APACHE II score cannot predict successful weaning from prolonged mechanical ventilation

Machine learning prediction of the total duration of invasive and non-invasive ventilation During ICU Stay

Predicting the duration of ventilation in the ICU helps in assessing the risk of ventilator-induced lung injury, ensuring sufficient oxygenation, and optimizing resource allocation. Prior models provided a prediction of total duration without distinguishing between invasive and non-invasive ventilation. This work proposes two independent gradient boosting regression models for predicting the duration of invasive and non-invasive ventilation based on commonly available ICU features. These models are trained on 2.6 million patient stays across 350 US hospitals between 2010 to 2019. The mean absolute error (MAE) for the prediction of duration was 2.08 days for invasive ventilation and 0.36 days for non-invasive ventilation. The total ventilation duration predicted by our model had MAE of 2.38 days, which outperformed the gold standard (APACHE) with MAE of 3.02 days. The feature importance analysis of the trained models showed that, for invasive ventilation, high average heart rate, diagnosis of respiratory infection and admissions from locations other than the operating room were associated with longer ventilation durations. For non-invasive ventilation, higher respiratory rates and having any GCS measurement were associated with longer durations.

Analysis of risk factors of prolonged mechanical ventilation in patients with severe burn injury

BACKGROUND

Mechanical ventilation is an essential means of life support for patients with severe burns. However, prolonged mechanical ventilation (PMV) increases the incidence of complications and length of hospital stay. Therefore, studying the risk factors of mechanical ventilation duration is of great significance for reducing the duration of mechanical ventilation, reducing related complications, and improving the success rate of severe burn treatment.

METHOD

This study was a retrospective study of patients with burns ≥30% of the area admitted to the BICU of Guangzhou Red Cross Hospital affiliated with Jinan University from January 2016 to January 2023 who were mechanically ventilated. Patients were classified into the prolonged mechanical ventilation group if they were mechanically ventilated for ≥21 days. Then, independent risk factors for prolonged mechanical ventilation were determined by logistic regression analysis of the collected data.

RESULT

Of all the 112 enrolled patients, 79 had prolonged mechanical ventilation, with an incidence of 70.5%. Logistic regression analysis revealed that including abbreviated burn severity index (ABSI%) (P < 0.001), moderate and severe inhalation injury (P = 0.005, P = 0.044), albumin (P = 0.032), lactic acid (P < 0.001) were independent risk factors for prolonged mechanical ventilation. In addition, ventilator-related complications were 44% in the PMV group and 21% in the non-PMV group.

CONCLUSION

ABSI%, inhalation injury, albumin, and lactic acid on admission are the risk factors for PMV in severe burn patients. In addition, ventilator-related complications were higher in group PMV than in group non-PMV in our study.

Prediction prolonged mechanical ventilation in trauma patients of the intensive care unit according to initial medical factors: a machine learning approach

The goal of this study was to develop a predictive machine learning model to predict the risk of prolonged mechanical ventilation (PMV) in patients admitted to the intensive care unit (ICU), with a focus on laboratory and Arterial Blood Gas (ABG) data. This retrospective cohort study included ICU patients admitted to Rajaei Hospital in Shiraz between 2016 and March 20, 2022. All adult patients requiring mechanical ventilation and seeking ICU admission had their data analyzed. Six models were created in this study using five machine learning models (PMV more than 3, 5, 7, 10, 14, and 23 days). Patients' demographic characteristics, Apache II, laboratory information, ABG, and comorbidity were predictors. This study used Logistic regression (LR), artificial neural networks (ANN), support vector machines (SVM), random forest (RF), and C.5 decision tree (C.5 DT) to predict PMV. The study enrolled 1138 eligible patients, excluding brain-dead patients and those without mechanical ventilation or a tracheostomy. The model PMV > 14 days showed the best performance (Accuracy: 83.63-98.54). The essential ABG variables in our two optimal models (artificial neural network and decision tree) in the PMV > 14 models include FiO₂, paCO₂, and paO₂. This study provides evidence that machine learning methods outperform traditional methods and offer a perspective for achieving a consensus definition of PMV. It also introduces ABG and laboratory information as the two most important variables for predicting PMV. Therefore, there is significant value in deploying such models in clinical practice and making them accessible to clinicians to support their decision-making.

