Rthe reliability of their clinical measures.10 Reliability

SCIENTIFIC/CLINICAL ARTICLES

This study formed part of an MSc in Rehabilitation of the UpperLimb at Curtin University of Technology.is one of the most basic requirements of evaluationtools and is established when it has been demon-strated that repeated measurements are consistent.3

To enhance a tools reliability it is necessary to exam-ine and develop measurement protocols. Reliabilityand instrument precision are essential requirementsto track changes in clinical practice.11,12

This study was funded jointly by the Westminster Medical SchoolResearch Trust, the Chelsea & Westminster Medical SchoolResearch Trust, and the Chelsea & Westminster Healthcare NHSTrust Charity.

This study was presented at the IFSHT congress in June 2004.

Correspondence and reprint requests to Fiona Coldham, BSc, MSc,MCSP, Physiotherapy Department, Chelsea & Westminster Hospital,London SW10 9NH, UK; e-mail: .JHT READ FO

The Reliability of Onein Symptomatic and A

Fiona Coldham, BSc (Hons), MSc, MCSPyJeremy Lewis, PhD, MCSP, MMPAHoe Lee, PhD, MSc, PG Dip, Prof DipPhysiotherapy DepartmentChelsea & Westminster HospitalLondon, UK

Grip strength testing is commonly used both inclinical and research settings to evaluate hand andupper limb function, level of impairment, and phys-ical performance.16 It is a rapid and easy test to per-form producing reliable results, which are easy torecord and interpret.7 There are various testing

y0894-1130/$ see front matter 2006 Hanley & Belfus, an imprintof Elsevier Inc. All rights reserved.

doi:10.1197/j.jht.2006.04.002

318 JOURNAL OF HAND THERAPYCREDIT #038

vs. Three Grip Trialssymptomatic Subjects

ABSTRACT: Grip strength is used in the assessment of hand andupper limb function. Current recommendations state that takingthemean of three repeated grip trials providesmore reliable resultsthan only one trial. A repeated measures, crossover design wasused. Sixty-six subjects were recruited (22 asymptomatic subjects,22 following carpal tunnel decompression, and 22 following flexortendon repair). Grip strength testing was performed on a Jamardynamometer using a standardized testing protocol. Pre- andpost testing pain levels were recorded using a verbal analoguescale. Each subjects grip strength was tested four times, twiceusing a single trial protocol and twice using three grip trials in ran-dom order. Intraclass correlation coefficients (ICC) (2,1), 95% con-fidence intervals, and standard error of measurements werecalculated. A two-tailed paired samples t-test was used to investi-gate the difference between the grip strength values obtained andthe changes in verbal analogue scale. High levels of testretestreliability (ICC$ 0.85) were found for the three methods of gripstrength testing (one trial, the mean of three trials, and the bestof three trials). The mean values of grip strength generated foreach method of grip strength testing produced comparable results.A significant difference (p$ 0.0001) was observed in the verbal an-alogue scale scores following one trial and three in all three samplegroups. Clinically acceptable levels of reliability ($0.91) were dem-onstrated by all three methods of grip strength testing other thanthe mean of three trials for the asymptomatic group. Distributionof the ICC results and the elevated verbal analogue scales associ-ated with three trials suggest that the use of one grip trial maybe appropriate. This study suggests that one maximal trial is asreliable as and less painful than either the best of, or, mean of threetrials.

J HAND THER. 2006;19:31827.

protocols available and the most appropriate shouldbe chosen in light of the assessment requirements andsubject involved. The use of a standardized protocolhas repeatedly been shown to maximize reliability.8,9

In the current climate of clinical governance in theUnited Kingdom, with the need to demonstrateclinical effectiveness, it is essential that the physio-therapy and occupational therapy professions proveThe American Society of Hand Therapists (ASHT)currently recommends using the mean of three trialsas it has been reported to produce more reliable

results than only one grip strength trial.13 These rec- In clinical practice, therapists may only be able to

ommendations have apparently been derived fromunpublished work by Fess on hand position andnormal grip strength.14

Investigations have been performed to ascertainthe most advantageous number of trials to maximizereliability. Bohannon and Saunders15 investigated thereliability and magnitude of the first (which they feltwas comparable to a single trial), the maximal, andthe mean of three maximal voluntary elbow flexiontrials. A digital hand held dynamometer was usedin this study to measure elbow flexion force. Theyfound that the maximal and mean force measure-ments were significantly different from one another(p. 0.001). All three measurement methods werefound to demonstrate comparable high levels ofreliability between days (intraclass correlation co-efficient [ICC] 3,1 0.9770.987). The authorsconcluded that if their findings were also found tobe true in a symptomatic population then a singletrial rather than a maximal or mean value of multipletrials may be adequate to accurately represent musclestrength.MacDermid et al.16 established that the reliability

coefficients were similar regardless of whether a sin-gle measurement or the mean of the three trials wasused. They concluded that satisfactory results maybe obtained using only one trial. This was the onlystudy found which involved a symptomatic subjectgroup and addressed the issue of the reliability ofthe number of grip strength trials performed.Hamilton et al.17 also investigated the reliability of

