Serum lactate as a predictor of mortality in patients with infection

Objective: To determine the utility of an initial serum lactate measurement for identifying high risk of death in patients with infection. Design and setting: Post-hoc analysis of a prospectively compiled registry in an urban academic hospital. Participants: Patients with (a) a primary or secondary diagnosis of infection and (b) lactate measurement who were admitted over the 18 months following hospital-wide implementation of the Surviving Sepsis Campaign guideline for lactate measurement in patients with infection and possible severe sepsis. There were 1,177 unique patients, with an in-hospital mortality of 19%. Measurements and results: Outcome measures included acute-phase (<= 3 days) death and in-hospital death. We defined lactate ranges a priori (low, 0.0-2.0; intermediate, 2.1-3.9; high, 4.0 mmol/l or above)-and tested for linear associations with mortality by one-way analysis of variance. We determined sensitivity/specificity, odds ratios, and likelihood ratios for a lactate >= 4.0 mmol/l and performed a Bayesian analysis to determine its impact on a full range (0.01-0.99) of hypothetical pretest probability estimates for death. In-hospital mortality was 15%, 25%, and 38% in low, intermediate, and high lactate groups, respectively. Acute-phase deaths and in-hospital deaths increased linearly with lactate. An initial lactate >= 4.0 mmol/l was associated with sixfold higher odds of acute-phase death; however, a lactate level less than 4 >= mmol/l had little impact on probability of death. Conclusions: When broadly implemented in routine practice, measurement of lactate in patients with infection and possible sepsis can affect assessment of mortality risk. Specifically, an initial lactate >= 4.0 mmol/l substantiallyincreases the probability of acute-phase death.