A C T A O P H T H A L M O L O G I C A VOL. 5 7 1 9 7 9

Drpnrtmrnt of Experimrntal Ophthalmology, Univwyity Eyr Clinic (Hrad: C. E . T . Kmknu) nnd thr Dnlh? Communzb Rrsecirc-h Crntrp (HPncl: Akr Nor&), Lund. S7uedrn

AUTOMATIC PERIMETRY IN A POPULATION SURVEY BY

BO BENGTSSON and C. E. T. KRAKAU

Automatic perimetry was performed in 2998 eyes of 1511 subjects comprising 78% of all persons born 1907-1921 and resident in a certain small area. Unreliable or abnormal tests were repeated. T h e average number of tests per person was 2.25. About 90% of all tests in normal eyes were performed in less than 3 min. T h e screening was considered negative in 2887 eves, incomplete in 20 eyes a,id positive in 91 eyes (3%).Eighteen of the positives were previously unknown glaucomatous defects. There was only one unexplained “false positive”. We concluded that the method is quick, sensitive, specific and dependable. T h e apparatus is simple to manage and cheap to run. Kry 7oord~:defined population glaucoma.

-

virual field defects

-

automatic perimetry

-

During the last few years an automatic perimeter has been developed and tested at the Department of Experimental Ophthalmology in Lund (He$ & Krakau 1975a,b; Krakau 1978). A great deal of clinical experience was gained i.a. the instrument was used on a material of glaucomas and glaucoma suspects (Heijl 1976). A trial in a population survey of the logic for detection of glaucomatous visual field defects was considered appropriate. A preliminary report comprising about one third of the present material was given at the Nara Symposium (Bengtsson & Krakau 1979). Received March 5 , 1979.

929 4cta ophthal.

57. 5

i 9

B . B m g h w n ant1 C. E. 1 . K m k m i

Material All persons born 1907- 192 1 and resident in the district provided for by the Dalby Community Care Centre were listed in December 1976. The list was arranged according to the residential addresses and kept up to date by means of weekly reports from the County Council on removals and deaths. Following this directory the inhabitants were contacted in rotation during 1977 and 1978 and offered repeated ophthalmological examinations. Attempts at persuasion were avoided and only persons able and willing to attend within a few weeks were included in the survey. One person known to be blind and 24 patients subjected to antiglaucomatous therapy were not invited. There were only five cases with field defects (in one congenital and four secondary glaucomas) in this group. Out of 1938 invited persons, 1511 (78%) took part in the survey. The rate of attendance was largely independent of age and sex. Six persons had lost one eye, 16 were blind in one eye and 2 were unable to look straight ahead with one eye. Automatic perimetry was attempted, however, in each one of the 2998 seeing and mobile eyes of the 15 1 1 subjects in the survey.

Methods Sphygmomanometric measurement of the systemic blood pressure, determination of the visual acuity, subjective refraction, automatic perimetry, fundus photography, indirect ophthalmoscopy, slit lamp examination and Goldmann tonometry were attempted in every case. Perimetry and photography were conducted by two alternating assistants - ophthalmoscopy, slit lamp examination and tonometry by one of the authors (BB). Perimetric data handling was entirely automatic and other data were immediately recorded on special forms. Transfer to magnetic tape and further processing were performed at the Computer Centre in Lund. In the present report we have confined ourselves to the results of automatic perimetry. Pprimeter. The perimeter described by Heijl & Krakau (1975) was used. Fifty-six points are fixed in a pattern of concentric circles at 5, 10 and 15 degrees’ eccentricity. At 20 degrees there are eight points, placed as shown in Fig. 1. At each test point a LED (light emitting diode) can be made to emit light at anyone of 16 levels. T h e relation of luminance between two consecutive levels is 1: 2 . One further light spot is projected on to the perimeter screen, adjusted so as to fall in the blind spot area, and lit at random intervals on an average l / 9 of all trials. If the patient keeps

930

Azrtomotzc prrzmrtr?

0

0

*

.

0

Fig. I . 1.ocation of test points. T h e threshold is first determined at the four places pointed at by arrows. When one or two of the points ( :0: ) and, eventually, one or tw o o f the points ( a r e not seen, the blind spot is denoted normal. (The right group refers to the right eye, and vice versa.) T h e central point is a red fixation light.

0)

his fixation it cannot be seen. The patient signals that he has perceived a light by pressing a button.

Trst l q i c . The test procedure is similar to the programmes for glaucoma screening by manual perimetry as used by Armaly, Drance and others. However, the points tested are chosen in random order in the computerized version. Minor changes only have been made in the programme used by Heijl(l976). The threshold is at first determined at four points on the 10 degree circle (Fig. 1). If, at a certain point, a target is seen, next time it will be shown on a one-step fainter level and vice versa. The target perceived is accepted as the threshold when the process has changed its direction three times. A supraliminal test value is calculated from the thresholds of the first four points and used as a starting level for the remaining test points. If, at any point, this level is seen, this point is not further tested; if not, the intensity is increased by one step. If this level is seen, this point is again not tested; if not seen, the LED i s lit at its very highest intensity level. If still not seen it is not tested again. If the highest intensity level is seen the process returns to the supraliminal starting level, passes step by step to higher intensities as long as the light is not seen, and ends as soon as the light is seen again (Fig. 2). 93 1

R Rrngtt,ron arid C E T Knlkmi

4 POINTS TESTED, 3 CHANGES OF ISIGN

1

SPREAD OF

E N D LEVEL

"seen"

seen?

