- Email: [email protected]
Effect of Patient Experience on the Results of Automated Perimetry in Glaucoma Suspect Patients
Effect of Patient Experience on the Results of Automated Perimetry in Glaucoma Suspect Patients ELLIOT B. WERNER, MD,t THEODORE KRUPIN, MD,2 ANDREW ADELSON, MD/ MARIANNE E. FEITL, MD2
Abstract: The first four Octopus-automated visual field examinations of 29 patients with elevated intraocular pressure but apparently normal optic discs and Goldmann visual fields were studied for the presence of a learning effect on the visual field parameters of mean sensitivity, number of disturbed test locations, total field loss, and short-term fluctuation. A learning effect, if present, would manifest itself as an improvement in the visual field as patients become more experienced with the test. There was no apparent effect of patient experience on the mean sensitivity of the whole visual fields or the mean sensitivity of the test locations within 20° of fixation. There was a significant (P = 0.Q12) increase in mean sensitivity for the test locations outside 20° of fixation. There were significant (P < 0.01) improvements in short-term fluctuation, total loss, and number of disturbed points between the first and second visual field examinations. The results indicated that there was a learning effect between the first and second automated visual field in glaucoma suspect patients who had previous experience with manual perimetry. It was not, however, very large in most patients and seems to be present in the peripheral portions of the visual field only. In most cases, it was not necessary to obtain more than two "baseline" examinations unless a patient demonstrated unusually high short-term fluctuation or had visual field defects inconsistent with the remainder of their clinical examination. Ophthalmology 1990; 97:44-48
Originally received: April 20, 1989. Revision accepted: July 17,1989. 1
2
Department of Ophthalmology, Hahnemann University, Philadelphia. Glaucoma Service, Scheie Eye Institute, University of Pennsylvania School of Medicine, Philadelphia.
Portions of this material were presented at the Annual Meeting of the Association for Research in Vision and Ophthalmology, Sarasota, May 1988. Supported in part by the Charles E. Goetz teaching and research fund, Scheie Eye Institute. Reprint requests to Elliot B. Werner, MD, Hahnemann University, Department of Ophthalmology, 216 North Broad St, Mail Stop 209, Philadelphia, PA 19102.
44
The learning effect is an important artifact in many psychophysical tests. I,2 It appears as an improvement in performance as a subject gains experience with a test. Several authors have reported small learning effects in subjects undergoing perimetry,3-1O whereas others have been unable to confirm this findingy,I2 In a previous study, we showed that there is a small learning effect on short-term fluctuation in patients with established glaucomatous visual field defects who had previous experience with manual (Goldmann) perimetry.13 Automated perimetry is time-consuming, expensive, and physically demanding on the patient. If it is necessary to perform multiple "baseline" examinations to obtain reliable data because there is a significant learning effect,
WERNER et al
•
EFFECT OF EXPERIENCE IN AUTOMATED PERIMETRY
this would have a major effect on the management of patients with glaucoma. In the current study, we evaluated the effect of patient experience with automated perimetry on the visual field in a sample of glaucoma suspect (ocular hypertensive) patients who have previously had normal results of manual visual field examinations.
MATERIALS AND METHODS The charts of all patients followed in the glaucoma service of the Scheie Eye Institute were reviewed. Glaucoma suspect patients were defined as follows: an intraocular pressure over 22 mmHg on at least two consecutive examinations; a normal appearing optic disc with cup-todisc ratio less than 0.5 based on examination of the fundus photograph with a uniform, healthy-appearing neural rim and no retinal nerve fiber layer defects by direct ophthalmoscopy; and a normal visual field as measured with a Goldmann perimeter before the first automated visual field examination. A normal visual field was defined as the I2e isopter at least 20° from fixation and the I4e isopter at least 40° from fixation, no nasal step greater than 5° width in any isopter and no nasal step of any width in more than one isopter, and absence of a scotoma or localized depression of sensitivity on suprathreshold static screening of mUltiple test locations within the central 20°. All glaucoma suspect subjects who met the following criteria were included in our study: (1) a minimum of four automated visual field examinations with at least 12 months offollow-up between the first and last automated visual field examination; (2) visual acuity of 20/25 or better with no change in visual acuity greater that one Snellen line during the follow-up period; (3) a clinically stable appearance of the optic nerve head during the follow-up period; and (4) unchanged medical glaucoma therapy during the follow-up period if the patient was