- Email: [email protected]
Association between Intraocular Pressure Peaks and Progression of Visual Field Loss
Association between Intraocular Pressure Peal{s and Progression of Visual Field Loss RAN c. ZEIMER, PhD,· JACOB T. WILENSKY, MD,t DAVID K. GIESER, MD,t,2 MARLOS A. G. VIANA, PhD l ,3
Abstract: Little is known about the relation between diurnal variations of intraocular pressure (lOP) and prognosis for glaucomatous visual field damage. The association between apparent progressive loss of visual field and the occurrence of lOP peaks was studied. Pressure peaks were detected by a self-tonometer in the natural environment of patients with glaucoma. The study groups consisted of patients with and without a strong indication of progressive visual field losses, all with lOPs of 22 mm Hg or less obtained in the ophthalmologist's office. Patients apparently undergoing progressive visual field loss were found to have significantly more frequent lOP peaks than patients with stable visual fields. A statistical evaluation indicated that, in a population with a 30% prevalence of progressive loss of visual field, 75% of the patients with peaks have progressive loss and 75% of those without peaks do not have visual field progression. Intraocular pressure peaks were thus shown to have an association with the apparent progression of vision loss independent of the mean lOP. Home tonometry appeared to be a promising tool for identifying patients at increased risk of developing visual field loss who may require intensified follow-up and an alteration in clinical management. However, the present study must be complemented by a prospective study. Ophthalmology 1991; 98:64-69
Originally received: April 17, 1990. Revision accepted: August 21, 1990. Department of Ophthalmology, UIC Eye Center, University of Illinois at Chicago, College of Medicine, Chicago. 2 Wheaton Eye Clinic, Wheaton. 3 School of Public Health, University of Illinois at Chicago, Chicago. 1
Supported in part by a University of Illinois Scholar Award (Dr. leimer), Public Health Service research grant EY03841 , and ophthalmic research core grant EY1792 from the National Eye Institute, Bethesda, MD. The self-tonometer described herein will be available commercially in the near future, at which time Dr. lei mer will have a proprietary interest in it. Reprint requests to Ran C. leimer, PhD, Applied Physics Laboratory, UIC Eye Center, University of Illinois at Chicago, 1855 W. Taylor Street, Chicago, IL 60612.
64
Home tonometry has been advocated by a number of ophthalmologists as a method of improving the diagnosis and treatment of glaucoma_ I - 4 To facilitate the use of home tonometry, we developed a self-tonometerS- 7 that can be used by the patient alone to monitor the diurnal variation in intraocular pressure (lOP) in the patient's natural environment. During the clinical application of home tonometry, we noticed that half of a group of patients with glaucoma considered to have a well-controlled lOP had peaks higher than 22 mmHg. Half of the time these peaks occurred outside of regular office hours. 8 Moreover, within this group, peaks were recorded more often in patients with suspected or documented progression of glaucoma damage than in either those believed to have stable visual fields or in normal subjects. 8 The pur-
ZEIMER et al
•
INTRAOCULAR PRESSURE PEAKS AND VISUAL FIELD LOSS
pose of this article is to investigate further the association between lOP peaks and the progression of glaucomatous damage. We did this by comparing the diurnal lOP curves of patients undergoing progressive visual field loss to those with stable visual fields.
PATIENTS AND METHODS The patients were selected from a list of all the patients at our institutions who had performed home tonometry before 1989. The patients were referred for home tonometry during an effort to gather data on a variety of patients. Some were referred for research reasons only, others because of a history ofIOP fluctuations, and some for clinical reasons due to evidence or suspicion of progressive visual field defects. Two groups were identified within this data base: 14 patients with progression of glaucomatous damage and 19 without progressive damage. To insure that determination of progressive damage would be objective, only information on visual fields was used. Cases were included only if home tonometry was performed within 4 weeks of the diagnosis of visual field progression. The visual fields of each patient were assessed solely by either kinetic isopter perimetry, with the use of the Haag Streit Goldmann perimeter, or static threshold perimetry, with the use of automated computerized perimeters (Octopus, Interzeag, Inc, Northboro, MA, with program 32 or 34 or Humphrey, Allergan Humphrey, SmithKline Beckman Co, San Leandro, CA, with program 30-2). To determine whether the visual field was demonstrating progressive loss, we compared only identical techniques for the same patient. The definitions of glaucomatous damage and its progression are far from absolute as there are other conditions that can cause defects similar to those of glaucoma. In addition, in recent years there has been considerable debate concerning the appropriate criteria for indicating visual field loss. Nonetheless, the following types of defects are strongly suggestive of glaucomatous damage: isolated paracentral scotomata, defects in the arcuate bundle area (more commonly superior than inferior), defects in the nasal step area and, to a much lesser extent, in the temporal step area. The evaluation is more ambiguous when the sensitivity is depressed over the whole visual field due to opacity of the ocular media. While keeping in mind the above-mentioned considerations, we defined progression of visual field loss for kinetic perimetry as (1) a greater than 5 degree enlargement of a nasal step on two isopters, (2) a greater than 10 degree enlargement of a nasal step on one isopter, (3) a greater than 5 degree enlargement of a previous scotoma, or (4) the deepening of a relative scotoma by more than half a log unit. The patterns have been much less established for static threshold perimetry. We adopted the criterion of a 5-dB deviation from the age-adjusted normal used in the delta program of the Octopus perimeter in deciding when to consider a given spot abnormal. Apparent visual field progression was defined as (1) the deepening of a pre-existing scotoma by more than 6 dB, or (2)
