DVT

A

D C L A

A r M

a e (

s a

w p

m C

l

t t m l p e

d p b p

P

t

H P M P P

S

a U

a t

s o

F

609

RTICLES

evelopment of a Clinical Prediction Rule for the Diagnosis of arpal Tunnel Syndrome

tCol Robert S. Wainner, PT, PhD, Julie M. Fritz, PT, PhD, James J. Irrgang, PT, PhD,

nthony Delitto, PT, PhD, COL Stephen Allison, PT, PhD, Michael L. Boninger, MD

C t d o a 1 a p s d a o b

s o s p c C a H p d m a t a h c

e i d ( a s C n p r e m O r D t l v i l

a o u

ABSTRACT. Wainner RS, Fritz JM, Irrgang JJ, Delitto A, llison S, Boninger ML. Development of a clinical prediction

ule for the diagnosis of carpal tunnel syndrome. Arch Phys ed Rehabil 2005;86:609-18.

Objectives: To develop a clinical prediction rule (CPR) and to ssess the reliability and diagnostic accuracy of individual clinical xamination items for the diagnosis of carpal tunnel syndrome CTS).

Design: Prospective diagnostic test study with blind compari- on to a reference criterion of a compatible clinical presentation nd abnormal electrophysiologic findings.

Setting: Multicenter medical center and community hospital ith patient referrals from ambulatory primary care and specialty ractice settings.

Participants: Eight-two consecutively referred patients (50% en; mean age, 45�12y) with suspected cervical radiculopathy or TS referred for electrophysiologic examination. Interventions: Not applicable. Main Outcome Measures: Sensitivity, specificity, and like-

ihood ratios. Results: The CPR identified in this study consisted of 1 ques-

ion (shaking hands for symptom relief), wrist-ratio index greater han .67, Symptom Severity Scale score greater than 1.9, reduced

edian sensory field of digit 1, and age greater than 45 years. The ikelihood ratio for the CPR was 18.3 when all 5 tests were ositive. Interrater reliability was acceptable for all but 2 clinical xamination items.

Conclusions: The CPR identified was more useful for the iagnosis of CTS than any single test item and resulted in posttest robability changes of up to 56%. Further investigation is required oth to validate the test-item cluster and to improve point-estimate recision.

Key Words: Carpal tunnel syndrome; Physical examination; rojections and predictions; Rehabilitation. © 2005 by American Congress of Rehabilitation Medicine and

he American Academy of Physical Medicine and Rehabilitation

From the US Army-Baylor Graduate Program in Physical Therapy, Fort Sam ouston, TX (Wainner); University of Utah, Salt Lake City, UT; Department of hysical Therapy, University of Pittsburgh, Pittsburgh, PA (Irrgang, Delitto); Rocky ountain University of Health Professions, Provo, UT (Allison); and Department of

hysical Medicine and Rehabilitation, University of Pittsburgh Medical Center, ittsburgh, PA (Boninger). Presented in part to the American Physical Therapy Association’s Combined

ections Meeting, 2001, San Antonio TX. Supported by the Orthopaedic Section of the American Physical Therapy Associ-

tion and the Foundation for Physical Therapy’s Clinical Research Center at the niversity of Pittsburgh. The opinions or assertions contained herein are the private views of the authors and

re not to be construed as official or as reflecting the views of the US Department of he Air Force, Department of the Army, or the Department of Defense.

No commercial party having a direct financial interest in the results of the research upporting this article has or will confer a benefit on the author(s) or on any rganization with which the author(s) is/are associated. Reprint requests to LtCol Robert S. Wainner, PT, PhD, 3151 Scott Rd, Ste 1303,

ort Sam Houston, TX 78234-6138, e-mail: [email protected].

c 0003-9993/05/8604-9146$30.00/0 doi:10.1016/j.apmr.2004.11.008

ARPAL TUNNEL SYNDROME (CTS) is an important cause of pain, neurologic symptoms, and functional limi-

ation of the hand. It is the most common nerve compression isorder of the upper extremity, with reported prevalence rates f 3% among women and 2% among men. Peak prevalence is mong women aged 55 years or older.1 As many as 2%2 to 5%3 of workers in high-risk industries are affected by CTS. In ddition to the frequency of occurrence, the percentages of atients experiencing treatment complications or recurrent ymptoms are sobering. Approximately 200,000 patients un- ergo surgical release of the volar carpal ligament annually,4

nd 7% to 20% of these surgically treated patients may fail to btain relief.5 Clearly CTS has a significant societal impact on oth an individual and economic level. Because the electrophysiologic examination for CTS is highly

pecific in a clinical setting, reasonably sensitive,6 and helps rule ut other neuropathic and comorbid conditions as a cause of hand igns and symptoms,7,8 both associated symptoms and electro- hysiologic findings (nerve conduction, needle electromyography) ontinue to be the mainstay for the laboratory diagnosis of TS.9-11 Furthermore, some third-party payers require electrodi- gnostic confirmation of CTS before compensating claims.12

owever, as with many laboratory diagnostic tests,13 electro- hysiologic procedures are not perfect, and there is continued ebate regarding their precise role in the diagnosis and manage- ent of CTS.11,12,14 Although other laboratory procedures, such

s computed tomography scan, magnetic resonance imaging, ul- rasonography, and magnetic resonance neurography may be vailable, the diagnostic accuracy and utility of these procedures ave yet to be determined, and they are therefore not considered linically useful at this time.15,16

The clinical examination, which consists of history, physical xamination, and manual test procedures, is once again increas- ngly relied on in this era of medical cost-cutting.17 Despite the emonstrated diagnostic value of a clinical prediction rule CPR) for many disorders,18-20 investigations of the precision nd accuracy of the clinical examination have lagged behind imilar studies of laboratory tests,21 as has the development of PRs. Although a definitive CPR for the diagnosis of CTS has ot been identified, a number of clinical examination items are urported to be useful, with provocative procedures and neu- ologic findings being the most frequently studied.22,23 How- ver, estimates of diagnostic test accuracy vary widely, and in ost cases, rater reliability for test measurements is unknown. ther tests that may be potentially useful, such as the wrist-

atio index24,25 and flick sign,26 have not been well studied. espite the demonstrated value of the history in many condi-

ions,27,28 useful data about its diagnostic value for CTS are imited. Considering the findings of 2 recent systematic re- iews,22,23 it is clear that high-quality studies of the value of ndividual elements of the clinical examination for CTS are acking, as is a methodologically sound CPR.

Although a CPR for the diagnosis of CTS is highly desirable nd would be extremely useful, the most accurate combination f clinical examination items for diagnosing CTS is currently nknown. A review by D’Arcy and McGee22 suggested that a

luster of test items would offer greater diagnostic accuracy for

Arch Phys Med Rehabil Vol 86, April 2005

t o C a m s p

g o e p t d f c t

P

D v B n 7 f t A w p h a a b w k m c i m d p ( a s j t

P

i m

p p l V u

S H t e

d c p w c s s b F f

S p C f “ “ h r r d f r

S

a n f p t r n o u o u l g w m m

m m v a b g l t s g i a e

g p p a l d C

610 CTS CLINICAL PREDICTION RULE, Wainner

A

he diagnosis of CTS. Unfortunately, these researchers and thers who have recommended the development of a CPR for TS did not offer any prospective data for analysis.22,23 Katz et l4 conducted original research and reported the diagnostic easurement properties for selected combinations of the Tinel

ign, Phalen test, and a hand diagram rating but did not incor- orate other clinical tests. Currently, no studies have followed recently published

uidelines29,30 for the development of a CPR for the diagnosis f CTS. A CPR considers multiple factors from the clinical xamination to improve a clinician’s diagnostic accuracy. The rimary purposes of our study were (1) to develop a CPR for he diagnosis of CTS and (2) to assess the reliability and iagnostic accuracy of individual clinical examination findings or the diagnosis of CTS using an electrophysiologic reference riterion. The secondary purpose of this study was to examine he influence of disease severity on diagnostic test accuracy.

