Impact of prolonged requirement for insulin on 90-day mortality in critically ill patients without previous diabetic treatments: a post hoc analysis of the CONTROLING randomized control trial

Abstract

Abstract Background Stress hyperglycemia can persist during an intensive care unit (ICU) stay and result in prolonged requirement for insulin (PRI). The impact of PRI on ICU patient outcomes is not known. We evaluated the relationship between PRI and Day 90 mortality in ICU patients without previous diabetic treatments. Methods This is a post hoc analysis of the CONTROLING trial, involving 12 French ICUs. Patients in the personalized glucose control arm with an ICU length of stay ≥ 5 days and who had never previously received diabetic treatments (oral drugs or insulin) were included. Personalized blood glucose targets were estimated on their preadmission usual glycemia as estimated by their glycated A1c hemoglobin (HbA1C). PRI was defined by insulin requirement. The relationship between PRI on Day 5 and 90-day mortality was assessed by Cox survival models with inverse probability of treatment weighting (IPTW). Glycemic control was defined as at least one blood glucose value below the blood glucose target value on Day 5. Results A total of 476 patients were included, of whom 62.4% were male, with a median age of 66 (54–76) years. Median values for SAPS II and HbA1C were 50 (37.5–64) and 5.7 (5.4–6.1)%, respectively. PRI was observed in 364/476 (72.5%) patients on Day 5. 90-day mortality was 23.1% in the whole cohort, 25.3% in the PRI group and 16.1% in the non-PRI group (p < 0.01). IPTW analysis showed that PRI on Day 5 was not associated with Day 90 mortality (IPTWHR = 1.22; CI 95% 0.84–1.75; p = 0.29), whereas PRI without glycemic control was associated with an increased risk of death at Day 90 (IPTWHR = 3.34; CI 95% 1.26–8.83; p < 0.01). Conclusion In ICU patients without previous diabetic treatments, only PRI without glycemic control on Day 5 was associated with an increased risk of death. Additional studies are required to determine the factors contributing to these results.