a new defect either greater than 8 dB and adjacent to a previous defect, or greater than 10 dB and in a previously normal area. Finally, to compensate in part for the variation in visual fields, we included only patients having sets of at least three consecutive fields. The determination of progression or stability was made before home tonometry was performed, thereby eliminating any bias from the results of the self-tonometry. Patients with applanation tonometry readings averaging 22 mmHg or higher during the three office visits preceding home tonometry were eliminated from the study. This elimination was deemed necessary because any ophthalmologists would consider this level of lOP to be uncontrolled and thus would not require further diagnostic inquiry. Home tonometry was performed with the self-tonometer, which has been previously described. 6 •7 In brief, the instrument consists of a pneumatically driven plunger, fitted with an elastic membrane, which slowly comes forward and flattens the corneal surface. Applanation is detected by an internal optic sensor, and the pressure necessary to achieve applanation is registered automatically. The lOP is then displayed as a digital read-out. To operate the self-tonometer, the patient instills a drop of topical anesthetic, places the orbital rim on a cup, centers the eye by aligning a visual target, and activates the instrument by depressing a knob. The reading is displayed within 2 seconds, and the patient records the value. If the eye is not properly aligned, no reading is displayed. For each eye, the procedure is repeated until four readings are obtained or six attempts have been made. Most patients need no more than four or five attempts to obtain four valid readings. 6 We previously found that the reproducibility of four consecutive readings was ± 1.4 mmHg and that the self-tonometer and the Goldmann applanation tonometer yielded values that were either similar or within ±2.4 mmHg and that were well correlated. 6•7 After the procedure had been explained and an informed consent had been obtained, the patients were trained in the use of the self-tonometer. Subjects who were able to obtain readings were given an instrument and a bottle of 0.5% proparacaine hydrochloride to use as a topical anesthetic. They were instructed to perform self-tonometry in both eyes upon waking and at noon, 4 p.m., dinner, and bedtime over a 4- to 8-day period (average, 6.5 days). Paired home tonometry and applanation tonometry values were obtained when the patients picked up and returned the self-tonometer. The data were entered into a computerized data base. A software program was designed to eliminate occasional abnormal readings (defined as values deviating from the mean of the four [or fewer] readings by more than twice the instrument's standard deviation [SD]); the software program was also used to average the values and plot the results versus time. The lOP peaks were defined as values 6 mmHg above the average of the office applanation readings obtained before and after home tonometry. This definition oflOP peaks, based on a relative elevation, was considered more adequate than one relying on an absolute lOP. Indeed, a 65
OPHTHALMOLOGY
•
JANUARY 1991
definition based on an absolute cutoff would introduce a bias as more peaks would be observed in subjects with a higher average lOP. The results, therefore, could be more affected by the level of lOP than by the degree of diurnal fluctuation. The number of lOP peaks divided by the number of days (referred to as the frequency of peaks) was used to neutralize the effect of the length of monitoring. For patients with bilateral progression or stability of visual fields, the two frequencies were averaged to yield a single value. A frequency of at least 0.5, i.e., a peak every second day, was considered to be a positive test outcome, whereas a lower frequency was considered a negative outcome.
STATISTICAL ANALYSIS The data were analyzed statistically to provide a preliminary indication on the predictive value ofIOP peaks. A number of parameters were used9 as specified in the appendix. The test specificity is defined as the probability that the test will be negative (T- ) among patients without the diagnosis of visual field progression (0- ). The test sensitivity is defined as the probability that the test will be positive (T+) among patients with the diagnosis of visual field progression (0+). The prevalence is defined as the probability of visual field progression (0+). Ultimately the goal was to evaluate the predictive value of a positive test (PVP) to identify correctly those patients with visual field progression and to evaluate the predictive value of a negative test (PVN) to identify correctly the lack of disease progression.
RESULTS According to the definition of visual field loss given in the Patients and Methods section, nine patients with open angle glaucoma were identified as having progressive visual field defects in one eye, five had such a worsening in both eyes, and 21 had well-documented stable visual fields. Table 1 gives the clinical data for these patients. The mean age (±SO) of the patients with a progressive visual field defect was 70 ± 7 years and for those with controlled glaucoma was 64 ± 11 years. The mean office applanation pressures (±SO) were 18.2 ± 2.4 and 17.6 ± 2.2 mmHg for the above two groups, respectively. Of the 14 patients with progressive visual field defects, 4 (29%) had at least one lOP peak every 2 days (0.5 peaks per day) compared with 1 of21 patients with stable visual fields (Table 1). Results of the examination, with the Kolmogorov test, of the cumulative frequency distribution for the two populations indicated there was a significant difference (P < 0.03). To determine if the frequency of peaks provided information in addition to that supplied by office tonometry, 66
•
VOLUME 98
•
NUMBER 1
we performed a number of tests. The average office applanation pressures were compared and found to be 17.8 ± 1.5 mmHg (mean ± SEM) for patients with more than one peak every 2 days and 17.9 ± 0.4 mmHg for those with fewer peaks. The difference, determined by Student's t test, was not statistically significant (P> 0.95). Moreover, the correlation between the frequency of peaks and office tonometry was derived, and no statistically significant correlation was found (r = 0.12, P > 0.46). With the use of the values and the formulas derived above, the test specificity was found to be 0.95 (with a 95% credibility interval of 0.71 to 0.98) and the sensitivity to be 0.29 (with a 95% credibility interval of 0.11 to 0.55). The predictive value of a positive test and predictive value of a negative test, with the aid of the above values, were plotted as a function of prevalence (Figure I). Inspection of the curves indicates that the two predictive values were optimal at a prevalence of 30%. Under this condition, the presence of lOP peaks at least 6 mmHg higher than office tonometry at least every second day had a 75% probability of accompanying a progressive loss in visual field.
