Clinical Neurology and Neurosurgery 112 (2010) 752757

Contents lists available at ScienceDirect

Clinical Neurology and Neurosurgery

journal homepage: www.elsevier.com/locate/clineuro

Accuracy of physical signs for detecting meningitis: A hospital-based

diagnostic accuracy study
Swati Waghdhare, Ashwini Kalantri, Rajnish Joshi, Shriprakash Kalantri
Department of Medicine, Mahatma Gandhi Institute of Medical Sciences, Sevagram, Dist. Wardha, 442102 Maharashtra, India

a r t i c l e

i n f o

Article history:
Received 5 November 2009
Received in revised form 6 June 2010
Accepted 9 June 2010
Available online 7 July 2010
Keywords:
Meningitis
Physical examination
Sensitivity and specicity
Likelihood ratios

a b s t r a c t
Objectives: To evaluate accuracy of physical signs for detecting meningitis.
Patients and methods: We enrolled patients aged 12 years or more, admitted with acute encephalitis
syndrome (fever, headache, altered mental status, vomiting, seizures, neurodecit) to a rural teaching
hospital. The design was a double-blind, cross-sectional analysis of consecutive patients, independently
comparing signs of meningeal inammation (nuchal rigidity, head jolt accentuation of headache, Kernigs
sign and Brudzinskis sign) elicited by internal medicine residents against an established reference standard (cerebrospinal uid white cell count >5 white cells/L). Diagnostic accuracy was measured by
computing sensitivity, specicity and likelihood ratios (LRs) and their 95% condence interval (CI) values.
Results: Of 190 patients (119 men, 71 women; ages 1381 years; mean 38(SD 18) years) CSF analysis
identied meningitis in 99 (52%; 95% CI 44, 59%) patients. No physical sign of meningeal irritation could
accurately distinguish those with and without meningitis: nuchal rigidity (LR+ 1.33 (0.89, 1.98) and LR
0.86 (0.70, 1.06)), head jolt accentuation of headache (LR+ 5.52 (0.67, 44.9) and LR 0.95(0.89, 1.00)),
Kernigs sign (LR+ 1.84 (0.77, 4.35) and LR 0.93(0.84, 1.03)) and Brudzinskis sign (LR+ 1.69 (0.65, 4.37)
and LR 0.95 (0.87, 1.04)).
Conclusion: Physical signs of meningeal inammation do not help clinicians rule in or rule out meningitis
accurately. Patients suspected to have meningitis should undergo a lumbar puncture regardless of the
presence or absence of physical signs.
2010 Elsevier B.V. All rights reserved.

1. Introduction
Meningitis is an important cause of morbidity and mortality,
worldwide. Bacterial meningitis is among the 10 most common
infectious causes of death and kills estimated 135,000 people
throughout the world each year [1]. It is estimated that about a
quarter of adults with bacterial meningitis [2,3], and a third of all
patients with tuberculous meningitis die [4]. Another one-fourth
of survivors develop transient or permanent neurologic morbidity
[2,3]. Early recognition of this serious infection in primary care settings is important; so as to initiate timely life saving treatments
and appropriate referrals. This in turn can reduce mortality and
morbidity in meningitis.
For over 100 years, clinicians have used three physical signs
nuchal rigidity, Kernigs and Brudzinskis signs to help diagnose
meningitis at bedside and to decide need for lumbar puncture,
or more intensive care. Although Verghese and Gallemore [5]
argued that the physical signs of meningeal irritation may aid
in early diagnosis and treatment of meningitis and are excellent

Corresponding author. Tel.: +91 9970168766.

