The importance of blood lactate clearance as a predictor of early mortality following the modified Norwood procedure

State of the Art of Machine Learning-Enabled Clinical Decision Support in Intensive Care Units: Literature Review

Background

Modern clinical care in intensive care units is full of rich data, and machine learning has great potential to support clinical decision-making. The development of intelligent machine learning–based clinical decision support systems is facing great opportunities and challenges. Clinical decision support systems may directly help clinicians accurately diagnose, predict outcomes, identify risk events, or decide treatments at the point of care.

Objective

We aimed to review the research and application of machine learning–enabled clinical decision support studies in intensive care units to help clinicians, researchers, developers, and policy makers better understand the advantages and limitations of machine learning–supported diagnosis, outcome prediction, risk event identification, and intensive care unit point-of-care recommendations.

Methods

We searched papers published in the PubMed database between January 1980 and October 2020. We defined selection criteria to identify papers that focused on machine learning–enabled clinical decision support studies in intensive care units and reviewed the following aspects: research topics, study cohorts, machine learning models, analysis variables, and evaluation metrics.

Results

A total of 643 papers were collected, and using our selection criteria, 97 studies were found. Studies were categorized into 4 topics—monitoring, detection, and diagnosis (13/97, 13.4%), early identification of clinical events (32/97, 33.0%), outcome prediction and prognosis assessment (46/97, 47.6%), and treatment decision (6/97, 6.2%). Of the 97 papers, 82 (84.5%) studies used data from adult patients, 9 (9.3%) studies used data from pediatric patients, and 6 (6.2%) studies used data from neonates. We found that 65 (67.0%) studies used data from a single center, and 32 (33.0%) studies used a multicenter data set; 88 (90.7%) studies used supervised learning, 3 (3.1%) studies used unsupervised learning, and 6 (6.2%) studies used reinforcement learning. Clinical variable categories, starting with the most frequently used, were demographic (n=74), laboratory values (n=59), vital signs (n=55), scores (n=48), ventilation parameters (n=43), comorbidities (n=27), medications (n=18), outcome (n=14), fluid balance (n=13), nonmedicine therapy (n=10), symptoms (n=7), and medical history (n=4). The most frequently adopted evaluation metrics for clinical data modeling studies included area under the receiver operating characteristic curve (n=61), sensitivity (n=51), specificity (n=41), accuracy (n=29), and positive predictive value (n=23).

Conclusions

Early identification of clinical and outcome prediction and prognosis assessment contributed to approximately 80% of studies included in this review. Using new algorithms to solve intensive care unit clinical problems by developing reinforcement learning, active learning, and time-series analysis methods for clinical decision support will be greater development prospects in the future.

Outcomes of Norwood procedure with hypoplastic left heart syndrome: Our 12-year single-center experience

Background

In this study, we aimed to analyze the predictors and risk factors of mortality in patients who underwent Norwood I procedure with the diagnosis of hypoplastic left heart syndrome.

Methods

Between January 2009 and December 2020, a total of 139 patients (95 males, 44 females) who underwent Norwood I procedure with the diagnosis of hypoplastic left heart syndrome in our center were retrospectively analyzed.

Results

The median birth weight was 3,200 (range, 3,000 to 3,350) g and the median age at the time of operation was seven (range, 5 to 10) days. Pulmonary flow was achieved with a Sano shunt in the majority (72%) of patients. Survival rate was 41% after the first stage. Reoperation for bleeding (p=0.017), reoperation for residual lesion (p=0.011), and postoperative peak lactate level (p=0.029), were associated with in-hospital mortality. Nineteen (33%) of 57 patients died before the second stage. Thirty-three (58%) patients underwent second stage, and survival after the second stage was 94%. Thirteen patients underwent third stage, and survival after the third stage was 85%. Estimated probability of survival at six months, and one, two, three, and four years were 33%, 33%, 25%, 25%, and 22% respectively.

Conclusion

Hospital and inter-stage mortality rates are still high and this seems to be the most challenging period in term of survival efforts of the patients with hypoplastic left heart syndrome. Early recognition and reintervention of anatomical residual defects, close follow-up in the inter-stage period, and the accumulation of multidisciplinary experience may help to improve the results to acceptable limits.

Postoperative Serum Lactate Levels for In-Hospital Mortality Prediction Among Heart Transplant Recipients

Background

Hyperlactatemia is a common phenomenon following cardiac surgeries and is associated with prolonged ICU stay and higher morbidity and mortality rates, but such analyses have never focused on patients undergoing heart transplantation (HTX), in whom hyperlactatemia defined with the traditional threshold is observed in nearly every individual. The present study aimed to assess the prognostic value and clinical usefulness of postoperative serum lactate level measurements for in-hospital mortality prediction following HTX.

