What factors predict length of stay in the intensive care unit? Systematic review and meta-analysis

Protracted fibrinolysis resistance following cardiac surgery with cardiopulmonary bypass: A prospective observational study of clinical associations and patient outcomes

BACKGROUND

Surgery on cardiopulmonary bypass (CPB) elicits a pleiomorphic systemic host response which, when severe, requires prolonged intensive care support. Given the substantial cross-talk between inflammation, coagulation, and fibrinolysis, the aim of this hypothesis-generating observational study was to document the kinetics of fibrinolysis recovery post-CPB using ClotPro® point-of-care viscoelastometry. Tissue plasminogen activator-induced clot lysis time (TPA LT, s) was correlated with surgical risk, disease severity, organ dysfunction and intensive care length of stay (ICU LOS).

RESULTS

In 52 patients following CPB, TPA LT measured on the first post-operative day (D1) correlated with surgical risk (EuroScore II, Spearman's rho .39, p < .01), time on CPB (rho = .35, p = .04), disease severity (APACHE II, rho = .52, p < .001) and organ dysfunction (SOFA, rho = .51, p < .001) scores, duration of invasive ventilation (rho = .46, p < .01), and renal function (eGFR, rho = -.65, p < .001). In a generalized linear regression model containing TPA LT, CPB run time and markers of organ function, only TPA LT was independently associated with the ICU LOS (odds ratio 1.03 [95% CI 1.01-1.05], p = .01). In a latent variables analysis, the association between TPA LT and the ICU LOS was not mediated by renal function and thus, by inference, variation in the clearance of intraoperative tranexamic acid.

CONCLUSIONS

This observational hypothesis-generating study in patients undergoing cardiac surgery with cardiopulmonary bypass demonstrated an association between the severity of fibrinolysis resistance, measured on the first post-operative day, and the need for extended postoperative ICU level support. Further examination of the role of persistent fibrinolysis resistance on the clinical outcomes in this patient cohort is warranted through large-scale, well-designed clinical studies.

The impact of obesity on ventilator-associated pneumonia, a US nationwide study

BACKGROUND

Ventilator-associated pneumonia (VAP) is one of the leading causes of mortality in patients with critical care illness. Since obesity is highly prevalent, we wanted to study its impact on the outcomes of patients who develop VAP.

METHODS

Using the National Inpatient Sample (NIS) database from 2017 to 2020, we conducted a retrospective study of adult patients with a principal diagnosis of VAP with a secondary diagnosis with or without obesity according to 10th revision of the International Statistical Classification of Diseases (ICD-10) codes. Several demographics, including age, race, and gender, were analyzed. The primary endpoint was mortality, while the secondary endpoints included tracheostomy, length of stay in days, and patient charge in dollars. Multivariate logistic regression model analysis was used to adjust for confounders, with a p-value less than 0.05 considered statistically significant.

RESULTS

The study included 3832 patients with VAP, 395 of whom had obesity. The mean age in both groups was around 58 years, and 68% of the group with obesity were females compared to 40% in females in the group without obesity. Statistically significant comorbidities in the obesity group included a Charlson Comorbidity Index score of three and above, diabetes mellitus, hypertension, chronic kidney disease, and sleep apnea. Rates and odds of mortality were not significantly higher in the collective obesity group 39 (10%) vs. 336 (8.5%), p-value 0.62, adjusted odds ratio 1.2, p-value 0.61). The rates and odds of tracheostomy were higher in the obesity group but not statistically significant. Obese patients were also found to have a longer hospitalization. Upon subanalysis of the data, no evidence of racial disparities was found in the care of VAP for both the obese and control groups.

CONCLUSIONS

Obesity was not found to be an independent risk factor for worse outcomes in patients who develop VAP in the intensive care unit.

