The number of total knee arthroplasties (TKAs) performed annually in the United States has doubled during the last decade.1 Although the potential for local infiltration analgesia (LIA)2 and adductor canal block3 techniques await to be fully elucidated,4 in many centres femoral nerve block (FNB) remains a mainstay of multi-modal analgesic management following TKA.5-9 Despite a functioning FNB, however, TKA continues to be associated with moderate-to-severe postoperative pain in the majority of patients.10 To improve pain control, it has been suggested that proximal sciatic nerve block (SNB) performed between the parascacral and mid-thigh regions should be added to FNB in patients undergoing TKA.11-13 Unfortunately, the relevant evidence examining the role of SNB had been limited to observational studies and low-quality randomized controlled trials (RCTs), thus precluding a valid and reliable estimation of effect.14 Recently, however, nine RCTs have been separately published, with seemingly mixed results.15-23 In this updated meta-analysis, we aimed to quantify the analgesic effects of adding proximal single-shot or continuous SNB to single-shot or continuous FNB in the absence of LIA on analgesic consumption and pain scores during the first 48 hr postoperatively in adult patients undergoing unilateral TKA.

Methods

The Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines24 were followed while preparing this systematic quantitative review. Randomized and quasi-randomized controlled trials examining the analgesic effects of adding SNB to FNB following TKA were retrieved and reviewed with an a priori designed but unregistered systematic review protocol.

Literature search

Two authors (F.W.A., R.B.) separately searched the US National Library of Medicine database (MEDLINE), Excerpta Medica database (Embase), Medline In Process, Cochrane Central Controlled Trials Database Register, Cochrane Systematic Reviews Database, and other non-indexed databases of citations. The search was conducted for medical subject headings (MeSH), controlled terms, and text words relevant to the components of the study question of this review. They encompassed sciatic nerve block, knee arthroplasty or replacement, postoperative pain, and analgesia. The terms were used individually and in combinations (Appendix). We also hand-searched the bibliography sections of the manuscripts retrieved to identify any additional studies that met our eligibility criteria. Trials involving human adults (18 yr or older) published from January 195025,26 to March 2015 were considered. We also sought and reviewed related abstracts of meetings of the following medical professional societies: American Society of Regional Anesthesia (2005-2014), American Society of Anesthesiologists (2000-2014), and the European Society of Regional Anesthesia (2006-2014). In addition, the Internet-based registries of clinical trials including www.clinicaltrials.gov, www.ISRCTN.org, https://www.clinicaltrialsregister.eu, and http://www.anzctr.org.au were examined for ongoing relevant trials.

Eligibility criteria

We retrieved complete manuscripts and published abstracts of trials that evaluated the impact of single-shot or catheter-based infusion proximal SNB (i.e., parasacral, Labat, infragluteal, anterior, and mid-femoral approaches) (SNB group) compared to placebo or systemic intravenous analgesia (Control group) on postoperative analgesic outcomes in patients undergoing unilateral TKA. Only RCTs in which single-shot or continuous FNB was administered to both groups were included. RCTs examining blocks performed for surgical anesthesia and/or postoperative analgesia were also eligible. We excluded trials that (1) did not assess analgesic outcome measures (i.e., pain severity scores or postoperative analgesic consumption); (2) were performed only as distal (i.e., popliteal) SNB; (3) made it impossible to evaluate the analgesic impact of SNB separately from other co-analgesic interventions that were simultaneously administered (e.g., intra-articular injection, epidural anesthesia, wound infiltration); (4) performed additional surgery on areas other than the knee joint (e.g., the ankle); (5) undertook knee procedures that did not include arthroplasty (e.g., cruciate ligament repair); (6) used local anesthetic adjuvants, except epinephrine, to prolong postoperative analgesia; and (7) included fewer than 10 study subjects to limit the possibility of detecting a treatment effect. In the absence of resources to support the translation of RCTs published in foreign languages, inclusion was limited to RCTs published in English, although English abstracts published in foreign-language journals were reviewed for potential relevance. Two authors (F.W.A., R.B.) separately evaluated the search results. Decisions on inclusion of qualifying studies were made by consensus between the two authors. The opinion of a third author (C.M.) was obtained when agreement could not be reached.

