
T州中0m=w。学号方 线L1作I=0eI四rd ORIGINAL ARTICLE Evaluation of an Automated Subjective Refractor JAMES SHEEDY,OD,PhD,FAAO,PAUL SCHANZ,BS, and MARK BULLIMORE,MCOptom,PhD,FAAO TrOh品ania的Codirgofopr Colr,OMo ABSTRACT:Purpose.To test the validity and repeatability of an automated subjective refractor compared with subjective refraction performed by clinicians.Corrected visual acuity was used to compare the validity of each technique.Methods.On each of two visits separated hy at least 1 week,the following were measured in 60)naive subjects (mean age,3311 yearsk:automated objective refraction (AOR)and automated subjective refraction (ASR) with the Topcon BV-1000,subjective refraction by one of two doctors (DR),and monocular visual acuity with the ASR and DR findings.Repeatability of refraction was calculated as the 95%limits of agreement (LuA)between same-mude measurements from the two visits.The validity of the ASR compared with DR was determined two ways:as the 95% LoA between the two measures and as the mean visual acuity measured with each prescription.Results.The DR and AOR had similar repeatability:95%LoA =-0.49 to +0.46 D and-0.51 to +0.45 D,respectivvely.The ASR had poorer repeatability with 95'%LoA of -0.70 to 0.71 D.Because of a calibration (offset)error in the right eye path of the BV-1000,the ASR of the right eye yielded refractions +0.26+0.41 D higher than the DR.For the right eye,visual acuity was 3.42 6.09 letters better with the DR than with the A5R,consistent with this calibration error.For the left eye,visual acuity was 1.20=5.20 letters better for the DR.Conclusions.Doctor-performed refractions are more valid and repeatable than the ASR performed with the BV-1000,although the differences are small.When calibrated,the BV-1000 gives similar visual acuity values to the DR,although the visual acuity for the DR is more repeatable.(Optom Vs5c2004:81334-340 Key Words:auto-refraction,automated refraction,refraction,visual acuity utamatod objecrive nfrarrinn (ACR)ingruments hae Most imporntly,A(ORin时ments have bcen小own to pro- been available for more than 25 years.The carly devices vide dinicaly vaid and repeatable measurrmenes..In moot in- such as the Bausch Lomb Ophthalmetron'were crude stances,the tepeatabilicy of AOR measureitsents is equal oe supe- and sow compand with the rapid instnments of todry.Farhy rior ta trditionall netimnscopy ar suhjcctive nfracionAOR machines showed goed agrement with retinosenpy and subjective mearm-nts are even as valid as a dactor'g suhjective refraction. refraction for spherical ametropi,but did not have the same per- These peevious tudeshowever.ssume that the doctor refrac- formance for astigmatism.Pappas et al.'demrmincd,in a stuy of tiun is the oorct rfraction.Another appeuach oward cetermin- 56 patients the AOR measitements wrre as doee to the final ing the validiry of the AOR is to ptient sarisfaction as the fimal sbjecsive prescription as were the prvious puair ofger.They arbiter of perfoemance.Bullimare t al.had prientwar a pair suggested the AOR finding could be used as a starting paint for of glaises based on a doctot's refraction and a pair buased on an fracinGradully,AOR instruments became mare rpeatable AOR for 3 wocks cach.They found subjocts prferred the glasas and wrrr shown tn have the xme npeatability as momentional made fmom thecktair'xnfraizion moer aften thain the glsaes mal refractive techniques when tested on 376 cyes.By 1984.the from the autoenated refractians hy a 2-to-1 ratio.This finding Humphrey autommatic refractor produced resules that were within along with a literature review by Goss and Grosvenot,ss 0.50 D of a doctur's refraction for sphere and within 037 D for th球hhough AO求sare第npcatable as docturs refraciuns they cylinde 100%af the time Reeent AOR instruments incorporate should not he the sole hasis an which a final persrzipeinn i such features as foggng techniques to ras accoemedation.con- determined. current meauremeat of comeal curvarure,and targets for visual Clearly.AOR has gained aocepcance into ophthalmic practice in acity and garc toting.Thocand uther advancements in technol- the Unitod States In many practias,thecurent AOR intrments ogy have belped impnwe the acrracy and repeatability of prvide the starting point for the shjeetive esamintion and are auto-refractor refined by the clinician hefore the final prescription is isued. wny aud Vi名aVd,5,Mgm时
ORIGINAL ARTICLE Evaluation of an Automated Subjective Refractor JAMES SHEEDY, OD, PhD, FAAO, PAUL SCHANZ, BS, and MARK BULLIMORE, MCOptom, PhD, FAAO The Ohio State University, College of Optometry, Columbus, Ohio ABSTRACT: Purpose. To test the validity and repeatability of an automated subjective refractor compared with subjective refraction performed by clinicians. Corrected visual acuity was used to compare the validity of each technique. Methods. On each of two visits separated by at least 1 week, the following were measured in 60 naive subjects (mean age, 33 11 years): automated objective refraction (AOR) and automated subjective refraction (ASR) with the Topcon BV-1000, subjective refraction by one of two doctors (DR), and monocular visual acuity with the ASR and DR findings. Repeatability of refraction was calculated as the 95% limits of agreement (LoA) between same-mode measurements from the two visits. The validity of the ASR compared with DR was determined two ways: as the 95% LoA between the two measures and as the mean visual acuity measured with each prescription. Results. The DR and AOR had similar repeatability: 95% LoA 0.49 to 0.46 D and 0.51 to 0.45 D, respectively. The ASR had poorer repeatability with 95% LoA of 0.70 to 0.71 D. Because of a calibration (offset) error in the right eye path of the BV-1000, the ASR of the right eye yielded refractions 0.26 0.41 D higher than the DR. For the right eye, visual acuity was 3.42 6.09 letters better with the DR than with the ASR, consistent with this calibration error. For the left eye, visual acuity was 1.20 5.20 letters better for the DR. Conclusions. Doctor-performed refractions are more valid and repeatable than the ASR performed with the BV-1000, although the differences are small. When calibrated, the BV-1000 gives similar visual acuity values to the DR, although the visual acuity for the DR is more repeatable. (Optom Vis Sci 2004;81:334–340) Key Words: auto-refraction, automated refraction, refraction, visual acuity Automated objective refraction (AOR) instruments have been available for more than 25 years. The early devices such as the Bausch & Lomb Ophthalmetron1 were crude and slow compared with the rapid instruments of today. Early machines showed good agreement with retinoscopy and subjective refraction for spherical ametropia, but did not have the same performance for astigmatism.2 Pappas et al.1 determined, in a study of 56 patients, the AOR measurements were as close to the final subjective prescription as were the previous pair of glasses. They suggested the AOR findings could be used as a starting point for refraction. Gradually, AOR instruments became more repeatable and were shown to have the