Prevalence of diabetic retinopathy and its associated risk factors among adults in Ethiopia: a systematic review and meta-analysis

Abstract

Diabetic retinopathy is a complication of diabetes mellitus and a leading cause of blindness and visual impairment globally. Limited information existed on the epidemiology of diabetic retinopathy at the national level in Ethiopia. Thus, the objective of this review was to determine the pooled prevalence of diabetic retinopathy and its associated risk factors in Ethiopia. A systematic review and meta-analysis was conducted using previous primary studies that were found in electronic databases such as Web of Science, Scopus, PubMed, CINHAL, Google Scholar, and online African journals. We evaluated the quality of the included studies using the Newcastle-Ottawa Assessment Scale. The random-effects model was applied because heterogeneity was expected. I-Square and the Cochrane Q statistics were used to evaluate heterogeneity. Publication bias was examined using Egger’s test and a funnel plot. A random-effect meta-analysis was applied to pool the odds ratios of risk factors to determine the association between the independent and dependent variables. After 598 articles were found, 22 studies that met the eligibility requirements were included. The pooled prevalence of retinopathy among patients with diabetes in Ethiopia was 24.35% (95% CI: 18.88–29.83), with considerable heterogeneity (I² = 98.18%, p < 0.001). Ten years and longer with diabetes (AOR = 4.36, 95% CI: 1.71–7.01), hypertension (AOR = 2.54, 95% CI: 1.45–3.63), poor glycemic control (AOR = 3.83, 95% CI: 1.62–6.04), and positive proteinuria (AHR = 1.55, 95% CI: 1.02–2.07) were risk factors for diabetic retinopathy. Retinopathy affects one in four patients with diabetes. Diabetic patients with longer duration, hypertension, poor glycemic control, and positive proteinuria should receive special care.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-024-78596-9.

Introduction

The diabetes epidemic is spreading throughout the world¹. Diabetes affected 529 million people globally in 2021, representing a 6.1% prevalence of the disease². In Ethiopia, the prevalence of diabetes mellitus ranged from 5 to 10%^3–5. A contributing factor to the increasing prevalence of type 2 diabetes mellitus is increasing obesity⁶. Diabetic retinopathy is the most common microvascular complication of diabetes mellitus⁷. Diabetic retinopathy can cause vision-threatening damage to the retina, eventually leading to blindness⁸. Effective treatment is only available when the condition is detected early; therefore, routine screening is crucial for early detection⁹. With the increasing prevalence of diabetes, diabetic retinopathy is the most common cause of preventable blindness globally¹⁰. The increased prevalence of type 2 diabetes, especially in low- and middle-income countries, has significantly contributed to the increased prevalence of visual impairment and blindness caused by diabetic retinopathy, which has increased internationally¹¹. Proliferative diabetic retinopathy is the leading cause of blindness among working-age people in industrialized regions¹². As of 2010, over 100 million patients worldwide were affected by this condition, and the number is projected to increase to over 190 million by 2030¹³. The global prevalence of diabetic retinopathy was 25.2%. The prevalence of diabetic retinopathy was 20.6% in Europe and 12.5% in Southeast Asia. Comparable magnitudes of diabetes retinopathy were observed in Africa (33.8%), Middle East, North Africa (33.8%), and Western Pacific region (36.2%)¹⁴. The prevalence of diabetic retinopathy was lowest in South and Central America (13.37%) and highest in Africa (35.9%)¹⁵. The prevalence of diabetic retinopathy in a population-based study ranged from 30.2 to 31.6%, whereas in diabetic clinics, it ranged from 7.0 to 62.4%¹⁶. Sub-Saharan Africa is expected to have 47.1 million diabetes mellitus cases by 2045, up from 19.4 million cases in 2019¹⁷. An estimated 35% of people with diabetes mellitus in Sub-Saharan Africa have some type of diabetic retinopathy, and 10% have vision-threatening diabetic retinopathy¹⁸. In Ethiopia, the overall prevalence of retinopathy among diabetic patients was 19.5%¹⁹. Diabetic retinopathy is more common in people with type 1 diabetes than in those with type 2 diabetes²⁰. Approximately 75% and 50% of people with type 1 and type 2 diabetes develop retinopathy, respectively²¹.