the use of one trial, the mean of two and three trialsand the highest score of three trials to measure gripstrength on two occasions for 33 asymptomaticsubjects. They found no statistically significant differ-ence between the reliability of the various combina-tions of trials, thereby concluding that grip strengthscores obtained from a single grip measure will beas reliable as those obtained from the mean of twoor three repeated measures.Jansen et al.18 studied the reliability of the first,

maximal, and mean values of three trials using apinch dynamometer. They found the maximal scoreranked highest followed by the first and then themean of three trials. However, they concluded thatthey would continue to use the mean of three trialsdue to the wealth of literature supporting this mea-surement methods reliability.The use of three trials was supported by

Mathiowetz et al.14 who reported that the mean ofthree trials was more reliable than one for asymptom-atic subjects. However, the subject group used in thisstudy included asymptomatic, female, occupationaltherapy students between the ages of 20 and 39 andtherefore is not generalizable to a symptomatic popu-lation. Subsequently, there is little agreement regard-ing the optimal number of trials that should be used.use a restricted number of grip strength repetitions.This may be due to either the inherent nature of apatients condition and associated pain or time andfinancial constraints.16,18 Thus, if one repetition isfound to provide acceptable reliability in sympto-matic subjects the clinical implications are potentiallywide ranging. It is also hypothesized that discomforton gripping may yield submaximal grip force pro-duction and thus render the mean of three trialsless reliable than one maximal grip which may elicitless pain. This is supported by Stratford et al.19 whoinvestigated the reliability of certain outcome mea-sures in symptomatic subjects with extensor carpi ra-dialis tendonitis. They found that the presence ofpain affected the testretest reliability of the gripstrength measure. Schreuders et al.11 felt pain wasone factor that may influence measurement errorand subsequently in their study excluded patientswho complained of pain. It was hypothesized thatpainmay inhibit the subjects from exerting their max-imum grip strength. It was, therefore, suggested thatthe measurements would more probably reflect theamount of pain the subject was experiencing ratherthan their grip strength.11 This has pertinent clinicalimplications as grip strength is frequently evaluatedin subjects with pain.

Research Objectives

The specific objectives of this research were to:

1. Establish the testretest reliability of taking oneand three maximal isometric grip strength read-ings in symptomatic and asymptomatic subjects.

2. Compare the testretest reliability of one vs. themean of three trials and the highest of three trialsin symptomatic and asymptomatic subjects.

3. Evaluate the level of pain experienced by subjectsduring the grip tests.

RESEARCH METHODOLOGY

Sixty-six subjects were tested (22 asymptomaticsubjects, 22 post carpal tunnel decompression, and 22post flexor tendon repair)(Table 1). Participants wererequired to attend one assessment session, whichlasted one hour. In this session they carried out eightgrip trials on the dynamometer set in the secondhandle position, with the total testing time lastingno more than 10 minutes.Symptomatic subjects were recruited from patients

attending the Physiotherapy and Hand Therapydepartments at the Chelsea & Westminster NHSTrust, London, UK. All subjects were required togive informed consent prior to commencing the study.Approval for this study was granted from CurtinUniversitys Human Research Ethics Committee,

JulySeptember 2006 319

TABLE 1. Subject Group Demographics

o

mehm

oestrength measurement each subject was requiredto have returned his or her baseline score on theverbal analogue scale.

5. After this time grip strength was then tested again,but the subject was required to only produce onemaximal contraction. Again the standard instruc-tions were read.

6. At the end of this evaluation the subject was askedagain to rate his or her pain on the numerical scale.

strength.13 Interrater reliability and testretestreliability of the Jamar dynamometer have beendemonstrated on asymptomatic subjects using astandardized protocol and standard verbalinstructions.14

The same dynamometer was used throughout thestudy as measures between Jamar dynamometershave been shown to differ.24 It was purchased newprior to the study commencing and was initially

320 JOURNAL OF HAND THERAPYand the Riverside Research and Ethics Committeefor the Chelsea & Westminster Hospital, London.A computer-generated randomization series was

generated to determine in which order the tests wereconducted for each subject in an attempt to eliminateany sequential bias, thereby meaning half were ran-domized to do three trials followed by one with theother half doing one trial followed by three.The assessment protocol was comprised of the

following:

1. The subject was asked to rate his or her level ofpain on a verbal analogue scale from 0 to 10, where0 equals no pain and 10 the worst pain imaginable.

