Neuraxial anesthesia is frequently used for both labor analgesia and cesarean deliveries. Traditional landmark-based techniques for epidural procedures show a 5% to 8% failure rate when epidural catheters are placed for labor analgesia and then used for cesarean delivery.1 In patients who are morbidly obese, this failure rate may be as high as 42%.2 Patients with a high body mass index (BMI) are particularly susceptible to complications such as inadvertent dural or vascular puncture, and failed neuraxial anesthesia.3 Some of these complications are related to the inability to identify the anatomical landmarks of the spine by palpation, and to predict the distance from the skin to the epidural space.
The use of spinal ultrasound (US) has shown to increase the success rate of epidural catheter placement when compared with a blind technique.4,5 One study in normal-weight pregnant women found excellent correlation between the US-estimated depth (UD) to the epidural space and the actual needle depth (ND) using a single-screen method, with the US assessment performed in the transverse median (TM) plane.6 However, using the same imaging technique in obese women, a second study identified a trend toward underestimation of the ND by US as the depth to the epidural space increased.7 This finding was attributed to increased soft tissue compression by the US probe in women with a high BMI.7 Although the authors acknowledge that less compression would probably lead to more precise measurements, the image quality of the ligamentum flavum–dura mater unit in the TM plane is frequently inferior to that in the paramedian sagittal oblique (PSO) plane.8,9 Therefore, changes to the US scanning technique may be necessary if less subcutaneous compression is warranted. The logical step would be to pursue the assessment in the PSO plane, which has been shown to provide excellent imaging quality and a greater visibility of the relevant structures because of the larger acoustic window.8
Although the use of the PSO plane seems the logical choice, a midline rather than a paramedian needle approach is most frequently used during lumbar epidural catheter placement in obstetric anesthesia. It is unknown whether the estimated UD to the epidural space obtained in the PSO plane can be used for performing midline punctures. Borges et al.9 showed that the depth to the epidural space measured in the PSO plane is similar to that obtained in the TM plane and suggested that the measurements could be used interchangeably. However, this impression has not been tested in a clinical study while performing actual epidural needle punctures.
We are unaware of published data comparing the US measurements in the PSO plane and the actual needle distance to the epidural space in midline punctures in women who are obese. We hypothesized that the superior acoustic window in the PSO plane compared with the TM plane allows for a more precise estimate of the distance to the ligamentum flavum–dura mater while performing midline punctures in obese pregnant women.
After approval by the Research Ethics Board at Mount Sinai Hospital (Toronto, Canada), we conducted an observational cohort study. Laboring women requesting epidural analgesia or women undergoing scheduled cesarean delivery under combined spinal-epidural (CSE) anesthesia were recruited. Written informed consent was obtained from all enrolled patients. The inclusion criteria included a current BMI of ≥30 kg/m2. Patients were excluded if they had marked spinal deformities or a history of spinal instrumentation. Patient recruitment occurred only when the investigators were available. Obesity was classified according to the World Health Organization (WHO) categories (class I = BMI 30–34.9 kg/m2, class II = BMI 35–39.9 kg/m2, and class III = BMI ≥40 kg/m2). We planned to recruit 20 patients in each WHO category of obesity.
All patients underwent US imaging of the lumbar spine just before the epidural or CSE technique placement. The US scanning was performed in a nonsterile manner with the patient in the sitting position by one of the investigators (JCAC, MB, CA) who have 7 years experience in the use of spinal US. A portable US system with a 5-2 MHz curved array probe (M-Turbo™; SonoSite Canada Inc., Markham, ON, Canada) was used for US assessment, and the built-in caliper on the US machine was used for all measurements. Through a systematic scanning protocol, the PSO plane was scanned first followed by the TM plane.
