Published Ahead of Print on December 12, 2019 as 10.1212/WNL.

0000000000008749
ARTICLE

Sacral anatomical interspace landmark for lumbar

puncture in pregnancy
A randomized trial
Carlo Pancaro, MD, Baskar Rajala, MD, Christie Vahabzadeh, MD, Ruth Cassidy, MA, Thomas T. Klumpner, MD, Correspondence
Joanna A. Kountanis, MD, Madeline McCabe, Dana Rector, Casey Aman, Keerthana Sankar, Robert Schoenfeld, Dr. Pancaro
and Milo Engoren, MD cpancaro@med.umich.edu

®
Neurology 2019;00:1-10. doi:10.1212/WNL.0000000000008749

Abstract
Objective
To determine whether the sacral anatomical interspace landmark (SAIL) technique is more
accurate than the classic intercristal line (ICL) technique in pregnant patients and to assess the
percentage of clinical determinations above the third lumbar vertebra.

Methods
In this prospective, randomized, open-label trial, there were 110 singleton pregnant patients
with gestational age greater than 37 weeks included. Selection procedure was a convenience
sample of pregnant patients who presented for office visits or vaginal or cesarean delivery
between March 15 and July 31, 2018, at a single-center obstetric tertiary care university hospital.
Both techniques were evaluated by 2 physicians independently assessing each method. Before
data collection, we hypothesized that the SAIL technique would be more accurate than the ICL
technique in determining the L4-L5 interspace, and that the SAIL technique would produce
more estimations below the third lumbar vertebra than the ICL technique. Therefore, the
primary outcome was accuracy in identifying the L4-L5 lumbar interspace with SAIL vs ICL.
The secondary outcome was difference in clinical assessments above the third lumbar vertebra.
Both outcomes were measured via ultrasonography.

Results
Patients were 31 ± 5 years of age (mean ± SD) and had body mass index of 31.8 ± 5.7 kg/m2 and
gestational age of 38.8 ± 1.1 weeks. A total of 110 patients were analyzed. SAIL correctly
identified the L4-L5 interspace 49% of the time vs 8% using ICL (p < 0.0001). Estimations
above L3 were 1% for SAIL vs 31% for ICL (p < 0.0001).

Conclusions
Our study shows improved accuracy in identifying intervertebral space using the SAIL tech-
nique; this may prevent direct mechanical trauma to the conus medullaris when lumbar
punctures are performed in pregnancy.

Clinicaltrials.gov identifier
NCT03433612.

From the Department of Anesthesiology, University of Michigan, Ann Arbor.

Go to Neurology.org/N for full disclosures. Funding information and disclosures deemed relevant by the authors, if any, are provided at the end of the article.