2. Grip strength was tested using a Jamar dynamom-eter in a standardized test position, using the sec-ond handle setting on the dynamometer. A meanof three attempts was calculated, with a 15-secondrest in between each of the three contractions asrecommended by ASHT. Standard instructionswere read from a card:

Please hold the handle like this and squeeze as hard as youcan while the procedure was being demonstrated. Then thesubjectwaspositionedand instructed Areyouready?Squeezeas hard as you can. Harder.Harder.Harder.Relax. Theseinstructions are similar to those used in the study byMathiowetz et al.14

3. Following the test, the subject was asked to ver-bally rate his or her level of pain during the gripstrength test.

4. The subjects were requested to rest for 15 minutes.A period of 15minutes was thought to be sufficientto allow for recovery from any fatigue and settleany discomfort, which may have been elicitedthrough the assessment.1,20 Similar studies usedmuch shorter rest periods between repeated griptests.3,21 Prior to commencing the next grip

(Number ofCases 66)

Asymptomatic(Group 1) Dec

Age (years) 40.36 mean, range (2372) 60.50Gender 11 Males, 11 females 6 MalHandiness 21 Right, 1 left 20 RigSymptomatic hand Dominant hand

tested in each occasion15 Do

Duration postsurgery 20.5 mBaseline verbalanalogue scale

0 1.18 m7. A rest of 15 minutes was then given and steps 16repeated. If the subject completed three trialsfollowed by one they would now do one trialfollowed by three and vice versa.

8. No knowledge of results was given between thetest and retest. Subjects were also blinded to thedial of the dynamometer to eliminate visual feed-back and incentive.22

Testing Position

All subjects were seated in a straight backed chairwith the feet flat on the floor and positioned in astandardized position as described by Mathiowetz1

and advocated by the ASHT, with the shoulder ad-ducted and neutrally rotated, elbow flexed at 90,forearm in a neutral rotation, and the wrist between0 and 30 extension and between 0 and 15 ulnar de-viation. The arm was not supported by the examineror by an armrest. The dynamometer was then pre-sented vertically and in line with the forearm tomaintain the standard forearm and wrist positions.1

Tester

One hand therapist (FC) with five years handtherapy experience tested all subjects.

Instruments

The Jamar dynamometer (Promedics Ltd.,Blackburn, Lancashire, UK) is an adjustable, portablehydraulic dynamometer, which is commonly used inthe clinical setting.22,23 It was selected for investiga-tion in this study due to its recommendation by theAmerican Society for the Surgery of the Hand andthe ASHT as the tool of choice to measure grip

Carpal Tunnelmpressions (Group 2)

Flexor TendonRepairs (Group 3)

ean, range (3893) 39.91 mean, range (2072)s, 16 females 16 Males, 6 femalest, 2 left 20 Right, 2 leftinant, 7 nondominant 11 Dominant, 11 nondominant

nths, range (336 months) 5.7 months, range (313.2 months)an, range (08) 0.41 mean, range (03)

calibrated by themanufacturer prior to the start of the 2,1 was selected due to its use in similar studies3,31study. Calibration was also checked manually withknown weights according to the method describedby Fess12 and was retested regularly throughout thecourse of the data collection period as recommended.The dynamometer maintained its calibrationthroughout the duration of the study.The ASHT guidelines on the testing protocol of the

dynamometer recommending setting the handle ofthe dynamometer in the second position (3.8 cm)were followed in the study. This is commonly consid-ered to be the most effective position where both theintrinsic muscles and extrinsic flexors contribute tothe effort produced.25

The verbal analogue scale or numerical ratingscale26 was selected tomeasure the level of pain expe-rienced by the subject due to its frequent use in theclinical setting.27 It has also been shown to be a validmeasure of pain correlating highly with the visualanalogue scale27 and is the pain assessment toolmore commonly favored by patients.28

Research Design

The research followed a repeated measures, cross-over design.

Sample Size

Based on published recommendations for truereliability exceeding 0.7 a sample size of 19 wasrequired in this study.29 This was increased to 22 foreach of the sample groups in case of missing data,subject withdrawal, or an underestimation of thesample size required. Subsequently, the total samplesize was 66 (22 asymptomatic subjects, 22 followingcarpal tunnel decompression, and 22 posttendon re-pairs). These are larger sample groups than those ofprevious studies evaluating symptomaticsubjects.11,30

Statistical Analysis

The raw data collected were analyzed using theStatistical Package for the Social Sciences (SPSS Ltd.,Woking, Surrey, UK) version 12 software. The datawere entered manually into the data files. To confirmthe accuracy of the data entry, the data files werechecked to ensure that incorrect data had not beeninadvertently entered.The testretest reliability (intrarater reliability) was

investigated by analyzing the results of one trial, themean of three trials, and the highest of three trials,over both the test sessions and for each subject groupusing ICC 2,1, the 95% confidence interval (95% CI),and the standard error of measurement (SEM). Areview of the literature was undertaken to determinethe most appropriate method bywhich to analyze thecorrelations between the sets of data. The ICC modeland its recommendation for use in testretest andintrarater reliability studies.29

A two-tailed paired samples t-test was applied tothe results to establish the difference in mean gripstrengths.Similarly for the final part of the study, to ascertain

whether there was a significant difference betweenverbal analogue scale scores of one trial and three, atwo-tailed paired samples t-test was also applied tothese results. Box plots were then produced for thedifferences in verbal analogue scale scores followingone and three trials to visually depict the magnitudeof change in verbal analogue scale score and tohighlight any outliers, which are depicted as num-bers correlating to the subject number in theseries.29,32