DISCUSSION It has been documented repeatedly that, because the lOP varies diurnally, office tonometry has a limited probability of detecting the daily peaks. lO Because they are aware of this limitation, several ophthalmologists have advocated home tonometry with the Schiotz tonometer. To improve the safety, practicality, and accuracy of the measurements, we developed a self-tonometer to be used by the patient alone in his or her natural environment. 5,6 We have reported on the application of the self-tonometer in patients with glaucoma and have confirmed that home tonometry yields information on the lOP that could not be predicted by office tonometry.7,S However, the major clinical issue remains whether or not these occasional elevated values should concern the ophthalmologist. Little is known about the relationship between the diurnal lOP variations and prognosis. Katavisto " found that, despite therapy, deterioration was fastest in eyes with high diurnal lOP variations. Sampaolesi l 2 noted that isolated lOP measurements greater than 24 mmHg, mean lOPs greater than 19 mmHg, or variabilities in lOP larger than ±2 mmHg always led to damage. Recently, Alpar4 reported that patients with ocular hypertension with home tonometry values 4 to 13 mmHg above their office readings exhibited progressive glaucoma damage within 2 years. Conversely, those who had office readings that were above the home readings did not develop damage during a 5year follow-up. Unfortunately, these striking results do not lend themselves to scrutiny because the report lacks sufficient details. Only a prolonged prospective study, which we have begun, can answer this question unequivocally. As an intermediate step toward determining the clinical importance of occasional lOP peaks, we studied the as-
ZEIMER et al
•
INTRAOCULAR PRESSURE PEAKS AND VISUAL FIELD LOSS
Table 1. Clinical Data of Glaucoma Patients and Results of Home Tonometry Case No./Sex/ Diagnosis
Age (yrs)
Perimetry
Visual Field Code*
No. of Peaks per Day
Office lOP (mmHg)
0 0 0.38 1.88 0.63 0.60 0 0 0.14
20.0 19.0 19.5 20.5 21.0 12.5 21.0 20.0 19.5
0.31 0 1.58 0 0 0.39 ± 0.59
16.5 16.0 17.0 17.0 15.5 18.2 ± 2.4
0 0 0.42 0.25 0 0 0.29 0.25 0 0 0.81 0 0 0 0 0 0.12 0 0.8 0 0 0.11 ± 0.20
15.1 21.0 21.8 17.5 19.8 17.4 18.8 18.2 14.2 15.8 18.0 17.8 16.0 16.0 20.3 15.3 16.5 15.5 14.8 21.0 19.3 17.6 ± 22
Unilateral Progression of Visual Field Loss l/M/POAG 2/M/POAG 3/M/POAG 4/M/POAG 5/M/POAG 6/M/exfoliation 7/F/POAG 8/M/exfoliation 9/M/POAG
77
65 66 75 80 83 62 57 70
A G G G G G A G
2 2 2 2, 3, 9 2, 4 2, 4, 6 2 9 4
Bilateral Progression of Visual Field Loss 1O/M/low tension 11/F/Iow tension 12jM/POAG 13/F/POAG 14/F/Iow tension Mean ± SO
75 73 63 62 69
70 ± 7
G G A A A
2, 4/4, 5 4, 5/2, 4 4/4 1,9{1 7, 8, 9/4, 9 NA
Bilateral Controlled Glaucoma l/F/ACG 2/M/pigmentary 3/M/exfoliation 4/M/POAG 5/F/POAG 6/M/POAG 7/F/POAG 8/M/POAG 9/F/exfoliation 10/M/ACG 11/F/ACG 12/M/POAG 13/F/POAG 14/M/POAG 15/M/POAG 16/M/POAG 17/F/POAG 18/M/ACG 19/M/POAG 20/F/POAG 21/F/POAG Mean ± SO
54 44 76 64 72 79 63
77
69 78 75 63 76 59 69 65 52 47 47 54 61
64 ± 11
G G G G G G G G A A A G G A A A A A A A
2, 9/2, 9 4/2 2, 9/2, 6 2, 3/3 9/3 2{1 2/2, 3 1/3, 4 2, 5/2, 9 2, 4,6/9 2{1 4/4 1{1 2/2 4/1 2/2 4/9 1/1 4/2 9/9 1/1 NA
lOP = intraocular pressure (average of 4 readings); POAG = primary open-angle glaucoma; A = automated static perimetry; G = Goldmann kinetic perimetry; SO = standard deviation; NA = not applicable; ACG = angle-closure glaucoma. * 1= normal; 2 = nasal step; 3 = paracentral scotoma; 4 = Bjerrum's scotoma; 5 = split fixation; 6 = contraction of visual field; 7 = central island preservation; 8 = temporal island preservation; and 9 = other. Values for patients with bilateral progression and those with controlled glaucoma are right eye/left eye.