Heritability is associated with proliferative diabetic retinopathy, accounting for 50% of cases²². The type and duration of diabetes, blood pressure, blood glucose, and lipid levels are correlated with the development and progression of diabetic retinopathy^23–25. Untreated or poorly managed hypertension was significantly correlated with any type of diabetic retinopathy²⁶. Although reducing inflammation may help prevent retinopathy even in the context of hyperglycemia, maintaining good glycemic control is still the most effective method for preventing diabetic complications²⁷. Hyperglycemia may account for just about 10% of the risk of diabetic retinopathy; the simultaneous presence of dyslipidemia and hypertension could increase the risk²⁸. Diabetic retinopathy is more common in men than in women²⁹. The most significant socio-demographic risk factors for the onset of diabetic retinopathy were age at diagnosis, education level, monthly income, medication habits, medication type, and regular exercise³⁰. Independent risk factors for diabetic retinopathy included older age, illiterate, not taking medications as prescribed, high systolic blood pressure, family history of diabetes mellitus, other micro vascular complications, poor glycemic control, poor cholesterol control, and anemia³¹.

Despite the fact that diabetic retinopathy is increasing currently at the global level^7–11, including Africa^11–13, little is known about the epidemiology of diabetic retinopathy at the national level in Ethiopia. Thus, the aim of this study was to determine the pooled prevalence of diabetic retinopathy and its predictors in Ethiopia. The information generated from this study aids in the development of intervention programs targeted specifically at diabetic patients in order to improve diabetic complications such as retinopathy and risk factors of diabetic retinopathy in Ethiopia.

Methods

Reporting and protocol registration

This systematic review and meta-analysis was reported based on the PRISMA guidelines for reporting systematic reviews and meta-analyses (PRISMA)³² (S1 File). This protocol was registered in the International Prospective Register of Systematic Reviews (PROSPERO) with the identification number CRD42023440895.

Searching strategies

We used Web of Science, Scopus, PubMed, CINHAL, Google Scholar, African journals online in parallel using search strings adapted to the requirements of each database. The search strategies were updated from August 6 to 9, 2024. We used the following keywords and phrases for all databases to ensure that the search was sensitive: (i) diabetic retinopathy; (ii) adults; and (iii) Ethiopia (S2 File).

Study selection

Articles found in the pre-specified database searches were combined and exported to EndNote X8.1 software. Duplicate articles were removed using EndNote X8.1 software. Based on the titles and abstracts of the selected papers, TGW and JAM independently determined whether or not they were relevant to this review. Then, the full texts of all articles that seemed important in the first screening were independently assessed by two authors (TGW and JAM) for eligibility. Disagreements between the authors were resolved by reasonable scientific consensus.

Outcome measurement

To determine the outcome of interest for this study, the prevalence of diabetic retinopathy reported in the original articles, either as a percentage or as the number of cases of diabetic retinopathy (n), as well as the total number of adult diabetes patients (N), was necessary. Thus, the number of adult diabetes patients with retinopathy divided by the total number of adult diabetic patients (sample size) multiplied by 100 was used to calculate the prevalence of diabetic retinopathy.

The criterion for selecting studies for review

Criteria for inclusion according to population, exposure, outcome, and context (PEOCo)

• Population: Adults aged 18 years or older were eligible to participate.

• Exposure: Diabetics.

• Outcome: retinopathy in accordance with the outcome measurement described.

• Setting: Studies conducted in Ethiopia.

• Type of study: Descriptive and observational study designs (i.e., cross-sectional and cohort) reporting the prevalence of retinopathy consistent with the PEOCo framework of Boolean systematic review operators were included.

• Publication types: published and unpublished full texts.

• Only studies published in English were reviewed.

• There is no restriction on the publication year.

Exclusion criteria

• Studies that did not provide definite information about the prevalence of retinopathy.

• Despite contacting the corresponding author, articles with no full text or extraction difficulties were excluded because the lack of full text made it difficult to evaluate the quality of the studies.

• Review, qualitative research, editorials, and commentary were not considered.

• Randomized controlled trials (RCTs) and quasi-experimental studies were not used since such types of designs cannot be consistence with the PEOCo framework of the Boolean systematic review operators.

• This review excluded articles with methodological weaknesses as well as review articles.

Data extraction process

TGW and JAM independently extracted all necessary information from the primary articles. Disagreements were settled by consensus. The data were extracted as a summary table in a standardized data extraction format created from Microsoft Excel. For each included study, the first author’s name, the study year, the study country, measurement tools, types of diabetes mellitus, data collection methods, the sample size, the number of cases with retinopathy, and the prevalence of retinopathy (S3 File).

