Prevalence and associated factors of diabetic nephropathy at Tikur Anbessa Comprehensive Specialized University Hospital, Addis Ababa, Ethiopia

introduction: Given the global prevalence of diabetes, diabetic nephropathy and its consequences are among the major causes of morbidity and mortality in diabetic populations. However, the prevalence and determinants of diabetic nephropathy in Ethiopia are little studied, and were the main objectives of this study. Methods: A cross-sectional study design was followed among 340 randomly selected diabetic patients attending the national diabetes referral clinics at the diabetes centre of Tikur Anbessa Specialized Hospital, Addis Ababa, using an interviewer-administered structured questionnaire. A total of 340 patients were involved, of whom 200 (59%) were females and 256 (75%) had type 2 diabetes mellitus. Urine and blood samples were drawn from the study population and the corresponding biochemical analyses were conducted at the Ethiopian Public Health Research Institute. results: The mean age of the participants was 51.6 years (range 18–94 years). The median duration of their diabetes was 11 years (range 1–40 years). Forty-eight pecent of the patients were hypertensive. Only half of the hyperte nsive cases (53%) were using angiotensin-converting enzyme inhibitors, either alone or in combination with other anti-hypertensive medicines. Eighty-two percent of the participants had poorly controlled diabetes, with glycated haemoglobin >7%. None was using Sodium-glucose cotransporter 2 (SGLT2) inhibitors or glucagon-like peptide agonists. Some (109, 32%) of the participants were diagnosed with diabetic nephropathy in addition to reduced estimated glomerular filtration rate and albuminuria. Age, dyslipidaemia, educational status, presence of diabetic retinopathy, and elevated triglyceride levels were found to be significant predictors of the condition (P < 0.05). conclusion s : Diabetic nephropathy was present in nearly one-third of the diabetics in the study population. The management of diabetes with renoprotective agents, such as renin–angiotensin–aldosterone system inhibitors and SGLT2 inhibitors, are likely to be very importan t in this context.

Microvascular complications of diabetes are related to the duration of the disease and the degree of hyperglycaemia [4,6]. Other factors, such as cigarette smoking, genetics, dyslipidaemia and high blood pressure, contribute to the progression of these complications [4,6]. Patients with diabetic nephropathy have a higher mortality and morbidity than those without these complications [7]. Poor glycaemic control is associated with the development of microvascular complications [8].
The macrovascular complications of diabetes are coronary artery disease, peripheral arterial disease, and cerebrovascular disease [5]; they are also strongly associated with diabetic nephropathy [7,10]. Indeed, the common cause of morbidity in patients with diabetic nephropathy is cardiovascular diseases (CVD) [10]. The risk factors for CVD, including hypertension (HTN), dyslipidaemia, smoking and albuminuria, also affect the progression of kidney disease [10].
Diabetic nephropathy is clinically characterized by a progressive increment in protein excretion and deterioration of the estimated glomerular filtration rate [7]. Overall, diabetic nephropathy occurs in 20-40% of all diabetics [7]. The prevalences of DN in type 1 and type 2 diabetics are between 25-40% and 30-50%, respectively [11]. After 15 years, approximately one-third of diabetics showed microalbuminuria and less than half develop overt nephropathy [7]. The steps in the progression of diabetic nephropathy from microalbuminuria to overt nephropathy are not the same for type 1 and type 2 diabetics [7]. The condition also varies in people according to racial background [12].
The development and progression of diabetic nephropathy are associated with glycaemic control [8,13]. In the Diabetes Control and Complications Trial, patients assigned to the intensive treatment arm had a lower rate of development of albuminuria and progression to advanced chronic kidney disease (CKD) [8]. In the UK Prospective Diabetes Study, in newly diagnosed type 2 diabetic patients, good glycaemic control led to reduced albuminuria and doubling of creatinine [13].
Angiotensin-converting enzyme inhibitors (ACEIs) reduced the risk of progression to end-stage renal disease and mortality in type 1 diabetic patients [14]. ACEIs can also prevent the development of microalbuminuria in hypertensive patients with type 2 diabetes [14]. Blood pressure should be kept below 130/80 mmHg in diabetic patients with albuminuria and CKD [5,15,16].
Reduced progression of diabetic nephropathy was observed with the use of SGLT2 inhibitors in recently reported trials [17]. These inhibitors -empagliflozin, dapagliflozin and canagliflozin -have favourable cardiovascular outcomes [17]. In the CREDENCE trial, with participants with relatively wide eGFR, canagliflozin showed better renal and cardiovascular outcomes [18]. GLP agonists, such as liraglutide, also showed better renal outcome in type 2 DM patients [19].
A study conducted in our hospital showed that glycaemic control of patients with diabetes was poor [20]. The burden of microvascular complications, including diabetic nephropathy, was therefore expected to be high. There have been only relatively few studies of the prevalence of DN conducted in Ethiopia. The prospective cross-sectional study reported here examined the prevalence of diabetic nephropathy in patients subject to follow-up at the national diabetic referral centre in Addis Ababa.

