295
ARTICLES
Piriformis Syndrome: Diagnosis, Treatment, and Outcome—
a 10-Year Study
Loren M. Fishman, MD, George W. Dombi, PhD, Christopher Michaelsen, MD, Stephen Ringel, MD,
Jacob Rozbruch, MD, Bernard Rosner, PhD, Cheryl Weber, MD
ABSTRACT. Fishman LM, Dombi GW, Michaelsen C,
Ringel S, Rozbruch J, Rosner B, Weber C. Piriformis
syndrome: diagnosis, treatment, and outcome—a 10-year
study. Arch Phys Med Rehabil 2002;83:295-301.
Objectives: To validate an operational definition of pirifor-
mis syndrome based on prolongation of the H-reflex with hip
flexion, adduction, and internal rotation (FAIR) and to assess
efficacy of conservative therapy and surgery to relieve symp-
toms and reduce disability.
Design: Before-after trial of cohorts identified by opera-
tional definition.
Setting: Outpatient departments of 2 hospitals and 4 physi-
cians’ offices. Surgery performed at 3 hospitals.
Patients: Consecutive sample of 918 patients (1014 legs)
with follow-up on 733.
Intervention: Patients with significant (3 standard devia-
tions [SDs]) FAIR tests received injection, physical therapy,
and serially reported pain and disability assessments. Forty-
three patients (6.47%) had surgery.
Main Outcome Measures: Likert pain scale. Subjective
estimates of disablement in activities of daily living and instru-
mental activities of daily living.
Results: At 3 SDs, the FAIR test had sensitivity and spec-
ificity of .881 and .832, respectively. Seventy-nine percent
(514/655) of FAIR test positive (FTP) patients improved 50%
or more from injection and physical therapy at a mean fol-
low-up of 10.2 months. Average improvement was 71.1%. Of
385 FTP patients with disability data, mean disability fell from
35.37% prestudy (SD ⫽ .2275) to 12.96% poststudy (SD ⫽
.1752), a 62.8% improvement. Twenty-eight surgical FTP pa-
tients (68.8%) showed 50% or greater improvement; mean
improvement was 68% at a mean follow-up of 16 months.
Surgery reduced the mean FAIR test to 1.35 ⫾ 2.17 months
postoperatively. FTP patients generally improved 10% to 15%
more than others after conservative treatment.
Conclusions: The FAIR test correlates well with a working
definition of piriformis syndrome and is a better predictor of
successful physical therapy and surgery than the working def-
From the Department of Physical Medicine and Rehabilitation, New York Flushing
Hospital and New York–Presbyterian Medical Center, New York, NY (Fishman);
University of Detroit, Detroit, MI (Dombi); Department of Orthopedic Surgery, New
York–Presbyterian Medical Center, New York, NY (Michaelsen); Department of
Orthopedic Surgery, Texas Tech University Health Sciences Center School of Med-
icine, Lubbock, TX (Ringel); Department of Orthopedic Surgery, Beth Israel Medical
Center, New York, NY (Rozbruch); Department of Biostatistics, Deaconess and
Brigham & Women’s Hospital, Harvard Medical School, Boston, MA (Rosner); and
Department of Medical and Surgical Neurology, Texas Tech University, Lubbock,
TX (Weber).
Accepted in revised form March 28, 2001.
No commercial party having a direct financial interest in the results of the research
supporting this article has or will confer a benefit upon the authors(s) or upon any
organization with which the author(s) is/are associated.
Reprint requests to Loren M. Fishman, MD, 3 E 83rd St, New York, NY 10028,
e-mail: Loren@arcon-inc.com.
0003-9993/02/8303-6567$35.00/0
doi:10.1053/apmr.2002.30622
inition. The FAIR test, coupled with injection and physical
therapy and/or surgery, appears to be effective means to diag-
nose and treat piriformis syndrome.
Key Words: Electrodiagnosis; Physical therapy; Reflex,
monosynaptic; Rehabilitation; Sciatica; Treatment outcome.
© 2002 by the American Congress of Rehabilitation Medi-
cine and the American Academy of Physical Medicine and
Rehabilitation
INCE MIXTER AND BARR’S classic paper in 1934, 1
S diagnostic attention has focused on intramedullary and fo-
raminal causes of sciatica. Progressive advances in imaging
technology, anaesthesiology, antibiotics, and spinal surgery
have given further impetus to this trend. Piriformis syndrome
has been an elusive diagnosis, often a diagnosis of exclusion,
invoked by clinicians only when adequate inquiry suggested no
spinal cause of significant sciatic pain.2-6 Yet, recent studies
confirm that imaged abnormalities may cause no pain,7 and
nearly every clinician has found significant, even unbearable
sciatica in individuals with normal computed tomography,
magnetic resonance imaging (MRI), myelogram, and conven-
tional electromyography.
The term sciatica was coined in Florence in the 15th century
for leg pain thought to originate at the ischium. Contemporary
use of piriformis syndrome began with Robinson in 1947,8 who
delineated 5 salient characteristics: (1) history of local trauma;
(2) pain localized to the sacroiliac joint, greater sciatic notch,
and piriformis muscle, which extends along the course of the
sciatic nerve and presents difficulty in walking; (3) acute pain
brought on by stooping or lifting and relieved somewhat by
traction; (4) palpable spindle or sausage-shaped mass at the
anatomic location of the piriformis muscle; and (5) positive
Lasègue’s sign. Robinson also found gluteal atrophy more
likely in piriformis syndrome.
Ten years earlier, Freiberg,9 a surgeon, succinctly gave 3
