Current issues of ACP Journal Club are published in Annals of Internal Medicine


Diabetologists made appropriate referrals for diabetic retinopathy

ACP J Club. 1991 May-June;114:86. doi:10.7326/ACPJC-1991-114-3-086

Related Content in the Archives
• Editorial: Confidence intervals in research evaluation

Source Citation

Nathan DM, Fogel HA, Godine JE, et al. Role of diabetologist in evaluating diabetic retinopathy. Diabetes Care. 1991;14:26-33. [PubMed ID: 1991432]



To evaluate the ability of diabetologists to screen diabetic patients for retinopathy.


Independent examination of fundi by staff diabetologists and ophthal- mologists compared with the criterion standard of 7-field stereoscopic fundus photography.


Diabetes Clinic of the Massachusetts General Hospital.


Patients with insulin-dependent and non-insulin-dependent diabetes were selected if they had had no ophthalmologic examination in the previous year and had no results of a previous ophthalmologic examination in their charts.

Description of test and criterion standard

Fundus examinations were done by diabetologists during regular office visits in a darkened room in patients with undilated pupils. Corrected visual acuity was assessed with a hand-held or wall-mounted Snellen chart. Patients were reexamined within 4 weeks by 1 or 2 ophthal- mologists, using indirect ophthal- moscopy through dilated pupils, and by stereoscopic fundus photography (criterion standard) graded by 2 retinal ophthalmologists, masked to the patients' identities. The serious error rate was defined as the percentage of patients with moderate or serious retinopathy in whom the examiner underestimated the degree of retinopathy by ≥ 2 grades.

Main outcome measure

Patients were classified as having no, minimal, moderate, or severe retinopathy on the basis of stereoscopic fundus photography.

Main results

67 patients were examined by ≥ 1 diabetologist and had stereoscopic photography. The more severe retinal lesions went undetected more frequently. Of 23 patients who were classified by diabetologists as having no or insignificant retinopathy 1 (4%) had moderate retinopathy as was documented by stereoscopic photography. Of 24 patients classified by diabetologists as having minimal retinopathy, 7 (29%) had moderate retinopathy. The serious error rate was 1 in 20 (5%). Every patient with reduced visual acuity had moderate or severe retinopathy. For ophthalmologists, no significant differences existed in classification of severity of retinopathy compared with the diabetologists' assessments (P > 0.10), but ophthalmologists' examinations were more sensitive for perimacular lesions (P < 0.05) and macular edema (P < 0.001).


Diabetologists can make ophthalmology referrals with an acceptably low error rate. Patients without detectable retinopathy need not be referred; all others should be referred for stereoscopic fundus photography. All patients with impaired corrected visual acuity should be referred.

Sources of funding: Diabetes Association of Greater Fall River, Inc.; Canon U.S.A.; Polaroid Corporation.

Address for article reprint: Dr. D.M. Nathan, Diabetes Unit, Massachusetts General Hospital, 32 Fruit Street, Boston, MA 02114, USA.


Nathan and colleagues investigated whether physicians can provide adequate ophthalmologic screening for diabetic retinopathy. This is important because patients with diabetis are currently not receiving adequate screening (1, 2). Providing this screening in physicians' offices may remove an important barrier.

Unfortunately, certain elements in the design of this study make it difficult to generalize its findings to most physicians providing care for patients with diabetes. First, the sample of just 2 diabetologists is very small, and their good performance could be because unusually high interest and skill. Second, most patients with diabetes are cared for by general internists or family practitioners. Third, age is the strongest predictor of pupil size (my own unpublished data) and many older patients with non-insulin-dependent diabetes cannot achieve adequate nonpharm- acologic dilation. This study with 57% young patients with insulin- dependent diabetes may have overestimated the success of the undilated examination and is probably not generalizable to the much older and larger population of patients with non-insulin-dependent diabetes.

This study challenges the primary care providers of patients with diabetes to improve and use their fundoscopic examination skills, but it does not provide sufficient data to warrant changing the official recommendations for routine screening of patients with diabetes by an ophthalmologist. The study does suggest that visual acuity testing should also become one of the routine screening examinations done for patients with diabetes.

Jacqueline A. Pugh, MD
Audie L. Murphy Memorial Veterans HospitalSan Antonio, Texas, USA.


1. Witkin SR, Klein R. Ophthalmologic care for persons with diabetes. JAMA. 1984;251:2534-7.

2. Herman W, Halpern M, Pack BE, Beasley J, Callaghan C. Improving care for persons with diabetes mellitus — Michigan. MMWR. 1985;34:697-9.

Author's Response

The study results also underscore the need to refer patients with any retinopathy for stereoscopic fundus photography because ophthal- mologists miss important lesions as well.

David M. Nathan, MD