METHODS
The study was conducted at the Tikur Anbessa Comprehensive Specialised Hospital (TASH), in Addis Ababa, which hosts a national endocrine referral clinic. It is the only specialized clinic in the country for comprehensive diabetes care. The study was conducted from 1 May to 30 July 2019, to determine the prevalence and associated risk factors of diabetic nephropathy in diabetic patients attending the national diabetic referral clinic. All diabetic patients, aged 18 and above who came to the clinic during this period, were invited to participate in the study. Those who gave consent were included. Potential participants with mental health problems, hearing impairments, who were unable to provide appropriate information, were unwilling to participate in the study, or were are pregnant or suffered from gestational diabetes were excluded.
The required sample size for this study was determined using a single population proportion formula, on the basis of the prevalence indicated in a recent study that assessed the prevalence of diabetic nephropathy in diabetic patients in sub-Saharan Africa. The pooled overall prevalence of DN was found to be 35.3% for the subcontient, whereas in sub-group analyses the prevalence in eastern Africa was 29.7% [21]. Thus, the present study considered the latter prevalence for sample size determination (that is, P = 0.297). Additionally, assuming a 5% marginal error and 95% confidence level and a non-response rate of 5%, the estimated. The optimum minimum population sample size as 340. A systematic random sampling technique was used to select patients. Currently, there are about 3,200 diabetic patients who are registered and have a regular follow-up at the diabetic clinic in the study. Patients had appointments at intervals of three to four months. Nearly 70% of these patients receive medication and consultations through limited, government-based health insurance. Overall, 20-25% of patients' medical expenses were covered by their employers. Few patients buy medications themselves. The diabetic clinics provide services three days per week, mainly on Mondays and Wednesdays, but some patients with gestational diabetes mellitus and foot ulcers have follow-up on Tuesdays and Fridays, respectively. On average, about 200 patients are treated per week, representing 800 monthly. Based on the decision to collect data over the course of three months, the sampling interval was determined by dividing the expected number of diabetic patients per three months (2,400) into the sample size (340), which gives a sampling interval of 7. Thus, every seventh patient coming to the clinic for a follow-up service was interviewed until the target sample size was reached Respondents were recruited at the diabetes follow-up clinic; consent was obtained for participation, and then demographic data were obtained using a well-structured questionnaire. Patient chart review and anthropometric measurements were recorded by trained nurses. The questionnaire was prepared in English and translated into the commonly spoken local language, Amharic, and then translated back into English to check for consistency. The serum levels of total cholestrol, HDL-cholestrol, LDLcholestrol and triglyceride levels were measured using a COBAS INTEGRA 400 analyser (Roche Diagnostics, Mannheim, Germany) at the Ethiopian Public Health Institute. Triglycerides and total TC were evaluated with an enzymatic colourimetric method and HDL-c and LDL-c were analysed by a homogeneous enzymatic colorimetric method.
The study was conducted after obtaining ethical approval from the ethical review committee of the Department of Internal Medicine (Addis Ababa University). Informed consent was obtained from each respondent who participated in the study. A detailed explanation of the objectives, purpose and benefits of the study was given to the respondents, all of whose responses were kept confidential.

Data analysis
The collected data were cleaned, checked for completeness, compiled and analysed. Standard descriptive methods (reporting means/percentages and standard deviations) were used to record results. The significance for the association between dependent and independent variables was carried out using a chi-squared test where necessary.
Variables that showed significant association on bivariate analyses were fitted into a multivariable logistic regression model. All statistical tests were two-sided and significance was set at a P value of <0.05.
Diabetic nephropathy is defined as a reduction in kidney function, usually with an estimated glomerular filtation rate (eGFR) <60 mL/min/1.73 m 2 , and kidney damage, usually by estimation of albuminuria >30 mg/dL [7,22]. Chronic kidney disease is defined as abnormalities of kidney structure or function, is present for three months, which was classified based on cause, GFR reduction, and urinary albumin.