Quality assessment

The quality of the included studies in this systematic review and meta-analysis was assessed using an adaptation of the Newcastle-Ottawa Scale for evaluating the quality of observational studies, which was adapted to evaluate the quality of the included studies in this systematic review and meta-analysis. The qualities of the included studies were independently assessed by TGW and JAM. Subjectivities between the two authors were resolved through reasonable scientific consensus. The following criteria were used to assess the quality of the screened articles for cross-sectional studies³³: Selection (sample representativeness, non-response rate, and exposure ascertainment), comparability (controlling confounding factors), and outcome (assessing the outcome and statistical tests). Lastly, articles scoring at least half (50%) of the total score were considered to be of good quality and were included in the meta-analysis.

Statistical analysis

The required information was extracted using a Microsoft Excel spreadsheet and then exported into the statistical software STATA/SE version 15 for analysis. Since heterogeneity between studies was inevitable, the random-effects model was applied³⁴. Cochran’s Q test and I² statistics were used to evaluate heterogeneity. The level of statistical heterogeneity between studies was assessed using I² statistics; percentages of 25, 50, and 75% were classified as low, medium, and high, respectively³⁵. I² ≥ 50% along with p-value < 0.05 was considered as statistically significant heterogeneity³⁶. The subgroup analysis was conducted based on region, study design, publication year, and sample size in order to reduce random variation in the point estimates of the primary articles. The asymmetry of the funnel plot³⁷ and the statistical significance of Egger’s regression test³⁸ (P-value < 0.05) were used to assess the presence of publication bias. Publication bias was adjusted using the trim-and-fill technique³⁴. The odd or hazard ratio of each risk factor was used to calculate pooled odds ratios with 95% confidence intervals to determine the association between risk factors and diabetic retinopathy. Sensitivity analysis was employed to examine the impact of a single study on the total estimate. Finally, selected articles and findings were descriptively presented in tables and forest plots.

Results

Search results

Five hundred ninety-eight primary articles were found after searching six databases: Web of Science, Scopus, PubMed, CINHAL, Google Scholar, and African journals online. Three hundred twenty-two duplicate articles and 239 irrelevant articles were eliminated by observing their titles and abstracts during evaluations. The remaining 15 articles on diabetes were excluded because none specifically addressed diabetic retinopathy (n=12) and reviews (n=3). Finally, 22 studies were eligible for inclusion in this systematic review and meta-analysis (Fig.1).

Fig. 1.

PRISMA flow diagram which shows the selection of articles for systematic review and meta-analysis.

Characteristics of the included studies

In this meta-analysis, 8623 adult patients with diabetes were identified from 22 primary studies. Retinopathy was detected in 1803 of 8623 patients with diabetes. The sample size ranged from 111 at the minimum in Addis Ababa³⁹ to 1134 at the maximum in the Amhara region⁴⁰. In this study, four regions and one city administration were represented. All studies meeting the eligibility criteria for this systematic review and meta-analysis were published between 2017 and 2024. Ten studies were located in the Amhara region. Four studies were conducted in each of the Oromia region and Addis Ababa administrative city. There were three studies in the SNNP region and one study in the Harar region. Fourteen studies were cross-sectional and eight were cohort. The maximum percentage of retinopathy was reported in Addis Ababa^41,42, whereas the minimum percentage was reported in Amhara region⁴⁰. Nine studies were conducted on patients with type 2 diabetes. A comparable number of studies have used retinal cameras and slit-lamps to diagnose diabetic retinopathy. In 14 studies, data were collected via face-to-face review, while eight studies only used chart/ record review (Table 1).

Table 1.

Characteristics of the included studies.

