Epidural needle placement is performed most commonly using an anatomical landmark-based technique for identifying the chosen intervertebral level and site of insertion. This procedure can be difficult for some subjects, such as the obese. Typically, needle insertion is performed using the loss-of-resistance to saline/air, which is a “feel as you go” method to ensure a proper track toward the ligamentum flavum and epidural space. Loss-of-resistance to saline/air is the gold standard method for identifying entry into the epidural space.1

Ultrasound pre-puncture scanning can identify the intervertebral level2,3 and the midline. As a means to improve successful placement, it can also provide information about best angle, direction of approach, and depth to the epidural space.4-7 The ultimate goal is to plan the trajectory of the needle to achieve successful needle placement in one single attempt without redirections. This is true for both epidural and combined spinal-epidural (CSE) anesthesia. For simplicity in this article, we use the term epidural needle insertion for both cases, although small differences exist as explained in the methods.

In their efforts to achieve the ultimate goal, investigators have described a two-operator real-time ultrasound-guided technique of epidural needle placement.8,9 In this scenario, one operator would perform a midline epidural needle insertion while observing its entry into the epidural space on a paramedian ultrasound image from a second operator. A single-operator in-plane ultrasound-guided epidural needle insertion method has been described,10 but it is a free-hand ultrasound technique, i.e., the needle trajectory cannot be superimposed on the ultrasound image for planning before insertion. The free-hand technique allows flexibility in scanning but requires considerable operator expertise.11A fixed needle angle facilitates pre-puncture determination of the depth of the needle insertion from a simple measurement along this path, as shown in the ultrasound image as a superimposed line. This superimposed needle trajectory line also helps the anesthesiologist to choose a needle insertion site and angle suitable for a direct path toward the epidural space.

Herein, we present the findings of a single-arm prospective observational cohort study of a systematic method both for ultrasound pre-puncture scanning to identify the lumbar level for selection of the puncture site and for real-time guidance of needle insertion. The primary objective was to determine the feasibility and practicality of performing this aim-and-insert technique with a single operator. Secondary objectives were to compare ultrasound and palpation-based identification of intervertebral space count, to study the effect of transducer pressure on measured distances in the lumbar paramedian plane, to compare the straight-down depth to the epidural space with the diagonal needle trajectory and the relationship to the needle guide geometry, and, finally, to compare the site of insertion selected in the pre-puncture examination with the actual insertion site. We expected the secondary objectives would provide insight into potential limitations of the proposed method.

Methods

Subject selection

This study was approved by the Children and Women’s Health Centre of British Columbia and the University of British Columbia’s clinical research ethics boards (CW05-0262 / H05-70409). The subjects were recruited throughout March 2008 to May 2008 using informed written consent. Exclusion criteria included the inability to speak English and the standard contraindications for spinal anesthesia. The subjects’ age, weight, and height were collected and the body mass index was calculated. As this is a feasibility study, test subjects were those scheduled for Cesarean delivery so that labour pains did not disrupt the time or subject stability needed to learn the new technique. A CSE procedure was used for such subject. A sample size of 20 was chosen because a minimum of ten subjects was found to give an 85% success rate4 for other ultrasound guidance studies, and given the nature of our study, we recruited 20 subjects to allow for a more robust statistical analysis.

One sonographer performed all pre-puncture measurements for consistency, and one experienced anesthesiologist (A.K.) performed all epidural needle insertions using ultrasound guidance. All complications were recorded during the course of the procedure (insufficient spinal block, failure to obtain loss-of-resistance, paresthesia, failure to obtain cerebrospinal fluid (CSF) aspirations, and patient discomfort).

Intervertebral level identification

In the pre-surgical assessment area, the anesthesiologist identified and marked the lumbar interspaces, L2-3 and L3-4, using traditional palpation-based techniques and Tuffier’s line.12

Using our designed systematic approach described below, the sonographer (V.L.) also identified and marked the L2-3 and L3-4 interspaces using a curvilinear 1-5 MHz ultrasound transducer (Model C5-1/60) (Ultrasonix Medical Corporation, Richmond, BC, Canada).

