Proteinuria and Blood Glucose Levels in a Population With Diabetic Retinopathy

B e n g t J e r n e l d , M . D . , a n d P e e p A l g v e r e , M . D .

In a populat ion-based study of all insul in-treated diabetic patients on the Swedi sh is land of Gotland, we compared the prevalence and severity of retinopathy with those of nephrop-athy as measured by proteinuria and serum creatinine leve ls . Of 365 diabetic patients, 66 (18%) had proteinuria. Of these 66, 39 (59%) had retinopathy. Proteinuria and serum creati­nine correlated wi th increasing severity of retinopathy (P < .001). Of 47 patients with proliferative retinopathy, 19 (40%) had pro­teinuria. Of 124 patients with retinopathy of other grades of severity, 20 (16%) had protein­uria. Visual acuity in the best eye was nega­tively correlated to proteinuria, which was present in 17 of 203 (8.4%) patients with a visual acuity of 20/20, compared with eight of 15 (53%) of those wi th a visual acuity of 20/200 or less . Blood g lucose , determined two hours postprandially, was satisfactory (<10 mmol/1) in 162 patients (44%), unsatisfactory (10 to 14 mmol/1) in 89 (24%), and poorly regulated (>14 mmol/1) in 114 (31%). Increasing mean blood glucose correlated to retinopathy (P < .05).

N E P H R O P A T H Y has been repor ted in 47% to 60% of pa t i en t s wi th d iabe tes mel l i tus w h o have advanced proliferat ive r e t i nopa thy . 1 2 Pro­teinuria is a risk factor for diabetic r e t i nopa thy , since the preva lence of proliferat ive re t inopa­thy in juveni le -onse t i n s u l i n - d e p e n d e n t d iabe­tes is cons iderably h ighe r a m o n g pa t i en t s wi th as compared to those w i t h o u t diabetic ne­p h r o p a t h y . 3 Diabetic r e t i nopa thy is bel ieved to

Accepted for publication June 5, 1987. From the Department of Ophthalmology, Karolinska

Institute and Hospital, Stockholm, Sweden. This study was supported by grants from the Clas Groschinsky Memorial Foundation, the Carmen and Bertil Regner Foundation, and Praktikertjanst, Inc., Stockholm.

Reprint requests to Bengt Jerneld, M.D., Department of Ophthalmology, Karolinska Hospital, Box 60500, 104 01 Stockholm, Sweden.

be more c o m m o n t h a n clinical n e p h r o p a t h y , a l t h o u g h there are few recent popu la t ion-based s tud ies on the correla t ion of these two condi­t ions . 4

We per fo rmed a s tudy of the re la t ionship b e t w e e n n e p h r o p a t h y , as d e t e r m i n e d by pro­te inur ia a n d s e r u m crea t in ine levels, a n d dia­betic r e t i nopa thy in a popu l a t i on compr i s ing all insu l in - t rea ted diabet ics on the Swedish is land of Got land . Blood glucose va lues de t e rmined two h o u r s pos tp rand ia l ly were also found to be corre la ted to the sever i ty of r e t inopa thy . Be­cause the preva lence of diabetic r e t inopa thy w a s prev ious ly d e t e r m i n e d in this popula t ion , 5

t he corre la t ion to n e p h r o p a t h y was feasible.

Patients and Methods

There were 399 insu l in- t rea ted diabetics census - reg i s te red in Got land (popula t ion 55,623, Dec. 31 , 1981) d u r i n g the observa t ion per iod from 1981 to 1982. The m e t h o d s for regis t ra t ion and tracing these pa t i en t s have b e e n descr ibed previous ly . 5 Before t r ea tment , all 399 pa t i en t s wi th a d iagnos i s of diabetes mell i tus had had e i ther a fast ing glucose value of > 7 mmol/1 or of s l f j mmol/1 in capil lary w h o le blood tes ted r a n d o m l y d u r i n g the day . W h e n the resul t s were a m b i g u o u s , glucose tol­e rance tes ts were pe r fo rmed; a v e n o u s whole blood glucose va lue of &10 mmol/1 two hour s after oral admin i s t r a t ion of 75 g of glucose was cons ide red pathologic . 6

Of the 399 insu l in - t rea ted pa t i en t s , 14 were ch i ld ren u n d e r the age of 15 years w h o had a normal f u n d u s and normal visual acuity. They were not inc luded in the s tudy . Ten pa t ien ts d ied before examina t ion and six refused to par t ic ipa te . O n e pa t ien t had a d e n s e cataract a n d a d iagnos i s by ocular u l t r a s o n o g r a p h y could not be m a d e . Three o ther pa t ien t s did not u n d e r g o labora tory tes ts . Therefore , the labo­ratory test resul ts were correla ted wi th the

