Dural Tissue Trauma and Cerebrospinal Fluid Leak after Epidural Needle Puncture: Effect of Needle Design, Angle, and Bevel Orientation

Click on the links below to access all the ArticlePlus for this article.

Please note that ArticlePlus files may launch a viewer application outside of your web browser.

• https://links.lww.com/ALN/A228

• https://links.lww.com/ALN/A229

• https://links.lww.com/ALN/A230

• https://links.lww.com/ALN/A231

This article is featured in “This Month in Anesthesiology.” Please see this issue of Anesthesiology, page 5A.

Additional material related to this article can be found on the Anesthesiology Web site. Go to the following address, click on Enhancements Index, and then scroll down to find the appropriate article and link. http://www.anesthesiology.org

POSTDURAL puncture headache (PDPH) is a consequence of cerebrospinal fluid leak, 1–2occurring after both spinal and epidural placement. However, measures used to prevent headache differ between techniques. By addressing the effect of needle gauge and tip design on leak, spinal needle redesign has greatly reduced the incidence of PDPH after spinal needle puncture. 3–5 

Headache prevention during epidural placement continues to focus on techniques of avoiding puncture rather than reducing cerebrospinal fluid leak. However, the wide variability in unintentional dural puncture rates (0.4–6%) across North America, 6the high incidence of PDPH after puncture, 7–8and the severe, refractory nature of some headaches 8–10suggest that other, potentially additive methods of prevention merit exploration in an effort to further reduce PDPH.

Epidural needle selection and technique of placement represent two obvious and easily modifiable factors likely to impact on cerebrospinal fluid leak and headache. Unfortunately, little research has been done in either area, with most of this information extrapolated from spinal needle studies. 11–16Given the differences that exist between epidural and spinal needle design, the validity of such assumptions requires investigation. This in vitro  study examined the effect of epidural needle design, angle of puncture, and bevel orientation on cerebrospinal fluid leak using human cadaveric dura mounted on a physiologically pressurized dural sac model. Dural trauma patterns were examined using scanning electron microscopy.

After institutional research ethics approval (University of Toronto, Toronto, Ontario, Canada), fresh human cadaveric spinal cords with intact dural tissue were obtained at autopsy. Before the study, specimens were kept in cooled Ringer's lactate solution. Inclusion criteria were consent for autopsy and medical research; age older than 18 yr; absence of known or suspected infections such as meningitis, HIV, hepatitis, or Creutzfeldt-Jakob disease; and absence of spinal cord trauma or spinal cord malignancy.

Dura was dissected from the lumbar dural sac from L1–L2 to L4–L5 and cut into approximately 2-cm square pieces. Specimens were mounted, in order of harvest (cephalad to caudal), over a 1-cm aperture in a cylindrical human dural sac model, preserving the anatomic orientation of the tissue (fig. 1A). A wet seal was achieved using a customized gasket and hose clamps. The OD of the model (2.4 cm) closely approximated the ID of the adult human vertebral canal (dural sac) at L3–L4 and L4–L5 measured in five cadavers before the study onset.

The model was pressurized to physiologic levels (fig. 1B) with artificial cerebrospinal fluid (147 mm Na, 2.88 mm K, 127 mm Cl, 1.0 mm phosphate, 1.15 mm Ca, 1.10 mm Mg, 1.10 mm SO⁴, 23.19 mm HCO³, 5,410 mg/L glucose; 300 mOsm/kg) prepared by the hospital pharmacy. Pressures were measured using a standard in-line manometer. Three milliliters methylene blue dye was added to each 3-l bag of fluid to allow visualization of fluid levels.

