Table Of ContentMacionisBMCMusculoskeletalDisorders2013,14:17
http://www.biomedcentral.com/1471-2474/14/17
RESEARCH ARTICLE Open Access
Reliability of the standard goniometry and
diagrammatic recording of finger joint angles:
a comparative study with healthy subjects and
non-professional raters
Valdas Macionis1,2
Abstract
Background: Diagrammatic recording of finger joint angles by using two criss-crossedpaper stripscan be a quick
substitute to the standard goniometry. As a preliminary step toward clinical validationof the diagrammatic
technique, thecurrentstudy employed healthy subjects and non-professional raters to explore whether reliability
estimates of the diagrammatic goniometry are comparable with those of thestandard procedure.
Methods: The study included two procedurally different parts, which were replicatedby assigning24 medical
students to act interchangeably as 12 subjects and 12 raters. A larger component ofthe study was designed to
compare goniometers side-by-side inmeasurement of finger joint angles varying from subjectto subject. In the
restof the study, theinstrumentswere compared by parallel evaluations of joint angles similar for all subjects in a
situation ofsimulated changeof joint range of motion over time. The subjectsused special guides to position the
jointsof their left ringfinger at varying angles of flexion and extension.The obtained diagrams of joint angles were
converted to numericalvalues by computerizedmeasurements. The statistical approachesincluded calculation of
appropriate intraclass correlationcoefficients, standard errors of measurements, proportionsof measurement
differences of5 or less degrees, and significant differences between paired observations.
Results: Reliability estimates were similar for both goniometers. Intra-raterand inter-rater intraclass correlation
coefficients ranged from 0.69 to 0.93. The corresponding standard errors of measurements ranged from 2.4to 4.9
degrees. Repeated measurements ofa considerable number of raters fell within clinically non-meaningful5 degrees
ofeach other inproportions comparablewith a criterion value of 0.95. Data collected with both instrumentscould
be similarly interpreted ina simulated situationof change of joint range ofmotion over time.
Conclusions: The paper goniometer and the standard goniometercan be used interchangeably by
non-professional raters for evaluation ofnormal finger joints. The obtained results warrant further research to assess
clinical performance of thepaper strip technique.
Correspondence:[email protected]
1ClinicofRheumatology,OrthopaedicsandTraumatology,and
ReconstructiveSurgery,VilniusUniversityFacultyofMedicine,Vilnius,
Lithuania
2DepartmentofPlasticandReconstructiveSurgery,CentreofPlasticand
ReconstructiveSurgery,VilniusUniversityHospitalSantariskiuKlinikos,Vilnius,
Lithuania
©2013Macionis;licenseeBioMedCentralLtd.ThisisanOpenAccessarticledistributedunderthetermsoftheCreative
CommonsAttributionLicense(http://creativecommons.org/licenses/by/2.0),whichpermitsunrestricteduse,distribution,and
reproductioninanymedium,providedtheoriginalworkisproperlycited.
MacionisBMCMusculoskeletalDisorders2013,14:17 Page2of11
http://www.biomedcentral.com/1471-2474/14/17
Background objectives of the present investigation, this study chose
Graphical presentation of finger range of motion by to include non-professional evaluators with a similar
means of malleable wire tracing is a recognized adjunct medicalbackground.
to the standard goniometry [1]. This method, however,
has been shown to be of inadequate reliability [2]. The Ethicsstatement
range of motion of the finger joints can also be diagram- The study was approved by Vilnius regional ethics
matically visualized by tracing the arms of an ad hoc committee for biomedical research. Written informed
goniometer obtained by criss-crossing two folded paper consents of the participants were obtained before the
strips [3]. This simple tool can be a quick substitute to study.
the standard goniometer in clinical situations when the
latter is unavailable and allows evaluation of finger joint Participantsandstudydesign
positions, where application of the standard goniometer Twenty-four healthy, third- and fourth-year medical stu-
is impossible (Additional file 1). It has been suggested dentswereincludedin the study inthe order oftheir re-
that the performance of the diagrammatic goniometry sponse to an advertisement inviting to participate in a
should be comparable with that of the standard pro- goniometric reliability study for reimbursement. The key
cedure because both methods are technically similar. criterion in selecting the participants was similarity of
As an initial step to test this supposition, the current their academic and practical background. None of the
investigation determined measurement reliability [4,5] participants had considerable experience in goniometry;
for the paper and standard goniometer in non-clinical however, all of them were familiar with the concept of
imitated situations when there was no change in finger goniometry through their earlier study. The age of the
range of motion and when the range of motion changed participants ranged from 20 to 24 years. Additionally, 2
overtime. fourth-year medical students were invited to participate
in a separate reliability study of computerized evaluation
Methods of jointanglediagrams.
Rationaleofthestudydesign The 24 participants were randomly assigned to the rater
Search for apossible researchmodel revealed 28reliabil- orsubjectgroup,eachincluding12people.Theraterswere
ity studies involving standard finger goniometry [2,6-37]. randomly subdivided into subgroups of 10 and 2 to per-
A larger part of the studies was carried out on subjects form different tasks. The study consisted of 2 procedurally
with normal hands [2,6-10,14-16,23,26,29-31,33,35,36]. identical replicate stages, stage I and stage II. In the study
Taking into account the novelty of the diagrammatic stage II the participants changed their roles (Figure 1,
technique and difficulties in carrying out a comparative Additionalfile2).Eachreplicatestageofthestudyincluded
reliability study of a considerable extent on patients, the procedurally different parts A and B according to the sub-
current exploration opted for healthy subjects as well. divisionoftheratersintosubgroupsof10and2.Thus,the
The present investigation chose static finger position studyincludedreplicatepartsI-AandII-Awith10different
model, since in healthy subjects, only a few finger joint ratersineachandreplicatepartsI-BandII-Bwith2differ-
postures can be obtained by using standard types of mo- entratersineach.Allratersofthesamestageevaluatedthe
tion [9,12,14,33,36]. Previous researchers ensured stabil- same remaining 12 participants acting as subjects under
ity of the desired finger positions by employing various evaluation. The replicate study parts A were designed to
palmar blocks [7,10,16,29,30,35] and splints [2,15,23,26]. compare the goniometers side-by-side in measurement of
Due to skin mobility and suppleness, however, it seems the metacarpophalangeal (MCP), proximal interphalangeal
difficult to achieve steadiness of the joint angle with pal- (PIP), and distal interphalangeal (DIP) joints set at angles
mar supports alone. The use of hand cast [24], transarti- varyingfromsubjecttosubject(Additionalfile2).Therep-
cular pinning of the cadaver finger joints in various licate study parts B were planned to compare the instru-
degrees of flexion [31], and wooden finger joints [35] mentsbyparallelevaluationsofthePIPjointanglessimilar
is arguably too artificial. Only one study used change for all subjects in a situation of simulated change of joint
of finger motion due to a treatment to test inter- rangeofmotionovertime(Additionalfile2).
