ORIGINAL CONTRIBUTION

Neurovisual Rehabilitation in Cerebral BlindnessGeorg Kerkhoff, PhD; Udo Münßinger, MA; Elke K. Meier, MA

Ob jective: The efficacy of a systematic trainingof saccadiceye movements was evaluated in hemianopic patients withthree main objectives: (1) to determine the role of visualfield recovery, (2) to assess the transfer of treatment gainsto functional outcome measures, and (3) to evaluate thepatients' subjective experience throughout therapy.

Design: Within-subject repeated measures design. Themean follow-up interval was 3 months (ränge, l to 10months).

Setting: Outpatients of a day clinic for the treatment ofneuropsychological disorders that is associated with a cityhospital.

Patients: A consecutive sample of 22 hemianopic pa-tients without neglect after unilateral stroke. Follow-upwas possible in all cases.

Interventions: Saccadic eye movement strategies weretreated regularly (30-minute daily sessions 5 days perweek; 25 to 27 total treatment sessions).

Main Outcome Measures: Visual perimetry results,

visual search field within the scotoma, visual search onprojected slides with wide eccentricity, search times foridentifying objects visually on a table (table test), and stan-dardized rating of the degree of subjective visual impair-ment due to the field defect. All outcome measures wereplanned before initiation of the study.

Results: (1) Increase in visual search field size (mean,30°). (2) Training-related visual field increases in 12 (54%)of 22 patients (mean increase, 6.7°; ränge, 2° to 24°). (3)Transfer of treatment gains to functional measures (tabletest) and improvement after training in patients' subjec-tive rating of their visual impairments. (4) Stabilityofim-provements at the 3-month follow-up visit. (5) Returnto part-time work in 20 (91%) of 22 patients. All men-tioned results were significant (nonparametric tests; etlevel, .05; two-sided; adjusted for the number of tests).

Conclusions: Training of compensatory eye movementstrategies restores oculomotor functions, improves per-formance in functional visual activities, and reintegrateshemianopic patients into vocational life.

(Arch Neurol 1994;51:474-481)

From the EKNEntwicklungsgruppe KlinischeNeuropsychologie, CityHospital Bogenhausen, Munich,Germany.

H OMONYMOUS VISUAL fielddefects (VFDs) are one ofthe most frequent conse-quences of brain damage.]

Approximately 20% to30% of all patients with cerebrovascular in-farction requiring therapy in a rehabilitationcenter have homonymous VFDs.2 Amongthese, 70% show a visual field sparing of lessthan 5° of visual angle.1 Patients with VFDshave repeatedly been found to have a poorrehabilitation outcome, äs measured by analy-sis of activities of daily living (ADL), functionalStatus outcome indicated by the Barthel In-dex,3 or vocational rehabilitation success.4'5

In addition, Uzzell et al6 found that patients

with head trauma with VFDs exhibit morewidespread neuropsychological defects thancomparable patients without VFDs. Althoughother factors might also contribute to the un-favorable outcome of patients with VFDs inrehabilitation, it is surprising that VFDs havenot been a major target for rehabilitation pro-grams, äs has been the case for aphasia6-7 orvisual neglect.8'10 Quantitative studies of vi-

See Patients and Methodson next page

ARCH NEURO17VOL 51, MAY 1994474

PATIENTS AND METHODS

PATIENTS

Twenty-two patients with VFDs without visual neglect (16male and six female) with a mean age of 46 years (ränge, 16to 77 years) were included in the sample. Seven patients hadleft-sided homonymous hemianopia and 15 had right-sidedhemianopia. All patients had a unilateral vascular brain lesiondocumentedby clinical (field defect) andneuroradiologic (mag-netic resonance imaging) findings. Mean time after onset was7.5 months (ränge, l to 37 months). None of the patients hadhemiparesis. All patients underwent detailed neuropsycho-logical examinations (for details, see von Cramon and Zihl24).Patients with inadequate fixation during visual perimetry orvisual search field testing and patients with oculomotor nervepalsies or central oculomotor disorders (see Leigh and Zee25)were excluded. All patients had normal function of the an-terior visual pathways, äs evaluated by orthoptic and ophthal-mologic tests (fundus and slit-lamp examinations). Correctedbinocular visual acuity (Snellen letter chart) was at least 20/30for the near (40-cm) and far (6-m) distances. All patients un-derwent detailed assessment of their visual complaints by asystematic questionnaire before and after visual training (seebelow). Patients with visual neglect were excluded from thestudy by testing horizontal line bisection and drawing of sym-metrical figures from memory (for details, see Kerkhoff et al11).Because patients with neglect deviate to the ipsilesionalhemispace in horizontal line bisection26 and patients with VFDsdeviate to the contralesional scotomatous hemifield,27'30 thisdeviation was taken äs one criterion to differentiate betweenpatients with VFDs and those with neglect. All patients showeda displacement of the truncation point to the hemianopic hemi-field beyond the cutoff scores. None of the 22 patients showedevidence of visual neglect in the drawing task.

