Giant Cell Arteritis:

Validity and Reliability

of Various Diagnostic Criteria
SOHAN SINGH HAYREH, MD, PHD, DSC, PATRICIA A. PODHAjSKY, BSN,
REMA RAMAN, MSc, AND BRIDGET ZIMMERMAN, PHD

PURPOSE: To ascertain the validity, reliability,

sensitivity, and specificity of various signs and
symptoms of and diagnostic tests for early diagno
sis of giant cell arteritis.
METHODS: From 1973 to 1994, we studied 363
patients who had temporal artery biopsy for sus
pected giant cell arteritis. All patients underwent
detailed clinical evaluation and had erythrocyte
sedimentation rates determined; since 1985, 223
patients had their C-reactive protein values esti
mated. Erythrocyte sedimentation rate and Creactive protein levels were also estimated in 749
and 138 control subjects, respectively. Signs and
symptoms of giant cell arteritis, erythrocyte sedi
mentation rate, and C-reactive protein levels
among patients with positive and negative biopsies
were compared.
RESULTS: Of the 363 patients, temporal artery
biopsy was positive in 106 and negative in 257.
The odds of a positive biopsy were 9.0 times
greater with jaw claudication (P < .0001), 3.4
times greater with neck pain (P = .0085), 2.0
times greater with an erythrocyte sedimentation
rate of 47 to 107 mm/hour (P = .0454), 3.2 times
greater with C-reactive protein above 2.45 mg/dl
Accepted for publication Aug 6, 1996.
From the Departments of Ophthalmology (Dr Hayreh and Ms Podhajsky) and Preventive Medicine and Environmental Health (Division of
Biostatistics) (Ms Raman and Dr Zimmerman), College of Medicine,
University of Iowa, Iowa City, Iowa. Supported by grant EY-1151 and
RR-59 from the National Institutes of Health and in part by unrestricted
grants from Research to Prevent Blindness, Inc, New York, New York. Dr
Hayreh is a Research to Prevent Blindness Senior Scientific Investigator.
Reprint requests to Sohan Singh Hayreh, MD, PhD, DSc, Department
of Ophthalmology, University Hospitals and Clinics, 200 Hawkins Dr,
Iowa City, IA 52242-1091; fax: (319) 353-7996; e-mail: sohan-hayreh@
uiowa.edu

VOL.123, No. 3

(P = .0208), and 2.0 times greater for age 75 years

or more (P = .0105).
CONCLUSIONS: Clinical criteria most strongly
suggestive of giant cell arteritis include jaw claudi
cation, C-reactive protein above 2.45 mg/dl, neck
pain, and an erythrocyte sedimentation rate of 47
mm/hour or more, in that order. C-reactive protein
was more sensitive (100%) than erythrocyte sedi
mentation rate (92%) for detection of giant cell
arteritis; erythrocyte sedimentation rate combined
with C-reactive protein gave the best specificity
(97%).

IANT CELL ARTERITIS (ALSO CALLED TEMPORAL

arteritis or cranial arteritis) is an ophthalmic

emergency because it presents a high risk of
severe visual loss.1 Kearns2 rightly stressed that giant
cell arteritis "ranks as the prime medical emergency
in ophthalmology, there being no other disease in
which the prevention of blindness depends so much
on prompt recognition and early treatment." Because
giant cell arteritis is a well-known masquerader and
its early diagnosis is the key to correct management of
patients and prevention of its complications, it is
essential to ascertain the validity, reliability, sensitivi
ty, and specificity of the signs and symptoms of and
diagnostic tests for giant cell arteritis.

See also pp. 392-395.

The literature is contradictory. For example, in a

review of the literature, Keltner3 pointed out that, in

AMERICAN JOURNAL OF OPHTHALMOLOGY 1997;123:285-296

285

THE DATA FOR THIS PROSPECTIVE STUDY CAME FROM 363

range, 20 to 95 years) seen in different departments of

the University of Iowa Hospitals and Clinics and
referred to the ophthalmology department for tem
poral artery biopsy from 1973 to 1994. Because all
temporal artery biopsies for the entire medical center
are done in our department, our sample represents
not only patients first seen for ophthalmic disorders
but also those of a much wider spectrum. Some of
these patients came first to our department with
suspicious unilateral or bilateral ocular lesions and
associated visual disturbance, but many were from
other departments (mainly the rheumatology, inter
nal medicine, and neurology clinics), where they had
been found to have systemic signs and symptoms
suggestive of giant cell arteritis.
The decision to perform temporal artery biopsy was
made either by ophthalmologists in our department
or by the referring physician. At biopsy, a segment of
at least 1 inch of the temporal artery was removed,
processed, and examined. Complete serial sectioning
of the biopsied artery was done to make sure that
"skip areas" of involvement by the giant cell arteritis
did not result in false-negative biopsies. Criteria for a
positive temporal biopsy included fibrinous necrosis of
media, focal or generalized destruction of internal
elastic lamina, a predominance of mononuclear cell
infiltration, or a granulomatous process with multinucleated giant cells. Histopathologists generally agree
that the absence of giant cells does not preclude the
histologic criteria of giant cell arteritis. W h e n the
temporal artery biopsy results on one side were
equivocal or negative for giant cell arteritis but there
was a very strong index of suspicion for giant cell
arteritis, the biopsy was repeated on the other side (of
76 patients who underwent second biopsy, the second
biopsy was positive in seven). The index of suspicion
was based on various clinical findings, including signs
and symptoms strongly suggestive of giant cell arteri
tis,1 arteritic anterior ischemic optic neuropathy (as
denned in detail by one of us [S.S.H.] elsewhere4), or
both. The erythrocyte sedimentation rate or Creactive protein values by themselves did not influ
ence this decision. T h e remaining 257 patients with
negative temporal artery biopsy on follow-up showed
no stigmata of giant cell arteritis at all, indicating that
a second temporal artery biopsy in them would not
have been positive.

patients (123 men, 240 women; mean age, 72 years;

