nomogram for predicting the need for sciatic nerve block after total knee arthroplasty

16

A nomogram for predicting the need for sciatic nerve block after total knee arthroplasty

Rovnat Babazade1,2 · Thilak Sreenivasalu2,3 · Pankaj Jain4 · Matthew T. Hutcherson4 · Amanda J. Naylor4 · Jing You4,5 · Hesham Elsharkawy4,6 · Ali Sakr Esa Wael4,6 · Alparslan Turan4

Abstract

Purpose Sciatic nerve block (SNB) is commonly per-formed in combination with femoral nerve block (FNB) for postoperative analgesia following total knee arthroplasty (TKA). Despite the fact that 10–20 % of TKA patients require SNB for postoperative posterior knee pain, there are no existing studies that suggest a model to predict the need for SNB. The aim of our study was to develop a pre-diction tool to measure the likelihood of patients undergo-ing TKA surgery requiring a postoperative SNB.

Methods With institutional review board approval, we obtained data from the electronic medical record of patients who underwent TKA at the Cleveland Clinic. A multivari-able logistic regression was used to estimate the probabil-ity of requiring a postoperative SNB. Clinicians selected potential predictors to create a model, and the potential

This report was previously presented, in part, at the American Society of Anesthesiologists, 2014 Meeting, New Orleans.nonlinear association between continuous predictors and SNB was assessed using the restricted cubic spline model. Results In total 6279 TKA cases involving 2329 patients with complete datasets were used for building the predic-tion model, including 276 (12 %) patients who received a postoperative SNB and 2053 (88 %) patients who did not. The estimated C statistic of the prediction model was 0.64. The nomogram is used by first locating the patient position on each predictor variable scale, which has corresponding prognostic points. The cut-off of 11.6 % jointly maximizes the sensitivity and specificity.

Conclusion This is the first study to be published on SNB prediction after TKA. Our nomogram may prove to be a useful tool for guiding physicians in terms of their deci-sions regarding SNB.

Keywords Nomogram · Sciatic nerve block · Postoperative pain · Total knee arthroplasty

Introduction

Total knee arthroplasty (TKA) is a commonly performed procedure in the USA, with the primary aims of relieving knee pain and improving the quality of life of patients with end-stage arthritis. An estimated 718,000 procedures were carried out in 2011 [1, 2], and the demand for primary TKA is expected to grow in the coming years [3]. TKA causes severe postoperative pain in 60 % of patients and moderate pain in 30 % of patients [4]. Therefore, pain management remains one of the major challenges for anesthesiologists following TKA.

Postoperative knee pain interferes with physical ther-apy, mobilization [5], and patient satisfaction, which in turn results in delayed recovery and higher hospital costs

[6]. The incidence of immediate clinically significant postoperative pain in this patient population has been estimated to be as high as 80 % despite the placement of

  1. well-functioning femoral nerve block (FNB) [7–10]. The commonly employed modalities for postoperative pain relief in these patients are intravenous opioids [11], local anesthetic infiltration, epidural anesthesia [12], and regional anesthesia techniques, including continuous FNB [13, 14]. Various combinations of the above-mentioned modalities offer sufficient control of anterior knee pain. However, posterior knee pain remains fairly common, and even with a multimodal approach, including femoral nerve catheter, 10–20 % of patients may still require single shot or continuous sciatic nerve block (SNB) in the postopera-tive period [15]. Randomized controlled trials and obser-vational studies have reported a reduction in postopera-tive pain scores and opioid consumption in the immediate postoperative period when SNB, administered either as a single shot or continuously, was combined with a FNB [8, 16–18].

To enhance rehabilitation and limit long-term postopera-tive pain, preemptive analgesia approaches may be prefer-able to more simple attempts to decrease postoperative pain [19]. Although the addition of SNB to the postopera-tive pain management regimens of all patients may still be clinically controversial, prediction of the preemptive anal-gesia approach to posterior knee pain may improve recov-ery after TKA for some patients [19]. Despite reports that 10–20 % of TKA patients require SNB for postoperative posterior knee pain, there are no existing studies that sug-gest a model to predict the need for SNB. Thus, the aim of our study was to develop a prediction tool to measure the likelihood of patients undergoing TKA surgery requiring a postoperative SNB.

Methods

With Cleveland Clinic Institutional Review Board approval (IRB# 13-128, Date: 2/07/2013, Cleveland, Ohio), we obtained data from the electronic medical record of patients who underwent TKA from 2007 to 2012 at the Cleveland Clinic Main Campus. The data extraction process was facilitated by an initial query of the Cleveland Clinic Perio-perative Health Documentation System registry to collect all available research data.