TESTS

In designing the different measures of visual performance inour patients, we referred to the World Health Organization clas-sification of levels based on which any disease may be analyzed(see Wade31)- These levels are pathology, impairment, disabil-ity, and Handicap. In this context, pathology is to be understoodäs the underlying brain damage, impairment äs the resultingscotoma, disability äs the disorder of visual search and the ac-companying visual problems (ie, bumping into obstacles) indaily life, and handicap äs the individual problems resultingfrom the disease (ie, loss of driver's license due to the scotoma).Pathology and handicap were not considered in this study.

VISUAL PERIMETRY

Binocular visual fields were mapped with a Standard Tübin-gen projection perimeter (Oculus, Wetzlar, Germany) .32 Eyeposition was monitored through a telescope. For kinetic pe-rimetry, a large (diameter, 116 minutes of arc of visual angle)

white circular target (brightness, 102 candelas/m2) was cho-sen to determine the extent of the scotoma (for further de-tails of perimetric measurement, see Kerkhoff et al11). Peri-metric measurement accuracy was 0.45° within the central15° of the visual field and 2.5° for the periphery, äs determinedin our previous study.11

VISUAL SEARCH FIELD

Visual search field was measured with the Tübingen pe-rimeter (for details, see Kerkhoff et al11). The search fieldwas defined äs the area in the perimeter (in degrees) that apatient could actively scan via eye movements but withouthead movements when searching for a bright, suprathresh-old Stimulus. Insufficient compensation for a VFD will bereflected in time-consuming and small-amplitude sac-cades in the scotoma20 and hence a small visual search field(eg, the dashed line in Figure l , top). Alternatively, goodoculomotor compensation for the VFD will result in a largersearch field (eg, the dashed and dotted line in Figure l, top).

The patient was required to fixate a small red spot in thecenter of the perimeter (diameter of the spot: 30 minutes ofarc) of visual angle while the perimetrist moved the target (iden-tical to that used during perimetry) with a velocity of approxi-mately 2°/s from the periphery to the center of the sphere. Thepatient was instructed to search for the target by actively ex-ploring the perimeter sphere using eye, but not head, move-ments. The patient rested his or her head on a chin rest, whilebis or her forehead leaned against a supporting headband. Thesearch field was measured 10 times along eight meridians; thesequence of the meridians was pseudorandom. For statisti-cal analyses, we used the median of all search field values mea-sured on the horizontal meridian located in the patient's blindfield. Measurement accuracy of visual search field testing was2.25° of visual angle (see Kerkhoff et al11 for details). Searchfield extension is 46° for each hemifield in healthy controlsubjects measured under exactly the same conditions.33

SEARCH ON PROJECTED SLIDES

We used three slides with geometric Symbols similar to thepaper and pencil tests developed by Weintraub and Mesu-lam34 and nearly identical to the visual search test of Chedruet al.18 The patient sät 86 cm away from a white wall ontowhich slides were rear-projected that subtended 40° hori-zontally and 25° vertically. The patient had to indicate witha hand-held pointer all Symbols of a specific class (circles onslide l, triangles on slide 2, and squares on slide 3). The ratioof targets to distractors was held at approximately 50%. Omis-sions and multiple pointing to a target were rated äs errors.

TABLE TEST

The table test was designed to measure transfer of treatmentstrategies on nontrained visually related ADL. The patient was

Continued on next page

ARCH NEUROL7VOL 51, MAY 1994475

seated in front of an 80X60-cm table containing 40 real ob-jects randomly distributed on it (eg, pencil, eraser, coin, comb;Figure 2, top). The patient was shown an object and was askedto find the same object on the table äs quickly äs possible us-ing eye and head movements but without changing his or herseatingposition. Search times were measured with a stop watchby the experimenter from the moment when the target objectwas shown to the patient. The table was invisibly divided intofour quadrants of equal size. Within each quadrant, five ob-jects had to be searched for while the other five served äs tar-gets when the test was administered a second time (Figure 2,top). The sequence of the objects that the patient had to findwas pseudorandom and thus unpredictable for the patient. Forrepeated measurements throughout the study, the table wasrotated by 180° so that four different test versions were avail-able (two kinds of targets, two orientations). Twenty healthycontrols showed nearly identical search times for the samequadrant in each of the four versions. Search times of the dif-ferent versions were significantly correlated (rbetween .93 and.98 for all four quadrants); the mean difference between anyquadrant of the different versions was small (0.95 second; ränge,0.1 to 2.0 seconds). No significant retest effect was seen.

RATING OF SUBJECTIVE VISUAL IMPAIRMENT

A questionnaire based on 11 items describing the most fre-quent visual impairments of patients with VFDs (eg, bump-ing into persons in the blind hemifield) was read to eachpatient before and after therapy. For each item, patients hadto judge on a five-point scale to what extent they experi-enced the problem in question. The scale was äs follows: 0,no problem; l, rare problem; 2, partially relevant problem;3, frequent problem; and 4, very frequent problem. The 11items referred to in the "Results" section are shown inFigure 3. To minimize the tendency of the patients to an-swer in a socially desirable manner after therapy, they werenot informed about their ratings at the beginning of therapy.