Scanning in the PSO Plane
All patients were initially scanned in the left PSO plane to identify the upper border of the sacrum (Fig. 1). The US probe was placed in a longitudinal orientation over the sacral area, 1 to 2 cm lateral to the neuraxial midline, and tilted slightly oblique to the midline to obtain a PSO view of the vertebral canal. The probe was then moved cephalad to identify the L3-4 interspace by counting consecutive interspaces. If the image quality at L3-4 was poor, the L2-3 or L1-2 interspace was used for epidural placement. In the PSO plane, we identified the following structural elements: (1) the posterior complex (a linear hyperechoic structure consisting of ligamentum flavum, epidural space, and posterior dura mater), (2) the anterior complex (a linear hyperechoic structure consisting of anterior dura mater, posterior longitudinal ligament, and posterior aspect of the vertebral body or intervertebral disk), and (3) the intrathecal space (dural sac), which appears uniformly anechoic between the posterior and anterior complexes.10 The posterior and anterior complexes are referred to by other authors as simply the ‘‘ligamentum flavum–dura mater unit” and the ‘‘vertebral body,’’ respectively.11 We have adopted this terminology throughout the article. We also sought to identify the laminae in the PSO view.
Once the optimum US image was obtained, the interspace was centered on the US screen. The tissue compression by the US probe was released gradually as much as possible until the structures were still visible and the image was then frozen. The distance from the skin to the inner aspect of the ligamentum flavum–dura mater unit was measured (US depth in the PSO plane, UD/PSO).
Scanning in the TM Plane
Thereafter, the scanning was performed at the same lumbar interspace by placing the US probe in transverse orientation and scanning in a plane perpendicular to the long axis of the spine to obtain a TM view of the vertebral canal (Fig. 2). The ligamentum flavum–dura mater unit, the vertebral body, the transverse processes, and the articular processes were identified. To determine the x-intercept of the insertion point, the tip of the spinous process was identified and centered on the screen, and the skin was marked at the midpoint of the cephalad aspect of the probe. The y-intercept of the insertion point was then determined by moving the probe cephalad or caudad to locate the interspace, and the skin was marked at the midpoint of the right side of the probe. The optimal insertion point for the puncture was determined by the intersection of the 2 skin marks extended medially (x-axis) and caudally (y-axis) in the horizontal and vertical planes, respectively. Upon identification of the interspace structures, the same procedure of releasing the tissue compression and measurements were performed. The distance from the skin to the inner aspect of the ligamentum flavum–dura mater unit was measured (US depth in TM plane, UD/TM), as well the distance from the skin to the tip of the spinous process. Finally, the depth from the skin to an imaginary line connecting the most posterior surface of the transverse processes (TP) on each side (US depth to TP, UD/TP) was measured. The distance from the skin to the tip of the spinous process was considered a direct surrogate of the thickness of the adipose tissue.
During the scanning protocol and at the time of measuring, images in both planes were obtained and recorded (PSO, TM) for later assessment. The image quality was rated as good (all structures visualized with clear demarcation of both the ligamentum flavum–dura mater unit and vertebral body), poor (most but not all structures visible with poor demarcation), or inconclusive (no discernible structures). The images were rated in pairs (TM, PSO) after completion of the scanning by at least 2 of the investigators (JCA, MB, CA), and any discrepancies were resolved by discussion. A research assistant recorded the total duration of the US scanning.
Either an experienced obstetric anesthesia fellow (with at least 6 months within the fellowship program) or a staff anesthesiologist performed the epidural procedure via a midline approach. They were offered information about the interspace and insertion point. The investigator disclosed an estimated depth to the epidural space, which was the shortest distance measured by the US assessment rounded to the lowest integer number in centimeters (e.g., if UD = 7.6 cm, the estimated depth disclosed was 7.0 cm). The physicians performing the epidural/CSE procedure were asked to insert the Tuohy needle via a midline approach perpendicular to the skin, or with minimal angle if required, to ensure consistent depth measurements with the US-measured depths.
The epidural/CSE procedures were executed in the conventional manner under sterile conditions using a 17-gauge Tuohy epidural needle after infiltration of the skin with 2% lidocaine. The patient was positioned reproducing the position used during the US assessment. The epidural space was confirmed by the loss-of-resistance technique to either air or saline. Once the epidural space was identified, a sterile marking pen was used to mark the Tuohy needle at the skin, and the ND was measured using a ruler to the nearest millimeter. In the CSE technique for cesarean delivery, the epidural space was identified as above, and a “needle-through-needle” technique was performed with a 27-gauge Whitacre needle. After intrathecal drug administration, the spinal needle was removed and a 19-gauge uniport, wire-embedded epidural catheter (Arrow FlexTip Plus; Arrow International, Reading, PA) was inserted approximately 5 cm into the epidural space.