E-mail address: sp.kalantri@gmail.com (S. Kalantri).
0303-8467/$ see front matter 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.clineuro.2010.06.003

demonstrations for medical students and house staff of the art of

the bedside examination, there is limited information about the
accuracy of these signs for detection of meningitis. A prospective
cross-sectional study [6] has concluded that physical signs do not
accurately discriminate between patients with and without meningitis (Kernigs sign (sensitivity, 5%; likelihood ratio for a positive test
result (LR+), 0.97)), Brudzinskis sign (sensitivity, 5%; LR+, 0.97), and
nuchal rigidity (sensitivity, 30%; LR+, 0.94). Another study [7] that
evaluated 54 patients with fever and new headaches reported that
jolt accentuation of headache may be the best sign for meningitis
(sensitivity, 97%; LR+, 2.4).
Despite poor accuracy, the physical signs of meningeal irritation continue to be part of bedside clinical teaching and practice.
In resource limited settings, cerebral malaria, tuberculous meningitis and bacterial meningitis are close differential diagnoses in
patients with fever and impaired mental status. In these settings
clinicians cannot always perform a lumbar puncture or lack access
to cerebrospinal uid (CSF) microscopy. They, therefore, use clinical signs to distinguish meningitis from encephalopathy such as
cerebral malaria. It is important for the clinicians to know if the
signs of meningeal irritation are accurate enough to obviate the
need for CSF examination. We carried out this study to nd out
the diagnostic accuracy of four physical signs nuchal rigidity, jolt

S. Waghdhare et al. / Clinical Neurology and Neurosurgery 112 (2010) 752757

accentuation of headache (hereafter called head jolt sign), Kernigs

sign and Brudzinskis sign to detect meningitis among hospitalized patients, 12 years of age or older, suspected to have meningitis
with cerebrospinal uid cell count as the reference standard.
2. Subjects and methods
2.1. Patients, setting and data collection
This study was performed at a 620 bed teaching hospital with
400,000 patient visits and about 10,000 patient admissions to the
internal medicine wards, each year. Typically about 200 patients
with fever, headache and altered mental status (clinically classied
as acute encephalitis syndrome or AES) seek care at this hospital every year [8]. These patients typically are usually tested for
malaria (microscopy and rapid diagnostic tests) and evaluated for
a metabolic encephalopathy (glucose, electrolytes and renal panel
tests). Patients testing negative for malaria and not found to have
a metabolic encephalopathy undergo a lumbar puncture and CSF
microscopy to determine cause of AES. Treating physicians use on
admission Glasgow coma score (GCS) to grade severity of disease,
and make a nal diagnosis based on clinical prole, results of CSF
based tests, clinical biochemistry, and neuro-imaging.
For a period of 1-year, beginning May 2008, we prospectively
enrolled all consecutive patients with AES (fever, headache, and
altered mental status, with or without seizures or focal neurological
decit) in whom treating physicians had ordered CSF examination.
Blind to the diagnosis, physical ndings and laboratory data, the
ICU residents, who were aware of the study, used standardized
examination techniques [1] to elicit the following physical signs
on each patient before they underwent a lumbar puncture.
Physical sign

Method of
elicitation

Positive test

Nuchal rigidity

With the patient in

the supine position,
the resident gently
exed the neck,
asking the patients
to touch their chin
to sternum
The resident asked
the patients to turn
their heads
horizontally at a
frequency of 23
rotations per
second
With the patient in
the supine position,
the resident lifted
the knee in exed
position until
maximal hip
exion was
obtained. The leg
was extended at
the knee and
resistance was
checked
With the patient in
the supine
position, the
resident exed the
neck, and looked
for exion of both
the lower limbs

