Cumulative incidence of retinopathy. Numbers at the bottom represent patients who developed retinopathy/total number of patients followed up by year.
Cohen O, Norymberg K, Neumann E, Dekel H. Complication-Free Duration and the Risk of Development of Retinopathy in Elderly Diabetic Patients. Arch Intern Med. 1998;158(6):641-644. doi:10.1001/archinte.158.6.641
Copyright 1998 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.1998
Determining which diabetic patients are at risk for complications and targeting these patients for intensive therapy may avoid the unwanted consequences of hypoglycemia in low-risk patients. Since aging is associated with a decrease in the incidence of diabetic retinopathy, we assessed whether long complication-free duration can define elderly patients at lower risk for future development of diabetic retinopathy.
In a 10-year clinic-based study, we studied 833 type 2 diabetic patients who were free of diabetic retinopathy and older than 50 years, followed up for more than 4 years. Data included demographic and clinical information on arrival, updated every 3 to 6 months, and yearly direct ophthalmoscopic examination after pupillary dilation by experienced ophthalmologists. All the data were prospectively compiled on relational databases. End points studied were presence of retinopathy, nephropathy, peripheral neuropathy, peripheral vascular disease, hyperlipidemia, and hypertension.
Of the patients without retinopathy at the age of 50 years, 10% developed retinopathy during 4 years of follow-up. These patients had longer duration and younger onset of diabetes than the group without retinopathy at the 4-year follow-up. Clustering of microvascular and macrovascular complications was noted. Discriminant analysis showed the following factors to be significant and independent predictors of the development of retinopathy in the elderly: duration of diabetes, body mass index, age, and glucose control.
A long complication-free period does not define elderly patients with type 2 diabetes who are at lower risk for future development of retinopathy. On the contrary, the increase in disease duration is significantly associated with the development of retinopathy in this age group, as described in younger patients.
THE GOAL of treatment in patients with type 2 diabetes mellitus is to achieve normoglycemia1 so as to delay or prevent hyperglycemia-related complications, as is practiced in type 1 diabetes.2 Attempts to achieve this goal are met with an increase in hypoglycemic episodes.2,3 Aging is a risk factor for drug-induced hypoglycemia4 and is associated with a more severe outcome of hypoglycemia.5 It is therefore of major interest to define the patients at risk for complications and to target these patients for intensive therapy, avoiding the unwanted consequences of hypoglycemia in low-risk patients.
Data from the extensive Wisconsin Epidemiological Study6,7 show a decline in the 10-year incidence and rate of progression of retinopathy with increasing age and duration of diabetes in older diabetic patients not using insulin. These data complemented information from the Joslin Clinic, Boston, Mass,8 that showed that greater age was protective against the progression to proliferative diabetic retinopathy. On the basis of the above information, we hypothesized that elderly patients who do not suffer from diabetic retinopathy at baseline but have a long duration of complication-free disease are at lower risk for the future development of retinopathy than patients with shorter disease duration. This would define a "retinopathy-resistant" population in whom less rigorous glucose control therapy would be justified. To verify our hypothesis, we analyzed the data of type 2 diabetic patients who did not have retinopathy at the age of 50 years or more and investigated risk factors for future development of retinopathy.
The study group included all type 2 diabetic patients seen at the regional diabetic outpatient clinic from January 1985 through December 1995, who did not have diabetic retinopathy at the age of 50 years or more. The clinic serves most of the diabetic patients in a community with an urban population of 150000. All the data were prospectively compiled on relational databases (Clipper 87, Ashton Tate, Torrance, Calif). Data included 5500 medical records with demographic and clinical information on arrival, updated every 3 to 6 months, and yearly direct ophthalmoscopic examination after pupillary dilation by experienced ophthalmologists.
Type 1 diabetes mellitus was diagnosed in insulin-treated patients with 2 or more of the following: (1) significant ketonuria or ketoacidosis; (2) age at onset less than 40 years; (3) weight less than 110% of ideal weight; and (4) treatment with insulin within 1 year from diagnosis. Type 2 diabetes mellitus was diagnosed in all patients who did not meet the criteria for type 1.
Retinopathy was defined as the presence of more than 1 exudate, microaneurysm, or hemorrhage (nonproliferative retinopathy) and/or proliferative retinopathy in either eye.9
Nephropathy was defined as a urinary albumin level of more than 300 mg/L (a positive result on the Combur Test strip [Boehringer Mannheim, Mannheim, Germany]) in 2 of 3 consecutive urine samples and/or serum creatinine level greater than 133 µmol/L (1.5 mg/dL) at the first year of follow-up.
Peripheral neuropathy was defined as the presence of neuropathic pain and/or paresthesias in 1 or both legs, and/or absence of Achilles tendon reflex or vibration sensation and/or polyneuropathy on electromyography at the first year of follow-up.
Peripheral vascular disease was the absence of 2 or more peripheral pulses at the first year of follow-up.
Hyperlipidemia was defined as follows: total cholesterol level greater than 6.21 mmol/L (240 mg/dL) and/or low-density lipoprotein cholesterol level greater than 4.65 mmol/L (180 mg/dL) and/or fasting serum triglyceride level greater than 2.30 mmol/L (180 mg/dL) at the first year of follow-up.
Hypertension was defined as a diastolic blood pressure greater than 90 mm Hg and/or systolic blood pressure greater than 160 mm Hg and/or the use of antihypertensive medication at the first year of follow-up.
Levels of hemoglobin A1c was determined immunologically by means of a 2-channel method (normal range, 4.5%-6.4%) (Tina-quant, Boehringer Mannheim). Data are presented as the mean hemoglobin A1c measurements during the last year of follow-up.