goniometer reliability [6]. However, none of the static
models have been employed for this purpose. The Equipment
current study designed a stabilization system for fingers For joint angle measurement, the study employed the
taking into account the experience and limitations of the improvisedpapergoniometer(twoapproximately10.5cm
previous explorations. Earlier method comparisons, as a by 3.8 cmrectangularpaper stripsobtainedbyfolding A4
rule, involved professional raters who must have been paper sheets) and a standard flexion-hyperextension plas-
more experienced with one of the techniques under tictransparentfingergoniometer(JamarE-ZRead)gradu-
evaluation. Therefore, and considering the extent of the atedin1degreeincrements(Figure2).Aplasticcoverwas
MacionisBMCMusculoskeletalDisorders2013,14:17 Page3of11
http://www.biomedcentral.com/1471-2474/14/17
Study part I-A I-B
15-20 W-1 W-3 W-5 W-7 W-9 W-11
min. W-2 W-4 W-6 W-8 W-10 W-12
e I5 min.
g
a W-1 W-3 W-5 W-7 W-9 W-11
St W-2 W-4 W-6 W-8 W-10 W-12
and so on and so on
W-1 W-3 W-5 W-7 W-9 W-11
W-2 W-4 W-6 W-8 W-10 W-12
1 h Break
Study part II-A II-B
Z-2 Z-4 Z-6 Z-8 Z-10 Z-12
Z-1 Z-3 Z-5 Z-7 Z-9 Z-11
e II
g
a Z-2 Z-4 Z-6 Z-8 Z-10 Z-12
St Z-1 Z-3 Z-5 Z-7 Z-9 Z-11
and so on and so on
Z-2 Z-4 Z-6 Z-8 Z-10 Z-12
Z-1 Z-3 Z-5 Z-7 Z-9 Z-11
Stations where individual sets of 6 try-angles were used. Evaluation of
MCP, PIP, & DIP joints of the left ring finger in 2 positions (1 for imitated
flexion and 1 for extension; each measured with 2 instruments in 2 trials;
angles, i.e. sub-positions, varied accross subjects)
Stations where shared sets of 12 try-angles were used. Evaluation of
PIP joint of the left ring finger in 2 positions (6 angles, i.e. sub-positions, for
imitated flexion and 6 for extension; each measured with 2 instruments in 1
trial only; angles the same for all subjects)
subject'sID Red font rater'sID(permanentposition)
Direction of subjects' movement for the next measurement session
Figure1Schemeofthestudy.
used to mask the pointer of the goniometer during the individualsetscontained3subsetsof2try-angles,onepair
evaluations. The raters entered the measurements into for each of the finger joints to imitate position of incom-
blanks unique for each rater-subject combination pair. plete flexion and extension. Similarly, in the shared sets
Plastic funnels and triangle rulers were used as supports there were 2 subsets of 6 try-angles, one subset for each
forsubjects’fingers.Tosetthefingerjointsinappropriate of the positions of imitated extension and flexion of the
postures,custommadewoodentry-angles(try-squaretype PIP joint only. The angles of the try-angles (or standard
guides) were applied over the dorsal aspect of the joint angles) were varied to produce different sub-positions of
(Figure 2a). There were 12 individual sets of 6 try-angles flexion and extension. Importantly, each of the 2 subsets
distributed to each subject and 2 shared sets of 12 try- of 6 try-angles in the shared sets enabled 6 different sub-
angles to be used by all subjects (Additional file 2). The positionsofthePIPjoint(Additionalfiles2and3).
Figure2(A,B)Simplifiedillustrationoftheevaluationstationandprocedure.(A)Stabilizationofsubject’shandbygraspingafunneland
applicationofatry-angletoobtainanappropriatepositionofthePIPjoint.(C)Evaluationoftheobtainedjointposturewiththepaperstripgoniometer.
MacionisBMCMusculoskeletalDisorders2013,14:17 Page4of11
http://www.biomedcentral.com/1471-2474/14/17
Procedures possible to the position of the anatomical axes of the ap-
Preparatoryprocedures propriate bones. Dorsal method of placement was used
A pilot exploration employing a healthy subject and 17 for both instruments. After aligning the arms of the
raters was performed to elucidate possible technical pro- standard goniometer, the rater removed the cover from
blems ofthestudy. the pointer and read the value together with the subject
Two weeks before the study, the participants were sent to exclude reading errors; the obtained value was en-
step-by-step instructions with the appropriate images of tered into the blank. The angles, obtained by proper
the procedure and the equipment. At least a week before alignment of the paper strips, were drawn onto the ap-
the study, the equipment and procedures were demon- propriate sections of the blanks by using edges of the
stratedtotheparticipantslive.Exampletry-angles,triangle paper strips as rulers. If the arms of the standard goni-
rulers, and paper strips were distributed for individual ometer or paper strips were inadvertently displaced du-
trainingathome.Takingintoaccounttheunusualmanipu- ringthe evaluation,themeasurementwasrepeated.
lative task of the diagrammatic goniometry, the partici- The procedure protocol also included relaxation of the
pants learned to copy printed angles by using the paper subject’s hand between the measurements and short
goniometerindividuallyorasparticipantsofanotherstudy. breaks between the evaluation sessions. As the length of
Two days before the study, the participants were required the evaluation sessions differed from rater to rater, the
toanswerashortquiztestingtheknowledgeoftheirtasks intervals between sessions also varied. The participants
inthestudy. were freetochooselongerbrakesifthey felttired.