same repeatability as conventional refractive techniques when tested on 376 eyes.3 By 1984, the Humphrey automatic refractor produced results that were within 0.50 D of a doctor’s refraction for sphere and within 0.37 D for cylinder 100% of the time.4 Recent AOR instruments incorporate such features as fogging techniques to relax accommodation, concurrent measurement of corneal curvature, and targets for visual acuity and glare testing. These and other advancements in technology have helped improve the accuracy and repeatability of auto-refractors. Most importantly, AOR instruments have been shown to provide clinically valid and repeatable measurements.5, 6 In most instances, the repeatability of AOR measurements is equal or superior to traditional retinoscopy or subjective refraction.7, 8 AOR measurements are even as valid as a doctor’s subjective refraction.9 These previous studies,5–9 however, assume that the doctor refraction is the correct refraction. Another approach toward determining the validity of the AOR is to use patient satisfaction as the final arbiter of performance. Bullimore et al.10 had patients wear a pair of glasses based on a doctor’s refraction and a pair based on an AOR for 3 weeks each. They found subjects preferred the glasses made from the doctor’s refraction more often than the glasses made from the automated refractions by a 2-to-1 ratio. This finding, along with a literature review by Goss and Grosvenor,5 suggests that although AORs are as repeatable as doctors’ refractions, they should not be the sole basis on which a final prescription is determined. Clearly, AOR has gained acceptance into ophthalmic practice in the United States. In many practices, the current AOR instruments provide the starting point for the subjective examination and are refined by the clinician before the final prescription is issued. 1040-5488/04/8105-0334/0 VOL. 81, NO. 5, PP. 334–340 OPTOMETRY AND VISION SCIENCE Copyright © 2004 American Academy of Optometry DOI: 10.1097/01.opx.0000134907.06027.b0 Optometry and Vision Science, Vol. 81, No. 5, May 2004

Aufomaled subicclive refrectur-Sheely el al.335 Rocent advances in technolgy have cnabled dvelupment of an ing fira at 32 foct and then at 20 foct.By using this appruach,the instrtment to perfurm automatrd subjective refeactien (ASR).The sat of ktters at the limits of th:cye's visial acuity wrrr different for BV-10100 refraction system by Topcon (Paramus,N])has the abil- each of the four measurements.Subjeets were ercoutaged toge iry to perform an AOR on bath eyes and alo a voice-guided ASR until an entire row ol's letters was mised.Credit was given for each indoding a bincoular balance. letter suocessfully identified and the total numaber of lerters read The objectives of this study were to issess the repeatabaliry oc the was recorded.Visaal acuiry soores were reoorded oa the Visual BV-1000's subjective refractice compared to a doctor-performed Acuiry Rating (VAR)scale.One point is given far every letter subjective refraction,to detenine the validiry of the BV-1000 suocesfully identified.A VARof100 comespoeds tovisual acuiry measarements in compison with a doctoc's subjective refraction. of 20/20,ind a VAR of 95 comrespouds to 20/25.VAR is directly and to compare the validry of the BV-1000 and doctor subjective rdated to lgMAR scores by the eqution:VAR-100-(50 X refractions by comparing visual xcuity measurements obtained logMAR). with bech tefractions. Data Analysis METHODS Right eve and left eye findings were analyoed stparateh.The Subjcctxwen recmitod to mect the followingcriteria:no known manoculr sphere,cylinder,andis findings fromn the balnced ocular diseise,age between 17 and 6years,nocurrent contact lens refractions were converted to power vetor valuess (spherical wear,no training in refractive technigues,corrected acuity of equivalent,J and ls for anahysis.Repeatability was assessed us 20/30 or better in cach eye,and no systeric discascs affocting ing the mechodsadvocatod by Blnd and Alman.The diffcrince nfractive erme,sch as diaberes.Sicty subjees (mean agr.33 11 between measurements from the two eapcrimental sessions was yrars)wereenrulled in the study.The Chin Seate University Office calealted for each subject and power veeror vale.Then the dis- of Responsible Reseach Pracrices approved the testing prococol. tribution of these differences wis described by cakulating,the and written intormed consent was obtainod fum all suhjocts be- mcan,the andard deviation (S1),and the5%limits of agr fure tegting Dita wrmr colleiand on two srpurate viits.At cach ncne nA及Ihs95%laA are caln由dl as the mran4 ifterenee± visit,an AOR.an ASR,and a docor's suhjective refraction (DR) 1.96 times the SD of the difference.The beradth of the LaA indi- were determined.The DR w perfoemed by one of wo licensed cates the repearability of the technique:the narrower the LaA,the optometrists.Seventeen subjects were refricted by doctor A(Dr. more repearable is the technique.The mean difererce represents A)an both visirc 20 wrre refractrd hy docnr H(Dr.B]on both the hias hetwren the twn sexsions.The validity of the ASR cam- visitx and 23 were refracted by different dotors (Dr.Mix)at the pred with the DR ws determined in a similar fahon.The dif. [wo visits ference berween the two values was calculated for each subject. The Topcon BV.1000 consists of dual automated objective in Then the distriburion of these differences wis described by calcu frared refracors.wich auto tacking of eye position and a mirror lating the mean.the SD.and the 95%LoA.Statistical significance chan projecion sycem.This is integrated with a computerized was determined using a rwo-tailed paired t-test and Levine's test of vision terer and approprate sofware,and houied withinatble oquality of variance. mounted instrumenc.The ASR peocedure includes preseating the patient with a series of computer voice-guided forced-choice ques RESULTS tions regarding visual stimui.Patints ropond by tilting ajoystick in the approrine dirceinn Targefare inchde Repeatability Landolt rings.duochrome targets,and various circr dot pat The repeatability of spherical equivalent for ASR and DR is terns.More complete operational decails have been previously shown in Figs.I and 2.repectivey.Data are shown for the right published. eye only,hecause data for left oye were similar.The data for the The AOR and ASR measurementx were all made by one af the clinicimn (DR)have a narower distribution,indicating bettern authors (PS)with the Topcon BV-1000.Ar each visit.the AOR peatahility than those for ASR.Summary data for rrpearabiliry af measurements were performed first.The testing order of ASR and spherical oquivalent.Jo.and Jas,are shown in Table 1.A positive DR alternated across subjects and betwen fint and scand visits an indica more positive valuc an the sxund vit,whrcas The aptometrist was pevidod with the AOR findings to svas the St)nf the difrrrrcrs (shnwn in purenthexrx)and the 95%LnA the staring point foe ther refraction using a Phoropior.The op- reflect the repeatahility.The mean values are almast all lexs than tometriss were