Ei-1 seen?

INTERPRETATION

YES

Ei-1

"seen"

"not seen',

"seen"

"seen"

Ei-n

'"notseen"

Fi,g. 2 Flow diagram, showing the course of testing one point E;: start level; E:highest intensity level.

Test conditions. As a rule, the blind spot stimuli for fixation control were directed towards a point about 2 cm temporally and inferior to the test diode 15" temporally to the fixation light. An individual adjustment was resorted to only when an anomalous position of the blind spot was suspected. Only spheric glasses were used to correct for near vision and no special steps were taken to secure dark adaptation. T h e instructions to the subjects (to fixate the red light and to press the button whenever a white light was seen) were kept as few and simple as possible. The assistant remained in the room for a while but left it as soon as she felt that cooperation was satisfactory. 932

Automatic perimetqi

Presentation of results. The results were presented by means of a “keyboard-printer” (Silent 700, TEXAS Instruments). The printout of one field chart required less than 1 min. Interpretation. A test point was considered seen if it had ended on a level not more than one step lower than the initial one (Fig. 2 ) . Up to four points not seen in the blind spot area (Fig. 1 ) were accepted as a normal blind spot. If the blind spot was missed and, in addition, the blind spot check light seen more than five times, the test was regarded as unreliable. A reliable test was considered normal if all points outside the normal blind spot were seen (otherwise abnormal). If a reliable test was normal the eye was classed as negutiue.

Fig. 3 Flow diagram used when classifying the field screening of an eye as positive or negative. The number of eyes following a certain course is given in the cartouches. The upper number refers to tests in which the blind spot was found - the lower to tests in which it was missed.

REP? = Repetition of test? TERM? = Termination of screenings? CONF? = Confirmation of defects? INC = Incomplete screenings NEG = Negative screenings POS = Positive screenings

U= N= A= B= C =

Unreliable tests Normal tests Abnormal tests Unconfirmed defects Confirmed defects

R. Rengtsson arid C. E . T. Kraknu

When a reliable, abnormal test was obtained in an eye with a known field defect or in which a defect could be attributed to an ophthalmoscopic finding, the screening was terminated and considered positzve. With this exception the screening of one eye was not considered complete unless all unreliable or abnormal tests were repeated. If the first reliable test was abnormal a proposed kfect was assumed. If such a defect was followed by a similar one in the same eye, the points outside the normal blind spot were compared. If at least one of those points was “not seen” in both tests the second test showed a confirmed defect, and the screening was considered positive. If a proposed defect was not reproduced at any point in the second test, the screening was considered negatiue. T h e final classification of screenings as “zncomplete”, “positive” or “negative” followed rules so strict that the computer was entrusted with it (Fig. 3). The operator had to initiate a retest but even this task could have been left to the computer. An ophthalmologist had to decide whether a termination of the screening was warranted (and to reclassify terminated screenings from incomplete to positive). As indicated in the preceding paragraph even the immediate interpretation of each test, ordering a retest according to the strategy adopted etc. could have been left to the computer. In the course of the survey a major part of the soft-ware necessary for a complete automation was developed for the purposes of supervising and arranging the material. For a start, however, we had to resort to a semiautomatic procedure involving a presentation of the results in a cruder form - and a human element - which was retained throughout the entire study.

Results Thp h n c k c ~ o m tllumznntzon d was in nine eyes with nuclear cataract 0.1 cd/m2, in all others 1 cd/m2. There was no difference in the threshold reached between the first (right) and the second (left) eye. Thus there was no indication of insufficient adaptation. The time needed for one test was less than 3 rnin in 90% of all normal eyes. The average number of tests per person was 2.25. Eleven unreliable and 9 abnormal tests were not repeated - by inadvertence (5 subjects) or because the subject was not motivated to continue (9 subjects) but the field screening was completed in 2978 eyes - 2887 negative and 91 (3%) positive (Fig. 3). T h e positive field screenings are summarized in Table I. Six knorcm (nonglaucomatous) field defects were all spotted by the automatic perimeter. We also soon learned to expect relative coecocentral defects or enlarged blind cpot, in highly mvopic eyes with posterior staphyloma and/or peripapillary atrophy. 934

Automntzc perimet?

Tahk I. Positive screenings No. of cases n = 1.505

No. of eyes n = 2978

From o u r point of view the actual yield of the field screening was 46 unknown nrul itrtrxpectrd tlrf&c - 18 glaucomatous and 28 “others” - described in Table 11. They were all reproduced and also verified by Goldmann perimetry, if not by ophthalmoscopy. Their extension was in general adequately mapped out by the “automatic” field chart.