Proceduresonthedayofstudy
The study was conducted in a spacious auditorium. The Evaluationofdiagrams
raters and subjects faced each other across a long narrow All the blanks withthe recorded angles of the joints were
table and sat along the table sides in checkmate pattern. scanned.Thescanneddiagramsweremagnified,andtheir
The raters’ locations were permanent, while the subjects, angles were measured to the nearest degree by the same
having completed an evaluation session, moved along the researcher with ImageJ program. Each diagram was mea-
table sides clockwise bypassing the neighboring raters to sured at least twice without reference to the previous
beevaluatedbythenextrateracrossthetable(Figure1). results. If the results of the two computerized measure-
The subjects’ task in all study parts was to stabilize ments were different, the diagram was remeasured again.
their left ring finger joints in postures set up by grasping If 2 identical measurements were not obtained, mean of
a funnel or a triangle ruler and by applying appropriate themeasurementswasfoundandroundedofftothenear-
try-anglesoverthedorsalaspectofthejoint(Figure 2a). est degree. To assess intra-rater and inter-rater reliability
In the replicate study partsA, the subjectsused their in- of the latter procedure, two invited medical students
dividualtry-anglesetsatthe10evaluationstations(Figure1, remeasured48randomlychosenscanneddiagrams.Com-
Additionalfile2).Thevaluesoftheanglesofindividualtry- puterized evaluation instead of a simple use of a trad-
angles were randomly distributed across the finger joints itional protractor was chosen to equalize varying sizes of
and across the subjects. The angles of the individual try- the hand drawn diagrams and to avoid errors of hand-
anglesofthesamesubjectwereofdifferentmagnitude,and donemeasurements.
noneofthesubjectshadthesamecombinationoftheangle
magnitudes (Additional file 3). Raters of the study parts A
had to obtain twice the MCP, PIP, and DIP joint angles in Independenceofobservations
each of the two positions (flexion and extension) by using Although dependency of observations is inherent to
bothgoniometers(Additionalfile2). within-subject designs, care was taken to ensure the
In the replicate study parts B, the subjects employed required rater related independence of measurements
thesharedtry-anglesetspermanentlyavailableattheap- [5,38]. To prevent any form of communication of the
propriate 2 evaluation stations (Figure 1, Additional file obtained angles, the current study design included
2). Both shared sets were almost identical in the magni- checkmate arrangement of the subject and rater pairs,
tude of the standard angles; however, the order of the alternating use of the instruments, proceeding to trial 2
try-angles in the sets was different (Additional file 3). only after completing trial 1 for all joints and both
The task of the two raters of the study parts B included instruments, irregular arrangement of the standard angle
only evaluation of the PIP joint in the 6 sub-positions of magnitudes, and masking the pointer of the standard
eachofthetwo positionswithboth instrumentsina sin- goniometer. Also, the participants were not allowed to
gletrial (Additionalfile2). share the results of their measurements and were made
When evaluating the joints, the raters were required aware that the standard angles varied widely across the
to do their best to align the instrument arms as close as subjectsandjoints.
MacionisBMCMusculoskeletalDisorders2013,14:17 Page5of11
http://www.biomedcentral.com/1471-2474/14/17
Statisticalapproaches synchrony of the evaluation sessions. Ten raters were ex-
Measuresofreliability pectedtoperform2repeatedmeasurements(trial1,trial2)
Measurement reliability has been expressed in relative ofthesamejointinthesamepositionwiththesameinstru-
and absolute measures [39,40]. In the current study, the ment, which summed up to 20 observations per subject.
reliability term was used as a hypernym for expressions TheICCswereexpectedtoreach0.9.However,takinginto
defining various aspects of measurement uncertainty account the conventionally acceptable lowest ICC values
[39,40], although some authors have used agreement [38], reliabilities of 0.7-0.75 could also be considered as
term for this purpose [41] or have understood reliability adequate for non-professional raters. Using an earlier pro-
inanarrowersense [4,5,41]. posed formula [48] with the above values and the recom-
For continuous variables, the most common measure of mendedlevelsofα=0.05andβ=0.2resultedinsamplesizes
relative reliability is intraclass correlation coefficient (ICC) between8and12(Additionalfile4).
accompaniedbyappropriateanalysisofvariance(ANOVA)
[40]. Differently from the previous studies, which used DataAnalysis
the popular models of ICCs described by Shrout and Each of the replicate study stages was analyzed separ-
Fleiss [42], the current investigation employed concurrent ately.Thesignificance levelwassetat p<0.05.
assessment of reliability proposed by Eliasziw et al. [43].
Unlike calculating the traditional ICCs, the method of Exploratorydataanalysis
concurrent assessment allows simultaneous estimation of Exploratory data analysis included obtaining descriptive
intra-raterandinter-raterreliabilityalongwiththehypoth- statistics, searching for outliers, and assessing the nor-
esistestingincaseswhenmultipleratersevaluatemultiple mality of distribution of the appropriate data sets by
subjects and perform more than one measurement per means of Shapiro-Wilk tests and the analysis of histo-
subject. In respect to the traditional models, the concur- gramsandQ-Q plots.
rent methodology has been cited as a more advantageous
approach[44]. AnalysisofthestudypartsA
For the clinician, however, reliability coefficients are less InthereplicatestudypartsA,2x2x10(trialxgoniometerx
important than measures of absolute reliability like the rater) and 2x10 (goniometer x rater) repeated measures
standarderrorofmeasurement(SEM),which(whenmulti- ANOVAs were run for each position-joint and trial-
plied by 1.96) indicates how far from the hypothetical true position–joint data set, respectively, to assess the main
value [38,39,45] the measurement obtained by a practi- effectsand interactionsofgoniometer, trial, and rater. The
tioner could be [40]. The SEM enables derivation of other sphericity assumption was tested by using Mauchly’s test
measures of absolute reliability including the limits of withappropriateepsilonadjustments.