rESULTS
The mean age of the participants was 51.6 years, ranging from 18 to 94 years; nearly one-third of the sample were older than 60 years; 69% were females. About 63% of the study population earned a monthly income below the national average per capita (that is, US$772.3 per annum for the year 2018, according to the World Bank, which is equivalent to 2,070 Ethiopian birrs monthly). About twothirds of them had received secondary education and above (Table 1).

biochemical characteristics of the participants
In 82% of the study population, glycated haemoglobin level was above the acceptable cut-off (>7%). Half of the participants demonstrated elevated triglyceride levels. Furthermore, 40% of the study group reported that they had lipid disorder, and all of those cases are currently using lipidlowering agents. However, the biochemical result indicates that dyslipidaemia was reported in about three-quarters (77%) of the participants as defined by any abnormality in either of the lipid profiles (TC, LDL, HDL or TG) ( Table 3).
The majority of the participants (75%) were type 2 diabetics. The median duration of their condition was 11 years, ranging from 1 to 40 years. In 51% of cases, the duration of diabetes was more than 10 years. Metformin was used to manage their diabetes either alone or in combination with either sulfonylurea (SU) or insulin by 55% of participants. Likewise, insulin was used by 60% of the study population ( Table 2). None of these patients reported the use of dipeptidyl peptidase-4 (DPP-4) inhibitors, sodium-glucose co-transporter 2 (SGLT2) inhibitors or GLP-1 agonists.
Nearly half (48.5%) of the participants were found to be hypertensive, and half of those cases (53%) had been using ACEIs either alone or in combination with other categories of anti-hypertensive medication (Figure 1), whereas only a single case was using angiotensin receptor blocker (ARB).
Among 231 patients (68%) screened for retinopathy in the preceding year, 74 (32%) were found to have diabetic retinopathy in a mild to severe degree ( Figure 2).

Diabetic nephropathy in ethiopia
In respect of the two indicators of nephropathy, 40% of the sample had mild to severely decreased eGFR and a quarter of them fell into categories of moderate to severely increased albuminuria (Table 3). About 32% of the study population were diagnosed for diabetic nephropathy with a stage of moderate to severely increased risk of CKD, using a composite measure of eGFR and albuminuria categories as presented in Figure 3. The prevalence of DN at a stage of moderate to severely increased risk of CKD using a composite measure of eGFR and albuminuria categories in type 1 diabetics and type 2 diabetics was 23.8% and 34%, respectively ( Table 5).

Comparison of various variables with respect to type 1 and type 2 diabetes mellitus
Age, BMI and triglyceride levels were significantly higher among type 2 diabetics, whereas mean eGFR was significantly higher in type 1. Glycaemic control in both groups was poor (glycated haemoglobin > 7%); however, the mean glycated haemoglobin was relatively higher in type 1 cases (Table 4).   Triglyc-eride levels were significantly higher than the desired cut-off for those with type 2 than with type 1 ( Table 5).
Sixty percent of type 2 diabetics were hypertensive whereas the prevalence was significantly lower in type 1 cases (20%). There was no significant difference between groups in respect of monthly income. However, a significantly higher proportion of type 1 diabetics (82%) had attended education at secondary and higher levels.
The number of participants who were overweight and obese was significantly higher in type 2 (62%) than type 1 (32%) cases. The estimate of eGFR (>90 mL/min/1.73 m 2 ) was significantly higher in type 1 diabetics (Table 6). However, no significant association was observed with respect to albuminuria, diabetic retinopathy and total cholesterol.
Among participants screened for diabetic retinopathy, 32% presented with the condition in various stages (20% in mild and 12% in moderate to severe form). The latter condition was significantly associated with advanced age (Table 6). Three-quarters of the study population with moderate and severe diabetic retinopathy had eGFR <60 mL/min/1.73 m 2 .

bivariate and multivariate analysis of factors associated with diabetic nephropathy
In the bivariate analysis, DN was significantly associated with age, diabetic retinopathy, triglyceride level, dyslipidemia, eGFR and albuminuria. Diabetic patients with reduced eGFR (<60 mL/min/1.73 m 2 ) and with increased albuminuria had an odds chance of being presented with diabetic nephropathy of 4 and 11 times greater probability, respectively (Table 7).
In multivariate analysis, after adjusting for other factors, the only variables that were associated were educational status, presence of diabetic retinopathy and elevated triglyceride level. Diabetic patients with hypertriglyceridemia (≥150

