Introduction

Diabetes mellitus is a serious chronic metabolic disorder characterized by persistently elevated blood glucose levels resulting from impaired insulin secretion, insulin action, or both [1]. Globally, more than one in ten adults lives with diabetes, a condition responsible for 6.7 million deaths worldwide in 2021. Approximately 80.6% of people with diabetes reside in low- and middle-income countries (LMICs), where 44.7% of cases remain undiagnosed, placing immense strain on overburdened health systems [1].

Against this global backdrop, Ethiopia is undergoing a rapid epidemiological transition, with non-communicable diseases (NCDs) accounting for a growing share of morbidity and mortality. According to 2021 International Diabetes Federation (IDF) estimates, the prevalence of diabetes among Ethiopian adults was 3.3%, with 57.6% of cases undiagnosed [1]. Data from the 2015 National STEPS Survey indicated a prevalence of raised blood glucose of 3.2%, with a higher prevalence among males (3.5%) than females (3.0%) [2]. Despite this, comprehensive and comparable national and subnational trend estimates over time remain limited. Previous subnational Global Burden of Disease (GBD) analyses for Ethiopia (based on GBD 2019) provided baseline estimates; however, the GBD 2023 iteration incorporates updated data sources, refined modeling frameworks, and newly delineated regional boundaries (the creation of the Sidama and South West regions) [3].

This study aims to provide the most up-to-date, granular assessment of the diabetes burden in Ethiopia from 1990 to 2023 using GBD 2023 data. By analyzing subnational heterogeneity and temporal trends, this study provides novel insights into the specific impact of regional disparities and upstream risk factors, offering a critical evidence base to inform localized NCD policy, planning, and monitoring.

Materials and methods

Results

Prevalence and incidence of diabetes mellitus

In 2023, Ethiopia’s age-standardized prevalence of all forms of diabetes mellitus was 2,996.4 cases per 100,000 population (95% UI: 2,704.8–3,269.5). T2DM made up nearly all cases (2,967.3 per 100,000; ~99% of total), whereas T1DM had a uniformly low prevalence (~29.0 per 100,000). There was substantial regional variation: Sidama had the highest prevalence (4,066.4 per 100,000), while the Somali region had the lowest (2,243.1 per 100,000) about 45% lower than Sidama (Figure 1). The age-standardized incidence rate increased modestly between 1990 and 2023 in both sexes. Among males, the incidence rose from 163.5 (95% UI: 147.4–179.4) to 171.4 (95% UI: 153.9–187.7) per 100,000 population, while among females it increased from 145.1 (95% UI: 130.0–160.6) to 151.8 (95% UI: 138.2–165.2) per 100,000 population (Table 1).

Display full size

Figure 1. Age-standardized prevalence rates of diabetes mellitus [Central Statistical Agency (CSA) standard population referenced for national alignment] per 100,000 population by region/city administration, Ethiopia, 2023 (both sexes). Software: QGIS 3.28. Data sources: GBD 2023 [4] and CSA 2021 [8].

Table 1. Prevalence and incidence of type 2 diabetes mellitus by sex in 1990 and 2023, with absolute change (per 100,000 population).

Indicator	Sex	1990	2023	Absolute change (1990–2023)
Prevalence	Male	2,817.8 (2,505.6–3,121.2)	3,013.7 (2,710.7–3,311.3)	+195.9
Prevalence	Female	2,709.1 (2,417.2–3,005.6)	2,912.9 (2,621.0–3,189.6)	+203.8
Incidence	Male	163.5 (147.4–179.4)	171.4 (153.9–187.7)	+7.9
Incidence	Female	145.1 (130.0–160.6)	151.8 (138.2–165.2)	+6.7

Display full size

Values are age-standardized rates per 100,000 population with 95% uncertainty intervals (UI).

From 1990 to 2023, the age-standardized prevalence of T2DM increased steadily (EAPC: +0.2%, p < 0.05), with this growth driven primarily by accelerations after 2015. In 2023, incidence rates showed a modest but consistent rise (Figure 2). Males had consistently higher prevalence and incidence than females throughout the study period, though the gap narrowed slightly by 2023 (Table 2).

Display full size

Figure 2. Age-standardized incidence rates per 100,000 population for all forms of diabetes mellitus and by type, Ethiopia and its regions/city administrations, both sexes, 2023. Software: QGIS 3.28. Data sources: GBD 2023 [4] and CSA 2021 [8].

Table 2. Age-standardized prevalence and incidence of type 2 diabetes mellitus (T2DM) by sex, with annual absolute change, 1990–2023.