TRAINING PROCEDURES

Training procedures have been described in detail else-where.n The training comprised three major Steps: (l) per-forming large saccades to the blind field, (2) improving vi-sual search on projected slides, and (3) transfer of both toADL. During the first part of training, the targets were pre-sented on a large Computer screen. The patient's task wasto direct a large saccadic eye movement to the target shownin the blind field without head movements. Target onsetwas signaled acoustically. Target eccentricity, duration, pre-dictability, and position within the patient's scotoma couldbe manipulated with the Software by the therapist. Im-provement in saccadic training was checked regularly bymeasurements of the visual search field (see "Results").

Because patients with VFDs, like those with neglect,34

show disordered visual searching strategies in blind and in-tact field regions,18'19 the patients were then taught a sys-tematic search strategy enabling them to scan the blind and

the intact visual fields without omission of relevant targets. Asystematic scanning strategy (either horizontally row by rowor vertically column by column) was demonstrated to the pa-tient repeatedly, and he or she was encouraged to adopt oneof these strategies when searching for targets on slides. In ad-dition, all patients were urged always to begin their search inthe periphery of their scotoma because this was the most likelyarea in daily living in which they would fail to notice personsor objects. Approximately 200 different slides were designedspecifically for this stage of training.35 The slides varied accord-ing to the overall number and size of Stimuli, the target-foilratio, the similarity among Stimuli, and systematic vs unsys-tematic spatial arrangement of Stimuli (rows and columns vsmixed). Improvement in visual search was measured with thetest "search on projected slides." None of the slides used in thesearch on projected slides task were used for training.

In the last part of the training, the patient was urgedto use the scanning strategies in situations of everyday life.Application of the basic scanning strategies was trained di-rectly in those situations that were rated äs being most prob-lematic in the patient's individual Situation (eg, crossing astreet). Possible improvements following this treatmentstepwere evaluated by the table test and the subjective ratingsof visual disabilities before and after therapy.

Treatment sessions took place at least once daily for30 minutes 5 days each week. Ten of the 22 patients re-ceived therapy in other areas, mostly memory training. Vi-sual training lasted from 4 to 12 weeks, depending on theindividual progress. Detailed therapy cutoff criteria werepreviously defined (see Kerkhoff et al11).

TREATMENT DESIGN

All patients were treated in a baseline design to separate train-ing-related improvements from possible effects of sponta-neous recovery, adaptation to tests, and effects of medica-tions or other therapies. Visual search field extension wasdefined äs the relevant target variable because it could be re-liably and repeatedly measured in all patients with VFDs andhas been shown to be a sensitive measure for the evaluationof the outcome of saccadic training in patients with VFDs.11

In addition, binocular visual fields were repeatedly mea-sured before, during, and after training for the same reason.

Measurements were performed weekly during therapyand three times before and after training (interval betweeneach of the three measurements, at least l week) to estab-lish a baseline period. Similarly, the table test was per-formed twice before training, directly after training, and onceat follow-up to obtain baseline data. The search-on-slidetest was administered directly before and after initiation oftherapy and at follow-up (mean follow-up period, 3 months;ränge, l to 12 months). The subjective rating of disabilitywas performed directly before and after training.

Nonparametric tests were performed with an a level of.05 (two-sided). Adjustment of the a level in case of multiplepairwise comparisons was performed according to the Bon-ferroni procedure (see Holm36).

ARCH NEUROL7VOL 51, MAY 1994476

sual field recovery11'14 suggest that some spontaneous recov-ery occurs in approximately 10% to 20% of patients and oc-curs within 2 to 3 months after brain damage in most cases.The extent to which visual fields can be recovered by system-atic training, however, is controversial.12"16 At present, themajority of patients with VFDs probably have to face long-term visual problems associated with their VFDs14 becauserecovery of visual fields does not occur spontaneously or re-mains very limited even after extensive training.11>13>I5 ThisSituation has far-reachingconsequences for the ability of diesepatients to adapt to vocational and private life. Poppelreuter17

and later researchers, using detailed eye movement record-ing techniques,18"20 have generally confirmed the view thatpatients with VFDs show disordered and slowed visual searchin their blind hemifield. Consequently, they bump into ob-stacles, other persons, and door frames and collide with bi-cyclesor cars.1'21

Although a few studies have dealt directly with therestoration of visual field regions in patients with VFDswith rather limited success,2-12-16 the question äs to whetherand how such patients can be successfully rehabilitatedusing compensatory mechanisms is completely open. It isnot known whether systematic training in patients withVFDs leads to a transfer of training effects to visually re-lated ADL, which is one of the major aims of cognitive re-habilitation.22

In a recent study,1 * we were able to show that saccadiccompensation training leads to a significant improvementin visual search in patients with homonymous VFDs and pa-tients with additional left-sided visual neglect. However, be-cause that study employed a design whereby visual searchwas measured before and after training and during a follow-upperiod, treatment-related gains in visual performance couldhave been influenced by factors not related to treatment. Fur-thermore, the transfer of these improvements to visually re-lated ADL was not addressed in that study. Finally, the patients'subjective experience of their visual impairments was not as-sessed throughout therapy. Because the transfer of treatment-related gains is one of the major aims of cognitive rehabili-tation22 and is unlikely to occur on its own, it is necessary todemonstrate that the therapy in question leads to measur-able improvements in relevant outcome measures. However,most of the available outcome scales score visual performanceon a very coarse level and are too nonspecific to measure thera-peutic changes.23 Therefore, new, sensitive measures had tobe designed to evaluate the transfer of treatment gains.