Material/Methods

Forty-six consecutive patients who underwent HTX in the Department of Cardiovascular Surgery and Transplantology between 2010 and 2015 were enrolled into a retrospective analysis. Serum lactate level measurements within the first 48 h after HTX were obtained from arterial blood gas analyses, that were routinely conducted every 6 h. Lactate clearance was determined for each patient individually throughout 3 different time frames: the first 24-h (Lac clear 0–24) and second 24-h period (Lac clear 24–48), and the first 48 h after surgery (Lac clear 0–48).

Results

The ICU admission serum lactate levels differed between the deceased and survivors (7.6 vs. 4.3 mmol/L; p=0.000). Among all tested postoperative lactate level measurements, only the measurement taken upon ICU admission predicted in-hospital mortality (OR 1.94 95% CI [1.09–3.43]; p=0.024). The receiving operating characteristic (ROC) curve for in-hospital mortality was constructed for ICU admission measurement, with the optimal cut-off point estimated at 7.0 mmol/L.

Conclusions

Serum lactate level measurement upon ICU admission can be used as a predictive parameter for in-hospital mortality among heart transplant recipients. Values greater than 7.0 mmol/L can predict in-hospital mortality with 90% accuracy.

Prognostic Value of Blood Lactate and Lactate Clearance in Refractory Cardiac Arrest Treated by Extracorporeal Life Support

OBJECTIVE

During cardiac arrest (CA) resuscitation, an 'ischaemia-reperfusion' syndrome occurs leading to multiorgan failure reflected by an increase in blood lactate. Blood lactate is a diagnosis and prognosis biomarker in extracorporeal life support (ECLS), but its kinetic appears more informative to assess a patient's outcome. The aim of the present study was to describe the prognostic value of blood lactate and lactate clearance (LC) 3 (H3) and 6 h (H6) after the initiation of ECLS in the treatment of refractory CA.

METHODS

Patients admitted to the intensive care unit for refractory CA were included. Lactate measurements were performed at the initiation of ECLS (H0) and at H3 and H6 upon the initiation of ECLS. LC was measured from 0 to 3 h (LC03), 0 to 6 h (LC06) and 3 to 6 h (LC36). The primary endpoint was in-hospital mortality within 28 days.

RESULTS

Sixty-six patients were enrolled in the study. Lactate levels were higher in deceased patients. Increased mortality was observed with increasing levels of lactate at H3 and H6 and with decreasing LC03. Using logistic regression, an association was observed between mortality and lactate at H3 with an odds ratio (OR) of 1.21 (95% confidence interval (CI) 1.05-1.42); LC03, OR of 0.93 (95% CI 0.87-0.99) and LC06, OR of 0.96 (95% CI 0.92-0.99).

CONCLUSION

Blood lactate and LC within the first 3 h of ECLS in refractory CA are associated with mortality. LC is a more relevant parameter than blood lactate, taking into account both the production and elimination of lactate. We suggest to preferentially use LC to assess the patient's outcome.

Effect of acid-base balance on postoperative course in children with hypoplastic left heart syndrome after the modified Norwood procedure

Hypoplastic left heart syndrome (HLHS) is a congenital heart defect that requires 3-stage cardiac surgical treatment and multidirectional specialist care. The condition of newborns in the first postoperative days following the modified Norwood procedure is characterized by considerable hemodynamic instability that may result in a sudden cardiac arrest. It is believed that the most important cause of hemodynamic instability is the fluctuations in redistribution between pulmonary and systemic blood flow.The paper analyzes the postoperative course in 40 neonates with HLHS following the modified Norwood procedure performed under deep hypothermic cardiopulmonary bypass hospitalized in Cardiac Surgical Intensive Care Unit (CSICU) in the years 2014-2015. For all hospitalized children, the arterial blood acid-base balance (ABB) parameters (pH, pO2, pCO2, HCO3, base excess (BE), and lactic acid) were measured 2 times a day during the first 5 postoperative days. The main goal of the studies is to analysis of ABB parameters and their influence on the clinical state of newborns with HLHS. Several descriptors were concerned to describe the neonates clinical state: the date of the surgery (the day of life when the child was operated on), the duration (number of days) of mechanical ventilation employment, the time of hospitalization in intensive care unit, and the total duration of treatment in CSICU.The statistical analysis of the particular ABB parameters revealed a significant dependence (P < .001) between the values of pH, pO2, pCO2, HCO3, BE, lactic acid, and all concerned descriptors of the newborn clinical state.The article shows that monitoring the ABB parameters, proper interpretation of the results, and appropriate modification of pharmacotherapy and respiratory treatment are crucial for therapeutic results and survival rates in neonates with HLHS after the modified Norwood procedure.

