Development of a Clinical PredictionRule for Guiding Treatment of aSubgroup of Patients With Neck Pain:Use of Thoracic Spine Manipulation,Exercise, and Patient EducationJoshua A Cleland, John D Childs, Julie M Fritz, Julie M Whitman, Sarah L Eberhart
Background and PurposeTo date, no studies have investigated the predictive validity of variables from theinitial examination to identify patients with neck pain who are likely to benefit fromthoracic spine thrust manipulation. The purpose of this study was to develop aclinical prediction rule (CPR) to identify patients with neck pain who are likely toexperience early success from thoracic spine thrust manipulation.
SubjectsThis was a prospective, cohort study of patients with mechanical neck pain whowere referred for physical therapy.
MethodsSubjects underwent a standardized examination and then a series of thoracic spinethrust manipulation techniques. They were classified as having experienced a suc-cessful outcome at the second and third sessions based on their perceived recovery.Potential predictor variables were entered into a stepwise logistic regression modelto determine the most accurate set of variables for prediction of treatment success.
ResultsData for 78 subjects were included in the data analysis, of which 42 had a successfuloutcome. A CPR with 6 variables was identified. If 3 of the 6 variables (positivelikelihood ratio�5.5) were present, the chance of experiencing a successful outcomeimproved from 54% to 86%.
Discussion and ConclusionThe CPR provides the ability to a priori identify patients with neck pain who arelikely to experience early success with thoracic spine thrust manipulation. However,future studies are necessary to validate the rule.
JA Cleland, PT, PhD, OCS,FAAOMPT, is Assistant Professor,Department of Physical Therapy,Franklin Pierce College, 5 ChenellDr, Concord, NH 03301 (USA);Research Coordinator, Rehabilita-tion Services, Concord Hospital,Concord, NH; and Faculty, Man-ual Therapy Fellowship Program,Regis University, Denver, Colo.Address all correspondence to DrCleland at: [email protected].
JD Childs, PT, PhD, MBA, OCS,FAAOMPT, is Assistant Professorand Director of Research, USArmy-Baylor University DoctoralProgram in Physical Therapy, SanAntonio, Tex.
JM Fritz, PT, PhD, ATC, is AssociateProfessor, Division of PhysicalTherapy, University of Utah, SaltLake City, Utah, and Clinical Out-comes Research Scientist, Inter-mountain Health Care, Salt LakeCity, Utah.
JM Whitman, PT, DSc, OCS,FAAOMPT, Assistant Faculty,Department of Physical Therapy,Regis University.
SL Eberhart, PT, MPT, is PhysicalTherapist and Clinical II, Rehabili-tation Services, Concord Hospital.
[Cleland JA, Childs JD, Fritz JM,et al. Development of a clinicalprediction rule for guiding treat-ment of a subgroup of patientswith neck pain: use of thoracicspine manipulation, exercise, andpatient education. Phys Ther.2007;87:9–23.]
January 2007 Volume 87 Number 1 Physical Therapy f 9
Neck pain is a common occur-rence with a lifetime inci-dence ranging from 22% to
70%.1,2 Over a third of patients willdevelop chronic symptoms lastingmore than 6 months in duration,3
representing a serious health con-cern.4 Over 50% of patients withneck pain are referred for physicaltherapy and comprise approximately25% of all patients seeking physicaltherapy services.5,6 Although cervi-cal spine thrust manipulation hasbeen advocated as an interventionappropriate for the care of patientswith neck disorders, clinicians mustconsider the benefits relative to thepotential risks, especially vertebralartery insult.7,8 The lack of evidencefor premanipulative screening toidentify which patients may be atrisk has caused some authors to sug-gest that serious complications,although rare, are unpredictable andthat the potential benefits of cervicalspine thrust manipulation do notoutweigh the inherent risks.8,9
Clinical experience and preliminaryevidence suggest that thoracic spinethrust manipulation may be useful inthe management of patients withneck pain.10 The biomechanical linkbetween the cervical spine and thethoracic spine suggest that distur-bances in joint mobility in the tho-racic spine may serve as an under-lying contributor to the developmentof neck disorders. In addition, it hasbeen demonstrated that a significantassociation exists between decreasedmobility of the thoracic spine and thepresence of patient-reported com-plaints associated with neck pain.11
With inherently lower risk of seriouscomplications, thoracic spine thrustmanipulation might be a suitable alter-native, or supplement, to cervicalspine thrust manipulation. Perhapsthis accounts for why some cliniciansperform thoracic spine thrust manipu-lation rather than cervical spine thrustmanipulation at much higher rates inpatients with neck pain.12
Although widely used in patientswith neck pain, there are currentlyno decision-making strategies toidentify individual patients withneck pain who are most likely tobenefit from thoracic spine thrustmanipulation.10,13,14 Classificationprovides a means of breaking downa larger entity into more homoge-nous subgroups of patients based onexamination data.15,16 Moreover,classification is most helpful forphysical therapists when it is basedon signs and symptoms that matchinterventions to the subgroup ofpatients most likely to benefit fromthem (ie, treatment-based classifica-tion).17
Clinical prediction rules (CPRs) con-sist of combinations of variablesobtained from self-report measuresand the historical and clinical exam-inations and assist with subgroupingpatients into specific classifications.Recently, CPRs have been shown tobe useful in classifying patients withlow back pain (LBP) who are likelyto benefit from a particular treat-ment approach.18–20 Although atreatment-based classification systemfor the management of neck pain hasrecently been proposed,21 no studieshave investigated the predictivevalidity of variables from the initialexamination to identify patients withneck pain who are likely to benefitfrom thoracic spine thrust manipula-tion. Therefore, the purpose of thisstudy was to develop a CPR to iden-tify patients with neck pain who arelikely to benefit from thoracic spinethrust manipulation based on a ref-erence standard of patient-reportedimprovement.
Materials and MethodsWe conducted a prospective cohortstudy of consecutive patients withmechanical neck pain who werereferred for physical therapy at oneclinical site (Rehabilitation Services,Concord, Hospital, Concord, NH).Inclusion criteria required subjects
to be between the ages of 18 and 60years, with a primary complaint ofneck pain with or without unilateralupper-extremity symptoms and abaseline Neck Disability Index (NDI)score of 10% or greater. Exclusioncriteria were as follows: identifica-tion of any medical “red flags” sug-gestive of a nonmusculoskeletal eti-ology of symptoms, history of awhiplash injury within 6 weeks ofthe examination, a diagnosis of cer-vical spinal stenosis, evidence of anycentral nervous system involvement,or signs consistent with nerve rootcompression (at least 2 of the follow-ing had to be diminished to be con-sidered nerve root involvement:myotomal strength, sensation, orreflexes). All subjects reviewed andsigned a consent form approved bythe Institutional Review Board atConcord Hospital, Concord, NH.