Inspiratory Muscle Training Strategies in Tracheostomized Critically Ill Individuals

BACKGROUND

Inspiratory muscle training (IMT) strategies can reduce ICU length of stay and optimize recovery in critically ill patients. Our objective was to compare IMT combined with spontaneous breathing with T-piece in tracheostomized subjects.

METHODS

Tracheostomized critically ill subjects who were ready to wean were selected and randomly allocated to one of 2 groups: electronically-assisted IMT (EIMT) or spontaneous breathing with T-piece. Electronically assisted IMT was delivered using 30% of maximal inspiratory pressure (manual EIMT or automatically adjusted loads). The following variables were analyzed: ICU length of stay, weaning time, maximal inspiratory pressure, rapid shallow breathing index, pressure (cm H₂O), power (W), flow (L/s), volume (L), and energy (J).

RESULTS

A total of 132 patients were assessed; 104 subjects were enrolled with EIMT, n = 51 (automatic EIMT, n = 25 and manual EIMT n = 26), or spontaneous breathing with T-piece group, n = 53. The Acute Physiology and Chronic Health Evaluation II score was significantly higher (P = .02) in subjects in the manual EIMT group. Weaning time did not differ significantly between groups (8.55 ± 6.48 d and 10.86 ± 6.48 d, EIMT and spontaneous breathing with T-piece group, respectively; P = .23). Weaning success rates (75%) were lower in the manual EIMT group. Invasive mechanical ventilation time was longer but not significantly different (P = .21) in the spontaneous breathing with T-piece group. Maximal inspiratory pressure was significantly higher in the spontaneous breathing with T-piece and the automatic EIMT groups (P < .001 and P = .007, respectively). Pressure, power, and energy values were significantly higher in the manual EIMT group (P < .001, P = .003, and P = .003, respectively).

CONCLUSIONS

IMT modalities in this trial had no significant impacts on weaning time or successful weaning rates.

Predictive Value of Sequential Organ Failure Assessment (SOFA), Quick Sequential Organ Failure Assessment (qSOFA), Acute Physiology and Chronic Health Evaluation (APACHE II), and New Early Warning Signs (NEWS-2) Scores Estimate Mortality of COVID-19 Patients Requiring Intensive Care Unit (ICU)

Introduction

Various mortality predictive score models for coronavirus disease-2019 (COVID-19) have been deliberated. We studied how sequential organ failure assessment (SOFA), quick sequential organ failure assessment (qSOFA), acute physiology and chronic health evaluation II (APACHE II), and new early warning signs (NEWS-2) scores estimate mortality in COVID-19 patients.

Materials and methods

We conducted a prospective cohort study of 53 patients with moderate-to-severe COVID-19. We calculated qSOFA, SOFA, APACHE II, and NEWS-2 on initial admission and re-evaluated on day 5. We performed logistic regression analysis to differentiate the predictors of qSOFA, SOFA, APACHE II, and NEWS-2 scores on mortality.

Result

qSOFA, SOFA, APACHE II, and NEWS-2 scores on day 5 exhibited a difference between survivors and nonsurvivors (p <0.05), also between ICU and non-ICU admission (p <0.05). The initial NEWS-2 revealed a higher AUC value than the qSOFA, APACHE II, and SOFA score in estimating mortality (0.867; 0.83; 0.822; 0.794). In ICU, APACHE II score revealed a higher AUC value than the SOFA, NEWS-2, and qSOFA score (0.853; 0.832; 0.813; 0.809). Concurrently, evaluation on day 5 showed that qSOFA AUC had higher scores than the NEWS-2, APACHE II, and SOFA (0.979; 0.965; 0.939; 0.933) in predicting mortality, while SOFA and APACHE II AUC were higher in ICU admission than NEWS-2 and qSOFA (0.968; 0.964; 0.939; 0.934). According to the cutoff score, APACHE II on day 5 revealed the highest sensitivity and specificity in predicting the mortality (sensitivity 95.7%, specificity 86.7%).

