SPINE Volume 40, Number 7, pp 475-479

2015, Wolters Kluwer Health, Inc. All rights reserved.

DIAGNOSTICS

Hoffmann Sign
Clinical Correlation of Neurological Imaging Findings in the Cervical Spine and Brain

Ray A. Grijalva, MD,* Frank P. K. Hsu, MD, PhD, Nathaniel D. Wycliffe, MD, Bryan E. Tsao, MD,
Paul Williams, MS, Yusuf T. Akpolat, MD, and Wayne K. Cheng, MD

and 72% negative predictive value for cervical cord compression.

Study Design. Retrospective validity study.
For brain pathology, sensitivity was 71%, specificity 33%, positive
Objective. To investigate the relationship between Hoffmann sign
predictive value 10%, and negative predictive value 95%.
and radiographical evidence of cervical spinal cord compression
Conclusion. Hoffmann sign has too low a positive predictive value
and brain lesions.
to be relied upon as a stand-alone physical examination finding and
Summary of Background Data. Clinical significance of
is not a reliable screening tool for solely predicting the presence of
Hoffmann sign remains controversial with conflicting reports
cervical spinal cord compression or brain pathology.
regarding its sensitivity and specificity and its usefulness.
Key words: Hoffmann sign, clinical correlation, neurological
Methods. Patients were divided into 2 groups according to the
imaging, cervical spine, brain, sensitivity, specificity, positive
presence of Hoffmann sign on physical examination. Imaging studies
predictive value, negative predictive value, accuracy.
were blindly examined by 2 observers for possible cervical and
Level of Evidence: 2
brain lesions. The sensitivity, specificity, positive predictive value,
Spine 2015;40:475479
negative predictive value, as well as accuracy for Hoffmann sign as
it relates to cervical spinal cord compression and brain pathology,
were calculated.
Results. Of the 91 patients with a positive Hoffmann sign, 32 (35%)

A
showed severe cervical cord compression and/or myelomalacia. lthough its origin remains controversial, Hoffmann
Forty-seven of these patients had brain imaging studies, and 5 sign has been in clinical use for more than a century.
(10%) had positive findings. There were 80 patients in the negative The sign is attributed to Johann Hoffmann, a pupil of
Hoffmann sign or control group. Twenty-one (27%) of them had Erb and a Professor of Neurology at Heidelberg, Germany, in
severe cervical cord compression and/or myelomalacia. Twenty- the late 19th century. The sign was first reported in 1911 by
three of these control patients underwent neurological imaging of the one of his assistants, Hans Curschmann, who coined the mon-
brain, and 2 (8%) had positive findings. Hoffmann sign was found to iker.1 In response to an inquiry, Dr. Curschmann later wrote24:
have 59% sensitivity, 49% specificity, 35% positive predictive value,
The finger phenomenon mentioned by me originates
from Johann Hoffmann, Professor of Neurology at
From the *Department of Orthopaedic Surgery, Kaiser Permanente, Riverside, Heidelberg (died 1919). I learned it while his pupil and
CA; Department of Neurosurgery, School of Medicine, University of assistant from 1901 to 1904. He demonstrated it in his
California, Irvine; Department of Diagnostic Radiology, Loma Linda
University Medical Center, Loma Linda, CA; Department of Neurology,
classes and clinics as a sign of hyper-reflexia of the upper
Loma Linda University Medical Center, Loma Linda, CA; Department of extremity. So far as I know he never published it (p. 202).
Pharmaceutical Science, School of Pharmacy, Loma Linda University, Loma
Linda, CA; and Department of Orthopaedic Surgery, Loma Linda University
Medical Center, Loma Linda, CA.
Hoffmann sign was originally described as follows2,4:
Acknowledgment date: September 5, 2014. Revision date: December 1,
2014. Acceptance date: January 15, 2015. The test is performed by supporting the patients hand so that
The manuscript submitted does not contain information about medical it is completely relaxed and the fingers are partially flexed.
device(s)/drug(s). The middle finger is firmly grasped, partially extended, and
No funds were received in support of this work. the nail snapped by the examiners thumbnail. The snapping
Relevant financial activities outside the submitted work: board membership, should be done with considerable force, even to the point
consultancy, payment for lectures, employment, travel/accommodations/
meeting expenses.
of causing pain. The sign is present if quick flexion of both
Address correspondence and reprint requests to Wayne K. Cheng, MD,
the thumb and index finger results. Fingernails other than
Spine Services, Department of Orthopaedic Surgery, Loma Linda University the middle one are sometimes selected for the snapping.
Medical Center, 11406 Loma Linda Dr, Ste 213, Loma Linda, CA 92354;
E-mail: [email protected]
Currently, Hoffmann sign is used as a test for cortico-
DOI: 10.1097/BRS.0000000000000794 spinal pathway dysfunction. It has also been described as
Spine www.spinejournal.com 475
Copyright 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.