TherapistsFour physical therapists participatedin the examination and treatment ofsubjects in this study. All therapistsunderwent a standardized trainingregimen, which included studyinga manual of standard procedureswith the operational definitions andvideo clips demonstrating eachexamination and treatment proce-dure used in this study. All partici-pating therapists then underwent a1-hour training session in which theypracticed the examination and treat-ment techniques to ensure that allstudy procedures were performed ina standardized fashion. Prior to par-ticipating in data collection, thera-pists were visually observed by theprincipal investigator as being ableto successfully perform all examina-tion and treatment procedures on apatient with neck pain. Participatingtherapists had a mean of 12.3 years(SD�10.0, range�3–23) of clinicalexperience.
Examination ProceduresSubjects provided demographicinformation and completed a variety
Clinical Prediction Rule for Patients With Neck Pain
10 f Physical Therapy Volume 87 Number 1 January 2007
of self-report measures, followed bya standardized history and physicalexamination at baseline. Self-reportmeasures included a body diagramto assess the distribution of symp-toms,22 a numeric pain rating scale(NPRS),23 the NDI,24 and the Fear-Avoidance Beliefs Questionnaire(FABQ). Subjects recorded the loca-tion of their symptoms on the bodydiagram to determine the most distalextent of their symptoms.22
The FABQ was used to quantify thesubjects’ beliefs about the influenceof work and activity on their neckpain.25 The FABQ consists of a work(FABQW) subscale and a physicalactivity (FABQPA) subscale, both ofwhich have been shown to exhibit ahigh level of test-retest reliability.26
The FABQW subscale has beenshown to exhibit predictive validityin the identification of patients withLBP who are likely to respond tospinal manipulation,19,20 but the pre-dictive validity for patients withneck pain is unknown. For thisstudy, the FABQ was modified toreplace the word “back” with“neck.”27 Finally, the NDI was usedto capture the subjects’ perceivedlevel of disability as a result of theirneck pain.24
The historical examination includedquestions regarding the mode ofonset, nature and location of symp-toms, aggravating and relieving fac-tors, and prior history of neck pain.The physical examination beganwith a neurological screen28 fol-lowed by postural assessment.29 Theoperational definitions for posturalassessment used in this study wereas follows: a subject was identified ashaving a forward head if the sub-ject’s external auditory meatus wasanteriorly deviated (anterior to thelumbar spine),29 and the shoulderswere identified as protracted if theacromion was noted to be anteriorlydeviated (anterior to the lumbarspine).29 The examiners were
instructed to identify the contour ofthe spine for the following groups ofsegments: C7 through T2 (cervico-thoracic junction), T3 through T5,and T6 through T10. Each group wasrecorded as normal (no deviation),as having excessive kyphosis, or ashaving diminished kyphosis.30 Ex-cessive kyphosis was defined as anincrease in the convexity, anddiminished kyphosis was defined asa flattening of the convexity of thethoracic spine (at each segmentalgroup).30
The clinician next measured cervicalrange of motion and symptomresponse31 and assessed the length28
and strength (force-generatingcapacity)29 of the muscles of theupper quarter and endurance of thedeep neck flexor muscles.32 Theamount of motion and symptomresponse were recorded for bothsegmental mobility testing28 of thecervical spine and spring testing33 ofthe cervical spine and thoracic spine(C2–T9).
The physical examination culmi-nated with a number of special teststypically performed in the examina-tion of patients with neck pain,including the Spurling test,34 Roostest,35 Neck Distraction Test,36 andUpper Limb Neurodynamic Test.37
Specific operational definitions foreach test and criteria defining a pos-itive test are presented in the Appen-dix.
Of the 80 subjects who wereenrolled in the study, 22 underwenta second examination by an addi-tional therapist who was blind to thefindings of the first clinician. The 22subjects who underwent a secondevaluation were selected based onthe availability of a second clinicianto perform the examination. Thereliability analysis was performed toevaluate the reliability of the identi-fied potential predictor variables.
TreatmentAll subjects received a standardizedtreatment regimen, regardless of theresults of the clinical examination,because treatment outcome servedas the reference criterion.38 Eachsubject received 3 different thrustmanipulation techniques directed atthe thoracic spine during each ses-sion: a seated “distraction” manipu-lation, a supine upper thoracic spinemanipulation, and a middle thoracicspine manipulation. The first manip-ulation performed was the “distrac-tion” manipulation. The subject wasseated, and the therapist placed hisor her upper chest at the level ofthe subject’s middle thoracic spineand grasped the subject’s elbows. Ahigh-velocity distraction thrust wasperformed in an upward direction(Fig. 1).
The upper thoracic spine manipula-tion was performed with the subjectpositioned supine and clasping hisor her hands across the base of theneck. The therapist used his or hermanipulative hand to stabilize theinferior vertebra of the motion seg-ment (the therapist was instructed totarget between T1 and T4 with thistechnique) and used his or her bodyto push down through the subject’sarms to perform a high-velocity, low-amplitude thrust (Fig. 2).
The middle thoracic spine manipula-tion was performed in the identicalfashion as the upper thoracic tech-nique, except the subject graspedthe opposite shoulder with his orher hands and the therapist wasinstructed to target between T5 andT8 with the thrust (Fig. 3). Immedi-ately after performing a manipula-tion, the treating therapist recordedwhether a “pop” was heard. Regard-less of the presence of a “pop,” thetherapist again performed the identi-cal manipulation technique. There-fore, each subject received 6 manip-ulations per treatment session.
Clinical Prediction Rule for Patients With Neck Pain
January 2007 Volume 87 Number 1 Physical Therapy f 11
Following the manipulation tech-niques, all subjects were instructedin a cervical-range-of-motion (CROM)exercise (10 repetitions performed3–4 times daily)39 (Fig. 4) and wereadvised to maintain their usual activ-ity within the limits of pain. TheCROM exercise consisted of the sub-ject placing his or her fingers overthe manubrium and placing his orher chin on the fingers. The subjectwas instructed to rotate to one sideas far as possible and return to neu-tral. This was performed alternatelyto both sides within pain tolerance.The first treatment session wasalways performed on the day of theinitial examination, and the subjectwas scheduled for a follow-up visitwithin 2 to 4 days.