Authors, publication year	Region	Study design	Sample size	Cases		Outcome measurement	Reported types of DM	Data collection methods
				n	%
Zegeye AF, et al.43, 2023	Amhara	Cross-sectional	496	180	36.3	Retinal camera	Type 2	Interview
Yirdaw BE, et al.42,2023	Addis Ababa	Cross-sectional	191	98	51.3	Retinal camera	Type 2	Interview
Sahiledengle B, et al.44, 2022	Oromia	Cross-sectional	256	51	19.9	Slit-lamp	Type 1 and 2	Interview
Woyessa DN.39, 2020	Addis Ababa	Cross-sectional	111	24	21.6	Slit-lamp	Type 2	Interview
Abera F.45, 2021	Harar	Cross-sectional	210	58	27.6	Slit-lamp	Type 2	Interview
Debele GR, et al.46, 2021	Oromia	Cohort	402	81	20.15	Ophthalmoscopy	Type 1 and 2	Record review
Alemu Mersha G, et al.47, 2022	Amhara	Cross-sectional	331	113	34.1	Slit-lamp	Type 1 and 2	Interview
Gelcho GN, et al.48, 2022	Oromia	Cohort	373	154	41.3	Retinal camera	Type 1 and 2	Interview
Tilahun M, et al.49, 2020	Amhara	Cross-sectiona	302	57	18.9	Retinal camera	Type 1 and 2	Interview
Takele MB, et al.50,2022	Amhara	Cohort	494	142	28.74	Retinal camera	Type 1 and 2	Record review
Kebede SA, et al.51, 2022	Amhara	Cohort	466	80	17.17	Ophthalmoscopy	Type 2	Record review
Azeze TK, et al.52, 2018	Addis Ababa	Cohort	377	70	18.57	Retinal camera	Type 1 and 2	Record review
Ejigu T, et al.53, 2021	Amhara	Cross-sectional	223	95	42.6	Slit-lamp	Type 1 and 2	Interview
Tsegaw A, et al.54, 2021	Amhara	Cross-sectional	739	108	16	Retinal camera	Type 2	Record review
Shibru T, et al.41, 2019	Addis Ababa	Cross-sectional	191	98	51.3	Retinal camera	Type 2	Interview
Tassew CW, et al.55, 2023	Amhara	Cohort	400	78	19.5	Ophthalmoscopy	Type 2	Record review
Chisha Y, et al.56, 2017	SNNPR	Cross-sectional	270	36	13	Slit-lamp	Type 1 and 2	Record review
Barata TY, et al.57, 2024	SNNPR	Cohort	376	96	25.5	Retinal camera	Type 1 and 2	Record review
Bedada KT, et al.58, 2024	Oromia	Cross-sectional	428	53	12.4	Slit-lamp	Type 2	Interview
Alemayehu HB, et al.59, 2024	SNNPR	Cross-sectional	391	42	11%	Slit-lamp	Type 1 and 2	Interview
Abuhay HW, et al.60, 2024	Amhara	Cohort	462	54	11.7	Ophthalmoscopy	Type 1 and 2	Interview
Shumye AF, et al.40. 2024	Amhara	Cross-sectional	1134	35	3.1%	Slit-lamp	Type 1 and 2	Interview

Quality of the included study

New Castle-Ottawa was used by two authors to independently assess the quality of included studies. The quality was evaluated using seven scores. The quality of the included studies ranged from 5 to 7 scores. Of the 22 included studies, 18 achieved a score of seven out of seven, two studies obtained a score of six out of seven, and the remaining two studies obtained a score of five out of seven. Lastly, a quality score of 5 or higher was obtained for all 22 included studies (≥ 5/7 = 71.4%). The two authors jointly determined appraisals of each included study are presented in the S4 file.

Pooled prevalence of diabetic retinopathy

The random effect model of the meta-analysis revealed that the pooled prevalence of diabetic retinopathy was 24.35% (95% CI: 18.88–29.83). Significant heterogeneity exists in the studies (I² = 98.18%, p < 0.001) (Fig. 2).

Fig. 2.

Forest plots of the pooled prevalence of diabetic retinopathy.

Subgroup analysis

Addis Ababa and the Southern Nation Nationality of People Representatives (SNNPR) had the highest and lowest rates of diabetic retinopathy, at 35% (95% CI: 0.17–0.55) and 13% (95% CI: 0.1–0.18), respectively. Heterogeneity was observed in Amhara (I² = 95.6%, p < 0.001) and Addis Ababa (I² = 94%, p < 0.001). The prevalence of retinopathy in patients with type 2 diabetes was 28% (95%CI: 0.2–0.36). The prevalence of diabetic retinopathy was 29% (95% CI: 0.21–0.38) in the cross-sectional study and 24% (95% CI: 0.17–0.31) in the cohort study. In both study designs, substantial heterogeneity was observed. Using a retinal camera and slit lamp, 32% (95%CI: 0.23–0.4) and 20% (95%CI: 0.12–0.28) diabetic retinopathy were detected, respectively. About 27% and 20% of diabetic retinopathy were found using face-to-face interviews and recorded reviews data collection methods, respectively. Sample sizes larger or equal to 422 had a prevalence of 24% (95%CI: 0.14–0.34) of diabetic retinopathy with high heterogeneity (I² = 96.8%, p < 0.001), whereas sample sizes less than 422 had a prevalence of 28% (95%CI: 0.22–0.35) with significant heterogeneity (I² = 95.3%, p < 0.001). Studies published in 2020 and after had a 27% (95%CI: 0.22–0.33) prevalence of diabetic retinopathy with substantial heterogeneity (I² = 95.2%, p < 0.001) (Table 2).