Resistance to
exion

Jolt accentuation of
the patients
headache

Kernigs sign

Brudzinskis sign

Worsening of the
base line headache

753

recording-sheet was deposited in a box placed in the intensive care

unit of the hospital.
Since lumbar punctures are often done in the middle of night
and cells in the CSF need to be counted within 30 min of a lumbar puncture, a medicine resident (SW) underwent training to do
CSF cell counting using a haemocytometer (Neubauers chamber).
The methodology was pilot tested and cell counts performed by
the medical resident and an experienced microscopist were compared and shown to be reproducible in a set of 25 patients. It was
ensured that time-interval between recording of physical signs,
lumbar puncture, and CSF cell counting does not exceed 60 and
30 min, respectively. The resident (SW) was blind to the history
and results of the physical examination until after the CSF counts
were entered in data collection forms.
Meningitis was dened as >5 white blood cells/L of CSF.
Patients were excluded from the study if lumbar puncture was traumatic (dened as either grossly bloody CSF, or if red blood cell count
in CSF was >400 cells/L). A differential count was also obtained
from the wet smear. Additional tests (CSF sugars, proteins, and bacterial cultures) were performed in all samples as a standard of care.
Further CSF tests (e.g. mycobacterial cultures, specic viral diagnostics) were ordered based on discretion of the treating physician,
cost, and availability.
The study protocol was approved by the institutional ethics
committee, and consent was obtained from patients or their relations to elicit physical signs and to do lumbar punctures.
2.2. Statistical analysis
We assessed the accuracy of physical signs in diagnosing meningitis by calculating sensitivity, specicity, positive and negative
predictive values and positive and negative likelihood ratios. In
a post hoc analysis we estimated diagnostic accuracy of physical
signs by sub-classifying individuals with meningitis by [1] severity
of meningeal inammation: mild (CSF cells 6100/L), moderate
(1011000 cells/L) and severe (>1000 cells/L); [2] predominant
cells in CSF: lymphocytes, neutrophils, and mixed, using 75% cells
as a cutpoint; [3] tertiles of Glasgow coma score and [4] discharge
diagnosis of type of meningitis (aseptic, tuberculous or bacterial).
We used diagti command in STATA (version 10, Stata Corp. College
Road, TX, USA) to calculate point estimates of accuracy and their
95% condence intervals.
3. Results

Resistance to
extension at the
knee to >135 or
pain in the lower
back or posterior
thigh

Flexion of the
knees and hips

ICU resident recorded the results of these tests on a specially

designed sheet. This sheet was folded, and sealed to ensure that
index tests were interpreted blinded to and independent of the
reference standard. Before performing lumbar puncture the sealed

We have used STARD (Standards for Reporting Diagnostic Accuracy Study) guidelines to report this study. Figure shows the study
prole. Between May 2008 and July 2009, we enrolled 204 patients.
We could not use data from 14 patients because the lumbar puncture was traumatic. Thus, our nal sample consisted of 190 patients
(119 men, 71 women); ages 1381 years [mean 38 (SD 18) years].
CSF analysis identied meningitis in 99 of 190 (52%) patients. The
diagnosis based on clinical prole, cerebrospinal uid ndings, and
neuro-imaging for these 99 patients was aseptic meningitis (n = 62
(63%)), tuberculous meningitis (n = 30 (31%)) and bacterial meningitis (n = 7 (7%)). There were only seven patients with conrmed
bacterial meningitis, and 13 with a predominantly neutrophilic
leucocytosis. The nal discharge diagnosis of those classied in
non-meningitis group (n = 91) consisted of acute encephalitis of
undermined etiology, acute hepatic encephalopathy, metabolic
encephalopathy, alcoholic encephalopathy, cerebral malaria, brain
abscess, delirium, pesticide poisoning, seizure disorder, sepsis,
stroke and subdural haemorrhage.
Table 1 describes patient characteristics according to the presence (n = 99) or absence (n = 91) of meningitis. Also, shown in this

754

S. Waghdhare et al. / Clinical Neurology and Neurosurgery 112 (2010) 752757

Table 1
Characteristics of patients with suspected meningitis.
Characteristic

Patients without
meningitis (n = 91)

Patients with
meningitis

All patients
(n = 190)

Mild
inammation
(n = 33)

Moderate
inammation
(n = 50)

Severe
inammation
(n = 16)

Age (mean, SD) years

No. (%) of patients >60 years old
Female sex

39.6 (1.9)
14 (15.3)
32 (45.7)

40.5 (16.8)
3 (9)
16 (22.5)

35.5 (17.9)
7 (14)
18 (25.5)

36.37 (17.5)
1 (6)
5 (7.1)

38.4 (18.2)
25 (13.1)
71 (38)

CSF ndings
WBC count, median (IQR)
Protein level, median (IQR)
Glucose level, median (IQR)

0 (00)
30 (2042)
69 (5595)

52 (4080)
70 (30100)
68 (5395)

185 (115475)
80 (40180)
51 (4069)

1265 (11751632)
107 (57140)
74 (5694)

25 (0170)
40 (2590)
64.5 (4985)

Because of rounding, the sum of percentage may not be 100.

