ARTICLE OPEN ACCESS

Psychometric properties of the Friedreich AtaxiaRating ScaleChristian Rummey PhD Louise A Corben PhD Martin B Delatycki MD SH Subramony MD

Khalaf Bushara MD Christopher M Gomez MD Joseph Chad Hoyle MD Grace Yoon MD

Bernard Ravina MD Katherine D Mathews MD George Wilmot MD Theresa Zesiewicz MD

Susan Perlman MD Jennifer M Farmer MD and David R Lynch MD PhD

Neurol Genet 20195e371 doi101212NXG0000000000000371

Correspondence

Dr Lynch

lynchdmailmedupennedu

AbstractObjectiveTo investigate the psychometric properties of the Friedreich Ataxia Rating Scale neurologicexamination (FARSn) and its subscores as well as the influence of the modifications resulting inthe now widely used modified FARS (mFARS) examination

MethodsBased on cross-sectional FARS data from the FAndashClinical Outcome Measures cohort weconducted correlation-based psychometric analyses to investigate the interplay of items andsubscores within the FARSnmFARS constructs

ResultsThe results provide support for both the FARSn and the mFARS constructs as well asindividually for their upper limb and lower limb coordination components The omission of theperipheral nervous system subscore (D) and 2 items of the bulbar subscore (A) in the mFARSstrengthens the overall construct compared with the complete FARS

ConclusionsA correlation-based psychometric analysis of the neurologic FARSn score justifies the overallvalidity of the scale In addition omission of items of limited functional significance as created inthe mFARS improves the features of the measures Such information is crucial to the ongoingapplication of the mFARS in natural history studies and clinical trials Additional analyses oflongitudinal changes will be necessary to fully ascertain its utility especially in nonambulantpatients

From the Clinical Data Science GmbH (CR) Basel Switzerland Bruce Lefroy Centre for Genetic Health Research (LAC MBD) Murdoch Childrenrsquos Research Institute ParkvilleVictoria Australia Department of Paediatrics (LAC MBD) University of Melbourne Parkville Victoria Australia Department of Neurology (SHS) McKnight Brain Institute RoomGainesville FL University of Minnesota (KB) University of Chicago (CMG) Ohio State University (JCH) Divisions of Neurology and Clinical and Metabolic Genetics (GY)Department of Paediatrics the Hospital for Sick Children University of Toronto Ontario Canada Hospital University of Rochester (BR) University of Iowa (KDM) Emory University(GW) University of South Florida (TZ) Friedreichrsquos Ataxia Research Alliance (SP) Downingtown PA and Division of Neurology (DRL) Childrenrsquos Hospital of Philadelphia

Go to NeurologyorgNG for full disclosures Funding information is provided at the end of the article

The Article Processing Charge was funded by the authors

This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 40 (CC BY-NC-ND) which permits downloadingand sharing the work provided it is properly cited The work cannot be changed in any way or used commercially without permission from the journal

Copyright copy 2019 The Author(s) Published by Wolters Kluwer Health Inc on behalf of the American Academy of Neurology 1

Inherited ataxias vary in clinical presentation yet most doshare cerebellar pathology leading to loss of balance and limbcoordination cardinal features that need to be measured toevaluate the efficacy of an intervention Friedreich ataxia(FRDA) exemplifies this challenge of a complex neuropa-thology including degeneration of the spinal cord peripheralsensory nerves and the dentate nucleus of the cerebellum1

which demands a broad but sensitive measure to capture theeffect of neuroanatomical dysfunction on clinical progressionof affected patients

A first widely accepted such instrument was the InternationalCooperative Ataxia Rating Scale (ICARS)2 but its use inclinical studies in FRDA3ndash7 has diminished89 in favor of ad-vanced developments One the Scale for the Assessment andRating of Ataxia (SARA)1011 focuses on a compact structureand quick administration Concurrently the Friedreich AtaxiaRating Scale (FARS) combines timed measures of perfor-mance functional disability staging (FDS) and a patient-reported outcome (activities of daily living [ADL]) witha paramount neurologic examination the FARS neurologicexamination (FARSn) Both SARA and FARS are now gen-erally accepted and used in 2 large natural history studies inEurope12ndash15 and the United States1617

The relationships of the FARSn with other rating scales11

patient-reported outcomes18ndash20 disease duration and FDS21

age at onset and repeat length161821ndash24 are well establishedWe summarize the revisions of the FARSn examination afterits introduction18 and based on classical test theory25 (CTT)provide a psychometric analysis of the construct at its sub-scores the lack of which has been acknowledged112627

MethodsStandard protocol approvals registrationsand patient consentsInformed consent was obtained from all participants beforestarting evaluations The study was approved by the in-stitutional review board at all participating sites

Study design and participantsWe used data from all participants recruited via the Collabo-rative Clinical Research Network in Friedreich Ataxia28 into theFriedreich AtaxiandashClinical Outcome Measures Study (FA-COMS)17 Results from this ongoing and actively recruitingregistry study have been publishedwidely162224 Briefly the FA-COMShasminimal inclusionexclusion criteria (age 4ndash80 years

and clinical and genetic confirmation of FRDA) and the follow-up time now (as of 2019) is up to 15 years Participant char-acteristics cover the full spectrum of FRDAwith regard to age atdisease onset disease duration and disease severity There are12 participating sites (number of participants) ChildrenrsquosHospital of Philadelphia (417) University of California LosAngeles (191) Murdoch Childrenrsquos Research Institute (179Melbourne Australia) Emory University (73) University ofSouth Florida (36) University of Iowa (26) University ofFlorida (20) University of Chicago (20) Sick Kids Hospital(19 Toronto Canada) University of Minnesota (17) OhioState University (10) and University of Rochester (4)