All patients underwent a clinical evaluation in our

these patients, the reported incidence of headache

varied from 4% to 100%; of tender temporal artery,
from 28% to 9 1 % ; of myalgia and arthralgia, from
28% to 86%; of fever, from 30% to 100%; of weight
loss, from 16% to 76%; of jaw claudication, from 4%
to 67%; of anorexia, from 14% to 69%; of malaise,
from 12% to 97%; and of leg claudication, from 2% to
43%. Variable and sometimes conflicting information
in different studies is partly caused by selection bias in
patients (because of the different specialties of the
authors, for example, internal medicine, rheumatolo
gy, neurology, or ophthalmology), different diagnostic
criteria used to define giant cell arteritis (for example,
diagnoses based strictly on the presence of a positive
temporal artery biopsy or only on clinical manifesta
tions), and variations in demographics and sample
size of the patient population in the studies.
To rationalize this conflicting information, we
have, since 1973, conducted a prospective study of
giant cell arteritis in a patient population in our
medical center to obtain information on various
diagnostic criteria that may help in early and reliable
detection of giant cell arteritis. The first objective of
the present study was to establish the sensitivity and
specificity of the erythrocyte sedimentation rate and
C-reactive protein level in the diagnosis of giant cell
arteritis. Diagnostic cutoffs for the erythrocyte sedi
mentation rate and C-reactive protein were identified
from a control group at the Ocular Vascular Clinic
(University of Iowa Hospitals and Clinics, Iowa City,
Iowa) who had their erythrocyte sedimentation rates
and C-reactive protein levels evaluated as a part of a
routine systemic evaluation for other ocular studies
and who had no ocular or systemic reasons that would
cause an elevated erythrocyte sedimentation rate or
C-reactive protein level. The patients in the giant cell
arteritis group had positive temporal artery biopsies
for giant cell arteritis. The second objective was to
determine, among those with suspected giant cell
arteritis, the signs and symptoms most associated with
giant cell arteritis as confirmed by positive temporal
artery biopsy.

PATIENTS AND METHODS

286

AMERICAN JOURNAL OF OPHTHALMOLOGY

MARCH 1997

department and had an erythrocyte sedimentation

rate estimation (Westergren's method) done routinely
before the biopsy. At the initial visit of all patients,
detailed information on signs and symptoms of giant
cell arteritis were collected. Beginning about 1985,
when estimation of C-reactive protein became avail
able to us, we also routinely included this measure
ment for each patient in the study (223 [81 men, 142
women] of 363 patients). All studies had the approval
of the institutional review committee.
To serve as a control population for the erythrocyte
sedimentation rate and C-reactive protein level, we
obtained data on these two parameters from patients
in our prospective studies on systemic diseases associ
ated with retinal vein occlusion and nonarteritic
anterior ischemic optic neuropathy conducted from
1974 to 1990. The control group comprised 749
patients aged 15.9 to 99.7 years. We excluded all
patients with any systemic or other abnormality,
including diabetes mellitus, that might possibly ele
vate their erythrocyte sedimentation rates or Creactive protein values. Our detailed studies in more
than 1,200 patients with retinal vein occlusion and
more than 1,000 patients with nonarteritic anterior
ischemic optic neuropathy have disclosed that once
patients with various systemic and hematologic dis
eases are excluded, the two ocular diseases per se have
no effect on erythrocyte sedimentation rate and
C-reactive protein level; the values of the erythrocyte
sedimentation rate and C-reactive protein in other
wise healthy patients with nonarteritic anterior ische
mic optic neuropathy or retinal vein occlusion are
consistently normal for their respective age and
gender groups, as confirmed by several other studies5,6
based on much larger samples of normal healthy
persons. In our studies, we have found that both
retinal vein occlusion and nonarteritic anterior ische
mic optic neuropathy are also far more common in
the middle-aged and elderly than the young.7,8
For statistical analysis, the patients were divided
into three groups: patients with positive temporal
artery biopsy for giant cell arteritis (n = 106), patients
with negative biopsy (n = 257), and the control
group (n = 749).
Sensitivity and specificity of the erythrocyte sedi
mentation rate (Westergren's method) and C-reactive
protein value for diagnosis of giant cell arteritis were
evaluated using the erythrocyte sedimentation rate
VOL.123,

No. 3

and C-reactive protein data of the group with positive

temporal artery biopsy and the control group. We did
not use the patients with negative temporal artery
biopsy in the evaluation of specificity because one of
the reasons many of these patients were biopsied was
because of elevated erythrocyte sedimentation rate
and C-reactive protein; their inclusion would have
caused our sample distribution of erythrocyte sedi
mentation rates and C-reactive protein values for
individuals with no giant cell arteritis to have a higher
proportion of elevated erythrocyte sedimentation
rates and C-reactive protein values than that found in
the general population of persons with no giant cell
arteritis, and that would have biased the sample for
this parameter.
Sensitivity and specificity were estimated for the
following alternative cutoff criteria for the erythrocyte
sedimentation rate (Table 1). All the erythrocyte
sedimentation rates in our study were done by the
pathology laboratory of the University of Iowa Hospi
tals and Clinics. Their recommended normal values
are 0 to 15 mm/hour in men and 0 to 20 mm/hour in
women, which agrees with the recommendations of
most other pathology laboratories. Another com
monly advocated formula is the one derived empiri
cally by Miller and associates5 based on their study of
27,912 healthy adults aged 20 to 65 years. The value
of the erythrocyte sedimentation rate in 98% of their
subjects was as follows: for men, the age in years
divided by 2; for women, the age in years plus 10
divided by 2. A cutoff criterion as a function of age
and gender was derived from the 95% upper confi
dence limit for a single observation from the regres
sion of the erythrocyte sedimentation rate with ages
and genders of our control group (Figure 1, Table 1).
Empirical distribution of the observed erythrocyte
sedimentation rate values of the control group was
based on the 95 th percentile as the cutoff point of the
values in the control group (Table 1).
To evaluate the sensitivity and specificity of Creactive protein for diagnosis of giant cell arteritis, we
used less than 0.5 mg/dl as the cutoff, which is the
normal level for C-reactive protein as currently de
fined by the pathology laboratory of the University of
Iowa Hospitals and Clinics (Table 1). For patients in
whom both the erythrocyte sedimentation rate and
C-reactive protein level were obtained, sensitivity and
specificity were evaluated using both values (Table 2).