The standard anesthetic technique for TKA in our hospital is: (1) preoperatively patients receive FNB with catheter placement for postoperative pain management in the induction room; (2) intraoperatively, anesthesiolo-gists either administer a single-shot spinal block or gen-eral anesthesia, depending on patient and surgical factors. Administration of SNB perioperatively is not a standard

anesthetic technique for TKA in our hospital. The protocol for postoperative knee pain is as follows: (1) if available, first check and redose the femoral catheter; (2) administer rescue intravenous opioids; (3) administer SNB if poste-rior knee pain is rated ≥5 on the visual analog scale and numeric rating scale. A multivariable logistic regression was used to estimate the probability of requiring a post-operative SNB. The potential predictors selected by clini-cians included age, gender, race, body mass index (BMI), diabetes, peripheral vascular disease, anxiety disorder, psychiatric disorder, depression, previous knee surgery, chronic pain, use of chronic pain medications, American Society of Anesthesiology (ASA) physical status, emer-gency (vs. elective) surgery, primary (vs. revision) surgery, intra-operative anesthetic management (general/spinal/oth-ers), joint infiltration by the surgeon, duration of surgery, and tourniquet. All of the predictors included in the model were selected via the backward model selection (conserva-tive alpha-to-stay criterion at 0.20). The univariable asso-ciation between receiving SNB and each potential predic-tor was estimated and reported. The significance criterion was a P value of <0.05.

The discrimination (predictive accuracy) was measured by C-statistic. The C-statistic is equivalent to the probabil-ity that if a pair of patients are randomly selected, the one who received SNB would have a higher predicted prob-ability than the other who did not receive a block. The value ranges between 0.5 (random guess) and 1 (perfect prediction).

We assessed the potential nonlinear association between continuous predictors and SNB with the use of the restricted cubic spline model. Adding these nonlinear terms did not improve the predictive ability (measured by the C-statistic) of the model and, therefore, only linear terms were considered in the final model.

We also used the “Least Absolute Shrinkage and Selec-tion Operator” (LASSO) modeling approach. The penalty parameter was chosen to minimize the tenfold cross-vali-dation error (glmnet package in R: cv.glmnet). The same set of variables was retained as in the backward selection method. However, the LASSO approach gave a lower C-statistic; thus, we used the model selected via the back-ward selection as our final model.

The final model was subjected to 200 bootstrap resam-ples for minimizing overfitting bias and for internal vali-dation. The calibration of our final model was assessed graphically by plotting the bootstrapped calibration curve of observed proportions against the predicted probabilities arising from the model with smoothing.

A nomogram was constructed based on the fitted logistic regression model. For information purposes, we illustrate the measures of discrimination, that is, sensitivity, speci-ficity, positive predictive values, and negative predictive values associated with nomogram calculated probability of receiving SNB.

Sample size considerations

While there are a multitude of techniques which—depend-ing on the type of outcome and the experimental setup— can estimate the number of needed patients to adequately address research hypotheses comparing two or more expo-sures on an outcome, the literature on sample size plan-ning for prediction studies is rather sparse. One popular guideline for any regression model is to include at least 15–20 outcomes for every parameter to be estimated. With approximately 15 predictors, the guideline of at least 15 outcomes for every parameter being estimated would sug-gest that at least 225 outcomes are collected. We estimated conservatively that 15 % of TKA patients receive sciatic blockade after surgery. Thus, a sample of at least 1500 patients would be required to meet the guideline, as 225 is 15 % of 1500.

R statistical software version 2.15.1 for 64-bit Microsoft Windows (The R Foundation for Statistical Computing, Vienna, Austria) and SAS software version 9.4 (SAS Insti-tute, Cary, NC) were used for statistical analyses.

Results

There were 6279 TKA cases from 2007 to 2012 at the Cleveland Clinic Main Campus. There were no obvious changes during the study period in regards to surgical and anesthesia management of patients undergoing such sur-gery. We excluded patients who received preoperative SNB and who had undergone repeated/multiple TKA surger-ies during the queried time window, as well as those for whom the outcome was unknown, leaving 4744 eligible patients. Then, we collected data for 2372 randomly sam-pled patients (50 % of patients meeting the criteria), choos-ing each patient randomly and entirely by chance, such that each patient had the same probability of being chosen dur-ing the sampling process. The prediction model was con-structed using completed data on 2329 patients, among whom 276 (12 %) patients received a postoperative SNB and 2053 (88 %) patients did not (Table 1). None of the patients had hematoma at the back of the knee.

