The Added Value of Lactate and Lactate Clearance in Prediction of In-Hospital Mortality in Critically Ill Patients With Sepsis

Establishment of a mortality risk nomogram for predicting in-hospital mortality of sepsis: cohort study from a Chinese single center

Objective

To establish a mortality risk nomogram for predicting in-hospital mortality of sepsis patients in the Chinese population.

Methods

Data were obtained from the medical records of sepsis patients enrolled at the Affiliated Huadu Hospital, Southern Medical University, between 2019 and 2021. A total of 696 sepsis patients were initially included in our research, and 582 cases were finally enrolled after screening and divided into the survival group (n = 400) and the non-survival group (n = 182) according to the incidence of mortality during hospitalization. Twenty-eight potential sepsis-related risk factors for mortality were identified. Least absolute shrinkage and selection operator (LASSO) regression was used to optimize variable selection by running cyclic coordinate descent with k-fold (tenfold in this case) cross-validation. We used binary logistic regression to build a model for predicting mortality from the variables based on LASSO regression selection. Binary logistic regression was used to establish a nomogram based on independent mortality risk factors. To validate the prediction accuracy of the nomogram, receiver operating characteristic curve (ROC) analysis, decision curve analysis (DCA) and restricted cubic spline (RCS) analysis were employed. Eventually, the Hosmer-Lemeshow test and calibration curve were used for nomogram calibration.

Results

LASSO regression identified a total of ten factors, namely, chronic heart disease (CHD), lymphocyte count (LYMP), neutrophil-lymphocyte ratio (NLR), red blood cell distribution width (RDW), C reactive protein (CRP), Procalcitonin (PCT), lactic acid, prothrombin time (PT), alanine aminotransferase (ALT), total bilirubin (Tbil), interleukin-6 (IL6), that were incorporated into the multivariable analysis. Finally, a nomogram including CHD, LYMP, NLR, RDW, lactic acid, PT, CRP, PCT, Tbil, ALT, and IL6 was established by multivariable logistic regression. The ROC curves of the nomogram in the training and validation sets were 0.9836 and 0.9502, respectively. DCA showed that the nomogram could be applied clinically if the risk threshold was between 29.52 and 99.61% in the training set and between 31.32 and 98.49% in the testing set. RCS showed that when the value of independent risk factors from the predicted model exceeded the median, the mortality hazard ratio increased sharply. The results of the Hosmer-Lemeshow test (χ2 = 0.1901, df = 2, p = 0.9091) and the calibration curves of the training and validation sets showed good agreement with the actual results, which indicated good stability of the model.

Conclusion

Our nomogram, including CHD, LYMP, NLR, RDW, lactic acid, PT, CRP, PCT, Tbil, ALT, and IL6, exhibits good performance for predicting mortality risk in adult sepsis patients.

Early prediction of mortality at sepsis diagnosis time in critically ill patients by using interpretable machine learning

This study applied machine learning for the early prediction of 30-day mortality at sepsis diagnosis time in critically ill patients. Retrospective study using data collected from the Medical Information Mart for Intensive Care IV database. The data of the patient cohort was divided on the basis of the year of hospitalization, into training (2008-2013), validation (2014-2016), and testing (2017-2019) datasets. 24,377 patients with the sepsis diagnosis time < 24 h after intensive care unit (ICU) admission were included. A gradient boosting tree-based algorithm (XGBoost) was used for training the machine learning model to predict 30-day mortality at sepsis diagnosis time in critically ill patients. Model performance was measured in both discrimination and calibration aspects. The model was interpreted using the SHapley Additive exPlanations (SHAP) module. The 30-day mortality rate of the testing dataset was 17.9%, and 39 features were selected for the machine learning model. Model performance on the testing dataset achieved an area under the receiver operating characteristic curve (AUROC) of 0.853 (95% CI 0.837-0.868) and an area under the precision-recall curves of 0.581 (95% CI 0.541-0.619). The calibration plot for the model revealed a slope of 1.03 (95% CI 0.94-1.12) and intercept of 0.14 (95% CI 0.04-0.25). The SHAP revealed the top three most significant features, namely age, increased red blood cell distribution width, and respiratory rate. Our study demonstrated the feasibility of using the interpretable machine learning model to predict mortality at sepsis diagnosis time.

