Kidney function trajectories and their associated factors among diabetic ICU patients: a group-based trajectory modeling study

Abstract

Despite global efforts, kidney dysfunction among diabetic patients continues to rise. In low-resource settings such as Ethiopia, where access to renal replacement therapy is limited, diagnosis is often delayed, and critical care infrastructure is weak, the burden is particularly severe. However, no prior study has investigated kidney function trajectories and their associated factors among diabetic patients in the region. This study seeks to address this gap by providing baseline evidence to guide early intervention and improve renal outcomes in diabetic populations. A retrospective cohort study was conducted among 805 diabetic ICU patients aged 18 years and above. Simple random sampling was used to select medical records, and data extraction was carried out between January and March 2025. Kidney function was assessed using estimated glomerular filtration rate calculated with the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation. Group-based trajectory modeling was employed to identify distinct kidney function trajectories. Model adequacy and classification quality were evaluated using the Bayesian Information Criterion (BIC), posterior probabilities, and entropy values. Multinomial logistic regression was applied to identify associated factors of kidney function trajectories, while Cox proportional hazards regression was used to assess the association between eGFR trajectories and ICU mortality risk. Three distinctive kidney function trajectories were identified: stable trajectory (39.6%), gradual decline (25.0%), and rapid decline (35.4%). Patients aged 65 years and above (AOR = 2.70, 95% CI: 1.61–4.53), those with hypertension (AOR = 2.65, 95% CI: 1.56–4.53), proteinuria levels of + 3 and + 4 compared with negative proteinuria (AOR = 2.98, 95% CI: 1.61–5.53), and elevated potassium levels (AOR = 1.58, 95% CI: 1.12–2.24) were significantly associated with the gradual decline trajectory. Congestive heart failure (AOR = 3.45, 95% CI: 2.02–5.91), a history of acute kidney injury (AOR = 3.22, 95% CI: 1.65–6.30), hypovolemic shock (AOR = 3.96, 95% CI: 2.30–7.73), severe sepsis (AOR = 1.79, 95% CI: 1.03–3.11), and increased blood urea nitrogen (AOR = 1.54, 95% CI: 1.11–1.83) were significantly associated with the rapid decline trajectory. Furthermore, patients in the rapid decline group had a significantly higher risk of ICU mortality compared to those in the stable group (AHR = 1.84, 95% CI: 1.32–3.12). Identifying distinct kidney function trajectories highlights critical clinical patterns in diabetic ICU patients. Factors such as age, comorbidities, and key laboratory findings were significantly linked to worsening trajectories. These patterns were also associated with increased ICU mortality risk. These findings may support targeted monitoring and early risk stratification among diabetic ICU patients.