RESULTS

Formula to convert kilograms to pounds:

1kg 2:2046 lbIt is recommended that ICC results ranging from

0.75 to 0.9 indicate good reliability and ICCs between0.91 and 1.0 indicate adequate reliability for clinicalmeasures.33 The three measurement methods allachieved good reliability in the context of these rec-ommendations but only the highest of three trialsand one trial demonstrated single measure ICCs of0.91 and above suggesting adequate clinical reliabil-ity. The results of the mean of three trials in theasymptomatic group consistently demonstrated aslight reduction in measurement reliability (ICC0.85) (Table 2). The SEM for the mean of three trialsin the asymptomatic group was 4.84 kg, which repre-sents a larger measurement error, which may haveimportant clinical implications when carrying out atrial using the mean of three trials (Figure 1).In contrast to the other subject groups the mean of

three trials demonstrated the highest ICC (0.98) in theflexor tendon repair group. These results wouldsuggest that the mean of three trials showed a slightincrease in measurement reliability when comparedto both the highest value of three trials and one singletrial (Table 3).The mean magnitude of the grip strength measure-

ment from one trial, the mean of three trials, and thebest of three trials are comparable within subjectgroups. The range of difference between the valueswithin subject groups is 0.12.5 kg, which wouldmake it unrealistic to distinguish clinically. It is alsonoted that the mean of three trials consistentlyproduced lower mean values than either one gripor the best value of three trials.A significant difference was observed between the

values obtained from one trial and those from themean of three trials (p 0.005) in the asymptomatic

JulySeptember 2006 321

TABLE 2. TestRetest Reliability of Different Methods of Grip Strength Trials

D

C

000

pdpain and the test itself producing some discomfort.The SEMs and ICCs calculated for both the sympto-matic subject groups (carpal tunnel decompressionsand flexor tendon repairs) demonstrated little differ-ence between measurement methods.The mean of three trials demonstrated the lowest

testretest reliability out of the three measurement

Asym

Subject Group

Sta

n

FIGURE 1.Displaying the standard error of measurement(kg) of one trial, the mean of three, and the highest of threetrials in each subject group.

322 JOURNAL OF HAND THERAPYand flexor tendon repair groups (p 0.0001) butnot in the carpal tunnel decompression group(p 0.093). No significant difference was found forthe values of one trial and the highest of three trialsin the asymptomatic group (p 0.718) and carpaltunnel decompression group (p 0.309) but therewas a significant difference in the flexor tendonrepair group (p 0.033). Finally, a significant differ-ence was observed in all three subject groupsbetween the highest of three trials and the mean ofthree trials (p 0.0001). This difference observed inactual magnitude of grip strength values for themean of three and the highest of three trials raisesquestions regarding the validity of the use of themean of three trials as a measure of maximal gripstrength.

Pain Levels Associated with the OneTrial and Three Trials

A significant difference (p$ 0.0001) was found inthe reported pain levels after one trial and after threetrials in each sample group. Through examination ofFigures 24 it can be observed that in all subjectgroups, one trial produces lower verbal analoguescale scores than three trials (p$ 0.0001). This wascalculated using a paired t-test analysis. Due to thecontroversy regarding the nature of data generatedthrough analogue scales, a nonparametric Wilcoxon

(Number of Cases 66)

Asymptomatic(Group 1)

ICC 2,1 95% CI SEM (kg) IC

One trial 0.95 0.890.98 3.04Mean of three trials 0.85 0.670.94 4.84Highest of three trials 0.95 0.890.98 2.93

ICC intraclass correlation coefficient; CI confidence interval; SEM

0123456

tom

atics

CTD

FTR

ard

E

rro

r o

f M

easu

rem

en

t

(K

g)

1 Trial

Mean of 3 trials

Highest of 3 trialssigned rank test was also run.29 This produced verysimilar results and the differences between painlevels experienced following one trial and followingthree were significant to the 0.001 level.An unexpectedly large proportion of asymptom-

atic subjects can be seen to have experienced somediscomfort when undertaking three trials. All sub-jects who rated one trial as painful considered threetrials more painful, represented by a higher verbalanalogue scale score.

DISCUSSION

TestRetest Reliability of the ThreeMeasurement Methods

All three methods of grip strength measurementdemonstrated very high levels of reliability with highICCs of $0.85 suggesting that all grip strengthmeasurements were stable over test occasions.Comparable results were found for one single trial

and the highest of three trials with high ICCs ($0.95)found in each sample group. Similarly, the SEMs ofone trial and the highest of three trials across each ofthe subject groups did not differ more than 0.7 kg.This difference is not possible to detect clinically onthe dial of a Jamar dynamometer, which displays2-kg increments. It is interesting to note that theSEMs for all three measurements were the highestfor the asymptomatic group and displayed the mostvariation with measurement method. This could bedue to the fact that the subjects in the asymptomaticgroupwere observed to bemoremotivated to achievehigher results from the second set of measurementseven though they did not have any knowledge ofthe results between tests. The subjects in the twosymptomatic groups appeared to be unaffected bymotivation to increase their score during the secondset of measurements. This is possibly due to a per-centage of the sample groups having pre-existing

Carpal Tunnelecompressions (Group 2)

Flexor TendonRepairs (Group 3)

2,1 95% CI SEM (kg) ICC 2,1 95% CI SEM (kg)

.97 0.940.99 1.9 0.96 0.910.98 2.2

.94 0.800.98 2.6 0.98 0.960.99 1.6

.97 0.920.99 2.2 0.97 0.900.99 2.3

standard error of measurement.