masked co the ASR findings and to any previous 005Dand mot significantly different from zer.This indicates no fraciv findings.Theoptomctris prooxhres induded monoc- biaas h-two'n refractions framt the two visirs The 95%6 LoA,huw- ular sphere check Jackson cross cylinder tost,and hinocular blar evet.arr generally lower for DR and AOR than for ASR.epecially tulance. for the spherical equralent.This confirms the noticeably broader Using spherical and cylindrical til lenses intril frme,right distriburion of the daa poimts foe ASR repeatabaliry (Fig.1)com and left eye monocular vigl acuiry was mesured at each visit for pared with DR repeatabality(Fg 2)Forspherical oquivalent,ASR the ASR and DR findings with Bailey-Lovie logarithm of the min- repeatability was significantly poorer than the AOR repeatability imum nle of resolution (lgMAR)visl acuiry charts.The (Levent's test of oqualicy of variance,p<005)and DR repeat- night eve was measared firs,but the order of testing was alcernaced abiliry (p0001)for each eye.Although the AOR repeatabiliry between instrument and doctor.Chart memorization efects were was beter than the DR repestability for the right eye,the differ minimized by using two different Bailey-Lovie charts and by tes- ence did pot quite reach statistical significance (p =0.059). Oarowvemy and Virion Seiraer.Vol.81.Ne.5.Mar 204
Recent advances in technology have enabled development of an instrument to perform automated subjective refraction (ASR). The BV-1000 refraction system by Topcon (Paramus, NJ) has the ability to perform an AOR on both eyes and also a voice-guided ASR including a binocular balance. The objectives of this study were to assess the repeatability of the BV-1000’s subjective refraction compared to a doctor-performed subjective refraction, to determine the validity of the BV-1000 measurements in comparison with a doctor’s subjective refraction, and to compare the validity of the BV-1000 and doctor subjective refractions by comparing visual acuity measurements obtained with both refractions. METHODS Subjects were recruited to meet the following criteria: no known ocular disease, age between 17 and 60 years, no current contact lens wear, no training in refractive techniques, corrected acuity of 20/30 or better in each eye, and no systemic diseases affecting refractive error, such as diabetes. Sixty subjects (mean age, 33 11 years) were enrolled in the study. The Ohio State University Office of Responsible Research Practices approved the testing protocol, and written informed consent was obtained from all subjects before testing. Data were collected on two separate visits. At each visit, an AOR, an ASR, and a doctor’s subjective refraction (DR) were determined. The DR was performed by one of two licensed optometrists. Seventeen subjects were refracted by doctor A (Dr. A) on both visits; 20 were refracted by doctor B (Dr. B) on both visits; and 23 were refracted by different doctors (Dr. Mix) at the two visits. The Topcon BV-1000 consists of dual automated objective infrared refractors, with auto tracking of eye position, and a mirror chart projection system. This is integrated with a computerized vision tester and appropriate software, and housed within a tablemounted instrument. The ASR procedure includes presenting the patient with a series of computer voice-guided forced-choice questions regarding visual stimuli. Patients respond by tilting a joystick in the appropriate direction. Targets used for various tests include Landolt rings, duochrome targets, and various circular dot patterns. More complete operational details have been previously published.11 The AOR and ASR measurements were all made by one of the authors (PS) with the Topcon BV-1000. At each visit, the AOR measurements were performed first. The testing order of ASR and DR alternated across subjects and between first and second visits. The optometrist was provided with the AOR findings to serve as the starting point for their refraction using a Phoroptor. The optometrists were masked to the ASR findings and to any previous refractive findings. The optometrists’ procedures included monocular sphere checks, Jackson cross cylinder test, and binocular blur balance. Using spherical and cylindrical trial lenses in a trial frame, right and left eye monocular visual acuity was measured at each visit for the ASR and DR findings with Bailey-Lovie logarithm of the minimum angle of resolution (logMAR) visual acuity charts.12 The right eye was measured first, but the order of testing was alternated between instrument and doctor. Chart memorization effects were minimized by using two different Bailey-Lovie charts and by testing first at 32 feet and then at 20 feet. By using this approach, the set of letters at the limits of the eye’s visual acuity were different for each of the four measurements. Subjects were encouraged to guess until an entire row of 5 letters was missed. Credit was given for each letter successfully identified and the total number of letters read was recorded. Visual acuity scores were recorded on the Visual Acuity Rating (VAR) scale. One point is given for every letter successfully identified. A VAR of 100 corresponds to a visual acuity of 20/20, and a VAR of 95 corresponds to 20/25. VAR is directly related to logMAR scores by the equation: VAR 100 (50 logMAR). Data Analysis Right eye and left eye findings were analyzed separately. The monocular sphere, cylinder, and axis findings from the balanced refractions were converted to power vector values13 (spherical equivalent, J0, and J45) for analysis. Repeatability was assessed using the methods advocated by Bland and Altman.14 The difference between measurements from the two experimental sessions was calculated for each subject and power vector value. Then the distribution of these differences was described by calculating the mean, the standard deviation (SD), and the 95% limits of agreement (LoA). The 95% LoA are calculated as the mean difference 1.96 times the SD of the difference. The breadth of the LoA indicates the repeatability of the technique; the narrower the LoA, the more repeatable is the technique. The mean difference represents the bias between the two sessions. The validity of the ASR compared with the DR was determined in a similar fashion. The difference between the two values was calculated for each subject. Then the distribution of these differences was described by calculating the mean, the SD, and the 95% LoA. Statistical significance was determined using a two-tailed paired t-test and Levine’s test of equality of variance. RESULTS Repeatability The repeatability of spherical equivalent for ASR and DR is shown in Figs. 1 and 2, respectively. Data are shown for the right eye only, because data for left eye were similar. The data for the clinician (DR) have a narrower distribution, indicating better repeatability than those for ASR. Summary data for repeatability of spherical equivalent, J0, and J45, are shown in Table 1. A positive mean indicates a more positive value on the second visit, whereas the SD of the differences (shown in parentheses) and the 95% LoA reflect the repeatability. The mean values are almost all less than 0.05 D and not significantly different from zero. This indicates no bias between refractions from the two visits. The 95% LoA, however, are generally lower for DR and AOR than for ASR, especially for the spherical equivalent. This confirms the noticeably broader distribution of the data points for ASR repeatability (Fig. 1) compared with DR repeatability (Fig. 2). For spherical equivalent, ASR repeatability was significantly poorer than the AOR repeatability (Levene’s test of equality of variance,15 p 0.05) and DR repeatability (p 0.001) for each eye. Although the AOR repeatability was better than the DR repeatability for the right eye, the difference did not quite reach statistical significance (p 0.059). Automated subjective refractor—Sheedy et al. 335 Optometry and Vision Science, Vol. 81, No. 5, May 2004

