Incidence of diabetes retinopathy and determinants of time to diabetes retinopathy among diabetes patients at Tikur Anbessa Hospital, Ethiopia: a retrospective follow up study

Data regarding diabetes retinopathy and associated factors are currently lacking in Ethiopia. The study aims to determine the incidence and determinants of time to diabetes retinopathy among diabetes mellitus patients at Tikur Anbessa Specialized Hospital, Addis Ababa, Ethiopia. The incidence of diabetes retinopathy is a rapidly growing burden of disease in Ethiopia. The incidence rate of diabetes retinopathy was 2.65 (95% CI 2. 54, 4.05) per 1000 person-years observation. Moreover, 70 (18.57%, 95% CI 14.63, 22.5) DM patients developed diabetes retinopathy. The median time was 74.07 months (with IQR 53.60, 89.88). Male sex (AHR = 1.94, 95% CI = 1.10, 3.39), type 2 DM (AHR = 4.01, 95% CI = 1.34, 12.00), creatinine (AHR = 2.59, 95% CI = 1.91, 3.52), borderline triglyceride (AHR = 2.87, 95% CI 1.33, 6.21) and high triglyceride levels (AHR = 2.59, 95% CI = 1.31, 4.97) were positively correlated factors to diabetes retinopathy occurrence. Multisectoral, population-based approaches are needed to reduce type 2 DM complications.


Introduction
Diabetes mellitus (DM) is an important public health problem throughout the world and its prevalence has been steadily increasing over the past few decades [1]. Diabetes retinopathy (DR) is one of the life-threatening complications and is the most common cause of acquired blindness in adults [1][2][3].
Several studies in Ethiopia have shown that the presence and severity of complications related to DR are steadily increasing cause of premature death and disability [19,22,32]. The incidence of DR has been well documented and risk factors are known in developed countries. However, studies regarding the incidence of related to diabetes retinopathy are scarce in Ethiopia and among the greater Africa continent in general. Therefore, the aim of this study was to identify the risk factors leading to DR among patients with diabetes at Tikur Anbessa specialized Hospital (TAH), Addis Ababa, Ethiopia.

Methods
An institution-based retrospective follow up study was conducted at Tikur Anbessa Hospital, the largest referral public and teaching hospital in the country, located in the capital city of Ethiopia, Addis Ababa. The hospital receives approximately 370,000-400,000 patients referred from across Ethiopia yearly [32]. The hospital has had an established diabetes clinic since 1994. Nearly 10,000 patients have enrolled the clinic due to diabetes related retinopathy disease and over 869 DM patients visited the hospital in 2017.
Patients newly diagnosed with type 1 or type 2 DM from January 2009 to May 2017 were considered for this study. Those who had at least one follow up visit and adults (aged ≥ 15 years) were included. The required sample size (373) was determined using a single popu- where n is the sample size; z is the value of standard normal distribution corresponding to a significant level of α of 5%, which is 1.96; d is the margin of error which was taken as 5% and p is the estimated proportion of DR among diabetes patients 41.4% [33]. Data were collected by standard data extraction checklist, entered using Epi Info 7 and exported to STATA 14 software for analysis. The outcome variable was time to DR due to DM after dilated eye examination using tonometry and ophthalmologic instruments. Diabetic retinopathy diagnosis includes a medical history, an ophthalmic examination and screening with retinal photographs, and regular follow up. In short, diabetes retinopathy was defined by both direct and indirect ophthalmoscopy assessments done by retinal specialists confirmed by fundus photography. Diabetes retinopathy occurs when blood vessels in the retina and optic nerve are damaged using dilated eye exam (drops are placed in the eyes to widen or dilate the pupils) using a special magnifying lens. Diabetes retinopathy apparat if there is leaking blood in retinal vessels, Pale, fatty deposits on the retina indicating leaking blood vessels. The incidence of DR was determined from the start of diagnosis date of DM until the last follow-up visit or known date of DR occurrence.
We used Cox regression survival model, since time to DR is non-negative values often exhibiting highly skewed distributions and censors [34]. Variables in the bi-variable proportional hazard model with a p-value below 2% were subsequently included in the multivariable analysis.
The parsimonious model was selected using the log likelihood test and Akaike's information criterion (AIC). Schoenfeld residual and Cox-Snell residual plot were used to check the model's assumptions. The adjusted hazard ratio (AHR) with their respective 95% confidence intervals (CI) were reported to show the strength of the associations.

