Introduction

Knee osteoarthritis (KOA) is a degenerative disease involving the intra-articular tibiofemoral and patellofemoral cartilage1. It causes chronic pain, functional limitation, and emotional disturbance which may lead to disability and poor quality of life2. A knee arthroplasty (KA) is a reliable and suitable surgical procedure for end-stage OA patients to relieve pain, to recover function and to improve health related quality of life3. With the creative designs and improved skills, we are now entering a new phase in which partial osteoarthritic changes can be treated with partial resurfacing prosthetic solutions such as unicompartmental, bi-unicompartmental or patellofemoral arthroplasty4. However, no matter what type of KA is performed, this surgery has commonly been associated with severe postoperative pain as well as postoperative nausea and vomiting (PONV). This can result in immense discomfort, emotional distress, low satisfaction, and further leads to poor surgical outcomes and delayed postoperative recovery5,6. Therefore, effective management of postoperative pain is essential for early rehabilitation and better functional outcomes. Various procedures including continuous epidural anesthesia, local infiltration anesthesia, peripheral nerve block, and patient controlled analgesia (PCA) are taken to control pain.

As an orthopedic surgeon, we place great concerns on the multimodal cocktail periarticular or intra-articular injection (MCPI) strategy after KA for analgesia. Not only this treatment is proven to be a safe and cost-effectiveness measure7,8, but also this intervention is easy to perform. Although the gold standard for the cocktail formula has not yet been set up, the components always contains local anesthetics (ropivacaine and bupivacaine), epinephrine, steroids, opioids and nonsteroidal anti-inflammatory drugs (NSAIDs)9,10,11. However, whether those compositions are all necessary is still to be investigated considering the efficacy and safety.

Steroids have been extensively used in various perioperative settings due to their potent anti-inflammatory and antiemetic effects12,13. Clinical trials have indicated that the addition of steroids to MCPI might decrease edema and blood loss and achieve permanently better range of motion (ROM) due to reduced local inflammatory response following surgical trauma14,15. Some studies have advocated the use of steroids for the benefits of postoperative pain relief and reduced risk of POVN16,17,18, while others have found no significant decrease in pain and PONV with administration of steroids19,20. Moreover, periarticular steroids use may arouse concerns for possible postoperative infection and patellar tendon rupture21. To sum up, it remains inconclusive whether the addition of steroids in the MCPI is necessary.

Several meta-analyses have been launched to address the efficacy and safety of steroids application in knee joint arthroplasty. The major drawback of these studies is not distinguishing local usage of steroid from systemic application12,22,23,24,25,26,27,28, which may hamper the reliability of the conclusion. Only three studies focused on the periarticular application of steroids in KA26,27,29. In Tran’s study, only five studies were included and there was no overwhelming data to suggest the addition of steroids to MPCPI improves postoperative total knee arthroplasty (TKA) pain26. In Zhao’s study, only six RCTs met the inclusion criteria and it was found that the addition of steroids to MCPI could improve the analgesic effect and was proved to be highly safe in patients undergoing TKA, but it couldn’t increase the postoperative ROM of knee joint27. Though eight studies were evaluated in Xing’s meta-analysis, it didn’t assess the analgesia effect of the steroid in patients underwent joint arthroplasty, the results showed that intraarticular steroid injections may lead to increased deep infection rates of subsequent joint arthroplasty29. Moreover, all of these studies were confined to TKA with a limited number of studies included. Actually, for the patients suffering KA, no matter whether partial knee arthroplasty (PKA) or TKA are performed, the mechanisms of pain, stress and other reactions that caused by KA might be similar30,31,32. Consequently, the aim of this study is to systematically summarize the available scientific literatures on all kinds of KA to evaluate the efficacy and safety the addition of steroids addition to MCPI by conducting a quantitative meta-analysis including randomized controlled trials (RCTs).

Results

Search results

A total of 918 potentially relevant articles wereidentified from the databases, including 315 duplicated articles; 698 studies were excluded after screening titles and abstracts. 34 full-text articles were assessed for eligibility, of which 24 were excluded after assessment of the full-text articles. Eventually, ten14,21,33,34,35,36,37,38,39,40 articles that fulfilled the inclusion criteria were identified for synthetic evaluation. The flow diagram is presented in Fig. 1.

