ACADEMIC EMERGENCY MEDICINE • January 2001, Volume 8, Number 1
9
Data Analysis. Means, medians, ranges, inter-
quartile ranges, and 95% confidence intervals (95%
CIs) of means (the primary outcome analysis) were
calculated. These calculations describe the differ-
ences in each subject’s change in CSFp with each
of the following position changes:
1. change from flexed to extended position
2. change from extended to flexed position, and
3. change from the first to the third measurement
(same position).
M
ETHODS
Study Design. This was a prospective crossover
study. Patients were randomly assigned to one of
two sequences for three successive measurements
of CSFp. Sequence A was: flexed, extended, then
flexed. Sequence B was: extended, flexed, then ex-
tended. Our institutional review board approved
this study. The study was considered exempt from
informed consent.
Differences are expressed as cm H2O or percentage
changes from the first reading obtained. Because
the assumptions underlying the usual parametric
and the nonparametric CIs for the differences be-
tween the paired changes did not appear to be sat-
isfied, bootstrap 95% CIs (with 1,000 replications)
using the percentile method and two-sided p-val-
ues were constructed using intercooled Stata sta-
tistical software, version 6.0 (Stata Corp., College
Station, TX, 1999).
Study Setting and Population. The study was
performed at an urban emergency department
(ED) with an emergency medicine residency train-
ing program and an annual census of 47,000 and
enrolled a convenience sample of adults, deter-
mined by the attending physician to require a lum-
bar puncture during their ED evaluations. Pa-
tients were enrolled from April 1997 to December
1997. Exclusion criteria were: contraindication to
lumbar puncture, hemodynamic instability, preg-
nancy, bleeding diathesis, and inability to assume
the positions being studied.
RESULTS
Nineteen patients were studied in sequence A,
while 20 were studied in sequence B. Procedures
were performed by 18 different physicians. The fi-
nal pressure measurement was not taken for one
patient in sequence B. The range of patients’ ages
was from 16 to 84 years (mean = 39). Review of
data available retrospectively from medical records
revealed the following about the sample. Men com-
prised 66% (25/38). Lumbar punctures were per-
formed most commonly for suspected meningitis,
possible subarachnoid hemorrhage, or unexplained
alterations in mental status. Sedation or neuro-
muscular blockade occurred in 32% (12/38), using
the following drugs: lorazepam alone (n = 2), mid-
azolam alone (n = 2), haloperidol alone (n = 2), hal-
operidol and lorazepam (n = 2), morphine sulfate
with midazolam (n = 1), and vecuronium with mid-
azolam (n = 1). Information regarding which se-
dating or paralyzing drugs were used was una-
vailable for two patients. Only 22% of the patients
(7/32) were discharged home. Cerebrospinal fluid
was abnormal in 26% (10/39). Meningitis was
found in three, elevated CSF white blood cell
(WBC) count with no growth on cultures in five,
and elevated protein levels (not explained by he-
molysis and without elevated CSF WBC counts) in
two. The CSF was normal or had abnormalities
most likely resulting from being traumatic in the
other 29 patients.
Study Protocol. The patients’ physicians per-
formed lumbar punctures in the lateral decubitis
position using a 20-gauge needle (Adult Lumbar
Puncture Tray, Lot # 47885, Kendall Co., Mans-
field, MA), sterile technique, with knees and hips
flexed. The thoracolumbar spine was in the neutral
position. Standard sterile preparation was per-
formed at the lumbar puncture site. Lidocaine was
used for local anesthesia. Blood pressure and pulse
from the patient’s exposed upper extremities were
recorded with each lumbar puncture pressure
measurement. Upon entrance to the subarachnoid
space (confirmed by free flow of CSF), the lower
extremities were then positioned for serial CSFp
measurements according to either sequence A or B.
Using the manometer included in the lumbar
puncture kit, pressure measurements were re-
corded in three positions consecutively. Flexion
was defined as 30 degrees of neck flexion and 90
degrees of hip and knee flexion. Extension was de-
fined as neutral neck position and knees and hips
held at 180 degrees. Withdrawal of CSF was with-
held until after the three CSFp measurements
were recorded.
The main outcome measures were the changes
in CSF opening pressures obtained when moving
the lower extremities from flexed to extended, and
from extended to flexed positions. To determine
whether serial testing itself affected CSFp values,
differences between pressures obtained in the
same position (i.e., the first and third pressures
taken for each subject) were also calculated. Dif-
The CSFp values for the patients studied in se-
quence A (flexed, extended, flexed) are depicted in
Figure 1. Those for the patients studied in se-
quence B (extended, flexed, extended) are shown
ferences of about 2 cm H2O or 10%, were thought in Figure 2. Only two CSFp measurements were
unlikely to be clinically meaningful. obtained for one patient in sequence B. Descriptive