Epidural needles were coded and selected for use according to predetermined randomization tables made for each part of the study. Use of a micromanipulator enabled precise needle angulation and bevel orientation during dural puncture as well as controlled advancement. Needles were advanced by means of a hand screw at the highest rate possible. Each needle was used to puncture a specimen of dura obtained from the same cadaver. Each dural specimen was punctured only once. The model was pressurized to 15 ± 1 cm H²O pressure (left lateral decubitus position) during puncture. Dural puncture  was defined by free flow of artificial cerebrospinal fluid through the needle hub. After puncture, the needle was withdrawn, and the syringe model was rotated (puncture site downward) without disconnection of the system to allow for collection of fluid. Pressure was increased to 25 ± 1 cm H²O for fluid collection, simulating semisitting position pressure. 3Fluid was collected in preweighed, dry beakers for four 15-min periods for each puncture. Each 15-min collection was timed using a digital timer and weighed using a standard electronic pan balance sensitive to 0.1 g. Volumes were obtained from weights using a cerebrospinal fluid density of 1.0. Dural specimens were then mounted on paraffin and fixed with universal fixative, and scanning electron microscopy was performed. The trauma patterns produced by each needle type were photographed. The dura used in part 1 was derived from 10 cadavers. The dura used for parts 2 and 3 was derived from a second set of 10 cadavers.

Punctures were made using the following six epidural needles at an angle of 90°, bevel parallel (where applicable) to the dural long axis: 17-gauge Hustead (a modified Tuohy with a shorter, blunter tip; Portex/SIMMS, Keene, NH); 17-gauge Tuohy (Ballard Medical Products, Draper, UT); 18-gauge Tuohy-Schliff Perifix (B. Braun, Bethlehem, PA); 18-gauge Special Sprotte® (Pajunk, Geisingen, Germany); 18-gauge Crawford (Becton-Dickinson, Rutherford, NJ); and a 20-gauge Tuohy (Portex/SIMMS). Needle specifications are found in table 1.

The effect of needle angulation and tip design was studied using 18-gauge Tuohy and 18-gauge Special Sprotte® epidural needles. Punctures were performed at 90° and 30° angles to the dural sac in the horizontal plane using a fresh specimen for each puncture. Measurements were repeated for each needle and angle using a separate specimen from the same cadaver. The bevel of the Tuohy needle was oriented parallel to the dural long axis for all of these punctures.

An 18-gauge Tuohy needle was used to puncture specimens at a 90° angle with the bevel oriented perpendicular or parallel to the dural long axis.

Dural tissue texture appeared uniformly normal (opaque) in most cadavers. In one cadaver, the dura was found to be uniformly translucent throughout. Specimens from these cadavers were punctured as assigned, and cerebrospinal fluid leak rates were included in the main analysis.

In several cases, the dura was nonuniform in texture. This was noted as patchy translucency interspersed within grossly normal dura in the same cadaver. When this was found, tissue was handled in the following way: (1) the dura was dissected into approximately 2-cm square specimens in accordance with the standard study protocol; (2) both the translucent specimen and the next adjacent grossly normal-appearing specimen were punctured using the assigned needle and conditions. Only leak rates from grossly normal-appearing specimens were included in the main analysis, with puncture of the translucent dura done for comparative purposes only. This method of dealing with gross nonuniformity of tissue within the same cadaver was initiated with the first recognition of gross tissue nonuniformity and was applied consistently throughout the study.

When dural puncture was achieved but complete absence of leak was found after withdrawal of the needle, the result was noted but not included in the final analysis because this was thought to represent a separate phenomenon. The puncture was repeated using the next 2-cm square specimen in the sequence and the same needle. The leak rate from the second (leaking) specimen was entered into the analysis. Nonleaking puncture sites were examined with use of scanning electron microscopy.

Cerebrospinal fluid leak rates were assessed for part 1 using two-factor repeated measures analysis of variance (RMANOVA) for needle (six levels) and time (four levels); for part 2 using three-factor RMANOVA for needle (two levels), angle (two levels), and time (four levels); and for part 3 using two-factor RMANOVA for bevel orientation (two levels) and time (four levels). A P  value of less than 0.05 was considered statistically significant. Secondary analyses of part 1 and 2 results using RMANOVA were corrected for multiple testing and required P  values of less than 0.003 and less than 0.008, respectively, to reach statistical significance. The statistician responsible for the analysis was Dr. J. P. Szalai, Associate Professor, Director of Research Design and Biostatistics, Sunnybrook and Women's College Health Sciences Center.