ONE TRIAL THREE TRIALS 4444N =

tions (SD) Averaged from Both the Test Sessions (kg)eNumber of trials

FIGURE 2. Displaying the verbal analogue scale resultsfollowing one grip strength trial and three trials for theasymptomatic subjects. Numbers 4, 3, 22, and 43 representoutliers.

ONE TRIAL THREE TRIALSNumber of trials

FIGURE 3. Displaying the verbal analogue scale resultsfollowing one grip strength trial and three trials for sub-jects who have undergone a carpal tunnel decompression.

JulySeptember 2006 323methods except in the flexor tendon group. However,it must be noted that the difference in SEMs withinthis subject group is 0.7 kg, which is felt to be clini-cally undetectable and subsequently of little clinicalsignificance.The results demonstrate that larger bandwidths of

error are associated with the mean of three trialsalthough this is again not seen in the flexor tendongroup. Despite the degree of error observed with themean of three trials being clinically acceptable, theSEM associated with the mean of three trials inthe asymptomatic group was appreciably higherthan the SEMs of the other methods of measurementat 4.84 kg. This level of measurement error has clearclinical implications with a relatively larger increasein actual grip strength required to detect a genuinechange or improvement.When comparing the present data to previous

studies, discrepancies can be observed in the findingsand conclusions drawn. Mathiowetz et al.14 foundthe mean of three trials achieved the highest levelof reliability and one trial achieved the lowest levelof reliability, the reverse of the findings of the currentstudy. Asymptomatic subject groups of similar sizewere used in both studies. However, Mathiowetzet al.s14 subject groupwas all female with a relativelynarrow age range (2039) when compared to themixed population with a wider age range used in

4444N =

Verb

al a

nalo

gue

scal

e sc

ore

4

38252

44282253 43

-1

0

1

2

3

4

5

TABLE 3. The Mean Grip Values and Standard Deviafor One Trial, the Mean of Thre

(Number of Cases 66)Asymptomatic

(Group 1)

One grip trial (years) 39.7 (SD 13.47)Mean of three trials 37.6 (SD 12.35)Highest of three trials 39.6 (SD 12.96)the present study, which may account for some ofthe differences observed. Mathiowetz et al.14 alsodid not differentiate between one trial and three,merely taking the value of the one trial to be the firstscore of the three trials. It is hypothesized that moti-vational and psychological factors may have an influ-ence, with subject effort increasing if they know theyonly have to do one squeeze. Mathiowetz et al.14 alsoretested their subjects within a week compared to30 minutes later in the present study. It could be ar-gued that the period of time between testing sessionscould be a possible cause for the disparity betweenthe findings of the two studies.The findings of the present study also differ from

those presented by MacDermid et al.16 in their studyon interrater reliability in an asymptomatic popula-tion. They concluded that due to the similarity inICCs demonstrated by both one single trial and themean of three trials, acceptable scores could beobtained using only one trial. However, they alsosupported the use of the mean of three trials wherepossible. Although, it must be noted that the studyby MacDermid et al.16 did not examine testretestreliability, rather interrater reliability.The results presented in the study by MacDermid

et al.16 do not support those obtained in the presentstudy. Again, their results for one trial were foundto have a lower level of reliability (ICC$ 0.94) thanthe best of three trials (ICC$ 0.95). However, the

0

2

6

4

8

10

12

Verb

al a

nalo

gue

scal

e sc

ore

Trials, and Best of Three Trials

Carpal TunnelDecompressions (Group 2)

Flexor TendonRepairs (Group 3)

18.8 (SD 11.65) 20.7 (SD 13.55)18.4 (SD 11.41) 18.8 (SD 12.60)19.3 (SD 11.17) 20.6 (SD 13.56)

trials in their symptomatic group. This is in contrast8difference in ICCs was not found to be statisticallysignificant. Thus, leading the authors to draw thesame conclusion as the present study that gripstrength scores obtained from one single trial willbe as reliable as those obtained from the mean ofthree trials. In a similar manner to Mathiowetzet al.,14 Hamilton et al.17 did not conduct a separatetrial to obtain a value for one trial but used the firstvalue of three trials to represent this value. This toocould be a source of error and explain the discrep-ancy in order of ranking in results of the currentstudy and that conducted by MacDermid et al.16