on treatment, or no treatment initiated during the follow-up period if the patient was not on treatment. All subjects were phakic and had no other known ocular, neurologic, or systemic disease likely to affect the visual field. All visual fields were done on the Octopus 201 perimeter (Interzeag, Bern) using Program 32 with the appropriate refractive correction, and all examinations were done by trained technicians who had extensive experience in the use of the Octopus perimeter. All visual fields were performed with an un dilated pupil. None of the subjects in the sample was using topical miotics, and pupil size varied very little in individual patients from one examination to the next. Mean sensitivity, total loss, total number of disturbed test locations, and short-term fluctuation as measured by the root mean square were determined for the first four automated visual fields of each subject. Mean sensitivity was defined as the mean of the measured decibel thresholds for all test locations in the visual field. Total loss was defined as the sum of the differences between the measured threshold and the age-corrected normal threshold for each
test location where that difference exceeded 4 dB. A dis turbed test location was defined as one with a sensitivity reduced 5 dB or more below the age-corrected normal value stored in the Octopus computer for that test location. The root mean square was defined as the square root of the sum of the variances of the test locations where the threshold was determined twice. Values for mean sensitivity for the whole field, mean sensitivity for test locations between 0° and 10° from fixation, mean sensitivity for test locations between 10° and 20° from fixation, mean sensitivity for test locations between 20° and 30° from fixation, total loss, number of disturbed test locations, and the root mean square were obtained from the Delta Program of the Octopus perimeter. In addition, total loss and number of disturbed test locations were recorded manually from the individual visual field printouts for the test locations between 0° and 20° from fixation and 20° and 30° from fixation. The sample mean of each parameter for each of the first four available visual fields on each subject was calculated. Means were compared using a one-way analysis of variance with a repeated measures design. When any of the analysis of variance tests demonstrated a significant treatment effect (P < 0.05), the outlying means were identified with the Newman-Keuls multiple-range test. To determine if patient responses become more consistent with experience, the absolute difference for' each parameter for each subject between fields one and two was determined and compared with the absolute difference between fields three and four using a paired t test.
RESULTS Twenty-nine subjects were identified who met our study criteria. In subjects where both eyes qualified for the study, one eye was selected at random. Of the 29 eyes, there were 12 right eyes and 17 left eyes. Mean age of the subjects was 61 years (range, 34-79 years). There were 13 men and 16 women. The mean interval over which the first four automated visual fields were performed was 24 months (range, 12-38 months). Table 1 shows the sample means and standard deviations for the mean sensitivities, total loss, disturbed test locations, and root mean square for the four visual fields, and the F ratios and significance levels for each parameter. Highly significant effects (P < 0.01) were found for total loss for the whole field and for the test locations between 20° and 30° from fixation, number of disturbed test locations for the whole field and for the test locations between 20° and 30° from fixation, and short-term fluctuation as measured by the root mean square. There was no effect on total loss or disturbed test locations for the test locations within the central 20° of the visual field and no significant effect on mean sensitivity for the whole field or for the test locations within 10° of fixation or 10° to 20° from fixation. A significant effect (P = 0.012) on mean
45
OPHTHALMOLOGY
•
JANUARY 1990
•
VOLUME 97
•
NUMBER 1
Table 1. Results of Repeated Octopus Visual Field Examinations in Glaucoma Suspect Patients Visual Field Examination (mean ± 1 SD) 2
Parameter Mean sensitivity whole field (dB) Mean sensitivity 0° to 10° (dB) Mean sensitivity 10° to 20° (dB) Mean sensitivity 20° to 30° (dB) Total loss whole field (dB) Total loss 0° to 20° (dB) Total loss 20° to 30° (dB) Disturbed test locations whole field Disturbed test locations 0° to 20° Disturbed test locations 20° to 30° RMS (dB) SD *
25.3 ± 28.8 ± 26.0 ± 24.1 ± 41 ± 9 ± 30 ±
26.1 ± 29.0 ± 26.8 ± 25.0 ± 10 ± 4 ± 7 ±
1.8 1.7 1.8 1.9 64 22 47
6 ± 10
4 ± 7 1.7 ± 0.5
1.5 1.5 1.5 1.5 15 8 12
25.9 ± 28.5 ± 26.6 ± 24.8 ± 14 ± 3 ± 9 ±
1.9 1.8 2.1 1.9 30 7 23
2 ± 2
3 ± 6
±
±
1 ± 2 1.3 ± 0.4
1 ± 3 1.3 ± 0.8
± 4
F ratio*
P
2.37 1.39 2.07 3.88 6.70 1.80 5.63
0.08 0.25 0.11 0.012 0.0004 0.15 0.0012
6.39
0.0006
0 ±
1.68
0.18
1 ± 1 1.3 ± 0.6
5 5. 9 4.19
0.0013 0.008
4
3 25.9 28.4 26.5 25.0 6 2 3
± ± ± ± ± ± ±
1.8 2.0 1.9 1.8 10 4 8
± 2
= standard deviation; RMS = root mean square.