Evaluation of methodological quality

The Jadad score27 was used to evaluate the methodological quality of the retrieved trials. A score was separately assigned for the overall quality by two of the authors. The final quality score for each RCT was determined by consensus, although the opinion of a third author was obtained whenever agreement could not be reached. Quality scores were not used to exclude trials.

Definition of relevant outcome data

A standardized data collection form was developed and used to extract outcome results. When consensus could not be reached, discrepancies were resolved by re-examining the source data and asking the opinion of a third author. The authors also agreed pre hoc to refrain from extracting data from any RCTs that they had previously co-authored,28 leaving this task to the third co-author. Extracted data included the name of the first author, publication year, the type of surgical anesthetic received, sample size and number of patients per study group, the nature of comparative groups, and the designated primary outcome. The analgesic outcomes extracted were the opioid consumption for two postoperative intervals: 0-24 hr and 24-48 hr. Pain severity scores were obtained at rest and with motion at two, four, eight, 12, 24, 36, and 48 hr postoperatively. Also obtained were the time-to-first analgesic request (minutes); risk of opioid-related side effects (postoperative nausea and vomiting, pruritus, undesirable sedation); occurrence of block-related complications; patient satisfaction with analgesia received; extent of functional recovery (knee flexion, in degrees) four days following TKA; and time to hospital discharge (days). The cumulative opioid medications consumed during the first interval (0-24 hr) postoperatively, converted into equi-analgesic doses of intravenous morphine, was selected as a primary outcome. All of the other analgesic outcomes were designated secondary outcomes. This choice of the primary outcome reflects the subjective nature of the pain severity scores and the influence of anxiety, catastrophisation, use of patient-controlled analgesia, and aggressiveness of postoperative physiotherapy.29-31

For this review, verbal rating scale scores and pain numerical rating scale (NRS) scores were converted32 to 0-10 visual analogue scale (VAS) scores, where 10 = worst pain imaginable, and 0 = no pain. The amounts of all postoperative opioid analgesics consumed during the first 48 hr were converted to equianalgesic doses of morphine.33 The degree of patient satisfaction with the pain relief at 24 hr postoperatively was reported as a VAS score (10 = most satisfied, 0 = least satisfied).

A priori identified sources of heterogeneity

To explore the possible heterogeneity sources in our review, we a priori determined the clinical and design characteristics of RCTs that could potentially produce clinically important variations in the outcomes assessed. We subsequently evaluated the contribution of these characteristics to the heterogeneity of the results using subgroup analysis. Because evidence suggests that continuous SNB (i.e., catheter-based infusion via a pump) prolongs the duration of SNB analgesia,16 we hypothesized that catheter use would introduce significant heterogeneity into our results. Consequently, we planned to analyze the outcome data separately according to whether single-shot sciatic nerve block (SSNB) or continuous sciatic nerve block (CSNB) had been used. Other sources of clinical heterogeneity that were explored via subgroup analysis included (1) surgical anesthetic (general vs spinal); (2) nature of the local anesthetic used for SNB (intermediate vs long acting); (3) SNB guidance technique (nerve stimulator vs nerve stimulator and ultrasonography); (4) type of FNB (single-shot vs continuous); and (5) postoperative analgesic regimens used (multimodal vs unimodal).

Statistical analysis

The data analyzed were primarily gathered from the tables in each of the source manuscripts. When data were not available in tables, we contacted the corresponding authors of the respective studies. If the authors did not reply or supply the outcome results needed, we resorted to abstracting results from the figures published in the source manuscripts as a secondary source of data. We also contacted the authors of published abstracts that met the eligibility criteria for further explanation regarding any missing details deemed relevant.

