Upload
peep
View
212
Download
0
Embed Size (px)
Citation preview
Relationship of Duration and Onset of Diabetes toPrevalence of Diabetic Retinopathy
Bengt Jerneld, M.D., and Peep AIgvere, M.D.
In a population study of all registeredinsulin-treated diabetic patients on theSwedish island of Gotland, the prevalence ofdiabetic retinopathy was determined withophthalmoscopy, biomicroscopy, and colorphotography. Retinopathy was present in 173of 368 patients (470/"...) and reached a prevalenceof 100% after 30 years of diabetes. Proliferativeretinopathy was found in 48 subjects (13%)and was more common in females (17%) thanin males (9.4%) (P = .01). By simple logisticregression test, the prevalence of total andproliferative retinopathy was correlated withboth duration and age at onset of diabetes (P <.00l). However, on multiple regression analysis only the relationship with duration wasstatistically significant (P < .001); age at onsetwas not (P > .2). Age had an additional influence only on background retinopathy withhard exudates, which were more frequent inolder subjects (P < .01). Thus, age at onset ofdiabetes was not correlated with the prevalence of total or proliferative retinopathy.
POPULATION STUDIES indicate that the prevalence of diabetic retinopathy varies between24% and 70% in different countries.!" In Sweden, the prevalence of retinopathy in patientswith insulin-treated diabetes 40 years of age oryounger was found to be 66%9; in orally treateddiabetic patients of all ages it was 17%.10
The aim of this study was to determine theprevalence of retinopathy among insulintreated diabetic patients in a well-defined and
Accepted for publication June 25, 1986.From the Department of Ophthalmology, Karolinska
Institute and Hospital, Stockholm, Sweden. This studywas supported by grants from the Clas GroschinskyMemorial Foundation, the Carmen and Berti! RegnerFoundation, and Praktikertjanst, Inc., Stockholm, Sweden.
Reprint requests to Bengt [erneld, M.D., Departmentof Ophthalmology, Karolinska Hospital, Box 60500,10401 Stockholm, Sweden.
stable population. The area selected was theSwedish island of Gotland. From these data,which included all insulin-treated diabetic patients registered by census in this municipalarea, it was possible to study the relationshipof prevalence to the duration of diabetes and toage at onset in various age groups. Since thesetwo variables may influence each other, weused multiple logistic regression analysis.
SUbjects and Methods
Subjects-The study included 399 registeredinsulin-treated diabetic patients who were registered as permanent residents on the Swedishisland of Gotland. The total population was55,623 (December 1981), and thus the prevalence of insulin-treated diabetes was 0.72%,which is the average for Sweden. 11 The numberof diabetic patients was checked with the. reccords at the only Department of Internal Medicine (Visby County Hospital) and with theprescriptions for antidiabetic drugs received bythe five pharmacies on the island during asix-month period in 1981. A six-month list ofinsulin prescriptions is sufficient to traceinsulin-treated diabetic patients in a populatiori."
Fourteen of the patients were children underthe age of 15 years. All had normal ophthalmoscopic findings and they were not included inthe study. Ten patients died before examination, and six refused to participate. One subjecthad bilateral dense cataracts, and ultrasonography could not be performed. All nonparticipants (mean age, 75 years; mean duration ofdiabetes, 8.0 years) had adult-onset diabetes.
Figure 1 shows the age distribution of the 368participating patients (181 men and 187women) (mean age, ± S.D., 55 ± 19.2 years).Figure 2 shows the duration of diabetes (meanduration, 13 ± 9.5 years), and Figure 3 showsage at diagnosis. For comparison with otherstudies,S,13-l6 we divided the patients into two
©AMERICAN JOURNAL OF OPHTHALMOLOGY 102:431-437, OCTOBER, 1986 431
432 AMERICAN JOURNAL OF OPHTHALMOLOGY October, 1986
20
o 10 20 30 40 50 60 70 80 soAge at examination
Fig. 1 (Jerneld and Algvere). Distribution of age atexamination. The peak is at 60 to 75 years.
groups by age at onset of diabetes. The"younger" group of 171 subjects (onset of diabetes at 40 years of age or earlier) had a meanage of 38 ± 12.5 years (range, 15 to 64 years)and a mean duration of disease of 17 ± 11.2years (range, one to 50 years). The "older"group of 197 patients (onset of diabetes afterthe age of 40 years) had a mean age of 69 ± 9.6years (range, 44 to 93 years) and a mean duration of disease of 11 ± 6.6 years (range, one to29 years).