Factors Affecting the Length of Stay in the Intensive Care Unit among Adults in Saudi Arabia: A Cross-Sectional Study

This study aimed to assess patient-related factors associated with the LOS among adults admitted to the ICU in Saudi Arabia. The Ministry of Health provided a cross-sectional dataset for 2021, which served as the data source for this study. The data included data on adults admitted to different ICUs at various hospitals. The number of days spent in the ICU was the outcome variable of interest. The potential predictors were age, sex, and nationality, as well as clinical data from the time of admission. Descriptive statistics and bivariate analysis were used to analyse the association between the predictors and the ICU LOS and characterize how they were distributed. We used negative binomial regression to examine the relationship between the study predictors and the ICU LOS. A total of 42,884 individuals were included in this study, of whom 25,520 were men and 17,362 were women. The overall median ICU LOS was three days. This study showed that the ICU LOS was highly influenced by the patient's age, sex, nationality, source of admission, and clinical history. Several predictors that affect how long adults stay in the ICU in Saudi Arabian hospitals were identified in this study. These factors can be attributed to variances in health care delivery systems, patient demographics, and cultural considerations. To allocate resources efficiently, enhance patient outcomes, and create focused treatments to reduce ICU LOS, it is essential to comprehend these elements.

Machine Learning for Benchmarking Critical Care Outcomes

OBJECTIVES

Enhancing critical care efficacy involves evaluating and improving system functioning. Benchmarking, a retrospective comparison of results against standards, aids risk-adjusted assessment and helps healthcare providers identify areas for improvement based on observed and predicted outcomes. The last two decades have seen the development of several models using machine learning (ML) for clinical outcome prediction. ML is a field of artificial intelligence focused on creating algorithms that enable computers to learn from and make predictions or decisions based on data. This narrative review centers on key discoveries and outcomes to aid clinicians and researchers in selecting the optimal methodology for critical care benchmarking using ML.

METHODS

We used PubMed to search the literature from 2003 to 2023 regarding predictive models utilizing ML for mortality (592 articles), length of stay (143 articles), or mechanical ventilation (195 articles). We supplemented the PubMed search with Google Scholar, making sure relevant articles were included. Given the narrative style, papers in the cohort were manually curated for a comprehensive reader perspective.

RESULTS

Our report presents comparative results for benchmarked outcomes and emphasizes advancements in feature types, preprocessing, model selection, and validation. It showcases instances where ML effectively tackled critical care outcome-prediction challenges, including nonlinear relationships, class imbalances, missing data, and documentation variability, leading to enhanced results.

CONCLUSIONS

Although ML has provided novel tools to improve the benchmarking of critical care outcomes, areas that require further research include class imbalance, fairness, improved calibration, generalizability, and long-term validation of published models.

Modelling of intensive care unit (ICU) length of stay as a quality measure: a problematic exercise

BACKGROUND

Intensive care unit (ICU) length of stay (LOS) and the risk adjusted equivalent (RALOS) have been used as quality metrics. The latter measures entail either ratio or difference formulations or ICU random effects (RE), which have not been previously compared.

METHODS

From calendar year 2016 data of an adult ICU registry-database (Australia & New Zealand Intensive Care Society (ANZICS) CORE), LOS predictive models were established using linear (LMM) and generalised linear (GLMM) mixed models. Model fixed effects quality-metric formulations were estimated as RALOSR for LMM (geometric mean derived from log(ICU LOS)) and GLMM (day) and observed minus expected ICU LOS (OMELOS from GLMM). Metric confidence intervals (95%CI) were estimated by bootstrapping; random effects (RE) were predicted for LMM and GLMM. Forest-plot displays of ranked quality-metric point-estimates (95%CI) were generated for ICU hospital classifications (metropolitan, private, rural/regional, and tertiary). Robust rank confidence sets (point estimate and 95%CI), both marginal (pertaining to a singular ICU) and simultaneous (pertaining to all ICU differences), were established.

