O0488 Oct.4: 4:32pm - Clinical and Laboratory Predictors of Dehydration Severity in Children with Diabetic Ketoacidosis

Objectives
The degree of dehydration in children with diabetic ketoacidosis (DKA) is challenging to discern on presentation and its association with DKA outcomes is unclear. We aimed to characterize the degree of dehydration at presentation of DKA in a large cohort of children and to identify clinical and biochemical factors associated with severity of dehydration and outcomes.
Methods
This was a planned secondary analysis of data from the Pediatric Emergency Care Applied Research Network (PECARN) Fluid Therapies Under Investigation in DKA (FLUID) Study, a randomized clinical trial of fluid resuscitation protocols for children with DKA. Admission laboratory and physical examination data from children with DKA were compared for children who had pre- and post-treatment weights recorded. We used forward stepwise logistic regression to identify independent clinical and laboratory predictors of dehydration severity. We also used multivariable analyses to explore associations between dehydration severity at presentation and DKA outcomes.
Results
Among 1389 children with DKA episodes, 811 children had baseline and discharge weight data which permitted analysis. 379 (47%) children had dehydration classified as mild (0 - <5%), 343 (42%) as moderate (5 - 10%), (Table 1). In multivariable models including enrolling hospital, baseline characteristics and biochemical indicators of DKA severity, we found that more severe dehydration at presentation was significantly associated with new onset diabetes mellitus, higher blood urea nitrogen (BUN), lower pH and higher anion gap. Neither heart rate nor capillary refill time were associated with severity of dehydration in the stepwise multivariable model. (Table 2).
Patients with moderate dehydration received an average of 9.1 ml/kg additional intravenous (IV) fluid (95% CI 5.3-12.9) and severely dehydrated patients received an average of 8.7 ml/kg additional IV fluid (95% CI 2.6-14.8) compared to mildly dehydrated patients, adjusting for other factors. Average adjusted length of hospital stay (LOS) was 6.2 hours longer (95% CI 3.8-8.7) for moderately dehydrated patients and 11.0 hours longer (95% CI 7.1-14.9) for severely dehydrated patients compared to mildly dehydrated patients.
Conclusions
Most children with DKA had mild-to-moderate dehydration at baseline. Children with more severe dehydration received larger amounts of IV fluid and had longer hospital stays. Typical clinical features associated with dehydration (elevated heart rate and prolonged capillary refill time) were not reliably associated with degree of dehydration. New onset of diabetes, elevated BUN, more acidosis and larger anion gap, were associated with more severe dehydration at presentation. These data suggest that biochemical rather than clinical criteria are more reliable to assess dehydration severity in children with DKA.