The testretest reliability of the mean of three trialsusing the Jamar dynamometer with subjects follow-ing carpal tunnel decompression has been previouslyexamined in one other study.22 Excellent correlationscan be observed for testretest reliability in the carpaltunnel group in the current study (ICC 0.98) and inthe study conducted by Stephens et al.22 (ICC 0.96).The SEM in the current study was lower at 1.6 kgcompared to 2.3 kg found by Stephens et al.22

The study conducted by MacDermid et al.16 in-cluded 23 (62%) subjects with carpal tunnel syn-drome (CTS). A distinction should be madebetween CTS and carpal tunnel decompression, asthe subjects with CTS would not necessarily have apalmer scar and it may be argued that those withCTS may have moremotor deficit from ongoing com-pression of the median nerve, which may alter gripstrength response and the variability of fatigue. Thesame cautions as detailed earlier must be applied tothe comparison of results with those of MacDermidet al.16 due to the nature of the study and the discrep-ancies in subject group with the present study.Subsequently, comparisons are made only tenuously.MacDermid et al.16 observed higher levels of reliabil-ity from the mean of three trials than the best of three

9

0

6

4

2

Verb

al a

nalo

gue

scal

e sc

ore

4444N = ONE TRIAL THREE TRIALS

Number of trials

FIGURE 4. Displaying the verbal analogue scale resultsfollowing one grip strength trial and three trials forsubjects post flexor tendon repair. Number 9 representsan outlier.

324 JOURNAL OF HAND THERAPYwith the current study where the best of three trialsdisplayed higher levels of reliability than the meanof three trials. The ICCs again are very similarbetween the studies.It is interesting to note that the majority of previous

studies undertaken in this area found that the meanof three trials was more stable over test occasions,displaying the highest testretest reliability. Clearlythe results of this current study conflict with theprevious findings. It is hypothesized that this couldbe due to the fact that this study, in contrast to manyprevious studies, also investigated symptomatic sub-jects, with pain experienced on gripping possiblybeing the reason for the difference in the findings.

Difference in Values of Grip StrengthsObtained

The mean of three trials consistently yielded thelowest value in kg of the three measurementmethods. The values for one trial and the highest ofthree trials showed close correlations with no signif-icant difference between their values for the asymp-tomatic and carpal tunnel decompression subjectgroups. Conversely, significant differences were ob-served between the values obtained for the mean ofthree trials compared to one single trial and thehighest of three trials. These differences can possiblybe explained from both a mathematical and physio-logical perspective. By definition, the highest score ofthree trials is always going to be the highest value.The mean value of three trials may yield the lowestvalues due to inter trial fatigue. Mathiowetz1 found aslight decrease in strength when comparing the thirdtrial to the first of three repeated trials. The corre-lation between the highest of three trials and onesingle trial may be due to the sequencing order ofwhen the highest score is performed in the threetrials. If the first trial is frequently the highest asfound by Bohannon and Andrews34 then it is proba-ble that one single trial may also produce a similarvalue.The mean grip strength values from the first

measurement session and the second session are verysimilar for all of the three subject groups. Thissuggests that there was either little or no fatigueeffect nor learning effect influencing the results.Differences have been observed in previous stud-ies,20,35 which has led previous authors to questiontheir methodology.

Pain Levels Associated with theOne Trial and Three Trials

Across the subject groups three trials producedstatistically significantly larger changes in verbalanalogue scale scores than one trial. As expected,

those subjects who have undergone a carpal tunnel recommendations made by the ASHT and thoseprotocols in common clinical practice. By only con-decompression experienced more pain on gripping

the Jamar dynamometer than the asymptomaticsubject group. These findings are in accordancewith those of Stephens et al.22

It is important to note that even though the subjectsin the asymptomatic group have no known upperlimb pathology, pain was reported by a substantialnumber of participants (50%). This brings into ques-tion the construct or internal validity of the Jamardynamometer as a tool with which to measure gripstrength. If a patient reports pain while performing agrip strength test using the Jamar dynamometer, onecannot be sure that the pain is due to the pathologyalone rather than discomfort elicited through simplegripping of the device. It is reasonable to expect thehypothesis that discomfort with grip testing mayresult in submaximal grip-force production to beapplicable to an asymptomatic as well as a sympto-matic population.

Limitations of the Study

The sample groups were selected for investigationbecause they both presented with palmar or digitalscars and hence were thought to be more likely toexperience pain on grip strength testing. Any con-clusions drawn from this study can only be appliedwith confidence to the two conditions examined,though further studies may find that they mayapply to conditions other than those tested here.The subject groups were also not matched for age orgender.No optimal recovery time following a maximal

isometric grip contraction has been documented.Therefore, the provision of 15 minutes between con-tractions has been given here following the protocolsused in other studies. This factor of fatigue may havehad an impact on the results. However, it is proposedthat by randomizing the order of tests this effectwould have been equal throughout the test protocols.Similarly, if one examines the results, there is nodecline in grip strength between the two measure-ment sessions making the likelihood of fatigue small.It would have been valuable to have noted which

of the three trials most frequently generated thehighest value. It has been previously establishedthat the first of multiple trials is usually the greatestin magnitude.34 Knowledge of this would assist in as-certaining why the values on one trial and the highestvalue of three trials show strong correlations.

