Insulin Pump Therapy vs Multiple Daily Insulin Injections for Glycemic Control in Children With Type 1 Diabetes: A Systematic Review and Meta-Analysis

Abstract

Type 1 diabetes mellitus (T1DM), characterized by the autoimmune destruction of pancreatic beta cells and consequent insulin deficiency, leads to various complications. Management primarily focuses on optimal glycemic control through intensive insulin therapy, either via multiple daily injections (MDI) or continuous subcutaneous insulin infusion (CSII) using insulin pumps, which offer flexibility and improved basal insulin delivery. Despite the benefits of insulin pumps, such as reduced hypoglycemia risk and better mealtime insulin management, they pose challenges such as complexity in site changes and potential ketoacidosis due to tubing issues. This systematic review adheres to PRISMA guidelines and compares CSII with MDI in children and adolescents with T1DM, concentrating on outcomes such as glycemic control measured with HbA1c and glucose levels. The review includes studies meeting stringent criteria, encompassing a broad range of methodologies and geographies. The findings of this meta-analysis indicate the differences in glycemic control with CSII compared to MDI. However, significant heterogeneity in results and methodological variations across studies necessitate cautious interpretation. The study underscores the potential of CSII in offering better control for some patients, supporting a more personalized approach to T1DM management. It highlights the need for further research to understand the long-term effects and to refine treatment protocols, considering the variations in healthcare systems, treatment approaches, and patient demographics globally.

Keywords: bolus regimen of injecting insulin; children; continuous subcutaneous insulin therapy; diabetes mellitus type 1; glycemic control; hba1c; insulin pump; mdi; multiple insulin injections.

Figure 1. PRISMA flow chart

Figure 2. Risk of bias of randomized control trials with Risk of Bias 2.0 tool

Sources: [16,23-33] Risk of bias in each article. Eleven articles were assessed: one showed a high risk of bias, six showed some concerns, and the remaining four showed a low risk of bias.

Figure 3. Forest plot of the meta-analysis

A) Forest plot detailing the mean difference and 95% confidence intervals (CI) for the effect on HbA1c of CSII against MDI. B) Forest plot detailing the mean difference and 95% confidence intervals (CI) for the effect on glucose levels of CSII against MDI.

Figure 4. Funnel plot detailing publication bias in the included studies in the meta-analysis

A) Funnel plot detailing publication bias in the included studies in the meta-analysis of HbA1c. B) Funnel plot detailing publication bias in the included studies in the meta-analysis of glucose levels.

Figure 5. Post-hoc analysis

GOSH plots with different algorithms: A) K-means, B) DBSCAN, C) Gaussian mixture model, and D) leave-one-out analysis

Figure 6. Post-hoc meta analysis of the included studies of HbA1c

A) Post-hoc forest plot detailing the mean difference and 95% confidence intervals (CI) for the effect on HbA1c of CSII against MDI. B) Post-hoc funnel plot detailing publication bias in the included studies in the meta-analysis of HbA1c.