The present study was planned to evaluate the fol-lowing questions in the rehabilitation of patients withVFDs: (1) What is the relative contribution of restitu-tion of function (field recovery) vs acquisition of com-pensatory strategies (systematic scanning within the sco-toma)? (2) Does systematic treatment of patients withVFDs and their visual problems lead to a transfer of train-ing-related improvements in visual ADL? Do improve-ments remain stable during long-term follow-up after ces-sation of training? (3) Do treated patients experience

Baseline Treatment Follow-up

45-

40-

35-

30-

25-

20-

15-

10-

4 5 6 7

No. of Sessions

10 11

Figure 1. Top, Demonstration of a typical visual search field border in apatient with a left-sided homonymous hemianopia before training (dashedline) and after training (dashed and dotted line). Bottom, Visual searchfield size in the scotoma (in degrees of visual angle) at baseline andduring treatment and follow-up in 22 patients with unilateral visual fielddefects. Vertical bars indicate SEs.

subjectively the observed improvements in visual per-formance after therapy?

RESULTS

RESTITUTION OF VISUAL FIELD

Only visual field increases surpassing measurement errorwere considered valid. Partial restitution of visual field dur-ing systematic visual treatment was observed in 12 (54%)of 22 patients (Table l). Friedman analyses of ranks fordependent data37 showed a significant difference betweenthe measurement dates(x2=22.18, P=.0001). Pairedcom-parisons with Wilcoxon tests revealed a significant visualfield increase only during training from the second pre-test to posttest measurement (z=3.29, d/=2, P=.001). Noother comparisons reached statistical significance (small-est P=.20). Actual visual field increase was less then 8° innine of the 12 patients with enlarged visual fields after train-ing. However, the three remaining patients showed con-siderable increases of 11°, 17°, and 24° along the measuredmeridian.

ARCH NEUR017VOL 51, MAY 1994477

OA

oo

o

o

A

A

o!A l

A!o i

oA

A

O

A

Oco

A

0

oo

)A

A0Ao

CutoftforHealthyControls

UpperScotoma

Lower UpperScotoma Intact

Table Test

LowerIntact

Figure 2. Top, Outline of the table test. P indicates patient; dotted lines,the quadrants of the test, which are invisible to the patient; triangles,targets A; and circles, targets B. Bottom, Mean search times for the tabletest at two baseline measurements before treatment (pretreatment 1 andpretreatment 2), after treatment, and at follow-up in 22 patients with visualfield defects. Dashed line indicates mean+2.5 SDs of performance of 20healthy controls in this task; vertical bars, SEs.

VISUAL SEARCH FIELD

Figure l shows the results of the visual search field measure-ments during the baseline, treatment, and follow-up phasesfor all patients. Friedman tests revealed no significant differ-ence in search field size in the scotomatous field region be-tween the three baseline measurement dates (x2=3.025, d/=2,P=. 2204). As can be seen in Figure l, search field values rangedbetween 10° and 20° before training and were thus severelyreduced compared with normal values (46°). During treat-ment, a significant improvement was seen in visual searchfield size (Friedman test: x2=52.18, d/=4, P=.0001). To re-duce the number of statistical comparisons, only the last base-line measurement and the last measurement during train-ing were compared (Wilcoxon Test). A significant improve-ment in search field size was seen between these twomeasurements (e=3.91, P=.0001). After cessation of visualtraining, while other therapies (eg, memory training) still con-tinued in 10 of the 22 patients, no significant change was seencompared with the last treatment measurement (^=.24, P=.40).During follow-up, the search field values remained stable,ie, no further improvement or any significant decrease in vi-sual search field size was seen (Friedman test: x2= l -02, df=2,P=.59). In addition, the patients with VFDs scored signifi-

IVc

Notice Obstacles

Bump Into Obstacles

Using Public Traffic

Finding Way in Clinic

Find Therapist's Boom

Find Way at Home

Crossing Street

linimum Maximum) 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4 0

^^m

i• 1

•M

•, ... l

•^m ', i•

... i•• . |

•

Figure 3. Mean ratings of subjective disability/visual prob lern s resultingfrom the scotoma betöre (shaded bars) and after (black bars) visualtraining in 22 patients with visual field defects. See "Patients andMethods" section for an explanation of the scale.

cantly better on any follow-up measurement compared withany of the baseline measurements (all P<.001). In summary,all 22 patients showed a significant, training-related improve-ment in search field but no improvements in this task dur-ing baseline or follow-up. Visual search field gain was notcorrelated significantly to visual field restitution (Pearson cor-relation coefficient: r=.3353 [ not significant]). Therefore, im-provements in visual search field size cannot be explainedby enlargement of scotomatous field regions.