Table 2 shows the univariable and multivariable asso-ciation between predictors and receiving SNB. Age, race, BMI, depression, peripheral vascular disease, previous knee surgery, use of chronic pain medications, intra-oper-ative anesthetic management (general/spinal/others), joint infiltration, and duration of surgery were retained in the final prediction model. Although some variables were not statistically significant independent predictors of receiving

SNB, they were retained in the model because omitting them from the model decreases the accuracy.

The estimated C-statistic of the prediction model was 0.64 [95 % confidence interval (CI) 0.61–0.67], indicating a slightly moderate discriminative ability. The bias-cor-rected (based on bootstrap resampling) estimate of predic-tive discrimination was 0.62, the estimate of the maximum calibration error in predicting SNB was 0.13, and the cor-rected Brier score was 0.10; all of these estimates were quite satisfactory. Figure 1 shows a linear calibration func-tion estimate. The ideal calibration curve lies on the 45° line from the origin. The overfitting-corrected calibration is seen to be excellent everywhere, being only slightly worse than the apparent calibration.

Figure 2 graphically displays the prediction model in the form of a nomogram. The nomogram is used by first locat-ing the patient position on each predictor variable scale, which has corresponding prognostic points (top axis). Point values for all the predictor variables are determined con-secutively and summed to arrive at a total point axis and directly below is the predicted probability for receiving SNB postoperatively.

Table 3 further illustrates the measures of discrimination associated with nomogram-calculated probability of receiv-ing SNB (cut-offs; SNB is predicted if above the cut-off). The cut-off of 11.6 % jointly maximizes the sensitivity (0.70—proportion of patients above the cut-off among all patients who received SNB) and specificity (0.52—propor-tion of patients below cut-off among all patients who did not receive SNB). A lower cut-off increases the sensitiv-ity but decreases the specificity; the inverse is true with a higher cut-off.

Discussion

Our study allowed us to identify an association between prespecified demographic-, morphometric-, and periop-erative surgical/anesthesia-related variables and the neces-sity for postoperative SNB in patients undergoing TKA. Overall, we found that 276 (12 %) patients required post-operative SNB among 2329 patients who underwent TKA in our hospital during the study period. Among the factors considered in our study, we found that younger age, lower BMI, a state of depression, short duration of surgery under anesthesia other than general or spinal, and surgery with-out joint infiltration were significantly associated with increased probability of postoperative SNB requirement. For example, patients with a diagnosis of depression were 1.55-fold more likely (95 % CI 1.10–2.19) to receive SNB postoperatively than patients without this diagnosis. Fur-thermore, the probability tends to increase in non-white, patients without peripheral vascular disease, patients who

Table 1  Baseline and intraoperative characteristics
Variable Patient groups (total N = 2329)
Patients receiving SNB postoperatively (N = 276) Patients not receiving SNB postoperatively (N = 2053)
Age (years) 63 (56, 72) 65 (58, 74)
Gender (male) 106 (38) 879 (43)
Race (white) 219 (79) 1716 (84)
BMI (kg/m2) 30.8 (26.4–34.9) 31.1 (27.1–36.7)
Diabetes 54 (20) 402 (20)
Peripheral vascular disease 5 (2) 90 (4)
Anxiety disorder 0 (0) 1 (0)
Psychiatric disorder 7 (3) 50 (2)
Depression 50 (18) 260 (13)
Previous knee surgery 4 (1) 72 (4)
Chronic pain 8 (3) 33 (2)
Use of chronic pain medications 3 (1) 67 (3)
ASA physical status
I 4 (1) 23 (1)
II 114 (41) 765 (37)
III 150 (54) 1171 (57)
IV 8 (3) 94 (5)
Emergency surgery (vs. elective) 2 (1) 22 (1)
Primary surgery (vs. revision) 222 (80) 1616 (79)
Intra-operative anesthetic management
General 95 (34) 709 (35)
Spinal 180 (65) 1246 (61)
Others 1 (0) 98 (5)
Joint infiltration 68 (25) 688 (34)
Duration of surgery (min) 153 (127–189) 159 (130–204)
Tourniquet 261 (95) 1938 (94)

Median with 1st and 3rd quantiles are in parenthesis

SNB sciatic nerve block, BMI body mass index,ASA American Society of Anesthesiology

had not had previous knee surgery, patients who took medi-cations to treat chronic pain, and patients who underwent surgery under spinal versus general anesthesia.