The global rating of change (GROC)served as the reference criterion forestablishing a successful outcome.The GROC is a 15-point global ratingscale ranging from �7 (“a very greatdeal worse”) to 0 (“about the same”)to �7 (“a very great deal better”).40
Intermittent descriptors of worsen-ing or improving are assigned valuesfrom �1 to �7 and �1 to �7,respectively.41,42 It has beenreported that scores of �4 and �5are indicative of moderate changesin patient status and scores of �6and �7 indicate large changes inpatient status.40 It was determined apriori that subjects who rated theirperceived recovery on the GROC as“a very great deal better,” “a greatdeal better,” or “quite a bit better”(ie, a score of �5 or greater) at thesecond session were categorized ashaving a successful outcome, andtheir participation in the study wascomplete.
A high threshold for determining asuccessful outcome was establishedto maximize the likelihood that theclinical outcome was attributable tomeaningful improvements in symp-toms due to the intervention asopposed to the passage of time. Sub-
Figure 1.Seated thoracic spine distraction thrust manipulation used in this study. The therapistuses his or her sternum as a fulcrum on the subject’s middle thoracic spine and appliesa high-velocity distraction thrust in an upward direction.
Figure 2.Supine upper thoracic spine thrust manipulation technique used in this study. Thetherapist uses his or her body to push down through the subject’s arms to perform ahigh-velocity, low-amplitude thrust directed in the direction of the arrow toward T1through T4.
Clinical Prediction Rule for Patients With Neck Pain
12 f Physical Therapy Volume 87 Number 1 January 2007
jects whose scores on the GROC didnot exceed the �5 cutoff at the sec-ond session again received the thrustmanipulations as in the first treat-ment and were scheduled for afollow-up within 2 to 4 days. At thestart of the third session, subjectsagain completed the GROC andwere judged to have a successfuloutcome based on the previouslydescribed criterion. If the subjectsstill did not meet the threshold forsuccess, they were categorized ashaving a nonsuccessful outcome. Atthis point, their participation in the
study was complete, and furthertreatment was administered at thediscretion of their therapist.
In contrast to other studies identify-ing predictor variables for treatmentsuccess in patients with LBP,18,19 weelected to use perceived recoveryrather than a perceived level of dis-ability to determine success as theGROC. This decision is based on thefact that the GROC is considered tobe a valid reference standard foridentifying clinically importantchange.43–45 Perceived recovery also
was used as the reference criterionbecause the NDI has been criticizedfor not adequately capturing low lev-els of disability and for not beingresponsive to small, but clinicallyimportant, changes in patients withlow levels of initial disability.46 Inaddition, a measure of success ratebased on patient’s perceived recov-ery has previously been used in trialsof patients with neck pain and hasbeen shown to be responsive tochanges with physical therapy man-agement programs.42,46
Data AnalysisSubjects were dichotomized as hav-ing a successful outcome or as hav-ing a nonsuccessful outcome basedon the treatment response, as indi-cated on the GROC. The mean NDIand NPRS change scores (and 95%confidence intervals [CIs]) werecalculated for the both groups andanalyzed using an independent ttest to determine whether a differ-ence existed between groups. Indi-vidual variables from self-reportmeasures, the history, and thephysical examination were testedfor univariate relationship with theGROC reference criterion usingindependent-samples t tests forcontinuous variables and chi-square tests for categorical vari-ables. Variables with a significancelevel of P�.10 were retained aspotential prediction variables.47
This significance level was selectedto increase the likelihood that nopotential predictor variables wouldbe overlooked.
Figure 3.Supine middle thoracic spine thrust manipulation technique used in this study. Thetherapist uses his or her body to push down through the subject’s arms to perform ahigh-velocity, low-amplitude thrust directed in the direction of the arrow toward T5through T8.
Figure 4.Active-range-of-motion exercise performed by subjects in the study.
Clinical Prediction Rule for Patients With Neck Pain
January 2007 Volume 87 Number 1 Physical Therapy f 13
For continuous variables with a sig-nificant univariate relationship, sen-sitivity and specificity values werecalculated for all possible cutoffpoints and then plotted as a receiveroperating characteristic (ROC)curve.48 The point on the curve near-est the upper left-hand corner repre-sented the value with the best diag-nostic accuracy, and this point wasselected as the cutoff defining a pos-itive test.48 Sensitivity, specificity,and positive likelihood ratios (LRs)were calculated for potential predic-tor variables. Potential predictorvariables were entered into a step-wise logistic regression model todetermine the most accurate set ofvariables for prediction of treatmentsuccess. A significance level of .10was required for removal from theequation to minimize the likelihoodof excluding potentially helpful vari-ables.47 Variables retained in theregression model were obtained asthe CPR for classifying patients withneck pain who are likely to benefitfrom thoracic spine thrust manipula-tion, exercise, and patient educationfor this sample of subjects.
We further analyzed the data to deter-mine whether weighting individualpredictors according to the relativesize of the beta coefficients increasesthe prognostic accuracy of the model.Weights were calculated by taking thebeta coefficient for each variable inthe final model and dividing it by thelowest beta coefficient and thenrounding to the nearest integer.49
Once the weight was formulated, anROC curve was used to identify thecutoff value that represented the bestdiagnostic accuracy for the point-based system.48 Sensitivity, specificity,and positive LRs as well as corre-sponding 95% confidence intervalswere calculated for the cutoff pointthat maximized the diagnostic utilityof the weighting system.
The Cohen kappa (�)50 was used tocalculate the interrater reliability of
categorical data with only 2 possibleresponse options from the patienthistory and clinical examination. Aweighted kappa51 was used to calcu-late the reliability of categorical datawith 3 response options such asintersegmental mobility assessmenttechniques as well as the symptomresponse (increased pain, decreasedpain, no change). Intraclass correla-tion coefficients (ICC[2,1]) and the95% CIs were calculated to deter-mine the interrater reliability forcontinuous variables.52
Therapists were characterized byyears of experience to determine theeffect of experience on patient out-comes. Therapists were dichoto-mized as having 3 or fewer years ofexperience or more than 3 years ofexperience. Only one treating clini-cian had less than 3 years of experi-ence. The percentage of successfuloutcomes for each group (�3 yearsof experience or �3 years of expe-rience) was calculated and com-pared using a chi-square test of inde-pendence. The NDI change scoresalso were calculated and were com-pared between groups using inde-pendent t tests.