Information relating to cadaver sex, age, cause of death, and tissue age at the start of the experiment is available on the Anesthesiology Web site at http://www.anesthesiology.org. The mean age of the tissue from the time of death to the start of each experimental phase was (mean ± SD) 43 ± 17 h for part 1, 38.1 ± 19 h for part 2, and 33 ± 18 h for part 3. Differences in tissue age between parts 2 and 3 reflect the smaller number of cadavers used in part 2 (n = 7) compared to part 3 (n = 10).

Dural specimens from 10 different cadavers were examined. Two-factor RMANOVA (needle, time) showed highly significant needle (P = 0.0004) and time (P = 0.0007) main effects with no evidence of a needle by time interaction (P = 0.7665). The leak rates obtained for six different epidural needles after 90° puncture are shown in table 2. The greatest leak was found with the 17-gauge Hustead needle, and the least leak was found with the 20-gauge Tuohy. Subanalysis using RMANOVA showed that 20-gauge Tuohy puncture produced large, statistically significant reductions in leak rate (P = 0.0001–0.0024) compared with all of the needles studied except the Special Sprotte® (P < 0.003 required to reach statistical significance, adjusted for multiple testing). Cerebrospinal fluid leak was found to decrease over time for all needle types. The magnitude of this decrease ranged from 3 to 14 ml/h. These reductions reached statistical significance only when compared between the first 15-min interval and at 1 h. Dural trauma patterns for the 17-gauge Hustead, 17-gauge Tuohy, 18-gauge Special Sprotte®, and 18-gauge Crawford needles after 90° puncture are shown in figures 2A–D. Additional figures showing trauma patterns for 18-gauge Tuohy and 20-gauge Tuohy needles are available on the Anesthesiology Web site.

Dural specimens from seven different cadavers were examined. Three-factor RMANOVA (needle, angle, time) showed a nonsignificant effect for needle type (P = 0.0634), no effect for puncture angle (P = 0.4019), and a significant time effect (P = 0.0159). The magnitude of the differences in leak and the consistency of the direction of the reductions led to subanalysis using RMANOVA for the purposes of hypothesis-generation only. The results are presented in table 3. None of the comparisons reached statistical significance (P < 0.008 required, corrected for multiple testing). Figures demonstrating the variability in dural trauma found after standardized puncture of dural specimens from the same cadaver using the same needle are available on the Anesthesiology Web site. Photographs are provided for both 18-gauge Tuohy and 18-gauge Special Sprotte® needles after 30° punctures.

Dural specimens from 10 different cadavers were punctured with an 18-gauge Tuohy needle with the bevel oriented parallel versus  perpendicular to the dural long axis. RMANOVA of leak rates per 15-min interval showed nonstatistically significant reductions in leak (mean ± SD) with perpendicular versus  parallel punctures, respectively (367 ± 119 vs.  485 ± 216; P = 0.12). Statistically significant reductions in leak rate were found over time, regardless of the needle bevel orientation at the time of puncture (P = 0.0010). Perpendicular 18-gauge Tuohy bevel orientation was often associated with a more prominent flap in the dura on scanning electron microscopy (fig. 3) compared with punctures with the needle bevel parallel (fig. 2B). Additional figures illustrating the spectrum of trauma found after perpendicular puncture with an 18-gauge Tuohy needle are available on the Anesthesiology Web site.

Seven specimens were found to produce no cerebrospinal fluid leak after needle withdrawal despite free flow of fluid from the needle hub at the time of puncture. Incremental increases in pressure (up to 45 cm H²O) failed to produce any leak from puncture sites in these specimens. This was observed for the following needles: 17-gauge Tuohy (one time); 18-gauge Tuohy (one time) and 18-gauge Crawford (two times) after 90° punctures; and the Special Sprotte® (three times) after 30° punctures. Scanning electron microscopy of nonleaking samples showed plugs of debris in the puncture site. An example is shown in figure 4A. Examination of the debris within the puncture site at higher levels of magnification (figs. 4B and C) and comparison with nondisrupted dural tissue outside the puncture site on the same specimen (fig. 4D) suggests that the hole was plugged with dural tissue fragments.