First, with each subject placed in the sitting position, the transducer was placed (with the top of the transducer oriented cephalad) on the subject’s back, 5-10 cm lateral to the midline where the 12th rib could be seen ultrasonically superficial to the kidney. Next, the 12th rib was followed medially where the transverse process of L1 could be seen to arise 3-5 cm from the midline, and then the transducer was moved caudad parallel to the midline. The transverse processes (Figure 1a) of L2 to L5 could be identified, and the acoustic shadows appeared like fingers. On moving the transducer medially while maintaining the same orientation, the transverse processes became less defined and were replaced by a wider shadow. This shadow was the intervertebral facet joint (Figure 1b), which looked like a thumb. More medially, the facet joint shadow diminished and a deeper flatter “wavy” pattern was seen (Figure 1c). This was the laminae, situated where the intervertebral spaces were seen as gaps in the wavy line. Next, the transducer was positioned immediately paramedian to the midline spinous processes. By gently angling the transducer towards the midline by 5-10°, the ultrasound beam could be directed at the base of the spinous process where it intersected with the laminae. This technique enabled better definition of the laminae, the foramina, and the ligamentum flavum/dura mater complex (seen as a small white line deep to the foramina) (Figure 1c). This paramedian plane provided the optimum window for ultrasound images of the lumbar anatomy.13 The L3-4 vertebral interspace generally was situated directly medial to the L4 transverse process.

Fig. 1
figure 1

a Transverse processes (finger-like appearance), b facet joints (thumb-like appearance) with the interfacet distance measurement taken from the centre of one facet to the centre of the neighbouring facet, c ultrasound image with dashed needle guide line representing the predicted needle path. The wave-like structures are the laminae, and the ligamentum flavum can be seen as a bright reflector at the base of the laminae with the epidural space underneath. The needle guide line is aimed at the visible portion of the ligamentum flavum. Each large white dot represents a 10 mm marking

Full size image

The differences between the anesthesiologist’s landmarks and the ultrasound skin surface marks were measured using a flexible millimetric ruler. If the difference was greater than one full interfacet distance, it was considered a misidentification of the intervertebral level by the landmark technique, as interspace identification by ultrasound has been shown to be more reliable.3 Ultrasound remains a more reliable count than palpation, although sometimes it is incorrect in identifying the intervertebral space count (inaccurate by one space in 14.7% of cases). The measurement of the intervertebral spacing (Figure 1b) was calculated as the distance between the facets of L2-3 and L3-4 as measured with ultrasound using software callipers. The difference between the left and right facets of L2-3 and L3-4 were also used to assess the possibility of scoliosis.

Geometric measurement for aim-and-insert technique

A needle guide bracket (Ultrasonix Corp., Richmond, BC, Canada) for the curvilinear 1-5 MHz transducer was mounted (Figure 2). The needle guide sets the needle angle β (as defined in Figure 3) to 23° with respect to the transducer centreline.

Fig. 2
figure 2

Assembled sterile transducer with needle guide and epidural needle attached, showing top view. The needle bracket is attached to the ultrasound transducer and the needle guide is attached to the bracket. The epidural needle is then mounted on the needle guide at an angle of 23° with respect to the ultrasound transducer

Full size image
Fig. 3
figure 3

Schematic of probe and needle guide geometry, viewed from the side. The distance A is the distance from the felt tip pen mark, indicating the centre of the ultrasound transducer when it is properly positioned for needle insertion in the surgery preparation room, with the needle guide line aligned with the epidural space. The distance B is the distance seen on the ultrasound image along the needle guide line that extends from the ligamentum flavum to the edge of the image

Full size image

Careful analysis of the geometric measurements was required in terms of the use of the needle guide and the compression of tissue by the ultrasound transducer. With the transducer perpendicular to the skin surface, the depth to the epidural space (distance E in Figure 4) was measured when no pressure was applied to the ultrasound transducer, and the depth to the epidural space was measured again with a typical contact pressure required by an experienced sonographer (V.L.) to achieve an adequate image. The distance A is the distance between (1) the mark that the centre of the transducer made in the pre-puncture examination (where the needle guide line is aligned with the epidural space) and (2) the actual needle insertion site.