©AMERICAN JOURNAL OF OPHTHALMOLOGY 104:283-289, SEPTEMBER, 1987 283

284 AMERICAN JOURNAL OF OPHTHALMOLOGY September, 1987

severity of retinopathy in 365 patients. For correlation with visual acuity, 361 patients were accessible.7

The mean age of the 365 insulin-treated pa­tients was 54 years (range, 15 to 93 years; S.D., 19.1 years) and the duration of diabetes was 13 years (range, one to 50 years; S.D., 9.5 years). Of these patients, retinopathy and proliferative retinopathy were present in 171 (47%) and 47 (13%), respectively (Table 1).

Five laboratory tests were performed: serum creatinine; blood glucose two hours postpran-dially; and urinalysis for glucose, protein, and acetoacetate. The latter three were determined with Labstix. The reagent for the protein analy­sis was tetrabromphenol blue. Most patients were examined before noon. The following classes of protein excretion were determined on the stick: trace, 0.3, 1.0, 3.0, and <20 g/1. A trace was not considered as proteinuria. Pro-teinuria was classified into two groups: mod­erate (0.3 to <1.0 g/1) and heavy (>1.0 g/1). The levels of blood glucose were divided into three groups: satisfactory (<10 mmol/1), unsat-

TABLE 1 RESULTS OF LABORATORY TESTS AND PREVALENCE

OF RETINOPATHY (N = 365)

CHARACTERISTIC

Proteinuria None Moderate (0.3 to<1.0g/l) Heavy (> 1.0 g/l)

Serum creatinine < 133 jjimol/l > 133 (imol/l

Ketoacidosis Absent Present

Blood glucose < 10 mmol/l 10 to 14 mmol/l > 14 mmol/l

Urine glucose Absent Present

Retinopathy None Background Exudative Preproliferative Proliferative

NO.

299 43 23

333 32

349 16

162 89

114

136 229

194 61 50 13 47

%

82 12 6

91 9

96 4

44 24 31

37 63

53 17 14 4

13

isfactory (10 to 14 mmol/l), and poor control (>14 mmol/l). A serum creatinine value of a l 3 3 u.mol/1 was considered pathologic.

Best corrected visual acuity was measured at a distance of 5 m on a Monoyer-Granstrom decimal chart. After dilating each pupil, oph-thalmoscopy and biomicroscopy using a Gold-mann three-mirror lens were performed. Fun-dus photographs covering the retina in seven different areas were taken as described by the Diabetic Retinopathy Study Research Group,8

except that the procedure was modified to ac­commodate a 45-degree fundus camera. Eyes with media opacities obscuring the fundus were examined by A- and B-scan ultrasonog-raphy, as described previously.6,9 The eyes were classified according to the most severe lesions found by any method. Background reti­nopathy was defined as the presence of one or more of the following signs: microaneurysms, punctate or striate intraretinal hemorrhages, and fewer than five soft exudates. Exudative retinopathy was defined as any of the above signs combined with hard exudates. Pre­proliferative retinopathy was defined as five or more cotton-wool spots, venous headings, and intraretinal microvascular abnormalities.1011

Proliferative retinopathy was defined as neo-vascularization on the optic disk or in the retina, equalling or exceeding one fourth of the disk area.1213

Eyes with fibrous proliferations or neovascu-lar glaucoma, and those enucleated because of diabetic complications, were assigned to the category of proliferative retinopathy. Eyes with preretinal or intravitreal bleeding were classi­fied according to their other fundus lesions, except when the bleeding was combined with neovascularization on the optic disk covering less than one fourth of the disk diameter, in which case they were put into the pre­proliferative retinopathy group.13

The degree of retinopathy in each patient was classified according to the most severe fundus lesions found in either eye. Visual acu­ity was recorded in the better eye. The statisti­cal evaluation was based'on a one-way analysis of variance (F test) and simple and multiple logistic regression analysis.