Hierarchical Capability in Distinguishing Severities of Sepsis via Serum Lactate: A Network Meta-Analysis

Background: Blood lactate is a potentially useful biomarker to predict the mortality and severity of sepsis. The purpose of this study is to systematically review the ability of lactate to predict hierarchical sepsis clinical outcomes and distinguish sepsis, severe sepsis and septic shock. Methods: We conducted an exhaustive search of the PubMed, Embase and Cochrane Library databases for studies published before 1 October 2022. Inclusion criteria mandated the presence of case-control, cohort studies and randomized controlled trials that established the association between before-treatment blood lactate levels and the mortality of individuals with sepsis, severe sepsis or septic shock. Data was analyzed using STATA Version 16.0. Results: A total of 127 studies, encompassing 107,445 patients, were ultimately incorporated into our analysis. Meta-analysis of blood lactate levels at varying thresholds revealed a statistically significant elevation in blood lactate levels predicting mortality (OR = 1.57, 95% CI 1.48-1.65, I2 = 92.8%, p < 0.00001). Blood lactate levels were significantly higher in non-survivors compared to survivors in sepsis patients (SMD = 0.77, 95% CI 0.74-0.79, I2 = 83.7%, p = 0.000). The prognostic utility of blood lactate in sepsis mortality was validated through hierarchical summary receiver operating characteristic curve (HSROC) analysis, yielding an area under the curve (AUC) of 0.72 (95% CI 0.68-0.76), accompanied by a summary sensitivity of 0.65 (95% CI 0.59-0.7) and a summary specificity of 0.7 (95% CI 0.64-0.75). Unfortunately, the network meta-analysis could not identify any significant differences in average blood lactate values' assessments among sepsis, severe sepsis and septic shock patients. Conclusions: This meta-analysis demonstrated that high-level blood lactate was associated with a higher risk of sepsis mortality. Lactate has a relatively accurate predictive ability for the mortality risk of sepsis. However, the network analysis found that the levels of blood lactate were not effective in distinguishing between patients with sepsis, severe sepsis and septic shock.

Prediction of intensive care unit mortality: Interest of serum lactates

INTRODUCTION

Many prognostic indices have been developed to assess clinical status and predict the probability of death in the intensive care unit (ICU) but none have perfect sensitivity or specificity.

AIM

To evaluate the prognostic value of admission lactate in patients admitted to ICU.

METHODS

A cohort, observational, prospective study was carried out in the intensive care unit (ICU) of Mongi Slim Hospital, la Marsa, over 12 months period. Arterial blood lactate (ABL) was measured in ICU admission (H0), then 6 hours (H6), 12 hours (H12), 24 hours (H24) and 48hours (H48) after admission. Prognostic scores were calculated 24 hours after the admission. We also recorded biological data, hemodynamic parameters, and the evolution during the stay in intensive care. Primary endpoint was ICU mortality.

RESULTS

We included 135 patients. The average age was 47.22 ± 16.88 years with a sex-ratio of 1.75. ICU mortality was 48%. The mean ABL at admission was 3.05 ± 2.63 mmol/l, higher in the dead group with a statistically significant difference. Prognostic value of lactate at admission was less powerful than severity indices in this study but remains excellent with an AUC >0, 7 defining « cut-off » values with a good sensitivity and specificity. In multivariate analysis, initial lactate > 2 mmol/l was found to be an independent predictive factor of ICU mortality with an Odd Ratio [IC 95%] =1.16 [1.07 - 3.6]; p=0.04.

CONCLUSIONS

Monitoring lactatemia in ICU could allow better identification of patients at high risk of death and the reassessment of therapeutic efficacy.

Can lactate levels and lactate kinetics predict mortality in patients with COVID-19 with using qCSI scoring system?

INTRODUCTION

In this study, we aimed to investigate the relationship between blood lactate levels and lactate kinetics (lactate clearance and Δ lactate) for predicting mortality in patients with COVID-19 admitted to the emergency department.