GIANT CELL ARTERITIS

287

Table 1. Sensitivity, Specificity, and Positive and Negative Predictive Value of ESR and CRP Using Various Cutoff Criteria

> Cutoff
Criterion

For ESR
Used by pathology laboratory at UIHC
Males, 10 mm/hr; females, 20 mm/hr
Suggested by Miller and associates5
Males, age/2; females, (age + 10)/2
From regression of age on ESR
of our control group
Males, 17.3 + 0.18(age);
females, 22.1 + 0.18(age)
From empiric distribution of ESR
of our control group
Males, 33 mm/hr; females, 35 mm/hr
For CRP
Used by pathology laboratory at UIHC
0.5 mg/dl

Specificity

Sensitivity

s Cutoff

Negative

Positive
Predictive
Value

No. of Subjects

Value
95% Cl

95% Cl

95% Cl

95% Cl

97

92-99

67

64-72

29

25-34

99

98-100

703

86

78-92

94

92-96

66

58-74

98

97-99

692

92

86-97

92

90-94

63

55-71

99

98-100

57

692

92

86-97

92

90-94

63

55-71

99

98-100

25

113

100

93-100

82

74-88

63

51-75

100

97-100

GCA

Control

GCA

Control

103

247

502

91

46

15

98

57

98

43

Cl = confidence interval, CRP = C-reactive protein, ESR = erythrocyte sedimentation rate, GCA = giant cell arteritis, UIHC =
University of Iowa Hospitals and Clinics.
160

160

MALES

140
120-

oo

100

0C 8 0
CO

o o

lii

60
40

tt

CO
ill

>..":

,.-.

..

...

i>. *?. Cs? .

0%

80

*
o

60

40

120

o
o

100-

20-

FEMALES

140

o.

.--.-.-.-f r-^.-;:.

20
010

20

30

AGE

40

50

60

70

80

90

100

AGE

Figure 1. Graphs showing regression analysis of erythrocyte sedimentation rate (ESR) (in mm/hour) with age (in years)
and gender in our patients with positive temporal artery biopsy for giant cell arteritis (open circles) and in patients in the
control group (dots). The dotted line represents the 95% upper confidence limit for the control group. In places, empty
circles for various patients overlap each other, giving the erroneous impression that the number of open circles
corresponding to patients with giant cell arteritis is too low.

In our analysis of factors associated with positive

temporal artery biopsy for giant cell arteritis, the
association of age, erythrocyte sedimentation rate,
C-reactive protein level, and various systemic signs
288

and symptoms with positive biopsy was examined in

all patients suspected of having giant cell arteritis
(Tables 3 and 4). Stepwise logistic regression analysis
was performed to study the association of these

AMERICAN JOURNAL OF OPHTHALMOLOGY

MARCH

1997

Table 2. Sensitivity, Specificity, and Positive and Negative Predictive Value Based on Patients With Both ESR and CRP
Measured Using Cutoff Criteria for Both ESR and CRP
No. of Subjects
ESR and CRP
Above Cutoff
Criterion

Used by pathology laboratory at UIHC

Males, 10 mm/hr; females, 20 mm/hr
Suggested by Miller and associates5
Males, age/2; females, (age + 10)/2
From regression of age on ESR
of our control group
Males, 17.3 + 0.18(age);
females, 22.1 + 0.18(age)
From empiric distribution of ESR
of our control group
Males, 33 mm/hr; females, 35 mm/hr

GCA

Control

42

ESR or CRP
Below Cutoff

Sensitivity

Negative
Predictive

Predictive
Value

Specificity

Value

95% Cl

95% Cl

86-96

79

66-89

99

96-100

98

94-100

93

80-98

96

92-99

80-92

97

93-99

90

78-97

98

94-100

78-97

98

94-100

93

81-94

97

93-99

GCA

Control

95% Cl

11

127

98

88-100

92

38

135

88

75-96

40

134

93

39

135

91

95% Cl

Cl = confidence interval, CRP = C-reactive protein, ESR = erythrocyte sedimentation rate, GCA = giant cell arteritis, UIHC =
University of Iowa Hospitals and Clinics.

variables with positive biopsy in a multivariate model.

The criterion for entry into the model was P < .15
and, for removal, P > .15.
Because C-reactive protein values were not avail
able for all patients, we first performed the logistic
regression for all patients with erythrocyte sedimenta
tion rate data available. A second logistic regression
model was fitted that included only those subjects
with both erythrocyte sedimentation rate and Creactive protein values present. Age, erythrocyte sedi
mentation rate, and C-reactive protein level were first
included in the logistic regression as continuous
variables. The assumption of a linear relationship
between the logit and these continuous variables was
checked using the method described by Hosmer and
Lemeshow.9 For age, erythrocyte sedimentation rate,
and C-reactive protein level, this linear relationship
was not observed; therefore, these variables were not
used as continuous variables in the final model. The
quartiles of the distribution of these variables were
used initially to define the interval groups that were
included in the logistic regression model. This model
indicated that some of these intervals could be
combined. This resulted in the final logistic regression
with age defined as a dichotomous variable below or
above the median age of 75 years; with the erythro
cyte sedimentation rate divided into below 47 mm/
VOL.123, No. 3

Table 3. Incidence of Systemic Signs and Symptoms

in Patients With Positive and Negative Temporal
Artery Biopsy for Giant Cell Arteritis
Temporal Artery Biopsy
(No. [%])
Positive
Signs and Symptoms

(n = 106)

Negative
(n = 257)

P Value

Headache
Anorexia/weight loss
Jaw claudication
Malaise
Myalgia
Fever
Abnormal temporal artery
Scalp tenderness
Neck pain
Anemia

59 (55.7)
55 (51.9)
51 (48.1)
40 (37.7)
31 (29.2)
28 (26.4)
21 (19.8)
19(17.9)
17(16.0)
14(13.2)

117(45.5)
84 (32.7)
22 (8.6)
78 (30.4)
68 (26.5)
42(16.3)
33(12.8)
27(10.5)
11 (4.3)
31 (12.1)

.084
.0005
<.0001
.177
.606
.040
.105
.058
.0003
.730

hour, from 47 to 107 mm/hour, and above 107

mm/hour; and the C-reactive protein level categorized
as 2.45 mg or less or as greater than 2.45 mg.

RESULTS
FINDINGS FOR AGE, GENDER, ERYTHROCYTE SEDIMENTA-

tion rate, and C-reactive protein level in patients with

positive temporal artery biopsy, in patients with

GIANT CELL ARTERITIS

289

Table 4. Clinical Variables That Showed an

Association With Increased Odds of a Positive Biopsy
for Giant Cell Arteritis

Variable

Using sample with ESR data*

Jaw claudication
Neck pain
ESR 47-107 mm/hrt
ESR >107 mm/hrt
Age a 75 yrs
Using sample with both ESR
and CRP data*
Jaw claudication
Neck pain
ESR 47-107 mm/hrt
ESR >107 mm/hrt
Age a 75 yrs
CRP >2.45 mg/dl!