Implications for Practice

Clinically, the results observed in this studywould indicate that one trial is as reliable as andless painful to perform than either the best of or themean of three trials. This inference challenges theducting one trial valuable clinical time would besaved and patients would be subjected to relativelyless discomfort.The pain perceived by the asymptomatic subject

groups indicates that the construct validity of theJamar dynamometer is questionable, in that theresults produced may not necessarily be a truemeasure of grip strength alone being possibly influ-enced by pain. This is particularly pertinent consid-ering the widespread use of the Jamar dynamometer.In light of 50% of subjects in the asymptomatic groupreporting pain while performing the grip strengthtest it should always be acknowledged that painreported during a grip test on the Jamar dynamom-eter may be in part to discomfort being elicited bygripping the device rather than solely the presentingpathology.

Implications for Further Research

It would be valuable to carry out a study toexamine the relationship between pain and gripstrength. This was not included in the present studyas it was a reliability study.Research also needs to be conducted with other

patient groups with different pathologies to evaluatethe most appropriate method of grip strength testingfor these populations. In the authors opinion thepain elicited by the Jamar dynamometer in theasymptomatic sample group suggests that there is areal need to adapt the Jamar dynamometer to makethe handles more comfortable or to develop a newtool, which fulfils these criteria. Previous studieshave examined various alternatives,22 but to dateno other tool has been widely accepted in the clinicalsetting.

CONCLUSIONS

The results of this study demonstrate high levels oftestretest reliability (ICC$ 0.85) for the threemethods of grip strength testing (one trial, the meanof three trials, and the best of three trials) while usingthe Jamar dynamometer with a standardized proto-col. The mean values of grip strength generated foreach method of grip testing produced comparable re-sults. A statistically significant difference (p. 0.0001)was highlighted in the verbal analogue scale scoresfollowing one trial and three in all three samplegroups.These findings indicate that performing one grip

strength trial is as reliable, and less painful toperform than either the best of or the mean of threetrials. This inference challenges the recommenda-tions made by the ASHT and those protocols in

JulySeptember 2006 325

common clinical practice. By only conducting onetrial valuable clinical time could be saved and pa-tients would be possibly subjected to less discomfort.Further research is necessary to provide verification.

REFERENCES

1. Mathiowetz V. Effects of three trials on grip and pinch strengthmeasurements. J Hand Ther. 1990;3:1958.

2. Marion R, Niebuhr B. Effect of warm-up prior to maximal gripcontractions. J Hand Ther. (JulySeptember)1992:1436.

3. MacDermid JC, Alyafi T, Richards RS, Roth JH. Testretestreliability of isometric grip strength and endurance grip testsperformed on the Jamar and NK devices. Physiother Can.2001;53(1):4854.

4. Innes E. Handgrip strength testing: a review of the literature.Aust Occup Ther J. 1999;46:12040.

16. MacDermid JC, Kramer JF, Woodbury MG, McFarlane RM,Roth JH. Inter rater reliability of pinch and grip strengthmeasurements in patients with cumulative trauma disorders.J Hand Ther. 1994;7:104.

17. Hamilton A, Balnave R, Adams R. Grip strength testing relia-bility. J Hand Ther. 1994;7:16370.

18. Stegink-Jansen CW, Simper VK, Stuart HG Jr, Pinkerton HM.Measurement of maximum voluntary pinch strength: effectsof forearm position and outcome score. J Hand Ther. 2003;16:32636.

19. Stratford P, Levy DR, Gauldie S, Levy K, Miseferi D. Extensorcarpi radialis tendonitis: a validation of selected outcome mea-sures. Physiother Can. 1987;39:2505.

20. Trossman PB, Li PW. The effect of inter trial rest periods on iso-metric grip strength performance in young adults. Occup TherJ Res. 1989;9:36278.

21. Fess EE. The need for reliability and validity in hand assess-ment instruments. J Hand Surg. 1986;11-A:6212.

22. Stephens JL, Pratt N, Michlovitz S. The reliability and validity

5. Gilbertson L, Barber-Lomax S. Power and pinch grip strength

recorded using the hand-held Jamar dynamometer and B&Lhydraulic pinch gauge: British normative data for adults. Br JOccup Ther. 1994;57(12):4838.

6. MacDermid JC, Richards RS, Donner A, Bellamy N, Roth JH.Responsiveness of the short form-36, disability of the arm,shoulder and hand questionnaire, patient rated wrist evalua-tion, and physical impairment measurements in evaluating re-covery after a distal radius fracture. J Hand Surg. 2000;25-A:33040.

7. Fess EE. Essential elements of an upper extremity battery. In:Hunter JM, Mackin EJ, Callahan AD (eds). Rehabilitation ofthe Hand: Surgery and Therapy. St. Louis: Mosby, 1995, pp185214.

8. Mathiowetz V. Reliability and validity of grip and pinchstrength measurements. Crit Rev Phys Rehabil Med. 1991;2:20112.