METHODS

This study was performed as a retrospective study that included patients admitted to the emergency department between March 1st, 2020, and January 1st, 2022. Lactate levels were recorded at the first admission (0 h lactate) and the highest blood lactate levels in the first 24 h of follow-up (2nd highest lactate). Lactate kinetics were calculated. Clinical severity was determined according to the quick COVID Severity Index (qCSI).

RESULTS

300 patients were included in the study. Lactate levels at admission were similar in groups with or without mortality, but 2nd highest lactate levels were found to be significantly higher in the group with mortality (p < 0.001). Lactate clearance and ∆ lactate levels were also found to be lower in the mortality group (p < 0.001). Lactate kinetics in patients in the clinically low severity group were lower in the mortality group (p = 0.02 and p = 0.039, respectively). In the low-intermediate and high-intermediate groups, 0-h lactate and 2nd highest lactate levels were found to be higher in the mortality group, and lactate kinetics were similar in the groups with and without mortality. In the group with high clinical severity, 2nd highest lactate levels were found to be higher in the group with mortality (p = 0.010). Lactate kinetics were also found to be significantly lower in the mortality group (p < 0.001). In the high qCSI group, based on ROC analysis, the AUC for 2nd highest lactate levels predicting mortality was 0.642 (95% CI: 0.548-0.728). The optimal cut-off value for mortality was greater than >2.4 mmol/L (60.6% sensitivity, 67.4% specificity). The AUC for lactate clearance was 0.748 (95% CI: 0.659-0.824). The lactate clearance cut-off value was ≤ -177.78% (49.3% sensitivity, 100% specificity). The AUC for ∆ lactate was 0.707 (95% CI: 0.616-0.787). The optimal ∆ lactate cut-off was ≤ -2 mmol/L (45.1% sensitivity, 93.5% specificity).

CONCLUSION

In COVID-19, 2nd highest blood lactate and lactate kinetics were found to be prognostic indicators of the disease. High 2nd highest lactate levels and low lactate kinetics in patients with high clinical severity were guiding physicians regarding the outcome of the disease.

The prognostic value of delta-lactate in critically ill children

AIM

This study aimed to test delta-lactate (ΔL) as a short-term risk stratification method in critically ill children.

METHODS

An exploratory study of patients admitted to paediatric intensive care unit (PICU) was conducted. ΔL was calculated as the difference between the maximum lactate concentrations on Days 1 and 2. According to the ΔL cutoff, two groups were considered: low mortality risk (LMR) - ΔL ≥ 0.05 mmol/L - and high mortality risk (HMR) - ΔL < 0.05 mmol/L.

RESULTS

Mortality, both during PICU stay and at 28 days, was statistically associated with elevated serum lactate on D1 and D2, per se. For the 93 cases with elevated lactate on Day 1, and a ΔL cutoff of 0.05 mmol/L, the area under the ROC curve was 0.698 (95% confidence interval, 0.47-0.93). HMR patients scored higher PIM3, were not discharged home until 28 days, counted fewer ventilation-free days and needed renal replacement therapy more often.

CONCLUSION

Elevated lactate levels at admission, as well as applying the optimal cutoff for ΔL, allowed to predict short-term mortality: if an increase or minimal decrease in lactate maximum levels occurred from D1 to D2, death was almost eight times more probable. In critically ill children, delta-lactate predicts short-term outcome.

Association Between Lactate and 28-Day Mortality in Elderly Patients with Sepsis: Results from MIMIC-IV Database

INTRODUCTION

This study aimed to explore the association of serum lactate with clinical outcomes in elderly patients with sepsis based on data from the MIMIC-IV database.

METHODS

All elderly patients with sepsis (age ≥ 65 years) were included. Different models were constructed for exploring the relationships between lactate and 28-day mortality. A two-segment linear regression model was performed to verify the threshold effects of lactate on clinical outcomes and smooth curve fitting was performed.