336 Automated subjective refractor-Sheedy ct al. +1.0 but not for the right eye (mean.-0.07 D.t 1.12.p 0.30). ,+076 This indicates that,an averag.Dt.A had a tendency to refract approsimately 0.10 D mure minus than Dr.B. +00 +025 Validity 0.可 One method of asessing the validiry of the RV-10100 is to com- 0.2 pare the AORs and ASRs with the DRs.This was assexscd hy calculating the differenoe berwcen the BV-1000 refraction and the 0.50 docor refraction (ASR DR and AOR DR).Figs 3 and 4 Q.76 show the validiry of the ASR spherical equivalent on visit I for the right and lef cyes,respectively.The right eye ASR data are shiftod -1.00 800 6.C0 40m-280Q.0 +200 upward approximatdy +0.25 D,indicating that the ASR ison average,0.25 D more plus than the DR.No such hias was ubeervod Mean of Two Visits (D) for the left cye results Data from the second visit showed similar FIGURE 1. trends. Repedalilily ol aukeodl subjective nufractiun (A5R righl tye splria A summary of the validity danta for ASR and AOR and for both equivalent.The shaded anea repeesents the 95%limrs of agreemen.and visits is shuwn in Table 3.Consistent with the data in Figs.3 and 4.there wis a mean ASR bias of +0.26 D for the righ:eye and only +0.04 D for the kft eye for the finst visit.Thecorresponding values for the second visit were +0.31 D and +0.03 D.The bias in the +1.80 right eye was statistically significant for both visits (41.9.p 巨0.75 0.001 for buth).There also was a significant right eye AOR hias of 4050 +0.21 D and +0.14 D for the first and secund visits,respectively (t 3.6,p 0.001 for boch).For the left eye,the AOR was 4025 significanty more minus by 0.13 D on both visits (t2.9.p 030 0.01 for boch)compared with the DR:however.s reviewed above. the ASK for the left eye showed very little bias.No bias was found 02 for any of the astigmatic components of the refraction. 0.50 Visual Acuity 1.00 Visual acuiry with the ASRs and DRs was used as an impartial &.00 602 4.80-200 0.0 +200 comparison of the vaidity of the two techniques.The mean visual Mean of Two Visits(D) acuiry values for each prescription are shown in Table 4.Values are shown for each eye and for each visir.The difference berween che FIGUKE 2. visual acuiry scores (VAR)obtained with the ASRs and DRs are also shown in Table 4.The visual acuiry was significantly berter shaded area represents the 95'%limits of agreement.and the honzontal line reptnants the mun tifumnce btwemn viits,ur bia. with the DR than with the ASR for the right eye on both visits (t 4.1.p<0.001)and for the left eye on the first visit (t -2.91,p- 0.00).I lowever,there was no significant difference berween the Repearability was also assessed for the individual doctors by DR and the ASR fot the left eve visual acuiry on the second visit (t calculating values for the subaets of subjects exmined by the same =0.48.p =0.6).The visual acuiry difference for the right eye is doctor on hoch visits (17 for Dr.A and 20 foe Dr.B).The data are consistent with the difference of +0.25 D for the ASR right eye shown in Table 2.The mean within-docor difference between refraction because of the calibration error [Table 3). visits 1 and 2 was 0.03 D or less for refractive measures,indi- The repearahiliry of the vBual acuiry was also assessed by deter- cating no significant within-doctor beas berween sessions The mining,the difference berween the VAR scores for ASR and DR of 95%LoA for spherical oquivalent were nearly identical for the rwo each visit.The mean difference,the SD,and the 95%Lad are doctars,indicating no difference in repeatabiliry berween the wo shown in Table 5.The repeatability of visual acuity meurements doctors.Repearabiliry was also agsessed for the 2 subjects exam is bester with the doctor's prescription than with the prescripcio ined by a different doctor on each visit (Dr.Mix).To show beas by bised on the ASR as shown by the narrower 95%LoA.Levine's cest one doctor compared with the other.Dr.Mix differences were ofequality of variance showed the difference to be significant for always calealted as Dr.A-De.B.regandless of the clinician who the right eye (p 0.003)and lefi eye (p<0.001). performod the first visit.As might he espeerod,the 95%LoA were slightly broader and indicated slightly poorer berween-doctor re peatabiliry compared with within-doctor repeatabiliry.The mean Binocular Balance values in Table 2 show a significant difTerence in spherical euiv- The halnce herween the wo eyes was calculated as the differ- alent of-0.12D(t =2.74.p=0.01,paired t-test)foe the left eye ence in spherical epivalents hetween the right and left eye This Opeaueny aud Vaien Seieer.Val.81,No 5,May 204
Repeatability was also assessed for the individual doctors by calculating values for the subsets of subjects examined by the same doctor on both visits (17 for Dr. A and 20 for Dr. B). The data are shown in Table 2. The mean within-doctor difference between visits 1 and 2 was 0.03 D or less for all refractive measures, indicating no significant within-doctor bias between sessions. The 95% LoA for spherical equivalent were nearly identical for the two doctors, indicating no difference in repeatability between the two doctors. Repeatability was also assessed for the 23 subjects examined by a different doctor on each visit (Dr. Mix). To show bias by one doctor compared with the other, Dr. Mix differences were always calculated as Dr. A Dr. B, regardless of the clinician who performed the first visit. As might be expected, the 95% LoA were slightly broader and indicated slightly poorer between-doctor repeatability compared with within-doctor repeatability. The mean values in Table 2 show a significant difference in spherical equivalent of 0.12 D (t 2.74, p 0.01, paired t-test) for the left eye but not for the right eye (mean, 0.07 D, t 1.12, p 0.30). This indicates that, on average, Dr. A had a tendency to refract approximately 0.10 D more minus than Dr. B. Validity One method of assessing the validity of the BV-1000 is to compare the AORs and ASRs with the DRs. This was assessed by calculating the difference between the BV-1000 refraction and the doctor refraction (ASR DR and AOR DR). Figs. 3 and 4 show the validity of the ASR spherical