sociation between progression of visual field defects and the occurrence of lOP peaks immediately after the diagnosis of progression. To insure the study remained clinically relevant, we limited our population to subjects who had an apparently well-controlled lOP, i.e., a value historically considered to be less than 22 mmHg. This selection had the beneficial effect of yielding two subgroups
with similar office applanation readings, which tended to rule out the influence of office lOP readings in the comparison. This lack of influence was supported by the similarity in office readings between the subjects with and without frequent lOP peaks and by the lack of correlation between the frequency of peaks and office tonometry. As noted in the Patients and Methods section, diagnosis
67
OPHTHALMOLOGY
•
JANUARY 1991
of visual field progression was aided by the availability of quantitative parameters but ultimately was still a matter of clinical judgment. The lack of strict standardization present in our study was also inherent in the determination of progression of glaucoma damage. Nevertheless, because both groups studied were evaluated by the same criteria, the difference in diurnal lOP peaks between the groups was significant. The main finding of the study was that a large proportion of patients with a progressive visual field defect had a greater frequency of lOP peaks compared with patients with stable visual fields. The difference was evident on first impression (29% versus 5%) as well as under statistical scrutiny. This finding is clearly not conclusive evidence that lOP peaks are the sole cause of visual field loss. At this point, it is only an indication that peaks are associated with damage. The findings also do not imply that the presence of peaks is a sufficient and necessary condition for loss of vision. As is widely accepted, a high lOP should not be regarded as an inevitable precursor of visual field loss but only as an indication that the patient is at risk. The same applies to lOP peaks. It must be emphasized that all the results and predictive values in our study were based on home tonometry readings performed five times daily. More frequent monitoring may increase the efficacy of peak detection and thereby increase the sensitivity and possibly the specificity of the test. Currently, we prefer to sacrifice these possible benefits to minimize the demands on the subjects and the potential side effects of repeated instillation of local anesthesia. Moreover, home tonometry was performed as soon as the visual field progression was diagnosed. It is possible that more peaks may have been present earlier during the process. We evaluated the diagnostic value of home tonometry by calculating the predictive value of positive tests, namely lOP peaks, and predictive value of negative tests of home tonometry. The resulting data shed some light on the practical implications of home tonometry. In a population with a low prevalence of progressive loss of visual field, home tonometry will have a limited value because the predictive value ofIOP peaks is low. In a population with a very high risk of developing visual field loss, home tonometry will be limited because a lack of lOP peaks has a low predictive value in ruling out progression. However, if we consider a population with a 30% prevalence of visual field progression, the positive and negative predictive values are satisfactory because 75% of the patients with peaks have progressive loss of visual field and 75% of those without peaks do not have visual field progression. Thus, under these conditions, home tonometry may be a valuable tool for identifying patients in whom visual field losses are likely to be progressing. Obviously, the clinical value of home tonometry could be enhanced if the clinician detected the peaks before progression of damage. Even the knowledge that progression is occurring is important, however, because a number of careful follow-up examinations are necessary with the current methods of diagnosis to confirm progression. At this point, it is left to the clinician to judge what portion of patients would constitute a population with a 30%
68
•
VOLUME 98
•
NUMBER 1
1
0.8
0.6
~
:cas
0.4
.0
a:o
0.2 o~
__ ____ ____ __ ____ ~
o
~
~
0.4
0.2
0.6
~
0.8
~
1
Prevalence Fig 1. Predictive value of intraocular pressure peaks measured by home tonometry as a function of prevalence of progressive loss of visual field. PVP indicates predictive value of a positive test in correctly identifying patients with visual field progression; PVN indicates predictive value of a negative test in correctly identifying patients who do not have progression of visual field damage.
prevalence of visual field progression. Generally speaking, patients with office lOPs of less than 22 mmHg who do not have well-established recorded stable visual fields may fit in this category.
APPENDIX Test specificity: 0 = P[T -I 0-]. Test sensitivity: 1'/ Prevalence: 'Tr
=
=
P[T+ I0+].
P[O+].
PVP
=
P[O+ 1T+].
PVN
=
P[O-I r].
To estimate these values based on readily available information 0[0, 1'/, and 'Tr, one can use Bayes's theorem 13, 14 to obtain, PVP = P[O+IT+] = P[T+IO+]'Tr p[r] PVN = p[O-lr] = p[rIO-](l - 'Tr) p[r] Moreover, since P[T+]
=
P[T+IO+]P[O+]
=
1'/
'Tr
+ (1
+ P[T+IO-]P[O-]
- 0)(1 - 'Tr)
ZEIMER et al
•
INTRAOCULAR PRESSURE PEAKS AND VISUAL FIELD LOSS
and p[r] = p[r I D+]P[D+]
= 8(1
- 11")
+ (1
+ P[T-I D-]P[D-]
- 1/)11",
it follows that pvp=
PVN
=
1/11"
(1)
8(1 - 11") • 8(1 - 11") + (1 - 1/)11"
(2)
1/11"
+ (1
- 8)(1 - 11") ,
With the use of equations (1) and (2) and the sensitivity and specificity estimates from independent groups, the PVP and PVN can be estimated as a function of the prevalence. Moreover, approximate lower and upper confidence interval limits can be also determined based on the credibility intervals for the specificity and sensitivity.
REFERENCES 1. Radn6t M, Follmann P. Diagnosis of primary open·angle glaucoma. In: Bellows JG, ed. Glaucoma: Contemporary International Concepts. New York: Masson, 1979; 213-38.
2. Posner A. Home use of the applanometer as an aid in the management of glaucoma. Eye Ear Nose Throat Mon 1965; 44(8):64,66. 3. Armaly MF. The visual field defect and ocular pressure level in open angle glaucoma. Invest Ophthalrnol1969; 8:105-24. 4. Alpar JJ. The use of home tonornetry in the diagnosis and treatment of glaucoma. Glaucoma 1983; 5:130-2. 5. Zeimer RC, Wilensky JT. An instrument for self-measurement of intraocular pressure. IEEE Trans Biomed Eng 1982; 29:178-83. 6. Zeimer RC, Wilensky JT, Gieser OK, et al. Evaluation of a self tonometer for home use. Arch Ophthalmol 1983; 101: 1791-3. 7. Zeimer RC, Wilensky JT, Gieser OK, et al. Application of a self-tonometer to home tonometry. Arch Ophthalmol1986; 104:49-53. 8. Wilensky JT, Gieser OK, Mori MT, et al. Self-tonometry to manage patients with glaucoma and apparently controlled intraocular pressure. Arch Ophthalmol1987; 105:1072-5. 9. Gastwirth JL. The statistical precision of medical screening procedures: application to polygraph and AIDS antibodies test data. Stat Sci 1987; 2:213-38. 10. Zeimer RC. Circadian variations in intraocular pressure. In: Ritch R, Shields MB, Krupin T, eds. The Glaucomas, vol. 1. St. Louis: CV Mosby, 1989; 319-35. 11. Katavisto M. The diurnal variations of ocular tension in glaucoma. Acta Ophthalmol [Suppl) 1964; 78:1-131. 12. Sampaolesi R. Personal interview. In: Boyd BF, ed. Highlights of Ophthalmology: 1975-1976. Panama: Clinica Boyd [1976); 414-37. 13. Press JS. Bayesian Statistics. New York: Wiley, 1989. 14. Ingelfinger JA, Mosteller F, Thibodeau LA, Ware JH. Biostatics in Clinical Medicine, 2nd ed. New York: MacMillan, 1987.