RESULTS

The ICU cohort was of 94,361 patients from 125 ICUs (metropolitan 16.9%, private 32.8%, rural/regional 6.4%, tertiary 43.8%). Age (mean, SD) was 61.7 (17.5) years; 58.3% were male; APACHE III severity-of-illness score 54.6 (25.7); ICU annual patient volume 1192 (702) and ICU LOS 3.2 (4.9). There was no concordance of ICU ranked model predictions, GLMM versus LMM, nor for the quality metrics used, RALOSR, OMELOS and site-specific RE for each of the ICU hospital classifications. Furthermore, there was no concordance between ICU ranking confidence sets, marginal and simultaneous for models or quality metrics.

CONCLUSIONS

Inference regarding adjusted ICU LOS was dependent upon the statistical estimator and the quality index used to quantify any LOS differences across ICUs. That is, there was no "one best model"; thus, ICU "performance" is determined by model choice and any rankings thereupon should be circumspect.

What we talk about when we talk about COVID-19 vaccination campaign impact: a narrative review

Background

The lack of precise definitions and terminological consensus about the impact studies of COVID-19 vaccination leads to confusing statements from the scientific community about what a vaccination impact study is.

Objective

The present work presents a narrative review, describing and discussing COVID-19 vaccination impact studies, mapping their relevant characteristics, such as study design, approaches and outcome variables, while analyzing their similarities, distinctions, and main insights.

Methods

The articles screening, regarding title, abstract, and full-text reading, included papers addressing perspectives about the impact of vaccines on population outcomes. The screening process included articles published before June 10, 2022, based on the initial papers' relevance to this study's research topics. The main inclusion criteria were data analyses and study designs based on statistical modelling or comparison of pre- and post-vaccination population.

Results

The review included 18 studies evaluating the vaccine impact in a total of 48 countries, including 32 high-income countries (United States, Israel, and 30 Western European countries) and 16 low- and middle-income countries (Brazil, Colombia, and 14 Eastern European countries). We summarize the main characteristics of the vaccination impact studies analyzed in this narrative review.

Conclusion

Although all studies claim to address the impact of a vaccination program, they differ significantly in their objectives since they adopt different definitions of impact, methodologies, and outcome variables. These and other differences are related to distinct data sources, designs, analysis methods, models, and approaches.

Investigating the impact of physical activity interventions on delirium outcomes in intensive care unit patients: A systematic review and meta-analysis

Background

To investigate the impact of physical activity interventions, including early mobilisation, on delirium outcomes in critically ill patients.

Methods

Electronic database literature searches were conducted, and studies were selected based on pre-specified eligibility criteria. Cochrane Risk of Bias-2 and Risk Of Bias In Non-randomised Studies-of Interventions quality assessment tools were utilised. Grading of Recommendations, Assessment, Development and Evaluations was used to assess levels of evidence for delirium outcomes. The study was prospectively registered on PROSPERO (CRD42020210872).

Results

Twelve studies were included; ten randomised controlled trials one observational case-matched study and one before-after quality improvement study. Only five of the included randomised controlled trial studies were judged to be at low risk of bias, with all others, including both non-randomised controlled trials deemed to be at high or moderate risk. The pooled relative risk for incidence was 0.85 (0.62-1.17) which was not statistically significant in favour of physical activity interventions. Narrative synthesis for effect on duration of delirium found favour towards physical activity interventions reducing delirium duration with median differences ranging from 0 to 2 days in three comparative studies. Studies comparing varying intervention intensities showed positive outcomes in favour of greater intensity. Overall levels of evidence were low quality.

Conclusions

Currently there is insufficient evidence to recommend physical activity as a stand-alone intervention to reduce delirium in Intensive Care Units. Physical activity intervention intensity may impact on delirium outcomes, but a lack of high-quality studies limits the current evidence base.

