This analysis used the WHO STEPwise survey conducted in Kenya in 2015 to evaluate national diabetes testing coverage and to describe the burden of undiagnosed diabetes and progression through the diabetes care cascade. The authors defined prior testing as self-reported history of a blood glucose measurement by a health worker among people without known diabetes. Diabetes was defined by fasting plasma glucose (FPG) ≥7 mmol/L, a previous diagnosis, or current diabetes treatment.
The analytic sample comprised 4,167 participants after excluding individuals with missing FPG measurements and those outside the 18–69-year age range. The survey is nationally representative and cross-sectional. Weighted prevalence estimates were calculated with 95% confidence intervals to account for the survey design. The investigators used univariable and multivariable logistic regression to identify factors associated with diabetes screening, accounting for complex sampling.
The weighted prevalence of diabetes in the study population was 2.66% (95% CI: 2.06–3.42). Among participants meeting the study case definition for diabetes, 53.4% (95% CI: 39.0–67.4) were previously undiagnosed, indicating that more than half of prevalent cases had not been identified prior to the survey.
Overall, 11.9% (95% CI: 9.7–14.5) of participants reported ever having been tested for diabetes. Testing prevalence varied across demographic and clinical subgroups. Higher testing prevalence was observed among older adults, individuals with tertiary education, people with obesity, and those with hypertension. Lower testing prevalence was reported among younger adults, people with no formal education, and residents of the North Eastern region.
The analysed diabetes care cascade revealed substantial attrition at each step. Among individuals identified as having diabetes in the survey:
46.9% were aware of their diagnosis.
Of those aware, 40.8% reported being on treatment.
Of those on treatment, 26.4% had achieved glycaemic control (as defined by the survey's criteria).
These results indicate progressive losses from diagnosis awareness through treatment uptake to successful control.
In multivariable logistic regression accounting for survey design, several factors were independently associated with higher odds of reporting prior diabetes testing. Notable adjusted odds ratios reported in the source include:
Age 60–69 years: adjusted odds ratio (aOR) = 6.29 (95% CI: 3.66–10.79).
Tertiary education: aOR = 2.65 (95% CI: 1.33–5.27).
Poorest wealth quintile: aOR = 8.42 (95% CI: 4.11–17.24).
Obesity: aOR = 2.13 (95% CI: 1.41–3.21).
Hypertension: aOR = 1.64 (95% CI: 1.21–2.23).
The source presents these associations with reported confidence intervals; no additional inferential details beyond those reported were available in the article text.
The findings indicate low national coverage of diabetes testing and a substantial proportion of prevalent diabetes that remains undiagnosed. The observed care-cascade losses—limited awareness, modest treatment uptake among those aware, and low rates of treatment-associated glycaemic control—highlight gaps across early detection, linkage to care, and effective disease management.
The authors recommend strengthening integrated screening and diabetes care services, with attention to underserved populations and groups at elevated risk, to improve early detection and continuity of care. The report emphasizes the need to address barriers to screening, diagnosis, and effective long-term management to reduce the burden of uncontrolled diabetes in Kenya.
The 2015 Kenya WHO STEPwise Survey was conducted following ethical approval from the Kenya Medical Research Institute (KEMRI) Ethics Review Committee, with informed consent obtained from participants. The secondary analysis used de-identified survey data; the authors stated that no additional ethical approval was sought for the present analysis. Data used in the study are available from the Kenya National Data Archive upon request, as noted in the source.
The source article provides the primary methods, prevalence estimates, care-cascade metrics, and multivariable associations reported above. The source text does not provide additional contextual or follow-up data beyond what is summarized here.