RESULTS

A total of 4199 elderly patients with sepsis were included. The 28-day mortality was 32.22% (n = 1395). After adjustment for all potential cofounders, for each 1 mmol/l increment in lactate, the odds ratio (OR) of 28-day mortality was 1.23 (95% CI 1.18-1.28, P < 0.0001). Smooth fitting curves indicated a non-linear positive relationship between lactate and 28-day mortality. The turning point of lactate level was 5.7 mmol/l: at ≤ 5.7 mmol/l, with each 1 mmol/l increment in lactate, the risk of 28-day mortality increased significantly (OR 1.32, 95% CI 1.25-1.38, P < 0.0001); the significantly positive relationship was still present at lactate > 5.7 mmol/l (OR 1.10, 95% CI 1.04-1.18, P = 0.0019). The area under the ROC curve (AUC) of lactate was 0.618 (95% CI 0.599-0.635) and the cutoff value of lactate was 2.4 mmol/l with a sensitivity of 0.483 and a specificity of 0.687.

CONCLUSION

In elderly patients with sepsis, a non-linear positive relationship was discovered between serum lactate and 28-day mortality. Physicians should be alert to lactate assessment at admission and pay more attention to those patients with higher levels of lactate.

Impact of supplemental thiamin on lactate levels in adults with septic shock

BACKGROUND

Elevated lactate levels at 24 h are highly predictive of in-hospital mortality in adults with septic shock. Thiamin is closely involved in carbohydrate metabolism, and in thiamin-deficient states, increased lactic acid levels can be found, exacerbated by critical illness. This integrative literature review focused on the relationship between supplemental thiamin, lactate clearance, and impact on mortality in sepsis.

METHODS

A search in PubMed, Embase, and CINAHL was conducted for literature published between January 2016 and January 2021. We included observational studies and clinical trials with ≥10 participants. We excluded studies involving pediatric (<18 years old) populations, animal studies, case studies, dropout rate of >20%, nonhospitalized patients, or patients receiving comfort measures only.

RESULTS

A total of 48 full-text articles were assessed for eligibility, with 15 evaluated for this integrative review. Included were five retrospective, two prospective observational, and eight randomized controlled trials. In almost all retrospective studies, thiamin administration was associated with decreased mortality, and in observational studies, with decreased lactate and improved clinical outcomes. In clinical trials, thiamin with or without vitamin C/hydrocortisone did not impact clinical outcomes or mortality. However, four trials testing intravenous thiamin 200-500 mg two to three times daily for up to 3 days reported improved lactate clearance.

CONCLUSIONS

Thiamin supplementation may improve lactate clearance when administered in the first 24 h. Those deficient in thiamin may benefit more from supplementation. The combination of thiamin, vitamin C, and/or hydrocortisone may not be advantageous. Lactate reduction in response to thiamin needs further rigorous research.

28-day sepsis mortality prediction model from combined serial interleukin-6, lactate, and procalcitonin measurements: a retrospective cohort study

Sepsis is a global medical issue owing to its unacceptably high mortality rate. Therefore, an effective approach to predicting patient outcomes is critically needed. We aimed to search for a novel 28-day sepsis mortality prediction model based on serial interleukin-6 (IL-6), lactate (LAC), and procalcitonin (PCT) measurements. We enrolled 367 septic patients based on Sepsis-3 (Third International Consensus Definitions for Sepsis and Septic Shock). Serum IL-6, LAC, and PCT levels were measured serially. Results collected within 24 and 48-72 h of admission were marked as D1 and D3 (e.g., IL-6D1/D3), respectively; the IL-6, LAC, and PCT clearance (IL-6c, LACc, PCTc) at D3 were calculated. Data were split into training and validation cohorts (7:3). Logistic regression analyses were used to select variables to develop models and choose the best one according to the Akaike information criterion (AIC). Receiver operating characteristic curves (ROC), calibration plots, and decision curve analysis (DCA) were used to test model performance. A nomogram was used to validate the model. There were 314 (85.56%) survivors and 53 (14.44%) non-survivors. Logistic regression analyses showed that IL-6D1, IL-6D3, PCTD1, PCTD3, and LACcD3 could be used to develop the best prediction model. The areas under the curves (AUC) of the training (0.849, 95% CI: 0.787-0.911) and validation cohorts (0.828, 95% CI: 0.727-0.929), calibration plot, and the DCA showed that the model performed well. Thus, the predictive value of the risk nomogram was verified. Combining IL-6D1, IL-6D3, PCTD1, PCTD3, and LACcD3 may create an accurate prediction model for 28-day sepsis mortality. Multiple-center research with a larger quantity of data is necessary to determine its clinical utility.