Studies have proven that aging affects the anatomy in general, with a loss of both myelinated and unmyelinated nerve fibers [20, 21], over and above the structural and functional properties of the peripheral nervous system [22]. There is no available study focused on the immediate post-operative pain (including posterior knee) and age relation-ship. Nonetheless, there are two studies which followed TKA patients and showed that elderly patients have less overall postoperative knee pain than younger patients [23, 24]. These studies do not specifically discuss posterior knee pain, but the findings do support the results of our study which also show a decreasing probability of SNB with increasing age.

Published data on the association between length of sur-gery and postoperative pain are lacking, but tourniquet time

and postoperative outcomes have been examined. Fan et al. showed that limiting the use of a tourniquet in TKA results in a reduction of knee joint pain [25]. Ejaz et al. conducted a small prospective randomized study and found that TKA without the use of a tourniquet results in reduced pain and analgesic usage [26]. Two decades ago, Abdel-Salam and Eyres performed a small prospective randomized study; based on the results these authors concluded that postop-erative pain was less in the patients in whom a tourniquet had not been used [27]. Our findings do not agree with those of these previous studies, all which found that tourni-quet and tourniquet time impact patient postoperative knee pain. Rather, we found no association between tourniquet and pain, although interestingly we found length of sur-gery and previous knee surgery to be independent factors for SNB requirement. Longer surgery, knee manipulations, and longer tourniquet time increase the risk of damage to the sciatic nerve. One possible explanation for our result is

Table 2  Univariable and multivariable logistic regression analyses for predicting the need for a sciatic nerve block (SNB) after total knee arthroplasty

Predictors Univariable analysis (N = 2329) Multivariable analysis (N = 2329)
Full model Final modela
Odds ratio P value AUC Odds ratio P valueb Odds ratio P valueb
Age (per increase of 5 years) 0.92 (0.77, 0.95) 0.005 0.55 0.92 (0.86, 0.98) 0.007 0.92 (0.87, 0.97) 0.003
Gender (male vs. female) 0.83 (0.64, 1.08) 0.16 0.52 0.89 (0.68, 1.16) 0.37
Race (white vs. others) 0.75 (0.55, 1.03) 0.08 0.52 0.74 (0.53, 1.03) 0.07 0.74 (0.53, 1.02) 0.06
BMI (per increase of 5 kg/m2) 0.91 (0.83, 1.00) 0.04 0.54 0.87 (0.79, 0.96) 0.006 0.87 (0.79, 0.96) 0.005
Diabetes (yes vs. no) 1.00 (0.73, 1.37) >0.99 1.10 (0.78, 1.54) 0.60
Peripheral vascular disease (yes vs. no) 0.40 (0.16, 1.00) 0.05 0.51 0.46 (0.18, 1.16) 0.10 0.44 (0.18, 1.11) 0.08
Psychiatric disorder (yes vs. no) 1.04 (0.47, 2.32) 0.92 0.50 0.77 (0.34, 1.76) 0.53
Depression (yes vs. no) 1.53 (1.09, 2.13) 0.01 0.53 1.53 (1.08, 2.17) 0.02 1.55 (1.10, 2.19) 0.012
Previous knee surgery (yes vs. no) 0.41 (0.15, 1.12) 0.08 0.51 0.41 (0.15, 1.15) 0.09 0.41 (0.15, 1.15) 0.09
Chronic pain (yes vs. no) 1.83 (0.84, 4.00) 0.13 0.51 1.53 (0.68, 3.46) 0.30
Chronic pain medications (yes vs. no) 0.33 (0.10, 1.04) 0.06 0.51 0.38 (0.12, 1.24) 0.11 0.37 (0.11, 1.19) 0.09
ASA physical status(per increase of 1 level) 0.83 (0.67, 1.03) 0.09 0.53 0.95 (0.75, 1.22) 0.70
Emergency surgery (vs. elective) 0.68 (0.16, 2.88) 0.60 0.50 0.91 (0.20, 4.01) 0.90
Primary surgery (vs. revision) 1.11 (0.81, 1.52) 0.51 0.51 1.02 (1.45, 0.72) 0.91
Intra-operative anesthetic management 0.53
Spinal vs. general 1.08 (0.83, 1.41) 0.58 1.07 (0.80, 1.42) 0.64 1.08 (0.82, 1.42) 0.59
Others vs. general 0.08 (0.01, 0.55) 0.01 0.07 (0.01, 0.53) 0.009 0.07 (0.01, 0.54) 0.01
Joint infiltration (yes vs. no) 0.65 (0.49, 0.87) 0.003 0.54 0.62 (0.46, 0.84) 0.002 0.62 (0.46, 0.83) 0.001
Duration of surgery (per increase of 10 min) 0.97 (0.95, 1.00) 0.02 0.54 0.98 (0.95, 1.00) 0.04 0.98 (0.95, 1.00) 0.03
Tourniquet (yes vs. no) 1.03 (0.59, 1.80) 0.91 0.50 1.09 (0.62, 1.92) 0.77
AUC of the multivariable models 0.644 0.641