Table 2.

Subgroup analysis by region, publication year, study design, and sample size.

Categories	Subgroup	Sample size	Prevalence (95%CI)	(I2, p-value)	df
Region	Amhara	5047	0.23 (0.14,0.31)	95.6%,< 0.001	9
	Addis Ababa	870	0.36 (0.17,0.54)	94%, < 0.001	3
	Oromia	1459	0.23 (0.12, 0.35)	-	3
	Harar	210	0.28 (0.22,0.34)	-	0
	SNNPR	1037	0.16 (0.08,0.25)	-	2
Study design	Cross-sectional	5273	0.25 (0.18, 0.33)	98.5%,< 0.001	13
	Cohort	3350	0.23 (0.17, 0.28)	94.5%,< 0.001	7
Study subjects	Type 2 DM patients	3232	0.28 (0.2–0.36)	96.9%,< 0.001	8
Outcome measurement tools	Retinal camera	3539	0.32 (0.23–0.4)	96.9%,< 0.001	8
	Slit-lamp	3554	0.2 (0.12–0.28)	98%,< 0.001	8
	Ophthalmoscopy	1730	0.17 (0.13–0.21)	81.3%,< 0.001	3
Data collection methods	Interview	5099	0.27 (0.19–0.36)	98.7%,< 0.001	13
	Record review	3524	0.2 (0.16–0.23)	86.4%,< 0.001	7
Sample size	≥ 422	4219	0.18 (0.09, 0.26)	98.6%,< 0.001	6
	< 422	4404	0.28 (0.22, 0.33)	95.7%,< 0.001	14
Publication year	≥ 2020	7795	0.24 (0.18,0.30)	98.26%,<0.001	18
	< 2020	838	0.27 (0.09,0.45)	-	2

SNNPR: Southern Nation Nationality People Representative, df: degree of freedom, dash (-): no heterogeneity.

Publication bias

Egger’s test was used to check for publication bias in the included studies. There is evidence of publication bias using Egger’s test (β = 12.3 (95%CI: 9.7–14.9), P < 0.001 0). A shape of funnel plot indicated that the estimated effect has an asymmetric distribution (Fig. 3).

Fig. 3.

Funnel plot to assess publication bias.

As asymmetry was detected using the funnel plot and Egger’s test, trim-and-fill analysis was used to determine 11 missed studies. The combined number of missed and observed studies was 33. The pooled effect size decreased from observed studies (24.4%, 95% CI: 18.9–29.8) to combined studies (10.5%, 95% CI: 4.9–16.2) approximately by 14%. A trim and fill plot incorporating the missed studies and then mirroring them on the opposite side to determine the best estimate of the unbiased pooled effect size is shown in Fig. 4.

Fig. 4.

Trim and fill plot.

Sensitivity analysis

A leave-one-out sensitivity analysis was carried out to examine potential sources of single-study heterogeneity in the analysis of diabetic retinopathy. The results of the sensitivity analysis showed that no single study had an effect on the findings, and all of the leave-one-out point estimates fell within the pooled estimate’s confidence interval. No obvious differences were observed. It was stable (Fig. 5).

Fig. 5.

Sensitivity analysis of diabetic retinopathy.

Risk factors for diabetic retinopathy

The significant risk factors associated with diabetic retinopathy were hypertension, poor glycemic control, duration of diabetes mellitus and proteinuria.

The association between age and diabetic retinopathy

Three studies^46,50,52 were found to determine the association between age and diabetic retinopathy. Two studies did not show significant association^46,50, while one study showed significant association⁵². This meta-analysis showed no significant association between age and diabetic retinopathy (AHR = 1.01; 95% CI: 0.99–1.02). There was moderate heterogeneity between studies (I² = 51.3%, p = 0.128) (Fig. 6).

Fig. 6.

The association between age and diabetic retinopathy.