Fig. 4
figure 4

Geometry showing the measurement of epidural space depth a straight down (distance E) and b the measurement along the diagonal needle guide line (distance B) where there is a need to angle the transducer by an angle α accounting for tissue deformation at the transducer location and a needle angle β with respect to the transducer axis

Full size image

The depiction of the expected needle trajectory was superimposed digitally as a dashed line on the ultrasound image. The transducer was repositioned so the expected needle path echoed the targeted epidural space while ensuring that the bottom of the needle guide apparatus rested on the skin surface. This involved aiming the transducer slightly cephalad by an angle α while maintaining the same scanning plane (Figure 4). Distance B, the diagonal ultrasound-measured needle depth along the guide line, was measured by the software callipers and recorded. The total length required for needle insertion was the ultrasound-measured needle depth, plus the fixed length of the guide (31.9 mm), plus the blind region (8.7 mm), i.e., distances B + D + C, respectively.

As shown in Figures 3 and 4, the distances B, C, E and angles α, β are related geometrically by

$$ \cos (\alpha + \beta ) = {\frac{E}{B + C}} $$
(1)

or

$$ B = {\frac{E}{\cos (\alpha + \beta )}} - C $$
(2)

where the measurement of C is 8.7 mm; α is the 15° transducer tilting angle (the approximate transducer angle so the needle guide touches the subject’s skin); and β is the 23° angle of the needle relative to the transducer. Then, eq. 2 can be written as

$$ B = 1.27E - 8.7mm $$
(3)

The anesthesiologist then marked the vertical and central plane of the ideal transducer position on the subject’s skin (Figure 5). These marks were used later for faster placement of the ultrasound transducer at the correct level in the operating room where less time is available for scanning.

Fig. 5
figure 5

Schematic of measured distance A, viewed from the back

Full size image

Needle insertion and block procedure

The second part of the experiment was conducted in the operating room. Standard aseptic methods were used, i.e., use of a sterile transducer cover (Cone Instruments, Solon, OH, USA) and a sterile disposable needle guide (Protek Medical Products, Coralville, IA, USA). The skin and subcutaneous tissues were anesthesized at the predicted needle insertion site, about 30-40 mm (distance A) below the felt-tipped pen mark corresponding to the centre of the transducer (Figure 5). The ultrasound transducer was placed on the previously marked locations and was positioned and angled so that the superimposed needle trajectory guide line was aimed at the epidural space on the ultrasound image. A 125 mm New Gertie Marx® CSE-Set epidural needle (IMD Incorporated, Huntsville, UT, USA) was inserted in-plane with the paramedian plane of the ultrasound. The epidural needle was used because it is more rigid than the spinal needle and, consequently, the actual needle path would follow more closely the predetermined needle path on the ultrasound image. Using this arrangement, the needle could be observed during the entire course of the insertion. Care was taken to ensure that the predicted needle path that was superimposed on the scanned image targeted the epidural space throughout the procedure. With the stylet remaining inside the shaft, the epidural needle was inserted until the tip was seen to be approximately 10 mm away from the expected position of the epidural space. Then, a saline-filled syringe was mounted on the epidural needle, and loss-of-resistance to saline was used to advance the needle through the ligamentum flavum and into the epidural space. To avoid degrading the ultrasound image, care was taken not to inject saline until the needle was about 10 mm from the ligamentum flavum. This is the aim-and-insert technique. Although this study used a CSE, it is expected that the success or failure of single-operator ultrasound guidance would also apply to epidural needle insertion.