F ratio from one-way analysis of variance with repeated measures. Table 2. Comparison of the Sample Means of the Absolute Differences Between the First Two Visual Field Examinations and the Third and Fourth Examinations Fl-F2* Parameter Mean sensitivity whole field (dB) Mean sensitivity 0° to 10° (dB) Mean sensitivity 10° to 20° (dB) Mean sensitivity 20° to 30° (dB) Total loss (dB) Disturbed test locations RMS (dB)
F3-F4*
(mean absolute difference ± 1 SD) 1.2 ± 0.9 ± 1.3 ± 1.3 ± 33 ± 5 ± 0.6 ±
1.3 1.4 1.4 1.3 59 9 0.5
1.0 ± 0.8 1.1 ± 1.0 1.1± 0.8 1.0 ± 0.8 9 ± 21 2 ± 4 0.6 ± 0.8
Tvaluet
P
0.53 -0.37 0.41 0.93 2.12 2.01 -0.64
0.30 0.36 0.34 0.18 0.02 0.03 0.47
SD = standard deviation; RMS = root mean square. * F1-F2 is the absolute value of the difference between the first and second visual field . F3-F4 is the absolute value of the difference between the third and fourth visual field. t Paired ttest.
sensitivity, however, was found for the most peripherally measured points 20° to 30° from fixation. For all parameters in which there was a significant effect, the sample mean of the first visual field was significantly higher than the three subsequent fields (P < 0.05). No significant difference was found among fields two, three, or four for any of the parameters. Table 2 shows the sample means and standard deviations for the absolute differences between the first two automated visual field examinations compared with the last two for each of the performance indices generated by the Delta Program software. Total loss and number of disturbed test locations were the only parameters where the subjects showed a significant improvement in consistency between examinations.
46
DISCUSSION Response variability upon repeated testing is common with all psychophysical tests. Random variability or fluctuation in perimetry is due to the nature of the visual system and the testing situation. Nonrandom variability may be due to external factors such as pupil size, media opacities, glaucoma therapy, or progression of the patient's disease. Performance effects are an additional source of nonrandom variability. Performance effects are due to the nature of the response of the subject to the testing situation apart from the physiologic function being measured. They may be either positive or negative. Positive performance
WERNER et al
•
EFFECT OF EXPERIENCE IN AUTOMATED PERIMETRY
Table 3. Frequency Distribution of the Total Loss in Decibels for the Whole Field and for the Test Locations Outside 20° from Fixation Total Loss (dB) Whole Field
o
5-20 20-50 >50 Outside 20°
o
5-20 20-50 >50
No. of Subjects
10 10
4
5
10 12 2 5
effects are associated with an improvement in the subject's performance, while negative effects are associated with a deterioration in the subject's performance. The learning effect in psychophysical testing is by definition a type of positive performance effect that results in an improvement in response upon repeated testing. When a sample of subjects undergoes repeated testing, significant improvement will be seen in their average score if there is an important learning effect. It is difficult to infer a learning effect in a single subject because it is impossible to differentiate an improvement due to a learning effect from random fluctuation. Our findings indicate that there is some learning effect on the visual field as measured by the Octopus 201 automated perimeter in glaucoma suspect patients who have had at least one prior manual visual field examination. The learning effect was most significant on short-term fluctuation, total loss, and number of disturbed test locations. The effect on total loss and disturbed test locations was entirely accounted for by the effect on the test locations located more than 20° from fixation. There are two possible situations in which the means of the parameters studied could show a systematic improvement suggestive oflearning. One is a modest learning effect in most or all the subjects and the other is a large effect in a few subjects with no or minimal effect in most of the subjects. The second explanation seems to be the case in our sample. Looking at total loss, for example, we find that only 5 of the 29 subjects had a total loss greater than 50 dB. These subjects showed a fairly marked improvement with experience, while the majority of the subjects showed very little change after the first visual field examination. Table 3 shows the frequency distribution for total loss for the whole field and for the area outside 20° for the sample. The effect on mean sensitivity was much less marked and only detectable in the test locations outside 20° from fixation. In this regard, our results are similar to those reported by Wood et al lO and Heijl et al. 14 The learning effect seemed to occur between the first and second visual field examinations in the peripheral portion of the central 30° of the visual field. This effect persisted during the
