Serial absolute lactate value < 4 versus relative 10% reduction as a predictor of mortality in severe sepsis and septic shock

Introduction

Serial lactate measurements provide a valuable tool to risk-stratify patients, determine the presence of ongoing tissue hypoperfusion, and potentially evaluate efficacy of therapeutic interventions. Prior studies have shown lactate clearance to be a valuable predictor of patient outcomes, however some have defined adequate change in lactate by a decrease from baseline by 10% whereas others considered an absolute reduction to lactate < 4mmol/L. We hypothesize that a serial absolute lactate < 4mmol/L after initial resuscitation will be as effective a predictor of future mortality as a change by 10%.

Objectives

To compare the diagnostic characteristics of an absolute serial lactate < 4 mmol/L after resuscitation with the more traditional definition (>10% decrease in serum lactate).

Methods

Single-center retrospective study of patients presenting to an urban tertiary care Emergency Department (ED) with lactate >4mmol/L and suspected infection. Continuous data was analyzed using a one-way ANOVA, whereas categorical data was compared using Fisher's exact test. Patients were stratified by lactate clearance using the traditional definition (>10% decrease in serum lactate) as well as a novel definition (second lactate < 4mmol/L) and compared.

Results

Median initial lactate was 5.2mmol/L [IQR 4.4-6.8]. The average decrease in serum lactate was 26.7% [2.3-44.4]. In-hospital mortality was 26.7%. 109/161 (67.7%) patients had >10% decrease in serum lactate, whereas 76/161 (47.2%) cleared lactate to < 4mmol/L. The average amount of crystalloid fluid resuscitation received was 2,970 ml, and the average time between lactate values was 181 minutes. Among patients who cleared lactate by at least 10%, mortality was 21.1% while those who did not clear their lactate by 10% had 38.5% mortality (p = 0.02). The negative predictive value for in-hospital mortality among those who clear their lactate by 10% was 78.9% [95% CI 69.8-85.9], sensitivity was 46.5% [95% CI: 31.5-62.2] and specificity was 72.9% [95% CI: 63.8-80.5]. Among patients who cleared their lactate to < 4mmol/L, mortality was 14.5%, while those who did not clear their lactate to < 4mmol/L had a mortality of 37.6% (p = 0.001). The negative predictive value for in-hospital mortality among those who clear lactate to < 4mmol/L was 85.5% [95% CI 75.2-92.2], sensitivity was 74.4% [58.5-86.0] and specificity was 55.1% [45.7-64.2].

Conclusions

A serial lactate < 4mmol/L as compared to >10% change has improved negative predictive value for in-hospital mortality and may provide more utility for risk stratification, assessment of response to therapy, and potentially clinical decision-making.

Grant Acknowledgment

Dr. Donnino, the Primary Investigator, is supported by NHLBI 1K02HL107447-01A1 and Dr. Cocchi is supported by AHA 15SDG22420010.

Table 1 Baseline Characteristics

Table 2 Comparison based on Lactate Clearance