Dichotomous outcome results describing opioid-related side effects during a specified time interval were converted to frequencies (n/N). The single highest frequency was used to estimate the number of subjects who experienced a certain side effect at least once. Continuous outcome results were reported as the mean and standard deviation (SD). The median and interquartile range [IQR] was used as an estimate whenever the mean value was not explicitedly reported.34 In studies where the SD values were not provided in tables or graphs, the SD value was estimated as the value of the range/434 or the IQR/1.35, as appropriate.35 Where applicable, the 95% confidence interval (CI) was used as an estimate of the range, and the most extreme values were used to estimate the SD. When a specific outcome measure was reported more than once during a prespecified interval, we selected the most conservative available value.

Meta-analysis

The extracted results were separately entered into the systematic review statistical package (Revman 5.3, Cochrane Library, Oxford, England) and were cross-examined by two of the authors (F.W.A., R.B.). We employed meta-analytic statistics to pool and analyze the outcome results. Because clinically heterogeneous anesthetic techniques were used (e.g., general vs spinal anesthesia, varying doses of local anesthetics in SNB, different postoperative analgesic regimens), random effect modelling was selected. The weighted mean difference and 95% CI were calculated for continuous outcomes, and the odds ratio (OR) and 95% CI were calculated for dichotomous outcomes. Differences from the Control group were considered statistically significant when P < 0.05 (two-sided), and when 0 and 1 were not included in the 95% CI for continuous and dichotomous outcomes, respectively.

We used the I2 statistic to evaluate the degree of heterogeneity among the RCTs reviewed.36 When heterogeneity was significant (I2 > 50%), we explored the heterogeneity sources using subgroup analysis as per the preidentified potential sources of heterogeneity.37 An additional post-hoc sensitivity analysis was planned to examine the effects of any of the preidentified factors where a meaningful subgroup analysis was not feasible as well as the effects of any additional factors identified during the data analysis.

Results

Our literature search yielded 54 studies (Fig. 1), eight of which were published between January 1998 and March 2015 and met the eligibility criteria.16,19,21,28,38-41 One published abstract17 was excluded for lack of outcome results after contacting its authors. Five trials that compared SNB with various analgesic modalities, but lacked an appropriate FNB control group, were also excluded.15,18,20,22,23 A single relevant RCT comparing single-shot FNB to single-shot FNB + SNB was identified in the trial registries (clinical trials registry NCT02135120, relevant sample size = 30), but it was still in the recruitment phase with no available outcome results. One other RCT published in a foreign language was excluded.

Fig. 1
figure 1

Flowchart summarizes the study selection process and depicts retrieved, included, and excluded randomized controlled trials

Full size image

All reviewed trials were randomized, but only three reported appropriate randomization techniques.19,21,28 Five of the trials were double-blinded,19,21,28,38,39 but only four of them19,21,28,38 reported appropriate blinding techniques. Three19,21,28 of the eight trials achieved a quality score of 5 out of 5 (i.e., they were randomized, double-blinded, reported appropriate blinding and randomization techniques, and described complete patient disposition. The median [IQR] and range of the methodological quality score for the eight trials included in this review was 3.5 (3-5). None of the RCTs was excluded based on its methodological quality scores or because the number of patients was fewer than ten. Table 1 describes the methodological quality and the risk of bias of the trials that were included.

Table 1 Summary of trial outcomes
Full size table

The data from 386 patients were used in our meta-analysis, which included 192 patients in the SNB group and 194 in the Control group. The characteristics and outcomes sought in each trial are presented in Table 1. Each of the eight RCTs included analgesic outcomes, including analgesic consumption and pain severity scores. An SSNB was used in five trials19,21,28,38,39 and a CSNB in two.40,41 Although both techniques were used in a single trial,16 only the CSNB group was included in the analysis. The local anesthetic solution, volume, localization technique, and delivery method for all SNBs performed are summarized in Table 2. One trial included two patient groups that underwent SNB,28 but only one group met the eligibility criteria.