Figure 1
figure 1

Flow chart of studyselection.

Full size image

Study characteristics

Demographic characteristics, the details about the included studies are summarized in Table 1. Ten RCTs were performed in 7 countries, involving 820 participants. Eight21,33,34,35,36,37,38,39 of the ten trials performed TKA surgery, while the other two21,40 performed UKA. Both experimental and control groups were tightly matched in sample size, age, sex, and body mass index (BMI). Experimental groups received single-dose postoperative intraarticular injection or periarticular infiltration of steroids, while control groups received placebo or none. Four studies33,34,35,38 performed spinal anesthesia (SA), two36,37 opted for general anesthesia (GA), one21 received endotracheal anaesthesia (EA), and the other three14,39,40

Table 1 Characteristics of ten included RCTs.
Full size table

received SA or GA. PCA with opioid, was given to all participates for concomitant pain management. The follow-up period ranged from 1 week to 2 years.

Risk of bias assessment

Cochrane Collaboration tool was used to assess the RCTs (Table 2). Double blinding was provided in all RCTs. The randomization algorithm was mentioned in 6 trials14,21,33,37,38,39, including random number generator, block randomization, random numbers, and randomization table. Four trials34,35,36,40 recorded the allocation concealment information by sealed envelopes. Other assessments, such as blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, and selective reports, were all in good performance.

Table 2 Methodological quality assessment according to the Cochrane Collaboration’s Risk of Bias tool study.
Full size table

Visual analogue pain score (VAS)

All studies recorded pain score, but only 4 trials21,33,37,38 provided detailed data as means and standard deviations (SDs). Two studies33,38 reported both VAS at rest (RVAS) and during activity (AVAS). AVAS could not be collected for analysis due to unsuitable format of the data. Therefore, if not specifically mentioned, VAS at different time points were all considered as RVAS. Finally, postoperative day (POD) 1, 3 and postoperative week (POW) 2 were eligible for meta-analysis. The result showed that VAS was significantly lower in the steroids group than the control group on POD1 (MD = −1.52, 95% CI −2.94 to −0.10, P = 0.04). However, no significant difference was found between the two groups on POD3 and POW 2. Random-effect models were used to perform meta-analysis for POD1 (I2 = 82%, P = 0.004) and POD3 (I2 = 76%, P = 0.04) as the substantial heterogeneity was high (Fig. 2).

Figure 2
figure 2

Forest plot diagram showing VAS at POD1 (a), POD3 (b), POW2 (c).

Full size image

Range of motion (ROM)

Seven studies14,21,33,35,37,38,40 provided detailed data. Eleven time points were analyzed which included POD1, 2, 3, 4, 5, POW1, 2, 4, 6, and postoperative month (POM) 3, 6, respectively. The results showed that the steroids group had significantly better ROM on the first postoperative 5 days: POD1 (MD = 11.57, 95% CI 9.85 to 13.30, P < 0.00001), POD2 (MD = 9.03, 95% CI 6.67 to 11.38, P < 0.00001), POD3 (MD = 5.73, 95% CI 0.85 to 10.60, P = 0.02), POD4 (MD = 5.53, 95% CI 0.68 to 10.38, P = 0.03), and POD5 (MD = 5.90, 95% CI 0.87 to 10.93, P = 0.02). However, the difference was not significant after POW1. The substantial heterogeneity was high at POW4 (I2 = 94%, P < 0.00001), POM3 (I2 = 93%, P < 0.00001), and POM6 (I2 = 81%, P = 0.006). Therefore, the random-effect models were used for these outcomes. Fixed-effect models were applied at other time points (Fig. 3).

Figure 3
figure 3

Forest plot diagram showing ROM at various time points after knee joint arthroplasty.

Full size image

Morphine consumption

Three studies14,21,37 provided detailed data for postoperative morphine consumption and the total amount used during hospitalization was accumulated for analysis. A random-effect model was used due to significant heterogeneity (I2 = 68%, P = 0.02). The overall effect showed that morphine consumption for the steroid groups were significantly less than that for the control groups (MD = −7.94, 95% CI −14.35 to −1.53, P = 0.02; Fig. 4a).