Two-Step Approach for Occupancy Estimation in Intensive Care Units Based on Bayesian Optimization Techniques

Due to the high occupational pressure suffered by intensive care units (ICUs), a correct estimation of the patients' length of stay (LoS) in the ICU is of great interest to predict possible situations of collapse, to help healthcare personnel to select appropriate treatment options and to predict patients' conditions. There has been a high amount of data collected by biomedical sensors during the continuous monitoring process of patients in the ICU, so the use of artificial intelligence techniques in automatic LoS estimation would improve patients' care and facilitate the work of healthcare personnel. In this work, a novel methodology to estimate the LoS using data of the first 24 h in the ICU is presented. To achieve this, XGBoost, one of the most popular and efficient state-of-the-art algorithms, is used as an estimator model, and its performance is optimized both from computational and precision viewpoints using Bayesian techniques. For this optimization, a novel two-step approach is presented. The methodology was carefully designed to execute codes on a high-performance computing system based on graphics processing units, which considerably reduces the execution time. The algorithm scalability is analyzed. With the proposed methodology, the best set of XGBoost hyperparameters are identified, estimating LoS with a MAE of 2.529 days, improving the results reported in the current state of the art and probing the validity and utility of the proposed approach.

International variation in length of stay in intensive care units and the impact of patient-to-nurse ratios

OBJECTIVE

To assess variation in ICU length of stay between countries with varying patient-to-nurse ratios; to compare ICU length of stay of individual countries against an international benchmark.

DESIGN

Secondary analysis of the DecubICUs trial (performed on 15 May 2018).

SETTING

The study cohort included 12,794 adult ICU patients (57 countries). Only countries with minimally twenty patients discharged (or deceased) within 30 days of ICU admission were included.

MAIN OUTCOME MEASURE

Multivariate Cox regression was used to evaluate ICU length of stay, censored at 30 days, across countries and for patient-to-nurse ratio, adjusted for sex, age, admission type and Simplified Acute Physiology Score II. The resulting hazard ratios for countries, indicating longer or shorter length of stay than average, were plotted on a forest plot. Results by country were benchmarked against the overall length of stay using Kaplan-Meier curves.

RESULTS

Patients had a median ICU length of stay of 11 days (interquartile range, 4-27). Hazard ratio by country ranged from minimally 0.42 (95% confidence interval 0.35-0.51) for Greece, to maximaly1.94 (1.28-2.93) for Lithuania. The hazard ratio for patient-to-nurse was 0.96 (0.94-0.98), indicating that higher patient-to-nurse ratio results in longer length of stay.

CONCLUSIONS

Despite adjustment for case-mix, we observed significant heterogeneity of ICU length of stay in-between countries, and a significantly longer length of stay when patient-to-nurse ratio increases. Future studies determining underlying characteristics of individual ICUs and broader organisation of healthcare infrastructure within countries may further explain the observed heterogeneity in ICU length of stay.

Explainable time-series deep learning models for the prediction of mortality, prolonged length of stay and 30-day readmission in intensive care patients

Background

In-hospital mortality, prolonged length of stay (LOS), and 30-day readmission are common outcomes in the intensive care unit (ICU). Traditional scoring systems and machine learning models for predicting these outcomes usually ignore the characteristics of ICU data, which are time-series forms. We aimed to use time-series deep learning models with the selective combination of three widely used scoring systems to predict these outcomes.

Materials and methods

A retrospective cohort study was conducted on 40,083 patients in ICU from the Medical Information Mart for Intensive Care-IV (MIMIC-IV) database. Three deep learning models, namely, recurrent neural network (RNN), gated recurrent unit (GRU), and long short-term memory (LSTM) with attention mechanisms, were trained for the prediction of in-hospital mortality, prolonged LOS, and 30-day readmission with variables collected during the initial 24 h after ICU admission or the last 24 h before discharge. The inclusion of variables was based on three widely used scoring systems, namely, APACHE II, SOFA, and SAPS II, and the predictors consisted of time-series vital signs, laboratory tests, medication, and procedures. The patients were randomly divided into a training set (80%) and a test set (20%), which were used for model development and model evaluation, respectively. The area under the receiver operating characteristic curve (AUC), sensitivity, specificity, and Brier scores were used to evaluate model performance. Variable significance was identified through attention mechanisms.