The primary outcome was the precision of the estimated depth to the ligamentum flavum determined by US in the PSO plane (UD/PSO) compared with the depth to the epidural space during the actual needle insertion via midline (ND), as well as compared with the estimated depth by US in the TM plane (UD/TM). The differences between measurements were evaluated using the Bland-Altman analysis. Secondary outcomes included the following: (a) precision and bias of the US estimates evaluated through the concordance correlation coefficient (CCC); (b) quality of imaging (TM versus PSO); (c) duration of US scanning and epidural/CSE procedure from skin infiltration to catheter insertion; (d) accuracy of the insertion point as determined by the need to redirect (redirection defined as any change in needle insertion trajectory that did not involve complete withdrawal of the needle from the patient’s skin) or reinsert the needle in the same interspace (reinsertion defined as complete withdrawal of the Tuohy needle to the patient’s skin followed by a new attempt in the same interspace); (e) correlations among BMI, US distance from skin to the tip of the spinous process, and ND; (f) patient’s comfort during the epidural procedure (verbal rating scale: 0–10; where 0 = no pain, 10 = worst pain imaginable), assessed after labor pain had subsided in labor epidural procedures and before skin incision in CSE procedures for cesarean deliveries; and (g) complications (vascular or dural puncture, paresthesias). A research assistant not involved in the US scanning or the epidural/CSE placement recorded all outcomes.
Statistical Analysis and Sample Size Calculation
Descriptive statistics, including the mean and SD, range, or median and interquartile range were used for continuous variables, whereas frequency and percentages were used for discrete variables.
The Bland-Altman analysis was performed to place the magnitudes of the differences between 2 measurements in a more clinical context.12,13 In addition, we estimated the 95% limits of agreement (LOA) for the differences, which represents differences likely to arise between the 2 measurements with a 95% probability. In our study, instead of using the mean between ND and UD, we considered the actual ND as the true value of the measurement for the x-axis in the Bland-Altman analysis. Although we acknowledge that the ND might be related to the differences between such measurements (UD and ND), we also expected and assumed an increase in the variability of the differences as the magnitude of the ND measurement increases. In clinical context, as the distance to the epidural space (ND) increases, the error in the US measurement will increase. Under these circumstances, logarithmic (log) transformation of both measurements before analysis would validate this assumption. The LOA derived from log-transformed data were back-transformed (antilog) to give the limits for the ratio of the actual measurements. We compared LOA obtained from the antilog transformation with the predicted 95% confidence intervals (CIs) obtained from a linear regression model, which considered the ND as the outcome variable and the UD as the predictor variable. Furthermore, we determined the 95% prediction intervals as the 95% lower confidence limit for the 5th percentile and the 95% upper confidence limit for the 95th percentile of the differences. The assumption of normal distribution of the differences was verified with the Shapiro-Wilk W test for normal data.
We also used the CCC to estimate the precision and bias of UD/PSO.14,15 The CCC combines measurements of both precision and bias to determine how far the observed data deviate from the line of perfect concordance (the line at 45° on a scatterplot between UD/PSO and ND). The coefficient increases in value as a function of the nearness of the data’s reduced major axis to the line of perfect concordance (the bias of the data) and of the tightness of the data about its reduced major axis (the precision of the data). The precision corresponds to the Pearson correlation coefficient, which as a sole measure could potentially overestimate agreement; therefore, it is adjusted by the bias correction factor. Based on linear regression analysis, we also compared and tested the CCC across the 3 obesity groups (classes I, II, and III).
The association of ND with the obesity group, BMI, and US distance from the skin to the tip of the spinous process was modeled through multiple linear regression. The quality of the US images in the PSO and TM views was analyzed using Fisher exact test, and a P value <0.05 was deemed significant. We reduced image-quality data to 2 groups for this analysis (good and poor/inconclusive).
The sample size was based on data from the study by Balki et al.7 In that study, the correlation coefficient between the ND and UD using the TM approach was 0.85 (null hypothesis) in a subgroup of obese parturients. We hypothesized a coefficient of at least 0.7 (alternative hypothesis) when comparing ND and UD in the PSO approach in a cohort including a greater number of morbidly obese parturients. Using the Fisher z transformation of the correlations, a 2-sided test, 80% power, and an α error of 0.05, a sample size of 54 patients was deemed adequate. We decided to collect a balanced and representative sample of 60 obese parturients by recruiting 20 subjects in each WHO category of obesity. However, the primary outcome and main analysis was performed as 1 sample of 60 subjects. Statistical analysis was performed with STATA Statistical Software for Macintosh, release 12.1 (StataCorp, College Station, TX).