Friedreich Ataxia Rating ScaleWith the exception of the FARSn the components of theFARS examination ADL timed measures and FDS were notchanged since its original publication18 Eg FARS disabilitystaging (FDS) is derived from an ordinal score from 1 to 6(no minimal mild moderate severe and total disability)independent from FARS ratings It is graded in units of 05linked by descriptors to ambulation status and overall func-tion (5 being permanent wheelchair user)18

The revision of the FARSn examination was initially driven byitem reduction based on interrater reproducibility and in-sertion of 2 items related to stance without visual aid21 Theresulting version (total of 125 points figure 1) consists of 5subscales directed to bulbar function (FARS A maximum of11 points) upper limb coordination (FARS B 36) lower limbcoordination (FARS C 16) peripheral nervous system(FARS D 26) and upright stability (FARS E 36) This ver-sion is used in the FA-COMS17 since 2003 and has been usedas an end point in clinical studies29ndash31

More recently the need to focus solely on functional abilitiesled to a reduction in items specifically in the peripheral ner-vous system and bulbar components16 This resulted in thenew modified FARS (mFARS) score (mFARS 93) which isused in currently ongoing and recently finished clinicaltrials30ndash33 It is important that all items of the mFARS areincluded in the full FARSn examination only the bulbarsubscore (mA) was modified (now excluding items A1 facialatrophy and A2 and tongue atrophy) and the peripheralnervous system (D) subscore is now omitted The remainingsubscales upper limb coordination (B) lower limb co-ordination (C) and upright stability (E) are identical in boththe mFARS and the FARSn (see figure 1 for description andscoring for overall scales and individual subscales)

GlossaryADL = activities of daily livingCTT = classical test theory FA-COMS = Friedreich AtaxiandashClinical OutcomeMeasures StudyFARS = Friedreich Ataxia Rating Scale FARSn = FARS neurologic examination FDS = functional disability staging FRDA =Friedreich ataxia ICARS = International Cooperative Ataxia Rating Scale IRT = Item Response Theory mFARS = modifiedFARS PCA = principal component analysis SARA = Scale for the Assessment and Rating of Ataxia

2 Neurology Genetics | Volume 5 Number 6 | December 2019 NeurologyorgNG

Scope and statistical analysisThe goal of this work was to evaluate the function of themeasurein terms of CTT ie identify valid subscores and components inadvance of a longitudinal evaluation Therefore we used exclu-sively cross-sectional (baseline) data which ensure the mostdiverse coverage of all disease phases To investigate the FARSnmeasurement model and establish that items measure the same(or a similar) underlying constructs34 corrected item-total cor-relations (the correlation of an item with its own subscale whenexcluding that item) were calculated as well as subscale inter-correlations In addition we investigated the range of observedsubscale and total scores skewness of their distributions andfloor and ceiling effects To identify clusters of intercorrelatingitems and examine underlying concepts measured by specificitems as well as to obtain information on their clinical in-terpretation34 we performed exploratory principal componentanalysis (PCA) using varimax rotation If applicable Cronbach αwas calculated for sub- and total scales as a measure of internalconsistency and stability against random error

All statistical calculations were performed in R35 using thetidyverse environment36 and the psych package37 for psy-chometric analyses

Data availabilityPrimary data for the present study (and all other previousstudies of this cohort) are available at TheCritical Path Institute

as part of the Friedreich Ataxia Integrated Clinical Database(cpathorgprogramsdccprojectsfriedreichs-ataxia) Casereport forms and protocols are available through FARA (curefaorg) It is expected that such data will remain availableindefinitely

ResultsBaseline demographic results of the FA-COMS cohortAt the time of data cut for this study (April 2019) 1011 par-ticipants enrolled into the FA-COMS had at least 1 visit witha complete FARSn examination (no missing data) FDS andage at disease onset available The majority of these individualswere in the early-onset group1238 ie first symptoms of FRDAoccurred before age 15 years (N= 695 687) representing thegenetically most severely affected individuals as evident by thegreater GAA1 (mean 71801 SD 20929) and GAA2 (mean94151 SD 18653) repeat sizes in this group (table 1) At theirfirst presentation within this group 319 individuals (459)were still in an early disease phase as evident by FDS 1ndash2 Stages3ndash4 in the early-onset group were represented by 145 (209)participants a subgroup that is in particular high risk of losingambulation within the next 2ndash4 years39 and are currentlya group of high interest regarding focused clinical trials39 Of theremaining participants 209 (207) had intermediate onset

Figure 1 Measurement model of the neurologic examination of the FARSn and the modified FARS (mFARS)

Maximum scoresubscaleitem scores are shown in brackets Items in subscales B C and D are conducted separately on lateral sides items A1 and A2 areexcluded in the mFARS examination FARS = Friedreich Ataxia Rating Scale mFARS = modified FARS

NeurologyorgNG Neurology Genetics | Volume 5 Number 6 | December 2019 3

(15ndash24 years) and 107 (106) had late onset (ge25 years) Inthe 2 later-onset groups fewer patients were in an early diseasephase (stage 1ndash2 383 and 299) but there also were fewerpatients in later nonambulatory stages (stage 5ndash6 249 and224 respectively) Overall 1 third of participants (332 inthe early-onset group 304 of all participants) were alreadynonambulatory at their first visit In line with previous reportsabout 54 of the patients were carrying a point mutationBecause these patients show increased risk of atypical diseaseprogression4041 they are typically excluded from clinical studiesHowever the potential effect on the results was deemed mini-mal and we included these patients in all analyses