Results

A total of 33 variables for 40,083 patients were enrolled for mortality and prolonged LOS prediction and 36,180 for readmission prediction. The rates of occurrence of the three outcomes were 9.74%, 27.54%, and 11.79%, respectively. In each of the three outcomes, the performance of RNN, GRU, and LSTM did not differ greatly. Mortality prediction models, prolonged LOS prediction models, and readmission prediction models achieved AUCs of 0.870 ± 0.001, 0.765 ± 0.003, and 0.635 ± 0.018, respectively. The top significant variables co-selected by the three deep learning models were Glasgow Coma Scale (GCS), age, blood urea nitrogen, and norepinephrine for mortality; GCS, invasive ventilation, and blood urea nitrogen for prolonged LOS; and blood urea nitrogen, GCS, and ethnicity for readmission.

Conclusion

The prognostic prediction models established in our study achieved good performance in predicting common outcomes of patients in ICU, especially in mortality prediction. In addition, GCS and blood urea nitrogen were identified as the most important factors strongly associated with adverse ICU events.

Data-driven methodology to predict the ICU length of stay: A multicentre study of 99,492 admissions in 109 Brazilian units

PURPOSE

The length of stay (LoS) is one of the most used metrics for resource use in Intensive Care Units (ICU). We propose a structured data-driven methodology to predict the ICU length of stay and the risk of prolonged stay, and its application in a large multicenter Brazilian ICU database.

METHODS

Demographic data, comorbidities, complications, laboratory data, and primary and secondary diagnosis were prospectively collected and retrospectively analysed by a data-driven methodology, which includes eight different machine learning models and a stacking model. The study setting included 109 mixed-type ICUs from 38 Brazilian hospitals and the external validation was performed by 93 medical-surgical ICUs of 55 hospitals in Brazil.

RESULTS

A cohort of 99,492 adult ICU admissions were included from the 01^st of January to the 31^st of December 2019. The stacking model combining Random Forests and Linear Regression presented the best results to predict ICU length of stay (RMSE = 3.82; MAE = 2.52; R² = 0.36). The prediction model for the risk of long stay were accurate to early identify prolonged stay patients (Brier Score = 0.04, AUC = 0.87, PPV = 0.83, NPV = 0.95).

CONCLUSION

The data-driven methodology to predict ICU length of stay and the risk of long-stay proved accurate in a large multicentre cohort of general ICU patients. The proposed models are helpful to predict the individual length of stay and to early identify patients with high risk of prolonged stay.

Network analytics and machine learning for predicting length of stay in elderly patients with chronic diseases at point of admission

Background

An aging population with a burden of chronic diseases puts increasing pressure on health care systems. Early prediction of the hospital length of stay (LOS) can be useful in optimizing the allocation of medical resources, and improving healthcare quality. However, the data available at the point of admission (PoA) are limited, making it difficult to forecast the LOS accurately.

Methods

In this study, we proposed a novel approach combining network analytics and machine learning to predict the LOS in elderly patients with chronic diseases at the PoA. Two networks, including multimorbidity network (MN) and patient similarity network (PSN), were constructed and novel network features were created. Five machine learning models (eXtreme Gradient Boosting, Gradient Boosting Decision Tree, Random Forest, Linear Support Vector Machine, and Deep Neural Network) with different input feature sets were developed to compare their performance.

Results

The experimental results indicated that the network features can bring significant improvements to the performances of the prediction models, suggesting that the MN and PSN are useful for LOS predictions.

Conclusion

Our predictive framework which integrates network science with data mining can forecast the LOS effectively at the PoA and provide decision support for hospital managers, which highlights the potential value of network-based machine learning in healthcare field.