69
Abstract: Little is known about the relation between diurnal variations of intraocular pressure (lOP) and prognosis for glaucomatous visual field damage. The association between apparent progressive loss of visual field and the occurrence of lOP peaks was studied. Pressure peaks were detected by a self-tonometer in the natural environment of patients with glaucoma. The study groups consisted of patients with and without a strong indication of progressive visual field losses, all with lOPs of 22 mm Hg or less obtained in the ophthalmologist's office. Patients apparently undergoing progressive visual field loss were found to have significantly more frequent lOP peaks than patients with stable visual fields. A statistical evaluation indicated that, in a population with a 30% prevalence of progressive loss of visual field, 75% of the patients with peaks have progressive loss and 75% of those without peaks do not have visual field progression. Intraocular pressure peaks were thus shown to have an association with the apparent progression of vision loss independent of the mean lOP. Home tonometry appeared to be a promising tool for identifying patients at increased risk of developing visual field loss who may require intensified follow-up and an alteration in clinical management. However, the present study must be complemented by a prospective study. Ophthalmology 1991; 98:64-69
Originally received: April 17, 1990. Revision accepted: August 21, 1990. Department of Ophthalmology, UIC Eye Center, University of Illinois at Chicago, College of Medicine, Chicago. 2 Wheaton Eye Clinic, Wheaton. 3 School of Public Health, University of Illinois at Chicago, Chicago. 1
Supported in part by a University of Illinois Scholar Award (Dr. leimer), Public Health Service research grant EY03841 , and ophthalmic research core grant EY1792 from the National Eye Institute, Bethesda, MD. The self-tonometer described herein will be available commercially in the near future, at which time Dr. lei mer will have a proprietary interest in it. Reprint requests to Ran C. leimer, PhD, Applied Physics Laboratory, UIC Eye Center, University of Illinois at Chicago, 1855 W. Taylor Street, Chicago, IL 60612.
64
Home tonometry has been advocated by a number of ophthalmologists as a method of improving the diagnosis and treatment of glaucoma_ I - 4 To facilitate the use of home tonometry, we developed a self-tonometerS- 7 that can be used by the patient alone to monitor the diurnal variation in intraocular pressure (lOP) in the patient's natural environment. During the clinical application of home tonometry, we noticed that half of a group of patients with glaucoma considered to have a well-controlled lOP had peaks higher than 22 mmHg. Half of the time these peaks occurred outside of regular office hours. 8 Moreover, within this group, peaks were recorded more often in patients with suspected or documented progression of glaucoma damage than in either those believed to have stable visual fields or in normal subjects. 8 The pur-
ZEIMER et al
•
INTRAOCULAR PRESSURE PEAKS AND VISUAL FIELD LOSS
pose of this article is to investigate further the association between lOP peaks and the progression of glaucomatous damage. We did this by comparing the diurnal lOP curves of patients undergoing progressive visual field loss to those with stable visual fields.
PATIENTS AND METHODS The patients were selected from a list of all the patients at our institutions who had performed home tonometry before 1989. The patients were referred for home tonometry during an effort to gather data on a variety of patients. Some were referred for research reasons only, others because of a history ofIOP fluctuations, and some for clinical reasons due to evidence or suspicion of progressive visual field defects. Two groups were identified within this data base: 14 patients with progression of glaucomatous damage and 19 without progressive damage. To insure that determination of progressive damage would be objective, only information on visual fields was used. Cases were included only if home tonometry was performed within 4 weeks of the diagnosis of visual field progression. The visual fields of each patient were assessed solely by either kinetic isopter perimetry, with the use of the Haag Streit Goldmann perimeter, or static threshold perimetry, with the use of automated computerized perimeters (Octopus, Interzeag, Inc, Northboro, MA, with program 32 or 34 or Humphrey, Allergan Humphrey, SmithKline Beckman Co, San Leandro, CA, with program 30-2). To determine whether the visual field was demonstrating progressive loss, we compared only identical techniques for the same patient. The definitions of glaucomatous damage and its progression are far from absolute as there are other conditions that can cause defects similar to those of glaucoma. In addition, in recent years there has been considerable debate concerning the appropriate criteria for indicating visual field loss. Nonetheless, the following types of defects are strongly suggestive of glaucomatous damage: isolated paracentral scotomata, defects in the arcuate bundle area (more commonly superior than inferior), defects in the nasal step area and, to a much lesser extent, in the temporal step area. The evaluation is more ambiguous when the sensitivity is depressed over the whole visual field due to opacity of the ocular media. While keeping in mind the above-mentioned considerations, we defined progression of visual field loss for kinetic perimetry as (1) a greater than 5 degree enlargement of a nasal step on two isopters, (2) a greater than 10 degree enlargement of a nasal step on one isopter, (3) a greater than 5 degree enlargement of a previous scotoma, or (4) the deepening of a relative scotoma by more than half a log unit. The patterns have been much less established for static threshold perimetry. We adopted the criterion of a 5-dB deviation from the age-adjusted normal used in the delta program of the Octopus perimeter in deciding when to consider a given spot abnormal. Apparent visual field progression was defined as (1) the deepening of a pre-existing scotoma by more than 6 dB, or (2)