equivalent on visit 1 for the right and left eyes, respectively. The right eye ASR data are shifted upward approximately 0.25 D, indicating that the ASR is, on average, 0.25 D more plus than the DR. No such bias was observed for the left eye results. Data from the second visit showed similar trends. A summary of the validity data for ASR and AOR and for both visits is shown in Table 3. Consistent with the data in Figs. 3 and 4, there was a mean ASR bias of 0.26 D for the right eye and only 0.04 D for the left eye for the first visit. The corresponding values for the second visit were 0.31 D and 0.03 D. The bias in the right eye was statistically significant for both visits (t 4.9, p 0.001 for both). There also was a significant right eye AOR bias of 0.21 D and 0.14 D for the first and second visits, respectively (t 3.6, p 0.001 for both). For the left eye, the AOR was significantly more minus by 0.13 D on both visits (t 2.9, p 0.01 for both) compared with the DR; however, as reviewed above, the ASR for the left eye showed very little bias. No bias was found for any of the astigmatic components of the refraction. Visual Acuity Visual acuity with the ASRs and DRs was used as an impartial comparison of the validity of the two techniques. The mean visual acuity values for each prescription are shown in Table 4. Values are shown for each eye and for each visit. The difference between the visual acuity scores (VAR) obtained with the ASRs and DRs are also shown in Table 4. The visual acuity was significantly better with the DR than with the ASR for the right eye on both visits (t 4.1, p 0.001) and for the left eye on the first visit (t 2.91, p 0.004). However, there was no significant difference between the DR and the ASR for the left eye visual acuity on the second visit (t 0.48, p 0.65). The visual acuity difference for the right eye is consistent with the difference of 0.25 D for the ASR right eye refraction because of the calibration error (Table 3). The repeatability of the visual acuity was also assessed by determining the difference between the VAR scores for ASR and DR of each visit. The mean difference, the SD, and the 95% LoA are shown in Table 5. The repeatability of visual acuity measurements is better with the doctor’s prescription than with the prescription based on the ASR as shown by the narrower 95% LoA. Levine’s test of equality of variance showed the difference to be significant for the right eye (p 0.003) and left eye (p 0.001). Binocular Balance The balance between the two eyes was calculated as the difference in spherical equivalents between the right and left eye. This FIGURE 1. Repeatability of automated subjective refraction (ASR): right eye spherical equivalent. The shaded area represents the 95% limits of agreement, and the horizontal line represents the mean difference between visits, or bias. FIGURE 2. Repeatability of doctor refraction (DR): right eye spherical equivalent. The shaded area represents the 95% limits of agreement, and the horizontal line represents the mean difference between visits, or bias. 336 Automated subjective refractor—Sheedy et al. Optometry and Vision Science, Vol. 81, No. 5, May 2004

Automatod subjcctive refmactor-Shccdy ct al.337 TABLE 1. Repeatability of refraction for cach technique Right eve Lelt eye ASR DR AOR ASR DR AOR Spherical Equivalent +0.D5±0.37 +0.01±D27 -D06±1.22 -1.D4±0.35 -0.02±022 -D02±1.27 -0.03±0.14 -0.01±0.16 002±0.15 -0.01±0.17 +0.01±0.10 000±0.12 +0.01±0.13 -0.01±000 -002±0.10 +0.01±0.14 -0.01±0.10 0.00±0.9 Sphere 95%limits 069:1D78 0.52040.53 -0.4810+0.37 0721o+D64 -0460+0.42 0.561D+052 ot agreement Values represent the mean (D)differences between visit 1 and 2.The 95%limits of apreemen for the sphere are also shon All values are diaptcrs. TABLE 2. Repeatability of refraction for the clinicians Right pye Left eye DeA Dr B Dr Mix DrA Dr B Dr Mix Spherical Equivalent -0.01±025 -003±0.25 -007±0.30 -0.0川±0.21 -0.03±0.22 -012±0.21 +0.02±0.17 000±0.09 -003±0.19 -0.03±0.11 -0.03±0.07 000±0.11 0.D±Da7 -D02±0.4 -012±0.12 +0.02±.10 +0.01±0.16 DC0±0.13 Sphere 95%limits of -0.50 to +0.48 -0.521D+0.46-0.63o+0.52 -0.420+0.40 -046e+0.40-0.53D+0.29 agreement Values represent the mean SD)differences between visit and 2 for sbjects examined hwice by the same doctor and those examined by different doctors at the two visits(Dr Mig.The 95%limits of agreement for the sphere are abso shown.All values are diopters. 百+1.0 包+150 +1.00 +1.0 0 +05 Q.02 n.no 5段 -1.00 -1.400 -1.0 -1.60 -009 -6.80 400 -200 000 +200 000 600 -400 -20 00 +20可 DR Spherical Equivalent(D) DR Spherical Equivalent (D) FIGLRE 3. FIGURE 4. Validry of autnmated suhjertvve mraction (AS ripht mye sherical Validity of automalod sbjecthve relaction (ASR:let eve splerical oquiv oquivakent on vis 1.The shaded area mpresents the 5%limis of alent on vist 1.The ares repnsents the 99%lmits of agreement agneement,and the horiznneal liee mpresents the mean diffeeence he- and the horioontal line reprsents the meen ciflerence between auloranedl tween automaled subjective relraction and docor tdfraction.or bias. suhjective retraction and docer nrartion,nr hias. was calculted sepuratey for ASRs,DRs,and AOR&Mean and indicating similar repeatability of the balance with all three refrac standand devttions of the halance are shown in Table 6.The mean tion methndk valnes for DR are clse to zero,whereas the mean values for ASR and AOR arc approximately 0.25 D.The balnce differenoc of DISCUSSION 0.25 D with ASR and AOR is cunsisoent with the +0.25 D hias of the right eye (Table 3).The repearability of the halance was then Repeatability determined by caleulating the difference between the bilances The tepeatabiliry of refraction is very important in clinical pra measured on the two sessions The rosulting 95%InA for ASR. tice and rocarch.An instrment should pruvide the sme resulr. DR.and AOR halanoe shown in Tabl 6 an remarkably similar,within fairly narrow limits,on retesting.The repeatability of AOR Cpeauemy au Viion Scimare.Val.N1,No.5.May 2004
was calculated separately for ASRs, DRs, and AORs. Mean and standard deviations of the balance are shown in Table 6. The mean values for DR are close to zero, whereas the mean values for ASR and AOR are approximately 0.25 D. The balance difference of 0.25 D with ASR and AOR is consistent with the 0.25 D bias of the right eye (Table 3). The repeatability of the balance was then determined by calculating the difference between the balances measured on the two sessions. The resulting 95% LoA for ASR, DR, and AOR balance shown in Table 6 are remarkably similar, indicating similar repeatability of the balance with all three refraction methods. DISCUSSION Repeatability The repeatability of refraction is very important in clinical practice and research. An instrument should provide the same result, within fairly narrow limits, on retesting. The repeatability of AOR TABLE 1. Repeatability of refraction for each technique Right eye Left eye ASR DR AOR ASR DR AOR Spherical Equivalent 0.05 0.37 0.01 0.27 0.06 0.22 0.04 0.35 0.02 0.22 0.02 0.27 J0 0.03 0.14 0.01 0.16 0.02 0.15 