third and fourth examinations and was stable. It is clear, however, that there is a lot of variability from subject to subject. One must individualize the interpretation of visual fields in patients. Some patients will seem to show a much greater learning effect than others. The clinician should be very cautious about detecting any change in the visual field when only a few fields are available in an inexperienced patient. In addition to learning, there are other performance effects, such as motivation, fatigue, and effects of drugs. Some of these effects may have a negative impact on performance. It is possible that both negative and positive performance effects might be found in approximately equal numbers of subjects in our sample. In that case, analysis of the means would not reveal a significant change over the four visual fields. To determine whether there were other significant performance effects, we compared the absolute differences between the first two visual fields with the absolute differences between the last two visual fields. If there were larger changes over the first two fields in either a positive or negative direction, this technique would detect them. This did not seem to be the case except for total loss and number of disturbed test locations where this effect would have been expected from the analysis of variance results (Table 2). Thus, there do not appear to be any significant performance effects other than learning in our sample. Our results indicate that learning may play an important role in glaucoma suspect patients undergoing automated perimetry who have been screened previously with manual perimetry. However, it is probably not necessary to obtain more than two "baseline" examinations in the majority of such patients because the learning effect seems to occur between the first and second examinations. Two examinations, therefore, should be sufficient to exclude a learning effect. In addition, when interpreting a patient's visual fields, it is probably reasonable to ignore the effect of learning for the central 20° of the visual field in most subjects. It is possible that patients who have never had any sort of visual field examination may demonstrate a larger learning effect on automated perimetry. These results are somewhat different than those we reported in patients with established glaucomatous damage.13 In that study, the only effect of experience was a small improvement in short-term fluctuation, and there did not appear to be a learning effect on mean sensitivity, total loss, or number of disturbed points. Experience seemed to have a much more prominent effect in the sample of glaucoma suspect patients reported in this study. The most likely explanation is that the larger areas of visual field loss and the greater short- and long-term fluctuation seen in glaucoma patients with visual field defects may mask small learning effects.
REFERENCES 1. Aspinall PA. Some methodological problems in testing visual function. Mod Probl Ophthalmol1974; 13:2-7.
47
OPHTHALMOLOGY
•
JANUARY 1990
2. Breton ME, Fletcher DE, Krupin T. Clinical use of the 100-hue test: learning artifact. In: Optical Society of America. Topical Meeting on Noninvasive Assessment of the Visual System: a digest of technical papers presented March 24-26, 1986, Monterey, California. Washington, DC: OSA, 1987; 112-5. 3. Aulhom E, Harms H. Visual perimetry. In: Jameson D, Hurvich LM, eds. Handbook of Sensory Physiology. VII/4: Visual Psychophysics. Berlin: Springer-Verlag, 1972; 102-45. 4. Greve EL. Single and multiple stimulus static perimetry in glaucoma: the two phases of visual field examination. Doc Ophthalmol 1973; 36(thesis): 140-1. 5. Hodapp E. Computerized perimetry in glaucoma. In: Whalen WR, Spaeth GL, eds. Computerized Visual Fields: What They Are and How to Use Them. Thorofare, NJ: Slack Inc., 1985; 195-238. 6. Anderson DR. Perimetry With and Without Automation, 2nd ed. St. Louis: CV Mosby, 1987; 316-7. 7. Gloor BP, Schmied U, Fassler A. Changes of glaucomatous field defects: analysis of Octopus fields with programme Delta. Doc Ophthalmol Proc Ser 1981; 26:11-5. (4th International Visual Field Symposium, April 1980). 8. Gramer E, De Natale R, Leydhecker W. Training effect and fluctuations
48
•
VOLUME 97
•
NUMBER 1
in long-term follow-up of glaucomatous visual field defects calculated with program Delta of the Octopus-perimeter 201. New Trends Ophthalmol 1986; 1:219-28. 9. Heijl A. The implications of the results of computerized perimetry in normals for the statistical evaluation of glaucomatous visual fields. In: Krieglstein GK, ed. Glaucoma Update III. Berlin: Springer-Verlag, 1987; 115-22. 10. Wood JM, Wild JM, Hussey MK, Crews SJ. Serial examination of the normal visual field using Octopus automated projection perimetry: evidence for a learning effect. Acta Ophthalmol1987; 65:326-33. 11. Gloor BP, Dimitrakos SA, Rabineau PA. Long-term follow-up of glaucomatous fields by computerized (OCTOPUS-) perimetry. In: Krieglstein GK, ed. Glaucoma Update III. Berlin: Springer-Verlag 1987; 123-38. 12. Katz J, Sommer A. A longitudinal study of the age-adjusted variability of automated visual fields. Arch Ophthalmol1987; 105:1083-6. 13. Werner EB, Adelson A, Krupin T. Effect of patient experience on the results of automated perimetry in clinically stable glaucoma patients. Ophthalmology 1988; 95:764-7. 14. Heijl A. Lindgren G, Olsson J. The effect of perimetric experience in normal subjects. Arch Ophthalmol1989; 107:81-6.