Copyright © 2019 American Academy of Neurology 1

Copyright © 2019 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
 Glossary
BMI = body mass index; ICL = intercristal line; IQR = interquartile range; IRB = institutional review board; LP = lumbar
puncture; Pc = control proportion; Pt = treatment proportion; SAIL = sacral anatomical interspace landmark; STROBE =
Strengthening the Reporting of Observational Studies in Epidemiology.
In the United States, at least 360,0001 diagnostic lumbar
punctures (LPs) are performed annually, many without ra-
diologic guidance.2 In addition, nearly 1 million spinal anes-
thetics, also performed by LP without imaging guidance, are
administered in pregnant patients undergoing cesarean de-
livery.3 The intercristal line (ICL) clinical method is the most
widely employed method to identify the LP site.4,5 In preg-
nancy, the ICL method leads to misidentification of the L4-L5
interspace 86% of the time,6 causing the provider to enter
a higher than intended interspace,7 with consequent risk of
mechanical trauma to the conus medullaris,4,8 increasing the
risk of major morbidity.9 To our knowledge, there are no
studies evaluating a clinical identification technique alterna-
tive to the ICL method to perform LPs. The sacral anatomical
interspace landmark (SAIL) technique consists of identifying
the sacrum by palpation. The provider’s hand slides cephalad
along the dorsal surface of the sacrum in order to identify the
first interspace between the sacrum and the last lumbar ver-
tebra (L5-S1). Once the L5-S1 space has been identified, the
provider’s hand slides cephalad 1 more interspace to identify
the L4-L5 landmark space.
We hypothesized that the SAIL technique is more accurate in
identifying the L4-L5 interspace than the classic ICL tech-
nique in pregnancy. We also hypothesized that the SAIL
technique produces more estimations below the third lumbar
vertebra than the ICL technique. The primary outcome was
correct identification of the L4-L5 interspace. The secondary
outcome was the occurrence of measurements that fell
above L3.
Methods
Standard protocol approvals, registrations,
and patient consents
The University of Michigan institutional review board (IRB)
approved this prospective randomized observational study
(IRB HUM00109509). Before enrollment began, the study
was registered at clinicaltrials.gov (NCT03433612). All items
of the Strengthening the Reporting of Observational Studies
in Epidemiology (STROBE) statement checklist10 were fol-
lowed. The STROBE flow diagram is shown in figure 1. All
participants gave written consent to participate in the study.
Patients and enrollment
A total of 110 pregnant patients older than 18 years with
a singleton gestation and at ultrasound-determined gesta-
tional age of 37 weeks or greater were enrolled before
placement of neuraxial blockade for cesarean or vaginal
2 Neurology | Volume , Number  | Month ▪▪, 2019
Copyright © 2019 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
delivery. Patients who were not in labor and those in first stage
of labor were eligible for recruitment. Patients in labor who
were perceived by the nurse or the clinician to be unable to
cooperate with positioning were excluded. In addition,
patients who had undergone a previous spine surgery, who
had known spinal deformities, who had impaired decision-
making abilities, who had a body mass index (BMI) greater
than 45, or who were non-English-speaking were also
excluded.
Recruitment and randomization
The study was performed on days that authors C.P., B.R., and
C.V. were available. A convenience sample of eligible patients
was recruited during office visits or upon presentation to the
labor and delivery unit. Once written consent was obtained,
a study ID number was assigned and demographic in-
formation was collected: age, height, weight, BMI, and ges-
tational age. Both the operator order (C.P. or B.R. going first)
and the operator technique (whether C.P. or B.R. performed
ICL or SAIL) were randomized, using a computer-generated
randomization tool (randomizer.org/) in blocks of 10
patients. Using this procedure, both techniques were studied
in 110 patients by both providers. Each provider performed
55 ICL and 55 SAIL assessments. Each estimation technique
(ICL and SAIL) was performed by 2 obstetric anesthesiolo-
gists (C.P. and B.R.), who both have more than 10 years of
clinical experience performing LPs.
Clinical estimate technique
Patients were positioned sitting to one side of a level stretcher
with the neck, back, and hips flexed. The hips were positioned
with the weight distributed evenly between both sides. An
assistant stood facing the patient helping to maintain her
position, if needed.
The ICL technique consists of first palpating the posterior
iliac crests (figure 2A). Then, the provider’s hand slides me-
dially along an imaginary line connecting the 2 posterior iliac
crests (figure 2B). This line is said to cross the midline at the
L4 vertebrae or L4-L5 interspace. If the provider palpates
a spinous process, this would be regarded as L4 and the
provider’s nondominant hand would slide 1 interspace cau-
dally to the L4-L5 interspace. If the provider palpates an
intervertebral space, the palpated interspace is presumed to be
the L4-L5 intervertebral space.
The SAIL technique consists of first identifying the dorsal aspect
of the sacrum by palpation (figure 2C). Once the lowest part of
the sacrum has been palpated, the provider’s hand slides ceph-
alad to the dorsal surface of the sacrum in order to identify the
Neurology.org/N
Figure 1 CONSORT (Consolidated Standards of Reporting Trials) flow diagram shows recruitment numbers of eligible
patients, study participants, and number of techniques performed by each provider (C.P. and B.R.)
first interspace between the sacrum and the last lumbar vertebra,
the L5-S1 interspace. Then, once the L5-S1 intervertebral space
has been identified, the provider’s hand will slide up 1 more
interspace in order to identify the L4-L5 interspace (figure 2D).
Using these 2 techniques, 2 providers consecutively marked
the patient’s skin with an erasable pen at the estimated L4-L5
level as previously described.6 The ICL technique findings
were marked, with a short 1/2 inch horizontal line, on the skin
covering the areas where the left posterior axillary and the left
scapular lines intersect cranio-caudally: the latissimus dorsi,
external oblique, gluteus medius, and gluteus maximum. The
ICL technique mark was then concealed with a folded 40 × 40
woven gauze swab and tape. Likewise, the SAIL technique
findings were marked as above on the right side and the
markings concealed. The skin markings were aided by using
a 30-cm ruler with an embodied spirit level to ensure that
markings were at the same level of the palpation over the
spine. The folded 40 × 40 woven gauze swab and tape was used
so that each physician was blinded to the other’s estimation.
The first physician made sure that covering the marking did
not interfere with the next physician’s ability to accurately
assess and determine the interspace level with the other
technique. In half (n = 55) of the patients, 1 investigator
Neurology.org/N
Copyright © 2019 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
ICL = intercristal line method; SAIL = sacral ana-
tomical interspace landmark method.
performed the SAIL technique and the other investigator
performed the ICL technique. The investigator performing
each technique was reversed for the other 55 patients.
A third provider (C.V.), experienced with lumbar ultraso-
nography and blinded to the 2 clinical estimates, scanned with
ultrasound the lumbo-sacral area in the same flexed position
and marked with an erasable pen the vertebral levels L1, L2,
L3, L4, L5, S1 and the intervertebral levels L1-L2, L2-L3, L3-
L4, L4-L5, L5-S1, as previously described.6
Ultrasound equipment and technique
A portable SonoSite Inc. Edge (Bothell, WA) ultrasound
system, fitted with a 4-MHz C60X curved array probe, was
used to determine the vertebral and intervertebral spaces. The
probe was applied in the paramedian longitudinal plane to
visualize the sacrum and the interlaminar spaces individually.
The interlaminar space between the L5 vertebra and the sa-
crum was identified first. The L5 level was marked on the skin
at the midpoint of the probe by positioning the L5 lamina in
the center of the screen, as previously described.6 The L4 to
L1 levels were identified and marked moving cephalad. The
sacrum was marked in a similar manner moving caudally. The
marks for the ICL and SAIL clinical estimates were then
Neurology | Volume , Number  | Month ▪▪, 2019 3
 Figure 2 Clinical intervertebral space estimates of the classic and sacral anatomic interspace landmark (SAIL) techniques,
in term pregnancy