Results

Proteinuria was present in 66 of 365 patients (18%). Moderate proteinuria (0.3 to <1.0 g/1)

Vol. 104, No. 3 Diabetic Retinopathy, Proteinuria, and Blood Glucose 285

was found in 43 of 365 subjects (12%) and heavy proteinuria (>1 g/1) in 23 patients (6.3%). Impaired renal function as measured by serum creatinine ( s l 3 3 u.moI/1) was noted in 32 pa­tients (9%). No sex difference was found in the mean serum creatinine value or in the group of patients with impaired renal function. Ketoaci-dosis was demonstrated in 16 patients (4.4%), but no correlation to retinopathy was found (Table 1).

The renal function as evaluated by the serum creatinine correlated to the severity of retinop­athy (P < .001) (Fig. 1). With deteriorating retinopathy, the mean serum creatinine con­centration increased in both males and females.

Retinopathy was present in 39 of 66 patients with proteinuria (59%). The percentage of pa­tients with retinopathy increased with higher

^ 170 "o a. 160 4> C E 150 ra 0) o 140

5 130

120

110

100

90

80-j

70

60

TABLE 2 CROSS CLASSIFICATION OF PROTEINURIA AND

RETINOPATHY

— i — 0

194 61

3 4 Ret inopathy

50 13 47 Fig. 1 (Jerneld and Algvere). Serum creatinine

(mean ± S.D.) in groups with different degrees of retinopathy. 0, no retinopathy; 1, background retinopathy; 2, exudative retinopathy; 3, pre-proliferative retinopathy; 4, proliferative retinopa­thy.

PROTEINURIA

None Moderate Heavy

Total

NONE

NO. (%)

167 (56) 19(44) 8(35)

194

RETINOPATHY

BACKGROUND TO PREPROLIFERATIVE

NO.

104 17 3

124

(%) (35) (40) (13)

PROLIFERATIVE

NO.

28 7

12 47

(%) (9)

(16) (52)

TOTAL

299 43 23

365

levels of proteinuria (P < .05). This trend seemed to be caused by the increasing propor­tion of proliferative retinopathy in proteinuric patients (P<.001) (Table 2).

Unlike other laboratory variables, protein­uria showed a correlation to the duration of diabetes (r = 0.14). The duration of the disease had a significant impact on the severity of the retinopathy.5 When duration was factored into the multivariate analysis, the correlation be­tween proteinuria and proliferative retinopa­thy was still significant (P<.01). Furthermore, when the patients with different degrees of proteinuria were grouped according to the du­ration of their diabetes ( s 20 years and >20 years), the correlation was still significant (P < .05 and P < .001, respectively) (Fig. 2). More women (15 of 185, 8.1%) than men (eight of 180, 4.4%) had heavy proteinuria (>1 g/1), but this difference was not statistically significant (P = .15).

By simple logistic regression analysis, visual acuity of the better eye was found to be signifi­cantly correlated to proteinuria (P < .001). Generally, decreasing visual acuity was associ­ated with a higher prevalence of proteinuria. For example, proteinuria was present in 17 of 203 patients (8.4%) with a visual acuity of 20/20, but was found in eight of 15 (53%) of those whose visual acuity was less than 20/200. Of 64 patients with proteinuria, 16 (25%) had visual acuity of 20/100 or less (Table 3). Serum creatinine correlated to visual acuity in the same way (P < .001). When the degree of retinopathy was taken into account in the mul­tivariate analysis, the relationship became non­significant.

Blood glucose was determined two hours postprandially. The glucose levels were satis­factory (<10 mmol/1) in 162 of 365 patients (44%) and poorly controlled (>14 mmol/1) in

286 AMERICAN JOURNAL OF OPHTHALMOLOGY September, 1987

100

5 0

c 0) u Q.

100

Duration

241

F771

<20 years

3 2

i '

14

I %

5 0

c 0) u 0) Q.

0.3 -<1

Duration >20 years

11

>1

58

m W

m

P

I #

0 0.3 -< 1 it Protein g / l

Fig. 2 (Jerneld and Algvere). Percentage of patients with proliferative retinopathy in groups with differ­ent degrees of proteinuria. With heavy proteinuria (>1 g/l) and duration of diabetes £20 years (top) retinopathy was present in five of 14 patients (36%); at >20 years, seven of nine patients (78%) had retinopathy (bottom).

114 (31%) (Table 1). There were no differences in age, duration of diabetes, or age at onset between these groups. Statistically significant differences in blood glucose levels were found between various retinopathy groups (P < .05). Increasing grades of severity of retinopathy correlated with higher mean values of blood glucose, except in patients with proliferative retinopathy.