Abstract: The first four Octopus-automated visual field examinations of 29 patients with elevated intraocular pressure but apparently normal optic discs and Goldmann visual fields were studied for the presence of a learning effect on the visual field parameters of mean sensitivity, number of disturbed test locations, total field loss, and short-term fluctuation. A learning effect, if present, would manifest itself as an improvement in the visual field as patients become more experienced with the test. There was no apparent effect of patient experience on the mean sensitivity of the whole visual fields or the mean sensitivity of the test locations within 20° of fixation. There was a significant (P = 0.Q12) increase in mean sensitivity for the test locations outside 20° of fixation. There were significant (P < 0.01) improvements in short-term fluctuation, total loss, and number of disturbed points between the first and second visual field examinations. The results indicated that there was a learning effect between the first and second automated visual field in glaucoma suspect patients who had previous experience with manual perimetry. It was not, however, very large in most patients and seems to be present in the peripheral portions of the visual field only. In most cases, it was not necessary to obtain more than two "baseline" examinations unless a patient demonstrated unusually high short-term fluctuation or had visual field defects inconsistent with the remainder of their clinical examination. Ophthalmology 1990; 97:44-48
Originally received: April 20, 1989. Revision accepted: July 17,1989. 1
2
Department of Ophthalmology, Hahnemann University, Philadelphia. Glaucoma Service, Scheie Eye Institute, University of Pennsylvania School of Medicine, Philadelphia.
Portions of this material were presented at the Annual Meeting of the Association for Research in Vision and Ophthalmology, Sarasota, May 1988. Supported in part by the Charles E. Goetz teaching and research fund, Scheie Eye Institute. Reprint requests to Elliot B. Werner, MD, Hahnemann University, Department of Ophthalmology, 216 North Broad St, Mail Stop 209, Philadelphia, PA 19102.
44
The learning effect is an important artifact in many psychophysical tests. I,2 It appears as an improvement in performance as a subject gains experience with a test. Several authors have reported small learning effects in subjects undergoing perimetry,3-1O whereas others have been unable to confirm this findingy,I2 In a previous study, we showed that there is a small learning effect on short-term fluctuation in patients with established glaucomatous visual field defects who had previous experience with manual (Goldmann) perimetry.13 Automated perimetry is time-consuming, expensive, and physically demanding on the patient. If it is necessary to perform multiple "baseline" examinations to obtain reliable data because there is a significant learning effect,
WERNER et al
•
EFFECT OF EXPERIENCE IN AUTOMATED PERIMETRY
this would have a major effect on the management of patients with glaucoma. In the current study, we evaluated the effect of patient experience with automated perimetry on the visual field in a sample of glaucoma suspect (ocular hypertensive) patients who have previously had normal results of manual visual field examinations.
MATERIALS AND METHODS The charts of all patients followed in the glaucoma service of the Scheie Eye Institute were reviewed. Glaucoma suspect patients were defined as follows: an intraocular pressure over 22 mmHg on at least two consecutive examinations; a normal appearing optic disc with cup-todisc ratio less than 0.5 based on examination of the fundus photograph with a uniform, healthy-appearing neural rim and no retinal nerve fiber layer defects by direct ophthalmoscopy; and a normal visual field as measured with a Goldmann perimeter before the first automated visual field examination. A normal visual field was defined as the I2e isopter at least 20° from fixation and the I4e isopter at least 40° from fixation, no nasal step greater than 5° width in any isopter and no nasal step of any width in more than one isopter, and absence of a scotoma or localized depression of sensitivity on suprathreshold static screening of mUltiple test locations within the central 20°. All glaucoma suspect subjects who met the following criteria were included in our study: (1) a minimum of four automated visual field examinations with at least 12 months offollow-up between the first and last automated visual field examination; (2) visual acuity of 20/25 or better with no change in visual acuity greater that one Snellen line during the follow-up period; (3) a clinically stable appearance of the optic nerve head during the follow-up period; and (4) unchanged medical glaucoma therapy during the follow-up period if the patient was on treatment, or no treatment initiated during the follow-up period if the patient was not on treatment. All subjects were phakic and had no other known