Conclusion

All scores signify good predictive values on COVID-19 patients mortality following the evaluation on the day 5. Nonetheless, APACHE-II appears to be the best at predicting mortality and ICU admission rate.

How to cite this article

Asmarawati TP, Suryantoro SD, Rosyid AN, Marfiani E, Windradi C, Mahdi BA, et al. Predictive Value of Sequential Organ Failure Assessment, Quick Sequential Organ Failure Assessment, Acute Physiology and Chronic Health Evaluation II, and New Early Warning Signs Scores Estimate Mortality of COVID-19 Patients Requiring Intensive Care Unit. Indian J Crit Care Med 2022;26(4):464–471.

Ventilation Management and Outcomes for Subjects With Neuromuscular Disorders Admitted to ICUs With Acute Respiratory Failure

BACKGROUND

Patients with neuromuscular disorders (NMD) share the risk of acute respiratory failure (ARF) leading to ICU admissions. Noninvasive ventilation (NIV) is often proposed as an alternative to invasive ventilation. This study describes clinical features, ventilation management, and outcomes of subjects with NMD admitted to ICU and managed for ARF.

METHODS

We performed a multicenter retrospective study in 7 adult ICUs in the Auvergne-Rhone-Alpes area in France involving subjects with NMD admitted to the ICU for ARF. The primary end point was ICU mortality. Secondary end points were NIV failure, weaning from invasive ventilation, and long-term mortality. We hypothesized a poorer outcome in the case of bulbar musculature involvement.

RESULTS

A total of 242 subjects were included; 142 subjects had nonhereditary NMD (58.7%), and 100 had hereditary NMD (41.3%). Eleven subjects had home ventilation through a tracheostomy. While 112 were intubated at admission, 119 initially underwent NIV. NIV was successful in avoiding orotracheal intubation in 78 subjects (65.5%). ICU mortality was 13.6%. Factors associated with ICU mortality were nonhereditary NMD and requirement for invasive ventilation. The involvement of bulbar musculature in ARF and hereditary NMD were associated with NIV failure. After a median follow-up of 1.2 y, 53 of 209 subjects had died.

CONCLUSIONS

The ICU mortality of NMD subjects with ARF was low, with no impact of bulbar muscles involvement. NIV was proposed for approximately half of the subjects, and it was more effective when ARF was not attributed to bulbar musculature involvement. The long-term outcome was good.

Plasma concentrations of NOX4 are predictive of successful liberation from mechanical ventilation and 28-day mortality in intubated patients

Background

Nicotinamide adenine dinucleotide phosphate (NADPH) oxidase (NOX) enzymes play important roles in generating reactive oxygen species (ROS); in particular, NOX4 plays a distinct role in regulating lung inflammation and apoptosis.

Methods

We determined whether plasma NOX4 level can be used as a prognostic biomarker to guide weaning from mechanical ventilation and to predict mortality in intubated patients. Plasma levels of NOX4 were measured at days 1 (NOX4 D1) and 7 (NOX4 D7) after initiation of mechanical ventilation in 184 patients.

Results

With increase in day 7 NOX4 quartile, the success of weaning tended to decrease and 28-day mortality tended to increase. On multivariate logistic regression, Acute Physiology, Age, Chronic Health Evaluation II (APACHE II) [odds ratio (OR): 1.10; 95% CI, 1.02–1.18], duration of mechanical ventilation (OR: 1.12; 95% CI: 1.06–1.18), and NOX4 D7 levels >18.2 ng/mL (OR: 4.40; 95% CI: 1.91–10.06) were independently associated with weaning failure. Also, Cox-hazard proportional model showed that NOX4 D7 level >18.2 ng/mL (hazard ratio [HR], 2.29; 95% CI, 1.26–4.16), APACHE II (HR: 1.07; 95% CI: 1.02–1.14), Sequential Organ Failure Assessment (SOFA) (HR: 1.10; 95% CI: 1.01–1.20) and coexisting cancer (HR: 1.99; 95% CI, 1.01–3.94), were independently associated with 28-day mortality. The longitudinal trend of NOX4 level varied according to the clinical outcomes.