Psychometric properties of the FARSScaling characteristics of both full scores and all respectivesubscales are summarized in table 2 and the full corrected item-subscale correlations for FARSn subscores in table 3 (a fullcorrelation table for mFARS is provided as table e- 1 linkslwwcomNXGA189) Mean corrected item-total correlations forsubscales mA B C D and E which are forming the mFARSwere 058 or above whereas the A and the D subscale fromFARSn showed mean values of 040 and 042 already in-dicating that the step fromFARSn to themFARS leads tomorestable constructs An additional benefit from the exclusion ofA1 and A2 is the reduction of ceiling effects in the modifiedbulbar subscore (mA) leading to a less skewed distribution

Cronbach α values were satisfactory for the upper limb (A087) and lower limb (B 091) as well as for the upright

stability scale (E 087) Themodified bulbar subscale (mA) inthe mFARS showed an α of 065 which is improved from theinitial 4-item bulbar subscale (A) in FARSn but still has to beconsidered a low value clearly missing the suggested criterionof 094243 For the initial FARSn subscores A and D Cron-bach α (053 and 056 respectively) shows that the mod-ifications resulted in improved probabilities in mFARS Forthe complete examinations both the FARSn and the mFARSshowed a Cronbach α of 092 Constructs covered by sub-scales of a total score should be related (intercorrelated) butalso different to cover different aspects of a disease A range of030ndash070 has been suggested9 and this condition was met forall subscores in both overall scales

All subscales as well as the complete scores correlated withdisease duration and FDS and to a lesser extent with ageDetailed item-own subscale and item-other subscale corre-lations (table 3) further corroborate construct validities ofthe subscores and full scales All items in the mFARS sub-scales correlated with coefficients of gt03 with their sub-scores indicating that these items measure a commonconstruct and collect a similar amount of information934 Inaddition only 1 item (E1 sitting posture) of the FARSnmFARS examinations did not correlate higher with its ownsubscale than with other subscales This item correlated toa similar extent with all 3 subscales (table 3) in mFARS andalso with the peripheral nervous system subscale D inFARSn As a criterion for a margin between correlationsbetween its own subscale and others 2 SDs have been

Table 1 Demographic characteristics of the cohort percentages are based on the disease onset group

Mean (SD)

Disease onset group

Totallt15 y (early) 15-24 y (intermediate) gt24 y (late)

N (a) 695 (687) 209 (207) 107 (106) 1011

Male () 354 (509) 102 (488) 45 (421) 501 (496)

Age 1941 (1031) 3232 (1109) 5012 (1086) 2533 (1450)

Disease duration 1095 (967) 1440 (1070) 1518 (910) 1211 (997)

Symptom onset 846 (336) 1792 (254) 3494 (938) 1322 (941)

Age at diagnosis 1215 (508) 2296 (696) 4182 (1199) 1747 (1142)

GAA1b 71801 (20929) 50453 (19776) 27379 (19432) 62823 (25291)

GAA2c 94151 (18653) 85280 (24593) 79534 (28590) 90824 (21794)

Point mutation N () 40 (58) 10 (48) 5 (47) 55 (54)

Disease stage

1ndash2 319 (459) 80 (383) 32 (299) 431 (426)

3ndash4 145 (209) 77 (368) 51 (477) 273 (270)

5ndash6 231 (332) 52 (249) 24 (224) 307 (304)

a Based on the overall cohortb N = 939 (excluding point mutations)c N = 937

4 Neurology Genetics | Volume 5 Number 6 | December 2019 NeurologyorgNG

suggested934 which was fulfilled by all items but E1 (sittingposture) and E2B (stance feet apart with eyes closed)which also correlated well with subscale C (lower limb co-ordination) For the peripheral nervous system subscore inFARSn except for D2 (muscle weakness) none of the itemsin the peripheral nervous system subscale correlated wellwith any other subscale (table 3) Remarkably weaker item-subscale correlations were also found for items E3B (stancefeet together eyes closed) E4 (tandem stance) and E5(stance on dominant foot) (see below)

Floor and ceiling effectsCeiling effects were apparent in subscale C (heel-shin tap andheel-shin slide possible range 0ndash8) with 21 of observationsshowing the maximum score Notably even after loss of am-bulation patients can perform the heel-shin tests (subscale C)For subscale E (upright stability) we found that item E1 (sit-ting position possible range 0ndash4) is scored at 4 (indicatinggreatest impairment) in less than 5 of the observations whichmasks total scoreceiling effects for subscale E We alsoreported a ceiling effect at 90 of the maximum on this item

Table 2 Scaling characteristics of sub- and total scores in the FARS neurologic examination plus total FARSn and themFARSa

FARS subscaleitemsmAmdashBulbar(modified)

BmdashUpperlimb

CmdashLowerlimb

EmdashUprightstability mFARS FARSn AmdashBulbarb DmdashPeripheral

No of items 2 5 2 9 18 25 4 5

Corrected itemndashtotalcorrelations

Mean 058 067 084 060 065 057 040 042

Range 055ndash061 041ndash077 084ndash084 024ndash086 025ndash086 017ndash086 022ndash060 017ndash072

Subscalecharacteristics

Mean score (SD) 091 (090) 1366(697)

857 (473) 2506 (832) 4821(1852)

6034(2217)

110 (108) 1194 (464)

Observed score (max) 0ndash5 0ndash36 0ndash16 0ndash36 2ndash92 6ndash118 0ndash7 0ndash26

Floor () 33 2 2 0 0 0 25 0

Floor (10thpercentile )