Intraoperative change of lactate level is associated with postoperative outcomes in pediatric cardiac surgery patients: retrospective observational study

Background

A change of serum lactate concentrations appeared to be useful for predicting outcomes in various acute ill settings. However, there is little information on intraoperative change of lactate level in pediatric cardiac surgery patients.

Methods

We conducted a retrospective observational study of 459 children who received pediatric cardiac surgery to determine the association between change of lactate level after cardiopulmonary bypass (CPB) and patient prognosis (length of ICU stay and incidence of postoperative serious adverse events (SAEs)). We defined change of lactate level after CPB (LAC⊿) as (final lactate level measurement in the operating room) – (lactate level measured at the end of CPB). To study the independent association of LAC⊿ with length of ICU stay, we used linear regression model.

Results

There were 1145 lactate measurements after CPB in this study cohort. After weaning from CPB, the serum lactate levels significantly increased from 2.1 mmol/L to 2.5 mmol/L (p < 0.001). Patients with higher LAC⊿ had significantly longer stay in ICU (p = 0.017) and higher incidence of SAEs (p = 0.002). In multivariate linear regression analysis, higher LAC⊿ showed a significant independent association with longer length of ICU stay.

Conclusions

Increased lactate level after CPB was associated with the longer duration of ICU stay and increased risk of postoperative SAEs in pediatric cardiac surgery patients. Future studies should be conducted to determine the clinical utility of intraoperative trend of lactate levels.

Electronic supplementary material

The online version of this article (doi:10.1186/s12871-015-0007-y) contains supplementary material, which is available to authorized users.

Predictive value of lactate in unselected critically ill patients: an analysis using fractional polynomials

BACKGROUND AND OBJECTIVES

Hyperlactatemia has long been associated with poor clinical outcome in varieties of intensive care unit (ICU) patients. However, the impact of temporal changes in lactate has not been well established and there are some shortcomings in model building in previous studies. The present study aims to investigate the association of initial lactate and normalization time with hazard by using fractional polynomial Cox proportional hazard model.

METHODS

A large clinical database named Multiparameter Intelligent Monitoring in Intensive Care II (MIMIC-II) was employed for analysis. Demographics, comorbidities, laboratory findings were extracted and were compared between survivors and non-survivors by using univariable analysis. Cox proportional hazard model was built by purposeful selection of covariate with initial lactate (L0) and normalization time (T) remaining in the model. Best fit model was selected by using deviance difference test and comparisons between fractional polynomial regression models of different degrees were performed by using closed test procedure.

MAIN RESULTS

A total of 6,291 ICU patients were identified to be eligible for the present study, including 1,675 non-survivors and 4,616 survivors (mortality rate: 26.6%). Patients with lactate normalization had significantly reduced hazard rate as compared to those without normalization (log-rank test: P<0.05). The best powers of L0 in the model were -2 and -1 with the deviance of 19,944.51, and the best powers of T were 0.5 and 3 with the deviance of 7,965.63. The adjusted hazard ratio for the terms L0(-2) and L0(-1) were 1.13 (95% CI: 1.09-1.18) and 0.43 (95% CI: 0.34-0.54); and the adjusted hazard ratio for the terms T(0.5) and T(3) were 7.42 (95% CI: 2.85-19.36) and 3.06×10(-6) (95% CI: 3.01×10(-11)-0.31).

CONCLUSIONS

Initial lactate on ICU admission is associated with death hazard and the relationship follows a fractional polynomial pattern with the power of -2 and -1. Delayed normalization of lactate is predictive of high risk of death when it is measured within 150 hours after ICU admission.

Serum arterial lactate concentration predicts mortality and organ dysfunction following liver resection

BACKGROUND

The aim of this study was to determine if the post-operative serum arterial lactate concentration is associated with mortality, length of hospital stay or complications following hepatic resection.

METHODS

Serum lactate concentration was recorded at the end of liver resection in a consecutive series of 488 patients over a seven-year period. Liver function, coagulation and electrolyte tests were performed post-operatively. Renal dysfunction was defined as a creatinine rise of >1.5x the pre-operative value.