(A, B) Classic technique. The classical intercristal line technique consists of first palpating the posterior iliac crests (A). Then, the provider’s hand slides medially
along an imaginary line connecting the 2 posterior iliac crests to locate the estimated interspace (B). (C, D) The SAIL technique consists of first palpating the
lowest part of the dorsal aspect of the sacrum. The provider’s hand then slides cephalad along the dorsal surface of the sacrum in order to identify the first
interspace between the sacrum and the last lumbar vertebra, the L5-S1 interspace (C). Then, once the L5-S1 intervertebral space has been identified, the
provider’s hand slides cephalad 1 more interspace to identify the L4-L5 interspace (C, D).

uncovered and the corresponding vertebral and intervertebral
levels were noted. An assistant reported the levels as follows:
L1 = 1, L1-L2 = 1.5, L2 = 2, L2-L3 = 2.5, L3 = 3, L3-L4 = 3.5,
L4 = 4, L4-L5 = 4.5, L5 = 5, L5-sacrum = 5.5, sacrum = 6.
Statistical analysis
The primary outcome was the correct identification of the L4-
L5 interspace with each technique, ICL and SAIL. Observed
success rates were compared between techniques using the
McNemar test for marginal homogeneity. The primary null
hypothesis was that the success and failure rates for the ICL
and SAIL techniques are the same. The secondary outcome
was the frequency of measurements that fell above the third
lumbar vertebra. This secondary outcome was also assessed
for homogeneity across technique (ICL/SAIL) using the
McNemar test. The secondary null hypothesis was that the
probability of a measurement falling above the third lumbar
vertebra is the same for the ICL and SAIL technique.