METHODS

articipants and Design A total of 82 patients were enrolled in the study from

ecember 1998 to April 2000 from 4 medical facilities: Uni- ersity of Pittsburgh, Wilford Hall USAF Medical Center, rooke Army Medical Center, and Blanchfield Army Commu- ity Hospital. Consecutive patients between the ages of 18 and 0 years were recruited directly from the primary care clinic, rom the orthopedic department, and from patients referred to he electrophysiologic laboratories of participating facilities. ll had suspected cervical radiculopathy or CTS. Participants ere informed about the study by laboratory personnel. Only atients judged by the electrophysiologic laboratory provider to ave signs and symptoms compatible with cervical radiculop- thy or CTS were eligible to participate. The more symptom- tic limb was considered the involved limb for patients with ilateral symptoms. Patients with the following conditions ere disqualified from study participation: (1) systemic disease nown to cause a generalized peripheral neuropathy, (2) pri- ary complaint of bilateral radiating arm pain, (3) history of

onditions involving the affected upper extremity that resulted n a reduced level of function, (4) off work for more than 6 onths because of the condition, (5) history of surgical proce-

ures for pathologies giving rise to neck pain or for CTS, (6) revious needle electromyography and nerve conduction study NCS) testing of the symptomatic limb for cervical radiculop- thy and/or CTS, and (7) subjects receiving workers’ compen- ation or with pending litigation for their condition. All sub- ects gave informed consent for participation as approved by he respective facilities’ institutional review boards.

atient Self-Report Items Before undergoing a standardized electrophysiologic exam-

nation, each patient completed the following self-report instru- ents. Visual analog scale. Patients rated their hand and/or up-

er-extremity pain on a 10-cm visual analog scale (VAS). Each atient made 3 VAS ratings: worst pain over the last 24 hours, east pain over the last 24 hours, and current pain. Although the AS has been used extensively as an outcome measure,31-33 its se for diagnostic purposes has not been reported. Brigham and Women’s Hospital Hand Symptom Severity

cale and Function Status Scale. The Brigham and Women’s ospital Hand Symptom Severity Scale (SSS) and Function Sta-

us Scale (FSS) are condition-specific scales developed by Levine

t al.34 The SSS consists of 11 statement items related to 6 c

rch Phys Med Rehabil Vol 86, April 2005

omains thought critical for the evaluation of CTS. The FSS onsists of 8 items related to a variety of activities commonly erformed by a broad spectrum of patients (ie, young and elderly, orkers inside and outside the home). Both scales are scored by

alculating the mean of the individual items. A higher overall SSS core represents more severe symptoms. A higher overall FSS core represents greater disability. The psychometric properties of oth the SSS and FSS are acceptable.34-36 Although the SSS and SS have been used extensively as outcome measures, their use or diagnostic purposes has not been reported.

Hand diagram. The hand diagram developed by Katz and tirrat4,37,38 was used in this study and graded by classifying the atient as having “classic,” “probable,” “possible,” and “unlikely” TS based on the areas of the diagram that were marked. Results

rom the hand diagram were then dichotomized by classifying classic” and “probable” as a positive result and “possible” and unlikely” as a negative result.4 Intrarater and interrater reliability ave been reported as a percentage agreement of 91% and 84%, espectively.38 Prospective studies assessing the diagnostic accu- acy of the hand diagram alone and in combination with other iagnostic tests have reported sensitivities that range, respectively, rom .614 to .6438 and specificities that range from .714 to .73,38

espectively.

tandardized Electrophysiologic Examination Procedure A diagnosis of a compatible clinical presentation and

bnormal electrophysiologic findings as determined by a eurologist or physiatrist served as the reference criterion or CTS. All subjects underwent a standardized electro- hysiologic examination that was conducted by board-cer- ified personnel. NCSs consisted of palmar sensory39 and outine motor NCSs for both the median and ulnar erves.40-42 All median nerve NCS abnormalities were based n relative and absolute latency findings.39 Median and lnar nerve F-wave responses (minimum latency) were also btained. If abnormalities were observed in the median and lnar nerves of the same limb, nerves in the opposite upper imb and/or 1 lower limb were performed to rule out a eneralized peripheral neuropathy.6,43 All NCS procedures ere performed in accordance with guidelines for measure- ent, temperature, safety precautions, and electrode place- ent.42

After the NCS, needle electromyography of the following uscles was performed at rest and during contraction using a onopolar needle electrode: middle and lower cervical para-

ertebral muscles, deltoid, triceps brachii, extensor carpi radi- lis longus and brevis, flexor carpi radialis, abductor pollicis revis, and first dorsal interrosseous. In addition, electromyo- raphy and NCS providers sampled additional cervical and imb muscles when indicated by a patient’s clinical presenta- ion. For each muscle site sampled, the tester applied the tandard quadrant/level method for a total of 12 electromyo- raphic observations at each sampling site.42 Observations of nsertional activity, normal and abnormal spontaneous activity, nd motor unit firing frequency were recorded during needle lectromyography.42

All NCS and electromyography test results were judged and raded using published criteria.6,40,42,44 A board-certified hysiatrist or neurologist diagnosed all patients based on com- atible symptoms and electrophysiologic examination findings nd categorized them into 6 classifications according to estab- ished criteria (table 1).6,45-47 To examine the influence of isease severity on diagnostic test accuracy, all patients with TS (classifications 2– 4 and 6) were further categorized ac-

ording to the severity of their respective electromyographic

a s

f s d t

S

w s c i 2 B n c s

d q t

d a n s u s e t a t

s w C a m b v

i s t 2 e b

E

b e c r A t s

D

c t p n f f t i d i

F g

u i w P w ( t t v t c m a 9 m r l n r m w i

A ocity

611CTS CLINICAL PREDICTION RULE, Wainner

nd NCS findings using a modified classification system48 as hown in appendix 1.

Seven different electromyography and NCS providers per- ormed the NCSs, needle electromyography procedures, and ubsequent diagnostic classification of patients. At 1 center, 3 ifferent evoked potential technicians performed nerve conduc- ion procedures only.

tandardized Clinical Examination Procedure A standardized clinical examination consisting of 21 items

as performed by a physical therapist (examiner 1) after the tandardized electromyography and NCS examination was ompleted and after a 15- to 30-minute rest period. The exam- nation was repeated by a second physical therapist (examiner ) after a 10-minute rest period to assess interrater reliability. oth examiners were blinded to the patient’s suspected diag- osis, electromyography and NCS test results, and diagnostic lassification. Nine different physical therapists performed the tandardized clinical examination procedures.

History. All patients were asked 9 questions thought to be iagnostic of CTS; examiner 2 obtained responses to the same uestions 1 to 2 days later by telephone. The questions and heir respective response options are listed in appendix 2.