Conclusions

An increased plasma NOX4 D7 level was associated with weaning failure and 28-day mortality in patients with mechanical ventilation. Our results suggest that NOX4-directed management may lead to improved outcomes in patients with mechanical ventilation.

Association of weaning failure from mechanical ventilation with transthoracic echocardiography parameters: a systematic review and meta-analysis

BACKGROUND

Weaning from mechanical ventilation is a challenging step during recovery from critical illness. Weaning failure or early reintubation are associated with increased morbidity and mortality, exposing patients to life-threatening complications. Cardiac dysfunction represents the most common cause of weaning failure. We conducted a systematic review and meta-analysis to evaluate the association between transthoracic echocardiographic parameters and weaning failure.

METHODS

We performed a systematic search of MEDLINE and EMBASE screening for prospective studies providing echocardiographic data collected just before the beginning of spontaneous breathing trial and outcome of the weaning attempt. We primarily focused on parameters currently recommended for evaluation of left ventricular (LV) systolic or diastolic dysfunction.

RESULTS

We included 11 studies in our primary analysis, which included data on LV ejection fraction (LVEF, n=10 studies) and parameters recommended for the assessment of LV diastolic function (E/e' ratio n=10; E/A ratio n=9; E wave n=8; and e' wave n=7). Weaning failure was significantly associated to a higher E/e' ratio (standardised mean difference [SMD]=1.70, 95% confidence interval [CI; 0.78-2.62]; P<0.001), lower e' wave (SMD=-1.22, 95% CI [-2.33 to -0.11]; P=0.03), and higher E wave (SMD=0.97, 95% CI [0.29-1.65]; P=0.005). We found no association between weaning failure and LVEF (SMD=-0.86, 95% CI [-1.92-0.20]; P=0.11) and E/A ratio (SMD=0.00, 95% CI [-0.30-0.31]; P=0.98).

CONCLUSIONS

Weaning failure is associated with parameters indicating worse LV diastolic function (E/e', e' wave, E wave) and increased LV filling pressure (E/e' ratio). The association between weaning failure and LV systolic dysfunction as evaluated by LVEF is more unclear. More studies are needed to clarify this aspect and regarding the role of right ventricular function.

The Impact of Tracheostomy Timing on the Duration and Complications of Mechanical Ventilation

Background:

Mechanical ventilation is a life support for ICU patients and is indicated in case of acute or chronic respiratory failure. 75% of patients admitted to ICU require this support and most of them stay on prolonged MV. Tracheostomy plays a fundamental role in airway management, facilitating ventilator weaning and reducing the duration of MV. Early tracheostomy is defined when the procedure is conducted up to 10 days after the beginning of MV and late tracheostomy when the procedure is performed after this period. Controversy still exists over the ideal timing and classification of early and late tracheostomy.

Objective:

Evaluate the impact of timing of tracheostomy on ventilator weaning.

Method:

Single-center retrospective study. Patients were divided into three groups: very early tracheostomy (VETrach), intermediate (ITrach) and late (LTrach): >10 days.

Results:

One hundred two patients were included: VETrach (n=21), ITrach (n=15), and LTrach (n=66). ITrach group had lower APACHE II (p=0.004) and SOFA (p≤0.001). Total ICU length of stay, and incidence of post-tracheostomy ventilator-associated pneumonia were significantly lower in the VETrach and ITrach groups. The GCS and RASS scores improved in all groups, while the maximal inspiratory pressure and rapid shallow breathing index showed a tendency towards improvement on discharge from the ICU.

Conclusion:

Very early tracheostomy did not reduce the duration of MV or length of ICU stay after the procedure when compared to late tracheostomy, but was associated with low rates of ventilatorassociated pneumonia. Neurological patients benefitted more from tracheostomy, particularly very early and intermediate tracheostomy.