The combination of lactate level, lactate clearance and APACHE II score better predicts short-term outcomes in critically Ill patients: a retrospective cohort study

BACKGROUND

The mortality rate is high in critically ill patients due to the difficulty of diagnosis and treatment. Thus, it is very important to explore the predictive value of different indicators related to prognosis in critically ill patients.

METHODS

This was a retrospective cohort study of patients in the intensive care unit (ICU) of the Sixth People's Hospital in Shanghai, China. A total of 1465 ICU patients had lactate values > 2.1 mmol/L at least once within 24 h of ICU admission, and arterial blood gas was monitored more than twice during the ICU stay.

RESULTS

The predictive value of lactate clearance at 24 h was not high, and the sensitivity and specificity were lower. The predictive value of the lactate level at baseline and the APACHE II score was higher than that of lactate clearance at 24 h in critically ill patients. The predictive value of the lactate level at baseline combined with the APACHE II score was higher than that of the lactate level at baseline or the APACHE II score alone. In addition, the predictive value of lactate clearance at 24 h combined with the APACHE II score was also significantly higher than that of lactate clearance at 24 h or the APACHE II score alone. In particular, the area under the ROC curve reached 0.900, the predictive value was markedly higher than that of the ROC alone, and the sensitivity and specificity were better when these three indicators were combined.

CONCLUSIONS

The combination of lactate level, lactate clearance and APACHE II score better predicts short-term outcomes in critically ill patients.

A simple APACHE IV risk dynamic nomogram that incorporates early admitted lactate for the initial assessment of 28-day mortality in critically ill patients with acute myocardial infarction

BACKGROUND

Early risk stratification is important for patients with acute myocardial infarction (AMI). We aimed to develop a simple APACHE IV dynamic nomogram, combined with easily available clinical parameters within 24 h of admission, thus improving its predictive power to assess the risk of mortality at 28 days.

METHODS

Clinical information on AMI patients was extracted from the eICU database v2.0. A preliminary XGBoost examination of the degree of association between all variables in the database and 28-day mortality was conducted. Univariate and multivariate logistic regression analysis were used to perform screening of variables. Based on the multifactorial analysis, a dynamic nomogram predicting 28-day mortality in these patients was developed. To cope with missing data in records with missing variables, we applied the multiple imputation method. Predictive models are evaluated in three main areas, namely discrimination, calibration, and clinical validity. The discrimination is mainly represented by the area under the receiver operating characteristic curve (AUC), net reclassification improvement (NRI) and integrated discrimination improvement (IDI). Calibration is represented by the calibration plot. Clinical validity is represented by the decision curve analysis (DCA) curve.

RESULTS

A total of 504 people were included in the study. All 504 people were used to build the predictive model, and the internal validation model used a 500-bootstrap method. Multivariate analysis showed that four variables, APACHE IV, the first sample of admission lactate, prior atrial fibrillation (AF), and gender, were included in the nomogram as independent predictors of 28-day mortality in AMI. The prediction model had an AUC of 0.819 (95%CI 0.770-0.868) whereas the internal validation model had an AUC of 0.814 (95%CI 0.765-0.860). Calibration and DCA curves indicated that the dynamic nomogram in this study were reflective of real-world conditions and could be applied clinically. The predictive model composed of these four variables outperformed a single APACHE IV in terms of NRI and IDI. The NRI was 16.4% (95% CI: 6.1-26.8%; p = 0.0019) and the IDI was 16.4% (95% CI: 6.0-26.8%; p = 0.0020). Lactate accounted for nearly half of the total NRI, which showed that lactate was the most important of the other three variables.

CONCLUSION

The prediction model constructed by APACHE IV in combination with the first sample of admission lactate, prior AF, and gender outperformed the APACHE IV scoring system alone in predicting 28-day mortality in AMI. The prediction dynamic nomogram model was published via a website app, allowing clinicians to improve the predictive efficacy of the APACHE IV score by 16.4% in less than 1 min.