indications of piriformis-caused sciatica: (1) tenderness at the
sciatic notch, (2) positive Lasègue’s sign, and (3) improvement
with nonsurgical treatment.
But sacroiliac joint derangement and gluteal injections fit
Freiberg’s criteria, and Robinson’s exacting list selects out a
fraction of the piriformis syndrome cases identified clinically.
Finally, no rationale was given by either investigator for these
boundary rules.
While described as “difficult to substantiate,”3 the incidence
of piriformis syndrome has been estimated at 6% to 8% of low
back pain (LBP).4 It stands to reason that pain along the course
of the sciatic nerve would at times be caused by pathologic
involvement of the nerve itself and that rational diagnosis and
treatment would then focus on the site of the pathology. The
first purpose of the present study was to define operationally
piriformis syndrome according to its pathogenic mechanism.
Thereafter piriformis syndrome cannot remain a diagnosis of
exclusion.
Few diagnostic methods have been described for piriformis
syndrome,8-20 and only 1 includes statistical evaluation.21 Si-
Arch Phys Med Rehabil Vol 83, March 2002
296 PIRIFORMIS SYNDROME, Fishman
naki et al,12 Nainzadeh and Lane,17 and Syneck18 used somato-
sensory evoked potentials in a total of 7 cases to document
sensory nerve conduction delay in the region of the sciatic
nerve’s entry to the buttock adjacent to the piriformis muscle.
But no comparison with normals, no estimate of normal vari-
ations, and no measures of sensitivity or specificity were made.
Pace and Nagle,22 Beaton and Anson,23 and Freiberg and
Vinke24 described symptoms and (usually the absence of) signs
that seemed typical of piriformis syndrome. Yeoman,25 Pe-
cina,26 and Gotlin27 documented anatomic variation in different
parts of the world. But clinical correlation of the 15% to 20%
of variant sciatic nerve-piriformis muscle passages detected in
these cadaveric studies with actual cases of piriformis syn-
drome was naturally impossible.
Successful surgical series of Solheim et al,16 Freiberg and
Vinke,24 and Mizuguchi28 corroborate the remediability of the
syndrome. Still, without reliable means of diagnosis, one can-
not determine how small a subset of possible cures these cases
represent. Fishman and Zybert,21 who used delay of the H-
reflex with hip flexion, adduction, and internal rotation (FAIR)
to diagnose piriformis syndrome, showed sensitivity and selec-
tivity in 38 cases. The present study builds on that work: it
formulates working diagnostic criteria for piriformis syndrome,
validates electrophysiologic evidence of piriformis syndrome,
and assesses nonsurgical and surgical treatments of cases di-
agnosed by each criterion.
METHODS
Attention from the lay press from 1991 to 1994 drew to our
offices individuals from all parts of the globe who believed
they had piriformis syndrome. Approximately 75% lived in
New York, Connecticut, New Jersey, and Pennsylvania. An-
other 20% came from other American urban centers, and 5%
were from North and South America, Europe, Asia, Africa, and
Australia.
Serial patients presenting with LBP or sciatica were classi-
fied as having piriformis syndrome if they met at least 2 of 3
clinical criteria: (1) pain at the intersection of the sciatic nerve
and the piriformis muscle (the site of the pathology) on FAIR
(fig 1); (2) tenderness at the intersection of the piriformis
muscle and the sciatic nerve (mechanical pressure replicates
the pathogenetic mechanism); and (3) positive supine La-
sègue’s sign, applied as 15° reduction in straight-leg raise on
the affected side versus the unaffected side, or less than 65°
(which intensifies contact between the tendinous edge of the
piriformis muscle and the sciatic nerve).
Detailed history and thorough physical examination were
recorded on 918 consecutive patients complaining of LBP
Fig 1. The FAIR position. Simultaneous downward pressure at the
flexed knee and passive superolateral movement of the shin, with
both acetabula oriented vertically, maximizes adduction and inter-
nal rotation at the flexed thigh. This position is important in treating
piriformis syndrome as well. (Reprinted with permission.21)
Arch Phys Med Rehabil Vol 83, March 2002
and/or sciatica, in whom 1014 lower limbs were involved (96
cases were bilateral).
Each patient was then placed in the anatomic position, and
unilateral or bilateral posterior tibial and peroneal H-reflexes
were sought. We conducted posterior tibial H-reflex testing
according to the guidelines described by Hugon29 and Braddom
and Johnson.30 Peroneal H-reflexes were elicited by placing a
bar electrode 6cm distal to the fibular head and stimulating at
the lateral popliteal fossa. Gain was set at 500mV, and filters
were set at 100 to 10,000Hz, with sweep at 5ms/div. Limb
temperature, preexamination exercise, and timing of successive
H-reflex stimuli were taken into account throughout the work,
according to current guidelines of Bell and Lehmann31 and
Halar et al.32 Patients were then placed in the FAIR position
(fig 1). H-reflexes were elicited again, observing the same
guidelines. FAIR-position H-reflexes were elicited within 5
seconds of patients’ achieving the position, and the greatest
twice-replicable latencies were recorded. A rest interval of 5 or
more seconds separated 1 FAIR test positioning and testing
procedure from the next, minimizing the effect of repetitive
