosteroids for elective lower extremity arthroplasty have been demonstrated in several RCTs [7]. The 24-hour dosing regimen of IV hydrocortisone used in the treatment groups for the present study has been shown to durably suppress systemic IL-6, reduce the incidence of postoperative fever, improve visual analog scale pain scores, lessen patients’ postoperative analgesic requirements, improve predischarge knee ROM, and reduce urinary desmosine, a marker of lung injury, in patients undergoing BTKR [2]. However, little has been published comparing the long-term outcomes of patients receiving perioperative corticosteroids with the long-term outcomes of those who do not. Furthermore, it remains unknown whether these injections benefit patients who undergo BTKR, after which significant pain is anticipated and early rehabilitation can be challenging.

The current study found that for patients undergoing BTKR, the risk of infection was not increased with a 24-hour dosing regimen of IV hydrocortisone when compared with placebo. Specifically, there were no superficial or deep infections in patients who received CSIs, whereas there were 2 superficial infections in the group that received placebo. Although this was not a statistically significant difference given the infrequent number of occurrences relative to the total sample size, this finding does not support the belief that there is an increased risk of wound complications and infection with corticosteroid use [8], as these injections have been associated with immune suppression [4] and periprosthetic infection [6]. A recent retrospective study of 129 patients who underwent TKA and received a CSI into their replaced knee reported no periprosthetic joint infections within 1 year of the injection [8]. Mills et al [13], in their retrospective case series of 736 patients, demonstrated that the incidence of acute infection into knees with existing TKA was low, at 0.16% per injection. However, these findings oppose a recent nationwide private insurer database study, which found that patients who received a CSI less than 3 months before TKA had a 21% increased risk of periprosthetic joint infection 6 months postoperatively [14]. Similar to this study, there were no differences between the corticosteroid group and their control group (hyaluronic acid injections, which also increased the risk by 55%). These discrepancies may be accounted for by limitations inherent to national databases, which include uncertainty pertaining to corticosteroid dosage, patient selection, and definitions of infection. Furthermore, the results of the current study are derived from an RCT, whereas national database studies cannot account for confounding variables. Another previous RCT of patients with unilateral TKA compared periarticular methylprednisolone injections with injections of a mixture of ropivacaine, morphine, epinephrine, and ketoprofen and demonstrated no differences in surgical site infections and significantly reduced postoperative pain scores at 24 hours in the methylprednisolone group [16]. Although this group investigated periarticular injections and only investigated unilateral TKA, the results corroborate our findings and suggest that CSIs into or around knees with a previous TKA are safe. Given the significant level of anticipated pain and the low risk of subsequent infection, we recommend the use of early perioperative intra-articular or periarticular injections in patients with BTKR.

Complaints of knee stiffness and pain, diagnosis of postoperative arthrofibrosis, and rates of MUA or arthroscopy with MUA were more frequent in the placebo cohorts. Notably, despite these differences, the adjusted differences in knee ROM over the first postoperative year were not different. However, these trends are consistent with previous unilateral TKA literature that has investigated the influence of CSIs on postoperative clinical and functional outcomes. Tsukada et al [16] reported that patients who received periarticular methylprednisolone intraoperatively during unilateral TKA had significantly lower pain scores at 18, 20, and 22 hours postoperatively at rest, but not at other time points. Furthermore, mean pain scores on postoperative day 1 were significantly lower during activity in this group compared with controls. In terms of function, patients who received methylprednisolone also had better mean flexion angles at 1, 2, and 6 days and better mean extension at 1, 2, 3, and 10 days after surgery compared with controls. Based on these data, it appears that perioperative CSIs may confer early benefits that become statistically nonsignificant at later time points, as demonstrated by the current study and the results presented by Tsukada et al. Although not significant in the longer term, these trends toward less pain and better ROM in the early postoperative period may be clinically significant because to prevent stiffness, it is imperative to gain full extension and begin ROM exercises in the early perioperative period.

When examining patient-reported outcome scores, no statistically significant differences were observed between those who received a CSI and those who received placebo. This finding suggests that the propensity for improvement as measured by functional and clinical criteria is not inhibited by the early use of CSIs for pain. A recent meta-analysis of RCTs that sought to evaluate the efficacy of additional local injections of corticosteroids after TKA found that knee ROM and pain were significantly improved on postoperative days 1 and 2; however, no significant differences were found in the knee society score or ROM after postoperative day 2 [2]. In an early study of 76 patients who underwent unilateral TKA and were randomized to receive either periarticular CSIs or an analgesic mixture of bupivacaine, morphine, epinephrine, clonidine, and cefuroxime, there were no statistically significant differences in the Knee Society Score 12 weeks postoperatively [3]. Our study provides longer-term data using a higher risk BTKR patient population and demonstrates that CSIs are safe adjuncts that may reduce early pain and confer early benefits in function, though these benefits likely diminish with time.

In conclusion, low-dose corticosteroids were administered in select patients who underwent BTKR without increasing the risk of adverse events. At 3-year follow-up, administration of low-dose corticosteroids did not result in superior clinical outcome scores when compared with placebo.