DISCUSSION
About 63% of the study population earned a monthly income below the per capita national average This will create a problem of obtaining appropriate medications. Studies elsewhere show that monthly household income is inversely related to diabetic nephropathy prevalence rate. [23] The median duration of diabetes was 11 years, with a longer history seen in type 1 diabetic patients (ranging from 1 to 40 years). Poor glycaemic control is associated with microvascular complications including diabetic nephropathy. Glycated haemoglobin greater than 7% was found in four out of five participants. Patient, physician and healthcare-related factors might contribute to poor glycaemic control, which in this study was defined by glycated haemoglobin greater than 7% and was not associated with diabetic nephropathy. This may indicate that other factors are better predictors than glycaemic control for DN. No one in the study population was using agents with renoprotective potential such as SGLT2 inhibitors. This may be related to the lack of availability of these agents in the hospital and their associated cost, which may not be affordable by the majority of patients.
Only half of the hypertensive cases (53%) had been using ACEIs, which are widely available in Ethiopia, either alone or in combination with other categories of anti-hypertensives medicines. There is a lack of prescription of these agents in these circumstances. ACE inhibitors or ARBs have been found to delay the progress of diabetic nephropathy.
Dyslipidaemia was found in three-quarters of the participants even though only 40% of them were taking lipidlowering agents. Although statins are widely available and affordable, the low use of these lipid-lowering agents indicates a lack of routine lipid assessment and management in this population.
Regarding the two indicators of nephropathy, 40% of the participants had mild to severely reduced eGFR and a quarter of them demonstrated moderate to severely increased albuminuria. About 32% of the study population were diagnosed with diabetic nephropathy corresponding to moderate to severely increased risk on the basis of a composite measure of eGFR and albuminuria. Cases of advanced CKD with eGFR <30 mL/min/1.73 m 2 were relatively few in this group, which may be due to increased CVD-associated death or some patients may only continue their renal clinic follow-up.
The prevalence of diabetic kidney disease in Africa ranges from 11% in Tunisia to 83.7% in Tanzania [21]. Proteinuria was found in 5.3 % of diabetic patients in South Africa and in 53.1% in Cameroon [21]. Diabetic nephropathy using GFR criteria was seen in 4.6 % of patients in Tanzania and 43.1% in Nigeria [21].
A study in Butajira in southern Ethiopia recorded 18.2% and 23.8% of diabetic patients with eGFR < 60 mL/min/ 1.73 m 2 , based on MDRD and Cockcroft-Gault equation [24]. In the same study, 0.9% of patients had eGFR < 30 mL/min/1.73 m 2 [24]. In a systematic review conducted in Ethiopia, the prevalence of diabetic nephropathy   [4,5]. In northern Ethiopia, a casecontrol study showed that the major determinants of DM nephropathy were age, weight, poor glycaemic control, systolic blood pressure and non-adherence to self-monitoring glucose [27]. In Gonder, the major predictors of DM nephropathy were systolic hypertension and the duration of diabetes [26]. In southern Ethiopia, the major risk factors were greater age, longer duration of diabetes, family history of kidney disease, and poor glucose control [24]. In our study, in the bivariate analysis, diabetic nephropathy was significantly associated with age, diabetic retinopathy, triglyceride level, dyslipidaemia, eGFR and albuminuria.
In the multivariate analysis, after adjusting for other factors, the only variables that were associated significantly with diabetic nephropathy were educational status, the presence of diabetic retinopathy and elevated triglyceride levels.
Diabetic patients with hypertriglyceridemia (>150 mg/dL) are 3.64 times more likely to develop DN. Patients with diabetic retinopathy had a 5-times greater chance of presenting with diabetic nephropathy.
One of the limitations of this study is that we quantified albuminuria using an albumin dipstick, which may have low sensitivity and be prone to operator-dependent results.
Another limitation is that we assessed albuminuria just once, which may overestimate the prevalence of diabetic nephropathy.

CONCLUSIONS
Diabetic nephropathy was present in nearly one-third of the diabetic patients we examined. Age, educational status, dyslipidaemia, the presence of diabetic retinopathy and elevated triglyceride levels were significant predictors of the condition. We found that half of the diabetic patients were hypertensive; however, the use of ACEIs and ARBs was less than optimal. Despite SGLT2 inhibitors and GLP 1 agonists able to confer renoprotective advantage, none of the patients was on these drugs. Therefore, any intervention that promotes the availability of renoprotective agents, including health insurance, could provide benefits in the circumstances reported here.
Because of the known association between cardiovascular disease and diabetic nephropathy, we recommend further studies and interventions to mitigate their combined effect.