stimulation on H-reflex latency.31,32 H-reflexes and M waves of
similar configuration were recorded in anatomic and FAIR
positions for both nerves.
The legs of 44 asymptomatic individuals had been tested in
a previous study according to the same guidelines.21 An addi-
tional 44 legs of asymptomatic volunteers were tested, bringing
the total to 88. The standard deviation (SD) of these normal
legs’ H-reflex latency delay in the FAIR position was used to
gauge abnormality in suspected cases of piriformis syndrome.
For comparison, contralateral legs of 229 FAIR test positive
(FTP) patients were tested according to the same protocol.
Measurement of Delay
To account for the subcutaneous nerve movements that
occur when patients change from anatomic to the FAIR posi-
tion, we calculated 1 entire reflex arc’s latency by taking the
sum of the distal motor latency or M wave from the point of
stimulation to the soleus or peroneus longus muscle, plus the
H-reflex latency (H loop). With this approach, movement in
the point of nerve stimulation does not change the latency of
the reflex arc, except that the sensory limb is made shorter
(or longer) by the same distance that the path of the distal
motor latency is lengthened (or shortened).
Discrepancy Between Motor and Sensory Nerve
Conduction Velocity
One can estimate the range of difference in the FAIR test
assuming a 5-cm movement of the sciatic nerve and a 10-m/s
discrepancy between motor and sensory nerve conduction ve-
locity (NCV) with the following formula:
D/R ⫽ T:
.05m ⫼ 50m/s ⫽ .001s
.05m ⫼ 40m/s ⫽ .00125s
Difference in H-reflex latency ⫽ ⫺ .00025s
If a 10-m/s discrepancy between motor (50m/s) and sensory
(40m/s) NCV were present, then a 5-cm movement of the
sciatic nerve’s stimulation point would entail a .25ms change in
the sum of M ⫹ H latencies.
One SD in H-reflex latency for normal persons equals .62ms.
Because motor and sensory NCV are nearly equal in normal
circumstances, the point of stimulation is generally immaterial
to H-loop values (figs 1–3).
 PIRIFORMIS SYNDROME, Fishman
Fig 2. Calculating effects of moving the stimulation point from S1
to S2. If the nerve stimulus point changes from S1 to S2, then the
H-reflex is shortened by the distance (S1 ⴚ S2)/sensory NCV. How-
ever, the M-wave latency is increased by (S1 ⴚ S2)/motor NCV. In
the case of a sharp discrepancy between motor and sensory NCV,
this factor still amounts to less than 0.5 SD. (Adapted and reprinted
with permission from Ma DM, Liveson JA. Nerve conduction hand-
book. New York: Oxford University Press, Inc.; 1987. p 261.)
Treatment
Probands’ H-loop changes from anatomic to the FAIR po-
sition were compared with the differences seen in normal
persons. Patients whose posterior tibial or peroneal FAIR tests
were delayed more than 3 SDs beyond the mean established in
the 88 normals, and/or for whom at least 2 of 3 clinical criteria
were positive for piriformis syndrome, and/or where clinical
suspicion was high were treated according to the following
3-part protocol.
1. Injection. A solution containing 1.5mL of 2% lidocaine
and .5mL containing 20-mg triamcinolone acetonide was in-
jected at a point one third the distance from the greater tro-
chanter to the area of maximum tenderness in the buttock at a
depth of approximately 3 to 5cm (fig 4). This point, just medial
to the musculotendinous junction, approximates the motor
point of the piriformis muscle. Electromyographic localization
of the piriformis muscle was done only when the muscle’s
uncertain location or depth required it (7 cases).
Consenting patients received injections at the initial diag-
nostic visit. Repeat injections were done rarely on patients
297
Fig 3. Delay of H-reflex on NCV tracing. Progressive, reversible
delay in H-reflex as a function of solid angle made by shin on plinth
in figure 1. The M wave could also vary as a result of subcutaneous
movement of the nerve. Abbreviation: AIF, adducted, internally
rotated, flexed. (Reprinted with permission.21)
judged to have insufficient improvement with physical therapy
and on the average were given 6 weeks after the first injection.
2. Concentrated physical therapy. Patients with piriformis
syndrome were treated according to a standard protocol (table
1, fig 1).
3. Follow-up after 6, 12, 24, 36, and 48 months. In-person
follow-up, by telephone or through the mail, sought answers to
items in the same questionnaire used at the initial visit, along
with the visual analog scale. Telephone interviews replaced the
visual analog scale with the question, “What percentage, if any,
Fig 4. Injection of 1.5mL containing 1mL of 2% lidocaine and 20mg
triamcinolone acetonide uses a 3.5-in, no. 23–25 spinal needle. The
injection site is located approximately one third of the distance
from the greater trochanter to the point of maximum tenderness in
the buttock (which is generally where the piriformis muscle and
sciatic nerve intersect). The nerve-muscle junction is not injected,
rather the muscle’s motor point receives the lidocaine and steroid.
The needle is inserted oriented toward the navel to a depth of 1 to
2 inches. Patients generally feel immediate relief. Unfortunately,
without physical therapy, the pain returns, at least partially, within
a few weeks.
Arch Phys Med Rehabil Vol 83, March 2002
298 PIRIFORMIS SYNDROME, Fishman