Meningitis: dened as >5 WBCs/L of cerebrospinal uid.
SD, standard deviation; IQR, inter-quartile range; CSF, cerebrospinal uid; GCS, Glasgow coma scale.
Figures in parenthesis denote %, unless specied otherwise.
Meningeal inammation: mild = 5100 WBC/L of CSF, moderate = 1011000 WBC/L and severe = >1000 WBC/L.

table is age, distribution of symptoms, physical signs and CSF

chemistry in patients with meningitis, categorized according to
the severity of meningeal inammation. All patients were febrile
on admission. Compared to patients without meningitis, patients
with meningitis were more likely to report seizures (24% vs.16%;
p = 0.24) but were as likely to have altered sensorium (68% vs. 70%;
p = 0.80). Only 17 (17%) patients with meningitis had the classic
triad of fever, nuchal rigidity, and a change in mental status. Patients
with meningitis tended to have lower glucose and higher protein
compared to those without.
3.1. Accuracy of physical signs
All the four physical signs had a poor accuracy. While nuchal
rigidity was most sensitive (sensitivity of 39.4; 95% CI 29.7, 49.7%),
head jolt sign had highest specicity (98.9; 95% CI 94, 100%). The
accuracy estimates of Kernigs and Brudzinskis signs were intermediate between these two extremes. None of the positive or negative
likelihood ratios were statistically signicant (Fig. 1).
Of the 190 study patients, 33 (17.3%) had evidence of
mild meningeal inammation (5100 WBC/L of CSF), 50
(26.3%) had evidence of moderate meningeal inammation
(1011000 WBC/L) and 16 (8.4%) had evidence of severe
meningeal inammation (>1000 WBC/L). In a subset of patients
with severe meningeal inammation, no patient had a positive
head jolt sign, Kernigs sign or Brudzinskis sign. Nuchal rigidity
had a poor sensitivity (12.5%) and low positive likelihood ratio
(0.42) for diagnosis of severe inammation. The +LRs for those
with moderate meningeal inammation ranged between 0 and
1.69 while LRs for severe inammation ranged between 0.71 and
1.42. The degree of meningeal inammation did not inuence the
operating characteristics of the physical signs to detect meningitis.
(Table 2). We also stratied the test results according to types of
cells in CSF and three tertiles of Glasgow coma scale (data not
shown). In no subgroup did LRs swing signicantly farther from 1
and provided little change in probability that a patient has or does
not have meningitis. Among patients classied according to the
type of meningitis, except Kernigs sign (LR+ 5.57, 95% CI, 1.83, 17)
all physical signs were associated with unhelpful likelihood ratios
that also lacked statistical signicance (Table 3).
4. Discussion
The main nding of our study is that none of the physical signs
used to assess meningeal inammation are accurate enough to
condently rule in or rule out meningitis in adult inpatients sus-