AUC Area under the curve

a Selected via the backward model selection with alpha-to-stay of 0.20

b Significance criterion P < 0.05

that during the early postoperative period patients may not experience pain from posterior knee surgery due to nerve damage.

Cotter et al. performed a retrospective study involving 7160 patients and found that a high BMI is an independ-ent risk factor for block failure [28]. Due to comorbidities, deep location, difficult landmarks, and poor ultrasound visualization, physicians do not prefer to place SNBs in this particular patient population. Likewise, our prediction model showed the need for SNB in obese patients reaches a minimum at a BMI of 70 kg/m2. This may be a result of a selection bias, with clinicians being hesitant to perform a sciatic block in patients with higher BMI. In contrast with the results of Cotter et al. [28], we did not find a statistical association between the ASA status and SNB requirement. One possible explanation is that our institution is a tertiary care hospital and the majority of the patients have higher ASA status compared to those in some other institutions.

FNBs are still the gold standard for postoperative anal-gesia in TKA, but the routine addition of SNB to the pain

management regimen is controversial [12]. Recent ran-domized clinical trials, although limited by sample size, have demonstrated that the combination of FNB + joint infiltration provides sufficient analgesia [29] and a poten-tial alternative to the SNB + FNB combination [30]. Per-haps in the future, large-scale clinical trials will be able to demonstrate the advantage of infiltration. However, the best way to appropriately treat postoperative pain is to know in advance which patients are likely to need additional analge-sia, such as SNB or joint infiltration. Patients who received joint infiltration were given standard local anesthetics. Our finding that joint infiltration reduced the need for postoper-ative SNB may guide physicians to utilize joint infiltration in patients who are not good candidates for SNB.

Within our model, patients diagnosed with peripheral vascular diseases and those on medication to treat chronic pain were less likely to need SNB postoperatively. These findings were unexpected because most previous stud-ies have demonstrated that chronic pain medication use is associated with increased pain after surgery. The most

Fig. 1  Bootstrapped calibration curve. SNB Sciatic nerve block

Fig. 2  Nomogram predic-tion for receiving SNB after total knee arthroplasty (TKA) (N = 2329). The nomogram is used by first locating the patient position on each predictor variable scale, which has cor-responding prognostic points (top axis). For example, age of 50 years corresponds to 32.5 points; surgery of 100 min corresponds to 72.5 points; and so forth. Point values for all of the predictor variables are determined consecutively and summed to arrive at a total point axis; directly below is the predicted probability for receiving SNB postoperatively. For example, a total score of 335 corresponds to a predicted probability of 0.1 (10 %) for receiving SNB after TKA

probable explanation may be one in which the use of other pain medication(s) intraoperatively to control pain affects the postoperative need for SNB.

Our findings are consistent with those reported previ-ously which show that depression predicts higher pain lev-els after surgery [31, 32]. It is well known that patients with depression feel pain more than those without depression and also that they require more opioids [31, 32]. Therefore, it is not surprising that their pain scores were higher and they required SNB more frequently. Of particular impor-tance is that our results indicate that depression and race other than Caucasian are associated with an increased need of SNB. Additionally, our results did not identify anxiety disorder as an independent factor for SNB requirement, which is in contrast to the results of previous studies [31, 32].