Predictors of prolonged mechanical ventilation in patients admitted to intensive care units: A systematic review

OBJECTIVE

Although intensive care medicine has evidenced a significant growth in recent decades, the number of patients requiring prolonged mechanical ventilation (PMV) still represents a considerable burden on health-care expenditure. The prediction of the need for PMV seems to provide a plausible cost-effective intervention. The objective of this study is to systematically review the predictors of the need for PMV of adult patients admitted to intensive care units (ICUs) due to medical and surgical needs.

METHODS

We conducted a systematic search on three online databases (PubMed, Embase, and MEDLINE) till February 20, 2019. The search process employed several combinations of specific keywords and Boolean operators.

RESULTS

A total of 15 articles were included in the study. Based on pooling the outcomes of odds ratios (ORs) and their respective 95% confidence intervals (CIs) as reported from logistic regression analyses, the pooled PMV incidence in 8220 patients (69.59% males) was 17.67 cases per 100 ICU admissions (95% CI 13.69-21.65). We could not conduct a meta-analysis of ORs and 95% CIs due to the significant heterogeneity observed between the included studies (P < 0.001, I2 = 97%). Pre-operative/preadmission kidney dysfunction and chronic obstructive pulmonary disease were the most significant independent predictors of the need for PMV. Following cardiac surgeries, repeated or emergency surgery, prolonged cardiopulmonary bypass time, and the need for blood transfusion were predictors of the need for PMV.

CONCLUSION

Within the study limitations, several predictors were identified, which could be further investigated using a unified PMV definition. Successful prediction of the need for PMV would assist clinicians in identifying and adjusting a "weaning strategy" as well as improving patient care to reduce morbidity. Furthermore, establishing specialized weaning units could be warranted based on PMV incidence and prediction in the local settings.

Prognostic factors associated with mortality in mechanically ventilated patients in the intensive care unit: A single-center, retrospective cohort study of 905 patients

Data on outcomes of patients receiving mechanical ventilation (MV) in China are scarce.To investigate factors associated with the prognosis of patients given MV in the intensive care unit (ICU).A 12-year (January 1, 2006-December 31, 2017) retrospective cohort study.ICU of Beijing Geriatric Hospital, China.A total of 905 patients aged ≥16 years given MV during the study period.None.Among 905 patients included (610 men; median age, 78 years; Acute Physiology and Chronic Health Evaluation [APACHE]-II score, 27.3 ± 8.9), 585 survived (388 men; median age, 77 years; average APACHE-II score, 25.6 ± 8.4), and 320 died in the ICU (222 men; median age, 78 years; APACHE-II score, 30.6 ± 8.9). All-cause ICU mortality was 35.4%. In patients aged <65 years, factors associated with ICU mortality were APACHE-II score (odds ratio [OR], 1.108; 95% confidence interval [95% CI], 1.021-1.202; P = .014), nosocomial infection (OR, 6.618; 95% CI, 1.065-41.113; P = .043), acute kidney injury (OR, 17.302; 95% CI, 2.728-109.735; P = .002), invasive hemodynamic monitoring (OR, 10.051; 95% CI, 1.362-74.191; P = .024), MV for cardiopulmonary resuscitation (OR, 0.122; 95% CI, 0.016-0.924; P = .042), duration of MV (OR, 0.993; 95% CI, 0.988-0.998; P = .008), successful weaning from MV (OR, 0.012; 95% CI, 0.002-0.066; P < .001), and renal replacement therapy (OR, 0.039; 95% CI, 0.005-0.324; P = .003). In patients aged ≥65 years, factors associated with mortality were APACHE-II score (OR, 1.062; 95% CI, 1.030-1.096; P < .001), nosocomial infection (OR, 2.427; 95% CI, 1.359-4.334; P = .003), septic shock (OR, 2.017; 95% CI, 1.153-3.529; P = .014), blood transfusion (OR, 1.939; 95% CI, 1.174-3.202; P = .010), duration of MV (OR, 0.999; 95% CI, 0.999-1.000; P = .043), and successful weaning from MV (OR, 0.027; 95% CI, 0.015-0.047; P < .001).APACHE-II score, successful weaning, and nosocomial infection in the ICU are independently associated with the prognosis of patients given MV in the ICU.