0.01 0.17 0.01 0.10 0.00 0.12 J45 0.01 0.13 0.01 0.08 0.02 0.10 0.01 0.14 0.01 0.10 0.00 0.09 Sphere 95% limits of agreement 0.69 to 0.78 0.52 to 0.53 0.48 to 0.37 0.72 to 0.64 0.46 to 0.42 0.56 to 0.52 Values represent the mean ( SD) differences between visit 1 and 2. The 95% limits of agreement for the sphere are also shown. All values are diopters. ASR, automated subjective refraction; DR, doctor subjective refraction; AOR, automated objective refraction. TABLE 2. Repeatability of refraction for the clinicians Right eye Left eye Dr A Dr B Dr Mix Dr A Dr B Dr Mix Spherical Equivalent 0.01 0.25 0.03 0.25 0.07 0.30 0.01 0.21 0.03 0.22 0.12 0.21 J0 0.02 0.17 0.00 0.09 0.03 0.19 0.03 0.11 0.03 0.07 0.00 0.11 J45 0.00 0.07 0.02 0.04 0.02 0.12 0.02 0.10 0.01 0.06 0.00 0.13 Sphere 95% limits of agreement 0.50 to 0.48 0.52 to 0.46 0.63 to 0.52 0.42 to 0.40 0.46 to 0.40 0.53 to 0.29 Values represent the mean ( SD) differences between visit 1 and 2 for subjects examined twice by the same doctor and those examined by different doctors at the two visits (Dr Mix). The 95% limits of agreement for the sphere are also shown. All values are diopters. FIGURE 3. Validity of automated subjective refraction (ASR): right eye spherical equivalent on visit 1. The shaded area represents the 95% limits of agreement, and the horizontal line represents the mean difference between automated subjective refraction and doctor refraction, or bias. FIGURE 4. Validity of automated subjective refraction (ASR): left eye spherical equivalent on visit 1. The shaded area represents the 95% limits of agreement, and the horizontal line represents the mean difference between automated subjective refraction and doctor refraction, or bias. Automated subjective refractor—Sheedy et al. 337 Optometry and Vision Science, Vol. 81, No. 5, May 2004

338 Automatod subjoctive refmactor-Sheedy ct al. TABLE 3. Validity of refraction for ASK and AOR Righ城ee Left eye ASR AOR ASK AOR Firsl visit Sphcrical Equivalcnt +0.26±0.41“ +021±0.31 +0.4±039 -0.13±0.34 +0.01±0.16 -004±0.14 +0.1±017 +.01±0.12 -0.01±0.14 +002±0.1 -0.03±0.15 -0.02±0.10 Second visit Spherical Equivalent +.31±0.4学 +D.14±0.0 +0.03±0.42 -h13±0.3 0.00±0.12 -0.01±0.10 -0.01±0.10 0.00±0.11 las +0.01±0.08 +0.01±0.09 -0.01±0.11 +0.02±0.47 Values represert the mean SD)differences between the BV-1000 and the dotor refraction of each visit and for each eye.All values are diopters. hias is significant,PAlthough the DR The repeatability of refraction by the two individual doctors and AOR had similar 95%LaA of approximately0.50 D,the (Table 2)was similar.Dr.B had more repeatable cylinder mea Cydaweng au Vaisn Srimmen Val.81.No.5.May 2004
with the BV-1000 was very similar to that of the DR. This supports the results of previous studies that have found that the repeatability of AOR is equal to or superior to that of DR.5–9 Although the DR and AOR had similar 95% LoA of approximately 0.50 D, the ASR had a larger 95% LoA of approximately 0.75 D (Table 1). Therefore, the ASR was not as repeatable as the DR or AOR. The repeatability of refraction by the two individual doctors (Table 2) was similar. Dr. B had more repeatable cylinder meaTABLE 3. Validity of refraction for ASR and AOR Right eye Left eye ASR AOR ASR AOR First visit Spherical Equivalent 0.26 0.41a 0.21 0.31a 0.04 0.39 0.13 0.34a J0 0.01 0.16 0.04 0.14 0.01 0.17 0.01 0.12 J45 0.01 0.14 0.02 0.11 0.03 0.15 0.02 0.10 Second visit Spherical Equivalent 0.31 0.40a 0.14 0.30a 0.03 0.42 0.13 0.30a J0 0.00 0.12 0.01 0.10 0.01 0.10 0.00 0.11 J45 0.01 0.08 0.01 0.09 0.01 0.11 0.02 0.47 Values represent the mean ( SD) differences between the BV-1000 and the doctor refraction of each visit and for each eye. All values are diopters. a bias is significant, P 0.05. ASR, automated subjective refraction; AOR, automated objective refraction. TABLE 4. Mean visual acuity (SD) obtained with each prescription (visual acuity rating, in letters) 1st visit 2nd visit Right eye Left eye Right eye Left eye DR 103.6 3.8 103.7 4.3 103.6 4.2 103.1 4.1 ASR 99.9 5.9 101.6 5.4 100.4 6.3 102.8 4.8 Difference 3.7 6.1a 2.1 5.6a 3.2 6.0a 0.3 4.8 a difference is significant, P 0.05. Data are shown for ASR (automated subjective refraction) and DR (doctor subjective refraction) prescriptions, for each eye and for each study visit. The repeatability of refraction the visual acuity is also shown. Values represent the mean ( SD) differences between visit 1 and 2. The 95% limits of agreement for the sphere are also shown. All values are diopters. TABLE 5. Repeatability of visual acuity in VAR units (one point one letter): visit #2-visit #1 ASR DR Right eye Left eye Right eye Left eye Mean 0.6 6.3 1.3 4.9 0.0 4.3 0.6 3.5 Limits of agreement 11.7 to 12.9 8.4 to 11.0 8.5 to 8.5 7.5 to 6.3 VAR, visual acuity rating (in letters); ASR, automated subjective refraction; DR, doctor subjective refraction. TABLE 6. Balancea ASR AOR DR 1st visit 0.19 0.58 0.30 0.56 0.04 0.54 2nd visit 0.27 0.53 0.27 0.56 0.01 0.60 Mean difference ( SD) 0.08 0.30 0.03 0.25 0.03 0.23 95% LoA 0.52 to 0.68 0.53 to 0.46 0.42 to 0.47 a Right eye minus left eye spherical equivalent in diopters. ASR, automated subjective refraction; AOR, automated objective refraction; DR, doctor subjective refraction; LoA, limits of agreement. 338 Automated subjective refractor—Sheedy et al. Optometry and Vision Science, Vol. 81, No. 5, May 2004

Auromatod suboctive refractor-sheody et al.339 sures.The repeatabdlity of refractions meaured by the same doctor such as the BV-1000.In contrast,the infoemrion that is glaned on both visats was slightly becter than the repearabiliry of refrac. from the narure of che patient's response and is inteed into the tios mered by different doctors for the right eye.The within clinician's art of refraction would be significanty more difficult to doctor repeatabity and the between-docor repeatability are sim incorporate into an automated testing device. ilar for the left cye,but not foe the right eye.For subjects examined ono cach by the rwo docurs [Dr.Mix),Dr.A refractod about 0.10 D more minus than did Dr.B.A similar finding wis reported by Balance Bullimore et al.,who found one of their dinicians refracted,on The balance of the BV-1000 refraction was very simiar to that average.0.12 D mure minus than the ocher.The potential for of a doctor's refraction.The summary findings in Table 6 show herwon-clinician differenees cmphasxes the value of abjectie that the repeatahiliry of the balanee waimilar for ASR,AOR,and tochniees in studiex of refractive erur. DR.The mean differences show that the autamated halanee&are approximacdy+025 Ddifferent from docor balances,consisent Validity with and becau of the carfier noccd +0.25 D ingrument affoct (calihrarian errar)in the right eye. There was a calibeation (offet)error ofapprooimtely 01.2D in the righe eye pach of the instrument (Table 3 and Fg 3).This CONCLUSIONS afcocd the ASR and AOR of the right cye by yidding refractions that were appeuximately +0.25 D higher than the DR.The cali- The resulrx indicare that DRxare more valid and repeatahle than bration ermr was