The association between gender and diabetic retinopathy

We used five studies^{41–43,45,58} to determine the association between male sex and diabetic retinopathy. Three studies found a significant association between male sex and diabetic retinopathy (41 − 3), whereas two studies did not reveal a significant association^45,58. The meta-analysis results showed that males’ sex had no significant association with diabetic retinopathy (AOR = 1.91, 95% CI: 0.91–2.9). There was no heterogeneity among the studies (I² = 0, p = 0.657) (Fig. 7).

Fig. 7.

The association between gender and diabetic retinopathy.

The association between hypertension and diabetic retinopathy

The association between hypertension and diabetic retinopathy was assessed in six studies^{40,43,45,47,58,59}. Four studies^40,43,45,59 found a significant association between hypertension and diabetic retinopathy; the remaining studies did not show such a significant association^47,58. The findings of this meta-analysis indicated that diabetic patients with hypertension had a roughly three-fold increased risk of developing diabetic retinopathy compared with those without hypertension (AOR = 2.54, 95% CI: 1.45–3.63), with no evidence of heterogeneity (I² = 0, p = 0.829) (Fig. 8).

Fig. 8.

the association of hypertension and diabetic retinopathy.

The association between diabetic mellitus duration and retinopathy

Six primary studies^{40,44,49,53,56,59} were used to examine the association between diabetic retinopathy and the duration of diabetes. Two classifications regarding the duration of diabetes were reported in the primary studies. In three of the studies^40,44,49, the duration of diabetes was reported using a cut-off point of 10 years. In the other three studies^53,56,59, the duration of diabetes was reported using a cut-off point of 6 years. The results of the meta-analysis showed no significant association between diabetic retinopathy and six-year or longer patients with diabetes (AOR = 4.26, 95% CI: 0.82–7.69), with moderate heterogeneity (I² = 66.4%, p = 0.051). However, patients with greater or equal to 10 years of diabetes were 4.3 times more likely to develop diabetic retinopathy than those with less than 10 years of diabetes (AOR = 4.36, 95% CI: 1.71–7.01) (Fig. 9).

Fig. 9.

The association between duration of diabetic and occurrence of diabetic retinopathy.

The association between glycemic control and diabetic retinopathy

Three primary studies^47,49,59 were used to assess the association between poor glycemic control and diabetic retinopathy. All studies reported significant associations between poor glycemic control and diabetic retinopathy. The current meta-analysis indicated that patients with diabetes who had poor glycemic control were more likely to develop diabetic retinopathy than those with good glycemic control (AOR = 3.83, 95% CI: 1.62–6.04). There was no heterogeneity between studies (I² = 0, p = 0.706) (Fig. 10).

Fig. 10.

The association between glycemic control and diabetic retinopathy.

The association between type 2 diabetes mellitus and diabetes retinopathy

The association between type 2 diabetes and diabetes retinopathy was assessed using three studies^46,52,57. Two studies^46,52 reported significant association; one study reported no significant association⁵⁷. The findings of this meta-analysis showed that there was no significant association between type 2 diabetes mellitus and retinopathy (AHR = 1.59, 95% CI: -0.61-3.79) compared with type 1 diabetes mellitus, with moderate heterogeneity between studies (I2 = 55.3%, p = 0.107) (Fig. 11).

Fig. 11.

The association between type 2 diabetes mellitus and retinopathy.

The association between kidney disease and diabetes retinopathy

Using three studies^46,55,60, the association between kidney disease and diabetes retinopathy was assessed. The findings of two studies^46,60 did not reveal significant associations, whereas one study’s findings⁵⁵ revealed significant association. The results of this meta-analysis showed that no significant association was observed between kidney disease and diabetes retinopathy (AHR = 1.14, 95%CI: 0.33–1.95). There was high heterogeneity among the studies (I² = 73.7%, p = 0.022) (Fig. 12).

Fig. 12.

The association between kidney disease and diabetes retinopathy.

The association between proteinuria and diabetes retinopathy

Three studies^50,57,60 were identified to determine the association between positive proteinuria and diabetes retinopathy. Two studies^50,57 had a significant association between proteinuria and diabetes retinopathy, but one study did not report a significant association⁶⁰. In this study, diabetic patients with positive proteinuria were 1.5 times more likely to have diabetic retinopathy than those with negative proteinuria (AHR = 1.55, 95%CI: 1.02–2.07), with the absence of heterogeneity between studies (Fig. 13).