ResultsBetween March 2004 and September2005, 80 subjects were recruited forthe study. The total number of sub-jects screened, reasons for ineligibil-ity, and dropouts are shown in Fig-ure 5. Two subjects failed to returnfor the second treatment session,and their data were excluded fromthe analysis. Subject demographicsand initial baseline variables fromthe patient history and self-reportmeasures for the entire sample aswell as for the successful outcomeand nonsuccessful outcome groupsare presented in Table 1. Baselineclinical examination variables for theentire sample and for the successfuloutcome and nonsuccessful out-come groups are shown in Table 2for categorical data and in Table 3
for continuous data. Forty-two sub-jects were categorized as havingachieved a successful outcome, and36 subjects were categorized as hav-ing achieved a nonsuccessful out-come. Twenty-three subjects (55%)were classified as having a successfuloutcome after the initial treatment,and 19 subjects (45%) were classi-fied as having a successful outcomeafter 2 sessions. The mean numberof days between visit 1 and visit 2was 2.3 (SD�0.7) and 2.3 (SD�0.6)(P�.53) for the successful outcomeand nonsuccessful outcome groups,respectively. The mean number ofdays between visit 1 and visit 3 was6.3 (SD�1.2) and 6.2 (SD�1.2)(P�.99) for the successful outcomeand nonsuccessful outcome groupsrespectively. Analysis of NPRS andNDI change scores revealed that thesuccessful outcome group experi-enced significantly greater improve-ments (P�.001) in pain (NPRSchange score�2.2, 95% CI�1.4–2.9) and disability (NDI changescore�18.6%, 95% CI�13.3–25.0)over the nonsuccessful outcomegroup.
The 10 potential predictor variables(Tab. 4) that exhibited a significancelevel of less than .10 were enteredinto the logistic regression. The cut-off values determined by the ROCcurves were 11.5 for the FABQPAsubscale, 9.5 for the FABQW sub-scale, 30 days since the onset ofsymptoms, and 30 degrees of cervi-cal extension. In addition, the num-ber of prior episodes of neck painwas dichotomized into �3 episodesor �3 episodes. Accuracy statisticsfor all 10 variables (and 95% CIs) areshown in Table 4. The positive LRsranged from 1.1 to 6.4, with thestrongest predictor being symptomduration of �30 days.
The following 6 variables wereretained in the final regression mod-el: symptom duration of �30 days,no symptoms distal to the shoulder,
Clinical Prediction Rule for Patients With Neck Pain
14 f Physical Therapy Volume 87 Number 1 January 2007
Figure 5.Flow diagram showing subject recruitment and retention.
Clinical Prediction Rule for Patients With Neck Pain
January 2007 Volume 87 Number 1 Physical Therapy f 15
Table 1.Demographics, Baseline Self-report Variables, and Baseline Characteristics of Subjects
Variablea All Subjects(n�78)
SuccessfulOutcome Group(n�42)
NonsuccessfulOutcome Group(n�36)
P
Age, y
X 42.0 41.6 42.3
.79bSD 11.3 13.7 8.3
Range 20–60 20–60 22–56
Sex
Female, n (%) 53 (68%) 27 (64%) 26 (72%) .63c
Duration of symptoms, d
X 80 54.6 109.6
�.001bSD 70.6 39.6 86.4
Range 7–395 7–180 21–395
NPRS
X 4.7 4.6 4.8
.86bSD 1.8 1.8 1.8
Range 1–8 1–8 1–8
NDI
X 34.9 34.5 35.2
.80bSD 10.1 11.3 8.7
Range 20–58 20–58 22–54
FABQPA
X 12.6 11.8 14.2
.036bSD 4.1 3.8 3.8
Range 2–22 2–19 6–22
FABQW
X 13.1 10.3 16.2
.01bSD 10.1 8.8 10.8
Range 0–36 0–28 0–36
Symptoms distal to the shoulder, n (%) 35 (45%) 12 (29%) 23 (64%) .083c
Mode of onset
Traumatic, n (%) 32 (41%) 16 (38%) 16 (44%) .57c
Prior history of neck pain, n (%) 26 (33%) 16 (38%) 10 (27%) .34a
Symptoms (n [%]) aggravated by:
Turning right 52 (67%) 29 (69%) 23 (64%) .63c
Turning left 51 (65%) 28 (67%) 23 (64%) .80c
Looking up 42 (54%) 14 (33%) 28 (78%) �.001c
Looking down 54 (69%) 27 (64%) 27 (75%) .31c
Driving 64 (82%) 33 (79%) 31 (86%) .39c
a NPRS�numeric pain rating scale, NDI�Neck Disability Index, FABQPA�Fear-Avoidance Beliefs Questionnaire physical activity subscale, FABQW�Fear-Avoidance Beliefs Questionnaire work subscale.b Analyzed with independent-samples t tests.c Analyzed with chi-square tests.
Clinical Prediction Rule for Patients With Neck Pain
16 f Physical Therapy Volume 87 Number 1 January 2007
subject reports that looking up doesnot aggravate symptoms, FABQPAscore of �12, diminished upper tho-racic spine kyphosis (T3–T5), andcervical extension of �30 degrees(� 2�55.0, df�6, P�.001, Nagel-kerke R2�.682). These 6 variableswere used to form the most parsimo-nious combination of predictors foridentifying patients with neck painwho are likely to benefit from tho-racic spine thrust manipulation. Reli-ability data for these variables areshown in Table 4. The reliability val-ues for the remainder of the patienthistory and clinical examination arereported elsewhere.53
Fourteen out of 15 subjects whowere positive on at least 4 of thecriteria and 32 of 37 subjects whowere positive on at least 3 criteria
were in the successful outcomegroup. Of the 41 subjects with 2 orfewer variables, 31 were in thenonsuccessful outcome group(Tab. 5). Accuracy statistics werecalculated for the numbers of vari-ables present (Tab. 6). The pretestprobability for the likelihood ofsuccess with thoracic spine thrustmanipulation for this study was54% (42 out of 78 subjects). If asubject exhibited 4 out of the 6variables, the positive LR was 12.0(95% CI�2.3–70.8) and the post-test probability of successincreased to 93%. If a subject waspositive on 3 out of the 6 variables,the positive LR was 5.5 (95%CI�2.7–12.0) and the posttestprobability of success was 86%. Ifonly 2 of the 6 variables werepresent, the positive LR decreased
to 2.1 (95% CI�1.5–2.5) and theposttest probability of success was71%.
The analysis of the point-based sys-tem revealed a possible total of 10points (for the 6 variables). The cut-off that maximized the diagnosticaccuracy of the point-based systemwas 3.5 points. This resulted in asensitivity of .83 (95% CI�.69–.92),a specificity of .86 (95% CI�.71–.94), a positive LR of 5.9 (95%CI�2.6–13.0), and a posttest proba-bility of 87%.
