Association between normalized lactate load and mortality in patients with septic shock: an analysis of the MIMIC-III database

Abstract

Background: An index of dynamic lactate change that incorporates both the magnitude of change and the time interval of such change, termed "normalized lactate load," may reflect the hypoxic burden of septic shock. We aimed to evaluate the association between normalized lactate load and 28-day mortality in adult septic shock patients.

Methods: Patients with septic shock were identified from the Medical Information Mart for Intensive Care (MIMIC)-III database. Lactate load was defined as the sum of the area under the curve (AUC) of serial lactate levels using the trapezoidal rule, and normalized lactate load was defined as the lactate load divided by time. Receiver-operating characteristic curves were constructed to determine the performance of initial lactate, maximum lactate and normalized lactate load in predicting 28-day mortality.

Results: A total of 1371 septic shock patients were included, and the 28-day mortality was 39.8%. Non-survivors had significantly higher initial lactate (means ± standard deviations: 3.9 ± 2.9 vs. 2.8 ± 1.7 mmol/L), maximum lactate (5.8 ± 3.8 vs. 4.3 ± 2.2 mmol/L), lactate load (94.3 ± 71.8 vs. 61.1 ± 36.4 mmol·hr./L) and normalized lactate load (3.9 ± 3.0 vs. 2.5 ± 1.5 mmol/L, all p < 0.001). The AUCs of initial lactate, maximum lactate and normalized lactate load were 0.623 (95% confidence interval: 0.596-0.648, with a cut-off value of 4.4 mmol/L), 0.606 (0.580-0.632, with a cut-off value of 2.6 mmol/L) and 0.681 (0.656-0.706, with a cut-off value of 2.6 mmol/L), respectively. The AUC of normalized lactate load was significantly greater than both initial lactate and maximum lactate (all p < 0.001). In the multivariate logistic regression model, normalized lactate load was identified as an independent risk factor for 28-day mortality.

Conclusions: Normalized lactate load is an independent risk factor for 28-day mortality in adult septic shock patients. Normalized lactate load had better accuracy than both initial and maximum lactate in determining the prognosis of septic shock patients.

Keywords: Lactate; MIMIC-III; Mortality; Septic shock.

Fig. 1

Diagram describing the calculation of lactate load and normalized lactate load. Each “T” on the x-axis represents the chart time of lactate measurement, while the corresponding “Lac” represents the corresponding lactate values. Lactate load was calculated as: ((Lac₁ + Lac₀)/2)) * (T₁ - T₀) + ((Lac₂ + Lac₁)/2)) * (T₂ - T₁) + … + ((Lac₂₄ + Lac_N)/2)) * (T₂₄ - T_N). Where the T₀ represents ICU admission time, and the corresponding Lac₀ was defined as equals to Lac₁. Similarly, Lac₂₄ (lactate value at 24 h after admission) was defined as equals to Lac_N, where the Lac_N represents the last measured lactate value in the 24 h. Normalized lactate load was calculated as lactate load divided by (T₂₄ – T₀), which actually equals to 24 h

Fig. 2

Flowchart showing a step-by-step selection of patients included in the study

Fig. 3

Comparisons among receiver-operating characteristic curves. Panel a: Comparison of receiver-operating characteristic (ROC) curves between initial lactate, maximum lactate and normalized lactate load in all included patients. The AUC of normalized lactate load was significantly greater than the initial lactate and maximum lactate (all p <  0.001). There was no difference between initial lactate and maximum lactate (p = 0.207). Panel b: Comparison of ROC in patients with a maximum lactate ≥4 mmol/L. The AUC of normalized lactate load was significantly greater than both initial lactate and maximum lactate (p <  0.001 and p = 0.005, respectively). There was no difference between initial lactate and maximum lactate (p = 0.124)