45 6 3 1 1 0 71 1

Ceiling (90thpercentile )

0 1 22 31 2 1 0 2

Ceiling () 0 1 21 4 0 0 0 1

Skewness 102 052 040 minus042 016 034 156 056

Cronbach α 065 087 091 087 092 092 053 056

Subscaleintercorrelations

Bulbar (modified) mdash 063 057 055 068 067b mdash 049b

Upper limb mdash 068 066 088 086 064 057

Lower limb mdash 076 088 088 056 070

Peripheral mdash mdash 081 049 mdash

Upright stability 092 091 054 093

Correlation with otherparameters

Disease duration 024 028 036 037 042 043 024 029

Age 010 007 009 009 010 010 009 004

Disability staging 035 041 056 076 073 072 033 038

Abbreviations FARS = Friedreich Ataxia Rating Scale mFARS = modified FARSa Pearson correlation coefficients corrected for overlap and scale reliability37b Analyzed using FARSn subscales a complete table of subscale correlations for mFARS subscales is provided in table e-1 (linkslwwcomNXGA18)

NeurologyorgNG Neurology Genetics | Volume 5 Number 6 | December 2019 5

(table 2) Specifically for the E (upright stability) subscale theceiling effect corresponded well to the number of non-ambulatory patients in our sample (32 FDS gt 4 table 1)

Floor effects were only visible in the bulbar subscalemA and (andbulbarA in FARSn) The removal of A1 (facial atrophy) and A2

(tongue atrophy) reduced the ceiling effects to some extent still43of patients had scored 10or less of the total in themodified(mA) bulbar subscale We also analyzed the ceiling effects on anitem-based level All individual item scores in mFARS (groupedby subscales) are presented in figure 2 as heat maps from lowestscore (light blue) to maximum score (dark blue)

Table 3 Item-subscale correlations of the full FARSna

FARS subscaleitems AmdashBulbar BmdashUpper limb CmdashLower limb DmdashPeripheral EmdashUpright stability

Bulbar

Facial atrophyb A1 022 023 020 028 015

Tongue atrophyb A2 035 025 018 023 017

Cough A3 066 050 052 051 048

Speech A4 070 064 054 047 051

Upper limb coordination

Finger-finger B1 045 052 031 034 029

Nose-finger B2 069 082 063 060 057

Dysmetria B3 076 085 067 062 064

Rapid movements B4 060 079 059 055 062

Finger taps B5 072 082 067 063 061

Lower limb coordination

Heel-shin slide C1 069 069 089 079 076

Heel-shin tap C2 068 069 090 083 075

Peripheral nervous systemb

Muscle atrophy D1 035 030 043 053 032

Muscle weakness D2 072 061 074 073 062

Vibratory sense D3 022 020 033 037 034

Position sense D4 029 038 045 054 046

Deep tendon reflexes D5 004 016 012 020 016

Upright stability

Sitting posture E1 078 078 079 079 070

Stance feet apart E2A 064 060 078 076 083

Stance feet apart eyes closed E2B 040 050 054 053 072

Stance feet together E3A 050 056 063 060 080

Stance feet together eyes closed E3B 022 031 03 032 045

Tandem stance E4 020 025 026 027 044

Stance on the dominant foot E5 014 018 017 013 032

Tandem walk E6 048 045 052 049 074

Gait E7 069 065 081 077 090

Abbreviations FARS = Friedreich Ataxia Rating Scale mFARS = modified FARSa Corrected for item overlap and scale reliability37b A1 A2 and the peripheral nervous system subscore are not included in the mFARS scores A complete table of item-subscale correlations for mFARS isprovided in table e-1 (linkslwwcomNXGA18)Item-own subscale correlations are emphasized in bold

6 Neurology Genetics | Volume 5 Number 6 | December 2019 NeurologyorgNG

In general the heat maps for items in subscales A B and Creflect nicely the broad and balanced stages of function loss inthis population Floor effects discussed above are visible initems A3 (cough) and A4 (speech) and relevant but stillworkable ceiling effects in the lower limb coordination sub-scale C In the 9-item subscale E (upright stability) howeveralthough favorable validity and consistencyreliability prop-erties are achieved (α = 087 table 2) only items E1 (sitting)and E7 (gait) and to some extent E6 (tandem walk) showbalanced distributions Specifically all 6 stance-related itemsE2 to E5 reveal distinct bimodal distributions which meansthat a patient can either perform the test well (score 0) or willbe unable to complete the item rating a maximum score of 4This is most evident in items E2A (stance feet apart eyesopen) E2B (stance feet apart eyes closed) and E3A (stancefeet together eyes open) For the remaining items E3B(stance feet together eyes closed) E4 (tandem stance) andE5 (stance dominant foot) it is also true but only very fewpatients do not score the maximum

A grouped analysis shows that only 158 of all patients(n = 160) do not score the maximum result in all E3B (stancefeet apart eyes closed)E4 (tandem stance)E5 (stance ondominant foot) items or the other way around 985 ofpatients (n = 996) had the maximum possible score in at least 1of these 3 items In the ambulatory sub-population (FDSlt5N=704 696) these proportions were 225 (n = 159) and 979(n = 689)

Factor analysisFor principal component (PCA) and factor analysis severalsolutions for FARSn and mFARS were evaluated Using theKaiser criterion44 (retaining factors with an eigenvalue gt1)PCA found 5 factors for the complete FARSn A more detailedexamination of the factor loading structure did not result inoverall meaningful groupings Especially the items of the pe-ripheral nervous system (D) subscore loaded into componentstogether with items of other subscores (data not shown)