agreement[46]andtheminimaldetectablechange(MDC). For concurrent assessment of reliability, the pertinent
The MDC defines the difference that should be obtained mean squares were found by running two-factorial uni-
between 2 successive measurements on the same subject variate ANOVAs [43]. Subject and rater were random
overtimetostatethattherealchangehasoccurred.Inthis effects because the participants were selected randomly
studythe MDC, also referred to asminimal detectabledif- and there was no interest in particular raters. Homogen-
ference [38] or repeatability coefficient [45,47], was found eity of variances was tested with Leven’s test. The neces-
byusingformulaMDC=SEMx1.96x√2[4,40,45]. sary variance components were calculated using the
Additionally, following a previous suggestion [41], the obtained mean squares. The intra-rater and inter-rater
current study employed intuitive descriptive approaches. ICCs, their lower limits of 95% one-sided lower-limit
To facilitate interpretation of goniometric reliability, confidence intervals (LLs of 95% one-sided L-L CI), and
proportions of clinically non-meaningful ≤5-degree dif- SEMs were simultaneously calculated across all raters for
ferences between repeated measurements (here also each goniometer-position-joint data set. Following the
named “≤ 5-degree agreement”) were analysed [4]. Also, methodology for concurrent assessment of reliability [43]
in the smaller B component of this study, mean meas- and previous suggestions regarding meaningful ICC
urement differences and their standard deviations were values [38], the null hypothesis was that the ICCs were
employedtoreflectabsolute reliability [38,46]. less than or equal to 0.75, and the alternative hypothesis
wasthattheICCswouldbemorethan0.75.Thenullhy-
Samplesizeestimation pothesis was considered rejected, if the LLs of 95% one-
The main attention in this investigation was directed to- sidedL-LCIfortheICCswerelessthanorequalto0.75.
wards calculating intra-rater and inter-rater ICCs and Computationalgorithmsforconcurrentassessmentofre-
SEMs in the study parts A. The other components of the liabilityarepresentedintheAdditionalfile5.
study were designed as pilot investigations. Balanced For further reflection of intra-goniometer (i.e., intra-
numbers of subjects and raters were planned to ensure rater ) reliability, proportions ofclinicallynon-meaningful
MacionisBMCMusculoskeletalDisorders2013,14:17 Page6of11
http://www.biomedcentral.com/1471-2474/14/17
≤5-degreedifferencesbetweenthemeasurementsobtained the data is reflected in Additional file 8. In the study parts
with the same tool in the 2 trials were calculated for each A, the data arranged in trial-joint-position-goniometer sets
rater.Similarly,fortheassessmentofinter-goniometerreli- included 10 outliers with standard scores above 3.0
ability,proportionsof≤5-degreedifferencesbetweenmea- (Figure 3). The outliers were retained for the analysis to
surements of the same rater with different instruments preserve sufficient sample size. In the study part II-A, one
within the same trial were found. The observed propor- rater failed to perform 2 measurements with the standard
tionsofthe≤5-degreedifferencesweretestedagainstpro- goniometer, which necessitated sample size reduction of
portion of 0.95 for statistical significance by one sample the appropriate subgroups. Normality of distribution could
binomialtests.Thereferencevaluewasestimatedbycalcu- be assumed for almost 97% of the data sets of the study
lating the LL of 99% CI for population proportion [49] parts A arranged by the raters’ individual measurements.
using the largest previously employed sample sizes reach- Although larger data aggregates failed Shapiro-Wilk test,
ing 60 [32] and a generous assumption that the earlier normality could be assumed by analyzing the appropriate
sample proportion of ≤5-degree measurement differences histograms and Q-Q plots. Therefore, having confirmation
was 0.99. Counts of the raters who passed the binomial ofhomogeneityoftheappropriatevariancesbyLeven’stest,
testswereobtainedforintuitivecomparison.Toassessthe the analysis was continued with parametric tests based on
inter-goniometer ≤ 5-degree agreement, only the raters robust ANOVA [50]. In the study parts B, Shapiro-Wilk
who passed the binomial test in both trials were included. testconfirmednormaldistributioninupto90%ofthesets
Additionally,thebestraterswereselectedbymatchingthe of the differences between the appropriate subgroups of
individual successful raters across the three ≤ 5-degree measurements.
agreementsubgroups(i.e.,acrosstheinter-goniometerand
thetwointra-goniometersubgroups).
ResultsofthestudypartsA
AnalysisofthestudypartsB The repeated measures 2x2x10 ANOVAs revealed that
Tofindwhetherthetry-angleguidessignificantlychanged the main effect of goniometer was insignificant. The
the observed angles of the PIP joint, multiple Wilcoxon main effect of trial was significant for the MCP joint in
signed-rank tests with Bonferroni correction were per- imitated extension in study part I-A and in all studypart
formed for each rater-instrument-position-subposition II-A subgroups. The main effect of rater was significant
data setin respect tothebaseline joint angles obtained by for the MCP joint in study part I-A and in all study part
usingthe smallest standardangles. Then the standard dif- II-A subgroups. Trial by rater interaction effect was
ferences between the angles of the appropriate try-angles observed in all the subgroups except for that of the DIP
(i.e., between the standard angles) were calculated in re- joint in position of imitated flexion. Goniometer by trial
spectofthesmalleststandardangles.Next,thelowestsig- interaction was observed only in the DIP joint extension
nificant standard differences were found between the subgroup of the study part II-A. The 2x10 ANOVAs
smallest standard angles and the angles of the try-angles, showed that goniometer and rater effects were insignifi-
application of which produced significant changes in the cant in approximately half of the trial-position-joint data
observed PIP joint angles (Additional file 6). The lowest sets. Most of the two-way ANOVAs resulted in signifi-
significant standard differences were compared with each cant goniometer byrater interaction.Insignificance ofall
other and with the corresponding values of the MDC effects was observed only in the study part I-A, for the
derivedfromtheSEMsofthestudypartsA. first trial measurements of the DIP joint and for the sec-
ondtrialmeasurementsofthePIPjointinflexion.
