Investigators reported that the strongest factor associated with glycaemic outcomes was self-monitoring frequency. The study defined poor control as HbA1c >7.5% and applied both bivariate and multivariable methods to probe determinants. On bivariate testing, SMBG frequency correlated with control (p=0.018). When candidate variables were entered into a penalized logistic regression model to accommodate the small number achieving targets, SMBG ≥4 tests/day remained independently associated with lower odds of poor control (adjusted OR 0.19, 95% CI 0.04–0.84, p=0.028). This association persisted despite the cross-sectional design and the low overall prevalence of good control in the cohort.
Short, practical implication: more frequent SMBG aligned with better odds of achieving the defined HbA1c target. The study team observed that access to testing supplies and consistent use should be a programmatic priority, given this finding.
The cohort included 160 children and adolescents attending a paediatric diabetes clinic at Kenyatta National Hospital between September 2025 and May 2026. Only 9 participants (5.6%) met the HbA1c threshold of ≤7.5%; the mean HbA1c for the sample was 11.02% (SD ±2.29). The authors also reported glucometer-derived control: among 133 patients with available readings, just 5 (3.8%) had at least 70% of their SMBG results within the 3.9–10.0 mmol/L range. These numbers indicate a pervasive gap between current glycaemic control and international targets in this clinical population.
The study team described both measures—laboratory HbA1c and the percentage of in-range SMBG readings—as complementary indices, noting that each exposed a similarly low prevalence of target achievement.
Household income showed a statistically significant association with glycaemic control on bivariate analysis (p=0.026). The authors noted, though, that after adjusting for SMBG frequency and other candidate factors, the effect of household income was attenuated and no longer independently significant in the penalized regression model. In other words, income correlated with control in unadjusted comparisons but did not explain control when SMBG frequency was accounted for.
This pattern suggests that socioeconomic disadvantage may operate partly through constrained access to testing supplies or reduced monitoring frequency; however, the primary dataset did not include causal pathway measures, and the investigators did not report mediation testing. The study did not report additional quantitative measures of out-of-pocket costs or precise household income categories beyond the statistical association noted.
The authors used chi-square or Fisher’s exact tests for bivariate comparisons and selected variables with p<0.2 for multivariable modeling. Given the very small number of participants achieving the HbA1c target, the study team applied Firth’s penalized-likelihood logistic regression to reduce small-sample bias in the adjusted estimates. The source reported the adjusted odds ratio for SMBG ≥4/day as 0.19 (95% CI 0.04–0.84, p=0.028). The manuscript explicitly states that Firth’s method was chosen because of the limited number of outcomes (good control), and no additional modeling alternatives were reported.
Readers should note that penalized regression modifies likelihood estimation to reduce bias; the investigators used it appropriately for sparse outcome data but also described their multivariable analysis as exploratory in light of sample limitations.
The cohort received care within a setting where the Life for a Child (LFaC) programme provides free insulin, glucose-monitoring supplies, and specialist education. Despite these structured supports, glycaemic control remained poor across the study population. The authors point out that even in the presence of free insulin and education, consistent access to and use of testing supplies appears to be a limiting factor. They therefore recommend prioritising strengthening access to glucose-testing materials and interventions that promote regular SMBG adherence within existing programmes.
The source did not report on the intensity, duration, or content of the education provided, nor did it quantify distribution metrics for testing supplies under LFaC, so the precise programmatic gaps remain unspecified.
This was a cross-sectional analysis of data collected between September 2025 and May 2026; causality cannot be inferred from observed associations. The authors acknowledged the small number of participants who achieved the HbA1c target and accordingly described their multivariable analysis as exploratory; they used Firth’s penalized logistic regression to mitigate small-sample bias. The source did not report details about insulin regimens, adherence measures beyond SMBG frequency, or granular socioeconomic measures that could clarify pathways linking income to control.
Data availability is restricted because the dataset contains potentially identifying paediatric information; de-identified data may be shared with qualified researchers on request and with UoN-KNH Ethics Committee approval. The authors recommend longitudinal studies to confirm whether more frequent self-monitoring causally reduces the odds of poor glycaemic control and to evaluate interventions that expand reliable access to testing supplies in similar resource-constrained settings.
Ethical oversight was provided by the University of Nairobi–Kenyatta National Hospital Ethics Committee (Approval No. P8/01/2025). The authors declared no competing interests.