There was no significant differencein outcomes among therapists withvarying levels of experience foreither the percentage of successfuloutcomes or NDI change scores(P�.05). The group with �3 years of
Table 2.Categorical Variables From the Baseline Clinical Examination
Variable All Subjects(n�78)
SuccessfulOutcome Group(n�42)
NonsuccessfulOutcome Group(n�36)
P
Centralization during cervical motion testing (%) 27 33 20 .17a
Peripheralized during cervical motion testing (%) 36 31 42 .33a
No. of hypomobile levels identified during springtesting in the cervical spine
X 1.4 1.4 1.5.89b
SD 1.4 1.4 1.5
No. of hypomobile levels identified during springtesting in the thoracic spine
a Analyzed with independent-samples t tests.b Analyzed with chi-square tests.
Clinical Prediction Rule for Patients With Neck Pain
January 2007 Volume 87 Number 1 Physical Therapy f 17
experience achieved a success rateof 16/30 (53%), and the group thathad �3 years of experience demon-strated a success rate of 26/48 (54%).The NDI change scores were 12.8(SD�15.7) for the group with �3years of experience and 14.8(SD�14.6) for the group with �3years of experience.
DiscussionThe LR is the statistic often used todetermine the usefulness of a CPR.19
We selected to report the positiveLR because the purpose of thisstudy was to determine the changein probability that patients are likelyto experience a successful outcome
when they satisfy the criteria ofthe CPR. Based on the pretest prob-ability in this study (54%) that a sub-ject would respond positively to tho-racic spine thrust manipulation, ifthe subjects exhibited 4 of the 6 cri-teria (positive LR�12), the posttestprobability of success increased dra-matically to 93%. However, based onthe wide CI associated with positivefindings on 4 out of 6 tests (95%CI�2.28–70.8), clinicians can havegreater accuracy when determiningthe likelihood that a patient withneck pain will exhibit a rapidresponse to thoracic spine thrustmanipulation when using 3 out of 6variables (positive LR�5.5, 95%
CI�2.72–12.0) to guide decisionmaking (posttest probability�86%).
In some circumstances, assigning aweight to individual predictorsbased on the beta coefficientsincreases the accuracy of prognosticmodels.54 However, in someinstances, it is possible that translat-ing a prognostic model to a point-based scoring system can decreasethe discriminatory power of theindex.55 The cutoff point for thepoint-based system that maximizedthe diagnostic accuracy resulted in apositive LR of 5.9 and a posttestprobability of 87%, which onlyexceeded the posttest probability of
Table 3.Continuous Variables From the Baseline Clinical Examination
Variable All Subjects(n�78)
SuccessfulOutcome Group(n�42)
NonsuccessfulOutcome Group(n�36)
P a
Cervical flexion (°)b
X 42.5 41.6 43.5.49
SD 11.9 12.7 10.9
Cervical extension (°)b
X 33.9 28.8 39.8�.001
SD 12.6 9.4 13.3
Cervical side bending, right (°)b
X 31.4 31 31.9.76
SD 12.9 12.7 13.3
Cervical side bending, left (°)b
X 33.4 33.3 33.4.97
SD 15.5 14.3 17.0
Cervical rotation, right (°)c
X 59.6 60.7 58.4.38
SD 11.8 10.7 13.3
Cervical rotation, left (°)c
X 61.2 61.9 60.4.61
SD 12.2 12.3 12.3
Deep neck flexor muscle endurance (s)
X 6.8 6.0 7.6.21
SD 5.6 3.6 7.2
a Analyzed with independent-samples t tests.b Indicates measurement with a gravity inclinometer.c Indicates measurement with a standard dual-armed goniometer.
Clinical Prediction Rule for Patients With Neck Pain
18 f Physical Therapy Volume 87 Number 1 January 2007
the equal scoring system of the CPRby 1%. We therefore refrained fromusing the point-based system be-cause it does not add to the predic-tive accuracy of the rule and wouldincrease the complexity of the CPR,likely further detracting from theimplementation of the rule in clini-cal practice.56
The ability to a priori identifypatients with neck pain who arelikely to experience an early successwith thoracic spine thrust manipula-tion while avoiding the potential riskassociated with cervical spine thrustmanipulation is useful for guidingclinical decision making for individ-ual patients. The CPR also is usefulfor identifying patients with neckpain who should perhaps receiveother forms of treatment rather thanthoracic spine thrust manipulation.In our study, for example, if subjectsexhibited only one of the variables,the positive LR was only 1.2, suggest-ing that the posttest probability ofthese subjects achieving a successfuloutcome is not much larger thanchance, corresponding to a negligi-
Table 5.The 6 Variables Forming the Clinical Prediction Rule and the Number of Subjects inEach Group at Each Levela
y Symptoms �30 d
y No symptoms distal to the shoulder
y Looking up does not aggravate symptoms
y FABQPA score �12
y Diminished upper thoracic spine kyphosis
y Cervical extension ROM �30°
No. of PredictorVariables Present
SuccessfulOutcome Group
NonsuccessfulOutcome Group
6 2 0
5 3 0
4 9 1
3 18 4
2 7 11
1 3 14
0 0 6
a FABQPA�Fear-Avoidance Beliefs Questionnaire physical activity subscale, ROM�range of motion.
Table 4.Accuracy Statistics With 95% Confidence Intervals (CIs) for Individual Predictor Variables and Interrater Reliabilitya
Variable Reliability(95% CI)
Sensitivity(95% CI)
Specificity(95% CI)
PositiveLikelihood Ratio(95% CI)
Symptom duration �30 d NA .36 (.22–.52) .94 (.80–.99) 6.4 (1.60–26.3)
No symptoms distal to the shoulder NA .67 (.50–.80) .53 (.36–.69) 1.4 (0.94–2.2)
FABQPA score �12 NA .28 (.16–.45) .91 (.76–.98) 3.4 (1.05–11.20)
FABQW score �10 NA .55 (.39–.70) .69 (.52–.83) 1.8 (1.02–3.15)
a FABQPA�Fear-Avoidance Beliefs Questionnaire physical activity subscale, FABQW�Fear-Avoidance Beliefs Questionnaire work subscale, ROM�range ofmotion, NA�not applicable (subjects completed self-report measures only once [included the date of injury] and thus reliability data was not calculated).b Kappa.c Intraclass correlation coefficient.
Clinical Prediction Rule for Patients With Neck Pain
January 2007 Volume 87 Number 1 Physical Therapy f 19
ble increase of the posttest probabil-ity to 58% (Tab. 6).