Conventional neurologic examination. Strength of the ab- uctor pollicis brevis muscle was tested as described by Kendall nd McCreary49 and was graded as markedly reduced, reduced, or ormal, compared with the uninvolved extremity. Pinprick sen- ory testing of the median nerve cutaneous distribution was done sing the end of a straightened paper clip. Palmar cutaneous ensation of the pads of the thumb and index and middle fingers ach were compared with the proximal cutaneous sensation of the henar eminence. The result of each sensory test was graded as bsent, reduced, normal, or hyperesthestic, in comparison to the henar eminence.

Wrist-ratio index. The wrist-ratio index described by John- on et al50 is purported to be an indicator of carpal canal volume, ith larger ratios (�.70) suggested to be a predisposing factor for TS. A single pair of sliding calipers was used to measure both nteroposterior (AP) and mediolateral (ML) wrist width in centi- eters. Caliper jaws were aligned with the distal wrist crease for

oth measurements. The wrist ratio index was computed by di- iding the AP wrist width by the ML wrist width. Provocative tests. The following provocative tests were used

n this study: carpal compression test (CCT), Phalen test, Tinel ign (parts A, B), and the upper-limb tension test (parts A, B). The ests along with their operational definitions are listed in appendix . The reliability and validity of both conventional neurologic xamination items and provocative tests used in this study have

Table 1: Descriptive Statistics for Stu

Diagnosis Based on EMG/NCV Male

Normal 17 Unilateral CTS 3 Bilateral CTS 6 CTS with ulnar neuropathy 0 Cervical radiculopathy 12 Cervical radiculopathy with peripheral neuropathy 3

bbreviations: EMG, electromyography; NCV, nerve conduction vel

een summarized and previously reported.22,23 i

xaminer Training A videotape of all clinical examination procedures and hand-

ooks detailing the performance of each clinical examination and lectrophysiologic measure were distributed to each participating enter before data collection. All examiners viewed the tape and ead the handbooks to familiarize themselves with the procedures. ll examiners practiced all clinical examination measures at least

wice. Using a pinch gauge, examiners practiced applying the pecified amount of compression force required for the CCT.

ata Analysis Reliability was computed for each neurologic and provocative

linical examination measure using dichotomized findings from he involved limb obtained by 1 rater pair that examined 50 atients. Sensation and muscle test results were dichotomized into ormal or abnormal findings because of low observed base rates or responses of “increased” or “markedly reduced.” Reliability or neurologic and provocative tests was assessed with a � statis- ic.51 Reliability for the wrist-ratio index was reported as an ntraclass correlation coefficient (ICC2,1) and corresponding stan- ard error (SE) of measurement.52 Ninety-five percent confidence ntervals (CIs) were calculated for all reliability coefficients.

The following qualitative interpretation for � described by leiss53 was used in this study: excellent, .75 or higher; fair to ood, .40 to .74; and poor, less than .40. The clinical examination results obtained by examiner 1 were

sed for all computations of diagnostic test accuracy. Only find- ngs from the involved limb were used. Contingency tables (2�2) ere used to calculate sensitivity and specificity for each test item. atients diagnosed with CTS (classifications 2, 3), including those ith concomitant cervical radiculopathy or ulnar neuropathy

classifications 4, 6), formed the disease-positive group, and pa- ients classified as normal or as cervical radiculopathy (classifica- ions 1, 4) served as the disease-negative group. When a zero cell alue was encountered, .5 was added to all cell values in the table o permit calculation of likelihood ratios and their 95% CI. Re- eiver operator characteristic (ROC) curves were used to deter- ine cutoff values for self-report, age, and wrist-ratio index vari-

bles.54 Positive and negative likelihood ratios and their associated 5% CIs were computed for all clinical examination items.55 For ultilevel response items (questions 1–3) and the CPR, likelihood

atios were reported for each response level.56 The positive like- ihood ratio was calculated as sensitivity/(1–specificity) and the egative likelihood ratio as (1–sensitivity)/specificity. Likelihood atios are convenient summary measures of diagnostic test perfor- ance and indicate by how much a given diagnostic test or CPR ill raise or lower the pretest probability of the target disorder of

nterest.56,57 The diagnostic accuracy of individual clinical exam-

ubjects by Diagnostic Classification

(n) Age (y)

(mean � SD)

Symptom Duration (d)

Female Median Range

23 40�11.7 176 21–1439 4 48.7�14.7 184.5 93–7220 9 45.7�10.6 262 31–5415 1 NA NA NA 2 51�7.4 74.5 21–2555 2 58.7�6.1 60 42–80

; NA, not applicable.

dy S

Sex

nation variables was considered useful if the positive likelihood

Arch Phys Med Rehabil Vol 86, April 2005

r o o 2 p

d o h i a s t c c g t V v fi fi o p f p g 1 1 a i e

e a p d l

t p t 3 m 4 c 4 i b a m h c a t c

R

n s d ( h

D

a B p r h i

A

612 CTS CLINICAL PREDICTION RULE, Wainner

A

atio was 2.0 or greater or if the negative likelihood ratio was .50 r less.58 Based on the prevalence or pretest probability for CTS f 34% in this sample, positive likelihood ratio values greater than .0 and negative likelihood ratio values less than .5 result in osttest probability changes of at least 14%.

A stepwise binary logistic regression model was used to etermine a CPR for the diagnosis of CTS.59 Other than age, nly variables with useful accuracy (�2.0, or negative likeli- ood ratio �.50)58 were entered into the model. Age was ncluded in the model because CTS is associated with increased ge.4,60 A backward stepwise selection procedure was used to elect variables, with P values of .15 to exit the model and .10 o enter it. The method of entry and liberal P values were hosen to prevent potentially useful variables from being ex- luded from the model.61 The Hosmer-Lemeshow summary oodness-of-fit statistic was used to assess the fit of the model o the data and test the hypothesis that the model fits the data.62

ariables retained by the regression model were used to de- elop a CPR for the diagnosis of CTS. The sensitivity, speci- city, and positive likelihood ratio for the number of positive ndings in the CPR were calculated as previously described for ther dichotomous variables. Because all patients had at least 1 ositive CPR finding, diagnostic values were only calculated or 2 or more CPR findings. According to Jaeschke et al,58

ositive likelihood ratio values between 2.0 and 5.0 would enerate small shifts in probability, values between 5.0 and 0.0 would generate moderate shifts, and values greater than 0.0 would generate large and often conclusive shifts in prob- bility. We chose to focus on the positive likelihood ratio nstead of the negative likelihood ratio because we were inter- sted in the diagnosis of CTS based on positive test results.