Odds
Ratio

95% Cl
for Odds Ratio

P Value

9.1
3.4
2.0
2.7
2.0

5.0-16.7
1.4-8.3
1.0-4.1
1.2-5.9
1.2-3.4

<.0001
.0085
.0454
.0106
.0105

7.5
4.5
1.4
2.3
2,3
3.2

3.1-18.0
1.2-17.2
0.5-4.5
0.6-8.4
1.0-5.3
1.2-8.6

<.0001
.0263
.5326
.2085
.0412
.0208

*N = 363: 106 with positive biopsy and 257 with negative

biopsy.
tCompared with patients with ESR <47 mm/hr.
*N = 223: 43 with positive biopsy and 180 with negative
biopsy.
Compared with patients with CRP =s2.45 mg/dl.
Cl = confidence interval, CRP = C-reactive protein, ESR =
erythrocyte sedimentation rate.

negative biopsy, and in control subjects are given in

Table 5. Figure 1 shows the relation of the erythrocyte
sedimentation rate with age for men and women in
patients with giant cell arteritis and in the control
subjects. In Figure 2, the relation of the erythrocyte
sedimentation rate and C-reactive protein level is
presented as a scatterplot for these two groups and
shows that in the giant cell arteritis patients, a high
C-reactive protein value is associated with a high
erythrocyte sedimentation rate. Control subjects with
high erythrocyte sedimentation rates did not neces
sarily have high C-reactive protein values. Our
findings of the sensitivity and specificity of the
erythrocyte sedimentation rate and C-reactive pro
tein value in the detection of giant cell arteritis (that
is, positive temporal artery biopsy), using the differ
ent cutoff criteria defined above, are given in Tables 1
and 2.
Table 3 gives the incidence of various systemic
signs and symptoms in the 106 patients with positive
and the 257 patients with negative temporal artery
biopsies. A n occasional patient had other symptoms,
290

for example, tongue claudication and tenderness over

the facial artery.
The variables that showed an association with
increased odds for a positive temporal artery biopsy,
based on logistic regression analysis, are given in
Table 4. We performed the logistic regression analysis
using two samples: all patients with erythrocyte
sedimentation rate data available (363 patients) and
patients with both erythrocyte sedimentation rate and
C-reactive protein values available (223 patients). In
the sample with only erythrocyte sedimentation rate
data, there was a significant association of jaw claudi
cation, neck pain (mostly in the occipital or back part
of the neck), erythrocyte sedimentation rate of 47
mm/hour or more, and age of 75 years or more with
increased odds of a positive biopsy for giant cell
arteritis. When a sample with both the erythrocyte
sedimentation rate and C-reactive protein values was
used, as in the previous analysis, the presence of jaw
claudication, neck pain, and age of 75 years or more
were significantly associated with increased odds of a
positive biopsy for giant cell arteritis. With both the
C-reactive protein level and erythrocyte sedimenta
tion rate included in the model, only the C-reactive
protein value showed a significant association with
positive biopsy. Patients with C-reactive protein above
the observed median level (that is, a C-reactive
protein level of 2.45 mg/dl for all patients who had
undergone temporal artery biopsy) were 3.2 times
more likely to have a positive temporal artery biopsy
(95% confidence interval [Cl], 1.2 to 8.6; P = .0208).
The effect of the erythrocyte sedimentation rate was
not significant (P = .5326 for an erythrocyte sedi
mentation rate of 47 to 107 mm/hour; P = .2085 for
an erythrocyte sedimentation rate greater than 107
mm/hour). The two fitted models yielded similar
results for neck pain, jaw claudication, and age of 75
years of more, with overlapping 95% CIs for the odds
ratio. The second model, which was based on a
smaller sample size, had wider 95% CIs compared
with the first model, which included all patients with
an erythrocyte sedimentation rate.

DISCUSSION
THE ESSENTIAL REQUIREMENT TO PREVENT BLINDNESS

in patients with giant cell arteritis is early, correct

AMERICAN JOURNAL OF OPHTHALMOLOGY

MARCH

1997

Table 5. Descriptive Statistics

Temporal Artery Biopsy
Positive
Variable

Age (yrs)
Range
Median
Mean SD
Gender (no. [%])
Male
Female
Erythrocyte sedimentation
rate (Westergren) (mm/hr)
Range
Median
Mean SD
C-reactive protein (mg/dl)
n
Range
Median
Mean SD

140-

CONTROL
GCA

56-93.4
75.8
75 7.9

0
o

o
o

;
o

"l....!...!.
*i

:
0c0.05

. i

20* 1
fi

o o

UJ

o
o

o 0o
0

40-

o
o

60-

t 1 m i1

0.1

o
r

I I I I I I |

10

100

CRP

Figure 2. Scatterplot showing the relationship of the

erythrocyte sedimentation rate (ESR) (in mm/hour) and
C-reactive protein level (CRP) (in mg/dl) in patients
with positive temporal artery biopsy for giant cell arteritis (GCA) (open circles) and in patients in the control
group (dots).

diagnosis. The purpose of our study was to establish

the relative degree of validity, reliability, sensitivity,
and specificity of the various signs and symptoms of
V o t . 123, N o . 3

2-155
68
72 39

43
0.5-34.7
4.35
6.6 6.7

80-

94 (36.6)
163(63.4)

4-140
87.5
84.9 33.4

:
:

100-

20.2-95.4
72
71.3 9.5

29 (27.4)
77 (72.6)

i
120-

Negative
(n = 257)

(n = 106)

180
<0.5-24.6
2.05
3.3 4.3

Control Group
(n = 749)

15.9-99.7
63
61.6 14.3
441 (58.9)
308(41.1)