For mFARS 3 principal components with eigenvalues gt1were found and 1 additional with an eigenvalue of 083Factor analysis using 4 factors resulted in a separation ofindependent clinically meaningful factors Items in FARS E(upright stability) and C (lower limb coordination) loadedstrongest in component 1 upper limb coordination items(FARS B) loaded into component 2 and component 3 in-cluded solely the 2 bulbar function items of FARS A The 3remaining items of the FARS E (upright stability) with highamounts of ldquounablerdquo (E3B E4 and E5) loaded into the 4thcomponent Cumulatively these 4 factors explained 70 ofthe variance in the overall construct

DiscussionFor our psychometric analysis of the FARSn examination weused the FA-COMS cohort comprehensively representingthe 2 most important dimensions of FRDA being individualsfrom the complete severity spectrum of FRDA (as measuredby age at symptom onset or GAA1GAA2 repeat length) andat the same time including all disease durations Moreover thepredominance of early vs late onset and children vs adults inrecruitment is most likely representative of the incidence butnot necessarily prevalence of FRDA Affected individualsprogress fastest in the early phases of the disease20 we usedbaseline data only to achieve optimal coverage of diseasephases This population profile should be ideal for the anal-yses of inherently sample-dependent psychometric propertiesof any scale Still overall correlation parameters will revolvearound the proportion of early-late-stage patients and theratio of ambulantnonambulant patients

In the present study classical psychometric correlation anal-ysis (item-own vs item-other subscale correlations) confirmsthe validity and structure of the FARSn examination butspecifically endorses the modifications leading to the mFARSOverall both scales showed appropriate item-subscalegroupings inter-subscale correlations and internal consis-tency (Cronbach α gt090) However 2 sets of items withinthe complete 125-point FARSn scale show clear weaknesses

First items in the peripheral nervous system (D) subscoreitself have inefficient within subscale correlations and forma weak construct This may reflect the arbitrary grouping ofitems within FARS D (peripheral nervous system) In par-ticular muscle strength and atrophy items which have

Figure 2 Heat map of individual items within the modifiedFARS

Data are grouped by subscales Five blue hues (from light to dark) correspondto increasing item scores in steps of 20 FARS = Friedreich Ataxia RatingScale

NeurologyorgNG Neurology Genetics | Volume 5 Number 6 | December 2019 7

a slower decline (and thus lower scoring) and may be sec-ondary to both peripheral and CNS pathology are groupedwith deep tendon reflexes which being a hallmark feature ofFRDA are usually absent or diminished on first presentationtherefore receiving a maximum score Second the A1 (facialatrophy)A2 (tongue atrophy) items correlate weakly withother items in the bulbar (FARS A) subscore and other partsof the FARS examination The remaining items in the bulbarsubscore (related to cough and speech) have residual flooreffects yet they show considerable intercorrelation justifyingtheir inclusion in the mFARS Also the Cronbach α value isimproved for the modified bulbar (FARS mA) subscale

These results justify the exclusion of these 2 groups of items(FARS D and A1A2) in the newly formed modified mFARSscore Of interest the requirement to focus on more patient-driven tasks initiated this change without knowledge of thepsychometric properties of the items that quantify the func-tional ability of the individual with FRDA As a result intra-subscale correlations and reliability (Cronbach α) of themFARS score are improved whereas floor effects in FARS A1(facial atrophy) and A2 (tongue atrophy) are avoided dem-onstrating that the items reflecting functional ability match themost psychometrically robust in the mFARS In addition thisshows that progression captured by the FARS examination(s)does not match the decline of dysfunctional sensory pathwaysin FRDA and that sensory loss is less relevant for the patient

In addition the present analysis demonstrates the detailedfeatures of the FARSE (upright stability) subscale which drivesquantification of progression in the early phase of the diseaseIts 6 stance-related items (E2A B E3A B E4 E5) are scoredalmost exclusively with extreme values (0 or the maximum 4)This might indicate that in context of an otherwise slowlyprogressing condition these abilities are lost rapidly Alterna-tively the extreme values could be a function of the itemconstructs The easier initial 3 stance items (E2A E3A andE2B) show preferable interitem correlations and should func-tion well within the item response theory (IRT) On the otherhand the increasing difficulty of the remaining stance itemsresults in very few patients in this cohort to perform E3B E4and E5 In most situations these items thus contribute little tothe total scores besides addition of noise However they alsomay provide useful information to the overall mFARS score inclinical studies that target those individuals with FRDAwho arevery early in the disease process These results also indicate thatall 6 stance items could be useful in classifying patientsaccording to function and in longitudinal analysis of thoseindividuals with the mildest of FRDA symptoms Theremaining 3 items in subscore E show beneficial psychometricproperties noteworthy E7 (gait) correlates with all items inmFARS (and FARSn) and has the best overall correlation withthe total mFARS score and in itself directly defines individualambulatory ability Another crucial item for functional abilitythe sitting posture item (E1) showed the weakest inter-subscale correlation in FARS E (when disregarding E3B E4and E5) perhaps as scoring on this item can be improved by

scoliosis surgery or physiotherapy Overall although ceilingeffects in FARS E (upright stability) match the nonambulatoryfraction of the cohort it contributes well to the overall FARSnmFARS score in the ambulatory portion of the cohort

Subscales FARS B (upper limb coordination) and C (lowerlimb coordination) showed beneficial psychometric proper-ties for inclusion in the overall FARSnmFARS Of note thefinger-finger test item (B1) showed comparatively low item-subscale correlation similar to that found in the ICARSanalysis9 Still the remainder of these subscales provide usefulcontributions to the scale