Association between nonpharmacological strategies and delirium in intensive care unit

BACKGROUND

Several nonpharmacological strategies for the prevention and treatment of delirium have been increasingly used because the aetiology of delirium is multifactorial.

AIMS

To verify the association between nonpharmacological strategies (presence of companion, mobilization, absence of physical restraint and natural light) and the occurrence of delirium, and to identify risk factors for delirium in intensive care unit (ICU) patients.

STUDY DESIGN

The study was conducted in a Brazilian medical and surgical ICU. The sample included patients older than 18 years with length of ICU stay greater than 24 h and without delirium on admission. Delirium was identified by applying the Confusion Assessment Method for the Intensive Care Unit (CAM-ICU). The association between the variables and delirium was analysed using Mann-Whitney and chi-square tests, and multivariate logistic regression to identify the predictive factors.

RESULTS

Of the 356 patients, 64 (18%) had delirium. The presence of a companion, mobilization, and physical restraint were associated with delirium, and the first two were identified as protective factors. That is, the odds of delirium decreased by 88% when a companion was present and by 95% when the patient was mobilized. The risk factors of delirium were length of ICU stay and age.

CONCLUSIONS

The presence of a companion and patient mobilization were identified as protective factors against delirium, highlighting their importance as preventive actions, especially in patients with a higher risk of developing this disorder. The findings regarding physical restraint can also be considered evidence indicating the need for careful use of this measure in clinical practice until evidence of its relationship with delirium is confirmed.

RELEVANCE TO CLINICAL PRACTICE

The implementation of strategies such as early mobilization, presence of a companion and careful assessment for the use of physical restraint by the multidisciplinary team can help control the occurrence of delirium in the ICU.

Forecasting the length-of-stay of pediatric patients in hospitals: a scoping review

BACKGROUND

Healthcare management faces complex challenges in allocating hospital resources, and predicting patients' length-of-stay (LOS) is critical in effectively managing those resources. This work aims to map approaches used to forecast the LOS of Pediatric Patients in Hospitals (LOS-P) and patients' populations and environments used to develop the models.

METHODS

Using the Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) methodology, we performed a scoping review that identified 28 studies and analyzed them. The search was conducted on four databases (Science Direct, Scopus, Web of Science, and Medline). The identification of relevant studies was structured around three axes related to the research questions: (i) forecast models, (ii) hospital length-of-stay, and (iii) pediatric patients. Two authors carried out all stages to ensure the reliability of the review process. Articles that passed the initial screening had their data charted on a spreadsheet. Methods reported in the literature were classified according to the stage in which they are used in the modeling process: (i) pre-processing of data, (ii) variable selection, and (iii) cross-validation.

RESULTS

Forecasting models are most often applied to newborn patients and, consequently, in neonatal intensive care units. Regression analysis is the most widely used modeling approach; techniques associated with Machine Learning are still incipient and primarily used in emergency departments to model patients in specific situations.

CONCLUSIONS

The studies' main benefits include informing family members about the patient's expected discharge date and enabling hospital resources' allocation and planning. Main research gaps are associated with the lack of generalization of forecasting models and limited reported applicability in hospital management. This study also provides a practical guide to LOS-P forecasting methods and a future research agenda.

Methods and measures to quantify ICU patient heterogeneity

Patients in intensive care units are heterogeneous and the daily prediction of their days to discharge (DTD) a complex task that practitioners and computers are not always able to solve satisfactorily. In order to make more precise DTD predictors, it is necessary to have tools for the analysis of the heterogeneity of the patients. Unfortunately, the number of publications in this field is almost non-existent. In order to alleviate this lack of tools, we propose four methods and their corresponding measures to quantify the heterogeneity of intensive patients in the process of determining the DTD. These new methods and measures have been tested with patients admitted over four years to a tertiary hospital in Spain. The results deepen the understanding of the intensive patient and can serve as a basis for the construction of better DTD predictors.