Patients with CTS were subclassified based on severity of lectromyography and NCS findings as mild/moderate CTS or s pronounced/severe CTS, and diagnostic accuracy was com- uted for both CTS subgroups in the same manner previously escribed for the entire CTS cohort. Each subgroup was ana-

Table 2: Reliability of

Variable

Question 1 (Most bothersome symptoms . . .) Question 2 (Where most bothersome . . .) Question 3 (Symptom behavior . . .) Question 4 (Hand fat/swollen . . .) Question 5 (Fumbling/dropping . . .) Question 6 (Entire limb numb . . .) Question 7 (Night symptoms wake . . .) Question 8 (Hand shaking improves . . .) Question 9 (Worse with hand use . . .) Median sensory field 1 Median sensory field 2 Median sensory field 3 MMT abductor pollicis brevis Upper-limb tension test A Upper-limb tension test B Tinel Part A Tinel Part B CCT Phalen test Wrist AP Wrist ML

bbreviations: MMT, manual muscle testing.

yzed independently of the other. (

rch Phys Med Rehabil Vol 86, April 2005

RESULTS The descriptive statistics for age and duration of symptoms of

he 82 patients (41 men, 41 women; mean age, 45�12y) who articipated in the study are listed by diagnostic classification in able 1. The prevalences of CTS and cervical radiculopathy were 4% (n�28) and 23% (n�19), respectively. The mean age and edian symptom duration of patients with CTS (mean age,

8.4�11.5y; symptom duration, 183.5d) did not differ signifi- antly (��.05) from patients without CTS (mean age, 3.2�11.7y; symptom duration, 123d). The right extremity was nvolved in 17 patients. Of the 28 patients with CTS, 15 had ilateral involvement. Based on severity of the electromyography nd NCS findings, 14 patients were classified as having mild/ oderate CTS (mean age, 46�12y) and 14 were classified as

aving pronounced/severe CTS (mean age, 50�11y). One patient, lassified as having cervical radiculopathy with concomitant CTS nd ulnar neuropathy at the elbow, dropped out of the study after he standardized electrophysiologic examination and was not in- luded in the analysis of test measurement properties.

eliability The 32 subjects not included in the reliability analysis did

ot differ from the rest of the sample with regard to age, SSS core, FSS score, or pain ratings (P�.05). Seventeen of the 19 ichotomous variables had � values of at least fair or better ��.40). The interrater reliability coefficients for items of istory and clinical examination are in table 2.

iagnostic Accuracy The sensitivity, specificity, and likelihood ratios for each vari-

ble, and their associated 95% CIs, are listed in tables 3 and 4. ecause the ROC curve for the VAS pain scale showed no otentially useful cutoff points, no further analyses of VAS pain atings were performed. The following 7 variables were found to ave useful diagnostic accuracy: question 3 (Which of the follow- ng best describes the behavior of your symptoms?), question 5

al Examination Items

(95% CI) ICC (95% CI) SE

(.55–.93) NA NA (.68–.96) NA NA (.35–.79) NA NA (.68–1.0) NA NA (.85–1.0) NA NA (.26–.81) NA NA (.60–1.0) NA NA (.75–1.0) NA NA (.49–.95) NA NA (.23–.73) NA NA (.25–.75) NA NA (.12–.68) NA NA (.00–.80) NA NA (.51–1.0) NA NA (.65–1.0) NA NA (.21–.72) NA NA (.10–.60) NA NA (.58–.96) NA NA (.59–1.0) NA NA NA .77 (.62–.87) 2.1mm NA .86 (.75–.92) 2.1mm

Clinic

�

.74

.82

.57

.85

.95

.53

.83

.90

.72

.48

.50

.40

.39

.76

.83

.47

.35

.77

.79

Do you have trouble with fumbling or dropping objects from

y w m s

C

r w u i N

c c s g 1 t m o 4 h p a

N

N d p

N A d rat *

613CTS CLINICAL PREDICTION RULE, Wainner

our affected hand?), question 8 (Do your symptoms improve ith moving, “shaking,” or positioning your wrist or hands?), edian sensory field 1, wrist-ratio index greater than .67, an SSS

core greater than 1.9, and an FSS score greater than 2.5.

linical Prediction Rule The 7 variables listed above and age were entered into the

egression model as potential predictors for CTS. After list- ise deletion, a total of 78 subjects (26 CTS, 52 control) were sed in the analysis. The results of the Hosmer-Lemeshow test ndicated that the model fit the data (P�.65) and the agelkerke R2 equaled .56. The following 5 test variables were

Table 4: Validity of Physical Exa

Test Item Sensitivity (95% CI)

Age .64 (.47–.82) Median sensory field 1 (thumb) .65 (.47–.84) Median sensory field 2 (index finger) .52 (.32–.72) Median sensory field 3 (middle finger) .44 (.26–.63) MMT abductor pollicis brevis .19 (.04–.34) Wrist-ratio index �.67 .93 (.83–1.0) Hand diagram (classic or probable) .75 (.58–.92) SSS score �1.9 .89 (.77–1.0) FSS score �2.5 .37 (.19–.55) Upper-limb tension test A .75 (.58–.92) Upper-limb tension test B .64 (.45–.83) Tinel Part A .41 (.22–.59) Tinel Part B .48 (.29–.67) CCT .64 (.45–.83) Phalen test .77 (.61–.93)

OTE. Useful likelihood ratios appear in bold.

Table 5: CPR for

Criteria for a Positive Test Sensitivity (95% CI)

�2 positive tests .98 (.14–1.0) �3 positive tests .98 (.14–.10) �4 positive tests .77 (.61–.93) All 5 tests positive .18 (.03–.31)

OTE. Five tests are included in the rule: (1) question 8 (hand shakin

Table 3: Validity o

Question* Sensitivity (95% CI)

1 (Most bothersome symptoms . . .) .04 (�.04–.11) 2 (Where most bothersome . . .) .35 (.16–.53) 3 (Symptom behavior . . .) .23 (.07–.39) 4 (Hand fat/swollen . . .) .38 (.20–.57) 5 (Fumbling/dropping . . .) .73 (.56–.90) 6 (Entire limb numb . . .) .38 (.20–.57) 7 (Night symptoms wake . . .) .73 (.56–.90) 8 (Hand shaking improves . . .) .81 (.66–.96) 9 (Worse with hand use . . .) .77 (.61–.93)

OTE. Useful likelihood ratios are in bold. bbreviations: NLR, negative likelihood ratio; PLR, positive likelihoo See wording of entire question in appendix 2.

iminished sensation in median sensory field 1 (thumb); and (5) age �45 y robability values are based on a pretest probability of 34%.

hosen by the model and are therefore considered the best CTS linical prediction rule: a single question (shaking hands for ymptom relief), wrist-ratio index greater than .67, SSS score reater than 1.9, diminished sensation in median sensory field (thumb), and age greater than 45. The 5 test variables and

heir diagnostic properties according to the number of abnor- alities required for a positive test are listed in table 5. For 4

r more positive test findings, the positive likelihood ratio was .6; when all 5 test findings were positive, the positive likeli- ood ratio was 18.3. Twenty CTS patients had 4 or more ositive test results and 4 CTS patients had positive results for ll 5 tests. All 81 patients had at least 1 positive CPR test result.

tion and Self-Report Measures

Specificity (95% CI)

NLR (95% CI)

PLR (95% CI)

.59 (.47–.72) .60 (.35–1.0) 1.58 (.46–2.4)

.70 (.47–.84) .49 (.28–.46) 2.2 (1.3–3.6)

.67 (.32–.72) .72 (.52–1.1) 1.6 (.92–2.7)

.74 (.26–.63) .75 (.86–1.1) 1.7 (.92–3.2)

.89 (.81–.9) .91 (.74–1.1) 1.7 (.58–5.2)

.26 (.14–.38) .29 (.07–1.2) 1.3 (1.0–1.5)

.13 (.04–.22) 1.9 (.72–5.1) .86 (.67–1.1)

.36 (.23–.49) .31 (.10–.96) 1.4 (1.1–1.8)

.81 (.71–.92) .77 (.56–1.1) 2.0 (.95–4.2)

.13 (.04–.22) 1.9 (.72–5.1) .86 (.67–1.1)

.30 (.17–.42) 1.2 (.62–2.4) .91 (.65–1.3)