ocular, neurologic, or systemic disease likely to affect the visual field. All visual fields were done on the Octopus 201 perimeter (Interzeag, Bern) using Program 32 with the appropriate refractive correction, and all examinations were done by trained technicians who had extensive experience in the use of the Octopus perimeter. All visual fields were performed with an un dilated pupil. None of the subjects in the sample was using topical miotics, and pupil size varied very little in individual patients from one examination to the next. Mean sensitivity, total loss, total number of disturbed test locations, and short-term fluctuation as measured by the root mean square were determined for the first four automated visual fields of each subject. Mean sensitivity was defined as the mean of the measured decibel thresholds for all test locations in the visual field. Total loss was defined as the sum of the differences between the measured threshold and the age-corrected normal threshold for each
test location where that difference exceeded 4 dB. A dis turbed test location was defined as one with a sensitivity reduced 5 dB or more below the age-corrected normal value stored in the Octopus computer for that test location. The root mean square was defined as the square root of the sum of the variances of the test locations where the threshold was determined twice. Values for mean sensitivity for the whole field, mean sensitivity for test locations between 0° and 10° from fixation, mean sensitivity for test locations between 10° and 20° from fixation, mean sensitivity for test locations between 20° and 30° from fixation, total loss, number of disturbed test locations, and the root mean square were obtained from the Delta Program of the Octopus perimeter. In addition, total loss and number of disturbed test locations were recorded manually from the individual visual field printouts for the test locations between 0° and 20° from fixation and 20° and 30° from fixation. The sample mean of each parameter for each of the first four available visual fields on each subject was calculated. Means were compared using a one-way analysis of variance with a repeated measures design. When any of the analysis of variance tests demonstrated a significant treatment effect (P < 0.05), the outlying means were identified with the Newman-Keuls multiple-range test. To determine if patient responses become more consistent with experience, the absolute difference for' each parameter for each subject between fields one and two was determined and compared with the absolute difference between fields three and four using a paired t test.
RESULTS Twenty-nine subjects were identified who met our study criteria. In subjects where both eyes qualified for the study, one eye was selected at random. Of the 29 eyes, there were 12 right eyes and 17 left eyes. Mean age of the subjects was 61 years (range, 34-79 years). There were 13 men and 16 women. The mean interval over which the first four automated visual fields were performed was 24 months (range, 12-38 months). Table 1 shows the sample means and standard deviations for the mean sensitivities, total loss, disturbed test locations, and root mean square for the four visual fields, and the F ratios and significance levels for each parameter. Highly significant effects (P < 0.01) were found for total loss for the whole field and for the test locations between 20° and 30° from fixation, number of disturbed test locations for the whole field and for the test locations between 20° and 30° from fixation, and short-term fluctuation as measured by the root mean square. There was no effect on total loss or disturbed test locations for the test locations within the central 20° of the visual field and no significant effect on mean sensitivity for the whole field or for the test locations within 10° of fixation or 10° to 20° from fixation. A significant effect (P = 0.012) on mean
45
OPHTHALMOLOGY
•
JANUARY 1990
•
VOLUME 97
•
NUMBER 1
Table 1. Results of Repeated Octopus Visual Field Examinations in Glaucoma Suspect Patients Visual Field Examination (mean ± 1 SD) 2
Parameter Mean sensitivity whole field (dB) Mean sensitivity 0° to 10° (dB) Mean sensitivity 10° to 20° (dB) Mean sensitivity 20° to 30° (dB) Total loss whole field (dB) Total loss 0° to 20° (dB) Total loss 20° to 30° (dB) Disturbed test locations whole field Disturbed test locations 0° to 20° Disturbed test locations 20° to 30° RMS (dB) SD *
25.3 ± 28.8 ± 26.0 ± 24.1 ± 41 ± 9 ± 30 ±
26.1 ± 29.0 ± 26.8 ± 25.0 ± 10 ± 4 ± 7 ±
1.8 1.7 1.8 1.9 64 22 47
6 ± 10
4 ± 7 1.7 ± 0.5
1.5 1.5 1.5 1.5 15 8 12
25.9 ± 28.5 ± 26.6 ± 24.8 ± 14 ± 3 ± 9 ±
1.9 1.8 2.1 1.9 30 7 23
2 ± 2
3 ± 6
±
±
1 ± 2 1.3 ± 0.4
1 ± 3 1.3 ± 0.8
± 4
F ratio*
P
2.37 1.39 2.07 3.88 6.70 1.80 5.63
0.08 0.25 0.11 0.012 0.0004 0.15 0.0012
6.39
0.0006
0 ±
1.68
0.18
1 ± 1 1.3 ± 0.6
5 5. 9 4.19
0.0013 0.008
4
3 25.9 28.4 26.5 25.0 6 2 3
± ± ± ± ± ± ±
1.8 2.0 1.9 1.8 10 4 8
± 2
= standard deviation; RMS = root mean square.