Multiple regression analysis confirmed the correlation between higher blood glucose and a more severe stage of retinopathy, which was independent of the duration of diabetes. Pa­tients with proliferative retinopathy had the same mean blood glucose level (11.5 mmol/1) as those without retinopathy (11.3 mmol/1), but

TABLE 3 VISUAL ACUITY AND PROTEINURIA IN 361 DIABETICS

PROTEINURIA

ACUITY

20/20 20/25 to 20/40 20/50 to 20/60 20/100 to 20/200 < 20/200

NO.

17 26 5 8 8

(%)

(8) (24) (33) (44) (53)

TOTAL

203 110 15 18 15

this was still considered high and unsatisfac­tory (Fig. 3).

Glucose in the urine was found in 229 pa­tients (63%). Differences in the amounts of urinary glucose were observed between pa­tients with different stages of retinopathy (P < .05). In contrast to serum creatinine and pro­teinuria, glucose levels in the blood and urine showed no significant correlation to visual acu­ity.

Discussion

In this population-based study comprising all insulin-treated diabetics in a stable and well-defined area proteinuria was found in 18% and retinopathy in 47% of all patients. These results may be compared with those of a large multina­tional study comprising insulin-treated diabet­ics (aged 34 to 56 years), in which proteinuria was observed in 29% and retinopathy in 54%."

Thus, most diabetics with retinopathy, even of the proliferative type, lack proteinuria as determined by the dipstick test. However, when present, proteinuria was the systemic variable most significantly correlated to reti­nopathy (P < .001). Among patients with heavy proteinuria, a majority (65%) had reti­nopathy and 52% proliferative retinopathy. These findings are in accordance with those of another population-based study of American Indians by West and cowbrkers,4 who reported that 47% of patients with heavy proteinuria had retinopathy. A considerable proportion of the diabetic patients with proteinuria in our study had no retinopathy (41%), as was also found in a multinational study of patients se­lected from several diabetic clinics.15

Of 47 patients with proliferative retinopathy, 19 (40%) had proteinuria, which is an indicator

Vol. 104, No. 3 Diabetic Retinopathy, Proteinuria, and Blood Glucose 287

\ 25 n o E E CO o u 3

I 00

2 0

15

10

r 1 1 1 1 0 1 2 3 4

Retinopathy n 194 61 50 13 47

Fig. 3 (Jerneld and Algvere). Blood glucose (mean ± S.D.) in groups with different grades of retinopa­thy. 0, no retinopathy; 1, background retinopathy; 2, exudative retinopathy; 3, preproliferative retinopa­thy; 4, proliferative retinopathy.

of vascular damage to the kidney. More sensi­tive tests for detecting microalbuminuria16

would probably yield a higher correlation. The discrepancy between the microvascular dam­age of the retina and that of the kidney might be partly explained by local events inside the eye. The presence of pathologic vitreoretinal adhesions resulting in traction on retinal ves­sels, with consequent vitreous hemorrhage or retinal detachment, will dramatically convert a moderate retinopathy to a severe proliferative stage.1718 Conversely, some factors such as sub­stantial myopia,1920 conditions associated with reduced arterial pressure (as observed by oph-thalmodynamometry),21 and increased intraoc­ular pressure22 will probably retard the progres­sion of retinopathy. Additionally, genetic constitution cannot be ruled out as a modifying factor for the development of retinopathy under given levels of blood glucose and pro­teinuria.

Since there is a strong correlation between

proteinuria and proliferative retinopathy, there is an association between proteinuria and visu­al impairment, which is also correlated to pro­liferative retinopathy.5 Patients with prolifera­tive retinopathy show great variations in visual acuity.7 However, when we compared the visu­al acuity in such patients with and without pro­teinuria, no significant difference was found. Increasing proteinuria is generally accompa­nied by a deteriorating retinal status and even­tually by decreasing visual acuity. These pa­tients run a greater risk of developing proliferative retinopathy. Hence, proteinuria indicates a need for careful follow-up.

Although we tested blood glucose levels on only one occasion (two hours postprandially), the results indicate that patients with retinopa­thy have a significantly higher mean blood glucose level than those without (P < .05). Increasing degrees of severity of retinopathy were associated with rising mean values of blood glucose, except in the group with prolif­erative retinopathy. For example, in patients who had background retinopathy significantly higher mean blood glucose levels were found in those with hard exudates than those without.