Table 1: Physical Therapy Protocol for Patients
With Piriformis Syndrome*
Place patient in contralateral decubitus and FAIR position.†
1. Ultrasound 2.0 to 2.5W/cm2 applied in broad strokes
longitudinally along the piriformis muscle from the conjoint
tendon to the lateral edge of the greater sciatic foramen for
10 to 14 minutes.†
2. Wipe off ultrasound gel.‡
3. Hot packs or cold spray at the same location for 10 minutes.
4. Stretch the piriformis muscle for 10 to 14 minutes by
applying manual pressure to the muscle’s inferior border,
being careful not to press downward, rather directing
pressure tangentially toward the ipsilateral shoulder.§
5. Myofascial release at lumbosacral paraspinal muscles.
6. McKenzie exercises.
7. Use lumbosacral corset when treating patient in the FAIR
position.㛳
Duration: 2 to 3 times weekly for 1 to 3 months.
* Patients usually require 2 to 3 months of biweekly therapy for 60%
to 70% improvement.
†
Because it is painful, patients often subtly shift to prone. This must
be avoided because it places the affected leg in abduction, not
adduction, greatly reducing the stretch placed on the piriformis
muscle.
‡
Cavitation is unreported in more than 20,000 treatments.
§
Unless explicitly stated, therapists may tend to knead or massage
the muscle, which is useless or worse. The muscle must be stretched
perpendicular to its fibers, in a plane that is tangent to the buttock at
the point of intersection of the piriformis muscle and the sciatic
nerve, but approximately 1 to 1.5-in deep to the buttock (ie, just
below the gluteus maximus).
㛳
It is particularly important to avoid inducing lumbar hypermobility
in patients with histories of laminectomy, fusion, or spondylolisthe-
sis.
have you improved or worsened since your first visit to our
offices?” This question was also asked in each written ques-
tionnaire.
RESULTS
Although configuration of the H-reflex tracing was geomet-
rically similar for the anatomic position and FAIR position in
each nerve tested, H-reflex amplitude often varied and tended
toward reduction in piriformis syndrome patients in the FAIR
position.
Maximum FAIR test values obtained in 88 normal persons
(asymptomatic controls) showed mean delay of .01ms beyond
values obtained in the anatomic position, with an SD equal to
.62ms (table 2). Patients meeting 2 of 3 clinical criteria for
piriformis syndrome had mean overall FAIR test H-reflex
prolongation of 3.39ms and 3.11ms for the posterior tibial and
peroneal nerves, respectively. These values were 5.45 and 5.02
SDs beyond the mean for the FAIR tests of asymptomatic
individuals. Patients’ legs failing to meet 2 of 3 clinical criteria
for piriformis syndrome had average FAIR test prolongation of
.83ms, 1.34 SDs beyond the normal mean (P ⬍ .001) (table 2).
A delay of 3 SDs on the FAIR test (1.86ms) was seen in the
posterior tibial and/or peroneal nerves of 468 of 537 limbs of
patients meeting 2 of 3 clinical piriformis syndrome criteria
and in 22 of 151 limbs of patients who did not meet 2 of the 3
criteria. This showed a sensitivity of .881 and a specificity of
.832. A delay of 2 SDs (1.24ms) was seen in 518 of 537
patients’ legs meeting 2 of 3 criteria and in 44 of 151 patients’
legs that did not. At 2 SDs, the FAIR test had a sensitivity of
.968 and specificity of .686.
Notably, the contralateral posterior tibial and peroneal FAIR
tests of FTP patients (in unilateral cases) showed mean delay of
.93ms and .51ms, respectively. This finding was significant in
comparison with normal limbs (P ⬍ .001). The contralateral
posterior tibial branch was more frequently and more severely
affected according to the FAIR test (table 2).
On average of 10.2 months’ follow-up after conservative
treatment (SD ⫽ 11.7), 79% of FTP responders (n ⫽ 665) had
achieved at least 50% improvement in painful symptoms, and
the 385 FTP patients with disability that responded to our
questionnaire showed a 62.8% reduction in disability related to
their original complaints of low back, buttock, and/or sciatic
pain. The FAIR test negative responders improved 54.8% (n ⫽
209), with a 54.8% reduction in disability (table 3).
To date, 43 FTP patients who either were unsatisfied with
the results of conservative treatment or did not choose to
undergo conservative therapy have been taken to surgery, with
28 (68.8%) achieving 50% relief or more. Anatomic variations
were reported in 6 of these 43 cases (13.95%), approximating