Analysisofreliabilityofthediagramevaluation Concurrent assessment of intra-rater and inter-rater re-
Intra-rater (inter-trial) and inter-rater (intra-trial) reliability liability showed that both methods have similar reliability
of the computerized measurements of diagrams was as- parameters, which, however, tended to be slightly higher
sessed by calculating mean differences between the appro- forthestandardgoniometer(Table1).Inthestatisticalhy-
priatepairsofmeasurementsandtheirstandarddeviations. pothesistesting,mostoftheLLsof95%one-sidedL-LCIs
for the ICCs were above 0.75. In 5 out of 8 instances
Results where the paper goniometer failed the test, the standard
Resultsofexploratorydataanalysis goniometer performed similarly. In the other three cases
The data available for the analysis included 5758 measure- of failure to reject the null hypothesis for paper goniom-
ments from the study parts A and 1152 measurements eter,theLLsof95%one-sidedL-LCIswereabove0.7.All
from the study parts B. Additional file 7 presents the raw ICCs and SEMs for the MCP joint tended to be superior
data of the study to enable rerun of the analysis and thus to the corresponding estimates for the interphalangeal
facilitate interpretation of the findings obtained by the un- joints.Intra-raterICCsandSEMswerehigherthancorre-
commonstatisticalapproaches.Thedescriptivestatisticsof spondinginter-raterreliabilitymeasures.
MacionisBMCMusculoskeletalDisorders2013,14:17 Page7of11
http://www.biomedcentral.com/1471-2474/14/17
Figure3BoxplotsofthejointanglemeasurementsobtainedinthestudypartsA.TR=trial;PGn=paperstripgoniometer;
SGn=standardfingergoniometer;MCP=metacarpophalangealjoint;PIP=proximalinterphalangealjoint;DIP=distalinterphalangealjoint;
EXT=positionofimitatedextension;FLEX=positionofimitatedflexion.
The results of the binomial tests for significance of agreement test. The relative increase in the number of
observed proportions of the clinically non-meaningful raters who passed the binomial test for the inter-
differences of ≤ 5 degrees are illustrated in Figure 4. The goniometer ≤ 5-degree agreement was due to the
number of raters whose repeated measurements fell instances where the individual intra-goniometer inter-
within ≤ 5 degrees of each other in proportions compa- trial differences exceeded 5 degrees for both instru-
rablewiththe criterionvalueof0.95wassimilarforboth ments, but the inter-goniometer intra-trial differences of
tools. In all joint and position subgroups except for the same measurements were within 5 degrees. Very few
that of MCP extension, slightly more raters passed the raters passed the binominal tests for both the intra-
inter-goniometer than the intra-goniometer ≤ 5-degree goniometer and inter-goniometer ≤ 5-degree agreement.
Table1ReliabilityestimatesobtainedinthestudypartsA
Position,Studypart, ICC(LLof95%one-sidedL-LCI) SEMindegrees
Characteristics,
MCP PIP DIP MCP PIP DIP
Goniometer
EXT,I-A,Intra-R,PGn 0.88(0.87) 0.84(0.82) 0.89(0.88) 3.2 4.1 3.5
SGn 0.89(0.885) 0.86(0.84) 0.91(0.90) 3.1 4.2 3.3
II-A,Intra-R,PGn 0.90(0.89) 0.86(0.85) 0.85(0.83) 2.8 3.3 3.8
SGn 0.89(0.88) 0.90(0.89) 0.87(0.865) 2.9 3.3 3.6
I-A,Inter-R,PGn 0.87(0.78) 0.78(0.65)* 0.85(0.76) 3.3 4.8 4.1
SGn 0.86(0.77) 0.80(0.69)* 0.88(0.79) 3.5 4.9 3.8
II-A,Inter-R,PGn 0.86(0.77) 0.83(0.72)* 0.82(0.71)* 3.2 3.8 4.1
SGn 0.87(0.78) 0.84(0.74)* 0.86(0.77) 3.2 4.0 3.8
FLEX.I-A,Intra-R,PGn 0.89(0.88) 0.86(0.85) 0.83(0.81) 2.8 4.2 4.3
SGn 0.91(0.90) 0.89(0.88) 0.86(0.85) 2.4 3.6 3.8
II-A,Intra-R,PGn 0.90(0.89) 0.85(0.82) 0.82(0.77) 3.2 3.4 3.8
SGn 0.93(0.92) 0.87(0.85) 0.85(0.82) 2.8 3.2 3.4
I-A,Inter-R,PGn 0.87(0.78) 0.83(0.73)* 0.78(0.66)* 3.1 4.6 4.9
SGn 0.86(0.76) 0.86(0.76) 0.83(0.73)* 3.0 4.1 4.2
II-A,Inter-R,PGn 0.83(0.72)* 0.76(0.62)* 0.69(0.54)* 4.2 4.3 4.9
SGn 0.88(0.80) 0.80(0.67)* 0.75(0.61)* 3.5 3.9 4.4
ICC=intraclasscorrelationcoefficient;LL=lowerlimit;CI=confidenceinterval;SEM=standarderrorofmeasurement;EXT,FLEX=positionofimitated
incompleteextensionandflexion,respectively;R=rater;PGn=papergoniometer;SGn=standardfingergoniometer;*nullhypothesisretained.
MacionisBMCMusculoskeletalDisorders2013,14:17 Page8of11
http://www.biomedcentral.com/1471-2474/14/17
Figure4(A,B)Summaryofonesamplebinomialtestsfortheproportionsofmeasurementdifferencesof≤5°.N/S=notsignificant;
SGn=standardfingergoniometer;PGn=paperstripgoniometer;Gn=goniometer.Thepartsofthebarsbelowthehorizontallinesrepresent
stageIofthestudy.
The results of the binomial tests also showed that the standard deviations of the mean differences were below
MCP joints were evaluated more precisely than the 0.7 degrees. All measurement differences were within 1
interphalangealarticulations. degree except for one occasion for each of the invited
The results of the study part II-A tended to be slightly raters, where their trial-to-trial measurements disagreed
worse than those ofthe studypart I-A. by2degrees.