The endpoint for a successful procedure under ultrasound guidance is defined as a loss-of-resistance permitting either successful anesthesia or a successful threading of a catheter. A spinal neuraxial block is our standard for elective Cesarean delivery. The spinal anesthesia is then injected into the subarachnoid space. In case of absence of cerebrospinal fluid (CSF), an epidural catheter would be threaded into the epidural space to confirm successful epidural placement of the needle. Then, a standard midline spinal needle insertion would be performed to administer the spinal anesthesia. For regional anesthesia for Cesarean delivery, spinal anesthesia was used to deliver intrathecal hyperbaric bupivacaine (12 mg), fentanyl (10 μg), and morphine (100 μg). Successful subarachnoid block was verified through the standard tests for Cesarean delivery (ice at T4 and pin prick at T6).

Statistical analysis

The Bland-Altman analysis was used to compare the measurement of the actual distance B with the distance B calculated in eq. 3 using the measurement of distance E. The 95% limits of agreement are between 95% of the actual distance B and the calculated distance B according to eq. 3.

Results

Twenty subjects gave written consent; however, one subject did not participate in the study due to unrecognizable bony landmarks. The participating subjects (n = 19) were aged 35 ± 5.3 yr, weighed 80.3 ± 13.2 kg, were 160.0 ± 5.6 cm in height, and had a body mass index of 31.5 ± 5.9 kg·m−2. There were 11 obese, five overweight, and three normal subjects (Table 1).

Table 1 Subject biometrics and comments
Full size table

All subjects underwent attempted needle placement using the new real-time ultrasound-guided single-operator aim-and-insert technique. In 18 of the 19 subjects, the epidural needle was successfully guided into the epidural space, as defined by good loss-of-resistance. In one subject (subject 9), loss-of-resistance to saline could not be elicited despite good ultrasound and needle views. There was CSF aspiration in 14 of the 18 subjects confirming that the spinal needle had passed through the epidural space, giving further evidence, though no guarantee, that the needle tip was in the epidural space. In the four subjects with no CSF aspiration, a catheter was threaded easily into the epidural space. Spinal anesthesia provided a satisfactory block in all 18 subjects.

Table 2 shows the differences in measurements between the interspaces identified by palpation and the interspaces identified by ultrasound. In two subjects (subjects 5 and 7), a difference of 25 mm was found. The magnitude of this difference was compared with the respective interfacet distances. The differences for subject 5 (24.1 mm interfacet distance) and subject 7 (23.7 mm interfacet distance) were considered to be miscounts of intervertebral level.

Table 2 Palpation compared with ultrasound: difference (mm) between the anesthesiologist’s mark and the sonographer’s mark. Data missing due to time constraints are labelled “n/a”
Full size table

The ultrasound-measured epidural space depths E (Figure 4), with and without contact pressure, were compared in 17 of the 19 subjects. These measurements were not obtained from two of the subjects due to time constraints in the preparation room. The average difference was 2.8 (SD 1.1) mm, and the 95% limits of agreement were 0.4 mm to 5.2 mm according to Bland-Altman analysis.

The measured diagonal distance B (n = 15) is compared with the distance B calculated from the measurement of E in eq. 3, and the Bland-Altman plot is shown in Figure 6. The bias is 0.8 mm and the 95% limits of agreement are −4.7 mm to 6.4 mm.

Fig. 6
figure 6

Bland-Altman plot of the measured distance B vs the distance B calculated by eq. 3, i.e. the predicted distance B

Full size image

The distance A (n = 13), the distance between the actual insertion point and the mark of the middle of the transducer at the pre-puncture examination (Figure 5), is 34 (SD 6) mm on average, with the values ranging from 25 mm to 45 mm. The actual measurement of this distance (Figure 2) from the middle of the transducer to the tip of the needle guide is 35 mm, which agrees closely with the average of 34 mm.