Methods-After dilation of their pupils withtropicamide 1% and phenylepinephrine 10%,the patients underwent ophthalmoscopy andbiomicroscopy with the Goldmann threemirror lens. Fundus photography was performed as recommended by the Diabetic Retinopathy Research Croup." modified to suit a45-degree fundus camera. We photographedseven areas of the fundus and projected thecolor slides onto a screen (magnification, x 8)for assessment, in a masked fashion, by twoindependent observers. All findings were recorded and fed into a computer for evaluation.The patients were classified according to themost severe changes found by any method inthe worst eye. The fundus changes were divided into three categories: (1) Background retinopa-
~ Examined
EEl All retinopathy
_ Proliferative rp.
60
40
-(IlQ.
...~ 80E::lZ
-o
5 10 20 30 40 50Duration
Fig. 2 (Jerneld and Algvere). Duration of diabetesin patients examined and in those with retinopathyand proliferative retinopathy.
thy, defined as the presence of one or more ofthe following signs: microaneurysms, punctateor striate intraretinal hemorrhages, hard exudates, and fewer than five soft exudates. Therewere two subgroups-those with and thosewithout hard exudates. (2) Preproliferative retinopathy, defined as five or more cotton-woolspots, venous beading, and intraretinal microvascular abnormalities.P-" (3) Proliferative retinopathy, subdivided into neovascularization onor within 1 disk diameter of the optic disk areain extent and neovascularization elsewhere inthe retina. 20,21
Eyes with fibrous proliferations or neovascular glaucoma or those enucleated because of
~
--
r-- r-~
f0- r-r-
r- f-fo-
fo-
f-
nI I I
20
10
30
'"Co
~ 50c.,
~ 40E::lZ
Vol. 102, No.4 Diabetic Retinopathy 433
Results
Diabetic retinopathy was present in 173 ofthe 368 patients examined (47%). Backgroundretinopathy was present in 112 subjects (31%),preproliferative retinopathy in 13 (3.5%), andproliferative retinopathy in 48 (13%).
The prevalence of retinopathy varied considerably between groups of different ages atonset of diabetes (P < .001). The highest prevalence (62%) was found among patients withonset at 20 or less years of age, and the lowest(21%) when diabetes was diagnosed after theage of 60 years (P < .001). The difference in theprevalence of proliferative retinopathy between patients with early (20 or less years ofage) and late (more than 60 years of age) onsetwas even larger, 28% vs 5% (P < .001) (Table 1).
Subjects with onset of diabetes at 40 or lessyears of age had a significantly higher prevalence of total retinopathy (56%) and of proliferative retinopathy (18%) than patients with lateronset (40% and 8.1%, respectively). With longer duration of diabetes there was a gradualincrease in the prevalence of both total retinopathy (P < .001) and proliferative retinopathy(P < .001) (Fig. 4).
There was a correlation between age at onsetand duration of diabetes (r = -0.44; P < .001).This required a multiple logistic regressionanalysis which, again, showed that the duration of diabetes was significantly correlated tototal (P < .001) and proliferative (P < .001)retinopathy. However, multiple logistic regression analysis did not correlate the age at onsetof diabetes to total retinopathy (P > .2) or toproliferative retinopathy (P > .2) when duration was taken into account. The difference inthe prevalence of retinopathy between thegroups of different ages at onset thus resultedfrom differences in the duration of diabetes,and age had no significant influence on retinopathy (Table 1).
There was an increasing prevalence of retinopathy with longer duration of diabetes in allfour age groups at onset of diabetes (Fig. 5).This was particularly notable for proliferativeretinopathy in all but the oldest age group(more than 60 years of age). The increase wasmost pronounced in patients with early onsetof diabetes (20 or less years of age), in whom
tic regression analysis should be consideredapproximations .,......,
r- -I-
-""""r-~
I-- I-
- I-
r-~
--
h
10
o ro w ~ ~ ~ ~ ro ~ ~
Age at onset
Fig. 3 (Jerneld and Algvere). Distribution of age atonset of diabetes. Two peaks can be distinguished,approximately between 10and 30 years and between50 and 65 years.
