Establishing vitreous glucose and beta-hydroxybutyrate thresholds to assist in the diagnosis of hypothermia

Abstract

The post-mortem diagnosis of hypothermia is challenging to establish due to the lack of pathognomonic findings and the confounding problem that any comorbidity may account for death. A 4-year retrospective case-control study was performed to compare the vitreous glucose and beta-hydroxybutyrate (BHB) concentrations between hypothermia deaths and controls. Over the study period 34 cases of hypothermia and 39 controls were analyzed. Hypothermia deaths versus controls had higher mean vitreous glucose (2.93 mmol/L vs. 1.14 mmol/L; p < 0.0001), BHB (1.89 mmol/L vs. 1.35 mmol/L; p = 0.01), and combined glucose+BHB (4.83 mmol/L vs. 2.46 mmol/L; p < 0.0001). Receiver operating characteristic (ROC) curves showed that the best model for predicting hypothermia in all cases was a combined vitreous glucose+BHB threshold of 2.03 mmol/L (sensitivity 88.2 %; specificity 56.4 %). A sub-group analysis broken down by detectable levels of blood ethanol showed that cases of hypothermia with and without ethanol maintained higher median vitreous glucose relative to the controls (2.05 vs. 0.35 mmol/L and 2.70 vs. 0.65 mmol/L; p = 0.02), however median BHB was only significantly elevated when ethanol was absent (1.88 vs. 1.42 mmol/L; p < 0.0001). Subsequent ROC curve analysis demonstrated that a better model for predicting hypothermia was in cases when blood ethanol was absent. In those deaths vitreous BHB alone had the best area under the curve, with an optimum threshold of 1.83 mmol/L (sensitivity 83.3 %; specificity 96.3 %). This study shows that post-mortem vitreous glucose and BHB are useful ancillary studies to assist in the diagnosis of hypothermia. Ethanol however is a confounder and can alter the utility of vitreous BHB when diagnosing hypothermia in those who have consumed alcohol prior to death.