1-59
11
14 10.7
138
<0.5-3.3
<0.5
<0.5 <0.5

and diagnostic tests for giant cell arteritis. It was

therefore imperative that the patients from whom
data were drawn must meet the definitive criterion for
diagnosis of giant cell arteritis (positive temporal
artery biopsy). Using that criterion, the main findings
in our 106 patients with giant cell arteritis are given
in Tables 3 and 5.
In 1990, the American College of Rheumatology10
defined criteria to differentiate giant cell arteritis from
other types of vasculitis, based on a comparison of
214 cases diagnosed as giant cell arteritis and 593 as
other types of vasculitis. Of the 214 giant cell arteritis
patients, 18 had negative temporal artery biopsy, and
the diagnosis was based entirely on other criteria. The
American College of Rheumatology selected five
criteria to diagnose giant cell arteritis: age of 50 years
or more at disease onset, new onset of localized
headache, temporal artery tenderness or decreased
pulse, an elevated erythrocyte sedimentation rate
(Westergren's method) of 50 mm/hour or more, and
temporal artery biopsy showing necrotizing arteritis
characterized by a predominance of mononuclear cell
infiltrates or a granulomatous process with multinucleated giant cells. In that study,10 the presence of
three or more of these criteria was associated with a
sensitivity of 93.5% and a specificity of 91.2% for
giant cell arteritis. The authors used new headache

GIANT CELL ARTERITIS

291

and scalp tenderness or nodules as a surrogate for

temporal artery biopsy in those without positive
temporal artery biopsy.
Chmelewski and associates11 retrospectively com
pared features at initial examination of 30 patients
with positive temporal artery biopsy and 68 with
negative temporal artery biopsy. In their study, the
features of temporal artery biopsy-positive patients at
first examination were headache in 93%, polymyalgia
rheumatica in 57%, jaw claudication in 50%, fever in
37%, scalp abnormality in 47%, and neurologic
symptoms in 17%. Patients with positive temporal
artery biopsy had significantly (P = .003) increased
headache (93% vs 62%) and jaw claudication (50%
vs 18%) compared with patients with negative tem
poral artery biopsy. Polymyalgia rheumatica symptoms
were present in similar frequency in the two groups.
To differentiate the temporal artery biopsy-positive
from biopsy-negative cases, headache had low speci
ficity (40%), jaw claudication had more specificity
(56%), and polymyalgia rheumatica had low sensitivi
ty (23%) but higher specificity (78%).

positive biopsy. The signs and symptoms that showed

significant association were jaw claudication (P <
.0001), neck pain (P = .0003), anorexia or weight
loss (P = .0005), fever (P = .0400), and scalp
tenderness (P = .058). However, when we considered
a multivariate logistic regression model that included
erythrocyte sedimentation rate and age, the only
significant systemic symptoms were jaw claudication
and neck pain (mostly in the occipital and back parts
of the neck). This implies that jaw claudication and
neck pain were indicators of a positive biopsy, inde
pendent of erythrocyte sedimentation rate and age,
and that they were more highly correlated to the
outcome of a positive biopsy than were anorexia or
weight loss, fever, and scalp tenderness. Other varia
bles may also be helpful in the diagnosis of giant cell
arteritis but not to the same extent as these parame
ters. The conclusions of previous studies were mainly
based on univariate analysis.

Vilaseca and associates12 suggested that simultane

ous jaw claudication, abnormal temporal arteries on
examination, and headache of recent onset had a
specificity of 94-8% for positive temporal artery biopsy
and concluded that the presence of these three
clinical features together points to a definite diagnosis
of giant cell arteritis. Fernandez-Herlihy,13 in a com
parison of 29 temporal artery biopsy-positive with 68
biopsy-negative patients, found in biopsy-positive
cases significantly more frequent jaw claudication (P
< .0005), anorexia (P < .001), and anemia (P <
.01), as well as an erythrocyte sedimentation rate of
45 to 148 mm/hour (average, 102 mm/hour; P <
.001). He reported that a combination of headache,
jaw claudication, and abnormal temporal artery had a
high specificity (100%) but low sensitivity (17%) for
predicting a positive temporal artery biopsy for giant
cell arteritis.
Our data differ in some respects from those of the
American College of Rheumatology10 and many
other studies; unlike those authors, we included in
our giant cell arteritis group only patients who had a
positive temporal artery biopsy; our choice was not
based on other criteria. Table 3 provides the results of
a univariate analysis of the associations in our data
between various systemic signs and symptoms with

Our study showed that the odds of having a

positive temporal artery biopsy were 9 times greater
with jaw claudication, 3.3 times greater with neck
pain, 2.1 times greater with an erythrocyte sedimenta
tion rate of 47 to 107 mm/hour, 2.7 times greater with
an erythrocyte sedimentation rate greater than 107
mm/hour relative to an erythrocyte sedimentation
rate less than 47 mm/hour, and 2.0 times greater
when the patients were aged 75 years or older
compared with age below 75 years (Table 4). The
odds of having a positive temporal artery biopsy were
3.2 times greater with a C-reactive protein level above
2.45 mg/dl compared with a C-reactive protein level
of 2.45 mg/dl or less.
It is generally agreed that giant cell arteritis is a
disease of the elderly. The American College of
Rheumatology 10 defines it as a disease of persons aged
50 years or older. In our study, the youngest patient
was 56 years old (median, 75.8 years; mean, 75 7.9
years). However, in numerous reports, infants, juve
niles, and adults under 50 years are claimed to have
giant cell arteritis.14'22 Wagenvoort and associates22
identified their patients, all younger than 50 years of
age, as having "infantile giant cell arteritis." A critical
review of these patients shows that they do not meet
all the criteria of the classic giant cell arteritis
syndrome. Confusion is also caused by the entity
called "juvenile temporal arteritis," which affects the
branches of the external carotid artery, principally the