20
diabetic complications were assigned to thecategory of proliferative retinopathy. Eyes withpreretinal or intravitreal hemorrhages wereclassified according to their other funduschanges, except when combined with neovascularization on the optic disk of less than 25%of the optic disk diameter, which put them inGroup 3 according to Diabetic RetinopathyStudy." Eyes with media opacities that precluded inspection of the fundus were examinedby A- and B-scan ultrasonography. The ultrasonographic findings were categorized as vitreous bleeding or membranes, prepapillary proliferations, and retinal detachment.P Vitreoushemorrhages and membranes were classified asbackground retinopathy, and prepapillary orpreretinal proliferations or retinal detachmentas proliferative retinopathy. Eight eyes (of fivepatients) with photocoagulation scars wereclassified according to their current status.
The statistical evaluation was based on simple and multiple logistic regression analysis"with a variable indicating the presence or absence of retinopathy as the dependent variable,and age at examination, age at onset of diabetes, duration, and sex as the independent variables. In the logistic analysis, we used theindependent variables with their numeric values; in the illustrations and tables they wereclassified. The P values obtained from the logis-
..c: 40.!!:;a.'0~...E 30:IZ
434 AMERICAN JOURNAL OF OPHTHALMOLOGY October, 1986
Duration (years)Fig. 4 (Jerneld and Algvere). Prevalence of retinop
athy and duration of diabetes. Solid bars show proliferative retinopathy. After 30 years, retinopathy wasfound in 100% of patients and proliferative retinopathy in 65%; n, number of subjects in each durationgroup.
the prevalence of proliferative retinopathyreached 55% after a duration of more than 20years. In subjects with onset of diabetes between 41 and 60 years, this prevalence was 38%but the difference was not statistically significant (P>.05). Within the three groups ofpatients with different durations of diabetes,
100X
80
60
40
20
10
n=169
::::::: 19
::::::: 2
0-10
120
:::::::: 10
11-20
59
21-30
20
111111" '00
::::::: 65
31-
there was no significant change in the prevalence of retinopathy with increasing age atonset of diabetes (Fig. 5). The higher prevalence of retinopathy in those with durations ofzero to ten years and with the onset of diabetesat 41 to 60 years could have resulted fromchance.
There was no significant sex difference in theprevalence of total retinopathy. However, proliferative retinopathy was present in 31 of 187women (17%) and 17 of 181 men (9.4%) (P <.05), and this difference became even moresignificant when the duration was consideredin the analysis (P = .01).
Hard exudates were found in 87 patients(24%). The prevalence of hard exudates wasclearly related to the duration of diabetes (P <.001), and increased conspicuously after a duration of ten years (Table 2). There was no sexdifference. In contrast to total retinopathy, inpatients with hard exudates at examination ageprovided statistically significant additional information (P < .01). In the total group, age wasnot correlated to the duration of diabetes (r =
0.01), and a simple listing of the various agegroups therefore yielded a correct estimate ofthe additional (predictive) value of the age atexamination when the duration was given(Table 2). In subjects with retinopathy, the ratiobetween those with and those without hardexudates was 1:2 when diabetes was diagnosedat 20 or less years of age but 2:1 when diabeteswas diagnosed after the age of 60 years.
Generally, the status of the fundus in felloweyes was in good accordance. In 311 of 350subjects (89%) both eyes of each individual hadthe same retinal status; the inclusion of thosewith a difference of one degree in severity ofretinopathy increased this to 335 of 350 patients(96%).
TABLE 1PREVALENCE OF RETINOPATHY, AGE AT ONSET, AND DURATION OF DIABETES
NO. WITH RETINOPATHY BY AGE AT ONSET (YRS)
DURATION OF DIABETES (YRS) "'20 21 to 40 41 to 60 «61 T,OTAL (%)
o to 5 1 of 13 2 of 20 4 of 22 3 of 29 12
6 to 10 3 of 13 4 of 20 12 of 30 3 of 22 26
11 to 15 4 of 8 6 of 11 15 of 31 6 of 21 44
16 to 20 6 of 9 13 of 16 17 of 22 2 of 2 78
21 or more 39 of 42 17 of 19 15 of 16 1 of 2 91
Mean (± S.D.) duration 19.7 ±12.5 13.6 ± 8.8 12.5 ± 7.0 7.7 ± 4.8 13.4 ± 9.5
Prevalence (%)
Total retinopathy 62 49 52 21 47
Proliferative retinopathy 28 9 10 5 13
Vol. 102, No.4
Age0-20
at onset21-40 41-60
Diabetic Retinopathy
61-
100
435
=I=
n=26
=NI-- n =17
40
27
52
53
51
23
50
o
100
50
o
Fig. 5 (Jerneld and Algvere). Prevalence ofretinopathy. Number of patients, expressedas % of subjects (n) in each age group atonset (years) and duration of diabetes(years). Solid bars show proliferative retinopathy. There was a uniform increase inretinopathy with longer duration (readingvertically). Different ages at onset of diabetes yielded no significant differences inprevalence of retinopathy in various duration groups (depicted horizontally).
