Diabetes Burden in Sub-Saharan Africa

01 / The question

What is driving the growing diabetes burden?

Rates, counts, and disability tell different stories. This study separates them rather than treating “burden” as a single outcome.

The analysis covers Sub-Saharan Africa overall and its Central, Eastern, Southern, and Western subregions. Results are stratified by sex and diabetes type across prevalence, incidence, deaths, DALYs, years of life lost, and years lived with disability.

Analytical focus

Describe where rates changed, decompose why total counts changed, and compare forecast models before projecting to 2030.

02 / Methodology

Trends, decomposition, then forecasts.

01

Structure GBD extracts

Validate age-standardized rates, all-age counts, age-specific rates, and age-specific counts for 1990–2023.

02

Compare burden

Estimate rate changes, sex ratios, diabetes-type shares, YLL-to-YLD ratios, and subregional contrasts.

03

Decompose counts

Use symmetric decomposition to separate population growth, population aging, and age-specific rate effects.

04

Validate forecasts

Compare ARIMA, ETS, and LSTM using rolling-origin or held-out historical predictions and scale-free error.

05

Project to 2030

Refit the selected model for each headline series and report prediction intervals.

Age-standardized diabetes burden trends across Sub-Saharan African subregions — Age-standardized trends by subregion, 1990–2023. Source: full analysis output.

03 / Findings

Prevalence rose while Type 1 mortality fell.

Diabetes prevalence rate+27.6%

3,543.5 to 4,521.4 per 100,000

Total incidence rate+19.2%

189.9 to 226.2 per 100,000

Type 1 death rate−41.9%

0.84 to 0.49 per 100,000

Total diabetes death and DALY rates increased by 15.3% and 14.4%, respectively, between 1990 and 2023. For absolute counts, population growth was the dominant driver: it explained 86.5% of the increase in deaths and 87.8% of the increase in DALYs across Sub-Saharan Africa.

Headline burden and forecasts, rates per 100,000
Measure	1990	2023	Change	Selected model	2030 forecast (95% PI)
Prevalence	3,543.5	4,521.4	+27.6%	ARIMA	4,740.5 (4,687.2–4,793.8)
Incidence	189.9	226.2	+19.2%	ARIMA	237.4 (227.3–247.4)
Deaths	33.2	38.3	+15.3%	ETS	36.2 (32.1–40.3)
DALYs	1,115.4	1,275.8	+14.4%	ETS	1,245.3 (1,150.6–1,340.0)

04 / Figures & forecasts

Population growth dominates the count increase.

Decomposition of diabetes burden changes into population growth, aging, and rate effects — Symmetric decomposition of the change in total burden between 1990 and 2023.

Selected forecasts for major diabetes burden measures through 2030 — Observed headline series and selected-model projections through 2030.

05 / Interpretation

Modeled estimates require careful language.

Uncertainty intervals are not confidence intervals.

GBD inputs are modeled estimates with uncertainty from the source study. Forecast intervals represent model uncertainty and do not automatically propagate GBD uncertainty unless draws are included. The age-adjusted drift analysis is descriptive and should not be interpreted as a fully identified age-period-cohort model.