Figure 4
figure 4

Forest plot diagram showing postoperative morphine consumption (a), CRP at POD3 (b), the interval to perform SLR postoperatively (c), LOS (d), KSS knee score (e) and function score (f) at POW6.

Full size image

C-reactive protein (CRP)

Three studies37,38,40 provided detailed data for CRP level, but only one time point included two studies that were eligible for meta-analysis. Pooled results showed that the CRP level on POD3 in the steroid groups was significantly lower than that in control groups (WMD = −4.82, 95% CI −7.41 to −2.23, P = 0.0003). Random-effect models were used for this study (I2 = 84%, P = 0.01; Fig. 4b).

Straight leg raising (SLR)

The interval required to perform a SLR was collected from three studies14,35,37 involving 282 knees. The result indicated that the patients in the steroid groups could perform a SLR significantly earlier than those in the control groups (MD = −0.65, 95% CI −0.86 to −0.44, P < 0.00001). Statistical heterogeneity was not found in SLR (I2 = 43%, P = 0.17). A fixed-effects model was performed (Fig. 4c).

Length of stay (LOS)

Two studies21,40 provided detailed data with regard to length of stay in hospital. The results indicated that the patients in the steroid groups had significantly shorter LOS than that of the control groups (MD = −0.98, 95% CI −1.25 to −0.71, P < 0.00001). The fixed-effects model was selected as no significant heterogeneity was found (I2 = 0%, P = 0.78; Fig. 4d).

Knee society knee and function scores (KSS)

Two studies21,35 provided detailed data of KSS knee scores and function scores, respectively. Only 1 time point was eligible for meta-analysis. During the follow-up evaluation POM6, the patients in the steroid groups did not achieve higher KSS knee scores than the control groups (MD = 0.54, 95% CI −0.64 to 1.72, P = 0.37, Fig. 4e). A fixed-effects model was performed because of low heterogeneity (I2 = 0%, P = 0.55). Likewise, the KSS function scores were also similar between the two groups (MD = −2.47, 95% CI −11.62 to 6.68, P = 0.60). A random-effect model was used due to significant heterogeneity (I2 = 56%, P = 0.13; Fig. 4f).

Complications

All ten included studies reported the incidence of various complications such as PONV, infection, and wound oozing. There was no significant difference in total incidence of complications between the steroid and control groups (OR = −1.07, 95% CI 0.63 to 1.82, P = 0.81; Fig. 5a).

Figure 5
figure 5

Forest plot diagram showing total incidence of complications (a), incidence of infection (b) and PONV (c) after knee joint arthroplasty.

Full size image

There were only two types of complications mentioned in at least two papers. Three studies14,21,39 involving 225 patients reported the occurrence of postoperative infection, and the steroid groups did not have an increased number of patients with infection (OR = 1.29, 95% CI 0.31 to 5.38, P = 0.72; Fig. 5b). Four trials33,34,35,37 involving 352 patients were included that assessed the occurrence of PONV. According to the outcomes, the incidence of PONV in the two groups was also similar (OR = 0.78, 95% CI 0.46 to 1.34, P = 0.37; Fig. 5c). The fixed-effects models were applied in these 4 cases due to the low heterogeneity (I2 = 0%). All the results of this meta-analysis are presented in Table 3.

Table 3 Results of Meta-Analysis.
Full size table

Discussion

This meta-analysis assessed the efficacy and safety of the addition of steroids to MCPI for pain management and physical dysfunction of TKA and UKA patients. The current evidence shows that periarticular steroids injection is certainly effective in pain relief and improved ROM during the early stage after surgery, and is able to reduce morphine consumption, decrease CRP level, achieve earlier SLR, and shorten LOS, but the analgesic effect and functional outcomes were not significant compared to placebo groups during the late postoperative period. The most encouraging result is that the periarticular steroids injection did not increase the risk of postoperative infection, PONV, or other complications. Therefore, we conclude that steroids are more likely to play a positive role during the early postoperative period.