Fig. 13.

The association between proteinuria and diabetes retinopathy.

The association between diabetic medication type and diabetic retinopathy

The associations between insulin and retinopathy, as well as the association between oral medication tablets and retinopathy were assessed in three studies^40,45,47. The results of this meta-analysis indicated that diabetic patients who used insulin (AOR = 2.47, 95% CI: 0.27–4.66) and oral tablets (AOR = 0.57, 95% CI: -0.01–1.15) did not have a significant association with diabetic retinopathy compared with those who used both (Fig. 14).

Fig. 14.

The association between diabetic medication type and diabetic retinopathy.

Discussion

In recent years, the diabetes mellitus pandemic has spread worldwide^1–5,17 because of increased obesity⁶. This has led to a rise in diabetic complications, such as diabetic retinopathy^7,10,11. The increased prevalence of type 2 diabetes mellitus in low- and middle-income countries has a considerable impact on the prevalence of visual impairment and blindness owing to diabetic retinopathy^8,11,18. Determining the magnitude of disease burden and risk factors at the national level is important for establishing prevention programs. Although fragmented studies have been conducted on the prevalence of diabetic retinopathy and its associated factors at the district level, there is a lack of studies in Ethiopia regarding the pooled prevalence of diabetes mellitus and its associated factors. Some findings from these fragmented studies on factors associated with diabetic retinopathy have been inconsistent because of the interference of many factors. Thus, the aim of this study was to determine the pooled prevalence of diabetic retinopathy and associated factors in Ethiopia. The study findings provide adequate and updated information about the pooled prevalence of diabetic retinopathy and associated factors in Ethiopia.

This systematic review and meta-analysis included 22 eligible studies that reported the prevalence or incidence of diabetic retinopathy in Ethiopia. Diabetes retinopathy was found in 1803 of the 8623 adult with diabetes. In this study, the pooled prevalence of diabetic retinopathy was 24.35% (95% CI: 18.88–29.83) in Ethiopia. This implies that diabetic patients have retinopathy at a high prevalence. In order to tackle the challenges posed by diabetes, an integrated strategy that prioritizes adherence to medication and health education is required. Long-term measures, such as better diabetes and diabetic retinopathy care and a rise in early diabetes screening, should be taken to lower the prevalence of diabetic retinopathy⁹. The magnitude of this study is less than that of studies conducted in Africa, the Middle East, and the western Pacific but higher than that of Europe and south-west Asia^14,15. Moreover, it is higher than that of studies conducted in South and Central America¹⁵. This magnitude is lower than that of a study carried out in a population-based setting, but it falls within the range of a survey carried out in a diabetic clinic¹⁶. In addition, the present prevalence of diabetic retinopathy was lower in a study conducted in sub-Saharan Africa, where the estimated prevalence of diabetic retinopathy was 35%¹⁸. The current pooled prevalence of diabetic retinopathy was higher than the previous overall prevalence of diabetic retinopathy (19.5%) in Ethiopia¹³. This discrepancy could be the result of various factors, including study settings (population-based vs. clinic-based)¹⁶, economic status (developed vs. developing countries)³⁰, study design differences (cross-sectional vs. cohort), variations in healthcare facility services and quality of care, and sample size variations. The other variations may be due to genetics^22,31, medication adherence and regular exercise^30,31, variations in the time period of the studies, and variations in diabetic retinopathy diagnostic techniques (single-view ophthalmoscopy examinations may underestimate the prevalence of diabetic retinopathy when compared with multitier ophthalmoscopy examinations).

The administrative city of Addis Ababa had the highest rate of diabetic retinopathy. The increasing prevalence of diabetic retinopathy may be attributed to the higher prevalence of obesity in urban populations⁶, which in turn increases the number of diabetic patients^1–5. Nearly comparable diabetic retinopathy was observed in the cross-sectional and cohort studies. However, a higher rate of diabetic retinopathy was observed in sample sizes of less than 422 compared with sample sizes greater than 422. The high number of included studies with sample sizes of less than 422 enabled us to detect the outcome accurately, which led to a higher prevalence of retinopathy. The prevalence of diabetic retinopathy was lower in studies published after 2020 than in those published before 2020. These differences may be due to medication non-adherence among the study participants, which increases the prevalence of diabetic retinopathy. Patients with type 2 diabetes mellitus had 28% of diabetic retinopathy. This finding was lower than that of an earlier study in which 50% of retinopathy occurred in patients with type 2 diabetes mellitus²¹. About 32%, 20% and 17% of diabetes retinopathy were detected using retinal camera, slit-lamp and ophthalmoscopy, respectively. This indicates that the retinal camera is the most effective test, with high sensitivity⁶¹. In face-to-face data collection, 27% of diabetic retinopathy cases were detected, while 20% were detected in the record review data collection method. The possible reason for the discrepancy could be that primary data had more validity than secondary data. The existence of incomplete records in the secondary data reduced the validity of the information.