292

OPHTHALMOLOGY

AMERICAN JOURNAL

MARCH 1997

superficial temporal artery, and which is a localized

and benign condition. 23 Because of such reports,
some of our referring physicians often insist that we
performed a temporal artery biopsy in persons young
er than 50 years of age; of the four young (less than 56
years old) patients in this study who had a temporal
artery biopsy, none had a positive biopsy. Features of a
number of systemic vasculitides can mimic and over
lap giant cell arteritis24 and result in confusion and
misdiagnosis.
In giant cell arteritis, it is well established that the
erythrocyte sedimentation rate is elevated, and its
estimation is an important test in diagnosis. There is,
however, a good deal of confusion about the level of
the erythrocyte sedimentation rate that may be con
sidered as a diagnostic indicator for giant cell arteritis
because the normal value of the erythrocyte sedimen
tation rate (Westergren's method) in those aged 50
years or more is still debatable. There are many
reported normal values and several proposed formulae
for calculating normal values. All reports agree that
the erythrocyte sedimentation rate increases with age
and that women have a higher erythrocyte sedimen
tation rate than men do. For example, Sparrow and
associates25 determined the erythrocyte sedimentation
rate in 1,480 healthy men and reported a significant
(P < .001) correlation between increasing age and
higher erythrocyte sedimentation rate values. Based
on the erythrocyte sedimentation rate in 4,341
healthy normal persons (2,140 men, 2,201 women),
Gilbertsen 6 concluded that the range of a normal
erythrocyte sedimentation rate that includes 90% of
persons aged 45 to 79 years was 0 to 32 mm/hour in
men and 1 to 48 mm/hour in women. Bottiger and
Svedberg,26 in 2,500 subjects aged more than 50 years,
found the upper limits of "normal" to be 20 mm/hour
in men and 30 mm/hour in women. Miller and
associates,5 based on a study of 27,912 individuals
aged 20 to 65 years, derived an empiric formula for an
erythrocyte sedimentation rate that included 98% of
healthy persons: age in years divided by 2 for men,
and age in years plus 10 divided by 2 for women.
However, Hanger and associates,27 based on a survey
of 194 elderly subjects, advocated that in the formula
of Miller and associates,5 3 be used instead of 2 as the
dividing factor. There are numerous other conflicting
reports based on smaller series.
Although a high erythrocyte sedimentation rate
VOL.123,

No. 3

traditionally is emphasized as a sine qua non for the

diagnosis of giant cell arteritis, there are numerous
anecdotal case reports of "low" erythrocyte sedimen
tation rates in patients with positive temporal artery
biopsies, between 5 and 20 mm/hour1,18'28'34 or be
tween 22 and 30 mm/hour.30,33'35 The association of
"normal" or "low" erythrocyte sedimentation rates in
giant cell arteritis with positive biopsy has been
variously reported as being 5% to 30%. 36
In our study, the erythrocyte sedimentation rate in
patients with positive temporal artery biopsy varied
between 4 and 140 mm/hour (median, 87.5 mm/
hour; mean, 85 33 mm/hour; Westergren's meth
od) (Table 5). Table 1 gives the sensitivity and
specificity of an erythrocyte sedimentation rate using
the various advocated cutoff criteria as well as that
denned by regression of age on the erythrocyte
sedimentation rate in our data (Figure 1). Our data
suggest a cutoff criterion of 33 mm/hour or less in men
and of 35 mm/hour or less in women is normal and
has a sensitivity of 92% and specificity of 92%, This
meant that 8% of our patients with giant cell arteritis
had an erythrocyte sedimentation rate within normal
levels, the lowest being 4 mm/hour.
The erythrocyte sedimentation rate level can be
affected by a variety of conditions, including the
degree of rouleaux formation by the blood cells, and
the amount of fibrinogen, immunoglobulins, other
plasma proteins, lipids, and even drugs. It can also be
affected by the length of time between the venipuncture and testing of the erythrocyte sedimentation rate
in the laboratory. (In our study, this test was always
started within a few minutes.) The rate is increased in
anemia, inflammatory disease, malignancy, infection,
connective tissue diseases, pregnancy, hypercholesterolemia, and a host of other systemic conditions. A
very low erythrocyte sedimentation rate is seen in a
number of hematologic disorders (including polycythemia, hypochromic microcystic anemia, hereditary
spherocytosis, hemoglobinopathy, and hypofibrinogenemia), congestive heart failure, and treatment
with anti-inflammatory drugs. These factors make the
erythrocyte sedimentation rate a nonspecific indica
tor of disease. In our patients with giant cell arteritis,
no patient had an elevated erythrocyte sedimentation
rate that could be attributed to any disease other than
giant cell arteritis.
C-reactive protein, an acute-phase plasma protein

GIANT CELL ARTERITIS

293

of hepatic origin, has been shown to be elevated in

giant cell arteritis.37'40 The C-reactive protein level
rises with tissue necrosis, infectious diseases, inflam
matory diseases, surgical tissue injury, myocardial
infarction, rheumatic disease, transplantation, and a
host of other conditions. Concentration of C-reactive
protein can reach abnormal levels within 4 to 6 hours
after the insult and can increase as much as 1,000fold. Similarly, after the insult is removed, its concen
tration can drop rapidly. Unlike the erythrocyte
sedimentation rate, the C-reactive protein level is not
influenced by various hematologic factors or age; this
makes the C-reactive protein level a more sensitive
diagnostic test than the erythrocyte sedimentation
rate. According to Deodhar, 41 the C-reactive protein
test is highly sensitive, reproducible, quantitative, and
easy and rapid to perform (we get the results within 1
hour of drawing the blood), and C-reactive protein
testing is superior to the erythrocyte sedimentation
rate on clinical, scientific, and practical grounds. We
have also found it reliable for monitoring disease
activity. However, the C-reactive protein level, like
the erythrocyte sedimentation rate, is a nonspecific
test.
C-reactive protein levels in our patients with giant
cell arteritis varied between 0.5 and 34-7 mg/dl
(median, 4-35 mg/dl; mean SD, 6.6 6.7 mg/dl);
Figure 2 shows the relation of the erythrocyte sedi
mentation rate and C-reactive protein level in pa
tients with giant cell arteritis and control subjects. In
patients with giant cell arteritis, a high C-reactive
protein value is associated with a high erythrocyte
sedimentation rate; control subjects with high eryth
rocyte sedimentation rate values, however, do not
necessarily have a high C-reactive protein level. The
sensitivity and specificity of C-reactive protein in
detecting giant cell arteritis in men were 100% and
83% and in women 100% and 79%, respectively. O n
logistic regression analysis of the erythrocyte sedi
mentation rate and C-reactive protein level with
positive temporal artery biopsy, only C-reactive pro
tein showed a significant association with positive
biopsy (Table 4). Thus, we have found C-reactive
protein to be a highly useful test for diagnosing and
monitoring the activity of giant cell arteritis.
Some have argued that the erythrocyte sedimenta
tion rate and C-reactive protein value should not be
used as independent variables because they are mea294

suring the same phenomenon (acute-phase response)