Several comparative punctures, using the same needle and dura from the same cadaver, were found to demonstrate markedly different cerebrospinal fluid leak rates. Partial plugging of the puncture sites was found in these cases on scanning electron microscopy. This was observed for both 18-gauge Tuohy and 18-gauge Special Sprotte® needles at 90° and 30°. Scanning electron microscopy of one slowly leaking puncture site (Special Sprotte® epidural needle) shows a partial dural tissue plug extending from the edge of the puncture site into the hole as well as debris (fig. 5).

Several punctures were performed to compare leak rates between grossly translucent- and normal-appearing dura obtained from the same cadavers. When subjected to a standardized puncture, translucent specimens of dura exhibited as much as a 50% greater leak rate compared with normal-appearing specimens from the same cadaver. Leak rates after standardized puncture of specimens derived from a single cadaver with diffusely transparent dura also exhibited much higher leak rates (> 50%) than those found in grossly normal-appearing specimens from other cadavers.

Of all the measures used to prevent PDPH after spinal needle puncture, needle modification has proved most effective. Understanding the effects of needle design on cerebrospinal fluid leak and headache and clinical acceptance of modified needles has led to a dramatic reduction in PDPH, from 20% to 1–2% or less. 5,17On average, the risk of PDPH after deliberate puncture during spinal anesthesia is now similar to the overall risk of PDPH after epidural placement. 17However, important differences remain in the character of PDPH produced—in the nature of the headache, its responsiveness to therapy, and possibly in the potential for long-term sequelae. 9,10,18 

Although it is best to avoid dural puncture during epidural placement, reducing dural damage when it does occur is likely to reduce PDPH development and severity. Unfortunately, little direct information is available to describe the effects of epidural needle design or technique of insertion on cerebrospinal fluid leak or PDPH. This study examined the effects of epidural needle design, angle, and bevel orientation on cerebrospinal fluid leak and dural trauma after puncture.

As with spinal needle studies, the results of the current study suggest that epidural needle gauge is the most important predictor of cerebrospinal fluid leak. 3,5The findings show that leak is greatest with the 17-gauge Hustead and least with the 20-gauge Tuohy needle (table 2). Although it is not possible to comment on PDPH in this study, existing clinical work suggests that accidental punctures with 20-gauge Tuohy needles (notably at mixed anatomic levels using only loss of resistance to normal saline) are associated with large reductions in the incidence of PDPH compared to rates found with larger epidural needles. 7,8,19 

The variable impact of epidural needle tip design on dural trauma and cerebrospinal fluid leak is interesting. The slightly smaller-diameter 17-gauge Hustead needle (a modified Tuohy needle with a shorter, blunter tip) produced “punched-out” ovoid dural holes and greater leak rates than the slightly greater-diameter 17-gauge Tuohy needle (table 1). The latter was found to produce crescentic punctures on scanning electron microscopy, although occasional punched-out holes were also seen with this needle. Differences in leak rates between these two needles did not achieve statistical significance (table 2).

In contrast, 18-gauge Special Sprotte® and 18-gauge Crawford needles, needles of equivalent OD with markedly different tip designs, produced leaks of similar magnitude (table 2) despite very different trauma patterns (figs. 2C and D). The large crescentic flap produced by the Crawford needle may be a reason for the low leak rate obtained with this needle.

Similar to spinal needle studies, cerebrospinal fluid leak was found to be less with a noncutting-tip epidural needle (Special Sprotte®) than with a cutting needle (Tuohy) of the same gauge. 11,13,14The angle of puncture reduced leak only for the cutting-tip needle. Reductions did not reach statistical significance in either case (table 3).

Bevel orientation at the time of puncture has demonstrated inconsistent effects on cerebrospinal fluid leak and PDPH in spinal needle studies. 11,15,16Results with an 18-gauge Tuohy epidural needle showed larger mean leak rates after parallel puncture compared with perpendicular puncture, but these differences did not reach statistical significance. The individual data points consistently suggest similar or greater leak rates after parallel (rather than perpendicular) puncture. The larger, crescentic flap found with perpendicular punctures may help to explain the differences in leak rate related to bevel orientation. These results are not in accordance with those of a clinical study reporting a greater incidence of PDPH and epidural blood patch after perpendicular (vs.  parallel) punctures in parturients. 20It should be noted that our study examined the effect of bevel orientation on cerebrospinal fluid leak for the 18-gauge Tuohy needle only. These findings may vary between needle types. Further study is warranted.