a new defect either greater than 8 dB and adjacent to a previous defect, or greater than 10 dB and in a previously normal area. Finally, to compensate in part for the variation in visual fields, we included only patients having sets of at least three consecutive fields. The determination of progression or stability was made before home tonometry was performed, thereby eliminating any bias from the results of the self-tonometry. Patients with applanation tonometry readings averaging 22 mmHg or higher during the three office visits preceding home tonometry were eliminated from the study. This elimination was deemed necessary because any ophthalmologists would consider this level of lOP to be uncontrolled and thus would not require further diagnostic inquiry. Home tonometry was performed with the self-tonometer, which has been previously described. 6 •7 In brief, the instrument consists of a pneumatically driven plunger, fitted with an elastic membrane, which slowly comes forward and flattens the corneal surface. Applanation is detected by an internal optic sensor, and the pressure necessary to achieve applanation is registered automatically. The lOP is then displayed as a digital read-out. To operate the self-tonometer, the patient instills a drop of topical anesthetic, places the orbital rim on a cup, centers the eye by aligning a visual target, and activates the instrument by depressing a knob. The reading is displayed within 2 seconds, and the patient records the value. If the eye is not properly aligned, no reading is displayed. For each eye, the procedure is repeated until four readings are obtained or six attempts have been made. Most patients need no more than four or five attempts to obtain four valid readings. 6 We previously found that the reproducibility of four consecutive readings was ± 1.4 mmHg and that the self-tonometer and the Goldmann applanation tonometer yielded values that were either similar or within ±2.4 mmHg and that were well correlated. 6•7 After the procedure had been explained and an informed consent had been obtained, the patients were trained in the use of the self-tonometer. Subjects who were able to obtain readings were given an instrument and a bottle of 0.5% proparacaine hydrochloride to use as a topical anesthetic. They were instructed to perform self-tonometry in both eyes upon waking and at noon, 4 p.m., dinner, and bedtime over a 4- to 8-day period (average, 6.5 days). Paired home tonometry and applanation tonometry values were obtained when the patients picked up and returned the self-tonometer. The data were entered into a computerized data base. A software program was designed to eliminate occasional abnormal readings (defined as values deviating from the mean of the four [or fewer] readings by more than twice the instrument's standard deviation [SD]); the software program was also used to average the values and plot the results versus time. The lOP peaks were defined as values 6 mmHg above the average of the office applanation readings obtained before and after home tonometry. This definition oflOP peaks, based on a relative elevation, was considered more adequate than one relying on an absolute lOP. Indeed, a 65
OPHTHALMOLOGY
•
JANUARY 1991
definition based on an absolute cutoff would introduce a bias as more peaks would be observed in subjects with a higher average lOP. The results, therefore, could be more affected by the level of lOP than by the degree of diurnal fluctuation. The number of lOP peaks divided by the number of days (referred to as the frequency of peaks) was used to neutralize the effect of the length of monitoring. For patients with bilateral progression or stability of visual fields, the two frequencies were averaged to yield a single value. A frequency of at least 0.5, i.e., a peak every second day, was considered to be a positive test outcome, whereas a lower frequency was considered a negative outcome.
STATISTICAL ANALYSIS The data were analyzed statistically to provide a preliminary indication on the predictive value ofIOP peaks. A number of parameters were used9 as specified in the appendix. The test specificity is defined as the probability that the test will be negative (T- ) among patients without the diagnosis of visual field progression (0- ). The test sensitivity is defined as the probability that the test will be positive (T+) among patients with the diagnosis of visual field progression (0+). The prevalence is defined as the probability of visual field progression (0+). Ultimately the goal was to evaluate the predictive value of a positive test (PVP) to identify correctly those patients with visual field progression and to evaluate the predictive value of a negative test (PVN) to identify correctly the lack of disease progression.
RESULTS According to the definition of visual field loss given in the Patients and Methods section, nine patients with open angle glaucoma were identified as having progressive visual field defects in one eye, five had such a worsening in both eyes, and 21 had well-documented stable visual fields. Table 1 gives the clinical data for these patients. The mean age (±SO) of the patients with a progressive visual field defect was 70 ± 7 years and for those with controlled glaucoma was 64 ± 11 years. The mean office applanation pressures (±SO) were 18.2 ± 2.4 and 17.6 ± 2.2 mmHg for the above two groups, respectively. Of the 14 patients with progressive visual field defects, 4 (29%) had at least one lOP peak every 2 days (0.5 peaks per day) compared with 1 of21 patients with stable visual fields (Table 1). Results of the examination, with the Kolmogorov test, of the cumulative frequency distribution for the two populations indicated there was a significant difference (P < 0.03). To determine if the frequency of peaks provided information in addition to that supplied by office tonometry, 66
•
VOLUME 98
•
NUMBER 1
we performed a number of tests. The average office applanation pressures were compared and found to be 17.8 ± 1.5 mmHg (mean ± SEM) for patients with more than one peak every 2 days and 17.9 ± 0.4 mmHg for those with fewer peaks. The difference, determined by Student's t test, was not statistically significant (P> 0.95). Moreover, the correlation between the frequency of peaks and office tonometry was derived, and no statistically significant correlation was found (r = 0.12, P > 0.46). With the use of the values and the formulas derived above, the test specificity was found to be 0.95 (with a 95% credibility interval of 0.71 to 0.98) and the sensitivity to be 0.29 (with a 95% credibility interval of 0.11 to 0.55). The predictive value of a positive test and predictive value of a negative test, with the aid of the above values, were plotted as a function of prevalence (Figure I). Inspection of the curves indicates that the two predictive values were optimal at a prevalence of 30%. Under this condition, the presence of lOP peaks at least 6 mmHg higher than office tonometry at least every second day had a 75% probability of accompanying a progressive loss in visual field.
