P-337: Predictive factors for diabetic ketoacidosis: glycemic control and therapeutic engagement in pediatric patients in Rio de Janeiro
T P D R B Navarro, J A Parreiras, A S M F D Costa, M B Gomes, J M D A PedrosoAbstract
Introduction
Diabetic ketoacidosis (DKA) represents the most severe complication of type 1 diabetes (T1D). Understanding interactions between glycemic and modifiable factors is essential to prevent hospitalizations amid scarce Brazilian epidemiological data.
Objectives
To identify clinical-laboratory variables associated with hospitalizations for DKA in youths with T1D to develop a clinical risk predictive model.
Methods
Cross-sectional study with 151 patients (<19 years) with T1D from Rio de Janeiro in 2024. Dependent variable was the number of DKA hospitalizations (0 to ≥3). Independent variables included demographic data, Glycated Hemoglobin (A1c), and self-reported treatment adherence (0-10). ANOVA, Pearson test, and multivariate logistic regression were applied.
Results
Demographic factors did not predict hospitalizations (average age: 12 years, Human Development Index: 0.7, 48% male and 67% black race). A1c was the dominant predictor (p<0.001; η2=0.318), with a mean of 8.6% (zero hospitalizations) versus 12% (≥3 episodes). Treatment adherence served as a protective factor (p<0.001; η2=0.163), declining from a mean of 7.2/10 to 5/10 between groups (0 vs ≥3). In multivariate analysis, each point increase in A1c multiplied the chances of decompensation by 1.52 (OR=1.52; p<0.001). Each additional point in adherence reduced risk by 25% (OR=0.75; p=0.009). The stratification model (AUC=0.82) showed that patients with A1c of 12% and self-reported adherence of 3/10 had a 98.6% probability of hospitalization, whereas those with A1c of 7% and self-reported adherence of 9/10 had only a 9.3% probability of hospitalization.
Conclusions
Each A1c point increase elevates DKA risk 1.52-fold, while each adherence to treatment point reduces it by 25%, showing that glycemic control and therapeutic engagement are essential preventive tools. Hence, intensive monitoring with individualized HbA1c targets and structured insulin adherence programs represents the most impactful interventions.