Spinal needle studies have reported that cerebrospinal fluid leak decreases over time and may stop completely, a phenomenon attributed to viscoelastic properties of the dura and hole retraction. 11,12,14,21In all parts of this study, statistically significant decreases in leak rate were found over the first hour after puncture. However, in no instance was there complete cessation of leak related to this phenomenon.

A second phenomenon was found when cerebrospinal fluid leak was either completely absent after needle withdrawal or markedly lower than comparative punctures performed under standardized conditions. Scanning electron microscopy of these specimens showed complete or partial plugging of the puncture sites with what seems to be dural tissue fragments. This finding affords a plausible mechanism for the observed absence of PDPH development in 20–30% of patients after recognized epidural needle puncture.

Little attention has been given to the role of dural structure on cerebrospinal fluid leak after puncture. The observation that grossly transparent tissue leaks at a greater rate than grossly normal-appearing dura suggests that differences in dural stroma may also predispose some patients to greater leak and presumably PDPH.

This in vitro  investigation has limitations, including the relatively small number of cadavers used (n = 20) and the absence of an epidural space in the model. No attempt was made to examine the propensity of the needles themselves to pierce dura, a factor that probably varies between needle types. The absence of protein in the artificial cerebrospinal fluid used may have influenced absolute leak rate measurements, although relative leak rates are probably comparable. Minor differences in the rate of needle advancement at the time of puncture might have had some impact on dural trauma patterns and leak, although overall, these were performed in a uniform fashion.

Other elements of the model design suggest that it demonstrates face and content validity. These include (1) the human lumbar spine dimensions of the model used, (2) the preservation of in vivo  orientation of dural tissue specimens, (3) use of physiologic cerebrospinal fluid pressures at the time of puncture and during leak measurement, (4) observed dural tenting during puncture, and (5) findings consistent with earlier work related to spinal needle gauge and tip design.

The results suggest that use of a 20-gauge Tuohy needle significantly reduces cerebrospinal fluid leak rate compared with larger epidural needles. The feasibility of using such small needles for continuous epidural analgesia or anesthesia in adults and the impact on PDPH awaits further study.

The authors thank Susan Cromwell, P.A., and Anna Marie Moscoluk, B.A., P.A. (Pathology Assistants, Department of Pathology); Shan Balasubramanian, R.R.T. (Department of Anesthesia); Frank Brommecker, B.Sc.Phm., Michael Ritchie, B.Sc.Phm., and staff (Pharmacy Department); Michael Wong, L.B.I.P.P. (Audio Visual Department); Weddad Hannah, M.D., F.R.C.P.C. (Chief, Department of Pathology); and Lee Cyn Ang, M.D., F.R.C.P.C. (Neuropathologist, Department of Pathology), all from Women's College Campus, Sunnybrook and Women's College Health Sciences Center, Toronto, Ontario, Canada; Jack Butany, M.D., F.R.C.P.C. (Chief, Department of Pathology, Toronto Hospital, Toronto, Ontario, Canada); William Halliday, M.D., F.R.C.P.C. (Chief, Department of Neuropathology, Toronto Western Hospital, Toronto, Ontario, Canada); and Aina Tilups, B.Sc., R.T. (Electron Microscopy Laboratory, Hospital for Sick Children, Toronto, Ontario, Canada). The authors also thank Stephen Halpern, M.D., F.R.C.P.C. (Women's College Campus, Sunnybrook and Women's College Health Sciences Centre, Toronto, Ontario, Canada); and Joanne Douglas, M.D., F.R.C.P.C. (British Columbia Women's Hospital, University of British Columbia, British Columbia, Canada) for reviewing the manuscript and for their helpful comments and suggestions.