Table 2: Maximal FAIR Test

Mean FAIR Test No. of SDs† Above

Group (ms) (N)* SD Normal Mean P

Normals ⫺.01 (88) .62 0

Clinical PS posterior tibial 3.45 (345) 1.65 5.56 ⬍.001
Contralateral .93 (112) 1.69 1.50 ⬍.001§
Clinical PS peroneal 3.14 (320) 1.37 5.06 ⬍.001
Contralateral .51 (114) 1.78 .82 ⬍.001§
All clinical PS 3.34 (665) 1.84 5.28 ⬍.001
All contralateral to PS 1.27 (657) 3.0 2.11 ⬍.001㛳
Clinically negative PS‡ .83 (150) 1.17 1.34 ⬍.001
Contralateral .42 (39) 1.26 .68 ⬍.001

NOTE. The maximal FAIR test is the maximum prolongation of M ⫹ H that was replicably produced by the test in either the posterior tibial
or peroneal division of the sciatic nerve.
Abbreviation: PS, piriformis syndrome.
* Mean FAIR test prolongation of the H ⫹ M loop for all patients in that group (in ms).
†
Mean FAIR test prolongation of the H ⫹ M loop for all patients in that group, measured in the SDs observed in the FAIR test in normals (.62ms;
see top line of table).
‡
Symptomatic patients neither leg of which met the clinical criteria for piriformis syndrome.
§
Comparing proband leg to contralateral leg.
㛳
Comparing piriformis syndrome with nonpiriformis syndrome legs’ contralaterals.

Arch Phys Med Rehabil Vol 83, March 2002

 PIRIFORMIS SYNDROME, Fishman 299

Table 3: Weight, Age, Follow-Up Time, and Average Improvement of Piriformis, and Nonpiriformis Legs
After Conservative and Surgical Treatment

Average Symptomatic
Spinal Months Improvement
Surgery Follow-up 50% or More Average Pretreatment Posttreatment
Weight (lb) Age (y) Prestudy (n) (%) Improvement (%) Disability* (%) Disability (%)

Piriformis present 150.99 (671) 54 (671) 106 (616) 10.2 (665) 79.0% (665) 71.06% (665) 35.4% (385) 12.96% (385)
Piriformis absent 149.57 (332) 57 (331) 22 (316) 9.9 (339) 54.8% (209) 55% (339) 31% (209) 14.1% (209)
After surgery for
piriformis† 149.12 (43) 49 (43) 18 (41) 16 (43) 68.8% (43) 66% (43) 44.8% (43) 17.3% (34)

NOTE. Values are mean (n).

* Pre- and poststudy disability was estimated from subjective reports of disablement with regard to walking, sitting, standing, stair climbing,
transferring, work, recreation, traveling, sexual activity, and sleeping. (Maximal score for each of the 10 categories was 10%.)
†
Piriformis syndrome present by FAIR test; 60.2% of patients with a negative FAIR test improved 50% or more with surgery. Only 4 surgical
patients were neither FTP nor met 2 of 3 clinical piriformis syndrome criteria.