eo..C'O~ I:I _
Q N
100
50
c..e..o 0.
n=42 19 16 2
In 394 normal eyes, the fellow eye was normal in 378 (96%) and proliferative retinopathywas present in four (1%). The probability thatboth eyes of each individual would have thesame retinal status was higher when one eyewas normal or had only slight retinopathy(Table 3). Among 62 eyes with proliferativeretinopathy, 42 (68%) had the same conditionin the fellow eye, five (8%) had preproliferative
TABLE2PREVALENCE OF HARD EXUDATES
WITH
EXUDATESNO. OF
CLINICAL DATA PATIENTS NO. %
Diabetes duration (yrs)
o to 10 169 15 911 to 20 120 42 3521 and more 79 30 38
Age at examination (yrs)
<29 46 4 930 to 49 96 22 2350 to 69 138 33 2470 and older 98 28 29
and 11 (18%) had background retinopathy, andfour (6%) had a normal fellow eye.
Twenty-three patients had opaque ocularmedia that precluded examination of the fundus (bilaterally in five subjects) and were examined by ultrasonography. Twelve had changescompatible with proliferative retinopathy; inseven the fellow eye displayed clear media andproliferative retinopathy. Five subjects showedultrasonographic evidence of vitreous hemorrhage or membranes and were classified ashaving background retinopathy. Six patientshad normal ultrasonograms and were classified
TABLE 3AGREEMENT BETWEEN FELLOW EYES
FELLOW EYE (%)TYPE OF PROLIFERATIVE
RETINOPATHY BETTER SAME WORSE RETINOPATHY'
Normal 0 96 4 1Background 6 86 8 5Preproliferative 29 50 21 21Proliferative 32 68 0 68
'Cases of proliferative retinopathy included in the "same" and
"worse" categories.
436 AMERICAN JOURNAL OF OPHTHALMOLOGY October, 1986
as normals (the fellow eyes had clear media andno retinopathy).
Discussion
The prevalence of diabetic retinopathy wasdetermined in a well-defined and stable population. The rate of migration turnover of thepopulation was only 3.4%.24 The use of a double control (hospital records and the prescriptions received by the pharmacies) made it likelythat all insulin-treated diabetic patients weretraced. 12 Only 17 of 385 patients (4.4%) did notparticipate, and these were elderly people(mean age, 75 years).
The prevalence of retinopathy was 47% andthat of proliferative retinopathy 13%. In a previous study of the same population," allinsulin-treated patients with onset at 40 or lessyears of age were examined by fluoresceinangiography, and the prevalence of retinopathy was found to be 66% and that of proliferative retinopathy was 18%. The correspondingfigures in this study were 56% and 18%, respectively. Thus, fluorescein angiography enhanced the diagnosis of background retinopathy but not that of proliferative retinopathy.
Few epidemiologic studies on the prevalenceof diabetic retinopathy have been published.Reports from Australia.':" Denmark." Iceland,"and the United States"? showed large variationsin the prevalence of total (24% to 70%) andproliferative (4.5% to 22%) retinopathy, partlybecause they were not performed under thesame conditions. In addition, considerable differences exist as to the nonparticipating fraction of the study population (up to 20%), theannual migration rate of the population, theage distribution of the patients, dissimilaritiesin the treatment of diabetes, and genetic variations affecting the diabetes prevalence of thepopulation (0.12% to 0.72%).
