The great improvements made over the years in the equipment and drugs used for epidural anesthesia (1) have not been matched by a corresponding advancement in the technical aspects of the technique, parti cularly in the identification of the epidural space. Although epidurography (2) and nerve stimulation (3) have been used to evaluate the epidural catheter position, the palpation of anatomical landmarks and the loss-of-resistance (LOR) technique remain the standard procedures to locate the epidural space. As a result, epidural anesthesia continues to be a relatively blind anesthetic technique.
Ultrasound (US) imaging of the lumbar spine in different scanning planes facilitates the identification of the landmarks necessary for appropriate epidural space location in pregnant patients (4,5). Grau et al. (6) compared the quality of images obtained with the transverse and median longitudinal approaches with those obtained with the paramedian longitudinal approach, and suggested that the latter was superior, probably due to a wider acoustic window.
We believe that the transverse approach provides a good quality image, and is easier to perform and reliable enough for acquiring proficiency. The goal of our study was to estimate the diagnostic accuracy and precision of the transverse approach, used as a “single-screen” method, to facilitate labor epidurals.
We followed the Standards for Reporting of Diagnostic Accuracy checklist (7) to present our study.
The research ethics board of our institution approved the study, and patients were required to give written informed consent. Term laboring patients requesting epidural analgesia were approached for enrollment. Those with previous spinal surgery and marked spinal bony deformity (e.g., scoliosis, as confirmed clinically by two anesthesiologists) were excluded. Patient recruitment was prospective, and sampling occurred only during times the investigators were available. Data collection was planned as a cohort study.
To achieve epidural needle placement successfully, we performed prepuncture US diagnostic imaging (index test) to estimate the depth to the epidural space. The actual puncture depth during the needle insertion with the LOR technique was considered the reference standard test.
Imaging and Measurement
The US imaging was performed in a nonsterile manner with the patient sitting. The L3–4 interspace was identified by palpation, as per Tuffier's line. Spine imaging of that area was then performed using a portable Titan Ultrasound System equipped with a 2–5 MHz curved array probe (Sonosite Canada Inc.). The best possible image was captured by positioning the probe perpendicular to the long axis of the lumbar spine (transverse approach) (6). The spinous process, corresponding to the midline of the spine, was identified as a small hyperechoic (bright) signal, immediately underneath the skin, and continuing as a long triangular hypoechoic (dark) acoustic shadow (Fig. 1). The probe was then moved slowly cephalad or caudad to capture a view of the upper or lower intervertebral space, visualized as an acoustic window containing the vertebral body, dural sac, and ligamentum flavum–dura mater unit (Fig. 2). We decided to use the ligamentum flavum–dura mater unit as a reference, as opposed to the ligamentum flavum and the dura mater separately, because with this equipment, they are most often visualized as one structure. The image was frozen when the components of our single-screen method containing the structures of the interspace were seen. At that moment, with the transducer kept steady, two marks were drawn on the skin: one coinciding with the center of the upper horizontal surface of the probe (midline), and the other coinciding with the midpoint of the right lateral vertical surface of the probe (interspace). The puncture site was determined by the intersection of the two marks on the skin on the vertical and horizontal planes. With the aid of a built-in caliper, we measured the ultrasound depth (UD), i.e., the depth to reach the epidural space from the skin to the inner surface of the ligamentum flavum–dura mater unit (Fig. 3).
The epidural needle insertion was then performed in the conventional manner, under sterile conditions, using a 17-gauge × 8.89 cm epidural needle with cm markings. The needle was introduced at the predetermined insertion point obtained by US, on a perpendicular plane to the skin surface, reproducing the direction of the US beam. If necessary, the needle was redirected at a steeper angle. The epidural space was confirmed by a LOR to air or saline technique in the conventional manner. At this time, the actual distance to the epidural space was measured to the nearest half-centimeter of the marked epidural needle (needle depth/ND). The study investigators performed the US imaging, and experienced anesthesiologists who had full access to the US data inserted the epidurals.