confirmed during a poststudy service visit by a the ASR performed by the BV-1000.although the differences are Topcon technician.No sgnificant bias was observed fot the cylin- small.The paricular unic tesed in this study had a calibration der findings (o and Js in Tabl 3).Once the calibration eror in enor of+0.25 D for the right eye.This error was easily corrected, the righe eve is accounted for,the mean spherical equivalent of and with proper calibration.the BV-1000 can,on average,pcovide both eyes of the ASR are simlar to the refractions of the DR.A the sme tefractive resules as DR.For the left eye,which was notewoethy observation is that the left AOR was different than the properiy calibrated,the difference in visual acuiry between the DR DR on each visit,but the ASR was noc.This means that the ASR and ASR prescriptions was small(1.2 lecters)and would likely pot improved on the AOR for a resultant refraction closer to that o the be noticeable in dinical practice. DR With proper calibration,the BV-1000 ASR woukd,on aver- In evaluating instrument perfommnce,however.repeatability age.yied similar refrcions.in both theb and lefreyes tothose y be more impor.For the left (properly calibraed)eye,the found by a dinician. DR had a narrower 95%LaA range than ASR (abour+0.50 D compared with0.75 D)and a narrower LaA for visual acuiry Visual Acuity m3 urmints labour±l53 ciy lines comp四red with±2.0 acuiry lin).These LoA differences might he moticeable in a din- Visl acuiry was measured with the ASRs and DRs,enablinga ical setting comparison of their validity without the assumption that the doc- In this study,the DR and ASR started with the AOR measure- ur is oor For buth cyes and foe both visits,visual aity with menes of the BV-1000.It is uncertain whcther the ASR mfrattion the DR wats herter than that with the ASR refraction.The differ- was any better than the AOR mfractinn (i.c,whether the ASR ence was statistically significant foe all viies and cnnditions,except impenved on the AOR).AOR repeatbility does pot differ from for the left eye on the secood visit (Table )Visual acuicy with the doctor repeatabiliry but the ASR is not pearable.Nonetheless, DR was,on arag 34 ktters herter for the right cys.This is the data shawn in Table 3 indicate that for the lefr epe (ic,the ons mainly attributahle to the instnament calihration crror for this cye. for which there was noe a calihratian crror),the ASR was more For the lefi eye,the difference is smaller.averaging 1.2 letters for valid than the AOR.This study did not include visual acuiry mea the rwo visits.This small difference cocresponds tofracrion ofa surements with the AO;hence.data are not available to compare line and would probably pot be dinically meaningful in daily pa the validiry of ASR and AOR using visual acuiry. tient care,given the SD of visual acuity measurement. The repeatabdiry of visual acuiry with the DR was abo better ACKNOWLEDGMENTS than with the ASR.The 95%LoA for visual acuiry with the ASR were approuimaely2 lines (10 letters)whereas chey were ap Thur y r aprand fy Toen Crpareiae. prooimately1.5 lines (7 ltters)with the DR (Tabk 5).The Rw62I,2阅浮图wy222 narrower range of the DR indicatesa moee repeatable visual acuiry. but agin the difference is small and may not be dinically REFERENCES meaningful. A doctoe performing a refraction can adjust for various aspects 1.Pappas J.Amderson DR.Briete F Is the Aurorefacror reading of human behsrvior.For campk,akhough many of a piticnt's coiest co manifesc pefrsction?A oomparison of the patient's previous spectackes and the 6600 Auooeefracooe reading Arch Ophthalmol npones daring the esamination may he yes or no and I ar 2 the 1978:9k997-3. clinician can lisen to the speed and oonfidence with which the 2.Deance SM.Miihell Dw.Schuber M.Sodies of an aurematic paticnt responds.If a patient sm uncertain,the clinician can refraction machine Can JOphthalmnl 1975;10b462-8. npear any numher af the steps.The simpl rules that make up the S.Rannon RE.A new amomaned sbjecrive opeomerer.Am JOpoom eienee of rrfraction are rdatively casy tn ineorparate into a device %ia0pt1977:54433-8. (preoveng aw Virion Seimare,Vol.81.Ne.5.May 2001
sures. The repeatability of refractions measured by the same doctor on both visits was slightly better than the repeatability of refractions measured by different doctors for the right eye. The withindoctor repeatability and the between-doctor repeatability are similar for the left eye, but not for the right eye. For subjects examined once each by the two doctors (Dr. Mix), Dr. A refracted about 0.10 D more minus than did Dr. B. A similar finding was reported by Bullimore et al.,10 who found one of their clinicians refracted, on average, 0.12 D more minus than the other. The potential for between-clinician differences emphasizes the value of objective techniques in studies of refractive error. Validity There was a calibration (offset) error of approximately 0.25 D in the right eye path of the instrument (Table 3 and Fig. 3). This affected the ASR and AOR of the right eye by yielding refractions that were approximately 0.25 D higher than the DR. The calibration error was confirmed during a poststudy service visit by a Topcon technician. No significant bias was observed for the cylinder findings (J0 and J45 in Table 3). Once the calibration error in the right eye is accounted for, the mean spherical equivalent of both eyes of the ASR are similar to the refractions of the DR. A noteworthy observation is that the left AOR was different than the DR on each visit, but the ASR was not. This means that the ASR improved on the AOR for a resultant refraction closer to that of the DR. With proper calibration, the BV-1000 ASR would, on average, yield similar refractions, in both the right and left eyes, to those found by a clinician. Visual Acuity Visual acuity was measured with the ASRs and DRs, enabling a comparison of their validity without the assumption that the doctor is correct. For both eyes and for both visits, visual acuity with the DR was better than that with the ASR refraction. The difference was statistically significant for all visits and conditions, except for the left eye on the second visit (Table 4). Visual acuity with the DR was, on average, 3.4 letters better for the right eye. This is mainly attributable to the instrument calibration error for this eye. For the left eye, the difference is smaller, averaging 1.2 letters for the two visits. This small difference corresponds to a fraction of a line and would probably not be clinically meaningful in daily patient care, given the SD of visual acuity measurement. The repeatability of visual acuity with the DR was also better than with the ASR. The 95% LoA for visual acuity with the ASR were approximately 2 lines (10 letters) whereas they were approximately 1.5 lines (7 letters) with the DR (Table 5). The narrower range of the DR indicates a more repeatable visual acuity, but again the difference is small and may not be clinically meaningful. A doctor performing a refraction can adjust for various aspects of human behavior. For