the percentage seen in cadaveric studies of the general popu- position should tighten the piriformis muscle sufficiently to
lation.25-27 compress the sciatic nerve’s fibers and transiently slow NCV at
The average FTP patient saw 6.55 clinicians for sciatica over that point. The difference between the anatomic and FAIR
a period of 6.2 years before entering the present study. positions’ H loops is enlarged by the fact that both the afferent
and efferent limbs of the H-reflex cross the piriformis muscle:
Surgical Corroboration of the FAIR Test any prolongation in latency from pathology at that point will be
One of the authors’ surgical series (SR) tested posterior amplified in the full H loop by a factor of 2.
tibialis M wave ⫹ H-reflex delay in the FAIR test before and Conservative treatment that lengthens the piriformis muscle,
after neurolysis surgery for piriformis syndrome, finding the reducing the compressive component at any given angle,
delay decreased 1.35ms (SD ⫽ 2.18) in 8 patients (table 4). should relieve the cause of piriformis syndrome. However, just
These cases were not used for other calculations in the present as the median nerve may be injured in carpal tunnel syndrome,
report. the sciatic nerve may sustain structural damage as a result of
piriformis syndrome and may require time to heal before the
Anatomic Consideration patient becomes symptom free. In extreme cases, axonotmesis
To study the relationship between the sciatic nerve and the or neuronotmesis may require considerable time or may be
piriformis muscle, 76 cadaveric legs were dissected and exam- irreversible, respectively. Nevertheless, 79% of patients im-
ined. Although detailed description is out of place, it is impor- proved at least 50%, with average follow-up time of 10.2
tant to note that of the 11 (14.5%) anomalous passages that months, suggesting mostly reversible levels of nerve damage in
were found, 10 (91%) were bilateral. most cases.
Our experience is that the posterior tibial or peroneal H-
Linking Clinical Presentation With Successful reflex or both are absent in approximately 5% of piriformis
Conservative Treatment syndrome probands. If a patient has sufficiently severe entrap-
Over 79% of patients (279/353) meeting 2 or more of the ment of either division of the sciatic nerve, then no H-reflex
clinical piriformis syndrome criteria improved by 50% or more may be evocable, making it impossible to obtain a FAIR test
(average, 71.7%). FTP patients meeting 2 or more of the for that division. Somewhat paradoxically, in these more severe
clinical piriformis syndrome criteria showed 50% or more cases, the less involved branch of the nerve will be the only
improvement in 256 of 308 cases (83.1%). Of the 109 patients branch capable of displaying the functional entrapment re-
meeting less than 2 clinical piriformis syndrome criteria, 82 vealed by the FAIR test.
(75.2%) improved 50% or more (average, 57.9%). Only 30 of Contralateral limbs might be expected to show no significant
45 (67.7%) of patients meeting fewer than 2 criteria and having H-loop prolongation in the FAIR test, but they are significantly
negative FAIR tests improved 50% or more (average, 52.5%)
(table 5). Table 4: Pre- and Postsurgical FAIR Tests
To determine whether other factors significantly influenced
outcome, we evaluated 71 characteristics recorded in extensive Preop Postop Test
Test Test Change SD of Clinical
interview and physical examination (table 6). This table may be Patient (ms) (ms) (ms) Change Outcome
used to determine the clinical likelihood of piriformis syn-
drome in a given patient with his/her individual symptoms and 1 NO 1.46 NO NO Excellent
signs. For example, 86.5% (283/337) of the patients with 2 2.29 1.04 1.25 2.00 Excellent
sciatica, pain on FAIR, tenderness in the region where the 3 3.54 2.08 1.46 2.34 Excellent
piriformis muscle intersects with the sciatic nerve, and greater 4 0.83 0.00 0.83 1.34 Good
amount of pain when sitting versus standing had positive FAIR 5 NO 1.25 NO NO Excellent
tests. In follow-up, 308 of these patients reported their out- 6 2.08 NO N NO Excellent
comes. With conservative therapy, 259 (84.1%) of these pa- 7 2.49 0.83 1.86 3.00 Excellent
tients reported 50% or more improvement. The discussion 8 NO 1.04 NO NO Poor
section examines this finding further. Totals 11.23 7.70 5.40 8.68
Mean 2.25 1.28 1.35* 2.17
DISCUSSION
Abbreviation: NO, not obtained.
If the piriformis muscle exerts significant compressive force * Differences in pre- and postsurgical FAIR tests when both are
on the sciatic nerve, then placing these patients in the FAIR present, equal to 2.17 SDs greater than the mean in normals.

Arch Phys Med Rehabil Vol 83, March 2002

300 PIRIFORMIS SYNDROME, Fishman

Table 5: Usefulness of the FAIR Test

Chi-Square,
FAIR Test Degree of
FAIR Test Positive Negative Freedom

2 or more PS criteria met 50% improvement 83.1% (256/308) 75.5% (77/102)

P .12 2.44,1df
Mean improvement 70.6% (308) 63.3% (102)
P .05
Fewer than 2 PS criteria met 50% improvement 81.3% (52/64) 66.7% (30/45)
P .13 2.88,1df
Mean improvement 66.7% (64) 52.5% (45)
P .028
All patients 50% improvement 82.8% (308/372) 74.8% (107/147)
P .015 5.97,1df
Mean improvement 69.6% (372) 59.9% (147)
P .003

NOTE. The % values are mean (n/N). Positive FAIR test indicates better outcome in patients with and without 2 of the 3 clinical characteristics
of piriformis syndrome.