SEARCH ON PRQJECTED SLIDES

A comparison of all three measurements revealed that asignificant change in the number of errors (Table 2) wasseen in the hemianopic (Friedman test: x2=19.52, P=.0001)and intact (Friedman test: x2=14.70, P=.001) hemi-fields. Paired comparisons showed a significant reduc-tion of errors from pretreatment to posttreatment mea-surements in the hemianopic and in the intact hemifields(2=3.20 and P=.001 for both comparisons), while nochange occurred after cessation of treatment and fol-low-up measurement (P>.05 for both hemifields).

A similar reduction in overall search time on the threeslides was seen (Friedman test: x2=21.94,P=.0001). Pairedcomparisons showed a significant reduction from pre-treatment to posttreatment measurements (^=3.01,P=.001) but no further significant change compared withfollow-up (^=1.16, P>.05). In summary, the patients per-formed more accurately and quickly in visual search onprojected slides after training than before, although theywere still impaired in relation to the healthy controls.

TABLE TEST

No significant improvement in search time was seen in anyof the four quadrants of the table test between the first andsecond pretests (Figure 2, smallest P=.25). Friedman tests

ARCH NEUR017VOL 51, MAY 1994478

Table 1. Mean Visual Field Sparing B e t ö r e and AfterVisual Training and at Fol low-up and Mean Visual FieldIncrease in 22 Patients With Visual Field Defects

Condition Mean (Range), Degrees

Pretest

12

After therapy

At follow-up

Mean increase

Mean increase of patients withpartial field recovery (n=12)

4.7(1-13)4.9(1-13)7.9(1-30)8.0(1-30)3.0 (0-24)

6.6 (2-24)

revealed a significant difference between pretreatment, post-treatment, and follow-up measurements for the upper sco-tomatous quadrant (x2=26.46, d/=3, P< .0001), for the lowerscotomatous quadrant (x2=19.76, d/=3, P<.0001), and forthe intact upper quadrant (\2= 10.22, d/=3, P< .006) but notfor the intact lower quadrant (x2=2.22, d/=3, P>.30). Thelast result was due mainly to large variances. N o significantchange was seen between posttreatment and follow-up mea-surements in any of the quadrants (smallest P=.30). As canbeseen in Figure 2, bottom, the 22 patients with VFDsshoweda nearly 50% reduction in search time after training but nochange after cessation of treatment and follow-up.

RATING OF SUBJECTIVE DISABILITY

Figure 3 shows the ratings of the mean subjective dis-ability before and after visual training. In all items, sig-nificant improvements during therapy were reported bythe patients (Wilcoxon tests, largest P=.001).

TREATMENT VARIABLES

Patients with left-sided homonymous VFDs received 27.3(ränge, 18to40) treatment sessions, while those with right-sided homonymous VFDs received 25.5 (ränge, 10 to 65)sessions. This difference was not statistically significant (17=54,P>.05). Time after onset, side of field defect, age, and sexwere not correlated to recovery of performance in any ofthe visual tests (P>.30 in all correlations).

LEFT VS RIGHT HEMIANOPIA

Patients with left-sided and right-sided homonymous VFDsdiffered in visual field sparing before training (U test: 17=21,P=.025, corrected for ties) and after training (17=21, P=.025,corrected for ties); however, no differences were found be-tween the two groups regarding the size of the visual searchfield before, during, or after training (smallest P=.09) or searchtimes in the table test and during search on projected slides(smallest P=.26). Furthermore, no difference was seen insubjective ratings (smallest P=. 20), except that right-sided

Table 2. Results of the 'Search on Projected Südes'Test in 10 Controls and 22 Patients With VisualField Defects Before and After Treatment andat Follow-up After Treatment

Condition Mean (Range)

Errors, blind field

Before treatment

After treatment

Follow-up

Errors, intact field

Before treatment

After treatment

Follow-up

Overall search time, s

Before treatment

After treatment

Follow-up

Controls, overall search time

Controls, errors on left/right

4.6 (0-27)0.8(0-11)0.7 (0-10)

2.7 (0-27)

0.7(0-12)

0.6 (0-10)

96.2 (50-130)

75.8 (44-90)

74.4(41-87)

55.4 (35-60)

0.5 (0-2)/0.4 (0-2)

hemianopics reported bumping into obstacles on their rightside and left-sided hemianopics on their left side. Further-more, the groups did not differ in visual field gain duringtraining (17=49.5, P=.81),nordidtheydiffer in visual searchfield gain throughout treatment (U=24, P= l .0). In summary,left-sided and right-sided hemianopics performed entirelysimilarly, except for the difference in field sparing, and therewas no evidence that left-sided hemianopics performed likepatients with left-sided visual neglect.