pected to have meningitis. These physical signs also perform poorly

regardless of the severity of meningeal inammation, type of cells
in CSF, severity of disease, and type of meningitis.
Textbooks of medicine describe, and students routinely learn
and elicit, physical signs to detect meningitis. Of these signs, nuchal
rigidity and Kernigs signs are almost universally elicited in all
patients suspected to have meningitis; Brudzinskis sign and the
jolt accentuation of headache sign have not become as popular. In Brudzinskis original description of 42 cases of meningitis
(including 21 cases of tuberculous meningitis), Kernigs sign had
a sensitivity of 57%, while Brudzinski nape of the neck sign had a
sensitivity of 97% and the contralateral reex sign had a sensitivity
of 66% [5]. Interestingly, although Vladimir Kernig described the
physical sign in a German journal in 1884 that epitomizes him and
Josef Brudzinski described the physical sign in a Polish journal in
1909 that bears his name [5], clinicians accepted these signs on their
face value and made little attempt to evaluate their diagnostic accuracy in well-designed cross-sectional studies. Thus generations of
medical students have learnt these physical signs and hundreds of
thousands of patients had these signs performed on them, without
rm evidence on their accuracy.
Studies that have evaluated the accuracy of these signs are
retrospective [3,916], have a small sample size [914,17], have
evaluated people aged 60 and older [9,10,17], have included only
patients with bacterial meningitis [3,1012,14,16] or very small
number of patients with tuberculous meningitis [7,17] or aseptic meningitis [7,9,13] or have not stated the method of elicitation
of physical signs [3,1014,16]. Moreover, except two studies [6,7]
no study has enrolled patients with and without meningitis, and
therefore specicity of the physical signs in other studies cannot be calculated. The pooled sensitivity of nuchal rigidity in a
meta-analysis of 10 studies [18] (nine studies of retrospective
chart review) of 824 patients (845 episodes) was 70% (95% condence interval, 58, 82%). By contrast, Thomas et al. [6] in their
well-designed prospective diagnostic accuracy study reported that
nuchal rigidity (LR+ 0.94 and LR 0.92) was too poor a physical
sign to rule in or rule out meningitis. Our estimates of diagnostic
accuracy of nuchal rigidity (LR+ 1.33 [0.89, 1.98] and LR 0.86 [0.7,
1.06]) are consistent with these observations. The prior probability of meningitis in our study was 52%; nuchal rigidity, if present,
raised the probability of meningitis to 59%; a supple neck reduced
the probability to 48%. Thus the presence or absence of nuchal
rigidity did not result in clinically meaningful changes in post-test
probability of meningitis. Older people frequently suffer from cervical spondylosis which could result in nuchal rigidity despite the
absence of meningitis 35% of the geriatric patients on acute-care

S. Waghdhare et al. / Clinical Neurology and Neurosurgery 112 (2010) 752757

755

Fig. 1. Study ow and diagnostic accuracy of physical signs for detecting meningitis.

and rehabilitation wards had nuchal rigidity in one series [19]. In

our study only 25 (13%) patients were aged 60 and above; the diagnostic accuracy of nuchal rigidity might have been much lower had
our study population been older.
Previous studies have shown that Kernigs sign and Brudzinskis
signs occur infrequently among patients with suspected meningitis
only 5% each in 297 patients in a prospective cross-sectional study
from America [6] and 9% of 54 patients in a study from Japan [7].

Our observations are almost similar: Kernings sign was positive in

12% of 190 patients. The low sensitivity of Kernigs sign indicates
that most patients with meningitis could be missed if clinicians
use it as a screening test to order a lumbar puncture. In line with
a previous study [6], the specicity of these signs was close to 95%
in our study. If very few patients test positive for a physical sign,
despite high specicity, the positive predictive value of the sign
remains low as shown by our study.

Table 2
Diagnostic accuracy of physical signs in different categories of meningeal inammation.
Clinical sign

TP

FN

FP

TN

Sensitivity

Specicity

Mild inammation
Nuchal rigidity
Head jolt sign
Kernigs sign
Brudzinskis sign

12
0
2
2

21
33
31
31

27
1
7
6

64
90
84
85

36.4 (20.4, 54.9)

0 (0, 20.6)
6.1 (0.74, 20.2)
6.1 (0.74, 20.2)

70.3 (59.8, 79.5)

98.9 (94, 100)
92.3 (84.8, 96.9)
93.4 (86.2, 97.5)

1.23 (0.70, 2.13)

0.79 (0.17, 3.6)

0.92 (0.19, 4.33)

0.90 (0.67, 1.21)

1.01 (0.98, 1.03)
1.02 (0.91, 1.13)
1.01 (0.91, 1.11)

Moderate inammation
Nuchal rigidity
Head jolt sign
Kernigs sign
Brudzinskis sign

25
6
12
9

25
44
38
41

27
1
7
6

64
90
84
85

50 (35.5, 64.5)
12 (4.53, 24.3)
24 (13.1, 38.2)
18 (8.58, 31.4)

70.3 (59.8, 79.5)

98.9 (94, 100)
92.3 (84.8, 96.9)
93.4 (86.2, 97.5)

1.69 (1.11, 2.57)

10.9 (1.35, 88.2)
3.12 (1.31, 7.42)
2.73 (1.03, 7.23)

0.71 (0.52, 0.96)

0.89 (0.80, 0.98)
0.82 (0, 0.97)
0.87 (0.76, 1.01)