Since blood glucose tests had not been per­formed previously in association with ocular examination, bias from the patients' knowledge of the test, and consequent intensification of control, was eliminated. Ideal testing of blood glucose levels should consist of regular deter­minations of glycosylated hemoglobin.23 It was not possible to measure HbAlC at the time of our examination. However, Constable and as­sociates24 pointed out that HbAlC is, in a popu­lation, highly correlated to nonfasting plasma glucose. Even random plasma glucose values were found to show significant correlation to retinopathy, although at a somewhat lower statistical level than HbAlC.25

A correlation between high blood glucose (frequently >11 mmol/1) and proliferative reti­nopathy has been reported by Rand and associ­ates.2 Most of their patients had attended an educational program on diabetic care, which motivated the subjects to ensure careful treat­ment from the early stages of diabetes; this motivation for treatment would thus have af­fected the whole case material equally. In Got­land no such educational program had been started at the time of this study. However, patients with proliferative retinopathy were probably more interested in careful treatment, since they were affected by a serious complica-

288 AMERICAN JOURNAL OF OPHTHALMOLOGY September, 1987

t ion tha t had resu l ted in pho tocoagu la t ion in some eyes a n d in visual i m p a i r m e n t (=£20/100 in the be t te r eye) in 46%.7 This hypo thes i s is s u p p o r t e d by the resu l t s of a popu la t i on -based s t u d y by Klein and colleagues,2 6 w h e r e only one th i rd of the diabet ics w i th b a c k g r o u n d r e t inopa thy k n e w of their ret inal d i sease as c o m p a r e d wi th 7 1 % of those wi th proliferat ive r e t inopa thy .

The blood glucose test in ou r s t u d y shou ld be r ega rded as incomple te , since only one deter ­mina t ion was per formed in each pa t ien t . Neve r the l e s s , t he re w a s a significant correla­t ion be tween the severi ty of r e t i nopa thy (ex­cept the proliferative re t inopa thy) and the blood glucose level two h o u r s pos tp rand ia l ly .

These resul ts from our popu la t ion -based s tudy a d d to the accumula t ing ev idence tha t a re la t ionsh ip exists be tween the d e g r e e of glycemia a n d r e t inopa thy , bo th in insul in-dependen t 2 7 3 0 a n d n o n i n s u l i n - d e p e n d e n t dia­betes.2 5 3 1

References

1. Stein, G., Kresse, S., Dietze, U., and Deufrains, A.: Nierenveranderungen bei Patienten mit pro-liferativer diabetischer Retinopathie. Z. Gesamte Inn. Med. 39:309, 1984.

2. Rand, L. I., Prudhomme, G. J., Ederer, F., and Canner, P. L., and the Diabetic Retinopathy Study Research Group: Factors influencing the develop­ment of visual loss in advanced diabetic retinopathy. Invest. Ophthalmol. Vis. Sci. 26:983, 1985.

3. Deckert, T., Parving, H. H., Andersen, A. R., Sandahl Christiansen, J., Oxenboll, B., Aaby Svendsen, P., Telmer, S., Christy, M., Lauritzen, T., Frokjaer Thomsen, O., Kreiner, S., Andersen, J. R., Binder, C , and Nerup, J.: Diabetic nephropathy. In Eschwege, E. (ed.): Diabetic Epidemiology. Amster­dam, Elsevier Biomedical Press, 1982, pp. 235-243.

4. West, K. M., Erdreich, L. J., and Stober, J. A.: A detailed study of risk factors for retinopathy and nephropathy in diabetes. Diabetes 29:501, 1980.

5. Jerneld, B., and Algvere, P.: Relationship of duration and onset of diabetes to prevalence of diabetic retinopathy. Am. J. Ophthalmol. 102:431, 1986.

6. WHO Expert Committee on Diabetes Mellitus. WHO Tech. Rep. Ser. 646:9, 1980.

7. Jerneld, B., and Algvere, P.: Visual acuity in a diabetic population. Acta Ophthalmol. 65:170, 1987.

8. Diabetic Retinopathy Study Research Group: Manual of operations. Baltimore, Diabetic Retinopa­thy Coordinating Center, University of Maryland, 1972, chap. 9.

9. Jerneld, B., and Algvere, P.: The prevalence of retinopathy in insulin-dependent juvenile-onset dia­

betes mellitus. A fluorescein-angiographic study. Acta Ophthalmol. 62:617, 1984.

10. Berkow, J. W.: Diabetic retinopathy. Int. Oph­thalmol. Clin. 17:89, 1977.

11. Murphy, R. P., and Patz, A.: The natural history and management of non-proliferative diabet­ic retinopathy. In Little, H. L., Patz, A., Jack, R. L., and Forsham, P. H. (eds.): Diabetic Retinopathy. New York, Thieme-Stratton, 1983, pp. 225-241.