Six predictor variables were retainedin the logistic regression analysis asmaximizing the accuracy of predict-ing patients with neck pain who arelikely to respond to thoracic spinethrust manipulation. Although theduration of the current episode wasthe strongest individual predictor,we used a higher threshold for defin-ing success on the GROC than whathas been recommended40 to providea greater degree of distinctionbetween subjects who improveddramatically with manipulation andthose who were improving overtime simply due to natural history ofthe disorder. In addition, the magni-tude of the difference in changescores for both the NPRS and NDIfurther substantiates that an impor-tant clinical change occurred in thegroup that was identified as havingexperienced a successful outcome.
The duration of the current episodewas identified as the strongest pre-dictor in a CPR for identifyingpatients with LBP who are likely toexperience a rapid and dramaticresponse to spinal manipulation(positive LR�4.39).19 The validationof the CPR also demonstrated that ashorter duration of symptoms waspredictive for identifying patientswho would respond to manipulation(positive LR�4.4).20 However, dura-
tion of symptoms was not predic-tive of the outcomes associatedwith the comparison group whoreceived an exercise program (pos-itive LR�1.0), suggesting that ashorter duration is predictive ofresponse to manipulation and notthe natural history of the disor-der.20 Further validation studies areneeded to determine whether thisis also the case with the currentCPR.
The FABQ was a predictor variablefor identifying patients with LBPwho are likely to respond to eitherspinal manipulation (FABQW)19,20 orspinal stabilization (FABQPA).18 Incontrast to patients with LBP whoare likely to benefit from spinal sta-bilization who exhibited elevatedFABQPA scores (�8),18 our studyidentified lower FAPQPA scores(�12) as a predictor of a successfuloutcome. A correlation between dis-ability and the FABQPA was iden-tified by George et al27 and Neder-hand et al,57 suggesting that fear-avoidance beliefs exhibit predictivevalidity in identifying patients withneck pain who may be at risk forchronic disability. Further researchis necessary to clarify the role of fear-avoidance beliefs in patients withneck pain.
One common flaw in the develop-ment of CPRs is that researchersoften do not investigate the reliabil-
ity of the measures used in theirstudy and thus cannot determinewhether predictor variables provideadequate reproducibility to beincluded in the rule.58 We investi-gated the reliability of potential pre-dictor variables and, according tothe descriptive criteria provided byLandis and Koch,59 all variables inthe CPR exhibited fair to substantialreliability. We consider these reliabil-ity coefficients acceptable to guideclinical decision making in the man-agement of patients with neck pain.
The predictor variables of adecreased upper thoracic spinekyphosis from T3 through T5 anddecreased cervical extension may beassociated with the biomechanicallink between the thoracic spine andthe cervical spine. Recent literatureidentified a correlation betweenmobility at the cervicothoracic junc-tion and thoracic spine with neck-shoulder pain.11,60,61 It is also possi-ble that impaired mobility in thethoracic spine may be a contributorto mechanical neck pain.62–64
Patient reports of “looking up doesnot aggravate the symptoms” and“no symptoms distal to the shoul-der,” as recorded on a body diagram,also were identified as predictor vari-ables in the CPR. In contrast, thepopulation that has pain distal to theshoulder that is aggravated by lookingup could potentially be a subgroupof patients with cervical radiculopa-
Table 6.Combination of Predictor Variables and Associated Accuracy Statistics With 95% Confidence Intervals
a The probability of success is calculated using the positive likelihood ratios and assumes a pretest probability of 54%.
Clinical Prediction Rule for Patients With Neck Pain
20 f Physical Therapy Volume 87 Number 1 January 2007
thy rather than solely mechanicalneck pain.65,66 Although symptomsextending into the arm and radicularsigns are not associated with a worseprognosis,67 it has been suggestedthat patients with more distal symp-toms may be more responsive to adifferent treatment approach such ascervical traction and other distrac-tion-oriented interventions.21
We successfully achieved the pur-pose of developing a CPR that iden-tifies patients with neck pain whoare likely to exhibit early successafter thoracic spine thrust manipula-tion. However, this is only the firststep in the process of developingand testing a CPR.68 Although no dif-ference in outcomes occurredamong clinicians with varying levelsof experience, it should be recog-nized that data were collected atonly one clinical site by 4 physicaltherapists. Future studies are neces-sary to validate our results and deter-mine whether similar findings occurin a broader patient population withdifferent treating clinicians. Addi-tionally, a validation study shouldinclude a long-term follow-up anda comparison group to further inves-tigate the predictive value of thevariables in the CPR. If the rule isvalidated, an impact analysis ofimplementation of the rule on clini-cal practice patterns, outcomes, andcosts of care should be investigated.
Dr Cleland, Dr Childs, Dr Fritz, and Dr Whit-man provided concept/idea/research de-sign, writing, and fund procurement. Dr Cle-land and Ms Eberhart provided data collec-tion, subjects, facilities/equipment, and cler-ical support. Dr Cleland, Dr Childs, and DrFritz provided data analysis. Dr Cleland andDr Childs provided project management. Allauthors provided consultation (includingreview of manuscript before submission). DrCleland acknowledges Madeleine Hellman,PT, EdD, MHM, and Eric Shamus, PT, PhD,for their dissertation committee work atNova Southeastern University, Ft Lauder-dale, Fla. The authors also thank SherylCheney, PT, and Diane Olimpio, PT, Directorof Physical Therapy, Rehabilitation Services,
Concord Hospital, for their assistance withdata collection.
The study was approved by the institutionalreview boards at Concord Hospital and NovaSoutheastern University.
This study was supported by a grant fromthe Orthopaedic Section of the AmericanPhysical Therapy Association.
This article was received June 5, 2006, andwas accepted August 18, 2006.
DOI: 10.2522/ptj.20060155
References1 Cote P, Cassidy J, Carroll L. The factors
associated with neck pain and its relateddisability in the Saskatchewan population.Spine. 2000;25:1109–1117.
2 Palmer KT, Walker-Bone K, Griffin MJ,et al. Prevalence and occupational associ-ations of neck pain in the British popula-tion. Scand J Work Environ Health. 2001;27:49–56.
3 Cote P, Cassidy JD, Carroll LJ, Kristman V.The annual incidence and course of neckpain in the general population: a popula-tion-based cohort study. Pain. 2004;112:267–273.
4 Wright A, Mayer T, Gatchel R. Outcomesof disabling cervical spine disorders incompensation injuries: a prospective com-parison to tertiary rehabilitation responsefor chronic lumbar disorders. Spine. 1999;24:178–183.
5 Jette AM, Smith K, Haley SM, Davis KD.Physical therapy episodes of care forpatients with low back pain. Phys Ther.1994;74:101–110.
6 Borghouts J, Janssen H, Koes B, et al. Themanagement of chronic neck pain in gen-eral practice: a retrospective study. ScandJ Prim Health Care. 1999;17:215–220.