ResultsofthestudypartsB Discussion
According to the multiple paired Wilcoxon tests, a In the current study, reliability of diagrammatic and
significant change in the PIP joint angle was mostly standard finger goniometry was assessed by employing a
observed after application of the try-angles differing repeated measures design with replication, in which
from the baseline angle at least by 9 degrees non-professional participants acted as raters and sub-
(Table 2). The lowest standard significant differences jects. The diagrams of the joint angles were converted to
were similar for both goniometers and raters. The numerical values by computerized angle measurements.
obtained lowest standard significant differences were The measurement errors due to the conversion were
comparable to the corresponding MDCs from the below 0.7 degrees, which is not substantial in terms of
study parts A. the clinicallyacceptable5-degree error.
The results of all the analytical approaches support
Reliabilityofthediagramevaluation the suggestion that both goniometers can be used inter-
Themeanintra-rater(inter-trial)andinter-rater(intra-trial) changeably. Significance of goniometer effect apparent
differencesofthecomputerizedmeasurementsofthejoint fromsomeofthe2x10ANOVAsshouldbeinterpretedin
angle diagrams ranged from - 0.1 to 0.21 degrees. The conjunction with significant goniometer by rater inter-
meanabsolutedifferencesdidnotexceed0.4degrees.The action, indicating that the performance of the instrument
Table2Comparisonoftheminimaldetectablechangeswiththeloweststandardsignificantdifferences*
Position, MDC Lowestsignificantstandarddifference(indegrees)
Goniometer
I-A II-A Rater1(I-B) Rater2(I-B) Rater3(II-B) Rater4(II-Bcpa)
EXT,PGn 11.5 9.3 14 9 9 9
SGn 11.6 9.0 9 9 9 9
FLEX,PGn 9.5 11.5 9 9 9 9
SGn 9.9 8.7 13 9 9 5
MDC=minimaldetectablechangesobtainedfromthecorrespondingSEMsinTable1;I-A,II-A,I-B,andII-B=appropriatestudyparts;EXT,FLEX=positionof
imitatedincompleteextensionandflexion,respectively;PGn=papergoniometer;SGn=standardfingergoniometer;*thelowestdifferencebetweentheangles
ofthetry-anglesapplicationofwhichresultedinsignificantdifferencesbetweenthecorrespondingobservedPIPjointanglesaccordingtoWilcoxonsignedranks
test(seeAdditionalfile6fordetails).
MacionisBMCMusculoskeletalDisorders2013,14:17 Page9of11
http://www.biomedcentral.com/1471-2474/14/17
tended to be depended on which the rater was using it. the proportion analysis of the inter-goniometer ≤ 5-degree
The small magnitudesofthe differences betweenthe reli- differences. Performing the procedures in open stations
ability estimates of the techniques were not convincing may be regarded as a violation of independence of mea-
enough to state disparity of the methods. In the three surements, which, however, is unlikely to be substantial
cases of failure to reject the null hypothesis for paper considering the study design features listed in the related
goniometeralone,theLLsof95%one-sidedL-LCIslevels sectionabove.
above 0.7 can still be considered as an acceptable level of
reliabilityfornon-professionalnoviceraters.Interchange-
Conclusions
ability of goniometers was also demonstrated by the
It can be concluded that that the paper goniometer and
binomial tests, which involved assessment of the
inter-goniometer ≤ 5-degree agreement. It is notable, the standard goniometer can be used interchangeably by
non-professionalratersfortheevaluationofnormalfinger
that the results of the proportion analysis echo the
joints.Theobtainedresultswarrantfurtherresearchtoas-
outcomes of parametric assessments indicating that the
sessclinicalperformanceofthepaperstriptechnique.
measurement consistency was rater and joint dependent.
Parityof thegoniometerswasfurthershown bytheresults
ofthestudypartsB,indicatingthatdatacollectedwithboth Additional files
instruments can be similarly interpreted in an exploration
of simulated change in joint range of motion over Additionalfile1:Anadvantageofpaperstriptechniqueover
time. Decrease in the reliability estimates in the second standardgoniometry.ThisadditionalfileincludesFigureAshowing
situationwhenproperalignmentofthestandardfingergoniometeris
stage of the study part A may be due to the weariness of
impossibleandFigureBdemonstratingsolutionoftheproblemby
theparticipants. meansofthepaperstriptechnique.
Straightforward comparison of the obtained results Additionalfile2:Datacollectiondesign.Thisadditionalfilereflects
with those of the other explorations is complicated, as thekeyfeaturesofthestudydesignandarrangementofthetry-anglesin
thesets.
reliability studies differ in technical and statistical as-
Additionalfile3:Standardangles.Thisadditionalfileincludesangles
pects[39].Somemethodologicalissuesoftheearlierstud-
ofthetry-anglesandcalculationofthestandarddifferences.
iesoffingergoniometrywereaddressedintherationaleof
Additionalfile4:Algorithmforsamplesizecalculation.This
the study design. A more detailed reflection of the design additionalfileincludesacalculationalgorithmbasedontheformula
diversity and results of the previous explorations is given describedbyWalteratal.[48].
inAdditionalfile9.Mostoftheintra-raterandinter-rater Additionalfile5:Algorithmsforconcurrentassessmentofintra-
raterandinter-raterreliability.Thisadditionalfilecontainsthe
ICCs obtained inthe current studywereabove 0.8,which
followingworksheets.ConcurrentassessmalgorithmFx.Thisworksheet
indicates reliability [38] comparable with the previously includesanalgorithmforcalculationofinter-raterandintra-raterICCs
reportedvalues[6,10,11,17-20,25,27,29,33,37].Mostofthe andSEMsforthecaseoffixedratereffectsusingtheformulaedescribed
byEliasziwetal.[43];ConcurrentassessmalgorithmR.Thisworksheet
SEMs obtained in the current study are also in compari-
includesanalgorithmforcalculationofinter-raterandintra-raterICCs
son with the corresponding estimates reported by the andSEMsforthecaseofrandomratereffectsusingtheformulae
earlier researchers [9,29,33,37]. The SEM exceeding 1.8 describedbyEliasziwetal.[43].
degrees, however, indicates that the repeatability coeffi- Additionalfile6:Schemeofobtainingsignificantstandard
differencesinthestudypartsB.
cient (or MDC) is above the conventional 5-degree limit.