The prediction model we present here is the first such model, but it does not have a perfect prediction value, as the estimated C-statistic of the prediction model was 0.64 (95 % CI 0.61–0.67). Our study has a few other limitations, including that it is a retrospective study and that it was con-ducted at a single center. The variables were assessed for

Table 3  Systematic analyses of the updated nomogram-derived cut-offs used to discriminate between patients who received a sciatic nerve block and those who did not after total knee arthro-plasty

Nomogram-calculated Number of patients who Number of patients who Sensitivity (95 % CI)c Specificity (95 % CI)d Positive prediction value Negative prediction value
probability of receiving needed SNB according to did not need SNB accord- (95 % CI)e (95 % CI)f
SNB (cut-off %) the cut-offa ing to the cut-offb
5 2093 (90 %) 236 (10 %) 0.99 (0.97, 1.00) 0.11 (0.10, 0.13) 0.13 (0.12, 0.14) 0.98 (0.97, 1.00)
6 2027 (87 %) 302 (13 %) 0.97 (0.95, 0.99) 0.14 (0.13, 0.16) 0.13 (0.12, 0.15) 0.97 (0.96, 0.99)
7 1938 (83 %) 391 (17 %) 0.94 (0.91, 0.97) 0.18 (0.17, 0.20) 0.13 (0.12, 0.15) 0.96 (0.94, 0.98)
8 1803 (77 %) 526 (23 %) 0.90 (0.87, 0.94) 0.24 (0.22, 0.26) 0.14 (0.12, 0.15) 0.95 (0.93, 0.97)
9 1627 (70 %) 702 (30 %) 0.85 (0.81, 0.89) 0.32 (0.30, 0.34) 0.14 (0.13, 0.16) 0.94 (0.92, 0.96)
10 1466 (63 %) 863 (37 %) 0.80 (0.76, 0.85) 0.39 (0.37, 0.42) 0.15 (0.13, 0.17) 0.94 (0.92, 0.95)
11 1279 (55 %) 1050 (45 %) 0.72 (0.67, 0.77) 0.47 (0.45, 0.50) 0.16 (0.14, 0.18) 0.93 (0.91, 0.94)
11.6g 1173 (50 %) 1156 (50 %) 0.70 (0.64, 0.75) 0.52 (0.50, 0.54) 0.16 (0.14, 0.18) 0.93 (0.91, 0.94)
12 1104 (47 %) 1225 (53 %) 0.64 (0.58, 0.70) 0.55 (0.53, 0.57) 0.16 (0.14, 0.18) 0.92 (0.90, 0.93)
13 933 (40 %) 1396 (60 %) 0.54 (0.48, 0.60) 0.62 (0.60, 0.64) 0.16 (0.14, 0.18) 0.91 (0.89, 0.92)
14 750 (32 %) 1579 (68 %) 0.47 (0.41, 0.53) 0.70 (0.68, 0.72) 0.17 (0.14, 0.20) 0.91 (0.89, 0.92)
15 589 (25 %) 1740 (75 %) 0.38 (0.33, 0.44) 0.76 (0.75, 0.78) 0.18 (0.15, 0.21) 0.90 (0.89, 0.92)
16 454 (19 %) 1875 (81 %) 0.31 (0.25, 0.36) 0.82 (0.80, 0.84) 0.19 (0.15, 0.22) 0.90 (0.88, 0.91)
17 356 (15 %) 1973 (85 %) 0.26 (0.21, 0.32) 0.86 (0.85, 0.88) 0.21 (0.16, 0.25) 0.90 (0.88, 0.91)
18 279 (12 %) 2050 (88 %) 0.21 (0.16, 0.25) 0.89 (0.88, 0.91) 0.20 (0.16, 0.25) 0.89 (0.88, 0.91)
19 212 (9 %) 2117 (91 %) 0.18 (0.13, 0.22) 0.92 (0.91, 0.93) 0.23 (0.17, 0.29) 0.89 (0.88, 0.91)
20 162 (7 %) 2167 (93 %) 0.15 (0.11, 0.19) 0.94 (0.93, 0.95) 0.25 (0.19, 0.32) 0.89 (0.88, 0.90)

95 % CI 95 % confidence interval

a Number and percentage (in parenthesis) of patients above the cut-off

b Number and percentage (in parenthesis) of patients below the cut-off

c Percentage of patients above the cut-off out of all patients who received SNB

d Percentage of patients below the cut-off out of all patients who did not receive SNB

e Percentage of patients above the cut-off who received a SNB out of all patients above the cut-off

f Percentage of patients who did not receive a SNB out of all patients below cut-off

g The cut-off that jointly maximizes the sensitivity and specificit

individual association and not as a single predictive model, and additional patient variables not included in the analy-sis may be associated with SNB. We have not assessed the performance of the nomogram in an external dataset. How-ever, given similar predictive ability obtained from the bias-corrected discrimination estimation (bootstrap resampling) and the penalized estimation (LASSO model), we expect that our model has a similar prediction ability for new data.

In conclusion, our nomogram may prove to be a use-ful tool for guiding physicians in their decision-making on SNB. In addition to our variables, large-scale studies are needed to identify new potential variables to obtain a better accuracy.