DISCUSSION It has been documented repeatedly that, because the lOP varies diurnally, office tonometry has a limited probability of detecting the daily peaks. lO Because they are aware of this limitation, several ophthalmologists have advocated home tonometry with the Schiotz tonometer. To improve the safety, practicality, and accuracy of the measurements, we developed a self-tonometer to be used by the patient alone in his or her natural environment. 5,6 We have reported on the application of the self-tonometer in patients with glaucoma and have confirmed that home tonometry yields information on the lOP that could not be predicted by office tonometry.7,S However, the major clinical issue remains whether or not these occasional elevated values should concern the ophthalmologist. Little is known about the relationship between the diurnal lOP variations and prognosis. Katavisto " found that, despite therapy, deterioration was fastest in eyes with high diurnal lOP variations. Sampaolesi l 2 noted that isolated lOP measurements greater than 24 mmHg, mean lOPs greater than 19 mmHg, or variabilities in lOP larger than ±2 mmHg always led to damage. Recently, Alpar4 reported that patients with ocular hypertension with home tonometry values 4 to 13 mmHg above their office readings exhibited progressive glaucoma damage within 2 years. Conversely, those who had office readings that were above the home readings did not develop damage during a 5year follow-up. Unfortunately, these striking results do not lend themselves to scrutiny because the report lacks sufficient details. Only a prolonged prospective study, which we have begun, can answer this question unequivocally. As an intermediate step toward determining the clinical importance of occasional lOP peaks, we studied the as-
ZEIMER et al
•
INTRAOCULAR PRESSURE PEAKS AND VISUAL FIELD LOSS
Table 1. Clinical Data of Glaucoma Patients and Results of Home Tonometry Case No./Sex/ Diagnosis
Age (yrs)
Perimetry
Visual Field Code*
No. of Peaks per Day
Office lOP (mmHg)
0 0 0.38 1.88 0.63 0.60 0 0 0.14
20.0 19.0 19.5 20.5 21.0 12.5 21.0 20.0 19.5
0.31 0 1.58 0 0 0.39 ± 0.59
16.5 16.0 17.0 17.0 15.5 18.2 ± 2.4
0 0 0.42 0.25 0 0 0.29 0.25 0 0 0.81 0 0 0 0 0 0.12 0 0.8 0 0 0.11 ± 0.20
15.1 21.0 21.8 17.5 19.8 17.4 18.8 18.2 14.2 15.8 18.0 17.8 16.0 16.0 20.3 15.3 16.5 15.5 14.8 21.0 19.3 17.6 ± 22
Unilateral Progression of Visual Field Loss l/M/POAG 2/M/POAG 3/M/POAG 4/M/POAG 5/M/POAG 6/M/exfoliation 7/F/POAG 8/M/exfoliation 9/M/POAG
77
65 66 75 80 83 62 57 70
A G G G G G A G
2 2 2 2, 3, 9 2, 4 2, 4, 6 2 9 4
Bilateral Progression of Visual Field Loss 1O/M/low tension 11/F/Iow tension 12jM/POAG 13/F/POAG 14/F/Iow tension Mean ± SO
75 73 63 62 69
70 ± 7
G G A A A
2, 4/4, 5 4, 5/2, 4 4/4 1,9{1 7, 8, 9/4, 9 NA
Bilateral Controlled Glaucoma l/F/ACG 2/M/pigmentary 3/M/exfoliation 4/M/POAG 5/F/POAG 6/M/POAG 7/F/POAG 8/M/POAG 9/F/exfoliation 10/M/ACG 11/F/ACG 12/M/POAG 13/F/POAG 14/M/POAG 15/M/POAG 16/M/POAG 17/F/POAG 18/M/ACG 19/M/POAG 20/F/POAG 21/F/POAG Mean ± SO
54 44 76 64 72 79 63
77
69 78 75 63 76 59 69 65 52 47 47 54 61
64 ± 11
G G G G G G G G A A A G G A A A A A A A
2, 9/2, 9 4/2 2, 9/2, 6 2, 3/3 9/3 2{1 2/2, 3 1/3, 4 2, 5/2, 9 2, 4,6/9 2{1 4/4 1{1 2/2 4/1 2/2 4/9 1/1 4/2 9/9 1/1 NA
lOP = intraocular pressure (average of 4 readings); POAG = primary open-angle glaucoma; A = automated static perimetry; G = Goldmann kinetic perimetry; SO = standard deviation; NA = not applicable; ACG = angle-closure glaucoma. * 1= normal; 2 = nasal step; 3 = paracentral scotoma; 4 = Bjerrum's scotoma; 5 = split fixation; 6 = contraction of visual field; 7 = central island preservation; 8 = temporal island preservation; and 9 = other. Values for patients with bilateral progression and those with controlled glaucoma are right eye/left eye.