2
0
0
0

14
16
16
16

27
1
7
6

64
90
84
85

12.5 (1.55, 38.3)

0 (0, 20.6)
0 (0, 20.6)
0 (0, 20.6)

70.3 (59.8, 79.5)

98.9 (94, 100)
92.3 (84.8, 96.9)
93.4 (86.2,97.5)

0.42 (0.11, 1.6)

1.24 (0.99, 1.56)

1.01 (0.99, 1.03)
1.08 (1.02, 1.15)
1.07 (1.03, 1.16)

Severe inammation
Nuchal rigidity
Head jolt sign
Kernigs sign
Brudzinskis sign

LR+

LR

TP, true positive; FN, false negative; FP, false positive; TN, true negative; LR+, positive likelihood ratio; LR, negative likelihood ratio. Meningeal inammation:
mild = 5100 WBC/L of CSF, moderate = 1011000 WBC/L and severe = >1000 WBC/L.

756

S. Waghdhare et al. / Clinical Neurology and Neurosurgery 112 (2010) 752757

Table 3
Diagnostic accuracy of physical signs in different types of meningitis.
Clinical sign

TP

FN

FP

TN

Sensitivity

Specicity

Aseptic meningitis (n = 62)

Nuchal rigidity
18
Head jolt sign
1
Kernigs sign
4
Brudzinskis sign
3

44
61
58
59

27
1
7
6

64
90
84
85

29.0 (18.2, 41.9)

1.61 (0, 04,8.6)
6.45 (1.79,15.7)
4.84 (1.01,13.5)

70.3 (59.8, 79.5)

98.9 (94, 100)
92.3 (84.8, 96.9)
93.4 (86.2,97.5)

0.97 (0.59, 1.62)

1.47 (0.09,23.0)
0.83 (0.25, 2.74)
0.73 (0.19, 2.82)

1.01 (0.82, 1.24)

0.99 (0.95, 1.03)
1.01 (0.92, 1.11)
1.02 (0.94, 1.11)

Tuberculous meningitis (n = 30)

Nuchal rigidity
17
Head jolt sign
3
Kernigs sign
7
Brudzinskis sign
7

13
27
23
23

27
1
7
6

64
90
84
85

56.7 (37.4, 74.5)

10.0 (2.1,26.5)
23.3 (9.93, 42.3)
23.3 (9.93, 42.3)

70.3 (59.8, 79.5)

98.9 (94, 100)
92.3 (84.8, 96.9)
93.4 (86.2,97.5)

1.91 (1.22, 2.98)

9.1 (0.98,84.2)
3.03 (1.25, 7.95)
3.54 (1.29, 9,71)

0.61 (0.40, 0.94)

0.91 (0.80, 1.03)
0.83 (0.67,1.02)
0.82 (0.66, 1.01)

3
5
4
6

27
1
7
6

64
90
84
85

57.1 (18.4, 90.1)

28.6 (3.67,71.0)
42.9 (9.9,81.6)
14.3 (0.36, 57.9)

70.3 (59.8, 79.5)

98.9 (94, 100)
92.3 (84.8, 96.9)
93.4 (86.2,97.5)

1.93 (0.94,3.94)
26.00 (0.26,253)
5.57 (1.83,17)
2.17 (0.30,15.6)

0.60 (0.25,1.45)
0.72 (0.45, 1.15)
0.61 (0.32,1.18)
0.91 (0.67,1.25)

Bacterial meningitis (n = 7)
Nuchal rigidity
Head jolt sign
Kernigs sign
Brudzinskis sign

4
2
3
1

LR+

LR

TP, true positive; FN, false negative; FP, false positive; TN, true negative; LR + , positive likelihood ratio; LR, negative likelihood ratio. Meningeal inammation:
mild = 5100 WBC/L of CSF, moderate = 1011000 WBC/L and severe = > 1000 WBC/L.