LOCATION OF LESION

The location of lesions was rated according to the radiologicfindings äs predominantly occipital (n= 12), predominantlytemporal (n=7), or occipitotemporal (n=3). Neither visualfield gain throughout treatment (Kruskal-Wallis test: x2= l -02,P=.59) nor visual search field gain (x2=5.01,P=.08) differedamong these three groups. The same was found for thereduction in search time in the table test (x2=.53, P=.77),for the search on projected slides (x2=3.11, P=.21), and re-garding the number of treatment sessions (x2=2.91, P=.23).Hence, there was no evident relationship between locationof lesion and performance during training. However, it shouldbe mentioned that lesion size was not taken into accountdue to the varying optical quality of the computed tomo-graphic and magnetic resonance imaging scans.

EARLY VS LATE TRAINING

To test whether early vs late training after brain damage hadany differential outcome, we performed a median split ofthe total sample (median of 6 months for time after onset)and compared improvement on all tests, except for the sub-jective ratings. No significant differences were found be-tween the two groups in any of the comparisons (smallest

ARCH NEURO17VOL 51, MAY 1994479

P=.20). Inaddition,impairment in visual performance wasnot significantly correlated to time after onset, sex, or cause(Pearson correlation: largest r=.20 [notsignificant]).

RETURN TO WORK

Twenty of the 22 patients with unilateral VFDs returnedto work after their rehabilitation. All patients returnedto their previous Job, but 15 worked only part-time.

COMMENT

The efficacy of cognitive rehabilitation in stroke victimsand those surviving traumatic brain injury has been re-peatedly debated in recent years because of its growingprevalence and thus the increasing bürden that care andrehabil i tat ion of these patients impose on soci-ety 22,31,38-42 jo our knowiedgei our study is the first tocombine stringent baseline methods with a group de-sign in a larger sample of patients with VFDs than typi-cally reported in most single case studies.43

Before discussing our results, we will consider rel-evant methodologic issues. First, a randomized controlgroup design has been advocated in most critiques of thepresent state of the art in cognitive rehabilitation.38 Whilethese designs may have some advantages, especially whenthe patients under study remain in the clinic for only alimited time so that baseline designs are difficult to imple-ment, they have two clear disadvantages: (1) they re-quire large, homogeneous patient groups and (2) theynormally require at least two equally promising treat-ment approaches because allocation to a nontreatmentgroup or nonoptimal treatment group is ethically com-plicated. We chose the combination of a baseline designwithin a larger patient group because our patients re-mained in the clinic sufficiently long to allow for base-line and follow-up measurements, and a second well-established treatment technique, such äs visual search fieldtraining,11 was not available. In addition, strictly parallelsubgroups are extremely difficult to establish.

Another possible explanation for the observed improve-ments could be simply a placebo effect. This argument can-not be discarded easily, äs we had no untreated control groupfor ethical reasons. However, there is pertinent evidence fromour previous work.11 We measured two untreated visual co-variables throughout the study in some patients to monitorthe specificity of improvements after search field training.We found virtually no improvement in performance in fivepatients with disturbed light or dark adaptation through-out search field training. Moreover, two patients with severelydisturbed visual object identification showed no improve-ment in this ability, while they improved dramatically in vi-sual search field size during training. In summary, the speci-ficity of improvements in visual search while other untreatedvisual deficits remained unchanged throughout the courseof the study strongly argues against the nonspecific placebo

effect's being responsible for the observed improvements.As an aside, it is difficult to explain why the improvementsin visual search field clearly coincide with treatment onset.

With these considerations and caveats in mind, our re-sults show that measurement variability, adaptation to testprocedures, or spontaneous recovery cannot explain the factthat the gains in any of the measures clearly coincided withtreatment onset and ranged far beyond any test-retest effect,äs seen from the baseline measurements and the results fromthe healthy controls. Placebo effects cannot be ruled out en-tirely but seem unlikely, given the specificity of improve-ments. Furthermore, most of our patients with a mean timeafter onset of 7.5 months clearly had more chronic disease,which renders spontanous recovery even more unlikely. Sta-bility of improvements could be demonstrated in follow-upexaminations l to 10 months later in all tests.

No difference was found between patients receiv-ing "early" vs "late" visual training after their brain dam-age, äs has been shown for aphasia rehabilitation.7 Thisimplies that visual rehabilitation in our patient group im-proves compensatory saccadic eye movements even inpatients with a time after onset of more than l year.

Left-sided vs right-sided hemianopics did not differregarding their improvements, although the latter groupshowed additional visual field sparing. This may indi-cate that disorders of visual search are equally distrib-uted in left-sided and right-sided hemianopic patientswhen there is no evidence of additional neglect. Further-more, it means that both left-sided and right-sided hemi-anopics are good candidates for training of visual search.

Lesion location did not have a differential effect on train-ing success. However, patients with neglect were excludedso that no patients in our sample had parietal lesions.

With regard to the two aims of our study, we foundsignificant training-related improvements of saccadic eyemovement strategies in all trained patients, which may in-dicate that training of compensatory techniques to cope withthe disability in daily life seems to be a very promising formof neurovisual rehabilitation. In contrast, recovery of sco-tomatous field regions during therapy was achieved in 12of 22 patients but was quantitatively rather limited com-pared with the large improvements in visual search field.Morover, a priori prediction concerning in which patientsand to what extent visual fields would recover would havebeen impossible from our data. This problem could pos-sibly be overcome by using positron emission tomographyto indicate in which patients systematic training could re-store damaged field regions (see Bosley et al44).