more affected by the FAIR positioning than legs of entirely also possible that some persons may be predisposed to contract
sciatica-free individuals (fig 5, table 2). The most common piriformis syndrome, for example, persons with hereditary
causes of piriformis syndrome in our study were overuse (383/ neuropathic pressure palsy. Many myelin-related genetic vari-
876 ⫽ 43.5%) and trauma (164/892 ⫽ 18.3%), both of which ations, such as those associated with Charcot-Marie-Tooth
are often “shared” by both buttocks. At present, MRI studies disease,33 may predispose to piriformis syndrome and may
suggest bilateral effacement of the triangular fatty sciatic fo- explain some of the increased sensitivity of contralateral legs to
ramen in high-performance athletes such as football quarter- the FAIR test. These considerations might be promising ave-
backs and dedicated runners with piriformis syndrome (J. Zito, nues for research into piriformis syndrome and other functional
MD, oral communication, Aug 2000). entrapment syndromes.
Unfortunately, the present study did not include systematic It is unlikely that the anatomic variant is responsible for
analysis of comorbidities such as diabetes, common neuropa- causing piriformis syndrome because the peroneal division is
thies, or occupation or hip or lumbar procedures, despite their usually the aberrant division anatomically; yet, the posterior
probable bearing on the incidence of piriformis syndrome.3 It is tibial division is involved more frequently and more severely,
both on the affected and contralateral sides (see table 2). The
anatomic anomaly involves the peroneal division of the sciatic
Table 6: Characteristics of Patients With Successful Outcome nerve in more than 75% of recorded cases, and the peroneal
nerve is the only anomalous nerve in 67% of these (⬎50% of
P OR 95% CI
total cases). Yet, contralateral posterior tibial FAIR tests are the
Positive FAIR test ⬍.001 2.26 1.42–3.60 most positive tests in all analyses of the data (see table 3). In
SLRM* ⬍.001 2.83 1.69–4.73 addition, the anatomic abnormality is nearly invariably bilat-
Overuse .001 2.05 1.35–3.12 eral, yet piriformis syndrome cases are 90% unilateral. Finally,
Tender piriformis .003 1.97 1.27–3.06 no greater incidence of the anatomic variants has been encoun-
SLRM .007 2.03 1.22–3.40 tered at surgery than is seen in the general population.
Nonsciatic polyphasics .041 2.51 1.04–6.06
SLRM .048 3.00 1.01–8.92
Tenderness at piriformis .005 1.84 1.20–2.82
Sitting worse than standing .007 1.77 1.17–2.68
Male gender .017 1.67 1.10–2.50
Iliotibial band syndrome .028 2.87 1.12–7.38
SLRM .032 3.83 1.12–13.06
Injection .030 1.55 1.05–2.31
Nonsciatic PSW† .035 1.0 1.05–4.06
SLRM .013 3.29 1.28–8.42
L2 SLRM‡ .040 0.36‡ 0.13–0.95
Peroneal polyphasics .041 2.02 1.03–3.98
Clinical piriformis ⬍.001 3.32 2.25–4.90

NOTE. Values are from logistic model (univariate analyses) and

stepwise regression; receiver operating characteristic (ROC) ⫽ .692
(n ⫽ 651 patients, 514 successful outcomes).
Abbreviations: CI, confidence interval; OR, odds ratio; PSW, positive
sharp waves; SLRM, stepwise logistic regression model.
* ROC ⫽ .692. Fig 5. Frequency distribution of FAIR test values of patients with
†
Either radiculopathy or positive electromyographic findings. clinical piriformis syndrome, individual nerves of legs contralateral
‡
Electromyographic abnormalities at L2 probably imply a different to clinically piriformis syndrome legs, and normals. Distance on the
origin and are associated with a lower probability of 50% improve- vertical axis is a measure of frequency. The horizontal axis extends
ment. from ⴚ4.5 to ⴙ11.5ms.