.58 (.45–.72) 1.0 (.69–1.5) .98 (.56–1.7)

.67 (.54–.79) .78 (.52–1.2) 1.4 (.84–2.5)

.30 (.17–.42) 1.2 (.62–2.4) .91 (.65–1.3)

.40 (.26–.53) .58 (.27–1.3) 1.3 (.94–1.7)

iagnosis of CTS

pecificity (95% CI)

Likelihood Ratio

(95% CI)

Posttest Probability of CTS (%)

(.23–.23) 1.1 (1.0–1.3) 44 (.40–.67) 2.1 (1.6–2.8) 52 (.73–.93) 4.6 (2.5–8.7) 70 (.97–1.0) 18.3 (1.0–328.3) 90

roves symptoms); (2) wrist-ratio index �.67; (3) SSS score �1.9; (4)

torical Questions

Specificity (95% CI)

NLR (95% CI)

PLR (95% CI)

.91 (.83–.98) 1.1 (.94–1.2) .42 (.05–3.4)

.40 (.27–.54) 1.6 (1.1–2.5) .58 (.33–1.0)

.89 (.81–.97) .87 (.69–1.1) 2.1 (.74–5.8)

.63 (.50–.76) .98 (.68–1.4) 1.0 (.57–1.9)

.57 (.44–.71) .47 (.24–.92) 1.7 (1.2–2.5)

.80 (.69–.90) .77 (.55–1.1) 1.9 (.92–3.9)

.31 (.19–.44) .86 (.41–1.8) 1.1 (.79–1.4)

.57 (.43–.70) .34 (.15–.77) 1.9 (1.3–2.7)

.37 (.24–.50) .62 (.28–1.4) 1.2 (.91–1.6)

io.

mina

the D

S

.14

.54

.83

.99

g imp

f His

ears. Useful likelihood ratios appear in bold. The associated posttest

Arch Phys Med Rehabil Vol 86, April 2005

D

m a p w i q o s s

w d a C c c S ( b r

l C C c s ( h m i p c d T o p d A t

s f s ( s T e d a w c c a

n d w t F i u t t t 2 t r h a m c d n t r p i o l p p r

N *

614 CTS CLINICAL PREDICTION RULE, Wainner

A

iagnostic Accuracy for CTS Subclassifications The diagnostic properties of clinical examination items for ild/moderate and pronounced/severe CTS subclassifications

re listed in table 6 along with the results for all 28 CTS atients. There were 12 items with useful likelihood ratios hen subclassification analyses were considered. A wrist-ratio

ndex greater than .67, an SSS score greater than 1.9, and the uestion concerning shaking hands for symptom relief were the nly test items with likelihood ratios that remained useful and omewhat consistent when comparing the results of the entire ample with both subclassification categories (table 6).

DISCUSSION A CPR has been defined as “a tool used by clinicians to assist

ith medical decision making that provides either a probability of isease or outcome or suggest a diagnostic or therapeutic course of ction.”29(p488) Our study is the first to propose a clinically useful PR for the diagnosis of CTS that meets all major methodologic riteria for a level IV CPR.29,30 The rule derived in our study omprises age, a question (shaking hands for symptom relief), SS score, wrist-ratio index, and median sensory field 1 deficit see table 5 ). Of the test items D’Arcy and McGee22 reported to e helpful based on their review, only diminished sensation was epresented in our CPR.

The diagnostic usefulness of a CPR is best expressed by ikelihood ratio statistics. The positive likelihood ratio of the PR expresses the change in odds favoring the presence of TS, depending on the number of positive findings. Before any linical testing, the probability that a patient had CTS in our ample was 34%. With a CPR criterion of all 5 tests positive positive likelihood ratio�18.3), the probability of a patient aving CTS increases to 90% (fig 1). A CPR criterion of 4 or ore positive tests yields a positive likelihood ratio of 4.6 and

ncreases the probability of CTS to 70%. In contrast, the ositive likelihood ratio reported by Katz et al4 for various ombinations of the Phalen test, Tinel sign, and the hand iagram did not exceed 3.5 (calculated from reported data).4

he CPR derived in our study represents a reference criterion f provider diagnosis based on both a compatible clinical resentation and electromyography and NCS reference stan- ard, the shortcomings of which have already been discussed. lthough a better criterion standard may be found in the future,

his is arguably the best currently available.9

Table 6: Validity of Test Item

Test Item

Sensitivity

Total Mild/

Moderate Seve

Age �45y .64 .50 .79 Question 3 (Symptom behavior . . .) .23 .14 .33 Question 5 (Fumbling/dropping . . .) .73 .64 .83 Question 6 (Entire limb numb . . .) .38 .50 .25 Question 7 (Night symptoms wake . . .) .73 .86 .58 Question 8 (Hand shaking improves . . .) .81 .86 .75 Question 9 (Worse with hand use . . .) .77 .86 .67 SSS score �1.9 .89 .93 .85 FSS score �2.5 .37 .50 .23 Median sensory field 1 .65 .50 .83 Abductor pollicis brevis weakness .19 .14 .25 Wrist-ratio index �.67 .93 .93 .92

OTE. Useful likelihood ratios appear in bold. Specificity remained constant for all 3 comparisons, because the contro

rch Phys Med Rehabil Vol 86, April 2005

Our univariate results for the Phalen test, Tinel sign, and night ymptoms were very similar to those reported by Katz.4 Results or the hand diagram and age were also similar for sensitivity and pecificity, respectively. However, specificity of the hand diagram specificity�.13 in our study vs specificity�.714–.7338) and sen- itivity of age (sensitivity�.64 in our study vs .804,38) differed. hese differences may be attributable to spectrum of disease, xclusion criteria, recruitment source (primary care and orthope- ic specialty clinic), and use of an ROC curve–specified cutoff for ge 45 years or older (vs �40y4). Patients with systemic diseases ere excluded in our study, but some of our CTS patients had

oncomitant peripheral mononeuropathies, as did patients in the omparison group, of whom 49% had normal electromyography nd NCS findings.

We assessed several test items that have been infrequently or ot previously reported that appeared to be useful for the iagnosis of CTS. These include the SSS score, FSS score, rist-ratio index, and selected historical questions. All these

est items had useful likelihood ratios. Although the SSS and SS are disease-specific scales commonly used as evaluative

nstruments for patients with CTS,35,36 they appear to have seful diagnostic properties as well (negative likelihood ra- io�.21, positive likelihood ratio�.2.0, respectively). Ques- ions related to behavior of symptoms (question 3) and func- ional limitations (question 5) had a positive likelihood ratio of .1 and a negative likelihood ratio of .47, respectively. Ques- ion 8, which related to the flick sign, had a negative likelihood atio of .34. The flick sign, in which a patient shaking his/her ands relieves symptoms, has been reported by Pryse-Phillips26

s being highly sensitive and specific. The wrist-ratio index had uch higher sensitivity and lower specificity in our study

ompared with others,24,25,50 but we used an ROC curve to etermine an optimal cutoff value, whereas other studies did ot. The smallest negative likelihood ratio for any test item in he study belonged to the wrist-ratio index (negative likelihood atio�.29), which indicates it may be useful for screening urposes. Results for the Phalen test, Tinel sign, and the CCT n this study were comparable to those reported in other studies f higher methodologic quality, as were the results for neuro- ogic tests.22 The exception was the sensitivity of the abductor ollicis brevis manual muscle test, which was low and may be artially explained by disease spectrum and a low index of eliability. Similar to other studies, we found diminished sen-

r Subclassifications of CTS

Specificity*

NLR PLR

Total Mild/

Moderate Severe Total Mild/

Moderate Severe

.59 .60 .84 .36 1.6 1.2 1.9

.89 .87 .96 .75 2.1 1.3 3

.57 .47 .62 .29 1.7 1.5 2.0

.80 .77 .63 .94 1.9 2.5 1.2

.31 .86 .45 1.3 1.1 1.3 .85

.57 .34 .25 .44 1.9 2.0 1.7

.37 .62 .39 .90 1.2 1.4 1.1

.36 .31 .20 .43 1.4 1.5 1.3

.81 .77 .61 .94 2.0 2.7 1.3

.70 .49 .71 .24 2.2 1.7 2.8

.89 .91 .96 .84 1.7 1.3 2.4

.26 .29 .28 .30 1.3 1.3 1.3

s fo

re

l group did not change.