4 Neurology | Volume , Number  | Month ▪▪, 2019 Neurology.org/N

Copyright © 2019 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
Additional secondary analyses were conducted to examine the
difference between the estimate and the L4-L5 interspace for
each technique and the difference between the 2 estimates.
For the secondary analyses, we coded the levels as L1 = 1, L1-
L2 = 1.5, L2 = 2, L2-L3 = 2.5, L3 = 3, L3-L4 = 3.5, L4 = 4, L4-
L5 = 4.5, L5 = 5, L5-sacrum = 5.5, sacrum = 6. Medians and
interquartile ranges (IQRs = quartile 3 − quartile 1) of the
difference between the ICL and SAIL techniques were com-
pared for each group. Differences between the 2 techniques
were assessed using the Wilcoxon signed-rank test. We used
the Fisher exact test as a post hoc analysis, with a Šidák cor-
rection for multiple comparisons and overall significance level
of 0.05, to compare the 2 operators (C.P. and B.R.) on their
ability to correctly identify L4-L5 interspace with each of the 2
techniques.11
To examine the relationship between patient demographics
and each estimate’s distance to L4-L5 (calculated as the code
for the estimated level minus 4.5 [the code for L4-L5]), we fit
2 linear regression models for each technique with distance
and absolute distance as the response variables, and age,
gestational age, and BMI as explanatory variables. We also fit
linear regression models for the distance between the 2
techniques (ICL–SAIL) and the absolute distance |ICL–
SAIL|, with age, gestational age, and BMI as explanatory
variables. Corresponding adjusted slope estimates, 95% con-
fidence intervals, and p values were calculated.
For the power analysis, we first estimated the percentage of
participants in whom the L4-L5 space was correctly identified
according to the ultrasound measurement with the study
performed by Lee et al.6 as the control proportion (Pc) = 5%.
Pc represents the success rate of the classic technique. Pt
(treatment proportion) represents the success rate of the
treatment technique and was estimated as 0.2 above the
success rate of the classic technique. For a 2-sided McNemar
test with a significance level of 0.05, a difference of proportion
of 0.2, and a proportion of discordant pairs of 0.275 (a con-
servative estimate calculated by Pc [1 − Pt] + Pt [1 − Pc]),
a sample of 110 achieves 98.7% power.
A 0.05 significance level was used to determine statistical
significance throughout. Power calculations used PASS, ver-
sion 14 (NCSS Statistical Software, Kaysville, UT). All other
statistical analyses were performed using SAS, version 9.4
(SAS Institute Inc., Cary, NC).
Data availability
Anonymized data not published within this article will be
made available by request from any qualified investigator.
Results
From March 15 to July 31, 2018, 143 parturients were
screened for the study. A total of 110 patients were ultimately
recruited. No patients dropped out of the study. The 110
Neurology.org/N
Copyright © 2019 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
patients in our study sample had an age of 31 ± 5 years (mean
± SD), gestational age of 38.8 ± 1.1 weeks, and BMI of 31.8 ±
5.7 kg/m2. Sixty eight percent (75/110) self-identified as
Caucasian, 14% (15/110) self-identified as African American,
and the remaining indicated another race/ethnicity.
The L4-L5 interspace was correctly identified 49% of the time
using the SAIL technique and 8% of the time using the ICL
technique (p < 0.0001) (figure 3). The most common in-
terspace identified by the ICL technique was the L3-L4 in-
terspace (49% = 54/110) while the most common site
identified by the SAIL technique was the L4-L5 interspace
(49% = 54/110) (figure 3). The median level identified was 1
level cephalad to the targeted L4-L5 interspace by the ICL and
on target for the SAIL technique, with the ICL technique
having a median of L3-L4 interspace (IQR L2-L3 interspace
to L3-L4 interspace) and the SAIL technique having a median
of L4-L5 interspace (IQR L4 body to L4-L5 interspace), p <
0.0001 (figure 3).
While all SAIL clinical estimates except 1 (1% = 1/110) were
never higher than L3 (where the spinal cord is most likely to
end), the ICL estimate was higher than L3 in 31% of cases
(34/110). This difference was statistically significant, with p <
0.0001 (figure 3). The median distance of the SAIL techni-
que’s estimate from the actual L4-L5 interspace to the pre-
dicted L4-L5 interspace was 0. The ICL technique had
a median distance from the actual L4-L5 interspace to the
predicted L4-L5 interspace of 1 (table 1). Within each in-
dividual patient, the ICL technique identified an interspace 1
level higher than the level identified by the SAIL technique
(table 1).
When not correctly identifying the L4-L5 interspace, SAIL
estimates were most commonly 1 space higher at L3-L4 (24%
= 26/110) or 1 space lower at L5-S1 (17% = 19/110). In
contrast, when the classic ICL technique was inaccurate, the
estimate was almost always higher than L4-L5 and rarely
lower. When incorrect, the ICL technique estimate was at the
L1-L2 interspace for 5% (5/110), at L2-L3 for 23% (25/110),
and at L3-L4 for 49% (54/110) of cases (figure 3). There was
only 1 inaccurate case where the ICL estimate was lower than
the L4-L5 interspace, at L5-S1 (1% = 1/110).
The difference in each operator’s ability to correctly identify
L4-L5 interspace with each of the 2 techniques did not reach
statistical significance: one provider correctly identified the
L4-L5 interspace 2% (1/55) of the time using ICL and 47%
(26/55) using SAIL, while the other provider was correct 15%
(8/55) of the time using ICL and 51% (28/55) using SAIL
(adjusted p = 0.0622 and p = 0.9772).
BMI was significantly associated with the distance between
the L4-L5 level and the SAIL-estimated level, with the linear
regression b-estimate for BMI (in 10 kg/m2 units) of −0.3.
This suggests that an increase of 34 kg/m2 in BMI is associ-
ated with about a 1-interspace decrease in the SAIL-estimated
Neurology | Volume , Number  | Month ▪▪, 2019 5
 Figure 3 Distribution of the clinical estimates with the classic intercristal line (ICL) (red bars) and sacral anatomic interspace
landmark (SAIL) (blue bars) palpation techniques

The mode produced by the classic technique was the L3-L4 interspace while the mode produced by the SAIL technique was the L4-L5 interspace. Median level
identified (interquartile range): classic ICL: L3-L4 interspace (L2, L3-L4 interspace) vs SAIL: L4-L5 interspace (L4, L4-L5 interspace), p < 0.0001, Wilcoxon signed-
rank test.

distance from the L4-L5 level, on average (table 2). No de- associated with absolute distance |ICL–SAIL| between the 2
mographic factors were significantly associated with either techniques. For every decade (10-year) increase in age, the
distance or absolute distance between the true L4-L5 level and average absolute distance between the 2 estimates increased by
the ICL-estimated level. Only age was found to be significantly about 0.31. Therefore, an increase of 33 years of age is

6 Neurology | Volume , Number  | Month ▪▪, 2019 Neurology.org/N

Copyright © 2019 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
Table 1 Summary measures for the estimates, distance between estimates, and distance between the targeted L4-L5
interspace and the estimated L4-L5 site identified by the intercristal line (ICL) and sacral anatomical interspace
landmark (SAIL) techniques
Lower quartile Median (IQR) Upper quartile Minimum Maximum p Value

ICL estimate 2.5 L2-L3 3.5 (1); L3-L4 (1) 3.5 L3-L4 1 L1 5.5 L5-Sacrum

SAIL estimate 4 L4 4.5d (0.5); L4-L5 (0.5) 4.5d L4-L5 2.5 L2-L3 5.5 L5-Sacrum

Distance between ICL −2 −1 (1) −1 −3 2 <0.0001e

and SAIL estimatesa

Distance from L4-L5 −2 −1 (1) −1 −3.5 1 <0.0001e

to ICL estimateb

Distance from L4-L5 −0.5 0 (0.5) 0 −2 1 0.2131e

to SAIL estimatec

Abbreviation: IQR = interquartile range (upper quartile − lower quartile).