Sixty patients, 20 from each WHO obesity class, were recruited from August to December 2010. Data from all 60 patients were analyzed; there were no protocol violations. The women participating in this study presented a mean (SD) age of 33.2 (5.3) years, height of 164.9 (7.2) cm, weight of 107.9 (23.8) kg, and BMI of 39.6 (7.9) kg/m2 in a range of 30.4 to 66.2 kg/m2.
The characteristics of the neuraxial procedures and US measurements are presented in Table 1. Labor epidural analgesia and CSE anesthesia for cesarean delivery were performed in 32 patients (53.3%) and 28 patients (46.7%), respectively. The most prevalent lumbar vertebral interspace for the neuraxial procedures was L3-4 (n = 48), followed by L2-3 (n = 11), and L1-2 (n = 1). All CSE procedures for cesarean delivery were performed at the L3-4 interspace.
The Bland-Altman analysis showed a mean difference of −0.05 cm between UD/PSO and UD/TM. Similarly, a mean difference of 0.05 cm was obtained between ND and UD, in both the PSO and TM planes (Fig. 3). All the differences were normally distributed (Shapiro-Wilk test, all P > 0.27). In Table 2, we present the 95% LOA between the actual ND and UD, the 95% LOA after logarithmic transformation of the measurements, the 95% prediction interval analysis, and the 95% CI of the predicted interval obtained from a linear regression model. Whereas the LOA were approximately ±1 cm, the analysis of precision (95% prediction intervals) demonstrated wider limits (likely because of the small sample size). The LOA derived from log-transformed data were back-transformed to give limits for the ratio of the actual measurements. These limits showed that in 95% of the observations, UD (both PSO and TM) was between 0.86 and 1.17 times the ND. Thus, the UD measurement may be as much as 14% shorter to 17% longer than the ND measurement. The linear regression model predicted similar magnitudes of error for UD/PSO: 11% shorter and 16% longer.
The calculated CCC values between the different US planes and the ND were high and significant (UD/PSO: 0.90, asymptotic 95% CI: 0.85, 0.95; UD/TM: 0.90, asymptotic 95% CI: 0.86, 0.95; UD/TP: 0.87, asymptotic 95% CI: 0.81, 0.93; UD/PSO versus UD/TM: 0.96, asymptotic 95% CI: 0.94, 0.98; all P values <0.0001). The CCC remained constant across the 3 obesity groups (classes I, II, and III). The hypothesis test resulted in nonsignificant difference (F test; P = 0.18).
The quality of the US images in both the PSO and TM planes is presented in Table 3. The PSO and TM planes depicted good quality images in 52 of 60 cases (86.7%) and in 41 of 60 cases (68.3%), respectively (Fisher exact test, P = 0.028). The main discrepancy in image quality was noted in 17 poor-quality images seen in the TM plane, 13 of which were deemed of good quality when viewed in the PSO plane.
The duration of the US scanning and of the neuraxial procedure was less than 5 and 9 minutes, respectively, in 75% of the cases (Table 1). During the neuraxial procedure, the need for redirection was ≤2 in 66.7% of the cases. The reinsertion of the epidural needle at the same interspace was needed in 15% of the cases. One patient (obesity class III) who had poor-quality imaging in the PSO plane (UD/PSO: 7.47 cm) had a loss of resistance at 8.5 cm (ND), and required 4 reinsertions in the same interspace.
The US distance from the skin to the tip of the spinous process (mean [SD] = 2.77 [1.02] cm) and the obesity group were statistically significant predictors of the ND in a multivariable regression model (R2 = 0.70, P < 0.0001), which means that 70% of the variability of ND was explained by the obesity group and skin–spinous process distance. This US distance (skin–spinous process) demonstrated a moderate correlation/substantial relationship with BMI (Pearson r = 0.50, P < 0.0001).
All of the recruited patients had successful neuraxial blocks. There were no vascular punctures or paresthesias; 1 unintentional dural puncture occurred with the Tuohy needle in a patient with a BMI of 39.8 kg/m2 (ND: 8.2 cm; UD/PSO: 8.14 cm; UD/TM: 8.05 cm).