Additionalfile7:Rawdataofthestudy.Thisadditionalfilecontains
The other finger goniometric studies [2,13,15,19,23,26]
thefollowingworksheets.DataIA,IIA.Thisworksheetincludesa
have also observed intra-rater or inter-rater repeatability condensedversionofrawdataofthestudypartsA,appropriate
ofmorethan5degrees. measurementdifferences,andtheirdichotomizedscores;DataI-B,II-B.
Thisworksheetincludesacondensedversionofrawdataofthestudy
Thefinding of this studythat the measurements of the
partsB.
distal interphalangeal joint are relatively less consis-
Additionalfile8:Summaryofthedescriptivestatistics.Includesa
tent corresponds to the results of the earlier research summarytableoftheessentialdescriptivestatisticsofboththestudy
[2,26,33,37]. This phenomenon may be associated parts.
with the stabilization difficulty of the less powered Additionalfile9:Comparisonofearlierreliabilitystudiesof
standardfingergoniometry.Thisfileincludesatablewiththeessential
interphalangeal joints and limited phalangeal length
resultsandmethodologicalaspectsoftheearlierpertinentstudies.
available for the alignment of the arms of goni-
ometers. The results of the current study also corro-
borate the observations of the other researchers that Abbreviations
PIP:Proximalinterphalangeal;MCP:Metacarpophalangeal;DIP:Distal
intra-rater reliability is better than inter-rater reliability
interphalangeal;ICC:Intraclasscorrelationcoefficient;SEM:Standarderrorof
[2,6,7,23,25,26,28,33]. measurement;MDC:Minimaldetectablechange;ANOVA:Analysisof
Thelimitationsofthisexplorationincludetoosmallsam- variance;LLs:Lowerlimits;L-L:Lower-limit;CI:Confidenceintervals;TR:Trial;
PGn:Paperstripgoniometer;SGn:Standardfingergoniometer;N:Numberof
ple size for the concurrent assessment of inter-goniometer
measurementsacrossallratersandsubjects;EXT:Positionofimitated
reliability. This shortcoming was partly compensated by incompleteextension;FLEX:Positionofimitatedincompleteflexion;R:Rater.
MacionisBMCMusculoskeletalDisorders2013,14:17 Page10of11
http://www.biomedcentral.com/1471-2474/14/17
Competinginterests 20. Lefevre-ColauMM,PoiraudeauS,FermanianJ,Mayoux-BenhamouMA,
Theauthorhasnocompetingintereststodeclare. BargyF,RevelM:Reliabilityoftwogoniometersinassessingrheumatoid
fingermobility:Relationshipbetweenmobilityanddisability.
Authors’contributions EuraMedicophys2001,37:3–10.
VM:concept,design,acquisitionofdata,analysisandinterpretationofdata, 21. MacdermidJC,FoxE,RichardsRS,RothJH:Validityofpulp-to-palm
manuscriptpreparationandcriticalrevisons.Theauthorreadandapproved distanceasameasureoffingerflexion.JHandSurgBr2001,26:432–435.
thefinalmanuscript. 22. GeorgeuGA,MayfieldS,LoganAM:Lateraldigitalphotographywith
computer-aidedgoniometryversusstandardgoniometryforrecording
Acknowledgements fingerjointangles.JHandSurgBr2002,27:184–186.
TheauthorsincerelythanksDr.RūtaLevulienė,AssociateProfessorofthe 23. EllisB,BrutonA:Astudytocomparethereliabilityofcompositefinger
FacultyofMathematicsandInformaticsofVilniusUniversity,forcheckingthe flexionwithgoniometryformeasurementofrangeofmotioninthe
soundnessofthestatisticalanalysisofthisstudyandDanguolėStraižytė, hand.ClinRehabil2002,16:562–570.
LecturerattheInstituteofForeignLanguagesofVilniusUniversity,for 24. RoseV,NdukaCC,PereiraJA,PickfordMA,BelcherHJ:Visualestimationof
proofreadingthemanuscript.Theauthorisalsogratefultoallparticipantsof fingerangles:doweneedgoniometers?JHandSurgBr2002,27:382–384.
thestudy. 25. GlasgowC,WiltonJ,ToothL:Optimaldailytotalendrangetimefor
contracture:resolutioninhandsplinting.JHandTher2003,16:207–218.
Received:23May2012Accepted:28December2012 26. BurrN,PrattAL,StottD:Inter-raterandintra-raterreliabilitywhen
Published:9January2013 measuringinterphalangealjoints:comparisonbetweenthreehand-held
goniometers.Physiotherapy2003,89:641–652.
References 27. GlasgowC,JamesM,O‘SullivanJ,ToothLR:Measurementofjointstiffness
1. RobinsRH:Handassessmentcharts.JHandSurgBr1986,11:287–298. inthehand:apreliminaryinvestigationofthereliabilityandvalidityof
2. EllisB,BrutonA,GoddardJR:Jointanglemeasurement:acomparative torqueanglecurves.BrJHandTher2004,9:11–12.
studyofthereliabilityofgoniometryandwiretracingforthehand. 28. PrattAL,BurrN,StottD:Aninvestigationintothedegreeofprecision
ClinRehabil1997,11:314–320. achievedbyateamofhandtherapistsandsurgeonsusinghand
3. MacionisV:Atechniqueforgraphicalrecordingofrangeofmotionusing goniometrywithastandardisedprotocol.HandTher2004,9:116–121.
animprovisedpapergoniometer.JHandTher2011,24:374–377. 29. StamHJ,ArdonMS,denOudenAC,SchreudersTA,RoebroeckME:The
4. DeVetHC,TerweeCB,KnolDL,BouterLM:Whentouseagreement compangle:anewgoniometerforjointanglemeasurementsofthe
versusreliabilitymeasures.JClinEpidemiol2006,59:1033–1039. hand.atechnicalnote.EuraMedicophys2006,42:37–40.