When we divided our study population intogroups of different ages at onset of diabetes,the highest prevalence of retinopathy wasamong patients with diabetes onset at 20 or lessyears of age. This was statistically significantby a simple logistic regression test. However,this test also yielded significant differenceswhen retinopathy was correlated to the duration of diabetes. Patients with early onset ofdiabetes had also had diabetes the longest(Table 1). Obviously, there is a relationshipbetween age at onset and duration of diabetes(r = -0.44). Consequently, studying one ofthese variables requires the other to be taken
into consideration. By multiple logistic regression analysis it appears that the differences inthe prevalence of retinopathy with differentages at onset of diabetes were not statisticallysignificant when duration was taken into account (P > .2). However, when the relationshipbetween retinopathy and duration was analyzed controlling for age at onset, for a higherprevalence of retinopathy was still found to berelated to a longer duration (P < .001). Thus,the duration of diabetes was the predominantfactor determining the prevalence of both totaland proliferative retinopathy, and age at onsetwas subordinate. These results support thereport of Constable and associates."
Among the groups of different ages at onsetof diabetes, there were some variations in theprevalence of retinopathy. When diabetes duration was less than ten years, the prevalenceof retinopathy was somewhat higher in patients with onset between 41 and 60 years thanin other groups (Fig. 5). Such a tendency wasalso noted by Aiello and associates." whofound that in 67% of patients with backgroundretinopathy and a diabetes duration of lessthan ten years, the onset of diabetes was afterthe age of 40 years. This group probably included non-insulin-dependent patients whose realduration of diabetes may have been much longer than the diagnosed one. Kahn and Bradley"also suggested that age is correlated to retinopathy only when the duration of diabetes is lessthan ten years.
In the long-term, however, there is a uniformincrease in the prevalence of retinopathy regardless of the age at onset of diabetes (exceptafter the age of 60 years). All patients withjuvenile-onset diabetes presumably have insulin-dependent disease, whereas those withonsets between 41 and 60 years do not. Nevertheless, after a known duration of 11 years ormore, both groups exhibited the same prevalence of retinopathy. The highest prevalence ofproliferative retinopathy (55%) developed inthe group who were youngest at onset and whohad the longest duration of diabetes (19.7years). This observation was in accord withprevious studies." The results for patients withonsets after the age of 60 years were not conclusive; among these, few subjects had a longduration of diabetes and cataracts and senilitymade examination difficult.
Retinopathy with hard exudates correlated toduration of diabetes and the age of the patients. Hard exudates occurred more often inthose over 60 years of age, a finding also notedby Klein and associates." This indicated thatfactors other than the duration of diabetes also
Vol. 102, No.4 Diabetic Retinopathy 437
influence retinal disease. The pathogeneticmechanisms are still speculative. A correlationbetween a higher prevalence of diabetic maculopathy and adult-onset diabetes has been reported. 5,13,27 This was reflected in our study asthe prevalence of hard exudates.
References
1. Heriot, W, J., Borger, J. P., Zimrnet, P., King,H., Taylor, R., and Raper, 1. R.: Diabetic retinopathyin a natural population. Aust. J. Ophthalmol. 11:175,1983.
2. Danielsen, R., [onasson, F., and Helgascn, T.:Prevalence of retinopathy and proteinuria in type 1diabetics in Iceland. Acta Med. Scand. 212:277, 1982.
3. Constable, I. J., Knuiman, M. W., Welborn,T. A., Cooper, R. 1., Stanton, K. M., McCann, V. J.,and Grose, G. c.: Assessing the risk of diabeticretinopathy. Am. J. Ophthalmol. 97:53, 1984.
4. Nielsen, N. V.: Diabetic retinopathy. I. Thecourse of retinopathy in insulin-treated diabetics. Aone year epidemiological cohort study of diabetesmellitus. The island of Falster, Denmark. Acta Ophthalmol. 62:256, 1984.
5. Klein, R., Davis, M. D., Moss, S. E., Klein, B.,and DeMets, D. 1.: The Wisconsin epidemiologicstudy of diabetic retinopathy. A comparison of retinopathy in younger and older onset diabetic persons. In Vranic, M., Hollenberg, C. H., and Steiner,G. (eds.): Comparison of Type I and II Diabetes. NewYork, Plenum Press, 1985, pp. 321-335.
6. Klein, R., Klein, B. E. K., Moss, S. E., Davis,M. D., and DeMets, D. 1.: The Wisconsin epidemiologic study of diabetic retinopathy. II. Prevalenceand risk of diabetic retinopathy when age at diagnosis is less than 30 years. Arch. Ophthalmol. 102:520,1984.