The forest plots indicated that the addition of steroids could further improve the analgesic efficacy of MCPI during the hyper-acute phase as we observed VAS score significantly decreased on the POD1 in the steroids group. The suppression of the inflammatory response by steroids application can be explained by the signaling pathways that suppress the cytoactivity of immune cells by inhibiting the production of phospholipase A2 (PLA2) and hence decreasing the production of inflammatory cytokines and chemokines41. Previous studies had detected lowered interleukin (IL)-6 and serum CRP level in cases treated with systemic and local steroids37,40. Our results also found decreased serum CRP level on the third day after surgery, proving the anti-inflammatory effects of steroids. Consistently, with the relief of postoperative pain, morphine consumption during the hospitalization was significantly decreased, and the time required to perform a SLR and LOS were also shortened in the steroids group. However, analgesic effect in the steroids group was not significant at POD3 and POW2, which indicated that MCPI with steroids could not provide a sustained analgesic effect. Steroids might play a pivotal role, but the effect of local anesthetics would disappear gradually over time.

In this meta-analysis, significant improvement in ROM was observed during the early postoperative period in those patients who received steroid injections during surgery, but the effects subsequently vanished. We inferred that the analgesic effect is sufficient to suppress the postoperative pain caused by surgical trauma, which is consistent with achievement of early SLR and shorter LOS as verified by our results. The difference in ROM between steroids and the control groups was insignificant at the POW1, suggesting that the addition of steroidsto MCPI was unable to increase the ROM, which might be explained by its inability to eliminate postoperative fibrosis and scarring within their respective time frame. Likewise, we observed no improvement of KSS knee or functional scores, which is a comprehensive system rating both the knee prosthesis function and patients’ functional abilities after TKA, at POW6. Three included RCTs reported the KSS scores. No group differences were noted in studies conducted by Christian et al. and Kown et al.21,35. Yue et al. observed significant improvement in postoperative KSSscores in the steroids group as compared with that in control groupat POM1 and POM3. However, at POM6, the KSS scores showed no significant difference between these two groups36. All these results indicate that long term effects of steroids application on functional outcomes are still uncertain.

For safety concerns, many surgeons remain leery of using steroids that may increase the risk of catastrophic complications such as infection and patellar tendon rupture42,43. A series of complications, including postoperative infection, wound oozing, PONV, pruritis, and transient peroneal nerve palsy were mentioned in the included studies. Our results indicate there were no difference in the total incidence of complications between the two groups. Steroids addition to MCPI would not increase the incidence of postoperative infection and PONV. Furthermore, there were no differences in the incidences of complicationsbetweenthe steroids and placebo groups. No case of tendon rupture was reported in the included RCTs.

Limitations of the present meta-analysis should also be acknowledged. Firstly, the sample size is relatively small and some of the indexes were objectively judged which may reduce the reliability of the study. The second limitation is the lack of consensus on the type and doses of the drugs applied in MCPI which also undermine the consolidation of the analysis. Third, some data were missing or could not be extracted. Some of the results appeared heterogeneous, and couldn’t be eliminated by sensitivity or subgroup analyses. We included these high-quality studies and applied random-effect models for meta-analysis, which may mildly influence the reliability of the results. Finally, the reported side effect is scarce and duration of follow-up is relatively short, which might lead to underestimation of complications and uncertainty of long-term efficacy44. Despite the above limitations, this is the first meta-analysis that included RCTs to evaluate the efficiency and safety of periarticular steroids addition to MCPI in TKA and UKA. Large well-designed RCTs are still needed to validate this research moving forward.

In conclusion, this meta-analysis of RCTs suggested that the administration of periarticular steroids injection with MCPI appears to effectively relieve postoperative pain, improve functional outcomes, reduce morphine consumption, and decrease inflammatory reaction without causing a higher incidence of complications in the early postoperative period.

Methods

Search strategy

This quantitative meta-analysis was in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines45. Two researchers searched the relevant studies independently including MEDLINE/PubMed database, the Cochrane Central Register of Controlled Trials (CENTRAL), and EMBASE databases from inception to April, 2018 for relevant RCTs that compared postoperative periarticular injection of steroids with placebo in the knee arthroplasty. Search terms included “knee”, “arthroplasty”, “replacement”, “steroids”, “corticosteroids”, “cortisol”, “cortisone”, “dexamethasone”, “glucocorticoid”, “betamethasone”, “hydrocortisone”, “prednisone”, “prednisolone”, “ethylprednisolone”, “riamcinolone”, and “adrenal cortex hormone” (supplymentary files). Search terms were combined using the Boolean operators ‘AND’ or ‘OR’. No restrictions were imposed, and reference lists of retrieved articles and reviews were also searched. A third reviewer acted as a judge if there was any disagreement.