example, although many of a patient’s responses during the examination may be yes or no and 1 or 2, the clinician can listen to the speed and confidence with which the patient responds. If a patient seems uncertain, the clinician can repeat any number of the steps. The simple rules that make up the science of refraction are relatively easy to incorporate into a device such as the BV-1000. In contrast, the information that is gleaned from the nature of the patient’s response and is integrated into the clinician’s art of refraction would be significantly more difficult to incorporate into an automated testing device. Balance The balance of the BV-1000 refraction was very similar to that of a doctor’s refraction. The summary findings in Table 6 show that the repeatability of the balance was similar for ASR, AOR, and DR. The mean differences show that the automated balances are approximately 0.25 D different from doctor balances, consistent with and because of the earlier noted 0.25 D instrument offset (calibration error) in the right eye. CONCLUSIONS The results indicate that DRs are more valid and repeatable than the ASR performed by the BV-1000, although the differences are small. The particular unit tested in this study had a calibration error of 0.25 D for the right eye. This error was easily corrected, and with proper calibration, the BV-1000 can, on average, provide the same refractive results as DR. For the left eye, which was properly calibrated, the difference in visual acuity between the DR and ASR prescriptions was small (1.2 letters) and would likely not be noticeable in clinical practice. In evaluating instrument performance, however, repeatability may be more important. For the left (properly calibrated) eye, the DR had a narrower 95% LoA range than ASR (about 0.50 D compared with 0.75 D) and a narrower LoA for visual acuity measurements (about 1.5 acuity lines compared with 2.0 acuity lines). These LoA differences might be noticeable in a clinical setting. In this study, the DR and ASR started with the AOR measurements of the BV-1000. It is uncertain whether the ASR refraction was any better than the AOR refraction (i.e., whether the ASR improved on the AOR). AOR repeatability does not differ from doctor repeatability but the ASR is not as repeatable. Nonetheless, the data shown in Table 3 indicate that for the left eye (i.e., the one for which there was not a calibration error), the ASR was more valid than the AOR. This study did not include visual acuity measurements with the AO; hence, data are not available to compare the validity of ASR and AOR using visual acuity. ACKNOWLEDGMENTS This study was supported by Topcon Corporation. Received November 21, 2003; accepted February 22, 2004. REFERENCES 1. Pappas CJ, Anderson DR, Briese FW. Is the Autorefractor reading closest to manifest refraction? A comparison of the patient’s previous spectacles and the 6600 Autorefractor reading. Arch Ophthalmol 1978;96:997–8. 2. Drance SM, Mitchell DW, Schulzer M. Studies of an automatic refraction machine. Can J Ophthalmol 1975;10:462–8. 3. Bannon RE. A new automated subjective optometer. Am J Optom Physiol Opt 1977;54:433–8. Automated subjective refractor—Sheedy et al. 339 Optometry and Vision Science, Vol. 81, No. 5, May 2004

340 Automaled subjective refractor-Sheedy et al. Wong EKJs.Pandla VM.Prant MV,Myers Gaster RN,Leopold 11.Dave T.Topcon BV-1000 aumomared subjective refncrion syutems IH.(inical cauiun ef the Humghrey auumis reratur.Arck 0 plomeiry1edy2003:4:52-7. 0h山lnum198:102:870-5. 12.Baiky I Lovie IE.New deigs priaciples for visl acuiry leter 5.Ginu DA,Gimuvennr T.Relihility of rfraction-a lirerature review. charts Am Opioe Phynid Opt 1976;53:740-5. Am Cpwm Aou.199%67619-30. 15.Thib LN.Wheeker W,Hura:D.Borwer vectunc a aplicatiun ul i MeRrien NA,Milndor M.CTical evlruticn of the Canon Autnref Fourier analysis tn the dearrprion and xratigcal analysis of refractiv R-1.Am Opom Physiol (pt 1985:62-785-92. 年mr.0 pinm Vis5i197:74:367-75 2.Zadnik K.Mumi DO.Adams AJ.The repearabiliry of mearement 14.Bland JM.Alman DG.Srarisrical methois for assesing igreement of the ocubar componenu Invest Ophtslatul Vis Si 1992:33 between two mechods of diaiea measartmcnt.Lances 1956cl: 232-33. 3F=10. 8.BulEre MA.Facru RE Adanu (The repeeablity uf aut- 15.Levene H.Rubot tots loe the rrly ul varince In:Olin I.od mocod ad clinicbet pefraction.Opoom Vis Sci 1998:75:617-22. Centributions to Probabdiry and Statistis.Stnford.CA:Stanford 9.Allen PM,Rxhakrishean H,CyLrary D].Rrpeatability and validity Univenity Prex 1960:278-92 of che PowerRefractor and te Nidek AR6-A in an adult popula. ticn with healthy epes.Optom Via Sci 2003:80-245-51. 10.Bullimore MA.Adoms CW,Fusgo RE.Bouman M.Comeral RM James F.Sheedy Ng Sarver J.Twelber D.Graham AD.Parient acoepeance of auro- e Chio Statr Uniserit的 nd dinician precipi:omidtial.I College of Opaeery Voual Scienee and Io Applicatinm,vul 1.1996 OSA Technic 望uTw6Ae Dpot Setio.Waslingion.DC:Opioll Soticty of Amtris 1996c Cobenebwk (hio 43218 14-7. ail:uvrdy.2s0oev.rose Oproy and Virion Serare,Vol.1,No.5.May 204
4. Wong EK Jr, Patella VM, Pratt MV, Myers SW, Gaster RN, Leopold IH. Clinical evaluation of the Humphrey automatic refractor. Arch Ophthalmol 1984;102:870–5. 5. Goss DA, Grosvenor T. Reliability of refraction—a literature review. J Am Optom Assoc 1996;67:619–30. 6. McBrien NA, Millodot M. Clinical evaluation of the Canon Autoref R-1. Am J Optom Physiol Opt 1985;62:786–92. 7. Zadnik K, Mutti DO, Adams AJ. The repeatability of measurement of the ocular components. Invest Ophthalmol Vis Sci 1992;33: 2325–33. 8. Bullimore MA, Fusaro RE, Adams CW. The repeatability of automated and clinician refraction. Optom Vis Sci 1998;75:617–22. 9. Allen PM, Radhakrishnan H, O’Leary DJ. Repeatability and validity of the PowerRefractor and the Nidek AR600-A in an adult population with healthy eyes. Optom Vis Sci 2003;80:245–51. 10. Bullimore MA, Adams CW, Fusaro RE, Bauman M, Cotteral RM, Ng Sarver J, Twelker D, Graham AD. Patient acceptance of autorefractor and clinician prescriptions: a randomized clinical trial. In: Visual Science and Its Applications, vol. 1. 1996 OSA Technical Digest Series. Washington, DC: Optical Society of America, 1996: 194–7. 11. Dave T. Topcon BV-1000 automated subjective refraction system. Optometry Today 2003;4:52–7. 12. Bailey IL, Lovie JE. New design principles for visual acuity letter charts. Am J Optom Physiol Opt 1976;53:740–5. 13. Thibos LN, Wheeler W, Horner D. Power vectors: an application of Fourier analysis to the description and statistical analysis of refractive error. Optom Vis Sci 1997;74:367–75. 14. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1: 307–10. 15. Levene H. Robust tests for the equality of variance. In: Olkin I, ed. Contributions to Probability and Statistics. Stanford, CA: Stanford University Press, 1960:278–92. James E. Sheedy The Ohio State University College of Optometry 320 West Tenth Ave. Columbus, Ohio 43218 e-mail: sheedy.2@osu.edu 340 Automated subjective refractor—Sheedy et al. Optometry and Vision Science, Vol. 81, No. 5, May 2004