Only one study [7] has tested the accuracy of accentuation of

headaches by head jolt. In 1993 Uchihara and Tsukagoshi prospectively examined the clinical signs of 54 febrile patients with new
headaches. Jolt accentuation of headaches, a novel sign, had a
sensitivity of 97%, a specicity of 60%, a LR+ of 2.4 and LR of
0. The authors argued that all febrile patients with recent-onset
headaches and jolt accentuation should undergo a lumbar puncture even in the absence of nuchal rigidity or Kernigs sign and lack
of accentuation of headache on physical examination may essentially exclude meningitis (LR, 0). Noteworthy, this study sampled
only 54 relatively young patients (mean age, 39 years) at a time
enteroviral meningitis was prevalent in Japan. The low frequency
of this sign in our study (7 of 190 patients) and poor likelihood
ratios that the sign yielded indicate that clinicians cannot rely on
this sign to conrm or exclude meningitis.
We also wondered if severity of meningeal inammation could
inuence the frequency of positive physical signs. We used CSF
cell count to categorize meningeal inammation mild, moderate
and severe. Contrary to our assumption that patients with severely
inamed meninges would have a higher frequency of the physical
signs, our study data did not support our assumption. The diagnostic accuracy of all physical signs was poor regardless of the
severity of meningeal inammation. In fact, no patient with severe
meningeal inammation had a positive head jolt sign, Kernigs sign
or Brudzinskis sign. Our results are in line with those reported
by Thomas et al. [6] who also found that the diagnostic accuracy
of Kernigs sign and Brudzinskis sign was poor for patients with
moderate and severe meningeal inammation. We also found that
among subgroups classied by type of cellular response, severity of
disease, and type of meningitis, the diagnostic accuracy of physical
signs was too low to make large and clinically meaningfully shifts
between pretest and post-test probability.
When clinicians suspect meningitis, they use physical signs of
meningeal inammation not to rule in but to rule out meningitis. These signs generate likelihood ratios which can be used
to rene clinical diagnosis. A physical sign with LR close to 0
would almost rule out meningitis and assure clinicians that they
would not be missing meningitis by not ordering a lumbar puncture. By contrast, physical signs with LR close to unity would
hardly reduce the probability of meningitis. In our study, among a
subgroup of patients with meningitis and mild meningeal inammation (Table 2), 21 of 33 patients with meningitis (64%) did not
have nuchal rigidity, as did 64 of 91 patients without meningitis (70%). The LR for nuchal rigidity (64%/70% = 0.90) suggests
that absence of nuchal rigidity among patients suspected to have

meningitis cannot exclude meningitis because it is almost as likely

in someone who has meningitis than in a patient without meningitis. Other signs of meningeal inammation also yielded LRs not
small enough to rule out meningitis with condence.
Our study has several strengths. We did a prospective
cross-sectional study in which the results of physical signs on
consecutively recruited groups of patients were cross-classied
against meningitis status determined by the reference standard
(CSF microscopy). Both the index tests and the reference standard
were performed in a blind and independent manner on the same
group of patients. Our spectrum of patients includes patients with
mild, moderate and severe meningeal inammation and disorders
with which meningitis is usually confused at bedside. Because our
study was prospective, our test results were not inuenced by the
knowledge of laboratory tests. We standardized the elicitation and
interpretation of physical signs. Also, we excluded patients with
traumatic lumbar punctures.
We acknowledge that our study has several limitations. We
enrolled hospitalized patients aged 12 and older, and therefore
our study results cannot be generalized to community settings and
to children. Second, the prevalence of bacterial meningitis in our
study was only 3.6%, thus our results may not be generalizable to
those settings with a higher prevalence of bacterial or tuberculous
meningitis. Third, although we enrolled 190 patients in the study,
95% condence intervals around the point estimates are not tight,
indicating that the sample size was small. Finally, this study does
not address the issue of reproducibility.
In conclusion, none of the physical signs used to assess
meningeal inammation are accurate enough to condently rule
in or rule out meningitis in adult inpatients suspected to have
meningitis. These physical signs also perform poorly regardless of
the severity of meningeal inammation. Their presence or absence
does not lead to large and conclusive change from pretest to
post-test probability of meningitis. Clinicians should not trust
physical signs to assess the need for performance of lumbar puncture. Patients with suspected meningitis must undergo a lumbar
puncture regardless of presence or absence of physical signs of
meningeal irritation.
References
[1] Tunkel AR. Clinical features and diagnosis of acute bacterial meningitis in
adults. In: Basow DS, editor. UpToDate. Waltham, MA: UpToDate; 2009.
[2] Aronin SI, Peduzzi P, Quagliarello VJ. Community-acquired bacterial meningitis: risk stratication for adverse clinical outcome and effect of antibiotic
timing. Ann Intern Med 1998;129(December 1 (11)):8629.

