Point-of-care testing on admission to the intensive care unit: lactate and glucose independently predict mortality

The aim of the study was to retrospectively investigate whether parameters of routine point-of-care testing (POCT) predict hospital mortality in critically ill surgical patients on admission to the intensive care unit (ICU).

Arterial blood analyses of 1551 patients on admission to the adult surgical ICU of the Technical University Munich were reviewed. POCT was performed on a blood gas analyser. The association between acid-base status and mortality was evaluated. Metabolic acidosis was defined by base excess (BE) <-2 mmol/L and, wherever applicable, subdivided into lactic acidosis by lactate >50% of BE, anion gap (AG)-acidosis by AG >16 mmol/L, hyperchloraemic acidosis by chloride >115 mmol/L. Metabolic alkalosis was defined by BE ≥3 mmol/L. Logistic regression analysis identified variables independently associated with mortality.

Overall mortality was 8.8%. Mortality was greater in male patients (p=0.012). Mean age was greater in non-survivors (p<0.0005). Nine hundred and eighty-six patients showed no metabolic acid-base disorder (mortality 7.3%), thereof 26 patients with pCO(2)>55 mm Hg (mortality 23.1%). Three hundred and seventy-seven patients presented with acidosis (mortality 11.4%), thereof 163 patients with lactic acidosis (mortality 19%). Mortality for alkalosis (174 patients) was 12.1%. Mean blood glucose level for non-survivors was higher compared to survivors (p<0.0005). Logistic regression analysis identified lactate, glucose, age, male gender as independent predictors of mortality.

Lactate and glucose on ICU admission independently predict mortality. BE and AG failed as prognostic markers. Lactic acidosis showed a high mortality rate implying that lactate levels should be obtained on ICU admission. Prevalence of hyperchloraemic acidosis was low. Metabolic alkalosis was associated with an increased mortality. Further studies on this disturbance and its attendant high mortality are warranted.