12. Diabetic Retinopathy Study Research Group: Photocoagulation treatment of proliferative retinop­athy. The second report of Diabetic Retinopathy Study findings. Ophthalmology 85:82, 1978.

13. Diabetic Retinopathy Study Research Group: Photocoagulation treatment of proliferative diabetic retinopathy. Clinical application of Diabetic Retinop­athy Study (DRS) findings. DRS Report No. 8. Oph­thalmology 88:583, 1981.

14. West, K. M., Ahuja, M. M. S., Bennett, P. H., Grab, B., Graubauskas, V., Mateo-de-Acosta, O., Fuller, J. H., Jarrett, R. J., Keen, H., Kosaka, A. S., Miki, E., Schliack, V., and Teuscher, A.: Interrela­tionships of microangiopathy, plasma glucose and other risk factors in 3583 diabetic patients. A multi­national study. Diabetologia 22:412, 1982.

15. Jarrett, R. J., and Keen, H.: The WHO multina­tional study of vascular disease in diabetes. 3. Micro-vascular disease. Diabetes Care 2:196, 1979.

16. Barnett, A. H., Dallinger, K., Jennings, P., Fletcher, J., and Odugbesan, O.: Microalbuminuria and diabetic retinopathy. Lancet 1:53, 1985.

17. Davis, M. D.: Vitreous contraction in prolifera­tive diabetic retinopathy. Arch. Ophthalmol. 74:741, 1965.

18. Machemer, R.: Pathogenesis of proliferative neovascular retinopathies and the role of vitrectomy. Int. Ophthalmol. 1:1, 1978.

19. Jain, I. S., Luthra, C. L., and Das, T.: Diabetic retinopathy and its relation to errors of refraction. Arch. Ophthalmol. 77:59, 1967.

20. Rand, L. I., Krolewski, A. S., Aiello, L. M., Warram, J. H., Baker, R. S., and Maki, T.: Multiple factors in the prediction of risk of proliferative dia­betic retinopathy. N. Engl. J. Med. 313:1433, 1985.

21. Gay, A. J., and Rosenbaum, A. L.: Retinal ar­tery pressure in asymmetric diabetic retinopathy. Arch. Ophthalmol. 75:758, 1966.

22. Palmberg, P. F.: Diabetic retinopathy. Diabetes 26:703, 1977.

23. Koenig, R. J., Peterson, C. M., Jones, R. L., Saudek, C , Lehrman, M., and Cerami, A.: Correla­tion of glucose regulation and hemoglobin A1C in diabetes mellitus. N. Engl. J. Med. 295:417, 1976.

24. Constable, I. J., Knuitman, M. W., Welborn, T. A., Cooper, R. L., Stanton, K. M., McCann, V. J., and Grose, G. C : Assessing the risk of diabetic retinopathy. Am. J. Ophthalmol. 97:53, 1984.

25. Howard-Williams, J., Hillson, R. M., Bron, A., Awdry, P., Mann, J. I., and Hockadry, T. D. R.: Retinopathy is associated with higher glycemia in maturity-onset type diabetes. Diabetologia 27:198, 1984.

Vol. 104, No. 3 Diabetic Retinopathy, Proteinuria, and Blood Glucose 289

26. Klein, R., Klein, B., Moss, S. E., and DeMetz, D. L.: The validity of a survey question to study diabetic retinopathy. Am. J. Epidemiol. 124:104, 1986.

27. Tchobroutsky, G.: Relation of diabetic control to development of microvascular complications. Dia-betologia 15:143, 1978.

28. Frank, R. N.: On the pathogenesis of diabetic retinopathy. Ophthalmology 91:626, 1984.

29. Doft, B. H., Kingsley, L. A., Orchard, T. J., Kuller, L., Drash, A., and Becker, D.: The associa­

tion between long-term diabetic control and early retinopathy. Ophthalmology 91:763, 1984.

30. Monson, J. P., Koios, G., Toms, G. C., Kopelman, P. G., Boucher, B. ] . , Evans, S. J. W., and Alexander, W. L.: Relationship between reti­nopathy and glycaemic control in insulin-dependent and non-insulin-dependent diabetes. J. R. Soc. Med. 79:274, 1986.

31. Jerneld, B., and Algvere, P.: Prevalence of retinopathy in diabetes treated with oral antihyper-glycaemic agents. Acta Ophthalmol. 63:535, 1985.