Coded level numbers: L1 = 1, L1-L2 = 1.5, L2 = 2, L2-L3 = 2.5, L3 = 3, L3-L4 = 3.5, L4 = 4, L4-L5 = 4.5, L5 = 5, L5-sacrum = 5.5, sacrum = 6.
a
ICL estimate − SAIL estimate.
b
ICL estimate − 4.5.
c
SAIL estimate − 4.5.
d
Targeted location: 4.5 = L4-L5 interspace.
e
Wilcoxon signed-rank test.

associated with approximately a 1-interspace increase in the
absolute distance between techniques (table 2).
Discussion
The SAIL technique is more accurate than the ICL technique
in the identification of the L4-L5 interspace. More impor-
tantly, the current study shows that even when the identifi-
cation of the L4-L5 interspace is incorrect, the SAIL technique
is within 1 space 98% of the time. Moreover, 99% of in-
terspace identifications using the SAIL technique were below
L3, the lowest point where the conus medullaris may be
found,12 minimizing the theoretical risk of needle injury to the
spinal cord. Conversely, when the ICL technique estimate is
not accurate, the intervertebral space is almost always higher
than L4-L5, and 30% of the time is even more cephalad
than L3.
of the L4-L5 interspace. One possibility is that the cephalad
end of a prominent median sacral crest (the rudimentary
fusion of sacral spinous processes) could have been mistaken
for the lumbo-sacral interspace, and the second cephalad in-
terspace counting upward was therefore erroneously esti-
mated as L4-L5. Given sacral bone anatomical variations,15,16
we do not know whether the absence or the incomplete
genesis of the median sacral crest could lead to better accuracy
in determining the desired intervertebral space when the SAIL
method is used. Another hypothesis as to why the L5-S1
interspace was incorrectly identified as L4-L5 is that even
though the apex of sacral hiatus is most commonly located at
the S4 level, in 3%–15% of cases it can terminate as high as the
S1, S2, and S3 levels,17,18 potentially misleading the clinician
to feel this gap as the lumbosacral interspace, and sub-
sequently causing an erroneous, more caudal estimation of the
L4-L5 interspace as the palpation moves cephalad.

The position of the conus medullaris has been reported to be
lower in females than in males.13 A combination of a low
position of the conus medullaris and a higher clinical esti-
mation of the targeted interspace might increase the risk of
contact between the tip of the spinal needle and the spinal
cord, putting patients at higher risk of neurologic injury.9 Of 7
reported conus injuries, 6 occurred in pregnant patients re-
ceiving spinal or combined spinal epidural blocks, most be-
lieved to be at the L2-3 interspace.9 Furthermore, the
American Society of Anesthesiologists’ Closed Claims data-
base reported that 2 spinal cord injuries that resulted in
paraplegia were attributable to direct drug injection into the
spinal cord.14
In 19% of cases, the SAIL technique incorrectly identified L5-
S1 as L4-L5. We hypothesize that in those assessments there
was anatomical variability leading to more caudal estimation
Using the sacrum as the initial clinical point of reference could
theoretically lead the clinician to inaccurately make a lower
assessment than the lumbo-sacral interspace. However, using
the SAIL method, no assessments were made lower than the
L5-S1 interspace, implying that whenever a LP is to be per-
formed there is always a chance to reach a vertebral interspace
for the purpose of CSF sampling or drug delivery. However, in
contrast to the ICL method, the SAIL method is less likely to
inaccurately make a high assessment, leading to a potentially
dangerous puncture at the conus medullaris level.
In addition, lumbosacral transitional vertebrae such as the
sacralization of the fifth lumbar vertebra or the lumbarization of
first sacral vertebra can mislead the clinician to an improper
identification of the correct interspace.19 In either situation,
when using the SAIL technique, the first palpable interspace
would actually be the L4-L5 intervertebral space and the

Neurology.org/N Neurology | Volume , Number  | Month ▪▪, 2019 7

Copyright © 2019 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
 Table 2 Associations of demographic factors with distances from the L4-L5 level and with distances between techniques
Distance between estimate and L4-L5 level (estimate 2 4.5)