Acknowledgments Received from the Departments of Outcomes Research and Quantitative Health Sciences, Cleveland Clinic, Cleve-land, Ohio. No external funding and no competing interests declared.

Compliance with ethical standards

Funding This manuscript was written by the investigators. None of the authors has a personal financial interest in this research.

References

\ 1.\ Pfuntner A WL, Stocks C. Most frequent procedures performed in US Hospitals. HCUP Statistical Brief #165. October 2013. Agency for Healthcare Research and Quality, Rockville, MD. Available at: http://www.hcup-us.ahrq.gov/reports/statbriefs/ sb165.pdf. Accessed Oct 2013.

\ 2.\ Weiss AJ EA, Andrews, RM. Characteristics of operating room procedures in US Hospitals. HCUP Statistical Brief #170. February 2014. Agency for Healthcare Research and Qual-ity, Rockville, MD. Available at: http://www.hcup-us.ahrq.gov/ reports/statbriefs/sb170-Operating-Room-Procedures-United-States-2011.jsp. Accessed Feb 2014.

\ 3.\ Kurtz S, Mowat F, Ong K, Chan N, Lau E, Halpern M. Preva-lence of primary and revision total hip and knee arthroplasty in the United States from 1990 through 2002. J Bone Joint Surg Am. 2005;87:1487–97.

\ 4.\ Lee JJ, Choi SS, Lee MK, Lim BG, Hur W. Effect of continu-ous psoas compartment block and intravenous patient controlled analgesia on postoperative pain control after total knee arthro-plasty. Korean J Anesthesiol. 2012;62:47–51.

\ 5.\ Oldmeadow LB, McBurney H, Robertson VJ. Hospital stay and discharge outcomes after knee arthroplasty: implications for physiotherapy practice. Aust J Physiother. 2002;48:117–21.

\ 6.\ Indelli PF, Grant SA, Nielsen K, Vail TP. Regional anesthesia in hip surgery. Clin Orthop Relat Res. 2005;441:250–5.

\ 7.\ Allen HW, Liu SS, Ware PD, Nairn CS, Owens BD. Peripheral nerve blocks improve analgesia after total knee replacement sur-gery. Anesth Analg. 1998;87:93–7.

\ 8.\ Cook P, Stevens J, Gaudron C. Comparing the effects of femoral nerve block versus femoral and sciatic nerve block on pain and opiate consumption after total knee arthroplasty. J Arthroplasty. 2003;18:583–6.

\ 9.\ Ilfeld BM, Ball ST, Gearen PF, Le LT, Mariano ER, Vanden-borne K, Duncan PW, Sessler DI, Enneking FK, Shuster JJ, Theriaque DW, Meyer RS. Ambulatory continuous posterior lumbar plexus nerve blocks after hip arthroplasty: a dual-center,

randomized, triple-masked, placebo-controlled trial. Anesthesi-ology. 2008;109:491–501.

10\.\ Ilfeld BM, Mariano ER, Girard PJ, Loland VJ, Meyer RS, Dono-van JF, Pugh GA, Le LT, Sessler DI, Shuster JJ, Theriaque DW, Ball ST. A multicenter, randomized, triple-masked, placebo-con-trolled trial of the effect of ambulatory continuous femoral nerve blocks on discharge-readiness following total knee arthroplasty in patients on general orthopaedic wards. Pain. 2010;150:477–84.

\11.\ Weber A, Fournier R, Van Gessel E, Gamulin Z. Sciatic nerve block and the improvement of femoral nerve block analgesia after total knee replacement. Eur J Anaesthesiol. 2002;19:834–6.

12\.\ Bauer MC, Pogatzki-Zahn EM, Zahn PK. Regional analgesia techniques for total knee replacement. Curr Opin Anaesthesiol. 2014;27:501–6.

\13.\ Lee RM. Lim Tey JB, Chua NH. Postoperative pain control for total knee arthroplasty: continuous femoral nerve block ver-sus intravenous patient controlled analgesia. Anesth Pain Med. 2012;1:239–42.

14\.\ Al-Zahrani T, Doais KS, Aljassir F, Alshaygy I, Albishi W, Terkawi AS. Randomized clinical trial of continuous femoral nerve block combined with sciatic nerve block versus epidural analgesia for unilateral total knee arthroplasty. J Arthroplasty. 2015;30:149–54.