Total A. Others 1. Of unknown etiology* 2. Circumscribed paracentral corresponding with various small but obvious fundus lesions 3 . Homonymous paracentral 4. Ritemporal

No. of cases

No. of eyes

15

18

2

2

18

21

2

4

1

1** ~

23

Total

* **

28

One defect had disappeared at follow up - the other turned out to be part of a homonymous one. Automatic perimetry was not attempted in the left eye, in which the visual acuity was 2/60.

935

B Rmgt\\on and C. E. T Krtrkau

T h e 18 glaucomatous defects were found in eyes with pathological excavation of the disc. All patients with glaucomas are, of course, followed up and a detailed account of the characteristics associated with glaucomatous field defects in the present material is planned. T h e 28 “others” were caused by a wide variety of disorders among which occlusions of small retinal veins ( 5 cases) and scars in retina and choroid (4 cases) were most frequent. Apart from the glaucomas only two cases with unknown defects (one chiasma compression and one retinal detachment) needed medical care. T h e field screening was considered to befulwporrtzue in 14 cases (16 eyes). Four subjects could not be prevented from “overfixating”. One deaf person was an excellent lip-reader and therefore by mistake left without adequate instructions in the dark room. In six cases extraocular causes such as ptosis, blepharochalasis, smudged glasses and matted (super)cilia partly blocking the sight of one eye were overlooked and therefore not eliminated. Repeated telephone calls disturbed one individual and another suffered from migraine. To sum up: false positives could usually be given an adequate - though trivial - explanation. Only one single point had to be left unexplained and not verified by kinetic perimetry. T h e blind spot was “missed” in 12% of all reliable tests. This rate was 8% even if the fixation was good (i.e. the check light seen 5 2 times) and still less than 40% when the fixation check light was seen three to five times. T h e results in geographical subdivisions of the material were alike and the achievements reported at the beginning of the survey were well maintained - or improved - at the end. The rates of unreliable tests, false abnormal tests and missed blind spots in repeated tests were similar to those in initial tests. Unreliable tests were twice as common in men as in women. The frequencies of missed blind spots and false abnormal tests increased in the older men and decreased in the younger women. All types of true defects were more prevalent in the older half of the population.

Discussion When first recognizable, glaucomatous field defects are small, circumscribed, deep scotomas, which may appear anywhere in the central field (Aulhorn & Harms 1967). Any attempt to detect progressive defects smaller than the blind spot must obviously be unprofitable since perimetry cannot be performed incessantly. The perimeter treats the blind spot area in exactly the same way as the rest of the paracentral field. T h e fact that the probability of detecting the blind spot by automatic perimetry is as high as about 0.9 should therefore mean that the method 936

Automatic prrzmrt9

amply fulfils any reasonable demands on sensitivity. Surely the detection of 46 unknown and unexpected defects does not contradict this conclusion. The rate of verified defects not warranting medical care (2%)was greater than might be desired but, of course, a natural and inevitable consequence of the high sensitivity. Our main concern was, however, specificiq, since a high rate of false positives might easily become a major obstacle. We were therefore much relieved to find that after retesting very few unexplained false positives remained. We conclude that the procedure of automatic perimetry in the form applied is quick, sensitive, specific and dependable.

Acknowledgment This work was supported by the Swedish Medical Research Council (proj. No. B79-04X5202-02A) which is gratefully acknowledged.

References Aulhorn E. & Harms H. (1967) Early visual field defects in glaucoma. In: Gla7ccomn, Symposium Tutzing Castlr 1966, pp. 151-186. Basel-New York, Karger. Bengtsson B. 8c Krakau C. E. T. (1979) Automatic perimrtry in stconda? prrvmtion of gkiucomn. A preliminary report. (Nara Symposium 1978). In glaucoma update, Springer Verlag, Berlin. Heijl A . (1976) Automatic perimetry in glaucoma visual field screening. A clinical study. Alhrecht 11. Grnpfrs Arch. klin. exp. Ophthal. 200, 2 1-37. Heijl A. & Krakau C. E. T. (1975a) An automatic static perimeter, design and pilot study. Artn ophthal. (Kbh.)53,293-310. Heijl A. & Krakau C. E. T. (1975b) An automatic perimeter for glaucoma visual field screening and control. Construction and clinical cases. Alhrecht 7'. Graefrs Arch. klin. exp. Ophthnl. 197. 13-23. Krakau C. E. T. (1978) Aspects on the design of an automatic perimeter. Actn ophthal. (Khh.) 56,389-405.

Authorr' nddrrsscs: Bo Bengtsson, med. lic., Dalby Community Care Research Centre, Skolgatan 1,240 10 Dalbv, Sweden. C. E. T. Krakau, professor, M. D., Department of Experimental Ophthalmology, University Eye Clinic, 221 85 Lund, Sweden.

93 7

Automatic perimetry in a population survey.

A C T A O P H T H A L M O L O G I C A VOL. 5 7 1 9 7 9 Drpnrtmrnt of Experimrntal Ophthalmology, Univwyity Eyr Clinic (Hrad: C. E . T . Kmknu) nnd th...
357KB Sizes 0 Downloads 0 Views