Because transfer of treatment gains to daily life isnot likely to happen spontaneously, we incorporated thisäs a last Step in our treatment program and designed newmeasures to evaluate possible improvements. The re-sults showed significant transfer effects of training in thetable test, indicating that the training of saccadic com-pensatory eye movement strategies ameliorates the rel-evant visual problems in daily life resulting from the sco-

ARCH NEUR01WOL 51, MAY 1994480

toma, such äs finding objects on a table. The results ofthe table lest show another interesting effect: head move-ments are not suitable to compensate for the deficits invisual search in patients with VFDs, äs can be seen fromthe deficient performance in the table test during the base-line measurements despite free head movements.

The subjective improvements noted by the patients arequite impressive even if one attributes part of the improve-ment to the patient's tendency to give socially desirable an-swers to the therapist. Alternatively, the reduced scores onthe questionnaire after training might indicate reducedawareness of the visual problems in our patients after train-ing. This is extremely unlikely because information aboutdaily visual problems resulting from the scotoma and useof compensatory scanning strategies was part of each train-ing session. Moreover, successful return to work (in 20 of22 cases) would not have occurred in patients who were un-aware of their visual problems. Our study demonstrates thatsystematic training of compensatory strategies in patientswith VFDs without neglect leads to significant improvementsin basic oculomotor functions. The observed transfer to vi-sually related ADL, the patients' subjective improvementsin their visual impairments, and the return to work in 91%of patients demonstrate clearly the efficacy of the presenttreatment approach in visual neurorehabilitation.

Accepted for publication May 17, 1993.We are grateful to Elisabeth Stögerer, Elisabeth Haaf,

Gisela Eberle-Strauss, and Gabriele Ramgraberfor treatments;Charles Heywood, PhD, for improving the style ofthe manu-script; and two anonymous reviewers of the manuscript.

Reprint requests to EKN Entwicklungsgruppe KlinischeNeuropsychologie, City Hospital Bogenhausen, DachauerStr 164, D-80992 Munich, Germany (Dr Kerkhoff).

REFERENCES

1. Zihl J. Cerebral disturbances of elementary visual functions. In: Brown JW, ed.Neuropsychology of Visual Perception. Hillsdale, NJ: Lawrence Erlbaum As-sociates; 1988.

2. Rossi PW, Kheyfets S, Beding MJ. Fresnel prisms improve visual perceptionin stroke patients with homonymous hemianopia or unilateral visual neglect.Neurology. 1990:4:1597-1599.

3. Reding MJ, Potes E. Rehabilitation outcome following initial unilateral hemi-spheric stroke: life table analysis approach. Stroke. 1988:19:1354-1358.

4. Groswasser Z, Cohen M, Blankstein E. Polytrauma associated with traumaticbrain injury. Brain Inj. 1990:4:161-166.

5. Savir H, Michelson l, David C, Mendelson L, Najenson T. Homonymous hemianop-siaand rehabilitation in fifteen cases of CCI. Scand J RehabilMed. 1977:9:151-153.

6. Uzzell BP, Dolinskas CA, Langfitt TW. Visual field defects in relation to headinjury severity: a neuropsychological study. Arch Neurol. 1988:45:420-424.

7. Wertz RT, Weiss DB, Aten JL, et al. Comparison of clinical, home and deferredlanguage treatment for aphasia. Arch Neurol. 1986:43:653-658.

8. Weinberg J, Diller L, Gordon WA, et al. Visual scanning training effect on reading-related tasks in acquired right brain damage. Arch Phys Med Rehabil. 1977;58:479-486.

9. Robertson l, Gray J, McKenzie S. Microcomputer-based cognitive rehabilita-tion of visual neglect. Brain Inj. 1988:2:151-163.

10. Robertson l, Gray J, Pentland B, Waite LJ. A randomized controlled trial ofcomputer-based cognitive rehabilitation for unilateral left visual neglect. ArchPhys Med Rehabil. 1990:87:236-242.

11. Kerkhoff G, Münßinger U, Haaf E, Eberle-Strauss G, Stögerer E. Rehabilitation ofhomonymous scotomata in patients with postgeniculate damage of the visual sys-tem: saccadiccompensation training. Restorative Neurol Neurosci. 1992:4:245-254.

12. Zihl J, von Cramon D. Restitution of visual function in patients with cerebralblindness. J Neurol Neurosurg Psychiatry. 1979:42:312-322.

13. Zihl J, von Cramon D. Visual field recovery from scotoma in patients with post-geniculate damage. Brain. 1985;108:439-469.

14. Zihl J, von Cramon D. Recovery of visual fields in patients with postgeniculatedamage. In: Poeck K, Freund HJ, Gänshirt H, eds. Neurology. New York, NY:Springer-Verlag NY Ine; 1986:188-194.

15. Potthoff RD. Partial visual field rehabilitation in two various cases of homony-mous hemianopia. Eur J Neurosci Suppl. 1988:73:10. Abstract.