Results
A total of 377 adult diabetes patients were included in the final analysis. Of these, 194 (51.5%) were females and 315 (83.6%) resided in Addis Ababa.

Discussion
This study attempted to determine the incidence of diabetes retinopathy and the risk factors among DM patients at Tikur Anbessa specialized hospital, Addis Ababa, Ethiopia. The proportion of DR was 18.57% with 2.65 per 1000 adult-months observation. The study showed a higher incidence of DR than studies done in Bangladesh and China [13,29]. This inconsistency might be due to the difference in health care systems and the quality of care given to diabetes patients. However, the incidence rate in this study was lower than that of a study done at Arbaminch general hospital, Ethiopia [35]. This difference might also be due to the data collection period included in the study. Our study included 9 years of follow up data, whereas the Arba Minch study used 3 years follow up data. Moreover, has been a general health care improvement in Ethiopia recently [36].
The study identified that among DM patient's males had a shorter time than females to develop diabetes retinopathy. This result is in line with that of a study conducted in China [29,37]. This might be due to the fact that neuroretinal function is more abnormal in adult males than in adult females [38]. However, a study conducted in Ethiopia [39] showed that males had a lower risk of developing microvascular complications of DM than females. This study showed that the hazard of diabetes retinopathy was increased with increasing creatinine levels. This result is similar to that of a study done in New Zealand [40,41]. One reason for this may be that increasing creatinine values may decrease retinal fiber layer thickness [42].
The risk of developing DR among Type 2 DM patients was higher than among Type 1 DM patients in this study. The result is similar to the study conducted at Ayder referral hospital in Ethiopia and noted that type 2 DM patients had a shorter time to developing microvascular DM complications than Type 1 DM patients [39]. This may be due to the older age of most of the Type 2 patients and their inability to support themselves to receive appropriate eye screening services [43]. Furthermore, since the disease process of Type 2 DM is more gradual than Type 1 DM, these patients may seek health services later in the disease course. In contrast, a study conducted in Spain found that type 1 diabetes patients had a higher risk to develop DR than type 2 [44]. The most likely explanation for this difference may relate to the study design, shorter follow-up period or smaller sample size as well as early screening.
The result in this study indicated that DM patients who had borderline high (150-199 mg/dl) and high (200-499 mg/dl) triglyceride (TG) levels had a lower likelihood of DR than DM patients who had desirable (< 150 mg/dl) TG level. This finding is consistent with the study findings in France [45] and Birmingham [46]. This is because raised TG levels increase blood viscosity, which in turn leads to the formation of hard exudates. In addition, TG incorporates into the cell membrane, altering its fluidity and permeability which ultimately may results in hemorrhages [47].

Conclusion
The incidence rate of diabetes retinopathy remains a high public health concern as the median time to developing diabetes retinopathy was relatively short. Health care workers should bear in mind that in addition to using drugs, consistent and proper triglyceride and creatinine level monitoring, as well as early screening and treatment of diabetes complications would be an essential part of DM care and delaying the onset of DR.

Limitations
One of the major limitations of the current study is the absence of institutional and behavioral factors, which may underestimate the effects and individual variations in the development of diabetes retinopathy. Secondly, Even though the study was conducted in the largest national referral center in which many adults come from around the country for DM follow up, the findings could not be representative of the adult diabetes population in the country due to possible selection bias.
Because of the retrospective nature of the study, some important complications of diabetes that had a significant association with DR, like chronic kidney disease were missed since we used chart review to obtain the data.