Table 2 Characteristics of the SNB and analgesic regimen
Full size table

Analgesic consumption for the 0-24 hr interval

Data relating to postoperative analgesic consumption during the 0-24 hr interval were available from seven RCTs,16,21,28,38-41 which included a total of 338 patients (167 patients in the SNB group and 171 in the Control group). Compared to no Controls, both the SSNB and CSNB reduced the weighted mean difference [95% CI] postoperative intravenous morphine equivalent consumption by 10.6 [−20.9 to −0.3] mg (P = 0.042) and 20.5 [−28.6 to −12.4] mg (P < 0.001), respectively, during the 0-24 hr interval postoperatively (Fig. 2).

Fig. 2
figure 2

Forest plots of intravenous morphine equivalent consumption during the first 24 hr postoperatively (0-24 hr interval). The sample sizes, means, standard deviations, and pooled estimates of the mean difference are shown. The 95% confidence intervals (CI) are shown as lines for individual studies and as diamonds for pooled estimates. CSNB = continuous sciatic nerve block; SSNB = single-shot sciatic nerve block

Full size image

The primary outcome results were characterized by high heterogeneity for the SSNB and CSNB subgroups, with I2 = 97% (P < 0.001) and I2=86% (P < 0.001), respectively. Alternative subgroup analysis showed that the patients undergoing SNB continued to have reduced opioid consumption during the 0-24 hr interval compared with that of the Control group. Specifically, the reduction in cumulative postoperative morphine equivalent consumption was 22.1 [−27.8 to −16.3] mg (P < 0.001), I2 = 82% and 6.5 [−12.0 to −1.1] mg (P < 0.001), I2 = 92% for the general anesthesia vs spinal anesthesia subgroups, respectively; 17.9 [−32.7 to −3.1] mg (P = 0.022), I2 = 98% and 10.9 [−20.7 to −1.1] mg (P = 0.011), I2 = 96% for the nerve stimulator vs nerve stimulator and ultrasonography, respectively; 12.7 [−22.2, to −3.2] mg (P < 0.001), I2 = 98% and 18.8 [−25.5 to −12.0] mg (P < 0.001), I2 = 87% for the single-shot FNB vs continuous FNB, respectively. No trials used intermediate-acting local anesthetics, and only one38,39 used unimodal postoperative analgesic regimens, precluding a meaningful subgroup analysis for the respective a priori identified sources of heterogeneity. The test for subgroup differences was statistically significant for the general anesthesia vs spinal anesthesia subgroup analysis (P < 0.001) but not for the nerve stimulator vs nerve stimulator and ultrasonography analysis (P = 0.421) or for the single-shot FNB vs continuous FNB analysis (P = 0.212). This subgroup difference suggests that patients who undergo TKA under general anesthesia may benefit more from SNB than those who receive a spinal anesthetic.

Analgesic consumption during the 24-48 hr interval

Only three studies reported analgesic consumption during the 24-48 hr interval. Whereas one study38 suggested that the cumulative postoperative opioid analgesic (intravenous morphine equivalent) consumption during the 24-48 hr interval was greater by 4.1 [2.2 to 6.0] mg (P < 0.001) in the SSNB subgroup compared to that of the Controls (Table 3), the remaining two studies16,41 indicated that the difference was not significant for the CSNB subgroup.

Table 3 Summary of outcomes results
Full size table

Pain at rest

Compared with the Controls, SSNB decreased the pain severity at rest at the two-, four-, and eight-hour checkpoints postoperatively by 1.9 [−3.8 to 0.0] cm (P = 0.041), 1.2 [−2.3 to 0] cm (P = 0.04), and 0.7 [−1.3 to −0.1] cm (P = 0.023), respectively (Table 3). The eight-hour assessment of pain severity at rest was the last point at which we observed a significant reduction in pain attributed to SSNB. Figures 3 and 4 show forest plots of the weighted mean difference in pain at rest (VAS scores) at 2 and 8 hr, respectively.