Throughout the study, we documented the quality of the anatomical landmarks (good or poor) as determined by palpation during the needle insertion by the epidural performer, and the quality of US imaging as determined by the study investigator. In addition, the number of needle redirections (different angles) and reinsertions (different skin perforations) were recorded.
Descriptive statistics were calculated using means and standard deviations for continuous data and percentages for discrete variables.
We used the concordance correlation coefficient (CCC) to determine the degree of agreement between UD and ND (8,9). This coefficient estimates agreement between two methods by measuring the variation of their linear relationship from the 45° line through the origin (perfect agreement). It not only measures how far each observation deviates from the line that best fits the data (precision), but also how far this line deviates from the 45° line through the origin (accuracy) (10). The precision corresponds to the Pearson correlation coefficient which, as a sole measure, could potentially over-estimate agreement (11). To visually represent what CCC evaluates, UD was plotted against ND, and the line that best fitted the data and the line of perfect agreement were estimated.
The Bland–Altman analysis was performed to place the magnitudes of the differences between the two measurements in a more clinical context (12,13). This approach shows the graphical presentation of agreement, plotting the differences between UD and ND against the means of the two measures for each patient. In addition, we estimated the 95% limits of agreement for the differences, which represents differences likely to arise between the two measures with a 95% probability. The assumption of normal distribution of the differences was checked by the Shapiro-Wilk W test for normal data. Interobserver variability was not possible to quantify, as only one investigator assessed the UD in each patient.
The sample size required 48 patients to detect a maximal difference of 1% in the actual CCC between UD and ND, assuming 80% power and α error of 5%. We based our calculation on a minimal CCC of 0.85, and an estimated standard deviation of 0.03. We elected to evaluate 60 patients (i.e., 25% more) to compensate for possible protocol violations during the study period. All statistical calculations were performed with STATA statistical software, version 8.0 SE.
We prospectively collected the data of 61 patients between August and November of 2005. The patients ranged in age from 15 to 43 yr, and in body mass index (BMI) from 22.2 to 42.5 (mean 29.7 ± 4.79). Fifteen patients (24.6%) had a BMI ≥ 35 (Table 1).
The quality of anatomical landmarks was rated as good in 81% of the patients by palpation, and in 95% or more by US (Table 2). During the epidural insertion, which was done according to the US-determined insertion point, there were no reinsertions of the epidural needle in 91.8% of the patients, and there was no need to redirect the needle in 73.7%. Successful identification of the epidural space was accomplished with two or less redirections in 96.7% of the cases.
The UD was 4.66 ± 0.68 cm (range 3.43–6.91 cm), whereas the ND was 4.65 ± 0.72 cm (range 3.5–6.5 cm) (Table 1). The graphical representation of UD versus ND shows that the best-fit line deviates little from the perfect agreement (Fig. 4). The CCC was 0.881 (95% CI 0.820–0.942), with an accuracy of 0.999 and a precision of 0.882. The slope was 0.951, and the intercept was at 0.236.
The Bland–Altman analysis showed that the mean difference between UD and ND was 0.010 ± 0.345 cm. The 95% limits of agreement for the difference between the two measurements were −0.666 to 0.687 cm (Fig. 5, Table 3).
We observed a high success rate of the US-determined puncture site, and a very good agreement between the distances from the skin to the epidural space determined by both the US and the needle puncture.
Although anatomical landmarks were evaluated as good by palpation in only 81% of the patients, the puncture site as determined by US was successful in 91.8% of the cases. We suggest that US may be helpful in reducing the number of attempts during epidural insertion compared with the conventional palpatory technique; however, the current study was not designed to prove this hypothesis, and therefore, further studies are warranted.
The agreement between UD and ND was statistically significant and clinically important based on both components (accuracy and precision) of the CCC. From our results, the expected depth can be predicted within a range of ±7 mm with a 95% probability, similar to the findings of a previous study in the obstetric population (4). Although our data suggest that this agreement is not affected by BMI, our study was under-powered to make a conclusive statement in that respect.