SPINE141190_LR 475 09/03/15 8:46 AM

 DIAGNOSTICS Hoffmann Sign, Imaging-Clinical Correlation Grijalva et al

the digital reflex, the snapping reflex, Jakobson sign, and position. The test was deemed positive if there was flexion of
Tromner sign.3,5,6 When present, Hoffmann sign is thought the ipsilateral thumb and/or the index finger.
to be indicative of upper motor neuron disease, especially for All neurological images of the spinal canal were blindly
lesions affecting the cervical spinal cord.5,79 However, the and independently examined by an orthopedic spine surgeon
clinical significance of Hoffmann sign remains controversial, and a neuroradiologist. Only magnetic resonance images or
with conflicting reports regarding its sensitivity, specificity, computed tomographic (CT) myelograms were defined as cer-
and positive and negative predictive values.4 In a compre- vical neurological imaging and were reviewed for evidence of
hensive review, 3 general views were described by Malanga spinal cord compression.
et al.4,10 The first is that Hoffmann sign is a pathologic Cervical spinal cord compression was defined as complete,
sign, indicating pyramidal tract involvement. The second anterior and posterior, cerebrospinal fluid effacement and
is that the sign indicates pyramidal tract involvement but deformation of the cord contour at the level of cerebrospinal
that, owing to its frequent presence in other conditions, its fluid effacement, or T2 lengthening within the spinal cord.7,14
clinical value is doubtful. Finally, Hoffmann sign is not Imaging findings were only considered positive for cervical
pathologic of any clinical value. Moreover, Curschmann cord compression if both observers agreed.
did not think that Hoffmann sign had pathognomonic sig- The medical records of both sets of patients were retro-
nificance as a Babinski of the upper extremity, because he spectively reviewed for neurological imaging studies of the
also found the reflex in patients with non-neurological dis- brain. Only magnetic resonance images and CT scans were
orders such as hysteria and neurasthenia. He agreed with defined as brain neurological imaging and were reviewed for
others that the reflex may occur in healthy patients and in evidence of cerebral pathology. All neurological imaging stud-
particular nervous individuals without any organic disor- ies of the brain were blindly and independently reviewed by a
der.11,12 Although several previous studies have examined the neurosurgeon and a neuroradiologist.
relationship between Hoffmann sign and spinal pathology, Criteria for brain pathology were based on lesion loca-
little is known about the relationship of Hoffmann sign to tion, size, and number. Cerebral lesions must have involved
pathology in the brain.5,7,8,13 The goals of this study were first either the cortex or corticospinal tract and must be larger than
to examine the relationship between Hoffmann sign and cer- 2 mm. A study was deemed positive for cerebral pathology if
vical pathology in symptomatic patients. Second, to evaluate there were more than 5 lesions or a single lesion larger than
whether patients with Hoffmann sign without radiographi- 1 cm. As with the cervical spine, positive studies were agreed
cal evidence of cervical myelopathy require further workup upon by both observers.
for a provocative lesion in the brain. Specifically, we asked The presence or absence of Hoffmann sign, cord compres-
the question: What percentages of patients with a positive sion, and brain pathology, as well as age, sex, and whether or
Hoffmann sign without radiographical evidence of cervi- not cervical decompression or brain surgery was performed,
cal pathology have a lesion in the brain that could explain were recorded.
this reflex? Finally, to measure the specificity, sensitivity, and The sensitivity, specificity, positive predictive value, nega-
positive and negative predictive values of Hoffmann sign for tive predictive value, as well as accuracy for Hoffmann sign as
cervical and brain lesions. it relates to cervical spinal cord compression and brain pathol-
ogy, were calculated. The disagreement numbers and coeffi-
MATERIALS AND METHODS cient of correlation ( statistic) was also determined compar-
This was a retrospective analysis of patients with cervical ing the readings of the surgeon and the neuroradiologist in the
complaints who underwent neurological imaging present- cervical spine and brain groups. Statistical significance was set
ing to a single orthopedic spine surgeon from April 2007 to at P = 0.05. True positive rate (sensitivity) was plotted as the
July 2009. The positive Hoffmann (PH) group consisted of function of the false positive rate (100 specificity) to obtain
patients who had (1) neck pain or radicular arm complaints the receiver operating characteristic (ROC) curve.
at initial presentation, (2) a positive Hoffmann sign, and (3)
neurological images of their cervical spine available for review RESULTS
by the examining physician. The control group or negative There were 91 patients in the PH group and 80 controls who
Hoffmann (NH) consisted of patients who had (1) neck met our inclusion criteria. Sixty-nine were female (76%)
pain or radicular arm complaints at initial presentation, (2) and 22 were male (24%). The average age was 55 years. In
no Hoffmann sign elicited, and (3) cervical spine neurologi- the control group, 49 were female (61%) and 31 were male
cal images available for review by the examining physician. (39%). There was no significant difference in the age, male to
Systemic diseases that may cause hyporeflexia were looked in female ratio, or presence of cervical spinal cord compression
both the PH and NH groups. between the groups (Table 1). Systemic disease was present in
The physical evaluation consisted of a standard history and 28 patients in the PH group and 23 patients in the NH group.
physical examination. The Hoffmann test was conducted by There was no statistical difference between 2 groups (P =
a single orthopedic spine surgeon. The test was conducted by 0.64). The diseases and numbers are summarized in Table 2.
flicking the long finger from dorsal to volar with the patients Of the 91 patients with a positive Hoffmann sign, neuro-
hand supported by the examiner with the wrist in slight logical imaging consisted of 85 magnetic resonance images
dorsiflexion. The test was done with the neck in the neutral and 6 CT myelograms. Thirty-two (35%) of these patients
476 www.spinejournal.com April 2015
Copyright 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.