Fig. 3
figure 3

Forest plots of pain severity at rest using visual analogue scale (VAS) pain scores at 2 hr. The sample sizes, means, standard deviations, and pooled estimates of the mean difference are shown. The 95% confidence intervals (CI) are shown as lines for individual studies and as diamonds for pooled estimates. CSNB = continuous sciatic nerve block; SNB = sciatic nerve block; SSNB = single-shot sciatic nerve block

Full size image
Fig. 4
figure 4

Forest plots of the pain severity at rest using visual analogue scale (VAS) pain scores at 8 hr. The sample sizes, means, standard deviations, and pooled estimates of the mean difference are shown. The 95% confidence intervals (CI) are shown as lines for individual studies and as diamonds for pooled estimates. CSNB = continuous sciatic nerve block; SNB = sciatic nerve block; SSNB = single-shot sciatic nerve block

Full size image

Compared with the Controls, CSNB decreased pain severity at rest two, four, eight, 12, 24, and 36 hr postoperatively by 4.3 [−6.4 to −2.1] cm (P < 0.001), 2.8 [−3.4 to −2.3] cm (P < 0.001), 3.2 [−4.1 to −2.3] cm (P < 0.001), 3.0 [−3.5 to −2.4] cm (P < 0.001), 2.0 [−2.9 to −1.2] cm (P < 0.001), and 1.3 [−2.2 to −0.4] cm (P = 0.004), respectively (Table 3). The 36-hr assessment of pain severity at rest was the last point at which we observed a significant reduction in pain attributed to CSNB. The changes in pain at rest (VAS scores, weighted mean difference) for the Controls and the CSNB subgroup over time are shown in Fig. 4.

Pain with movement

Compared with the Controls, SSNB decreased pain severity with movement at two, four, and eight hours postoperatively by 1.5 [−2.7 to −0.2] cm (P = 0.023), 3.2 [−6.2 to −0.3] cm (P = 0.034), and 0.5 [−0.8 to −0.3] cm (P < 0.001), respectively (Table 3). We observed no benefit attributable to SSNB beyond eight hours. In contrast, CSNB reduced pain with movement at all measured time points, including 48 hr, compared with that in the Control group (Table 3).

Opioid-related side effects

The frequency of opioid-related side effects was selected as an outcome by a limited number of trials. Three RCTs - two in the SNB subgroup28,38 and one in the CSNB subgroup40 - reported the incidence of postoperative nausea and vomiting, and only one RCT in the SNB subgroup38 reported the incidence of pruritus and sedation. The frequency of opioid-related side effects in the SNB subgroup did not differ from that in the Controls.

However, CSNB reduced the odds of postoperative nausea and vomiting at 24 hr by 93% (OR [95% CI], 0.09 [0.01 to 0.59], P = 0.011) compared with the Controls (Table 3). Inconsistent reporting precluded evaluating the effect of CSNB on pruritus and sedation.

Other outcomes

The frequency of block-related complications was reported only in trials with an SSNB subgroup (Table 3). Two patients were reported to experience transient paresthesias lasting one week,28 but we found no difference in block-related complications when the SSNB group was compared with the Controls. Functional recovery was reported only in trials with a CSNB subgroup (Table 3). We found no difference in functional recovery when the CSNB group was compared with the Controls (Fig. 5). In addition, we found no differences in the discharge time when SNB - whether SSNB or CSNB - were compared with the Controls (Table 3). We were unable to examine quantitatively the remaining secondary outcomes (time to first analgesic request, patient satisfaction with pain relief) because of inconsistent reporting and the heterogeneous assessment procedures in the source trials.

Fig. 5
figure 5

Forest plots of postoperative functional recovery (range of motion in degrees). The sample sizes, means, standard deviations, and pooled estimates of the mean difference are shown. The 95% confidence intervals (CI) are shown as lines for individual studies and as diamonds for pooled estimates. CSNB = continuous sciatic nerve block; SNB = sciatic nerve block; SSNB = single-shot sciatic nerve block

Full size image

We performed a post hoc alternative subgrouping and sensitivity analysis to explore the potential sources of the heterogeneity of the results. Alternative subgrouping by incorporating the SSNB arm of the study by Wegener et al.16 (a three-arm RCT that included SSNB, CSNB, and Control groups) in the analysis, instead of the CSNB arm, maintained the significance of results but reduced the heterogeneity of the CSNB subgroup, with opioid consumption reduced by 11.2 [−19.9 to −2.6] mg (P = 0.013), I2 = 97% and 25.0 [−30.0 to −20.0] mg (P < 0.001), I2 = 16% in the SSNB and CSNB subgroups, respectively.