s m t p b c d

u v i m t c

b t t w g n w t i

r a t m f t i a

c n s i s m

s t A t o F s

d V a p b

H

F C

615CTS CLINICAL PREDICTION RULE, Wainner

ation to have useful diagnostic properties.25,63 However, our ethod of testing sensation— by comparing the pad of the

humb with the base of the thenar eminence— has not been reviously reported. We did not include sensory threshold tests, ecause time restraints and training limit their widespread linical use, nor did we include body mass index as a variable, espite its being a known risk factor for CTS.60,64

The diagnostic values reported for many test items considered seful for the diagnosis of CTS, particularly provocative tests, ary widely. Their diagnostic value is at best mediocre, and nterpretation of research findings is confounded by a number of ethodologic flaws contained in previous studies. These limita-

ions include lack of a challenging control group, poor or unac-

ig 1. Pretest (34%) to posttest (90%) probability change using the PR with positive results from all 5 tests.

w

ias.22,23,57 However, much of the variation of results between hese studies can probably be explained by spectrum bias. The 3 est items (SSS, question 8, wrist-ratio index) that showed some- hat consistent and useful likelihood ratios for the total CTS roup and both CTS subgroup classifications (see table 6) have ot been well studied or studied at all. Apparently, these items ere not affected by the spectrum of disease in our study. All 3

ests were included in the CPR and therefore provide unique nformation in the diagnosis of CTS.

A common problem with previous studies is that most do not eport the reliability of the measures they used.22,23 Laupacis et l29 found that, of the 30 reports dealing with CPRs that met heir inclusion criteria, only 1 reported indices of reliability for easures used in their study. The reliability of measurements

or all items as operationally defined in our study was greater han fair, except the Tinel B and muscle test of the APB, ndicating that most common clinical examination items have cceptable reliability for clinical use.

Several factors support the generalizability of our results, in- luding the large number of electrophysiologic laboratory person- el, clinical examiners, multicenter involvement, recruitment of ubjects from a primary care setting, and inclusion of a challeng- ng control group. The study was limited by our relatively small ample size, which resulted in wide 95% CIs of our point-esti- ates and an increased potential for regression model misfit. Our results suggest that future research assessing the SSS

core, wrist-ratio index, and flick sign question is likely to find hat these clinical items have useful diagnostic properties. dditional studies of provocative test items such as the Phalen

est and Tinel sign, whose values have yet to be clearly dem- nstrated, are likely to yield more of the same unfruitful results. uture studies should assess and report the influence of disease pectrum on test item performance.

CONCLUSIONS This study represents a level IV CPR and the first step in

eveloping a clinically sensible CPR for the diagnosis of CTS. alidation is required before this rule can be applied clinically,

nd it has yet to be determined whether patients presenting in rimary care, orthopedic, and other specialty clinic settings will e better off for undergoing these tests.

Acknowledgments: The lead author acknowledges LtCol oward Gill, MD, for the use of his laboratory and gracious assistance

ith this project as well as LtCol Manuel Domenich and Maj Monte

eptable test operational definition, and test and diagnosis review Wilson for their assistance with data collection and clinical support.

APPENDIX 1: CTS SEVERITY CLASSIFICATION SYSTEM

Severity Level Description

Mild Abnormal median sensory latency. All other sensory and motor NCS parameters normal. Moderate Abnormal sensory and distal motor latency. Sensory nerve action potentials (SNAP) or compound

nerve action potentials (CNAP) amplitudes may be diminished but are greater than or equal to 50% of normative values. Motor nerve conduction velocity (NCV) is normal but mild slowing of forearm NCV may be present (�45m/s).

Pronounced Abnormal sensory and distal motor latency. SNAP or CNAP amplitudes at 50% or less of normative values. Compound muscle action potential (CMAP) amplitude may be diminished but at 50% or more of normative values. Mild slowing of forearm NCV may be present (�45m/s) and spontaneous activity may be noted on electromyography examination.

Severe Absent SNAP or CNAP, abnormal distal motor latency, CMAP amplitude less than 50% of normative values or absent. Mild (�45m/s) slowing of forearm NCV may be present and electromyographic abnormalities are present.

Arch Phys Med Rehabil Vol 86, April 2005

616 CTS CLINICAL PREDICTION RULE, Wainner

A

APPENDIX 2: QUESTIONS OF HISTORY AND PROVOCATIVE TESTS

Questions of History and Provocative Tests Responses

1. Which of the following symptoms are most bothersome for you? Pain, numbness & tingling vs loss of feeling 2. Where are your symptoms most bothersome? Neck, shoulder or shoulder blade, arm vs

hands and/or fingers 3. Which of the following best describes the behavior of your

symptoms? Intermittent, variable (comes and goes) vs

constant 4. Does your affected hand feel “fat” or “swollen”? Yes or No 5. Do you have trouble with fumbling or dropping objects from

your affected hand? Yes or No

6. Does your entire affected limb and/or hand feel numb? Yes or No 7. Do your symptoms wake you during the night? Yes or No 8. Do your symptoms improve with moving, “shaking,” or

positioning your wrist or hands? Yes or No

9. Are your symptoms made worse when performing tasks that require a lot of grasping or hand and/or finger use?

Yes or No

Provocative Tests Positive Test Criteria

● Tinel part A was performed and interpreted as a test of neural regeneration as described by Tinel.65 With the patient sitting, the elbow flexed 0°–30°, and the forearm in a supinated position, the patient’s wrist and hand were supported in a neutral position. A tendon reflex hammer positioned �6 inches above the wrist was allowed to fall 4–6 times over the median nerve located between the tendons of the flexor carpi radialis and the palmaris longus at the proximal wrist crease. Part B was performed and interpreted as a provocative measure used to reproduce the patient’s symptoms.4 The patient was positioned and the test was performed as for part A above, except that the examiner brought the hammer down with mild to moderate force in an attempt to elicit symptoms.

Part A: Nonpainful tingling sensation radiating distally along the course of the nerve.

Part B: Discomfort or pain at the wrist or radiating distally along the course of the nerve that is related to the patient’s condition.

● Carpal compression test (CCT). With the patient sitting, the elbow flexed 0°–30°, and the forearm in a supinated position, the patient’s wrist and hand were supported in a neutral position. The examiner placed both thumbs over the transverse carpal ligament and applied approximately 6 pounds of pressure with each thumb. The pressure was maintained for a maximum of 30 seconds.66 The patient was questioned with regard to symptoms at 15-second intervals during the 30-second period.