16. Balliet R, Blood KMT, Bach-y-Rita P. Visual field rehabilitation in the corticallyblind? J Neurol Neurosurg Psychiatry. 1985:48:1113-1124.

17. Poppelreuter W. Die psychischen Schädigungen durch Kopfschuß im Kriege1914/1916. Vol l: Die Störungen der niederen und höheren Sehleistungen durchVerletzungen des Okzipitalhirns. Leipzig, Germany: Voss; 1917.

18. Ch6dru F, Leblanc M, Lhermitte F. Visual searching in normal and brain-damaged subjects. Cortex. 1973:9:94-111.

19. Ishiai S, Furukawa T, Tsukagoshi H. Eye-fixation patterns in homonymous hemi-anopia and unilateral spatial neglect. Neuropsychologia. 1987:25:675-679.

20. Meienberg 0, Zangemeister WH, Rosenberg M, Hoyt WF, Stark L. Saccadic eyemovement strategies in patients with homonymous hemianopia. Ann Neurol.1981:9:537-544.

21. Williams D, Gassei MM. Visual function in patients with homonymous hemi-anopia, l: the visual fields. Brain. 1962:85:175-250.

22. Levin HS. Cognitive rehabilitation. Arch Neurol. 1990:47:223-224.23. Mai N. Evaluation in constructing neuropsychological treatments. In: von Stein-

büchel N, von Cramon DY, Pöppel E, eds. Neuropsychological Rehabilitation.New York, NY: Springer-Verlag NY Ine; 1992:96-99.

24. von Cramon D, Zihl J. Neuropsychologische Rehabilitation. New York, NY: Springer-Verlag NY Ine; 1988.

25. Leigh RJ, Zee DS. The Neurology of Eye Movements. San Francisco, Calif:Davies; 1983.

26. Heilman KM, Watson RT, Valenstein E. Neglect. In: Fredericks JAM, ed. Hand-book of Clinical Neurology. Vol 1: Clinical Neuropsychology. Amsterdam, theNetherlands: Eisevier Science Publishers; 1985.

27. Axenfeld D. Eine einfache Methode: Hemianopsie zu constatieren. Neurol Zen-tralbl. 1894:13:437-438.

28. Liepman H, Kalmus E. Über eine Augenmasstörung bei Hemianopikern. BerlKlin Wochenschr. 1900;3:838-842.

29. Kerkhoff G. Displacement of the egocentric visual midline in patients with al-titudinal postchiasmatic scotomata. Neuropsychologia. 1993:31:261-265.

30. Schenkenberg T, Bradford TC, Ajax ET. Line bisection and unilateral visual ne-glect in patients with neurologic impairment. Neurology. 1980;3:509-517.

31. Wade DT. Neurological rehabilitation. In: von Steinbüchel N, von Cramon DY,Pöppel E, eds. Neuropsychological Rehabilitation. New York, NY: Springer-Verlag NY Ine; 1992.

32. Aulhorn E, Harms J. Visual perimetry. In: Jameson D, Hurvich L, eds. Handbook ofSensory Physiology. New York, NY: Springer-Verlag NY Ine; 1972;7:102-145.

33. Zihl J, von Cramon D. Zerebrale Sehstörungen. Stuttgart, Germany: Kohlham-mer Verlag; 1986.

34. Weintraub S, Mesulam MM. Visual hemispatial inattention: Stimulus Parametersand exploratory strategies. J Neurol Neurosurg Psychiatry. 1988:5:1481-1488.

35. Münßinger U, Kerkhoff G. Therapy for disturbed visual search in patients withpostchiasmatic visual field disorders and visual neglect. In: EKN Materialien fürdie Rehabilitation. Dortmund, Germany: Borgmann Publishing; 1994.

36. Holm S. A simple sequentially rejective multiple test procedure. Scand J Stat.1979:6:65-70.

37. Siegel S. Nichtparametrische Statistische Methoden. Frankfurt am Main, Ger-many: Verlagsbuchhandlung, Verlag für Psychologie; 1976.

38. Reding MJ, McDowell FH. Focused stroke rehabilitation programs improve out-come. Arch Neurol. 1989:46:700-701.

39. Dobkin BH. Focused stroke rehabilitation programs do not improve outcome.Arch Neurol. 1989;46:701-703.

40. Hachinski V. Stroke rehabilitation. Arch Neurol. 1989:46:703.41. Berrol S. Issues in cognitive rehabilitation. Arch Neurol. 1990:47:219-222.42. Volpe BT, McDowell FH. The efficacy of cognitive rehabilitation in patients with

traumatic brain injury. Arch Neurol. 1990;47:220-222.43. Gianutsos R, Matheson P. The rehabilitation of visual perceptual disorders at-

tributable to brain injury. In: Meier MJ, Benton AL, Diller L, eds. Neuropsy-chological Rehabilitation. New York, NY: Churchill Livingstone Ine; 1987.

44. Bosley TM, Rosenquist AC, Kushner M, et al. Ischemic lesions of the occipital cortexand optic radiations: positron emission tomography. Neurology. 1985;35:470-484.

ARCH NEUR017VOL 51, MAY 1994481