Our study did not confirm the superior precision of the estimated US depth to the epidural space obtained in the PSO plane as compared with the estimate in the TM plane. However, in the context of our modified technique of applying the least pressure to the subcutaneous tissues, our study confirmed that the distances obtained both in the PSO plane and in the TM plane can be used interchangeably for midline punctures in obese pregnant women. The quality of imaging in the PSO plane allowed a detailed ultrasonographic assessment of the spine landmarks in the majority of the cases, even though some of the images in the TM plane were poor or inconclusive. Consequently, the US assessment in the PSO plane may be of significant clinical value in the obese population when imaging in the TM plane is suboptimal.
The proposed modified scanning technique with alleviation of the subcutaneous tissue compression by the US probe proved to be effective in minimizing the underestimation of the distance to the epidural space in comparison to a previous study.7 Based on the Bland-Altman analysis, we obtained a mean difference of 0.05 cm between the ND and the UD in the PSO plane, which resembles the mean difference in non-obese pregnant women.6 From a clinical perspective, when performing the measurements in the PSO plane, we should expect on average an almost perfect agreement of the estimation with the actual needle distance (mean difference between ND and UD/PSO: 0.05 cm), but practically we should still consider a possible estimation error of ±1 cm (95% LOA: ±1 cm). Our data suggest that UD estimation of the depth to the epidural space from the skin may be 14% shorter to 17% longer than the actual ND. This analysis confirms our assumption of an increase in the US estimation error (wider 95% CI) as the actual distance to the epidural space (ND) increases.
Furthermore, we described that the imaginary line interconnecting the posterior aspects of the TP as seen in the TM plane can be used as a surrogate for the ligamentum flavum–dura mater unit. The possibility of using estimates in different planes and to different landmarks may be very useful in conditions of poor sonoanatomy.
The thickness of the subcutaneous layer (distance from the skin to the tip of the spinous process) and the obesity group were both predictive and able to explain 70% of the variability of the actual distance to the epidural space. The distance skin–spinous process was measured with the least possible compression that would still allow good visibility under US. The mean value was similar to a previous study performed in an obese population (2.7 cm).7 This finding reinforces the need for considering the compression factor while performing prepuncture US in obese pregnant women. Although this is a dynamic maneuver that aims to increase the precision of the US estimate, it remains to be evaluated systematically for generalizability.
The ultrasonographic imaging of the spine is greatly hindered by the vertebral bony structures surrounding the spinal canal with consequent reduction of signal quality. The spinous process and lamina cause an intense reflection and refraction of the US beam resulting not only in a reduction of the ratio between the interspace “acoustic window” and the “acoustic shadow” from the bony structure, but also in a decrease of the visibility of deeper soft tissue structures such as the ligamentum flavum.8 In our study, the quality of the images obtained in the PSO plane allowed a good assessment of the ligamentum flavum–dura mater unit as the key element for estimating the depth to the epidural space. However, in many of our patients, we were also able to capture good images in the TM view, which emphasizes the point that US scanning is a dynamic assessment that incorporates information from both planes.
Our study has some limitations. The US measurements exhibit 95% LOA as wide as previously described in obese pregnant women (±1 cm). The inability to precisely measure the angle of the US beam and reproduce it with the needle trajectory may account for some of this discrepancy. Deviation of the needle trajectory from that of the US beam can result in differences between the estimated US and the actual needle depth. This might become more relevant as the obesity increases from class I to class III. The “compression factor” may also account for the wide LOA. The distance from the skin to the spinous process may serve as an estimate of the compression factor. We did not measure with maximal compression, which would have possibly given more information on the relative effect of the compression. This hypothesis remains to be tested.
Although the anesthesiologists performing the neuraxial techniques were blinded to the actual US measurements, we did provide them with the location of the optimal insertion point and an approximate distance to the epidural space. This disclosure might have introduced the potential for bias; however, we think that it was ethically important to disclose this estimate based on the current knowledge of spinal US and its beneficial contribution in difficult epidural procedures. Another source of bias may be related to the nonblinded assessment of image quality, which was performed in pairs (PSO, TM) by at least 2 investigators. Although the classification is simple and has been used for describing the sonoanatomy of the lumbar spine in pregnancy, the process could have favored the PSO plane as being part of the study hypothesis.