S. Waghdhare et al. / Clinical Neurology and Neurosurgery 112 (2010) 752757

[3] Durand ML, Calderwood SB, Weber DJ, Miller SI, Southwick FS, Caviness Jr VS,
et al. Acute bacterial meningitis in adults. A review of 493 episodes. N Engl J
Med 1993;328(January 7 (1)):218.
[4] Thwaites G, Chau TT, Mai NT, Drobniewski F, McAdam K, Farrar J. Tuberculous
meningitis. J Neurol Neurosurg Psychiatry 2000;68(March (3)):28999.
[5] Verghese A, Gallemore G. Kernigs and Brudzinskis signs revisited. Rev Infect
Dis 1987;9(NovemberDecember (6)):118792.
[6] Thomas KE, Hasbun R, Jekel J, Quagliarello VJ. The diagnostic accuracy of
Kernigs sign, Brudzinskis sign, and nuchal rigidity in adults with suspected
meningitis. Clin Infect Dis 2002;35(July (1)):4652.
[7] Uchihara T, Tsukagoshi H. Jolt accentuation of headache: the most sensitive
sign of CSF pleocytosis. Headache 1991;31(March (3)):16771.
[8] Joshi R, Colford Jr JM, Reingold AL, Kalantri S. Nonmalarial acute undifferentiated fever in a rural hospital in central India: diagnostic uncertainty and
overtreatment with antimalarial agents. Am J Trop Med Hyg 2008;78(March
(3)):3939.
[9] Behrman RE, Meyers BR, Mendelson MH, Sacks HS, Hirschman SZ. Central
nervous system infections in the elderly. Arch Intern Med 1989;149(July
(7)):15969.
[10] Domingo P, Mancebo J, Blanch L, Coll P, Net A, Nolla J. Acute bacterial meningitis
in the elderly. Arch Intern Med 1990;150(July (7)), 1546, 8.
[11] Genton B, Berger JP. Meningitis in adults at the Vaudois University Hospital
Center. Diagnosis, management and development of 112 cases. Rev Med Suisse
Romande 1988;108(September (9)):795803.

757

[12] Gorse GJ, Thrupp LD, Nudleman KL, Wyle FA, Hawkins B, Cesario TC. Bacterial meningitis in the elderly. Arch Intern Med 1984;144(August (8)):1603
7.
[13] Magnussen CR. Meningitis in adults; ten-year retrospective analysis at community hospital. N Y State J Med 1980;80(May (6)):9016.
[14] Massanari RM. Purulent meningitis in the elderly: when to suspect an unusual
pathogen. Geriatrics 1977;32(March (3)):559.
[15] Ray P, Badarou-Acossi G, Viallon A, Boutoille D, Arthaud M, Trystram D, et al.
Accuracy of the cerebrospinal uid results to differentiate bacterial from non
bacterial meningitis, in case of negative gram-stained smear. Am J Emerg Med
2007;25(February (2)):17984.
[16] Sigurdardottir B, Bjornsson OM, Jonsdottir KE, Erlendsdottir H, Gudmundsson
S. Acute bacterial meningitis in adults. A 20-year overview. Arch Intern Med
1997;157(February 24 (4)):42530.
[17] Rasmussen HH, Sorensen HT, Moller-Petersen J, Mortensen FV, Nielsen B. Bacterial meningitis in elderly patients: clinical picture and course. Age Ageing
1992;21(May (3)):21620.
[18] Attia J, Hatala R, Cook DJ, Wong JG. The rational clinical examination.
Does this adult patient have acute meningitis? JAMA 1999;282(July 14 (2)):
17581.
[19] Puxty JA, Fox RA, Horan MA. The frequency of physical signs usually attributed
to meningeal irritation in elderly patients. J Am Geriatr Soc 1983;31(October
(10)):5902.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.