Post-hoc sensitivity analysis was used to examine the impact of several factors on the results’ heterogeneity, including the methodological quality scores (i.e., inappropriate blinding), mixing epinephrine with local anesthetics, timing of SNB, not assessing block success, and using unimodal vs multimodal postoperative analgesia. Eliminating the results of the four RCTs16,39-41 that achieved a Jadad score of 3 rendered the SSNB subgroup results insignificant, with a weighted mean difference in opioid consumption of 5.4 [−12.0 to 1.1] mg (P = 0.142). Also, all of the RCTs in the CSNB subgroup were eliminated. Furthermore, removing the data of the two RCTs28,38 that mixed their local anesthetic solution with epinephrine rendered the SSNB subgroup results insignificant, with a weighted mean difference in opioid consumption of 13.4 [−37.3 to 10.5] mg (P = 0.274) and 20.5 [−28.6 to −12.4] mg (P < 0.001), I2 = 86% for the SSNB and CSNB subgroups, respectively. Additionally, removing the results of the single RCT38 that performed SNB postoperatively rendered the SSNB subgroup results insignificant, with a weighted mean difference in opioid consumption of 13.6 [−28.6 to 1.4] mg (P = 0.082) and 20.5 [−28.6 to −12.4] mg (P < 0.001), I2 = 86% for the SSNB and CSNB subgroups, respectively. Moreover, excluding the data of the three RCTs21,39,40 that did not perform block assessment to confirm SNB success rendered the SSNB subgroup results insignificant, with a weighted mean difference in opioid consumption of 7.8 [−19.9 to 4.2] mg (P = 0.232) and 21.1 [−30.9 to −11.3] (P < 0.001), I2 = 93% for the SSNB and CSNB subgroups, respectively. Finally, eliminating the data of the two RCTs38,39 that used unimodal postoperative analgesia rendered the SSNB subgroup results insignificant, with a weighted mean difference in opioid consumption of 7.6 [−20.1 to 5.0] mg (P = 0.244) and 20.5 [−28.6 to −12.4] mg (P < 0.001), I2 = 86% for the SSNB and CSNB subgroups, respectively. As such, our post hoc analysis failed to reduce the extent of heterogeneity in the source data.

Discussion

This meta-analysis suggests that adding SNB to FNB can improve analgesia following TKA by reducing postoperative opioid consumption up to 24 hr, with a larger reduction associated with CSNB than SSNB. The magnitude of this improvement may be greater in patients who undergo TKA under a general anesthetic compared with spinal anesthesia. Unfortunately, the heterogeneity and intrinsic limitations of the source data reviewed limit the strength of our findings. The present quantitative review nonetheless points to level 1a− (minus) evidence42 (i.e., representing a systematic review with significant heterogeneity) in support of the SNB, a notable improvement from the level 2a− evidence supporting SNB that we observed in our earlier qualitative review of the literature.14

Our results also suggest that SSNB can reduce pain at rest and during movement up to eight hours postoperatively, whereas CSNB can reduce pain at rest up to 36 hr and pain with movement up to 48 hr. The relative superior analgesia associated with CSNB compared with SSNB during the first eight hours postoperatively may well be attributable to the extensive multi-modal analgesic regimen used in CSNB trials. Also, the increase in opioid consumption and opioid-related side effects during the 24-48 hr interval associated with SSNB compared with CSNB may be at least partially explained by the “rebound pain” phenomenon that has been described in TKA patients when the effect of the nerve block dissipates.43 Additional SNB benefits may include a reduction in the frequency of postoperative nausea and vomiting in patients who underwent CSNB. Importantly, our results also signal no delay in functional recovery or hospital discharge associated with SNB for TKA. Finally, the relatively small number of patients reviewed herein (n = 338), coupled with the inconsistent assessment of block-related complications, preclude any meaningful conclusions regarding SNB safety. Consequently, concerns regarding SNB-related nerve injury, masking of surgical injury to the peroneal nerve, and delayed mobilization still apply.44,45