F ratio from one-way analysis of variance with repeated measures. Table 2. Comparison of the Sample Means of the Absolute Differences Between the First Two Visual Field Examinations and the Third and Fourth Examinations Fl-F2* Parameter Mean sensitivity whole field (dB) Mean sensitivity 0° to 10° (dB) Mean sensitivity 10° to 20° (dB) Mean sensitivity 20° to 30° (dB) Total loss (dB) Disturbed test locations RMS (dB)
F3-F4*
(mean absolute difference ± 1 SD) 1.2 ± 0.9 ± 1.3 ± 1.3 ± 33 ± 5 ± 0.6 ±
1.3 1.4 1.4 1.3 59 9 0.5
1.0 ± 0.8 1.1 ± 1.0 1.1± 0.8 1.0 ± 0.8 9 ± 21 2 ± 4 0.6 ± 0.8
Tvaluet
P
0.53 -0.37 0.41 0.93 2.12 2.01 -0.64
0.30 0.36 0.34 0.18 0.02 0.03 0.47
SD = standard deviation; RMS = root mean square. * F1-F2 is the absolute value of the difference between the first and second visual field . F3-F4 is the absolute value of the difference between the third and fourth visual field. t Paired ttest.
sensitivity, however, was found for the most peripherally measured points 20° to 30° from fixation. For all parameters in which there was a significant effect, the sample mean of the first visual field was significantly higher than the three subsequent fields (P < 0.05). No significant difference was found among fields two, three, or four for any of the parameters. Table 2 shows the sample means and standard deviations for the absolute differences between the first two automated visual field examinations compared with the last two for each of the performance indices generated by the Delta Program software. Total loss and number of disturbed test locations were the only parameters where the subjects showed a significant improvement in consistency between examinations.
46
DISCUSSION Response variability upon repeated testing is common with all psychophysical tests. Random variability or fluctuation in perimetry is due to the nature of the visual system and the testing situation. Nonrandom variability may be due to external factors such as pupil size, media opacities, glaucoma therapy, or progression of the patient's disease. Performance effects are an additional source of nonrandom variability. Performance effects are due to the nature of the response of the subject to the testing situation apart from the physiologic function being measured. They may be either positive or negative. Positive performance
WERNER et al
•
EFFECT OF EXPERIENCE IN AUTOMATED PERIMETRY
Table 3. Frequency Distribution of the Total Loss in Decibels for the Whole Field and for the Test Locations Outside 20° from Fixation Total Loss (dB) Whole Field
o
5-20 20-50 >50 Outside 20°
o
5-20 20-50 >50
No. of Subjects
10 10
4
5
10 12 2 5
effects are associated with an improvement in the subject's performance, while negative effects are associated with a deterioration in the subject's performance. The learning effect in psychophysical testing is by definition a type of positive performance effect that results in an improvement in response upon repeated testing. When a sample of subjects undergoes repeated testing, significant improvement will be seen in their average score if there is an important learning effect. It is difficult to infer a learning effect in a single subject because it is impossible to differentiate an improvement due to a learning effect from random fluctuation. Our findings indicate that there is some learning effect on the visual field as measured by the Octopus 201 automated perimeter in glaucoma suspect patients who have had at least one prior manual visual field examination. The learning effect was most significant on short-term fluctuation, total loss, and number of disturbed test locations. The effect on total loss and disturbed test locations was entirely accounted for by the effect on the test locations located more than 20° from fixation. There are two possible situations in which the means of the parameters studied could show a systematic improvement suggestive oflearning. One is a modest learning effect in most or all the subjects and the other is a large effect in a few subjects with no or minimal effect in most of the subjects. The second explanation seems to be the case in our sample. Looking at total loss, for example, we find that only 5 of the 29 subjects had a total loss greater than 50 dB. These subjects showed a fairly marked improvement with experience, while the majority of the subjects showed very little change after the first visual field examination. Table 3 shows the frequency distribution for total loss for the whole field and for the area outside 20° for the sample. The effect on mean sensitivity was much less marked and only detectable in the test locations outside 20° from fixation. In this regard, our results are similar to those reported by Wood et al lO and Heijl et al. 14 The learning effect seemed to occur between the first and second visual field examinations in the peripheral portion of the central 30° of the visual field. This effect persisted during the