Discussion

This study demonstrates that a single-operator aim-and-insert real-time ultrasound-guided lumbar paramedian epidural technique is feasible, with success achieved in 18 of the 19 subjects. The one failure occurred within the first ten trials, whereas the latter cases were viewed as being considerably smoother and faster, as the operator (anesthesiologist) became increasingly familiar with the technique. We chose loss-of-resistance as the endpoint, similar to previous studies on ultrasound guidance. We used the presence of CSF or insertion of the catheter as additional evidence that the needle reached its target, but, admittedly, these are not perfect measures of successful placement. Our rates of absence of CSF in subjects where loss-of-resistance was felt (4/18) were comparable with previous studies, with the added difficulty of performing the needle-through-needle technique in the paramedian plane.14 In previous studies, CSEs using a Tuohy epidural needle were noted as having similar rates of failure, i.e., 16%,15 25%,16 and 18%10 rates of absence of CSF aspirations were reported.

By using a fixed guided needle instead of a free-hand technique, the described technique has the advantage of being performed by a single operator. The fixed needle guide allows the needle to be held in the plane of the ultrasound image and allows the operator to release her/his grip on the needle momentarily without losing its alignment in the plane. Conversely, the free-hand technique requires more operator expertise to see the needle tip in the ultrasound image.17

In our study, identification of the interspace level by palpation rather than by ultrasound differed in only two of the 19 subjects. However, since previous studies3 have suggested that palpation misidentifies the actual space in 58% of the cases, it is recommended that ultrasound guidance continue being used.3

In our study, the curved shape of the transducer and the size of footprint presented three practical problems. First, with the subject sitting flexed forward, a better image was obtained by pressing the transducer onto the skin surface rather than resting it lightly. This aspect caused a difference in depth to the epidural space of approximately 2.8 mm. A less curved transducer shape with better skin contact enabling light pressure should reduce this practical error.

Second, our projected needle tract was substantially longer than the more direct midline trajectory used in the blind technique, as suggested by the 27% longer distance B than distance E in equation. 3. Our measured skin-to-epidural space distance was 88-111 mm, requiring the use of a long 125 mm New Gertie Marx® CSE-Set epidural needle. Future developments will require a transducer with a smaller footprint to reduce the track distance through the subject’s tissues.

Third, distance A, the vertical distance between the mark made from the centre of the transducer face during the pre-puncture scan and the actual puncture site, (Figure 3) is 34 mm on average, ranging from 25-45 mm. The magnitude of this distance is important because the puncture site is 35 mm below the desired intervertebral space. This situation necessitates the choice of L2-3 instead of L3-4 to avoid contact with the pelvic bones, and it is entirely due to the limitation of the size of the transducer held parallel to the midline for real-time scanning.

Also, the variability of distance A shows that the pre-puncture ultrasound is useful, but, ultimately, only real-time guidance can ensure the best site and angles, as the transducer position found in the surgical preparation room differs slightly (up to 10 mm) from the position found in the operating room. We believe the pre-puncture ultrasound gives the anesthesiologist time to mark the approximate location of the transducer and to become familiarized, thus saving time in the operating room.

The current protocol requires extra equipment (125 mm CSE set, probe cover, biopsy needle guide, biopsy bracket, ultrasound machine, and curvilinear transducer), additional work for the nurse and the anesthesiologist (the ultrasound transducer must be covered by the sterile cover, and the sterile biopsy guide must be attached prior to the procedure), and extra time for the pre-puncture orientation and skin marking. The extra time may be mitigated by a reduction in the number of needle insertion attempts. Also, as the injection of saline into the subcutaneous tissues obscures the image, the anesthesiologist should minimize the amount of saline injected prior to the needle approaching the ligamentum flavum.

In conclusion, in 18 of the 19 subjects (including obese and overweight subjects), a real-time aim-and-insert ultrasound-guided epidural needle insertion was successfully performed by a single operator. This technique involved a systematic approach for both a pre-puncture scan and ultrasound guidance during needle insertion. Several limitations were also described, mostly related to the needle guide geometry. Current efforts are now underway to design a specialized transducer and needle guide that will improve the clinical practicality of a real-time aim-and-insert ultrasound-guided epidural needle insertion technique.