Absolute distance from L4-L5 Distance from L4-L5

Variable B 95% CI p Value B 95% CI p Value

ICL technique

Intercept −0.39 (−5.36 to 4.59) 0.8782 1.26 (−3.93 to 6.44) 0.6324

Age decade (10 years) 0.07 (−0.22 to 0.36) 0.6338 −0.06 (−0.36 to 0.24) 0.6917

Gestational age (weeks) 0.03 (−0.09 to 0.16) 0.6223 −0.05 (−0.18 to 0.08) 0.4191

BMI (10 kg/m ) 2

0.09 (−0.15 to 0.34) 0.4499 −0.10 (−0.35 to 0.15) 0.4444

SAIL technique

Intercept 1.16 (−2.24 to 4.57) 0.4990 4.15 (−0.25 to 8.55) 0.0641

Age decade (10 years) −0.07 (−0.27 to 0.13) 0.4709 0.23 (−0.03 to 0.48) 0.0791

Gestational age (weeks) −0.02 (−0.10 to 0.07) 0.7140 −0.10 (−0.21 to 0.01) 0.0683

BMI (10 kg/m2) 0.05 (−0.12 to 0.21) 0.5802 −0.30 (−0.51 to −0.08) 0.0073

Distance between techniques (ICL 2 SAIL)

Absolute distance |ICL–SAIL| Distance (ICL–SAIL)

Variable B 95% CI p Value B 95% CI p Value

Intercept 4.64 (−0.49 to 9.76) 0.0756 −2.90 (−8.48 to 2.69) 0.3059

Age decade (10 years) 0.31 (0.01 to 0.60) 0.0430 −0.29 (−0.61 to 0.04) 0.0801

Gestational age (weeks) −0.09 (−0.22 to 0.04) 0.1517 0.05 (−0.09 to 0.19) 0.4867

BMI (10 kg/m2) −0.21 (−0.46 to 0.04) 0.1011 0.20 (−0.07 to 0.47) 0.1522

Abbreviations: BMI = body mass index; CI = confidence interval; ICL = intercristal line; SAIL = sacral anatomical interspace landmark technique.

second palpable interspace would actually be the L3-L4 in-
terspace. Because of the use of ultrasound in this study to define
each interspace, it remains unknown whether the patients in-
cluded in this study had lumbosacral transitional vertebrae.
Our results are consistent with what other investigators have
reported when assessing the LP site chosen by the ICL
technique. Studying 51 pregnant patients, 2 experienced
anesthesiologists were able to accurately determine the L4-L5
interspace in only 14% of their assessments.6 In another study,
when physicians examined 45 pregnant patients, they found
that the palpated ICL was located above the L4-L5 interspace
in all women enrolled in their study.20 Similarly, other
investigators found by ultrasound confirmation after LP that 6
LPs in 99 patients were carried out at the L1-L2 interspace,
with a considerable risk of needle contact with the conus
medullaris.7 Moreover, when a team of 4 anesthesiologists
with more than 5 years of experience in performing routine
LPs studied 100 consecutive patients, the clinical estimation
of the correct interspace occurred 29% of the time, as con-
firmed by MRI findings, with the actual space identified being
higher than assumed 68% of the time.21 In another study,
palpation successfully identified the intended interspace in
only 30% of cases, and intervertebral spaces in up to 27% of
cases were incorrectly identified by more than 1 level.22
The results of these studies reflect the inherent variability of the
ICL technique. The ICL’s normal distribution ranges from as
low as the L5-S1 interspace to as high as the L3-4 interspace in
nonpregnant patients.23 The variability of ICL distribution in
pregnancy is greater than in nonpregnant patients, where the
ICL determined by radiograph coincides with the L4 spinous
process or the L4-5 interspace 78.6% of the time but is as high
as the L3-4 interspace in only 3.7% of patients.24 In addition,
the soft tissue residing on top of the iliac crests is likely in-
creased in pregnancy, leading the clinician to higher than an-
ticipated assessment when using the ICL method.
In obese patients, the ICL method can be challenging,25 and
palpation of sacral interspaces through the SAIL can be con-
sidered as an alternative. However, we found that a significant
increase in BMI was associated with a 1-interspace decrease in
the SAIL-estimated distance from the L4-L5 level, but the
clinical significance of this finding remains to be established.
This study has several limitations. First, even though clinicians
have been performing caudal anesthesia for many decades, the
anatomical landmarks for caudal blocks do not take into
consideration the palpation of sacral structures that are
cephalad to the sacral hiatus and cornua.26 Therefore, clini-
cians may not be familiar with the palpation of the area above