sociation between progression of visual field defects and the occurrence of lOP peaks immediately after the diagnosis of progression. To insure the study remained clinically relevant, we limited our population to subjects who had an apparently well-controlled lOP, i.e., a value historically considered to be less than 22 mmHg. This selection had the beneficial effect of yielding two subgroups
with similar office applanation readings, which tended to rule out the influence of office lOP readings in the comparison. This lack of influence was supported by the similarity in office readings between the subjects with and without frequent lOP peaks and by the lack of correlation between the frequency of peaks and office tonometry. As noted in the Patients and Methods section, diagnosis
67
OPHTHALMOLOGY
•
JANUARY 1991
of visual field progression was aided by the availability of quantitative parameters but ultimately was still a matter of clinical judgment. The lack of strict standardization present in our study was also inherent in the determination of progression of glaucoma damage. Nevertheless, because both groups studied were evaluated by the same criteria, the difference in diurnal lOP peaks between the groups was significant. The main finding of the study was that a large proportion of patients with a progressive visual field defect had a greater frequency of lOP peaks compared with patients with stable visual fields. The difference was evident on first impression (29% versus 5%) as well as under statistical scrutiny. This finding is clearly not conclusive evidence that lOP peaks are the sole cause of visual field loss. At this point, it is only an indication that peaks are associated with damage. The findings also do not imply that the presence of peaks is a sufficient and necessary condition for loss of vision. As is widely accepted, a high lOP should not be regarded as an inevitable precursor of visual field loss but only as an indication that the patient is at risk. The same applies to lOP peaks. It must be emphasized that all the results and predictive values in our study were based on home tonometry readings performed five times daily. More frequent monitoring may increase the efficacy of peak detection and thereby increase the sensitivity and possibly the specificity of the test. Currently, we prefer to sacrifice these possible benefits to minimize the demands on the subjects and the potential side effects of repeated instillation of local anesthesia. Moreover, home tonometry was performed as soon as the visual field progression was diagnosed. It is possible that more peaks may have been present earlier during the process. We evaluated the diagnostic value of home tonometry by calculating the predictive value of positive tests, namely lOP peaks, and predictive value of negative tests of home tonometry. The resulting data shed some light on the practical implications of home tonometry. In a population with a low prevalence of progressive loss of visual field, home tonometry will have a limited value because the predictive value ofIOP peaks is low. In a population with a very high risk of developing visual field loss, home tonometry will be limited because a lack of lOP peaks has a low predictive value in ruling out progression. However, if we consider a population with a 30% prevalence of visual field progression, the positive and negative predictive values are satisfactory because 75% of the patients with peaks have progressive loss of visual field and 75% of those without peaks do not have visual field progression. Thus, under these conditions, home tonometry may be a valuable tool for identifying patients in whom visual field losses are likely to be progressing. Obviously, the clinical value of home tonometry could be enhanced if the clinician detected the peaks before progression of damage. Even the knowledge that progression is occurring is important, however, because a number of careful follow-up examinations are necessary with the current methods of diagnosis to confirm progression. At this point, it is left to the clinician to judge what portion of patients would constitute a population with a 30%
68
•
VOLUME 98
•
NUMBER 1
1
0.8
0.6
~
:cas
0.4
.0
a:o
0.2 o~
__ ____ ____ __ ____ ~
o
~
~
0.4
0.2
0.6
~
0.8
~
1
Prevalence Fig 1. Predictive value of intraocular pressure peaks measured by home tonometry as a function of prevalence of progressive loss of visual field. PVP indicates predictive value of a positive test in correctly identifying patients with visual field progression; PVN indicates predictive value of a negative test in correctly identifying patients who do not have progression of visual field damage.
prevalence of visual field progression. Generally speaking, patients with office lOPs of less than 22 mmHg who do not have well-established recorded stable visual fields may fit in this category.
APPENDIX Test specificity: 0 = P[T -I 0-]. Test sensitivity: 1'/ Prevalence: 'Tr
=
=
P[T+ I0+].
P[O+].
PVP
=
P[O+ 1T+].
PVN
=
P[O-I r].
To estimate these values based on readily available information 0[0, 1'/, and 'Tr, one can use Bayes's theorem 13, 14 to obtain, PVP = P[O+IT+] = P[T+IO+]'Tr p[r] PVN = p[O-lr] = p[rIO-](l - 'Tr) p[r] Moreover, since P[T+]
=
P[T+IO+]P[O+]
=
1'/
'Tr
+ (1
+ P[T+IO-]P[O-]
- 0)(1 - 'Tr)
ZEIMER et al
•
INTRAOCULAR PRESSURE PEAKS AND VISUAL FIELD LOSS
and p[r] = p[r I D+]P[D+]
= 8(1
- 11")
+ (1
+ P[T-I D-]P[D-]
- 1/)11",
it follows that pvp=
PVN
=
1/11"
(1)
8(1 - 11") • 8(1 - 11") + (1 - 1/)11"
(2)
1/11"
+ (1
- 8)(1 - 11") ,
With the use of equations (1) and (2) and the sensitivity and specificity estimates from independent groups, the PVP and PVN can be estimated as a function of the prevalence. Moreover, approximate lower and upper confidence interval limits can be also determined based on the credibility intervals for the specificity and sensitivity.
REFERENCES 1. Radn6t M, Follmann P. Diagnosis of primary open·angle glaucoma. In: Bellows JG, ed. Glaucoma: Contemporary International Concepts. New York: Masson, 1979; 213-38.
2. Posner A. Home use of the applanometer as an aid in the management of glaucoma. Eye Ear Nose Throat Mon 1965; 44(8):64,66. 3. Armaly MF. The visual field defect and ocular pressure level in open angle glaucoma. Invest Ophthalrnol1969; 8:105-24. 4. Alpar JJ. The use of home tonornetry in the diagnosis and treatment of glaucoma. Glaucoma 1983; 5:130-2. 5. Zeimer RC, Wilensky JT. An instrument for self-measurement of intraocular pressure. IEEE Trans Biomed Eng 1982; 29:178-83. 6. Zeimer RC, Wilensky JT, Gieser OK, et al. Evaluation of a self tonometer for home use. Arch Ophthalmol 1983; 101: 1791-3. 7. Zeimer RC, Wilensky JT, Gieser OK, et al. Application of a self-tonometer to home tonometry. Arch Ophthalmol1986; 104:49-53. 8. Wilensky JT, Gieser OK, Mori MT, et al. Self-tonometry to manage patients with glaucoma and apparently controlled intraocular pressure. Arch Ophthalmol1987; 105:1072-5. 9. Gastwirth JL. The statistical precision of medical screening procedures: application to polygraph and AIDS antibodies test data. Stat Sci 1987; 2:213-38. 10. Zeimer RC. Circadian variations in intraocular pressure. In: Ritch R, Shields MB, Krupin T, eds. The Glaucomas, vol. 1. St. Louis: CV Mosby, 1989; 319-35. 11. Katavisto M. The diurnal variations of ocular tension in glaucoma. Acta Ophthalmol [Suppl) 1964; 78:1-131. 12. Sampaolesi R. Personal interview. In: Boyd BF, ed. Highlights of Ophthalmology: 1975-1976. Panama: Clinica Boyd [1976); 414-37. 13. Press JS. Bayesian Statistics. New York: Wiley, 1989. 14. Ingelfinger JA, Mosteller F, Thibodeau LA, Ware JH. Biostatics in Clinical Medicine, 2nd ed. New York: MacMillan, 1987.
69