Finally, although the investigators performing the US assessments have substantial experience in spinal US for neuraxial techniques, thus allowing robust internal validity, interrater reliability was not assessed.
In summary, our study adds an important piece of evidence to the current knowledge on the usefulness of prepuncture US in a challenging population such as obese pregnant women. We demonstrated that the ultrasonographic estimates of the depth to the epidural space obtained in the PSO plane can be used for midline neuraxial punctures and are comparable to those obtained in the TM plane. Although we attempted to demonstrate that the PSO view depicts better quality of imaging than the TM view, especially when the latter precludes the distinct visualization of the ligamentum flavum–dura mater unit, a larger sample size and blinding are needed to confirm this outcome.
Name: Jagpaul S. Sahota, MD.
Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript.
Attestation: Jagpaul S. Sahota has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.
Name: Jose C. A. Carvalho, MD, PhD.
Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript.
Attestation: Jose C. A. Carvalho has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.
Name: Mrinalini Balki, MBBS, MD.
Contribution: This author helped design the study, conduct the study, and write the manuscript.
Attestation: Mrinalini Balki has seen the original study data and approved the final manuscript.
Name: Niall Fanning, MD.
Contribution: This author helped conduct the study and write the manuscript.
Attestation: Niall Fanning has seen the original study data and approved the final manuscript.
Name: Cristian Arzola, MD, MSc.
Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript.
Attestation: Cristian Arzola has seen the original study data, reviewed the analysis of the data, approved the final manuscript, and is the author responsible for archiving the study files.
This manuscript was handled by: Cynthia A. Wong, MD.
1. Riley ET, Papasin J. Epidural catheter function during labor predicts anesthetic efficacy for subsequent cesarean delivery. Int J Obstet Anesth. 2002;11:81–4
2. Hood DD, Dewan DM. Anesthetic and obstetric outcome in morbidly obese parturients. Anesthesiology. 1993;79:1210–8
3. Whitty RJ, Maxwell CV, Carvalho JC. Complications of neuraxial anesthesia in an extreme morbidly obese patient for cesarean section. Int J Obstet Anesth. 2007;16:139–44
4. Grau T, Leipold RW, Conradi R, Martin E, Motsch J. Efficacy of ultrasound imaging in obstetric epidural anesthesia. J Clin Anesth. 2002;14:169–75
5. Vallejo MC, Phelps AL, Singh S, Orebaugh SL, Sah N. Ultrasound decreases the failed labor epidural rate in resident trainees. Int J Obstet Anesth. 2010;19:373–8
6. Arzola C, Davies S, Rofaeel A, Carvalho JC. Ultrasound using the transverse approach to the lumbar spine provides reliable landmarks for labor epidurals. Anesth Analg. 2007;104:1188–92
7. Balki M, Lee Y, Halpern S, Carvalho JC. Ultrasound imaging of the lumbar spine in the transverse plane: the correlation between estimated and actual depth to the epidural space in obese parturients. Anesth Analg. 2009;108:1876–81
8. Grau T, Leipold RW, Horter J, Conradi R, Martin EO, Motsch J. Paramedian access to the epidural space: the optimum window for ultrasound imaging. J Clin Anesth. 2001;13:213–7
9. Borges BC, Wieczorek P, Balki M, Carvalho JC. Sonoanatomy of the lumbar spine of pregnant women at term. Reg Anesth Pain Med. 2009;34:581–5
10. Chin KJ, Karmakar MK, Peng P. Ultrasonography of the adult thoracic and lumbar spine for central neuraxial blockade. Anesthesiology. 2011;114:1459–85
11. Carvalho JC. Ultrasound-facilitated epidurals and spinals in obstetrics. Anesthesiol Clin. 2008;26:145–58
12. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1:307–10
13. Bland JM, Altman DG. Measuring agreement in method comparison studies. Stat Methods Med Res. 1999;8:135–60
14. Lin LI. A concordance correlation coefficient to evaluate reproducibility. Biometrics. 1989;45:255–68
© 2013 International Anesthesia Research Society
15. King TS, Chinchilli VM. A generalized concordance correlation coefficient for continuous and categorical data. Stat Med. 2001;20:2131–47