and that it may be more appropriate to assay interleukin-6, the cytokine that directly stimulates C-reactive
protein synthesis and other protein production lead
ing to an elevated erythrocyte sedimentation rate. We
recently conducted a small pilot study in 23 patients
with giant cell arteritis on treatment to find out the
relation of interleukin-6 with the erythrocyte sedi
mentation rate and C-reactive protein level. We found
no significant linear relationship between the eryth
rocyte sedimentation rate and interleukin-6 (r =
.004, P = .982); however, there was a linear relation
between the C-reactive protein level and interleu
kin-6 (r = .533, P = .028). This suggests that
interleukin-6 may not provide any better information
than C-reactive protein does. Moreover, C-reactive
protein, unlike interleukin-6, is a quick, cheap, and
readily available test.
Other hematologic tests, including serum amyloid
A apolipoprotein, anticardiolipin antibodies, von
Willebrand's factor antigen, plasma viscosity,
antineutrophil cytoplasm antibodies, and HLA-DR,
HLA-DQ, and HLA-DP, have occasionally been
advocated for the diagnosis of giant cell arteritis.
However, none of them has so far been shown to be as
readily available and reliable as the erythrocyte sedi
mentation rate and C-reactive protein level.
Although temporal artery biopsy is considered the
definitive criterion for diagnosis of giant cell arteritis,
false-negative biopsies have been reported and attrib
uted to "skip areas" of arteritis in temporal arteries.
Hedges and associates42 found 5% false-negative tem
poral artery biopsies for giant cell arteritis; however,
they also found that age and erythrocyte sedimenta
tion rate alone or in combination with other signs or
symptoms of giant cell arteritis, lacked the sensitivity
and specificity of biopsy results. Granulomatous in
flammation in the temporal artery may also occur
because of other systemic vasculitides and may result
in a false-positive, misleading biopsy for giant cell
arteritis. The latter may be the reason why some
young patients are misdiagnosed as having giant cell
arteritis. We have occasionally seen patients who have
evidence of arteritis (for example, polyarteritis nodosa) in the temporal artery but who not suffer from
giant cell arteritis.
Some physicians routinely advise performing tem
poral artery biopsy on both sides in all patients

AMERICAN JOURNAL OF OPHTHALMOLOGY

MARCH 1997

suspected of having giant cell arteritis. In this context,

it is important to bear in mind that, contrary to the
prevalent impression, temporal artery biopsy is not an
entirely benign procedure; although it is usually safe, a
number of complications can develop from it, includ
ing severe hemorrhage, scalp necrosis, and infection.
However, these complications are rare and should not
dissuade a physician from doing temporal artery
biopsy on one side, because the usefulness of the
procedure heavily outweighs the minor risks. In this
study, we performed a biopsy on the second side only
when the results for the first side were equivocal or
negative for giant cell arteritis despite a very strong
index of suspicion clinically. It could be argued that
by not performing a second biopsy in the remaining
patients with negative biopsies, we might have missed
some of the patients with giant cell arteritis. A close
follow-up of all patients with negative temporal artery
biopsies showed that no stigmata of giant cell arteritis
had developed in any of them, indicating that we did
not miss any patients with giant cell arteritis. Patients
with negative temporal artery biopsy but with a high
erythrocyte sedimentation rate or C-reactive protein
level were referred to their internists for a thorough
systemic evaluation to find out the cause or causes of
the elevated erythrocyte sedimentation rate/C-reactive protein level and to institute appropriate treat
ment.
After reviewing the validity, reliability, sensitivity,
and specificity of various signs and symptoms of giant
cell arteritis, and of the erythrocyte sedimentation
rate and C-reactive protein level in the early diagnosis
of giant cell arteritis, we conclude that the set of
clinical criteria most strongly suggestive of giant cell
arteritis comprises jaw claudication, C-reactive pro
tein level above 2.45 mg/dl, neck pain, and an
erythrocyte sedimentation rate of 47 mm/hour or
more (Westergren's method), in that order of impor
tance. Using the 95th percentile as the cutoff point in
our control group, our study also showed that Creactive protein (at a level of 0.5 mg/dl or more) is
more sensitive (100% sensitivity) than the erythro
cyte sedimentation rate is (92% sensitivity) (Figure 1)
for the diagnosis of giant cell arteritis, and that the
erythrocyte sedimentation rate and C-reactive protein
level together provide the best specificity (97%). We
therefore recommend using both the erythrocyte
sedimentation rate and C-reactive protein level for
VOL.123, No. 3

diagnosis and management of patients with giant cell

arteritis.
ACKNOWLEDGMENTS

We wish to express our thanks to various members of

the Department of Ophthalmology for their help
with this study over the years, particularly Dr M.
Talibi, MD, and Dr H. Stanley Thompson, MD,
Director of the Neuro-ophthalmology Clinic, and to
acknowledge the late Dr F. C. Blodi, MD, and to
thank Dr Robert Folberg, MD, Directors of the
Ocular Pathology Laboratory. We also wish to thank
Dr Zuhair Ballas, MD, of the Department of Internal
Medicine for his help with the interleukin-6 estima
tion.

REFERENCES
1. Hayreh SS. Ophthalmic features of giant cell arteritis.
Baillieres Clin Rheumatol 1991;5:431-459.
2. Kearns TP. Collagen and rheumatic diseases: ophthalmic
aspects. In: Mausolf FA, editor. The eye and systemic disease.
St Louis: Mosby, 1975:105-118.
3. Keltner JL. Giant-cell arteritis: signs and symptoms. Oph
thalmology 1982;89:1101-1110.
4- Hayreh SS. Anterior ischaemic optic neuropathy: differentia
tion of arteritic from non-arteritic type and management. Eye
1990;4:25-41.
5. Miller A, Green M, Robinson D. Simple rule for calculating
normal erythrocyte sedimentation rate. Br Med J 1983;
286:266.
6. Gilbertsen VA. Erythrocyte sedimentation rates in older
patients: a study of 4,341 cases. Postgrad Med 1965;38:
A44-A52.
7. Hayreh SS, Zimmerman MB, Podhajsky P. Incidence of
various types of retinal vein occlusion and their recurrence
and demographic characteristics. Am J Ophthalmol 1994;
117:429-441.
8. Hayreh SS, Joos KM, Podhajsky PA, Long CR. Systemic
diseases associated with non-arteritic anterior ischemic optic
neuropathy. Am J Ophthalmol 1994;118:766-780.
9. Hosmer DW, Lemeshow S. Applied logistic regression. New
York: Wiley, 1989:88-97.
10. Hunder GG, Bloch DA, Michel BA, et al. The American
College of Rheumatology 1990 criteria for the classification
of giant cell arteritis. Arthritis Rheum 1990;33:1122-1128.
11. Chmelewski WL, McKnight KM, Agudelo CA, Wise CM.
Presenting features and outcome in patients undergoing
temporal artery biopsy: a review of 98 patients. Arch Intern
Med 1992;152:1690-1695.
12. Vilaseca J, Gonzalez A, Cid MC, Lopez-Vivancos J, Ortega
A. Clinical usefulness of temporal artery biopsy. Ann Rheum
Dis 1987;46:282-285.
13. Fernandez-Herlihy L. Temporal arteritis: clinical aids to
diagnosis. ] Rheumatol 1988;15:1797-1801.
14. Hollenhorst RW, Brown JR, Wagener HP, Shick RM.
Neurologic aspects of temporal arteritis. Neurology 1960;
10:490-498.