SPINE141190_LR 476 09/03/15 8:46 AM

 DIAGNOSTICS Hoffmann Sign, Imaging-Clinical Correlation Grijalva et al

TABLE 1. Descriptive Statistics for 171 Patients TABLE 3. Sensitivity, Specificity, and Positive
Who Composed the Cervical Spine and Negative Predictive Values
Group for Hoffmann Sign As It Relates to
Positive Cervical Spine Pathology
Hoffmann Control Spinal Cord No
(n = 91) (n = 80) P Compression Compression
Age, mean SD, yr 54.68 12.76 56.51 14.95 0.400 Positive Hoffmann 32 (A) 59 (B)
Female:Male 3 1.6 0.062 Negative Hoffmann 21 (C) 59 (D)
Stenosis, n (%) 32 (35) 21 (27) 0.761 Sensitivity A/A + C 59.2%
Surgery, n (%) 21 (23) 10 (12.5) 0.049 Specificity D/B + D 49.5%
Positive predictive value
showed severe cervical cord compression and/or myelo- 35.1%
A/A + B
malacia. Twenty-one (23%) had surgery during the 2-year
Negative predictive
follow-up. 72.5%
value D/D + C
In the control group, all 80 patients had a magnetic reso-
nance image of the cervical spine available for review. Twenty-
one (27%) patients with a negative Hoffmann sign had severe Forty-seven (52%) of the 91 patients with a positive
cord compression and/or myelomalacia. Seventeen (21%) Hoffmann sign were found to have neurological imaging of
patients had cervical spine surgery at 2 years. the brain (34 magnetic resonance images and 13 plain CT
scans) for review. Of these, 5 (10%) had positive findings,
2 had pathology due to leukoencephalopathy, and 3 had
TABLE 2. Systemic Diseases in Both the Positive pathology due to infarction. However, no patient had brain
Hoffmann and Negative Hoffmann surgery within 2 years. Of the 5 patients with a positive Hoff-
Groups mann sign found to have brain pathology, 3 did not have cer-
vical cord compression.
Positive Negative In the control group, 23 patients were found to have neu-
Hoffmann Hoffmann
rological imaging of the brain (15 magnetic resonance images
No systemic disease 63 56 and 8 plain CT scans). Of these, 2 patients (8%) had posi-
Endocrinopathy tive findings, both with pathology due to tumor, and a single
patient had surgery during the 2 years of the study.
Diabetes 9 10 The sensitivity of Hoffmann sign relative to cord compres-
Hypothyroid 14 10 sion was found to be 59%, specificity 49%, positive predic-
Systemic vasculitis tive value 35%, and negative predictive value 72% (Table 3).
For brain pathology, Hoffmann sign was found to have a
Rheumatoid arthritis 1 3 sensitivity of 71%, specificity 33%, positive predictive value
SLE 0 0 10%, and negative predictive value 95% (Table 4).
PAN 0 0
Renal disease
TABLE 4. Sensitivity, Specificity, and Positive
Chronic renal failure 3 0
and Negative Predictive Values for
Hematological disease Hoffmann Sign As It Relates to Brain
Vitamin B12 deficiency 1 1 Pathology
Alcoholism 0 0 Brain No Brain
Hepatic disease 0 0 Pathology Pathology
Fluid and electrolyte imbalance 0 0 Positive Hoffmann 5 (A) 42 (B)