Inclusion and exclusion criteria

Only RCTs comparing the clinical efficacy between postoperative periarticular or intraarticular injection of steroids with placebo among adults of any sex undergoing primary KA (including TKA, PKA, UKA, et al.) for any indication (osteoarthritis, rheumatoid arthritis, osteonecrosis or acute trauma) were included in our study. Studies that reported at least one outcome were included. The Primary outcomes being VAS and ROM. With secondary outcomes being morphine consumption, CRP, SLR, LOS, and KSS knee and function scores. Studies were excluded from present meta-analysis in view of incomplete data, letters, comments, editorials, case reports, conference abstracts, or review articles.

Data extraction

The relevant data was extracted from included studies by using a standard data extraction. Two researchers used this form to collect the information from studies independently. The main characteristics included the following items: first author’s name, publication year, country, sample size, mean age, sex ratio, BMI, steroids intervention (dose and type), anesthetic techniques, perioperative analgesia intervention, and duration of follow-up. Other relevant data were extracted from included studies. Differences were resolved by consensus. If a study reported the outcomes of multiple doses of steroids for treatment, only data of the low dose group was extracted for analysis. The first main index was VAS score that has 11 pain levels (0 = no pain, 10 = extreme pain). All other types of VAS were converted to this 11 levels score according to ratio. The second index was ROM, composed of both flexion and extension angles of the knee. The flexion angle was used to represent ROM if flexion and extension angles cannot be combined. The third index contained the postoperative morphine consumption. The total amount of morphine consumption during hospitalization was accumulated for analysis. The other indexes included CRP, SLR, LOS, KSS knee and function scores, and complications. CRP is an acute-phase protein synthesized by the liver which reflects the postoperative inflammation46. All the data on CRP were converted to the unit of mg/dl. SLR is a test-conducted post knee surgery to gauge how high a patient is able to elevate his/her leg off of an exam table, reflecting pain control as well as muscle strength recovery26. KSS, composed of knee score and function score, is considered as an objective scoring system to rate the knee and patient’s functional abilities such as walking and stair climbing before and after TKA47. The total incidence of complications after surgery was calculated. If any special type of complication was mentioned more than twice, its incidence was pooled for subgroup analysis48. For missing or incomplete data, we extracted information from diagrams or contacted the corresponding authors to ensure that the information was integrated. Otherwise, we estimated the data wherever possible using the recommendations in the Cochrane Handbook49. Formats that were not suitable included studies reporting means with interquartile ranges, suggesting the data were non-normally distributed making conversion to means and SDs controversial49. Furthermore, data only reporting means without SDs, standard errors (SEs) or confidence intervals (CIs) could not be extracted considering that one of these is required with the mean for meta-analysis.

Quality assessment

The risk of bias of the included studies was assessed using the Cochrane Handbook for Systematic Reviews of Interventions version 5.1.049, based on the following items: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, and selective reporting. The assessment items were either categorized as low risk of bias, high risk of bias, or unclear risk of bias. Discrepancies between assessments were resolved by discussion with a third reviewer when necessary.

Statistical analysis

For continuous data, like VAS scores, ROM and so on, we calculated mean differences (MDs) and 95% CIs if outcomes were measured in the same way between studies, or we used the standardized mean differences (SMDs). Continuous data reported as means and ranges were transformed into means and SDs using Hozo’s formula50. Dichotomous data such as incidence of complications were expressed as odds ratios (ORs) with 95% CIs indicating the effect on intervention. Subgroup analysis was performed for different time points. The Q and chi-squared test were used to assess statistical heterogeneity with the value of P and I2. If heterogeneity was low (P > 0.1, I2 < 50%), a fixed-effects model was used. If heterogeneity was significant (P < 0.1, I2 > 50%), a random-effects model was used. When possible, sensitivity and subgroup analyses were conducted to find the source of the heterogeneity. Publication bias was visually examined by funnel plots. We used Review Manager 5.2 software (Rev Man 5.2, The Cochrane Collaboration, Oxford, UK) and STATA, version 12.0 (Stata Corp LP, College Station, TX) to perform statistical analyses. P < 0.05 was considered statistically significant.