The use of IRT3445 to evaluate rating scales in FRDA has beenproposed2646 However although Rasch-based models havebeen extended to polytomous45 and continuous4748 ratingscales they are not easily applied to scales with complexsubscaleitem structures like FARS and ICARS Similarly theassumption of unidimensionality in IRT-basedmodels impedestheir implementation4950

Factor analysis for FARSn confirmed previous such findings forthe FARSn examination including the equivocal overlap be-tween factor loadings and subscores11 For the mFARSa 4-factor solution was the most clinically and statistically ap-propriate and PCA provided strong support of the subscorestructure This analysis additionally backs the exclusion of theFARS D and the 2 bulbar items and indicates that a lower limbcomponent (FARS C plus E) and the upper limb portion(FARS B) within the mFARS have value as individual scales forseparate evaluation

In this context the conceptual differences between a multi-subscore scale like the mFARS and a more compact scale(SARA) become obvious The latter has been designed es-pecially for rapid application and a strong central constructleading to its unidimensionality1011 These features shouldimprove interrater variability highly important eg in largemulticenter studies However SARA might be limited inmore complex studies In contrast the mFARS providesa more detailed evaluation of overall patient status and a morecomplex yet valid construct In addition subcomponents ofthe mFARS might separately support the evaluation of in-dividual domains eg upper and lower limb function inspecific circumstances This might prove valuable when in-vestigating temporal differences in loss of function in separatedomains or when focusing on later stage exclusively non-ambulatory patients Although the complete FARS may re-quire more time to complete much of the extra information iscollected in other instruments in typical clinical trials Thusoverall the different measures are relatively similar in practicaladministration allowing investigators to select the best mea-sure for the clinical approach being tested

As future clinical trials in FRDA may focus on either sub-group potentially targeted with specific therapeutic inter-ventions or more diverse cohorts with broader therapeutic

8 Neurology Genetics | Volume 5 Number 6 | December 2019 NeurologyorgNG

interventions the present analysis demonstrates that themFARS examination captures the complex features of FRDAeven in early stages of the disease

DisclosureDisclosures available NeurologyorgNG

AcknowledgmentThe authors thank the coordinators of the Friedreich AtaxiaClinical Outcome Measures Study (FACOMS) for collectionof patient visit data

Study fundingThis work was funded by the Friedreichrsquos Ataxia Researchalliance (FARA)

Publication historyReceived by Neurology Genetics June 7 2019 Accepted in final formSeptember 23 2019

References1 Koeppen AH Mazurkiewicz JE Friedreich ataxia neuropathology revised

J Neuropathol Exp Neurol 20137278ndash902 Trouillas P Takayanagi T Hallett M et al International cooperative ataxia rating

scale for pharmacological assessment of the cerebellar syndrome The ataxia neuro-pharmacology committee of the world federation of neurology J Neurol Sci 1997145205ndash211

3 Di Prospero NA Baker A Jeffries N Fischbeck KH Neurological effects of high-doseidebenone in patients with Friedreichrsquos ataxia a randomised placebo-controlled trialLancet Neurol 20076878ndash886

4 Lynch DR Perlman SL Meier T A phase 3 double-blind placebo-controlled trial ofidebenone in Friedreich ataxia Arch Neurol 201067941ndash947

5 Meier T Perlman SL Rummey C Coppard NJ Lynch DR Assessment of neuro-logical efficacy of idebenone in pediatric patients with Friedreichrsquos ataxia data froma 6-month controlled study followed by a 12-month open-label extension studyJ Neurol 2012259284ndash291

6 A study of efficacy safety and tolerability of idebenone in the treatment of Friedreichrsquosataxia (FRDA) patients (MICONOS) Available at clinicaltrialsgovct2showNCT00905268 Accessed January 4 2019

7 Long-term safety and tolerability of idebenone in Friedreichrsquos ataxia patients(MICONOS extension) Available at clinicaltrialsgovct2showNCT00993967Accessed Janaury 4 2019

8 Saute JAM Donis KC Serrano-Munuera C et al Ataxia rating scalesmdashpsychometricprofiles natural history and their application in clinical trials Cerebellum 201211488ndash504

9 Cano SJ Hobart JC Hart PE Korlipara LVP Schapira AHV Cooper JM In-ternational Cooperative Ataxia Rating Scale (ICARS) appropriate for studies ofFriedreichrsquos ataxia Mov Disord 2005201585ndash1591

10 Schmitz-Hubsch T du Montcel ST Baliko L et al Scale for the assessment and ratingof ataxia development of a new clinical scale Neurology 2006661717ndash1720

11 Burk K Malzig U Wolf S et al Comparison of three clinical rating scales in Friedreichataxia (FRDA) Mov Disord 2009241779ndash1784

12 Reetz K Dogan I Costa AS et al Biological and clinical characteristics of the Eu-ropean Friedreichrsquos Ataxia Consortium for Translational Studies (EFACTS) cohorta cross-sectional analysis of baseline data Lancet Neurol 201514174ndash182

13 Reetz K Dogan I Hilgers RD et al Progression characteristics of the EuropeanFriedreichrsquos Ataxia Consortium for Translational Studies (EFACTS) a 2 year cohortstudy Lancet Neurol 2016151346ndash1354

14 Reetz K Dogan I Hohenfeld C et al Nonataxia symptoms in Friedreich ataxia reportfrom the registry of the European Friedreichrsquos Ataxia Consortium for TranslationalStudies (EFACTS) Neurology 201891e917ndashe930

15 Patient registry of the European Friedreichrsquos Ataxia Consortium for TranslationalStudies (EFACTS) Available at clinicaltrialsgovct2showNCT02069509Accessed April 29 2019

