JMIR Publications

ABSTRACT

Background:

China has the world’s largest diabetic population, with suboptimal glycemic control and fragmented care posing severe challenges to primary health care (PHC) institutions. Barriers including uneven practitioner competence, clinical inertia, and disjointed clinical-public health data limit effective diabetes management, while clinical decision support systems (CDSS) show potential for improving chronic disease care but lack real-world evidence for comprehensive integration in Chinese PHC.

Objective:

To evaluate the effectiveness of a Clinical Decision Support System (CDSS) intervention for patients with poorly controlled type 2 diabetes in primary care settings and to establish a standardized model for intelligent diabetes management.

Methods:

A retrospective cohort study was conducted among diabetic patients with baseline glycated hemoglobin (HbA1c) ≥7.0 mmol/L who visited 16 community health service centers in Kunshan from January to September 2024. Patients were divided into a CDSS intervention group and a control group (routine diagnosis and treatment). 1:1 nearest neighbor propensity score matching (PSM) was used to balance baseline characteristics. After a 1-year follow-up, the glycemic achievement rates, changes in metabolic indicators and hospitalization incidence were compared. Multivariate regression and subgroup analyses were performed to determine the effect of the CDSS intervention and the subgroup heterogeneity.

Results:

A total of 27,494 patients were included, resulting in 6,136 matched pairs after PSM (all standardized mean differences <0.1). The average reduction in HbA1c in the intervention group was 0.90±1.67 mmol/L, significantly greater than the 0.20±1.97 mmol/L reduction in the control group (difference=-0.70 mmol/L, 95%CI: -0.76~-0.63, P<0.001). The glycemic control rate was significantly higher in the intervention group (63.9% vs 47.4%, difference=17.0 percentage points, 95%CI: 15.0~18.0, P<0.001). The intervention group also had better metabolic indicators and lower hospitalization rate. Multivariate regression identified CDSS as a significant predictor of HbA1c improvement (regression coefficient=-0.68, OR=2.40, both P<0.001), with greater benefits observed in older patients, those with longer disease duration, baseline HbA1c 8~9 mmol/L and existing complications.

Conclusions:

CDSS intervention can significantly improve glycemic control, optimize metabolic indicators and reduce hospitalization risk in primary care diabetic patients. which is suitable for grassroots management and can be used as a promotable and replicable standardized plan.