Statistical analysis was performed with SPSS/PC (SPSS Inc, Chicago, Ill), with Student t test when values are given as mean ± SD or χ2 when values are given as percentages. Discriminant analysis to define significant and independent variables for predicting retinopathy was performed by means of the process discriminant of SPSS.
A total of 833 patients with type 2 diabetes who were older than 50 years had no retinopathy at the beginning of follow-up. At the 4th year of follow-up, 57 patients had developed diabetic retinopathy (group 1) and 401 had not developed diabetic retinopathy (group 2). Of the 833 patients, 263 did not complete 4 years of follow-up or their data were incomplete; 112 patients developed retinopathy after the 4-year mark. Patient characteristics are given in Table 1. The cumulative incidence of retinopathy is shown in Figure 1.
Comparison between groups 1 and 2 (Table 2) showed that patients who developed retinopathy had a longer retinopathy-free interval (13.2 ± 6.1 years vs 6.1 ± 6.6 years; P<.001) and younger age at onset (48.7 ± 9.0 years vs 53.4 ± 9.9 years; P=.001). Patients who developed retinopathy had poorer glucose control, as assessed by hemoglobin A1c level (8.12 ± 1.4% vs 7.7 ± 1.4%; P<.04). There were small but statistically significant differences in body mass index and age (1.2 kg/m2 and 2.65 years; P<.08 and P<.02, respectively).
The incidence of retinopathy was significantly correlated with the prevalence of other diabetes-related complications (Table 2). Of the patients who developed retinopathy, 31.6% had nephropathy and 38.6% had neuropathy vs 11.6% and 17.3%, respectively, in group 2 patients (P<.01). No increase in the prevalence of hypertension or hyperlipidemia was noted in the group 1 patients. Correlation between the variables studied and the development of diabetic retinopathy after the age of 50 years is shown in Table 3. The factors shown to be significantly associated with retinopathy were duration of diabetes (P<.001), age at onset (P=.01), body mass index (P<.08), hemoglobin A1c (P<.04), age (P<.02), and the presence of nephropathy and/or neuropathy (P<.001).
Table 4 presents the results of the discriminant function analysis. The standardized canonical discriminant function coefficient of the duration of diabetes was 0.84, while glucose control, body mass index, and age coefficients were 0.30, −0.20, and 0.27, respectively. Thus, although these 3 variables were significantly and independently associated with development of retinopathy, the most important predictor of retinopathy was the duration of diabetes. The level of prediction based on these 3 variables was 78.9% (28.9% improvement over chance).
Aging without diabetic retinopathy did not manifest itself as a protective factor against future retinopathy. Our data showed that 10% of the patients older than 50 years who had not developed retinopathy after more than 13 (average) years of diabetes developed retinopathy during the 4 years of follow-up. The incidence increased further with follow-up. Data on 10-year follow-up showed an incidence rate of up to 46.9%, but high dropout rates after 4 years of follow-up and biases caused by the tendency of patients with retinopathy to stay for longer periods of follow-up compelled us to limit the analysis to 4 years of follow-up. The high percentage of retinopathy developing during 4 years is in line with previous reports6,10,11 on elderly patients with type 2 diabetes.
The focus of our study was on risk factors of retinopathy in elderly diabetic patients. Duration of diabetes is an independent risk factor in some studies of younger patients,9,12,13 but the data on elderly patients with type 2 diabetes are not consistent. Some cross-sectional studies show that duration of diabetes was significant as a univariate variable but not an independent risk factor on multivariate analysis10,11 with age and glucose control as the dominant independent risk factors. Other reports6,7 show a decline in the incidence and rate of progression of retinopathy with increasing age and duration of diabetes in older patients with type 2 diabetes. In an attempt to rationalize this fact, we hypothesized that a diabetic patient who reaches 50 years of age without retinopathy after a long retinopathy-free interval will be less prone to develop retinopathy than a patient with a shorter duration of diabetes. The results of our study, however, did not confirm our hypothesis. Duration of diabetes was still an independent and significant risk factor for the development of retinopathy in this group of patients, as were body mass index, glucose control, and age. The impact of diabetes duration on the prediction of retinopathy was shown to be greater than the level of diabetes control and age. As 112 patients developed retinopathy after 4 years of follow-up, we analyzed the data with these patients added to either group (group 1 or group 2). No significant changes in prediction of retinopathy was noted, and the duration of diabetes was the dominant factor.
Retinopathy was associated with the prevalence of other diabetic microvascular complications but not with hypertension, as described by others.10,11,14 Possibly, the lack of significant association between retinopathy and hypertension results from false categorization of patients receiving antihypertensive medications for nonhypertensive causes as hypertensive patients. All but 1 of the patients with retinopathy developed background retinopathy, attributed to the short duration of follow-up. Thus, we could not address the rate of progression of retinopathy in the patients. Because the risk of progression is strongly related to the presence of any retinopathy,15 we assume that the development of background retinopathy in our elderly patients increases their risk of developing a more severe retinopathy. This point has to be further investigated, as there are reports that the progression of retinopathy decreases with age.7,8
Analyzing the data for patients older than 60 years in a similar way showed that 42 patients developed, and 295 did not develop, diabetic retinopathy after 4 years of follow-up. The significant factors for prediction of retinopathy were similar to those associated with retinopathy in the 50-year-old patients (data not shown).
We conclude that diabetic patients older than 50 years with a long retinopathy-free interval are at significant risk for future development of retinopathy and thus should be treated with the best control of diabetes manageable in an attempt to prevent or delay diabetes-related complications.
Accepted for publication July 17, 1997.
Reprints: Ohad Cohen, MD, Institute of Endocrinology, Sheba Medical Center, 52621 Tel Hashomer, Israel.