5. KottnerJ,AudigéL,BrorsonS,DonnerA,GajewskiBJ,HróbjartssonA, 30. CookJR,BakerNA,ChamR,HaleE,RedfernMS:Measurementsofwrist
RobertsC,ShoukriM,StreinerDL:GuidelinesforReportingReliability andfingerpostures:acomparisonofgoniometricandmotioncapture
andAgreementStudies(GRRAS)wereproposed.JClinEpidemiol2011, techniques.JApplBiomech2007,23:70–78.
64:96–106. 31. KatoM,EchigoA,OhtaH,IshiaiS,AokiM,TsubotaS,UchiyamaE:The
6. HamiltonGF,LachenbruchPA:Reliabilityofgoniometersinassessing accuracyofgoniometricmeasurementsofproximalinterphalangeal
fingerjointangle.PhysTher1969,49:465–469. jointsinfreshcadavers:comparisonbetweenmethodsofmeasurement,
7. GiudiceML:Effectsofcontinuouspassivemotionandelevationonhand typesofgoniometers,andfingers.JHandTher2007,20:12–18.
oedema.AmJOccupTher1990,44:914–921. 32. SmithRP,DiasJJ,UllahA,BhowalB:Visualandcomputersoftware-aided
8. Breger-LeeD,VoelkerET,GiurintanoD,NovickA,BrowderL:Reliabilityof estimatesofDupuytren'scontractures:correlationwithclinical
torquerangeofmotion:apreliminarystudy.JHandTher1993,6:29–34. goniometricmeasurements.AnnRCollSurgEngl2009,91:296–300.
9. DijkstraPU,deBontLG,vanderWeeleLT,BoeringG:Jointmobility 33. LewisE,ForsL,TharionWJ:Interraterandintraraterreliabilityoffinger
measurements:reliabilityofastandardizedmethod.Cranio1994, goniometricmeasurements.AmJOccupTher2010,64:555–561.
12:52–57. 34. TorokKS,BakerNA,LucasM,DomsicRT,BoudreauR,MedsgerTAJr:
10. WeissPL,AugustS,PetersG,SampalisJ:UsingtheExosHandmasterto Reliabilityandvalidityofthedeltafinger-to-palm(FTP),anewmeasure
measuredigitalrangeofmotion:reliabilityandvalidity.MedEngPhys offingerrangeofmotioninsystemicsclerosis.ClinExpRheumatol2010,
1994,16:323–328. 28:S28–36.
11. FlowersKR,LaStayoP:Effectoftotalendrangetimeonimproving 35. LiK,ChenIM,YeoSH,LimCK:Developmentoffinger-motioncapturing
passiverangeofmotion.JHandTher1994,7:150–157. devicebasedonopticallinearencoder.JRehabilResDev2011,48:69–82.
12. ChiuHY:Amethodoftwo-dimensionalmeasurementforevaluating 36. OessNP,WanekJ,CurtA:Designandevaluationofalow-cost
fingermotionimpairment.Adescriptionofthemethodandcomparison instrumentedgloveforhandfunctionassessment.JNeuroengRehabil
withangularmeasurement.JHandSurgBr1995,20:691–695. 2012,9:2.
13. GoldsmithN,JuzlE:Inter-raterreliabilityoftwotrainedratersusinga 37. EngstrandC,KreversB,KvistJ:Interraterreliabilityinfingerjoint
goniometerforthemeasurementoffingerjoints.BrJHandTher1998, goniometermeasurementinDupuytren’sdisease.AmJOccupTher2012,
3:11–12.citedbyBurretal.[26]. 66:98–103.
14. ChiuHY,SuFC,WangST,HsuHY:Themotionanalysissystemand 38. PortneyLG,WatkinsMP:FoundationsofClinicalResearch:Applicationsto
goniometryofthefingerjoints.JHandSurgBr1998,23:788–791. Practice.3rdedition.NJ,PrenticeHall:UpperSaddleRiver;2008.
15. BrutonA,EllisB,GoddardJ:Comparisonofvisualestimationand 39. BrutonA,ConwayJH,HolgateST:Reliability:Whatisitandhowisit
goniometryforassessmentofmetacarpophalangealjointangle. measured?Physiotherapy2000,86:94–99.
Physiotherapy1999,85:2011–2018. 40. WeirJP:Quantifyingtest-retestreliabilityusingtheintraclasscorrelation
16. WilliamsNW,PenroseJM,CaddyCM,BarnesE,HoseDR,HarleyP:A coefficientandtheSEM.JStrengthCondRes2005,19:231–240.
goniometricgloveforclinicalhandassessment.Construction,calibration 41. BarnhartHX,HaberMJ,LinLI:Anoverviewonassessingagreementwith
andvalidation.JHandSurgBr2000,25:200–207. continuousmeasurements.JBiopharmStat2007,17:529–569.
17. BrownA,CramerLD,EckhausD,SchmidtJ,WareL,MacKenzieE:Validity 42. ShroutPE,FleissJL:Intraclasscorrelations:usesinassessingrater
andreliabilityoftheDexterhandevaluationandtherapysystemin reliability.PsycholBull1979,86:420–428.
hand-injuredpatients.JHandTher2000,13:37–45. 43. EliasziwM,YoungSL,WoodburyMG,Fryday-FieldK:Statistical
18. CatalanoLW3rd,BrowneRH,CarterPR,FrobishAC,EzakiM,LittlerJW: methodologyfortheconcurrentassessmentofinterraterandintrarater
TheLittlerlinemethodandtheareaunderaGaussiancurve:anew reliability:usinggoniometricmeasurementsasanexample.PhysTher
methodofassessingdigitalrangeofmotion.HandSurgAm2001, 1994,74:777–788.
26:23–30. 44. HayenA,DennisRJ,FinchCF:Determiningtheintra-andinter-observer
19. GrothGN,VanDevenKM,PhillipsEC,EhretsmanRL:Goniometryofthe reliabilityofscreeningtoolsusedinsportsinjuryresearch.JSciMed
proximalanddistalinterphalangealjoints,PartII:placementpreferences, Sport2007,10:201–210.
interraterreliability,andconcurrentvalidity.JHandTher2001,14:23–29. 45. BlandJM,AltmanDG:Measurementerror.BMJ1996,313:744.