The main difference between our study and those of Grau et al. (4–6) is that we used only one screen (transverse approach) to obtain the prepuncture information (UD and optimal puncture site), without exploring other approaches (median and paramedian longitudinal) and measurements (e.g., angle), as proposed by those authors. It is our impression that the use of many components in the US diagnostic imaging would make the technique less likely to be adopted by anesthesiologists, as the additional components make it more difficult and time-consuming to teach the beginner. Additionally, the preferred midline approach for needle insertion in obstetric lumbar epidurals makes the US transverse approach, rather than the paramedian approach, the logical choice.
Our results and conclusions are subject to potential bias (14). The information on the index test (US) was available to the performer of the reference test (needle insertion), whose measurement was only an approximation to the nearest half-centimeter on the epidural needle. However, it is important to note that although the sonoanatomy of the spine is a very important tool to facilitate epidural insertions, it does not preclude the need for the LOR technique.
With regard to the applicability of the results, the depth of the epidural space found in our study corresponds to the results found in the literature (15). With the exception of patients with spinal deformities (excluded from the study), the spectrum of our patients was similar to that usually encountered in a busy tertiary care hospital (16). The subgroup of Obesity Class II or more (17) was also represented.
In summary, the US single-screen method using the transverse approach of the lumbar spine provides reliable information regarding the landmarks required for labor epidurals. Although further studies are warranted, this technique is feasible, fast to execute, reproducible, and easy to teach and learn. We believe that it is a useful tool to facilitate identification of the epidural space.
We thank Vincent Chan, MD, FRCPC, for his valuable guidance in the early planning of this study.
1. Frolich MA, Caton D. Pioneers in epidural needle design. Anesth Analg 2001;93:215–20.
2. Collier CB. Why obstetric epidural fail: a study of epidurograms. Int J Obst Anesth 1996;5:19–31.
3. Tsui BC, Gupta S, Finucane B. Determination of epidural catheter placement using nerve stimulation in obstetric patients. Reg Anesth Pain Med 1999;24:17–23.
4. Grau T, Leipold RW, Conradi R, et al. Efficacy of ultrasound imaging in obstetric epidural anesthesia. J Clin Anesth 2002;14:169–75.
5. Grau T, Leipold RW, Conradi R, et al. Ultrasound imaging facilitated localization of the epidural space during combined spinal and epidural anesthesia. Reg Anesth Pain Med 2001;26:64–7.
6. Grau T, Leipold RW, Conradi R, et al. Paramedian access to the epidural space: The optimum window for ultrasound imaging. J Clin Anesth 2001;13:213–17.
7. Bossuyt PM, Reitsma JB, Bruns DE, et al; for the STARD Group. Towards complete and accurate reporting of studies of diagnostic accuracy: the STARD initiative. BMJ 2003;326:41–4.
8. Lin L. A concordance correlation coefficient to evaluate reproducibility. Biometrics 1989;45:255–68.
9. Lin L, Hedayat AS, Sinha B, Yang M. Statistical methods in assessing agreement: Models, issues, and tools. J Am Stat Assoc 2002;97:257–70.
10. King TS, Chinchilli VM. A generalized concordance correlation coefficient for continuous and categorical data. Stat Med 2001;20:2131–47.
11. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307–10.
12. Bland JM, Altman DG. Measuring agreement in method comparison studies. Stat Methods Med Res 1999;8:135–60.
13. Cepeda MS, Africano JM, Polo R, et al. Agreement between percentage pain reductions calculated from numeric rating scores of pain intensity and those reported by patients with acute or cancer pain. Pain 2003;106:439–42.
14. Whiting P, Rutjes A, Dinnes J, et al. A systematic review finds that diagnostic reviews fail to incorporate quality despite available tools. J Clin Epidemiol 2005;58:1–12.
15. Segal S, Beach M, Eappen S. A multivariate model to predict the distance from the skin to the epidural space in an obstetric population. Reg Anesth 1996;21:451–5.
16. Ehrenberg H, Dieker L, Milluzzi C, Mercer B. Prevalence of maternal obesity in an urban center. Am J Obstet Gynecol 2002;187:1189–93.
© 2007 International Anethesia Research Society
17. ACOG Committee on Obstetric Practice. Obesity in pregnancy. Obstet Gynecol 2005;106:671–5.