\15.\ Abdallah FW, Chan VW, Gandhi R, Koshkin A, Abbas S, Brull R. The analgesic effects of proximal, distal, or no sciatic nerve block on posterior knee pain after total knee arthroplasty: a dou-ble-blind placebo-controlled randomized trial. Anesthesiology. 2014;121:1302–10.

16\.\ Chan EY, Fransen M, Parker DA, Assam PN, Chua N. Femoral nerve blocks for acute postoperative pain after knee replacement surgery. Cochrane Database Syst Rev. 2014;5:CD009941.

\17.\ Hunt KJ, Bourne MH, Mariani EM. Single-injection femoral and sciatic nerve blocks for pain control after total knee arthroplasty. J Arthroplasty. 2009;24:533–8.

18\.\ PhamDang C, Gautheron E, Guilley J, Fernandez M, Waast D, Volteau C, Nguyen JM, Pinaud M. The value of adding sciatic block to continuous femoral block for analgesia after total knee replacement. Reg Anesth Pain Med. 2005;30:128–33.

\19.\ Sinatra RS, Torres J, Bustos AM. Pain management after major orthopaedic surgery: current strategies and new concepts. J Am Acad Orthop Surg. 2002;10:117–29.

20\.\ Chase MH, Engelhardt JK, Adinolfi AM, Chirwa SS. Age-dependent changes in cat masseter nerve: an electrophysiological and morphological study. Brain Res. 1992;586:279–88.

\21.\ Ceballos D, Cuadras J, Verdu E, Navarro X. Morphometric and ultrastructural changes with ageing in mouse peripheral nerve. J Anat. 1999;195:563–76.

22\.\ Hanks RK, Pietrobon R, Nielsen KC, Steele SM, Tucker M, Warner DS, King KP, Klein SM. The effect of age on sciatic nerve block duration. Anesth Analg. 2006;102:588–92.

\23.\ Singh JA, Gabriel S, Lewallen D. The impact of gender, age, and preoperative pain severity on pain after TKA. Clin Orthop Relat Res. 2008;466:2717–23.

24\.\ Fitzgerald JD, Orav EJ, Lee TH, Marcantonio ER, Poss R, Goldman L, Mangione CM. Patient quality of life during the 12 months following joint replacement surgery. Arthritis Rheum. 2004;51:100–9.

\25.\ Fan Y, Jin J, Sun Z, Li W, Lin J, Weng X, Qiu G. The limited use of a tourniquet during total knee arthroplasty: a randomized con-trolled trial. Knee. 2014;21:1263–8.

26\.\ Ejaz A, Laursen AC, Kappel A, Laursen MB, Jakobsen T, Ras-mussen S, Nielsen PT. Faster recovery without the use of a tour-niquet in total knee arthroplasty. Acta Orthop. 2014;85:422–6.

\27.\ Abdel-Salam A, Eyres KS. Effects of tourniquet during total knee arthroplasty. A prospective randomised study. J Bone Joint Surg Br. 1995;77:250–3.

28\ .\ Cotter JT, Nielsen KC, Guller U, Steele SM, Klein SM, Green- block in addition to femoral nerve block for total knee arthro-
grass RA, Pietrobon R. Increased body mass index and ASA plasty. J Arthroplasty. 2014;29:2462–7.
physical status IV are risk factors for block failure in ambula- 31\.\ Sullivan M, Tanzer M, Stanish W, Fallaha M, Keefe FJ, Sim-
tory surgery—an analysis of 9342 blocks. Can J Anaesth. monds M, Dunbar M. Psychological determinants of prob-
2004;51:810–6. lematic outcomes following total knee arthroplasty. Pain.
29\ .\ Gi E, Yamauchi M, Yamakage M, Kikuchi C, Shimizu H, Okada 2009;143:123–9.
Y, Kawamura S, Suzuki T. Effects of local infiltration analgesia 32\.\ Papakostidou I, Dailiana ZH, Papapolychroniou T, Liaropoulos
for posterior knee pain after total knee arthroplasty: comparison L, Zintzaras E, Karachalios TS, Malizos KN. Factors affecting
with sciatic nerve block. J Anesth. 2014;28:696–701. the quality of life after total knee arthroplasties: a prospective
\30.\ Tanikawa H, Sato T, Nagafuchi M, Takeda K, Oshida J, Okuma study. BMC Musculoskelet Disord. 2012;13:116.
K. Comparison of local infiltration of analgesia and sciatic nerve

 

1 Comment
  1. Ernest Baker says

    Hi, this is a comment.
    To get started with moderating, editing, and deleting comments, please visit the Comments screen in the dashboard.

Comments are closed.