Pregnancy 0 0 Negative Hoffmann 2 (C) 21 (D)

Sarcoidosis 0 0 Sensitivity A/A + C 71.4%

Transplant patients 0 0 Specificity D/B + D 33.3%

P = 0.64 Positive predictive value A/A + B 10.6%

SLE indicates systemic lupus erythematosus; PAN, poliarteritis nodosa. Negative predictive value D/D + C 95.4%

Spine www.spinejournal.com 477

Copyright 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.

SPINE141190_LR 477 09/03/15 8:46 AM

 DIAGNOSTICS Hoffmann Sign, Imaging-Clinical Correlation Grijalva et al

DISCUSSION
In this study, we examined the relationship between Hoffmann
sign in symptomatic patients and correlative cervical spine
and brain imaging. Although Hoffmann sign was more com-
mon in patients with cervical spinal cord compression and/
or myelomalacia than in controls, we found this sign to have
a low positive predictive value of 35% and to be absent in
40% of patients with confirmed cervical pathology. Thus, the
presence of Hoffmann sign should not be used as a singu-
lar surrogate for the presence of cervical cord compression.
Alternatively, the absence of Hoffmann sign does not exclude
the presence of significant cervical myelopathy. The sensitivity
of Hoffmann sign to detect severe cervical cord compression
in our study correlated with findings of the previous studies
where sensitivity ranged from 58% to 68% in symptomatic
patients.5,7,13 The specificity of Hoffmann sign was 50% in
our study, which is considerably lower than a reported speci-
Figure 1. ROC curve for Hoffmann sign for cervical spinal cord com- ficity of 84%.13 One reason for this discrepancy is that prior
pression. ROC indicates receiver operating characteristics.
studies excluded patients with other noncervical spondylotic
disorders capable of producing myelopathic signs, making
An ROC curve was obtained for Hoffmann sign as a test the specificity values artificially higher than they would be in
for cervical spinal cord compression (Figure 1). The area under a general population that includes such patients. Our study
the curve was calculated and found to be 0.519, which is not included such patients and is more likely to accurately repre-
significantly different than chance (P = 0.278; 95% confi- sent the true value in the clinical setting. The positive predic-
dence interval [CI], 0.4590.654). An ROC curve obtained tive value reported in our study is much lower than what was
for Hoffmann sign as a test for brain pathology (Figure 2) reported by Glasser et al,5 where radiographs were examined
showed the area under the curve to be 0.519, which, again, unblinded. However, when the examiner was blinded, the
was not significantly different from chance (P = 0.872; 95% value dropped significantly to 26%, which correlates more
CI, 0.2950.743). closely with our value of 35%. Likewise, the negative predic-
Cohen values were calculated to determine interob- tive value of 73% in our study correlated with the 75% value
server reliability of magnetic resonance imaging and CT reported by Glasser et al and this value only decreased slightly
interpretation. There were 10 of 171 disagreements between to 67% when the examiner was blinded.
the observers for interpretation of spinal imaging and 5 of In our examination of the relationship of Hoffmann sign
70 for brain imaging. For cervical spinal cord compres- to causative lesions in the brain, there were no previously
sion, Cohen showed 0.895 correlation (95% CI, 0.8243 reported results for comparison. Although the radiographical
0.9656). For brain pathology, Cohen was 0.8592 (95% CI, criteria for cervical spinal cord compression had been defined
0.66871.0496). previously,14 the radiographical criteria for a brain lesion that
could cause Hoffmann sign needed to be created. This was
done with the co-operative efforts of the neuroradiologist
(N.W.) and the neurosurgeon (F.H.).
One weakness in this study is that there are no previous
studies to validate these criteria. Even so, the purpose of this
examination was to elucidate whether significant lesions in
the brain are being missed in patients with Hoffmann sign.
Only 5 patients in the PH group had lesions in the brain that
could cause Hoffmann sign based on our criteria, none of
which required surgical intervention. Interestingly, of the 5
patients with Hoffmann sign, 3 did not have radiographical
evidence of cervical cord compression and/or myelomalacia.
Although these numbers are too low to draw definitive conclu-
sions, they are in keeping with others, including Curschmann,
who found that the reflex can be elicited in subjects without
a defined central nervous system disorder. This is not to say
that Hoffmann reflex should be discarded entirely: we agree
with the statement that Hoffmann sign may be indicative of
Figure 2. ROC curve of Hoffmann sign for brain pathology. ROC indi- a pyramidal tract lesion, especially in cases with asymmetric
cates receiver operating characteristics. findings and in the presence of other pathological reflexes.12
478 www.spinejournal.com April 2015
Copyright 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.