8 Neurology | Volume , Number  | Month ▪▪, 2019 Neurology.org/N

Copyright © 2019 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
the sacral cornua and below the lumbo-sacral interspace and
we do not know how easy it would be to implement the SAIL Appendix (continued)
technique in clinical practice. Second, since 2 experienced Name Location Role Contribution
anesthesiologists performed the study, our data cannot be
manuscript for intellectual
generalized to other clinicians in different subspecialties or at content
different levels of training.
Baskar University of Author Performed clinical
Rajala, MD Michigan, assessment, revised
Further studies are needed to determine whether our findings Ann Arbor manuscript for intellectual
can be replicated in other clinical settings. Our study was content

performed exclusively in pregnant patients, so the generaliz- Christie University of Author Assisted with data collection
ability of our findings to the pediatric, male, elderly, and Vahabzadeh, Michigan, and performed lumbo-sacral
MD Ann Arbor ultrasound, interpreted data,
nonpregnant population remains undetermined. It is possible revised manuscript for
that other anatomic characteristics, unique to these other intellectual content
populations, affect the accuracy of each estimation technique. Ruth Cassidy, University of Author Performed statistical
Nonetheless, our findings strongly suggest that the conven- MA Michigan, analysis, interpreted data,
Ann Arbor revised manuscript for
tional ICL method for estimation of the site of LP should be intellectual content
reevaluated. Future studies should determine if our findings
Thomas T. University of Author Interpreted data, revised
can be reproduced in different patient populations, by a vari- Klumpner, Michigan, manuscript for intellectual
ety of medical practitioners, across diverse clinical settings. MD Ann Arbor content

Joanna A. University of Author Interpreted data, revised

Our study found that the SAIL technique was more likely than Kountanis, Michigan, manuscript for intellectual
the ICL to correctly identify the L4-L5 interspace and more MD Ann Arbor content

likely to result in an estimation below L3. Whenever ultrasound Madeline University of Author Recruited patients, assisted
or radiology guidance is not available, neurologists, emergency McCabe Michigan, with data collection,
Ann Arbor interpreted data, revised
medicine physicians, neurosurgeons, primary care physicians, manuscript for intellectual
intensivists, anesthesiologists, and others who perform LPs content
might find an estimation technique that reduces the risk of Dana Rector University of Author Recruited patients, assisted
misidentification of a higher lumbar interspace, like the SAIL Michigan, with data collection,
Ann Arbor interpreted data, revised
technique, useful in avoiding potential needle contact with the manuscript for intellectual
conus medullaris. However, further research is needed before content
this novel technique can be implemented in clinical practice. Casey Aman University of Author Assisted with data collection,
Michigan, interpreted data, revised
Acknowledgment Ann Arbor manuscript for intellectual
content
The authors thank Virginia E. Fulford, MS, CMI, Alar
Illustration, for artwork. Keerthana University of Author Designed and acquired
Sankar Michigan, approval for study
Ann Arbor
Study funding
Robert University of Author Designed and acquired
This study was funded by the University of Michigan De- Schoenfeld Michigan, approval for study
partment of Anesthesiology. Ann Arbor

Milo University of Author Analyzed data, study design,

Disclosure Engoren, MD Michigan, statistical analysis, acquired
The authors report no disclosures relevant to the manuscript. Ann Arbor approval for study, revised
manuscript for intellectual
Go to Neurology.org/N for full disclosures. content

Publication history
Received by Neurology April 17, 2019. Accepted in final form References
August 12, 2019. 1. Vickers A, Donnelly JP, Moore JX, Barnum SR, Schein TN, Wang HE. Epidemiology of
lumbar punctures in hospitalized patients in the United States. PLoS One 2018;13:e0208622.
2. Kroll H, Duszak R Jr, Nsiah E, Hughes DR, Sumer S, Wintermark M. Trends in
lumbar puncture over 2 decades: a dramatic shift to radiology. AJR Am J Roentgenol
2015;204:15–19.
3. Martin JA, Hamilton BE, Osterman MJ, Driscoll AK, Mathews TJ. Births: final data
for 2015. Natl Vital Stat Rep 2017;66:1.
4. Straus SE, Thorpe KE, Holroyd-Leduc J. How do I perform a lumbar puncture and
Appendix Authors analyze the results to diagnose bacterial meningitis? JAMA 2006;296:2012–2022.
5. Boon JM, Abrahams PH, Meiring JH, Welch T. Lumbar puncture: anatomical review
Name Location Role Contribution of a clinical skill. Clin Anat 2004;17:544–553.
6. Lee AJ, Ranasinghe JS, Chehade JM, et al. Ultrasound assessment of the vertebral level
Carlo University of Author Designed and conceptualized of the intercristal line in pregnancy. Anesth Analg 2011;113:559–564.
Pancaro, MD Michigan, study, performed clinical 7. Schlotterbeck H, Schaeffer R, Dow WA, Touret Y, Bailey S, Diemunsch P. Ultraso-
Ann Arbor assessment, drafted nographic control of the puncture level for lumbar neuraxial block in obstetric an-
aesthesia. Br J Anaesth 2008;100:230–234.