7 Haldeman S, Kohlbeck FJ, McGregor M.Unpredictability of cerebrovascular isch-emia associated with cervical spine manip-ulation therapy: a review of sixty-fourcases after cervical spine manipulation.Spine. 2002;27:49–55.
8 Di Fabio RP. Manipulation of the cervicalspine: risks and benefits. Phys Ther. 1999;79:50–65.
10 Cleland JA, Childs JD, McRae M, et al.Immediate effects of thoracic manipula-tion in patients with neck pain: a random-ized clinical trial. Man Ther. 2005;10:127–135.
11 Norlander S, Nordgren B. Clinical symp-toms related to musculoskeletal neck-shoulder pain and mobility in the cervico-thoracic spine. Scand J Rehabil Med.1998;30:243–251.
12 Adams G, Sim J. A survey of UK manualtherapists’ practice of and attitudestowards manipulation and its complica-tions. Physiother Res Int. 1998;3:206–227.
13 Fernandez-de-las-Penas C, Fernandez-Carn-ero J, Fernandez AP, et al. Dorsal manipu-lation in whiplash injury treatment: a ran-domized controlled trial. Journal ofWhiplash and Related Disorders. 2004;3:55–72.
14 Savolainen A, Ahlberg J, Nummila H, Nissi-nen M. Active or passive treatment forneck-shoulder pain in occupational healthcare? A randomized controlled trial. OccupMed (Lond). 2004;54:422–424.
15 Leboeuf-Yde C, Lauritsen JM, Lauritzen T.Why has the search for causes of low backpain largely been nonconclusive? Spine.1997;22:877–881.
16 Petren-Mallmin M, Linder J. MRI cervicalspine findings in asymptomatic fighterpilots. Aviat Space Environ Med. 1999;70:1183–1188.
17 Rose S. Physical therapy diagnosis: roleand function. Phys Ther. 1989;69:535–537.
18 Hicks GE, Fritz JM, Delitto A, McGill SM.Preliminary development of a clinical pre-diction rule for determining whichpatients with low back pain will respondto a stabilization exercise program. ArchPhys Med Rehabil. 2005;86:1753–1762.
19 Flynn T, Fritz J, Whitman J, et al. A clinicalprediction rule for classifying patientswith low back pain who demonstrateshort term improvement with spinalmanipulation. Spine. 2002;27:2835–2843.
20 Childs JD, Fritz JM, Flynn TW, et al. Aclinical prediction rule to identify patientslikely to benefit from spinal manipulation:a validation study. Ann Intern Med. 2004;141:920–928.
21 Childs JD, Fritz JM, Piva SR, Whitman JM.Proposal of a classification system forpatients with neck pain. J Orthop SportsPhys Ther. 2004;34:686–696.
22 Werneke M, Hart DL, Cook D. A descrip-tive study of the centralization phenome-non: a prospective analysis. Spine. 1999;24:676–683.
23 Jensen MP, Turner JA, Romano JM. Whatis the maximum number of levels neededin pain intensity measurement? Pain.1994;58:387–392.
24 Vernon H, Mior S. The Neck DisabilityIndex: a study of reliability and validity.J Manipulative Physiol Ther. 1991;14:409–415.
25 Waddell G, Newton M, Henderson I, et al.Fear-Avoidance Beliefs Questionnaire andthe role of fear-avoidance beliefs inchronic low back pain and disability.Pain. 1993;52:157–168.
26 Jacob T, Baras M, Zeev A, Epstein L. Lowback pain: reliability of a set of pain mea-surement tools. Arch Phys Med Rehabil.2001;82:735–742.
27 George S, Fritz J, Erhard E. A comparisonof fear-avoidance beliefs in patients withlumbar spine pain and cervical spine pain.Spine. 2001;26:2139–2145.
Clinical Prediction Rule for Patients With Neck Pain
January 2007 Volume 87 Number 1 Physical Therapy f 21
28 Flynn TW, Whitman J, Magel J. Orthopae-dic Manual Physical Therapy Manage-ment of the Cervical-Thoracic Spine andRibcage. San Antonio, Tex: ManipulationsInc; 2000.
29 Kendall FP, McCreary EK, Provance PG.Muscles: Testing and Function. 4th ed.Baltimore, Md: Williams & Wilkins; 1993.
30 Griegel-Morris P, Larson K, Mueller-KlausK, Oatis CA. Incidence of common pos-tural abnormalities in the cervical, shoul-der, and thoracic regions and their associ-ation with pain in two age groups ofhealthy subjects. Phys Ther. 1992;72:425–431.
31 McKenzie RA. Cervical and ThoracicSpine: Mechanical Diagnosis and Ther-apy. Minneapolis, Minn: OrthopaedicPhysical Therapy Products; 1990.
32 Harris KD, Heer DM, Roy TC, et al. Reli-ability of a measurement of neck flexormuscle endurance. Phys Ther. 2005;85:1349–1355.
33 Maitland G, Hengeveld E, Banks K, EnglishK. Maitland’s Vertebral Manipulation.6th ed. Oxford, United Kingdom: Butter-worth-Heinemann; 2000.
34 Spurling RG, Scoville WB. Lateral ruptureof the cervical intervertebral discs: a com-mon cause of shoulder and arm pain. SurgGynecol Obstet. 1944;78:350–358.
36 Wainner R, Fritz J, Irrgang J, et al. Reliabil-ity and diagnostic accuracy of the clinicalexamination and patient self-report mea-sures for cervical radiculopathy. Spine.2003;28:52–62.
37 Elvey RL. The investigation of arm pain:signs of adverse responses to the physicalexamination of the brachial plexus andrelated tissues. In: Boyling JD, PalastangaN, eds. Grieve’s Modern Manual Ther-apy. New York, NY: Churchill LivingstoneInc; 1994:577–585.
38 Jaeschke R, Guyatt GH, Sackett DL. Users’guides to the medical literature, III: howto use an article about a diagnostic test, B.What are the results and will they help mein caring for my patients? The Evidence-Based Medicine Working Group. JAMA.1994;271:703–707.
39 Erhard RE. The Spinal Exercise Hand-book: A Home Exercise Manual for aManaged Care Environment. Pittsburgh,Pa: Laurel Concepts; 1998.
40 Jaeschke R, Singer J, Guyatt G. Measure-ment of health status: ascertaining theminimal clinically important difference.Controlled Clin Trials. 1989;10:407–415.
41 Koes BW, Bouter LM, van Mameren H,et al. The effectiveness of manual therapy,physiotherapy, and treatment by the gen-eral practitioner for nonspecific back andneck complaints: a randomized clinicaltrial. Spine. 1992;17:28–35.