Reproduction of the symptoms in the cutaneous distribution of the median nerve that is related to the patient’s condition.

● Phalen test. With the patient sitting, the elbow flexed 0°–30°, and the forearm in a supinated position, the patient’s wrist and hand were supported in a neutral position. The examiner placed the patient’s wrist in a position of maximal flexion for a maximum of 60 seconds.67 The patient was questioned with regard to symptoms at 15-second intervals during the 60-second period.68

Reproduction or exacerbation of paresthesias or anesthesia in the cutaneous distribution of the median nerve in the hand.67

● Upper-limb tension test. Part A was performed similar to Elvey’s description.69 With the patient supine, the examiner sequentially introduced the following movements to the symptomatic upper extremity: (1) scapular depression, (2) shoulder abduction, (3) forearm supination, wrist and finger extension, (4) shoulder lateral rotation, (5) elbow extension, and (6) contralateral then ipsilateral cervical side-bending. Part B. With the patient supine and the shoulder abducted 30°, the examiner sequentially introduced (1) scapular depression, (2) shoulder medial rotation, (3) full elbow extension, (4) wrist and finger flexion, and (5) contralateral then ipsilateral cervical side- bending. In both parts the patient was questioned regarding

Any 1 of the following: (1) patient’s symptoms reproduced (2) side-to-side differences (�10°) in elbow

extension (part A) or wrist flexion (part B) on completion of all motion sequences

(3) symptomatic limb side: contralateral neck side-bending increased symptoms or ipsilateral side-bending decreased symptoms.

symptom reproduction throughout the maneuver.

rch Phys Med Rehabil Vol 86, April 2005

1

1

1

1

1

1

1

1

1

1

2

2

2

2

2

2

2

2

2

2

3

3

3

3

3

3

3

3

3

3

4

4

4

4

4

4

617CTS CLINICAL PREDICTION RULE, Wainner

References 1. Atroshi I, Gummersson C, Johnsson R, Ornstein E, Ranstam J,

Ingmar R. Prevalence of carpal tunnel syndrome in a general population. JAMA 1999;282:153-8.

2. Erdil M, Maurer K, Dickerson O. The burden of cumulative trauma disorders. In: Erdil M, Dickerson OB, editors. Cumulative trauma disorders. New York: Van Nostrand Reinhold; 1997. p 3-4.

3. Masear VR, Hayes JM, Hyde AG. An industrial cause of carpal tunnel syndrome. J Hand Surg Am 1986;11:222-7.

4. Katz J, Larson M, Sabra A, et al. The carpal tunnel syndrome: diagnostic utility of the history and physical examination findings. Ann Intern Med 1990;112:321-7.

5. Mackinnon S. Secondary carpal tunnel surgery. Neurosurg Clin N Am 1991;2:75-91.

6. Jablecki CK, Andary MT, Floeter MK, et al. Practice parameter: electrodiagnostic studies in carpal tunnel syndrome. Report of the American Association of Electrodiagnostic Medicine, American Academy of Neurology, and the American Academy of Physical Medicine and Rehabilitation. Neurology 2002;58:1589-92.

7. Hurst L, Weissberg D, Carroll R. The relationship of double crush to carpal tunnel syndrome (an analysis of 1,000 cases of carpal tunnel syndrome). J Hand Surg [Br] 1985;10:202-4.

8. Morgan G, Wilbourn A. Cervical radiculopathy and coexisting distal entrapment neuropathies: double crush syndromes? Neurol- ogy 1989;50:78-83.

9. Jarvik JG, Yuen E. Diagnosis of carpal tunnel syndrome: electro- diagnostic and magnetic resonance imaging evaluation. Neurosurg Clin N Am 2001;12:241-53.

0. Jablecki C, Andary M, So Y, Wilkins D, Williams F. Literature review of the usefulness of nerve conduction studies and electro- myography for the evaluation of patients with carpal tunnel syn- drome. Muscle Nerve 1993;16:1392-414.

1. Rempel D, Evanoff B, Amadio PC, et al. Consensus criteria for the classification of carpal tunnel syndrome in epidemiologic studies. Am J Public Health 1998;88:1447-51.

2. Concannon M, Gainor B, Petroski G, Puckett C. The predictive value of electrodiagnostic studies in carpal tunnel syndrome. Plast Reconstr Surg 1997;100:1452-8.

3. Deyo R, Haselkorn J, Hoffman R, Kent D. Designing studies of diagnostic tests for low back pain or radiculopathy. Spine 1994; 19(18 Suppl):2057S-65S.

4. Gunnarsson L, Amilon A, Hellstrand P, Leissner P, Philipson L. The diagnosis of carpal tunnel syndrome. Sensitivity and speci- ficity of some clinical and electrophysiological tests. J Hand Surg [Br] 1997;22:34-7.

5. Jarvik JG, Yuen E, Haynor DR, et al. MR nerve imaging in a prospective cohort of patients with suspected carpal tunnel syn- drome. Neurology 2002;58:1597-602.

6. Cantatore FP, Dell’Accio F, Lapadula G. Carpal tunnel syndrome: a review. Clin Rheumatol 1997;16:596-603.

7. Simel D, Dujardin B. The clinical examination: an agenda to make it more rational. JAMA 1997;277:572-4.

8. Stiell IG, Greenberg GH, Wells GA, et al. Prospective validation of a decision rule for the use of radiography in acute knee injuries. JAMA 1996;275:611-5.

9. Stiell IG, Greenberg GH, McKnight RD, et al. Decision rules for the use of radiography in acute ankle injuries. Refinement and prospective validation. JAMA 1993;269:1127-32.

0. Buchsbaum DG, Buchanan RG, Centor RM, Schnoll SH, Lawton MJ. Screening for alcohol abuse using CAGE scores and likeli- hood ratios. Ann Intern Med 1991;115:774-7.

1. Sackett D, Rennie D. The science of the art of the clinical examination. JAMA 1992;267:2650-2.

2. D’Arcy CA, McGee S. The rational clinical examination. Does this patient have carpal tunnel syndrome? JAMA 2000;283:

3110-7.

3. Massy-Westropp N, Grimmer K, Bain G. A systematic review of the clinical diagnostic tests for carpal tunnel syndrome. J Hand Surg [Am] 2000;25:120-7.

4. Radecki P. A gender specific wrist ratio and the likelihood of a median nerve abnormality at the carpal tunnel. Am J Phys Med Rehabil 1994;73:157-62.

5. Kuhlman K, Hennessey W. Sensitivity and specificity of carpal tunnel syndrome signs. Am J Phys Med Rehabil 1997;76:451-7.

6. Pryse-Phillips W. Validation of a diagnostic sign in carpal tunnel syndrome. J Neurol 1984;47:870-2.

7. Sandler G. The importance of the history in the medical clinic and the cost of unnecessary tests. Am Heart J 1980;100:928-31.

8. Hampton JR, Harrison MJ, Mitchell JR, Prichard JS, Seymour C. Relative contributions of history-taking, physical examination, and laboratory investigation to diagnosis and management of medical outpatients. BMJ 1975;2:486-9.

9. Laupacis A, Sekar N, Stiell IG. Clinical prediction rules. A review and suggested modifications of methodological standards. JAMA 1997;277:488-94.