Year	Prev. (M)	Δ Prev (M)	Prev. (F)	Δ Prev (F)	Inc. (M)	Δ Inc. (M)	Inc. (F)	Δ Inc. (F)
1990	2,817.8	—	2,709.1	—	163.5	—	145.1	—
1991	2,821.1	3.3	2,713.8	4.7	163.2	−0.3	145.2	0.1
1992	2,824.6	3.5	2,719.4	5.6	163	−0.2	145.3	0.1
1993	2,828.2	3.6	2,725.5	6.1	162.8	−0.2	145.5	0.2
1994	2,831.8	3.6	2,731.9	6.4	162.6	−0.2	145.6	0.1
1995	2,835.3	3.5	2,738.5	6.6	162.5	−0.1	145.8	0.2
1996	2,840.3	5	2,746.9	8.4	162.5	0	146	0.2
1997	2,847.7	7.4	2,758	11.1	162.7	0.2	146.3	0.3
1998	2,855.8	8.1	2,770.1	12.1	162.9	0.2	146.6	0.3
1999	2,862.9	7.1	2,781.6	11.5	163.1	0.2	146.9	0.3
2000	2,867.4	4.5	2,790.7	9.1	163.3	0.2	147.1	0.2
2001	2,865.1	−2.3	2,799.3	8.6	163.2	−0.1	147.2	0.1
2002	2,856.5	−8.6	2,809.3	10	162.7	−0.5	147.2	0
2003	2,845.4	−11.1	2,819.3	10	162.2	−0.5	147.2	0
2004	2,835.7	−9.7	2,827.5	8.2	161.8	−0.4	147.1	−0.1
2005	2,831.3	−4.4	2,832.2	4.7	161.9	0.1	147.1	0
2006	2,832.2	0.9	2,832.8	0.6	162.3	0.4	147.1	0
2007	2,835.6	3.4	2,830.3	−2.5	162.7	0.4	147.2	0.1
2008	2,840.7	5.1	2,826.4	−3.9	163.3	0.6	147.2	0
2009	2,846.9	6.2	2,822.7	−3.7	163.9	0.6	147.3	0.1
2010	2,853.6	6.7	2,820.5	−2.2	164.6	0.7	147.3	0
2011	2,861.3	7.7	2,818.5	−2.0	165.3	0.7	147.4	0.1
2012	2,870.8	9.5	2,815.6	−2.9	166.3	1	147.5	0.1
2013	2,881.5	10.7	2,813.6	−2.0	167.2	0.9	147.6	0.1
2014	2,892.7	11.2	2,814.2	0.6	168	0.8	147.7	0.1
2015	2,904.1	11.4	2,819.2	5	168.6	0.6	148	0.3
2016	2,918.9	14.8	2,830.8	11.6	169.2	0.6	148.4	0.4
2017	2,938.7	19.8	2,847.9	17.1	170.1	0.9	149.2	0.8
2018	2,959.6	20.9	2,867.1	19.2	171	0.9	150.1	0.9
2019	2,977.7	18.1	2,884.9	17.8	171.8	0.8	150.9	0.8
2020	2,989.3	11.6	2,897.6	12.7	172.2	0.4	151.5	0.6
2021	2,988.5	−0.8	2,905.4	7.8	171.7	−0.5	151.7	0.2
2022	2,986.4	−2.1	2,910.2	4.8	170.9	−0.8	151.9	0.2
2023	3,013.7	27.3	2,912.9	2.7	171.4	0.5	151.8	0.2

Display full size

Δ indicates absolute annual change relative to the preceding year. M: male; F: female.

Mortality and DALYs

In 2023, the age-standardized death rate from all forms of diabetes was 37.2 per 100,000 people. Substantial regional variations existed in mortality and overall burden (Tables S1–4). Urbanized regions like Addis Ababa and agrarian regions like Sidama exhibited the highest prevalence and YLD rates. In contrast, premature mortality (YLLs) was disproportionately high in the South West region (1,139.6 per 100,000), Harari, and Afar.

The national age-standardized DALY rate was 1,124.2 per 100,000 (95% UI: 855.5–1,429.2), with YLDs at 295.7 per 100,000 and YLLs at 828.5 per 100,000. The spatial distribution of these burden metrics is summarized in Figure 3.

Display full size

Figure 3. Age-standardized summary of DALYs, YLDs, and YLLs rates per 100,000 population due to diabetes mellitus by region/city administration in Ethiopia, 2023.

