Background

Diabetes mellitus (DM) is associated with microvascular complications, such as diabetic retinopathy (DR). DR is one of the main causes of visual loss in individuals aged 20-64 years old [1] and is present in more than 77% of patients with DM type 2 who survive for over 20 years with the disease [2]. Many factors have been associated with the progression and severity of DR, such as DM duration [3, 4], the control of serum glucose levels [4–6], hypertension [4, 7] or gender [4]. So far the most solid evidence concerning the risk of DR pertains to chronic hyperglycemia; interestingly, in insulin users with DM type 2 (i.e., more severe cases), DR occurs in approximately 84.5% of patients after 15 years of the disease [6, 8]. In this study, we aim to investigate the independent associations between the stage of DR and a variety of risk factors, including years since DM diagnosis, HbA1c levels, the coexistence of hypertension, age and gender.

Materials and methods

120 consecutive patients with type 2 DM (56 men and 64 women) participated in the study (age range: 42-89 years old, median: 73.5 years). All consecutive patients were recruited (during their routine follow-up) in the Department of Internal Medicine, Veroia General Hospital, Veroia, Greece, during the time period April 2009 to September 2009. A person was considered as known diabetic, if there was an informational letter from a diabetologist. All participants underwent a comprehensive dilated fundus examination to detect DR by indirect ophthalmoscopy. DR was clinically graded in accordance with the International Clinical Diabetic Retinopathy guidelines [9]. The examined risk factors included demographic and clinical parameters. The demographic risk factors studied were age and gender. The clinical factors were the following: years since diagnosis of DM, HbA1c levels and coexistence of hypertension.

The severity of DR was treated as an ordinal variable. Specifically, 0: without findings, 1: mild non-proliferative DR, 2: moderate non-priliferative DR, 3: severe non-proliferative DR, 4: proliferative DR. Similarly, the variable concerning years since diagnosis of DM was treated as an ordinal variable (0: 0-5 years, 1: 5-10 years, 2: 10-15 years, 3: 15-20 years, 4: more than 20 years). Accordingly, HbA1c levels were treated as an ordinal variable (0: <7%, 1: 7-8.5%, 2: >8.5%).

The statistical analysis encompassed two steps: univariate and multivariate analysis. At the univariate analysis, the associations between the severity of DR and the study variables (age, gender, HbA1c levels, years since diagnosis of DM and hypertension) were appropriately evaluated through non-parametric statistics (Mann-Whitney-Wilcoxon test for independent samples, designated as MWW for reasons of brevity, or Spearman's rank correlation coefficient).

At the multivariate analysis, multivariate ordinal logistic regression was performed. The severity of DR was treated as the dependent variable. The variables which were proven significant at the univariate analysis were entered as independent variables in the model; backward selection of variables was performed so as to reach the most parsimonious model. The satisfaction of the proportionality-of-odds assumption was assessed with the appropriate likelihood ratio test. Statistical analysis was performed with STATA 8.0 statistical software (Stata Corporation, College Station, TX, USA).

This study is in accordance with the Declaration of Helsinki and has been approved by the local Ethics Committee. Written informed consent was obtained from all patients. The authors declare no conflict of interest.

Results

At the univariate analysis, the severity of DR was positively associated with patients' age (Spearman's rho = 0.4869, p < 0.0001), years since DM diagnosis (Spearman's rho = 0.6877, p < 0.0001, Table 1), HbA1c levels (Spearman's rho = 0.6315, p < 0.0001, Figure 1). Male patients exhibited more advanced DR (2.56 ± 1.41 vs. 2.05 ± 1.45 for female patients; p = 0.045, MWW). Patients with coexisting hypertension presented with more severe DR (2.47 ± 1.37 vs. 0.50 ± 0.80 for patients without hypertension; p < 0.0001, MWW).

Figure 1
figure 1

Graph presenting the positive association between the severity of DR and HbA 1c levels.

Full size image
Table 1 Detailed results on the association between the severity of DR and years since diagnosis of DM.
Full size table

The results of the multivariate ordinal logistic regression are presented in Table 2. All variables, except for age, retained their statistically significant associations with the severity of DR at the multivariate approach.

Table 2 Results of the multivariate ordinal logistic regression
Full size table

Discussion

The duration of DM appears as a meaningful predictor for DR and its severity [3, 4, 10] however, cases free of DR are occasionally reported despite long duration of DM [8]. Carefully examining the descriptive statistics of our study sample, it is astonishing that 45.8% of patients who were diagnosed with DM twenty or more years ago exhibit proliferative DR; on the contrary, 88.9% of patients who have been diagnosed with DM since 0-5 years are free of DR.

The role of chronic hyperglycemia in the development of DR has also been well established [4–6]. High HbA1c levels are closely associated with severe DR in our study. Decrease in HbA1c concentrations by 1% leads to an estimated reduction of 30% in the risk of microvascular complications [8].

Hypertension is an important risk factor for the onset and progression of the disease, and in most studies it is an independent risk factor for DR [4, 6, 7, 10]. The UK Prospective Diabetes Study (UKPDS) demonstrated that blood pressure control is associated with a reduction in DR incidence; the relative risk for DR is 1.5 concerning systolic pressure between 125-139 mmHg and 2.8 for systolic pressure higher than 140 mm Hg [6].

In our study, male patients exhibited more advanced DR; this finding seems distinct than that reported by previous studies, according to which male gender was related with the presence of DR, but not its severity [4, 10].

Concerning the role of age, it is worth commenting on our study findings. Age was associated with the severity of DR at the univariate analysis but lost its significance at the multivariate model. Most probably, this reflects a confounding effect of age, as older age and longer duration of DM are two factors closely associated with each other. As a result, age does not seem to represent an independent risk factor for severe DR.

A limitation of the present study was that DR grading was based on fundoscopy and not on fundus photography grading. This could have resulted in underestimation of the prevalence and severity of DR. In addition, a considerable limitation of this study pertains to the relatively small sample size; nevertheless, the fact that the results persisted at the multivariate approach points to the validity of the findings presented herein.

Conclusions

In conclusion, this study demonstrates that male gender, coexistence of hypertension, years since diagnosis of DM and HbA1c represent independent risk factors for severe DR.