42 Koes BW, Bouter LM, van Mameren H,et al. Randomised clinical trial of manipu-lative therapy and physiotherapy for per-sistent back and neck complaints: resultsof one-year follow-up. BMJ. 1992;304:601–605.
43 Farrar J, Young JJ, La Moreaux L, et al.Clinical importance of changes in chronicpain intensity measured on an 11-pontnumerical pain rating scale. Pain. 2001;94:149–158.
44 Hurst H, Bolton J. Assessing the clinicalsignificance of change scores recorded onsubjective outcome measures.J Manipulative Physiol Ther. 2004;27:26–35.
45 Bolton JE. Sensitivity and specificity ofoutcome measures in patients with neckpain: detecting clinically significantimprovement. Spine. 2004;29:2410–2417.
46 Hoving JL, Koes BW, de Vet HC, et al.Manual therapy, physical therapy, or con-tinued care by a general practitioner forpatients with neck pain: a randomized,controlled trial. Ann Intern Med. 2002;136:713–722.
47 Freedman DA. A note on screening regres-sion equations. The American Statisti-cian. 1983;37:152–155.
48 Deyo RA, Centor RM. Assessing theresponsiveness of functional scales to clin-ical change: an analogy to diagnostic testperformance. J Chronic Dis. 1986;39:897–906.
49 Concato J, Feinstein AR, Holford TR. Therisk of determining risk with multivariablemodels. Ann Intern Med. 1993;118:201–210.
50 Cohen J. A coefficient of agreement fornominal scales. Educ Psychol Meas. 1960;20:37–46.
51 Cohen J. Weighted kappa: nominal scaleagreement with provision for scaled dis-agreement or partial credit. Psychol Bull.1968;70:213–220.
53 Cleland JA, Childs JD, Fritz JM, WhitmanJM. Inter-rater reliability of the historicaland physical examination in patients withmechanical neck pain. Arch Phys MedRehabil. 2006;87:1388–1395.
54 Kuijpers T, van der Windt DA, Boeke AJ,et al. Clinical prediction rules for the prog-nosis of shoulder pain in general practice.Pain. 2006;120:276–285.
55 Lee SJ, Lindquist K, Segal MR, CovinskyKE. Development and validation of a prog-nostic index for 4-year mortality in olderadults. JAMA. 2006;295:801–808.
56 Redelmeier DA, Lustig AJ. Prognostic indi-ces in clinical practice. JAMA. 2001;285:3024–3025.
57 Nederhand MJ, Ijzerman MJ, Hermens HJ,et al. Predictive value of fear avoidance indeveloping chronic neck pain disability:consequences for clinical decision mak-ing. Arch Phys Med Rehabil. 2004;85:496–501.
58 Laupacis A, Sekar N, Stiell IG. Clinicalprediction rules: a review and suggestedmodifications and methodological stan-dards. JAMA. 1997;277:488–494.
59 Landis JR, Koch CG. The measurement ofobserver agreement for categorical data.Biometrics. 1977;33:159–174.
60 Norlander S, Aste-Norlander U, NordgrenB, Sahlstedt B. Mobility in the cervico-thoracic motion segment: an indicativefactor of musculo-skeletal neck-shoulderpain. Scand J Rehabil Med. 1996;28:183–192.
61 Norlander S, Gustavsson BA, Lindell J,Nordgren B. Reduced mobility in the cer-vico-thoracic motion segment—a risk fac-tor for musculoskeletal neck-shoulderpain: a two-year prospective follow-upstudy. Scand J Rehabil Med. 1997;29:167–174.
62 Greenman P. Principles of Manual Medi-cine. 2nd ed. Philadelphia, Pa: LippincottWilliams & Wilkins; 1996.
63 Johansson H, Sojka P. Pathophysiologicalmechanisms involved in genesis andspread of muscular tension in occupa-tional muscle pain and in chronic muscu-loskeletal pain syndromes: a hypothesis.Med Hypotheses. 1991;35:196–203.
64 Knutson GA. Significant changes in sys-tolic blood pressure post vectored uppercervical adjustment vs resting controlgroups: a possible effect of the cervico-sympathetic and/or pressor reflex.J Manipulative Physiol Ther. 2001;24:101–109.
65 Wainner R, Gill H. Diagnosis and nonop-erative management of cervical radiculop-athy. J Orthop Sports Phys Ther. 2000;12:728–744.
66 Daffner S, Hilibrand A, Hanscom B, et al.Impact of neck and arm pain on overallhealth status. Spine. 2003;2817:2035.
67 Borghouts JA, Koes BW, Bouter LM. Theclinical course and prognostic factors ofnon-specific neck pain: a systematicreview. Pain. 1998;77:1–13.
68 McGinn T, Guyatt G, Wyer P, et al. Users’guides to the medical literature, XXII: howto use articles about clinical decisionrules. JAMA. 2000;284:79–84.
Clinical Prediction Rule for Patients With Neck Pain
22 f Physical Therapy Volume 87 Number 1 January 2007
Appendix.Operational Definitions for Special Tests Used in the Study
Test Performance Criteria for Positive Test
Spurling test34 The patient is seated, and the neckis passively side bent toward thesymptomatic side. The examinerapplies approximately 7 kg offorce through the patient’s headwith a caudally directed force.
Reproduction of the patient’s upper-extremitysymptoms
Neck Distraction Test36 The patient is positioned supine,and the examiner grasps underthe patient’s chin and occiput.The examiner flexes the neck topatient comfort and then applies adistraction force of approximately14 kg.
Reduction or resolution of the patient’supper-extremity symptoms
Upper Limb Neurodynamic Test37 The patient is positioned supine,and the examiner places thepatient’s upper extremity into:(1) scapular depression,(2) shoulder abduction,(3) forearm supination and wrist
and finger extension,(4) shoulder external rotation,(5) elbow extension, and(6) contralateral then ipsilateral
cervical lateral flexion.
Any of the following constitute a positive test:(1) Symptom reproduction(2) Greater than 10° difference in elbow
extension from side to side(3) An increase in symptoms with
contralateral cervical side bending ordecrease in symptoms with ipsilateralside bending
Roos test35 The patient is positioned standingand abducts the arms to 90° withlateral rotation of the shoulder.The patient then opens and closesthe hands slowly for 3 min.
The test is considered positive if the patient isunable to maintain the position or reportsheaviness and tingling in the arm.
Clinical Prediction Rule for Patients With Neck Pain
January 2007 Volume 87 Number 1 Physical Therapy f 23
