[Skip to Content]
[Skip to Content Landing]
Letters
July 26, 2000

How Often Should Patients With Diabetes Be Screened for Retinopathy?—Reply

Author Affiliations
 

Stephen J.LurieMD, PhD, Contributing EditorIndividualAuthor

JAMA. 2000;284(4):437-439. doi:10.1001/jama.284.4.436

In Reply: These responses to our study raise interesting points, but mainly serve to reinforce our conclusions, which are that (1) eye screening is critically important, (2) high-risk patients should receive annual screening, and, (3) low-risk patients (those with reasonable risk factors and a previously normal eye examination) receive little to no benefit from screening annually vs every 2 to 3 years. The discussion about utility values highlights these points.

We are perplexed by Dr Javitt's criticism of our blindness utility measure, since our base model used the same utility value (0.69) that he and his colleagues used in their most recent model.1,2 However, we agree that estimates of utility of visual impairment vary substantially, and Dr Brown and colleagues cite interesting recent articles on this topic. Still, although lower utility measures would influence the estimates, especially for high-risk patients (whom we already agree should be screened annually), no assumptions in our sensitivity analyses had a major effect on the cost-effectiveness for low-risk patients, even if we set the utility value to 0. In contrast, the inclusion of any degree of dislike of retinal examination (such as a disutility value of 0.0014, reflecting one half day lost per year) makes annual screening very unfavorable for most patients. The important result of our study is the stable and robust finding of minimal benefit of annual vs every 2- to 3-year screening for low-risk individuals, not conclusions about average cost-effectiveness for the overall population. None of the concerns raised in the letters change this conclusion.

Dr Hoskins misunderstood how the information from the patient populations was used in the study. The UKPDS population was used solely to determine the association between glycemic control and retinopathy risk. Population attribute estimates were obtained from NHANES III. The UKPDS is an ideal source of estimates because it was a long-term study of patients with newly diagnosed diabetes, not a selected group of patients who had been chosen because of other characteristics.3 Dr Javitt's concern about the cameras used for photography in NHANES III is also a minor factor. Because more severe retinopathy is more likely to be detected by examination, the patients who are most often misclassified are those with mild retinopathy; this has a minimal effect on the cost-effectiveness of screening. In addition, the model actually has a small bias in favor of annual screening because we assume that patients with retinopathy did not carry a clinical diagnosis of retinopathy at entry to the model.

Hoskins is also concerned about the use of models to set policy. We note that the current recommendations for annual eye examinations for diabetic patients are also based on simulation models. However, our model is based on the best available experimental and cohort data.3-5 The results do not cease to be based on well-conducted prospective evidence simply because they are put into a simulation model. Taken to its logical (but clearly incorrect) extreme, this argument would suggest that we should not be screening at all, as there are no randomized trials of eye screening.

Retinal screening is extremely important. As we concluded in our article, we have no problem with recommending annual examination to patients as the "safest" course of action. However, we should inform patients with diabetes that the evidence suggests that screening every 2 to 3 years is almost as good if they achieve good glycemic and blood pressure control. In addition, quality standards should be changed so that inefficiency is not mandated, and policy efforts should increase emphasis on the importance of close follow-up of high-risk patients and those with early retinopathy. Current policies may not merely be inefficient, but probably also represent missed opportunities to devote more resources and effort to achieving close follow-up of patients at the highest risk and to improving screening for those who are currently not receiving eye examinations at all. Our study demonstrates that it is the high-risk patient in whom most of the risk of preventable blindness resides, and these patients must occupy more of our attention.

References
1.
Eastman  RCJavitt  JCHerman  WH  et al.  Model of complications of NIDDM, I: model construction and assumptions.  Diabetes Care. 1997;20:725-734.Google Scholar
2.
Eastman  RCJavitt  JCHerman  WH  et al.  Model of complications of NIDDM, II: analysis of the health benefits and cost-effectiveness of treating NIDDM with the goal of normoglycemia.  Diabetes Care. 1997;20:735-744.Google Scholar
3.
UK Prospective Diabetes Study (UKPDS) Group, Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33).  Lancet. 1998;352:837-853.Google Scholar
4.
Early Treatment Diabetic Retinopathy Study Research Group, Early photocoagulation for diabetic retinopathy: ETDRS report number 9.  Ophthalmology. 1991;98(suppl):766-785.Google Scholar
5.
Diabetic Retinopathy Study Research Group, Photocoagulation treatment of proliferative diabetic retinopathy: clinical application of Diabetic Retinopathy Study (DRS) findings, DRS report number 8.  Ophthalmology. 1981;88:583-600Google Scholar
×