7. : The Wisconsin epidemiologic study ofdiabetic retinopathy. III. Prevalence and risk of diabetic retinopathy when age at diagnosis is 30 or moreyears. Arch. Ophthalmol. 102:527, 1984.
8. Mitchell, P.: The prevalence of diabetic retinopathy. A study of 1300 diabetics from Newcastle andHunter Valley. Aust. J. Ophthalmol. 8:241, 1980.
9. [erneld, B., and Algvere, P.: The prevalence ofretinopathy in juvenile-onset diabetes mellitus. Afluorescein angiographic study. Acta Ophthalmol.62:617, 1984.
10. : Prevalence of retinopathy in diabetestreated with oral antihyperglycaemic agents. ActaOphthalmol. 63:535, 1985.
11. Bergman, U.: Utilization of antidiabetic drugsin the island of Gotland, Sweden. Agreement between wholesale figures and prescription data. Eur.J. Clin. Pharmacol. 14:213, 1978.
12. Green, A., Hauge, M., Holm, N. V., andRasch, 1.: Epidemiological studies of diabetes mellitus in Denmark. II. A prevalence study based oninsulin prescriptions. Diabetologia 20:468, 1981.
13. Aiello, 1. M., Rand, 1. I., Briones, J. c..Wafai, M. Z., and Sebestyen, J. G.: Diabetic retinopathy in Joslin clinic patients with adult-onset diabetes. Ophthalmology 88:619,1981.
14. Sigelman, J.: Diabetic macular edema injuvenile- and adult-onset diabetes. Am. J. Ophthalmol. 90:287, 1980.
15. Soler, N. G., Fritzgerald, M., Malins, J., andSummers, R.: Retinopathy at diagnosis of diabetes,with special reference to patients under 40 years ofage. Br. Med. J. 3:567, 1969.
16. Palmberg, P., Smith, M., Waltman, S., Krupin,T., Singer, P., Burgess, D., Wendtlant, T.,Achtenberg, J., Cryer, P., Santiago, J., White, N.,Kilo, C.; and Daughaday, W.: The natural history ofretinopathy in insulin-dependent juvenile-onset diabetes. Ophthalmology 88:613, 1981.
17. Diabetic Retinopathy Study Research Group:Manual of Operations. Baltimore, Diabetic Retinopathy Coordinating Center, University of Maryland,1972, ch. 9.
18. Berkow, J. W.: Diabetic retinopathy. Int. Ophthalmol. Clin. 17:89, 1977.
19. Murphy, R. P., and Patz, A.: The natural history and management of non-proliferative diabeticretinopathy. In Little, H. 1., Patz, A., Jack, R. 1., andForsham P. H. (eds.): Diabetic Retinopathy. NewYork, Thieme-Stratton, 1983, pp. 225-241.
20. Diabetic Retinopathy Study Research Group:Photocoagulation treatment of proliferative diabeticretinopathy. The second report of Diabetic Retinopathy Study findings. Ophthalmology 85:82, 1978.
21. : Photocoagulation treatment of pro-liferative diabetic retinopathy. Clinical application ofDiabetic Retinopathy Study (DRS) findings. DRS Report No.8. Ophthalmology 88:583, 1981.
22. Jerneld, B., Algvere, P., and Singh, G.: Anultrasonographic study of diabetic vitreo-retinal disease with low visual acuity. Acta Ophthalmol.58:193,1980.
23. Cox, D. R.: Analysis of binary data. London,Methuen, 1970, pp. 1-142.
24. Official Statistics of Sweden. Summary of VitalStatistics 1979-1981 by county. Stockholm, Statistiska Centralbyran, 1982, vol. 1, pp. 122-124.
25. Kahn, H. A., and Bradley, R. F.: Prevalence ofdiabetic retinopathy. Br. J. Ophthalmol. 59:345,1975.
26. Klein, R., Klein, B. E. K., Moss, S. E., Davis,M. D., and DeMets, D. 1.: The Wisconsin epidemiologic study of diabetic retinopathy. IV. Diabetic macular edema. Ophthalmology 91:1464, 1984.
27. Kohner, E. M.: The evolution and natural history of diabetic retinopathy. Int. Ophthalmol. Clin.18:1, 1978.