Arch Phys Med Rehabil Vol 83, March 2002

 PIRIFORMIS SYNDROME, Fishman
CONCLUSION
One practical use of the FAIR test is to identify patients who
will improve from the physical therapy we described in table 1.
More than 81% of FTP patients with 2 of 3 piriformis syn-
drome criteria will improve 50% or more with conservative
therapy. Research using the FAIR test may also be useful in
determining which patients will not benefit from that therapy.
The stepwise linear regression model in table 6 reveals that L2
radiculopathy itself is very weakly related to positive outcome,
despite its conspicuous presence in many patients with the
syndrome. This finding suggests that factors other than the
radiculopathy at L2 mediate piriformis syndrome in these
cases. Possibly, radiculopathy at that level reduces iliopsoas
activity, another source of external hip rotation, raising the
potential for piriformis muscle overuse. But treating the piri-
formis muscle alone will not remedy this situation.
Acknowledgments: The authors thank Todd R. Olsen, PhD, Di-
rector of Clinical and Developmental Anatomy, Albert Einstein Col-
lege of Medicine, and Ann Nevins, Krista D. Oldfield, and Ilana Katz
for technical assistance.
References
1. Mixter WJ, Barr JS. Ruptures of the intervertebral disc with
involvement of the spinal canal. N Engl J Med 1934;211:210-1.
2. Kirkaldy-Willis WH, Hill RJ. A more precise diagnosis for low
back pain. Spine 1979;4;102-8.
3. Dawson DM, Hallett M, Millender, LH. Entrapment neuropathies.
2nd ed. Boston: Little Brown; 1990. p 270.
4. Hallin RP. Sciatic pain and the piriformis muscle. Postgrad Med
1983;74:69-72.
5. Pace JB, Nagle D. Piriformis syndrome. West J Med 1976;124;
435-9.
6. Sridhara CR, Izzo KL. Peroneal nerve entrapment syndrome. Arch
Phys Med Rehabil 1985;66;789-91.
7. Jensen MC, Brant-Zawadzki MN, Obuchowski N, Modic MT,
Malkasian D, Ross JS. Magnetic resonance imaging of the lumbar
spine in people without back pain. N Engl J Med 1994;331:69-73.
8. Robinson D. Piriformis muscle in relation to sciatic pain. Am J
Surg 1947;73;355-8.
9. Freiberg AH. Sciatic pain and its relief by operations on muscle
and nerve. Arch Surg 1937;34:337-49.
10. Agency for Health Care Policy and Research. Acute low back
problems in adults. Clinical Practice Guidelines No. 14. Rockville
(MD): Department of Health and Human Services; 1994. AHCPR
Publication No. 95-0642.
11. Parziale JR, Hudgins TH, Fishman LM. The piriformis syndrome.
Am J Orthop 1996;25:819-23.
12. Sinaki M, Merritt JL, Stillwell GK. Tension myalgia of the pelvic
floor. Mayo Clin Proc 1977;52:717-22.
301
13. Reichel B, Gaerisch F Jr. Piriformis syndrome. A contribution to
the differential diagnosis of lumbago and coccygodynia. Zentralbl
Neurochir 1988;49:178-84.
14. Synek VM. Short latency somatosensory evoked potentials in
patients with painful dysaesthesias in peripheral nerve lesions.
Pain 1987;29;49-58.
15. Steiner C, Staubs C, Ganon M, Buhlinger C. Piriformis syndrome:
pathogenesis, diagnosis, and treatment. J Am Osteopath Assoc
1987;87;318-23.
16. Solheim LF, Siewers P, Paus B. The piriformis muscle syndrome.
J Orthop Scand 1981;52:73-5.
17. Nainzadeh N, Lane ME. Somatosensory evoked potentials follow-
ing pudendal nerve stimulation as indicators of low sacral root
involvement in postlaminectomy patient. Arch Phys Med Rehabil
1987;68:170-2.
18. Synek VM. The piriformis syndrome; review and case presenta-
tion. Clin Exp Neurol 1987;23:31-7.
19. Fishman LM. Electrophysiological evidence of piriformis syn-
drome [abstract]. Arch Phys Med Rehabil 1987;68:670.
20. Fishman LM. Electrophysiological evidence of piriformis syn-
drome II [abstract]. Arch Phys Med Rehabil 1988;69:300.
21. Fishman LM, Zybert PA. Electrophysiological evidence of piri-
formis syndrome. Arch Phys Med Rehabil 1992;73:359-64.
22. Pace JB, Nagle D. Piriformis syndrome. West J Med 1976;124;
435-39.
23. Beaton LE, Anson BJ. Sciatic nerve and the piriformis muscle;
their interrelation a possible cause of coccygodynia. J Bone Joint
Surg 1938;20;686-8.
24. Freiberg AH, Vinke TH. Sciatica and the sacro-iliac joint. J Bone
Joint Surg 1938;16:1.
25. Yeoman W. Relation of arthritis of sacro-iliac joint to sciatica.
Lancet 1928;1119-22.
26. Pecina M. Contribution to the etiological explanation of the piri-
formis syndrome. Acta Anat 1979;105:181-7.
27. Gotlin R. Clinical correlation of an anatomical investigation into
piriformis syndrome. Proc N Y Soc Phys Med Rehabil 1991;
24(6):11.
28. Mizuguchi T. Division of the piriformis muscle for the treatment
of sciatica. Arch Surg 1976;111:719-22.
29. Hugon M. Methodology of the Hoffmann reflex in man. In:
Desmedt JE, editor. New developments in electromyography and
chemical neurophysiology. Vol 3. Basel: Karger; 1973. p 227-93.
30. Braddom RI, Johnson EW. Standardization of H reflex and diag-
nostic use in S1 radiculopathy. Arch Phys Med Rehabil 1974;55:
161-6.
31. Bell KR, Lehmann JF. Effect of cooling on H- and T-reflexes in
normal subjects. Arch Phys Med Rehabil 1987;68:490-3.
32. Halar EM, Hammond MC, Dirks S. Physical activity; its influence
on nerve conduction velocity. Arch Phys Med Rehabil 1985;66:
605-9.
33. Kamholz J, Awatramani R, Menichella D, Jiang H, Xu W, Shy M.
Regulation of myelin-specific gene expression: relevance to
CMT1. Ann N Y Acad Sci 1999;883:91-108.
Arch Phys Med Rehabil Vol 83, March 2002