GIANT CELL ARTERITIS

295

15. Meyers L, Lord JW. Cranial arteritis: report of its occurrence

in a young woman. JAMA 1948;136:169-175.
16 Gilmour JR. Giant-cell chronic arteritis. J Pathol Bacteriol
1941;53:263-277.
17 Roberts AM, Askey JM. Temporal arteritis: relief of headache
by injection of procaine hydrochloride. JAMA 1948;137:
697-699.
18. Bethlenfalvay NC, Nusynowitz ML. Temporal arteritis: a
rarity in the young adult. Arch Intern Med 1964;114:
487-489.
19. Kent DC, Arnold H. Aneurysm of the aorta due to giant cell
aortitis: successful surgical correction. J Thorac Cardiovasc
Surg 1967;53:572-577.
20. Bugg El, Coonrad RW, Grim KB. Giant-cell arteritis: an
acute hand syndrome. J Bone Joint Surg 1963;45:1269-1272.
21. Gelfand M. Giant cell arteritis with aneurysmal formation in
an infant. Br Heart J 1955;17:264-266.
22. Wagenvoort CA, Harris LE, Brown AL, Veeneklaas GMH.
Giant-cell arteritis with aneurysm formation in children.
Pediatrics 1963;32:861-867.
23. Tomlinson FH, Lie JT, Nienhuis BJ, Konzen KM, Groover
RV. Juvenile temporal arteritis revisited. Mayo Clin Proc
1994;69:445-447.
24. Vidal E, Liozon F, Rogues AM, Cransac M, Berdha JF,
Liozon E. Concurrent temporal arteritis and Churg-Strauss
syndrome. J Rheumatol 1992;19:1312-1314.
25. Sparrow D, Rowe JW, Silbert JE. Cross-sectional and longitu
dinal changes in the erythrocyte sedimentation rate in men. J
Gerontol 1980;36:180-18426. Bottiger LE, Svedberg CA. Normal erythrocyte sedimenta
tion rate and age. Br Med J 1967;2:85-87.
27. Hanger WA, Sainsbury R, Gilchrist NL, Beard MEJ. Erythro
cyte sedimentation rates in the elderly: a community study. N
Z Med J 1991;104:134-136.
28. Cullen JF. Occult temporal arteritis. Trans Ophthalmol Soc
U K 1963;83:725-727.
29. Eagling EM, Sanders MD, Miller SJH. Ischaemic papillopathy. Br J Ophthalmol 1974;58:990-1008.
30. Kansu T, Corbett JJ, Savino P, Schatz NJ. Giant cell arteritis
with normal sedimentation rate. Arch Neurol 1977;34:
624-625.

31 Wong RL, Korn JH. Temporal arteritis without an elevated

sedimentation rate: case report and review of the literature.
Am J Med 1986;80:959-964.
32 Grodum E, Petersen HA. Temporal arteritis with normal
erythrocyte sedimentation rate. J Intern Med 1990;227:
279-280.
33 Jundt JW, Mock D. Temporal arteritis with normal erythro
cyte sedimentation rates presenting as occipital neuralgia.
Arthritis Rheum 1991;34:217-219.
34. Wise CM, Agudelo CA, Chmelewski WL, McKnight KM.
Temporal arteritis with low erythrocyte sedimentation rate: a
review of five cases. Arthritis Rheum 1991;34:1571-1574.
35. Neish PR, Sergent JS. Giant cell arteritis: a case with unusual
neurologic manifestations and a normal sedimentation rate.
Arch Intern Med 1991;151:378-380.
36. Zweegman S, Makkink B, Stehouwer CD. Giant cell arteritis
with normal erythrocyte sedimentation rate: case report and
review of the literature. Neth J Med 1993;42:128-131.
37. Small JM, Gavrilescu K. The serum protein changes in giant
cell arteritis. J Neurol Neurosurg Psychiatry 1963;26:
257-261.
38. Eshaghian J, Goeken JA. C-reactive protein in giant cell
(cranial, temporal) arteritis. Ophthalmology 1980;87:
1160-1166.
39. Espinoza L, Jara LJ, Silveira LH, et al. Anticardiolipin
antibodies in polymyalgia rheumatica-giant cell arteritis:
association with severe vascular complications. Am J Med
1991;90:474-478.
40. Berthelot J-M, Lamour A, Mottier D, et al. Anti-neutrophil
cytoplasm antibodies in patients with giant cell arteritis
and/or polymyalgia rheumatica. Ann Med Intern 1992;
143:429-432.
41 Deodhar SD. C-reactive protein: the best laboratory indica
tor available for monitoring disease activity. Cleve Clin J Med
1989;56:126-130.
42 Hedges TR, Gieger GL, Albert DM. The clinical value of
negative temporal artery biopsy specimens. Arch Ophthalmol
1983;101:1251-1254.

Authors Interactive
Comments and questions of 100 words or less regarding this article may be
addressed to authors via THE JOURNAL'S Web Site on the Internet. Selected
questions and comments along with the authors' statements and responses will
be posted on the Authors Interactive Bulletin Board, a feature of THE
JOURNAL'S home page. Readers may access THE JOURNAL'S home page and the
Authors Interactive Bulletin Board at http://www.ajo.com/

296

AMERICAN JOURNAL OF OPHTHALMOLOGY

MARCH

1997