Machine Learning Prediction Models for Mortality in Intensive Care Unit Patients with Lactic Acidosis

Background: Lactic acidosis is the most common cause of anion gap metabolic acidosis in the intensive care unit (ICU), associated with poor outcomes including mortality. We sought to compare machine learning (ML) approaches versus logistic regression analysis for prediction of mortality in lactic acidosis patients admitted to the ICU. Methods: We used the Medical Information Mart for Intensive Care (MIMIC-III) database to identify ICU adult patients with lactic acidosis (serum lactate ≥4 mmol/L). The outcome of interest was hospital mortality. We developed prediction models using four ML approaches consisting of random forest (RF), decision tree (DT), extreme gradient boosting (XGBoost), artificial neural network (ANN), and statistical modeling with forward stepwise logistic regression using the testing dataset. We then assessed model performance using area under the receiver operating characteristic curve (AUROC), accuracy, precision, error rate, Matthews correlation coefficient (MCC), F1 score, and assessed model calibration using the Brier score, in the independent testing dataset. Results: Of 1919 lactic acidosis ICU patients, 1535 and 384 were included in the training and testing dataset, respectively. Hospital mortality was 30%. RF had the highest AUROC at 0.83, followed by logistic regression 0.81, XGBoost 0.81, ANN 0.79, and DT 0.71. In addition, RF also had the highest accuracy (0.79), MCC (0.45), F1 score (0.56), and lowest error rate (21.4%). The RF model was the most well-calibrated. The Brier score for RF, DT, XGBoost, ANN, and multivariable logistic regression was 0.15, 0.19, 0.18, 0.19, and 0.16, respectively. The RF model outperformed multivariable logistic regression model, SOFA score (AUROC 0.74), SAP II score (AUROC 0.77), and Charlson score (AUROC 0.69). Conclusion: The ML prediction model using RF algorithm provided the highest predictive performance for hospital mortality among ICU patient with lactic acidosis.

Can capillary lactate improve early warning scores in emergency department? An observational, prospective, multicentre study

AIMS

To determine the prognostic usefulness of the National Early Warning Score-2 (NEWS2) and quick Sepsis-related Organ Failure Assessment (qSOFA) scores, in isolation and combined with capillary lactate (CL), using the new NEWS2-L and qSOFA-L scores to predict the 30-day mortality risk.

METHODS

Prospective, multicentre and observational study in patients across four EDs. We collected sets of vital signs and CL and subsequently calculated NEWS2, qSOFA, NEWS2-L and qSOFA-L scores when patients arrived at the ED. The main outcome measure was all-cause mortality 30 days from the index event.

RESULTS

A total of 941 patients were included. Thirty-six patients (3.8%) died within 30 days of the index event. A high CL level has not been linked to a higher mortality. The NEWS2 presented AUROC of 0.72 (95% CI: 0.62-0.81), qSOFA of 0.66 (95% CI: 0.56-0.77) (P < .001 in both cases) and CL 0.55 (95% CI: 0.42-0.65; P = .229) to predict 30-day mortality. The addition of CL to the scores analysed does not improve the results of the scores used in isolation.

CONCLUSION

NEWS2 and qSOFA scores are a very useful tool for assessing the status of patients who come to the ED in general for all types of patients in triage categories II and III and for detecting the 30-day mortality risk. CL determined systematically in the ED does not seem to provide information on the prognosis of the patients.

Dynamic Transitions of Pediatric Sepsis: A Markov Chain Analysis

Pediatric sepsis is a heterogeneous disease with varying physiological dynamics associated with recovery, disability, and mortality. Using risk scores generated from a sepsis prediction model to define illness states, we used Markov chain modeling to describe disease dynamics over time by describing how children transition among illness states. We analyzed 18,666 illness state transitions over 157 pediatric intensive care unit admissions in the 3 days following blood cultures for suspected sepsis. We used Shannon entropy to quantify the differences in transition matrices stratified by clinical characteristics. The population-based transition matrix based on the sepsis illness severity scores in the days following a sepsis diagnosis can describe a sepsis illness trajectory. Using the entropy based on Markov chain transition matrices, we found a different structure of dynamic transitions based on ventilator use but not age group. Stochastic modeling of transitions in sepsis illness severity scores can be useful in describing the variation in transitions made by patient and clinical characteristics.