Discussion

This study provides the most comprehensive and updated assessment of the diabetes burden in Ethiopia using GBD 2023 data. Our findings reveal several critical insights. First, the overwhelming dominance of T2DM (> 99%) aligns with the broader African context, where rapid urbanization, dietary transitions toward high-calorie processed foods, and decreasing physical activity outpace the genetic or autoimmune etiology of T1DM [9, 10]. Second, the stark regional disparities, such as Somali’s prevalence being 45% lower than Sidama’s, highlight profound inequalities. This 45% difference translates into approximately 1,823 excess cases per 100,000 population in Sidama compared to Somali, representing a substantial disparity that demands geographically differentiated health planning and resource allocation. These variations likely reflect divergent socio-demographic trajectories rather than genetic differences. Sidama and Addis Ababa have higher urbanization rates and greater exposure to obesogenic environments, consistent with subnational evidence showing that urban populations in Ethiopia exhibit significantly higher metabolic risk factors than rural populations [2, 3, 11]. Conversely, the lower prevalence in Somali may be influenced by pastoralist lifestyles associated with higher physical activity, though severe underdiagnosis in these remote, under-resourced areas cannot be ruled out [12]. Third, regarding sex differences, prevalence and incidence remained consistently higher in males than in females. This persistent male predominance is likely driven by biological factors (e.g., higher visceral obesity rates in males) and differential healthcare-seeking behaviors, though the narrowing gap by 2023 suggests female metabolic risk exposures are rising rapidly as their socio-economic roles and urbanization rates change [13, 14].

Interestingly, regions like the South West and Afar exhibited the highest YLLs despite not having the highest prevalence. This suggests a “mortality penalty” driven by poor healthcare access, delayed diagnosis, and a lack of essential diabetes medications (insulin and oral hypoglycemic) in these peripheral regions [15, 16]. Evidence from Ethiopian clinical settings indicates that a majority of patients present with advanced complications and poor glycemic control due to late detection, directly translating into the high premature mortality observed in these specific regions [15, 16].

In this context of limited access to conventional pharmacotherapy, it is also important to recognize the widespread reliance on herbal medicine and natural products for diabetes management. In Ethiopia and similar settings, patients frequently turn to traditional remedies due to cultural acceptability, affordability, and accessibility. Typical herbal medicines such as mulberry (Morus spp.) and tea (Camellia sinensis) have garnered attention for their potential anti-diabetic properties. Mulberry leaves contain bioactive compounds like 1-deoxynojirimycin (DNJ), which inhibit carbohydrate-hydrolyzing enzymes and may reduce postprandial hyperglycemia [17]. Similarly, polyphenols found in tea have been associated with improved insulin sensitivity and a reduced risk of type 2 diabetes development in systematic reviews [18]. Acknowledging the role of these natural products is vital for a comprehensive understanding of diabetes management strategies in resource-limited settings, although further rigorous evidence is needed to standardize their integration with standard care and avoid potential interactions.

The observed decline in age-standardized YLLs from 1990 to 2023 represents a nuanced epidemiological shift. Ethiopia has historically battled high infectious disease mortality; however, recent declines in these competing causes of death (e.g., malaria, tuberculosis) would theoretically unmask diabetes deaths, which would act to increase YLLs [19, 20]. Therefore, the declining age-standardized YLLs we observed likely reflect genuine improvements in diabetes survival in urbanized areas due to better detection and access to care, alongside the mathematical effects of age-standardization removing the demographic shift of a youthful population aging over time [20].

While high fasting plasma glucose is a definitional proximal risk factor for diabetes, the critical policy message lies in upstream drivers. High BMI contributed substantially to attributable DALYs, acting as the primary modifiable precursor to insulin resistance. Temporal analyses of GBD data indicate that age-standardized DALYs attributable to high BMI and dietary risks have increased concurrently with diabetes prevalence over the past three decades globally and within Ethiopia [7, 21]. The interaction between these metabolic risks is critical: urbanization drives sedentary behavior and dietary shifts, which increase BMI, ultimately precipitating high fasting plasma glucose [2, 22].

These findings necessitate a shift from uniform national strategies to localized interventions. High-prevalence regions requiring obesity prevention include Sidama, Addis Ababa, and Harari, while high-mortality regions such as South West, Afar, and Harari need urgent health system strengthening to improve diabetes diagnosis and management. Specifically, in high-prevalence areas, interventions should focus on urban health policies: regulating processed food marketing, creating walkable urban environments, and integrating NCD screening into primary care [23, 24]. Conversely, in high-mortality regions, the priority must be strengthening health system capacity: ensuring the availability of essential medicines, training healthcare workers in peripheral health posts, and implementing task-shifting strategies for diabetes management [11, 25].

This study has limitations. As a GBD analysis, it relies on predictive modeling, which may not capture recent, acute changes in healthcare policy [26]. Furthermore, the analysis is ecological; we did not directly analyze regional covariate data (e.g., regional food security indices or local health expenditure), which limits our ability to definitively quantify the drivers of regional heterogeneity.

The growing burden of diabetes mellitus in Ethiopia, driven predominantly by type 2 diabetes and modifiable risk factors like high BMI, highlights the need for evidence-based strategies. Addressing profound regional disparities through tailored interventions for obesity prevention in high-prevalence urban centers and enhanced clinical capacity in high-mortality pastoral regions is essential. These targeted actions will mitigate the future impact of diabetes and support Ethiopia’s alignment with global Sustainable Development Goal targets for NCDs [27, 28].