9. Richards L, Palmiter-Thomas PP. Grip strength measurement:a critical review of tools, methods and clinical utility. CritRev Phys Rehabil Med. 1996;8:87109.

10. Hammond R. Evaluation of physiotherapy by measuring out-come. Physiotherapy. 2000;86(4):1702.

11. Schreuders TA, Roebroeck ME, Goumans J, van Nieuwenhuij-zen JF, Stijnen TH, Stam HJ, et al. Measurement error in gripand pinch force measurements in patients with hand injuries.Phys Ther. (9)2003:83.

12. Fess EE. A method for checking Jamar dynamometer calibra-tion. J Hand Ther. 1987;1:2832.

13. Fess EE, Moran CA. American Society of Hand Therapists.Clinical Assessment Recommendations. Raleigh, NC: ASHT,1981.

14. Mathiowetz V. Reliability and validity of grip and pinchstrength evaluations. J Hand Surg. 1984;9-A:2226.

15. Bohannon RW, Saunders N. Hand-held dynamometry: a singletrial may be adequate for measuring muscle strength inhealthy individuals. Physiother Can. 1990;42:69.

326 JOURNAL OF HAND THERAPYof the Tekdyne hand dynamometer. Part 2. J Hand Ther. 1996;9(18):1826.

23. Niebuhr BR, Marion R, Fike ML. Reliability of grip strength as-sessment with the computerized Jamar dynamometer. OccupTher J Res. 1994;14:317.

24. Flood-Joy M, Mathiowetz V. Grip strength measurement: acomparison of three Jamar dynamometers. Occup Ther J Res.1987;7:23543.

25. Bear-Lehman J, Colon Abreu B. Evaluating the hand: issues inreliability and validity. Phys Ther. 1989;69(12):102533.

26. Von Korff M, JensenMP, Karoly P. Assessing global pain sever-ity by self-report in clinical and health services research. Spine.2000;25(24):314051.

27. Clark P, Lavielle P, Martinez H. Learning from pain scales:patient perspective. J Rheumatol. 2003;30:15848.

28. Briggs M, Closs JS. A descriptive study of the use of visual an-alogue scales and verbal rating scales for the assessment ofpostoperative pain in orthopaedic patients. J Pain SymptomManage. 1999;18:43846.

29. Sim J, Wright C. Research in Health Care. Concepts, Designsand Methods. Cheltenham: Stanley Thomas, 2000.

30. Gutierrez Z, Schechtman O. Effectiveness of the five-handleposition grip strength test in detecting sincerity of effort inmen and women. Am J Phys Med Rehabil. 2003;82(11):84755.

31. MacDermid JC, Evenhuis W, Louzon M. Inter-instrument reli-ability of pinch strength scores. J Hand Ther. 2001;14:3642.

32. Bland JM, Altman DG. Statistical methods for assessing agree-ment between two methods of clinical measurement. The Lan-cet. 1986;8:30710.

33. Portney LG, Watkins MP. Foundations of Clinical Research:Applications to Practice. Norwalk: Appleton & Lange, 1993.

34. Bohannon RW, Andrews AW. Inter rater reliability of hand-held dynamometry. Phys Ther. 1987;67:9313.

35. Shechtman O, Davenport R, Malcolm M, Nabavi D. Reliabilityand validity of the BTE-Primus grip tool. J Hand Ther. 2003;16:3642.

JHT Read for CreditQuiz: Article #038

Record your answers on the Return Answer Formfound on the tear-out coupon at the back of this is-sue. There is only one best answer for eachquestion.

#1. The results of this study compared to previousstudies were found to be:a. virtually identicalb. unrecognizablec. in frequent conflictd. in only occasional conflict

#2. In this study the reliability of the best of 3 trialswas than the mean of 3 trials:a. higherb. lowerc. the same

b. the mean of 3 trialsc. the 1st of 3 trialsd. 1 trial

#4. The pain level after 1 trial was thanafter 3 trialsa. the sameb. higherc. lowerd. none of the above

#5. (Prior to this study) the ASHTrecommends usingclinically:a. the best of 3 trialsb. the mean of 3 trialsc. the 1st of 3 trialsd. 1 triald. none of the above#3. The authors recommend using clinically:

a. the best of 3 trialsWhen submitting to the HTCC for re-certification,please batch your JHT RFC certificates in groupsof 3 or more to get full credit.JulySeptember 2006 327

The Reliability of One vs. Three Grip Trials in Symptomatic and Asymptomatic SubjectsOutline placeholderResearch Objectives

Research MethodologyTesting PositionTesterInstrumentsResearch DesignSample SizeStatistical Analysis

ResultsPain Levels Associated with the One Trial and Three Trials

DiscussionTest-Retest Reliability of the Three Measurement MethodsDifference in Values of Grip Strengths ObtainedPain Levels Associated with the One Trial and Three TrialsLimitations of the StudyImplications for PracticeImplications for Further Research

ConclusionsREFERENCESJHT Read for Credit