Clinically, the present results support the idea that the sciatic nerve contributes to the postoperative pain following TKA. Anatomically, however, our understanding of the innervation of the posterior knee is relatively incomplete as it is based primarily on anatomical descriptions dating back to more than six decades ago.46 Hence, designing trials to quantify the effect of SNB accurately when its distribution is not clearly delineated is difficult at best. Additionally, the combined innervation of the knee joint by both the femoral and sciatic nerves, and the overlap in their innervation of the anterolateral aspect of the knee,46 further complicate accurate assessment. In fact, few studies12 have conceived designs capable of reliably isolating pain transmitted by the femoral and sciatic nerves, which would permit accurate quantification of a block’s analgesic effects on the individual nerves.44 Furthermore, the human capacity to distinguish between pain arising from two adjacent sources as separate is subject to the interplay between spatial summation47 and discrimination.48 Summation may account for a lack of difference in perceived pain arising from the anterior knee vs that arising from both the anterior and posterior knee.49 Also, discrimination enables distinguishing pain from two sources only if they are ≥10 cm apart.50 Realistically, overall knee pain may be the best available surrogate measure of the analgesic effects of SNB. Finally, LIA is an emerging analgesic modality2; but whether its ultimate role in a post-TKA analgesic regimen is to complement nerve blocks51 or replace them altogether52-55 is to be determined by future research.

Our systematic review has several areas of strength. The literature search we conducted was exhaustive and included all of the relevant electronic databases. Our eligibility criteria restricted the evidence sought to that in RCTs and excluded several observational studies11-13 that had been included in earlier reviews.14 Additionally, all of the trials we included were of high quality (Jadad score ≥ 3). These factors thus reinforce the validity of our findings.

Our systematic review, however, also has some limitations. First, the outcome data originated from varying RCT settings with diverse anesthetic and analgesic practices. For example, the nerve localization techniques employed and the analgesic regimens used varied among the trials, leading to considerable heterogeneity. We also did not use a specific type of TKA surgery to stratify our results. Variations among TKA techniques could result in differences in the duration and severity of postoperative pain as well as the dose of postoperative morphine needed to treat the pain. Second, the majority of the trials we reviewed were characterized by their small sample size, with a maximum of 33 patients per group. Such small RCTs could increase the likelihood of detecting a treatment effect and/or misestimating its magnitude because of a combination of mere chance and publication bias. Third, analgesic techniques other than SNB (e.g., intra- articular local anesthetic infiltration,18,23 posterior knee capsule injection56 have achieved effective pain relief following TKA, whether in combination with or as substitutes for SNB. The influence of the aforementioned analgesic options on SNB analgesia is not within the scope of our systematic review. Fourth, it is not possible to exclude reporting bias because publishing the study protocols on trial registration sites has not been common practice. Fifth, the reviewed RCTs used varying and sometimes unclear definitions to describe reported outcomes, which questions the validity of combining these outcome results. Sixth, this review was limited to RCTs published in English, and we did not translate any foreign trials. Seventh, although the absolute difference in pain scores may seem modest at certain time points, including at four and eight hours for SSNB, differences in pain severity scores as small as 0.9 and 1.1 units on the VAS have been proven to be clinically important.57,58 Finally, none of the RCTs described any preexisting (i.e., preoperative) knee pain, which precluded analysis of the association between this postoperative pain predictor59 and the duration of post-TKA analgesia supplied by the SNB.

In conclusion, the available evidence supporting an analgesic benefit of adding SNB to FNB following TKA is marked by significant heterogeneity. Keeping this heterogeneity limitation in mind, our meta-analysis suggests that SNB can significantly reduce postoperative opioid consumption and diminish the knee pain following TKA compared to no SNB in the setting of FNB. Although analgesic strategies vary greatly across institutions,45,60 our results support a role for SNB in acute pain management following TKA.