16 Patel M Isaacs CJ Seyer L et al Progression of Friedreich ataxia quantitativecharacterization over 5 years Ann Clin Transl Neurol 20163684ndash694

17 FA Clinical Outcome Measures (FA-COMS) Available at clinicaltrialsgovct2showNCT03090789 Accessed April 29 2019

18 Subramony SH May W Lynch D et al Measuring Friedreich ataxia interraterreliability of a neurologic rating scale Neurology 2005641261ndash1262

19 Fahey MC Corben L Collins V Churchyard AJ Delatycki MB How is diseaseprogress in Friedreichrsquos ataxia best measured A study of four rating scales J NeurolNeurosurg Psychiatry 200678411ndash413

20 Tai G Corben LAGurrin L et al A study of up to 12 years of follow-up of Friedreich ataxiautilising four measurement tools J Neurol Neurosurg Psychiatry 201586660ndash666

Appendix Authors

Author Location Role Contribution

ChristianRummeyPhD

Clinical DataScienceGmbH Basel

Author Drafted the initialmanuscript revised themanuscript performed dataanalysis and conceived thestudy

LouiseCorben PhD

MurdochChildrenrsquosResearchInstitute

Author Performed data collectionand critical revision

MartinDelatyckiMD

MurdochChildrenrsquosResearchInstitute

Author Performed data collectionand critical revision

SHSubramonyMD

University ofFlorida

Author Performed data collectionand critical revision

Jennifer MFarmer MD

FriedreichrsquosAtaxiaResearchAlliance

Author Critical revision

David RLynch MDPhD

ChildrenrsquosHospital ofPhiladelphia

Author Performed data collectionsupervised collaborationand critical revision

Grace YoonMD

University ofToronto

Author Performed data collection

TheresaZesiewiczMD

University ofSouth Florida

Author Performed data collection

Joseph ChadHoyle MD

Ohio StateUniversity

Author Performed data collection

KatherineMathewsMD

University ofIowa

Author Performed data collection

Appendix (continued)

Author Location Role Contribution

GeorgeWilmot MD

Emory Author Performed data collection

ChristopherM GomezMD

University ofChicago

Author Performed data collection

KhalafBusharaMD

University ofMinnesota

Author Performed data collection

SusanPerlmanMD

UCLA Author Performed data collection

BernardRavina MD

University ofRochester

Author Performed data collectionand critical revision

NeurologyorgNG Neurology Genetics | Volume 5 Number 6 | December 2019 9

21 Lynch DR Farmer JM Tsou AY et al Measuring Friedreich ataxia complementaryfeatures of examination and performance measures Neurology 2006661711ndash1716

22 Friedman LS Farmer JM Perlman S et al Measuring the rate of progression inFriedreich ataxia implications for clinical trial design Mov Disord 201025426ndash432

23 Milne SC Murphy A Georgiou-Karistianis N Yiu EM Delatycki MB Corben LAPsychometric properties of outcome measures evaluating decline in gait in cerebellarataxia a systematic review Gait Posture 201861149ndash162

24 Regner SR Lagedrost SJ Plappert T et al Analysis of echocardiograms in a large hetero-geneous cohort of patients with Friedreich ataxia Am J Cardiol 2012109401ndash405

25 Spector P Summated Rating Scale Construction Newbury Park SAGE PublicationsInc 1992

26 Delatycki MB Evaluating the progression of Friedreich ataxia and its treatmentJ Neurol 200925636ndash41

27 Marelli C Figoni J Charles P et al Annual change in Friedreichrsquos ataxia evaluated bythe Scale for the Assessment and Rating of Ataxia (SARA) is independent of diseaseseverity Mov Disord 201227135ndash139

28 FARAmdashCollaborative Clinical Research Network in Friedreichrsquos Ataxia [online]Available at curefaorg Accessed April 29 2019

29 Marcotulli C Fortuni S Arcuri G et al GIFT-1 a phase IIa clinical trial to test the safety andefficacy of IFNγ administration in FRDA patients Neurol Sci 201637361ndash364

30 Zesiewicz T Heerinckx F De Jager R et al Randomized clinical trial of RT001 earlysignals of efficacy in Friedreichrsquos ataxia signals of Efficacy in Friedreichrsquos Ataxia MovDisord 2018331000ndash1005

31 Zesiewicz T Salemi JL Perlman S et al Double-blind randomized and controlledtrial of EPI-743 in Friedreichrsquos ataxia Neurodegener Dis Manag 20188233ndash242

32 Lynch DR Farmer J Hauser L et al Safety pharmacodynamics and potential benefitof omaveloxolone in Friedreich ataxia Ann Clin Transl Neurol 2018615ndash26

33 Lynch DR Hauser L McCormick A et al Randomized double-blind placebo-controlledstudy of interferon-γ 1b in Friedreich ataxia Ann Clin Transl Neurol 20196546ndash553

34 Hobart J Cano S Improving the evaluation of therapeutic interventions in multiple scle-rosis the role of new psychometric methods Health Technol Assess 2009131ndash177

35 R Core Team R A Language and Environment for Statistical Computing [online]Vienna R Foundation for Statistical Computing 2018 Available at R-projectorgAccessed April 29 2019

36 WickhamH Tidyverse easily install and load the ldquotidyverserdquo [online] 2017 Availableat CRANR-projectorgpackage=tidyverse Accessed April 29 2019

37 Revelle W Psych Procedures for Psychological Psychometric and Personality Re-search [online] Evanston Northwestern University 2018 Available at CRANR-projectorgpackage=psych