SPINE141190_LR 478 09/03/15 8:46 AM

 DIAGNOSTICS Hoffmann Sign, Imaging-Clinical Correlation Grijalva et al

In conclusion, we feel that Hoffmann sign has too low a References

positive predictive value to be relied upon as a stand-alone 1. Curschmann H. ber die diagnostische Bedeutung des Babin-
physical examination finding for predicting the presence of skischen Phnomen im prurmischen Zustand. Munch Med
Wschr 1911:20547.
cervical spinal cord compression or brain pathology. If used, 2. Echols DH. The Hoffmann sign: its incidence in university stu-
it should be interpreted in the context of other pathological dents. J Ment Nerv Dis 1936;84:42731.
reflexes and signs. Furthermore, routine brain imaging in 3. Bendheim OL. On the history of the Hoffmann sign. Bull Hist Med
patients with a Hoffmann sign without evidence of structural 1937:5:6846.
4. Malanga GA, Landes P, Nadler SF. Provocative tests in cervical
spinal myelopathy is of low yield. Therefore, appropriate clin- spine examination: historical basis and scientific analyses. Pain
ical judgment should be used in the decision for imaging the Physician 2003;6:199205.
brain in patients with a Hoffmann sign without myelopathy. 5. Glaser JA, Cure JK, Bailey KL, et al. Cervical spinal cord compres-
sion and the Hoffmann sign. Iowa Orthop J 2001;21:4952.
6. Brain L, Wilkinson M, eds. Cervical Spondylosis and Other Disor-
ders of the Cervical Spine. Philadelphia: WB Saunders; 1967.
Key Points 7. Houten JK, Noce LA. Clinical correlations of cervical myelopathy
and the Hoffmann sign. J Neurosurg Spine 2008;9:23742.
8. Sung RD, Wang JC. Correlation between a positive Hoffmanns
The clinical significance of Homann sign remains reflex and cervical pathology in asymptomatic individuals. Spine
controversial, with conflicting reports regarding 2001;26:6770.
its sensitivity and specificity. 9. Denno JJ, Meadows GR. Early diagnosis of cervical spondylotic
myelopathy. A useful clinical sign. Spine 1991;16:13535.
For cervical cord spinal lesions, Homann sign 10. Madonick MJ. Statistical control studies in neurology, part III:
was found to have a sensitivity of 59%, specificity the Hoffman sign. AMA Arch Neurol Psychiatry 1952;68:
49%, positive predictive value 35%, and negative 10915.
predictive value 72%. 11. Brian R. Diseases of the Nervous System. 2nd ed. London: Oxford
University Press; 1940.
For brain lesions, the sensitivity of Homann sign 12. Krouwer HGJ, Barkhaus PE, Antuono PG. The reflexes of Hoff-
was 71%, specificity 33%, positive predictive value mann, Tromner, and Mayer. In: Koehler PJ, Bruyn GW, Pearce JMS,
10%, and negative predictive value 95%. eds. Neurological Eponyms. London: Oxford University Press;
Homann sign has too low a positive predictive 2000:12735.
13. Rhee JM, Heflin JA, Hamasaki T, et al. Prevalence of physical
value to be relied upon as a stand-alone physical signs in cervical myelopathy: a prospective, controlled study. Spine
examination finding and is not a reliable screen- 2009;34:8905.
ing tool for predicting the presence of cervical 14. Singh A, Crockard HA, Platts A, et al. Clinical and radiological
spinal cord compression or brain pathology. correlates of severity and surgery-related outcome in cervical spon-
dylosis. J Neurosurg 2001;94:18998.

Spine www.spinejournal.com 479

Copyright 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.

SPINE141190_LR 479 09/03/15 8:46 AM