third and fourth examinations and was stable. It is clear, however, that there is a lot of variability from subject to subject. One must individualize the interpretation of visual fields in patients. Some patients will seem to show a much greater learning effect than others. The clinician should be very cautious about detecting any change in the visual field when only a few fields are available in an inexperienced patient. In addition to learning, there are other performance effects, such as motivation, fatigue, and effects of drugs. Some of these effects may have a negative impact on performance. It is possible that both negative and positive performance effects might be found in approximately equal numbers of subjects in our sample. In that case, analysis of the means would not reveal a significant change over the four visual fields. To determine whether there were other significant performance effects, we compared the absolute differences between the first two visual fields with the absolute differences between the last two visual fields. If there were larger changes over the first two fields in either a positive or negative direction, this technique would detect them. This did not seem to be the case except for total loss and number of disturbed test locations where this effect would have been expected from the analysis of variance results (Table 2). Thus, there do not appear to be any significant performance effects other than learning in our sample. Our results indicate that learning may play an important role in glaucoma suspect patients undergoing automated perimetry who have been screened previously with manual perimetry. However, it is probably not necessary to obtain more than two "baseline" examinations in the majority of such patients because the learning effect seems to occur between the first and second examinations. Two examinations, therefore, should be sufficient to exclude a learning effect. In addition, when interpreting a patient's visual fields, it is probably reasonable to ignore the effect of learning for the central 20° of the visual field in most subjects. It is possible that patients who have never had any sort of visual field examination may demonstrate a larger learning effect on automated perimetry. These results are somewhat different than those we reported in patients with established glaucomatous damage.13 In that study, the only effect of experience was a small improvement in short-term fluctuation, and there did not appear to be a learning effect on mean sensitivity, total loss, or number of disturbed points. Experience seemed to have a much more prominent effect in the sample of glaucoma suspect patients reported in this study. The most likely explanation is that the larger areas of visual field loss and the greater short- and long-term fluctuation seen in glaucoma patients with visual field defects may mask small learning effects.
REFERENCES 1. Aspinall PA. Some methodological problems in testing visual function. Mod Probl Ophthalmol1974; 13:2-7.
47
OPHTHALMOLOGY
•
JANUARY 1990
2. Breton ME, Fletcher DE, Krupin T. Clinical use of the 100-hue test: learning artifact. In: Optical Society of America. Topical Meeting on Noninvasive Assessment of the Visual System: a digest of technical papers presented March 24-26, 1986, Monterey, California. Washington, DC: OSA, 1987; 112-5. 3. Aulhom E, Harms H. Visual perimetry. In: Jameson D, Hurvich LM, eds. Handbook of Sensory Physiology. VII/4: Visual Psychophysics. Berlin: Springer-Verlag, 1972; 102-45. 4. Greve EL. Single and multiple stimulus static perimetry in glaucoma: the two phases of visual field examination. Doc Ophthalmol 1973; 36(thesis): 140-1. 5. Hodapp E. Computerized perimetry in glaucoma. In: Whalen WR, Spaeth GL, eds. Computerized Visual Fields: What They Are and How to Use Them. Thorofare, NJ: Slack Inc., 1985; 195-238. 6. Anderson DR. Perimetry With and Without Automation, 2nd ed. St. Louis: CV Mosby, 1987; 316-7. 7. Gloor BP, Schmied U, Fassler A. Changes of glaucomatous field defects: analysis of Octopus fields with programme Delta. Doc Ophthalmol Proc Ser 1981; 26:11-5. (4th International Visual Field Symposium, April 1980). 8. Gramer E, De Natale R, Leydhecker W. Training effect and fluctuations
48
•
VOLUME 97
•
NUMBER 1
in long-term follow-up of glaucomatous visual field defects calculated with program Delta of the Octopus-perimeter 201. New Trends Ophthalmol 1986; 1:219-28. 9. Heijl A. The implications of the results of computerized perimetry in normals for the statistical evaluation of glaucomatous visual fields. In: Krieglstein GK, ed. Glaucoma Update III. Berlin: Springer-Verlag, 1987; 115-22. 10. Wood JM, Wild JM, Hussey MK, Crews SJ. Serial examination of the normal visual field using Octopus automated projection perimetry: evidence for a learning effect. Acta Ophthalmol1987; 65:326-33. 11. Gloor BP, Dimitrakos SA, Rabineau PA. Long-term follow-up of glaucomatous fields by computerized (OCTOPUS-) perimetry. In: Krieglstein GK, ed. Glaucoma Update III. Berlin: Springer-Verlag 1987; 123-38. 12. Katz J, Sommer A. A longitudinal study of the age-adjusted variability of automated visual fields. Arch Ophthalmol1987; 105:1083-6. 13. Werner EB, Adelson A, Krupin T. Effect of patient experience on the results of automated perimetry in clinically stable glaucoma patients. Ophthalmology 1988; 95:764-7. 14. Heijl A. Lindgren G, Olsson J. The effect of perimetric experience in normal subjects. Arch Ophthalmol1989; 107:81-6.