RESULTS

The median lactate was 2.8 mmol/L (0.6 to 16 mmol/L) and was elevated (≥2 mmol/L) in 72% of patients. The lactate concentration was associated with peak post-operative bilirubin, prothrombin time, renal dysfunction, length of hospital stay and 90-day mortality (P < 0.001). The 90-day mortality in patients with a post-operative lactate ≥6 mmol/L was 28% compared to 0.7% in those with lactate ≤2 mmol/L. Pre-operative diabetes, number of segments resected, the surgeon's assessment of liver parenchyma, blood loss and transfusion were independently associated with lactate concentration.

CONCLUSIONS

Initial post-operative lactate concentration is a useful predictor of outcome following hepatic resection. Patients with normal post-operative lactate are unlikely to suffer significant hepatic or renal dysfunction and may not require intensive monitoring or critical care.

Development of the Post Cardiac Surgery (POCAS) prognostic score

Introduction

The risk of mortality in cardiac surgery is generally evaluated using preoperative risk-scale models. However, intraoperative factors may change the risk factors of patients, and the organism functionality parameters determined upon ICU admittance could therefore be more relevant in deciding operative mortality. The goals of this study were to find associations between the general parameters of organism functionality upon ICU admission and the operative mortality following cardiac operations, to develop a Post Cardiac Surgery (POCAS) Scale to define operative risk categories and to validate an operative mortality risk score.

Methods

We conducted a prospective study, including 920 patients who had undergone cardiac surgery with cardiopulmonary bypass. Several parameters recorded on their ICU admission were explored, looking for a univariate and multivariate association with in-hospital mortality (90 days). In-hospital mortality was 9%. Four independent factors were included in the POCAS mortality risk model: mean arterial pressure, bicarbonate, lactate and the International Normalized Ratio (INR). The POCAS scale was compared with four other risk scores in the validation series.

Results

In-hospital mortality (90 days) was 9%. Four independent factors were included in the POCAS mortality risk model: mean arterial pressure, bicarbonate ratio, lactate ratio and the INR. The POCAS scale was compared with four other risk scores in the validation series. Discriminatory power (accuracy) was defined with a receiver-operating characteristics (ROC) analysis. The best accuracy in predicting in-hospital mortality (90 days) was achieved by POCAS. The areas under the ROC curves of the different systems analyzed were 0.890 (POCAS), followed by 0.847 (Simplified Acute Physiology Score (SAP II)), 0.825 (Sepsis-related Organ Failure Assessment (SOFA)), 0.768 (Acute Physiology and Chronic Health Evaluation (APACHE II)), 0.754 (logistic EuroSCORE), 0.714 (standard EuroSCORE) and 0.699 (Age, Creatinine, Ejection Fraction (ACEF) score).

Conclusions

Our new system to predict the operative mortality risk of patients undergoing cardiac surgery is better than others used for this purpose (SAP II, SOFA, APACHE II, logistic EuroSCORE, standard EuroSCORE, and ACEF score). Moreover, it is an easy-to-use tool since it only requires four risk factors for its calculation.

Stage one Norwood procedure in an emerging economy:Initial experience in a single center

Objective:

The evolution of surgical skills and advances in pediatric cardiac intensive care has resulted in Norwood procedure being increasingly performed in emerging economies. We reviewed the feasibility and logistics of performing stage one Norwood operation in a limited-resource environment based on a retrospective analysis of patients who underwent this procedure in our institution.

Methods:

Retrospective review of medical records of seven neonates who underwent Norwood procedure at our institute from October 2010 to August 2012.

Results:

The median age at surgery was 9 days (range 5-16 days). All cases were done under deep hypothermic cardiopulmonary bypass and selective antegrade cerebral perfusion. The median cardiopulmonary bypass (CPB) time was 240 min (range 193-439 min) and aortic cross-clamp time was 130 min (range 99-159 min). A modified Blalock-Taussig (BT) shunt was used to provide pulmonary blood flow in all cases. There were two deaths, one in the early postoperative period. The median duration of mechanical ventilation was 117 h (range 71-243 h) and the median intensive care unit (ICU) stay was 12 days (range 5-16 days). Median hospital stay was 30.5 days (range 10-36 days). Blood stream sepsis was reported in four patients. Two patients had preoperative sepsis. One patient required laparotomy for intestinal obstruction.

Conclusions:

Stage one Norwood is feasible in a limited-resource environment if supported by a dedicated postoperative intensive care and protocolized nursing management. Preoperative optimization and prevention of infections are major challenges in addition to preventing early circulatory collapse.