0. McGinn TG, Guyatt GH, Wyer PC, Naylor CD, Stiell IG, Rich- ardson WS. Users’ guides to the medical literature. XXII: how to use articles about clinical decision rules. Evidence-Based Medi- cine Working Group. JAMA 2000;284:79-84.

1. Downie W, Leatham P, Rhind V. Studies with pain rating scales. Ann Rheum Dis 1978;37:378-81.

2. Langley G, Sheps S. The visual analogue scale: its use in pain measurement. Rheumatol Int 1985;5:145-8.

3. Wilkie D, Lovejoy N, Dodd M, Tesler M. Cancer pain intensity measurement; concurrent validity of three tools—finger dyname- ter, pain intensity number scale, visual analogue scale. Hosp J 1990;6:1-13.

4. Levine D, Simmons B, Koris M, et al. A self-administered ques- tionnaire for the assessment of severity of symptoms and func- tional status in carpal tunnel syndrome. J Bone Joint Surg Am 1993;75:1585-92.

5. Amadio P, Silverstein M, Ilstrup D, Schleck C, Jensen L. Out- come assessment for carpal tunnel surgery: the relative respon- siveness of generic, arthritis-specific, disease specific, and physi- cal examination measures. J Bone Joint Surg Am 1996;21:338-46.

6. Katz J, Gelberman R, Wright E, Lew R, Liang M. Responsiveness of self-reported and objective measures of disease severity in carpal tunnel syndrome. Med Care 1994;32:1127-33.

7. Katz J, Stirrat C. A self-administered hand diagram for the diag- nosis of carpal tunnel syndrome. J Hand Surg 1990;15A:360-3.

8. Katz J, Stirrat C, Larson M, Fossel A, Eaton H, Liang M. A self-administered hand symptom diagram for the diagnosis and epidemiologic study of carpal tunnel syndrome. J Hand Surg [Am] 1990;17:1494-8.

9. Jackson DA, Clifford JC. Electrodiagnosis of mild carpal tunnel syndrome. Arch Phys Med Rehabil 1989;70:199-204.

0. DiBenedetto M, Mitz M, Klingbeil G, Davidoff D. New criteria for sensory nerve conduction especially useful in diagnosing car- pal tunnel syndrome. Arch Phys Med Rehabil 1986;67:586-9.

1. Kinura I, Ayyar DR. The carpal tunnel syndrome: electrophysio- logical aspects of 639 symptomatic extremities. Electromyogr Clin Neurophysiol 1985;25:151-64.

2. Dumitru D. Electrodiagnostic medicine. Philadelphia: Hanley & Belfus; 1995. p 111-60.

3. Jablecki C. Practice parameter for carpal tunnel syndrome. Neu- rology 1993;43:2406-9.

4. Kimura J. The carpal tunnel syndrome: localization of conduction abnormalities within the distal segment of the median nerve. Brain 1979;102:619-35.

5. Burke D, Burke M, Bell R, Stewart G, Mehdi R, Kim H. Subjec- tive swelling: a new sign for carpal tunnel syndrome. Am J Phys

Med Rehabil 1999;78:504-8.

Arch Phys Med Rehabil Vol 86, April 2005

4

4

4

4

5

5

5

5

5

5

5

5

5

5

6

6

6

6

6

6

6

6

6

6

618 CTS CLINICAL PREDICTION RULE, Wainner

A

6. Practice parameter for electrodiagnostic studies in ulnar neurop- athy at the elbow: summary statement. American Association of Electrodiagnostic Medicine, American Academy of Neurology, American Academy of Physical Medicine and Rehabilitation. Muscle Nerve 1999;22:408-11.

7. Committee AQA. Practice parameter for needle EMG evaluation of patients with suspected cervical radiculopathy: summary state- ment. Muscle Nerve 1999;22:S209-11.

8. Padua L, LoMonaco M, Aulisa L, et al. Surgical prognosis in carpal tunnel syndrome: usefulness of a preoperative neurophys- iological assessment. Acta Neurol Scand 1996;94:343-6.

9. Kendall F, McCreary E. Muscles, testing and function. Baltimore: Williams & Wilkins; 1983.

0. Johnson E, Gatens T, Poindexter D, Bowers D. Wrist dimensions: correlation with median sensory latencies. Arch Phys Med Rehabil 1983;64:556-7.

1. Shrout P, Spitzer R, Fleiss J. Quantification of agreement in psychiatric diagnosis revisited. Arch Gen Psychiatry 1987;44: 172-7.

2. Shrout P, Fleiss J. Intraclass correlations: uses in assessing rater reliability. Psychol Bull 1979;86:420-8.

3. Fleiss J. The measurement of interrater agreement: statistical methods for rates and proportions. New York: John Wiley & Sons; 1981. p 213-35.

4. Hanley J, McNeil B. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982; 143:29-36.

5. Simel D, Samsa G, Matchar D. Likelihood ratios with confidence: sample size estimation for diagnostic test studies. J Clin Epide- miol 1991;44:763-70.

6. Dujardin B, Van den Ende J, Gompel A, Unger J, Van der S. Likelihood ratios: a real improvement for clinical decision mak- ing? Eur J Epidemiol 1994;10:29-36.

7. Jaeschke R, Guyatt G, Sackett D. Users’ guides to the medical literature. III. How to use an article about a diagnostic test. A. Are the results of the study valid? The Evidence-Based Medicine

Working Group. JAMA 1994;271:389-91.

rch Phys Med Rehabil Vol 86, April 2005

8. Jaeschke R, Guyatt G, Gordon C, Sackett D. Users’ guides to the medical literature. III. How to use an article about a diagnostic test. B. What are the results and will they help me in caring for my patients? The Evidence-Based Medicine Working Group. JAMA 1994;271:703-7.

9. Holleman D, Simel D. Quantitative assessments from the clinical examination: how should clinicians integrate the numerous re- sults? J Intern Med 1997;12:165-71.

0. Werner RA, Franzblau A, Albers JW, Armstrong TJ. Influence of body mass index and work activity on the prevalence of median mononeuropathy at the wrist. Occup Environ Med 1997;54:268- 71.

1. Freedman D. A note on screening regression equations. Am Stat- istician 1983;37:152-5.

2. Hosmer D, Lemeshow S. Applied logistic regression. New York: John Wiley & Sons; 1989. p 142-4.

3. Golding D, Rose D, Selvarajah K. Clinical tests for carpal tunnel syndrome. Br J Rheumatol 1986;25:388-90.

4. Werner RA, Albers JW, Franzblau A, Armstrong TJ. The rela- tionship between body mass index and the diagnosis of carpal tunnel syndrome. Muscle Nerve 1994;17:632-6.

5. Tinel J. Le signe du fourmillement dans les lesions des nerfs peripheriques. Presse Med 1915;47:388-99.

6. Durkan J. A new diagnostic test for carpal tunnel syndrome. J Bone Joint Surg Am 1991;13:535-8.

7. Phalen G. The carpal tunnel syndrome: seventeen years experi- ence in diagnosis and treatment of six hundred and fifty four hands. J Bone Joint Surg Am 1966;48:211-28.

8. Kaplan S, Glickel S, Eaton R. Predictive factors in the non- surgical treatment of carpal tunnel syndrome. J Hand Surg [Br] 1990;15:106-8.

9. Elvey R. The investigation of arm pain: signs of adverse responses to the physical examination of the brachial plexus and related tissues. In: Boyling JD, Palastanga N, editors. Grieve’s modern manual therapy. New York: Churchill Livingstone; 1994. p 577-

85.