38 Metz G Coppard N Cooper JM et al Rating disease progression of Friedreichrsquosataxia by the International Cooperative Ataxia Rating Scale analysis of a 603-patientdatabase Brain 2013136259ndash268

39 Rummey C Loss of Ambulation in the CCRN-FANatural History Study Italy Pisa 201740 Lynch DR Deutsch EC Wilson RB Tennekoon G Unanswered questions in Frie-

dreich ataxia J Child Neurol 2012271223ndash122941 Galea CA Huq A Lockhart PJ et al Compound heterozygous FXN mutations and

clinical outcome in Friedreich ataxia insights from Frataxin structure and functionAnn Neurol 201679485ndash495

42 Lance CE Butts MM Michels LC The sources of four commonly reported cutoffcriteria what did they really say Organ Res Methods 20069202ndash220

43 Nunnally JC Psychometric Theory (2nd ed) New York NY McGraw-Hill 197844 Kaiser HF The application of Electronic computers to factor Analysis Educ Psychol

Meas 196020141ndash15145 Rasch G On General Laws and the Meaning of Measurement in Psychology

Copenhagen Denmark Danmarks paeligdagogiske Institut 196146 Lynch DR Farmer JM Wilson RL Balcer LJ Performance measures in Friedreich

ataxia potential utility as clinical outcome tools Mov Disord 200520777ndash78247 Andrich D A rating formulation for ordered response categories Psychometrika

197843561ndash57348 Muller H A Rasch model for continuous ratings Psychometrika 198752165ndash18149 Mair P Hatzinger R Extended Raschmodeling The eRm package for the application

of IRT models in R J Stat Softw [online serial] 200720 Accessed at jstatsoftorgv20i09 Accessed August 2 2019

50 Hohensinn C pcIRT an R package for polytomous and continuous Rasch modelsJ Stat Softw [online serial] 201884 Accessed at jstatsoftorgv84c02 AccessedAugust 2 2018

10 Neurology Genetics | Volume 5 Number 6 | December 2019 NeurologyorgNG

DOI 101212NXG000000000000037120195 Neurol Genet

Christian Rummey Louise A Corben Martin B Delatycki et al Psychometric properties of the Friedreich Ataxia Rating Scale

This information is current as of October 29 2019

reserved Online ISSN 2376-7839Published by Wolters Kluwer Health Inc on behalf of the American Academy of Neurology All rightsan open-access online-only continuous publication journal Copyright Copyright copy 2019 The Author(s)

is an official journal of the American Academy of Neurology Published since April 2015 it isNeurol Genet

ServicesUpdated Information amp

httpngneurologyorgcontent56371fullhtmlincluding high resolution figures can be found at

References httpngneurologyorgcontent56371fullhtmlref-list-1

This article cites 36 articles 1 of which you can access for free at

Citations httpngneurologyorgcontent56371fullhtmlotherarticles

This article has been cited by 1 HighWire-hosted articles

Subspecialty Collections

httpngneurologyorgcgicollectionspinocerebellar_ataxiaSpinocerebellar ataxia

httpngneurologyorgcgicollectionmitochondrial_disordersMitochondrial disorders

httpngneurologyorgcgicollectiongait_disorders_ataxiaGait disordersataxia

dy_cohort_case_controlhttpngneurologyorgcgicollectionclinical_trials_observational_stuClinical trials Observational study (Cohort Case control)

httpngneurologyorgcgicollectionclinical_neurology_examinationClinical neurology examinationfollowing collection(s) This article along with others on similar topics appears in the

Permissions amp Licensing

httpngneurologyorgmiscaboutxhtmlpermissionsits entirety can be found online atInformation about reproducing this article in parts (figurestables) or in

Reprints

httpngneurologyorgmiscaddirxhtmlreprintsusInformation about ordering reprints can be found online

reserved Online ISSN 2376-7839Published by Wolters Kluwer Health Inc on behalf of the American Academy of Neurology All rightsan open-access online-only continuous publication journal Copyright Copyright copy 2019 The Author(s)

is an official journal of the American Academy of Neurology Published since April 2015 it isNeurol Genet

ServicesUpdated Information amp

httpngneurologyorgcontent56371fullhtmlincluding high resolution figures can be found at

References httpngneurologyorgcontent56371fullhtmlref-list-1

This article cites 36 articles 1 of which you can access for free at

Citations httpngneurologyorgcontent56371fullhtmlotherarticles

This article has been cited by 1 HighWire-hosted articles

Subspecialty Collections

httpngneurologyorgcgicollectionspinocerebellar_ataxiaSpinocerebellar ataxia

httpngneurologyorgcgicollectionmitochondrial_disordersMitochondrial disorders

httpngneurologyorgcgicollectiongait_disorders_ataxiaGait disordersataxia

dy_cohort_case_controlhttpngneurologyorgcgicollectionclinical_trials_observational_stuClinical trials Observational study (Cohort Case control)

httpngneurologyorgcgicollectionclinical_neurology_examinationClinical neurology examinationfollowing collection(s) This article along with others on similar topics appears in the

Permissions amp Licensing

httpngneurologyorgmiscaboutxhtmlpermissionsits entirety can be found online atInformation about reproducing this article in parts (figurestables) or in

Reprints

httpngneurologyorgmiscaddirxhtmlreprintsusInformation about ordering reprints can be found online

reserved Online ISSN 2376-7839Published by Wolters Kluwer Health Inc on behalf of the American Academy of Neurology All rightsan open-access online-only continuous publication journal Copyright Copyright copy 2019 The Author(s)

is an official journal of the American Academy of Neurology Published since April 2015 it isNeurol Genet