Neurology.org/N Neurology | Volume , Number  | Month ▪▪, 2019 9

Copyright © 2019 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
 8. Soleiman J, Demaerel P, Rocher S, Maes F, Marchal G. Magnetic resonance imaging
study of the level of termination of the conus medullaris and the thecal sac: influence
of age and gender. Spine 2005;30:1875–1880.
9. Reynolds F. Damage to the conus medullaris following spinal anaesthesia. Anaesthesia
2001;56:238–247.
10. Vandenbroucke JP, von Elm E, Altman DG, et al. Strengthening the reporting of
observational studies in Epidemiology (STROBE): explanation and elaboration.
Epidemiology 2007;18:805–835.
11. Sidak Z. Rectangular confidence regions for means of multivariate normal dis-
tributions. J Am Stat Assoc 1967;62:626–633.
12. Saifuddin A, Burnett SJ, White J. The variation of position of the conus medullaris in
an adult population: a magnetic resonance imaging study. Spine 1998;23:1452–1456.
13. Kim JT, Bahk JH, Sung J. Influence of age and sex on the position of the conus
medullaris and Tuffier’s line in adults. Anesthesiology 2003;99:1359–1363.
14. Davies JM, Posner KL, Lee LA, Cheney FW, Domino KB. Liability associated with
obstetric anesthesia: a closed claims analysis. Anesthesiology 2009;110:131–139.
15. Bagheri H, Govsa F. Anatomy of the sacral hiatus and its clinical relevance in caudal
epidural block. Surg Radiol Anat 2017;39:943–951.
16. Kim DH, Park JH, Lee SC. Ultrasonographic evaluation of anatomic variations in
the sacral hiatus: implications for caudal epidural injections. Spine 2016;41:
E759–E763.
17. Sekiguchi M, Yabuki S, Satoh K, Kikuchi S. An anatomic study of the sacral hiatus:
a basis for successful caudal epidural block. Clin J Pain 2004;20:51–54.
10 Neurology | Volume , Number  | Month ▪▪, 2019
Copyright © 2019 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.
18. Aggarwal A, Kaur H, Batra YK, Aggarwal AK, Rajeev S, Sahni D. Anatomic consid-
eration of caudal epidural space: a cadaver study. Clin Anat 2009;22:730–737.
19. Nakagawa T, Hashimoto K, Tsubakino T, Hoshikawa T, Inawashiro T, Tanaka Y.
Lumbosacral transitional vertebrae cause spinal level misconception in surgeries for
degenerative lumbar spine disorders. Tohoku J Exp Med 2017;242:223–228.
20. Margarido CB, Mikhael R, Arzola C, Balki M, Carvalho JC. The intercristal line
determined by palpation is not a reliable anatomical landmark for neuraxial anesthesia.
Can J Anaesth 2011;58:262–266.
21. Broadbent CR, Maxwell WB, Ferrie R, Wilson DJ, Gawne-Cain M, Russell R. Ability
of anaesthetists to identify a marked lumbar interspace. Anaesthesia 2000;55:
1122–1126.
22. Furness G, Reilly MP, Kuchi S. An evaluation of ultrasound imaging for identification
of lumbar intervertebral level. Anaesthesia 2002;57:277–280.
23. Hogan QH. Tuffier’s line: the normal distribution of anatomic parameters. Anesth
Analg 1994;78:194–195.
24. Render CA. The reproducibility of the iliac crest as a marker of lumbar spine level.
Anaesthesia 1996;51:1070–1071.
25. Halpenny D, O’Sullivan K, Burke JP, Torreggiani WC. Does obesity preclude lumbar
puncture with a standard spinal needle? The use of computed tomography to measure
the skin to lumbar subarachnoid space distance in the general hospital population. Eur
Radiol 2013;23:3191–3196.
26. Senoglu N, Senoglu M, Oksuz H, et al. Landmarks of the sacral hiatus for caudal
epidural block: an anatomical study. Br J Anaesth 2005;95:692–695.
Neurology.org/N
 Sacral anatomical interspace landmark for lumbar puncture in pregnancy: A
randomized trial
Carlo Pancaro, Baskar Rajala, Christie Vahabzadeh, et al.
Neurology published online December 12, 2019
DOI 10.1212/WNL.0000000000008749

This information is current as of December 12, 2019

Updated Information &
Services
Subspecialty Collections
Permissions & Licensing
Reprints
including high resolution figures, can be found at:
http://n.neurology.org/content/early/2019/12/12/WNL.0000000000008
749.full
This article, along with others on similar topics, appears in the
following collection(s):
All Clinical trials
http://n.neurology.org/cgi/collection/all_clinical_trials
Patient safety
http://n.neurology.org/cgi/collection/patient__safety
Spinal cord trauma; see Trauma/spinal cord trauma
http://n.neurology.org/cgi/collection/spinal_cord_trauma-see_trauma-s
pinal_cord_trauma
Ultrasound
http://n.neurology.org/cgi/collection/ultrasound
Information about reproducing this article in parts (figures,tables) or in
its entirety can be found online at:
http://www.neurology.org/about/about_the_journal#permissions
Information about ordering reprints can be found online:
http://n.neurology.org/subscribers/advertise

Neurology ® is the official journal of the American Academy of Neurology. Published continuously since
1951, it is now a weekly with 48 issues per year. Copyright © 2019 American Academy of Neurology. All
rights reserved. Print ISSN: 0028-3878. Online ISSN: 1526-632X.