The effect of morphologic subtype on outcomes following the Sano-Norwood procedure

OBJECTIVES

Controversy exists concerning outcomes of patients with different morphologic subtypes of hypoplastic left heart syndrome undergoing the Norwood procedure, in particular, aortic atresia-mitral stenosis (AA-MS) patients receiving a systemic-pulmonary modified Blalock-Taussig (mBT) shunt. We sought to determine the influence of known risk factors and morphology on early survival in our cohort of Sano-Norwood patients with right ventricle-pulmonary artery (RV-PA) conduits as the source of pulmonary blood flow.

METHODS

We studied early survival in patients with Sano-modified Norwood procedures from 2002 to 2010 and included both typical and atypical (including unbalanced atrioventricular canal defect) morphologic variants. We included a comparison group composed of classical Norwood patients with mBT shunts.

RESULTS

Of 264 Sano-Norwood patients, 49 had AA-MS and 30 had atypical morphologies. Weight ≤ 2.5 kg was associated with a poorer 1-year survival (P = 0.0007), though ascending aorta (AscAo) size <2.0 mm was not. We did not observe a difference in 30-day or 1-year Kaplan-Meier (K-M) survival across typical morphologic variants for either a Sano or mBTS group. Atypical variants exhibited a trend towards lower 30-day and 1-year survival in both Sano and mBTS groups though this was not significant. Direct comparison of 30-day K-M survival for Sano versus mBTS in AA-MS patients showed similar outcomes (P = NS).

CONCLUSIONS

Use of the RV-PA conduit results in good early survival, even in those with a small AscAo size. Atypical morphologic variants seem to do worse irrespective of the Sano or mBTS group. Further studies will be required to determine conclusively whether the RV-PA shunt confers protective effects in the AA-MS subgroup compared with the mBTS.

Arterial switch in the first hours of life: no need for Rashkind septostomy?

OBJECTIVES

We describe our experience of arterial switch operation (ASO) in the first hours of life in children with D-transposition of the great arteries (TGA) intact ventricular septum and restrictive patent foramen ovale (PFO) with severe metabolic disorders. The Rashkind procedure is the usual strategy to stabilize critically ill patient before surgery. We assume that primary ASO is also the safe option.

METHODS

Between 2008 and 2010, 44 neonates underwent ASO. Of these, eight with simple D-TGA had severe acid-base disorders and form the subject of this report. All were operated on within the first 24 h of life. Median time from birth to admission to the hospital was 4 h 52 min. All patients had prenatal diagnoses. In six cases, childbirths were held in a maternity hospital near the heart surgery centre. In two cases, patients were transported from remote hospitals. All patients received prostaglandin E(l), three of them required ventilator support. All patients had small PFO (1-3 mm), oxygen saturation 60-85%. All had metabolic acidosis (pH ranging 6.96-7.2); lactate was increased in all patients (ranging 5.5-17.0 mmol/l).

RESULTS

All patients underwent primary ASO. We used normothermic cardiopulmonary bypass (CPB). Median CPB time was 143 min, and median ischaemic time was 68 min. All patients had uneventful postoperative periods. There were no in-hospital deaths and no late deaths after hospital discharge.

CONCLUSIONS

ASO for D-TGA in the first hours of life is a good option for the treatment of critically ill neonates with metabolic disorders. In our experience, there is no need for the Rashkind procedure to stabilize the patient before surgery.

Temporary restriction of right ventricle-pulmonary artery conduit flow using haemostatic clips following Norwood I reconstruction: potential for improved outcomes

Improved outcomes of the Norwood procedure in hypoplastic left heart syndrome have been achieved by the manipulation of the pulmonary:systemic flow ratio (Qp:Qs) in the early post-operative period, with focus on improving systemic perfusion. As an extension of this Qp:Qs-limiting strategy, we evolved a novel surgical technique to achieve transient flow restriction in the right ventricle-pulmonary artery (RV-PA) conduit for the first 48 h, using haemostatic clips, in a cohort of patients and compared the early outcomes with a non-clipped cohort. Clips were subsequently removed at 48 h at the time of definitive chest closure. We performed RV-PA shunt flow clipping in 37 patients; 75 historical controls had not received clips. Groups were matched for weight, size of ascending aorta, anatomy and circulatory arrest times. Thirty-day mortality was lower in the clipped cohort (2 of 37; 5.4%) versus the unclipped cohort (10 of 75; 13.3%). The minimum blood lactate levels within the first 24 h post-surgery in the unclipped group were significantly higher (P = 0.049), with a significantly lower Qp:Qs in the first 6 h in the clipped patients. These data suggest that limiting Qp:Qs in the early post-operative period following the Norwood procedure may help in the post-operative management of these difficult patients. These results warrant further study.