Diabetic patients with hypertension had a higher risk of developing retinopathy than those without hypertension. This evidence is in line with the findings of earlier studies^{23–26,28,31}. Hypertension may increase the risk of diabetic retinopathy because of increased vascular damage⁶². Longer durations (≥ 10 years) of patients with diabetes were more likely to develop retinopathy than shorter durations of patients with diabetes. This indicates that retinopathy can develop over a longer period. This evidence is consistent with that of previous studies^23–25. Diabetes patients with positive proteinuria were 1.5 times more likely to develop retinopathy than those without proteinuria. This evidence is in line with that of a previous study reporting that proteinuria and elevated blood urea nitrogen levels are excellent predictors of retinopathy; both conditions are caused by DM-related microangiopathies⁶³. Diabetic patients with poor glycemic control were more likely to develop retinopathy than those with good glycemic control. This finding is consistent with those of previous studies’ findings^23–25,31. Maintaining good glycemic control is the most effective method for preventing diabetic complications²⁷. The prognosis of diabetic retinopathy depends on glycemic control, duration of diabetes, associated comorbid conditions, and compliance with the appropriate treatment line⁶⁴.

In this study, older age was not significantly associated with diabetic retinopathy. This finding contradicts those of previous studies^30,31. The justification for the nonoccurrence of diabetic retinopathy could be the inclusion of small number of studies, resulting in an inadequate sample size to determine the association. In previous studies, there was a significant association between the types of diabetes and medication type with the occurrence of diabetes retinopathy^23–25,30. However, in this meta-analysis, a type of diabetes mellitus and medication were not significantly associated with retinopathy. The reason could be that the relationship between diabetes mellitus type and diabetic retinopathy, as well as between medication type and diabetic retinopathy, had an imprecise confidence interval, suggesting that there was not an adequate sample size to detect the relationship.

This study’s drawback is that most of the studies were cross-sectional, exhibiting unclear temporal or epidemiological cause-and-effect relationships. Another drawback is the occurrence of selection bias because all studies conducted in healthcare facilities have findings that cannot be generalized to the general population. This study’s shortcoming was the high degree of heterogeneity among the studies. The causes of the heterogeneity in the prevalence of diabetic retinopathy may include variations in the data validity between cross-sectional and cohort studies, regional variations in healthcare access and quality, and participant differences in terms of regular exercise and dietary habits. The availability of small studies conducted in certain regions restricted the study’s scope or might have revealed confounding or conflicting results because of the small sample size. Lack of study in other regions of Ethiopia is another problem. The existence of publication bias in this study indicates that a small study affected the overall pooled prevalence of diabetic retinopathy. This publication bias can be caused by different factors, such as variations in study methodology; journals may tend to publish studies that show significant results; and editors and reviewers may have a bias towards accepting studies with positive results. Our findings will help in preparing for and establishing significant health campaigns, such as diabetic retinopathy screening programs. This study can also assist policymakers and decision makers in developing and implementing potential mitigation plans for the risks of and advancement of diabetic retinopathy.

Conclusion

One in four patients with diabetes developed retinopathy. Maintaining good glycemic control and managing hypertension are the most effective ways to prevent diabetic retinopathy. Diabetic patients with longer disease duration and positive proteinuria should be given special care to reduce diabetic retinopathy.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

AOR

Adjusted Odds Ratio

CI

Confidence Interval

S

Supplementary

Fig

Figure

Author contributions

TGW and JAM conceived the idea and were fully involved in the identification, article review, data extraction, quality assessment, analysis, draft writing, and manuscript revision. TGW was heavily involved in the analysis, preparation, and revision of the manuscript. The final version of the manuscript to be considered for publication was read and approved by TGW and JAM. TGW and JAM also agreed to share equal responsibility for all aspects of this research project.

Data availability

All data generated or analyzed during this study are included in this article.

Declarations

Competing interests

The authors declare no competing interests.