A systematic review of tools for predicting severe adverse events following patient discharge from intensive care units

Oncology and intensive care doctors' perception of intensive care admission of cancer patients: A cross-sectional national survey

INTRODUCTION

Prognosis in oncology has improved with early diagnosis and novel therapies. However, critical illness continues to trigger clinical and ethical dilemmas for the treating oncology and intensive care unit (ICU) doctors.

OBJECTIVES

The objective of this study was to investigate the perceptions of oncology and ICU doctors in managing critically ill cancer patients.

METHODS

A cross-sectional web-based survey exploring the management of a fictitious acutely deteriorating case vignette with solid-organ malignancy. The survey weblink was distributed between May and July 2022 to all Australian oncology and ICU doctors via newsletters to the members of the Medical Oncology Group of Australia, the Australian and New Zealand Intensive Care Society, and the College of Intensive Care Medicine inviting them to participate. The weblink was active till August 2022. The six domains included patient prognostication, advanced care plan, collaborative management, legal/ethical/moral challenges, ICU referral, and protocol-based ICU admission. The outcomes were reported as the level of agreement between oncology and ICU doctors for each domain/question.

RESULTS

184 responses (64 oncology and 120 ICU doctors) were analysed. Most respondents were specialists (78.1% [n = 50] oncology, 78.3% [n = 94] ICU doctors). Oncology doctors more commonly reported managing cancer patients with poor prognosis than ICU doctors (p < 0.001). Oncology doctors less commonly referred such patients for ICU admission (29.7% [n = 19] vs. 80.8% [n = 97], p < 0.001; odds ratio [OR] = 0.07; 95% confidence interval [CI]: 0.03-0.16) and infrequently encountered patients with prior goals of care (GOC) in medical emergency team escalations (40.6% [n = 26] vs. 86.7% [n = 104]; p < 0.001; OR = 0.06; 95% CI: 0.02-0.15; p < 0.001). Oncology doctors were less likely to discuss GOC during medical emergency team calls or within 24 h of ICU admission. More oncology doctors than ICU doctors thought that training rotation in the corresponding speciality group was beneficial (56.3% [n = 36] vs. 31.7% [n = 38]; p = 0.012; OR = 2.07; 95% CI: 1.02-4.23; p = 0.045).

CONCLUSION

Oncology doctors were less likely to encounter acute patient deterioration or establish timely GOC for such patients. Oncology doctors believed that an ICU rotation during their training may have helped manage challenging situations.

Nursing care complexity as a predictor of adverse events in patients transferred from ICU to hospital ward after general surgery

OBJECTIVES

Predicting the likelihood of adverse events following discharge from the intensive care unit (ICU) can contribute to improving the quality of surgical care. This study aimed to evaluate the impact of nursing care complexity as a predictor of adverse event development in general surgery patients transferred from the ICU to the hospital ward.

METHODS

A prospective observational study was conducted with 100 patients in the ICU and general surgical inpatient unit of a training and research hospital in Istanbul, Turkey. The Nursing Care Complexity tool was used by ICU and hospital ward nurses to measure nursing complexity.

RESULTS

A total of 65 adverse events developed in 51 patients during hospital ward hospitalization after discharge from the ICU. Nursing care complexity evaluations by the ICU nurses predicted overall and some specific adverse events, while hospital ward nurses' evaluations predicted ICU readmission and some follow-up abnormalities such as patients' blood pressure, pulse rate, and laboratory results.

CONCLUSION

The results of the current study validate that nursing care complexity can serve as a valuable tool for predicting the risk of adverse events and ICU readmission following discharge from the ICU.

IMPLICATIONS FOR CLINICAL PRACTICE

The use of the Nursing Care complexity tool by the ICU and even hospital ward nurses after ICU discharge may have a significant impact on patient outcomes and contribute to the recognition of nursing efforts.

Current perspectives on the use of artificial intelligence in critical patient safety

Intensive Care Units (ICUs) have undergone enhancements in patient safety, and artificial intelligence (AI) emerges as a disruptive technology offering novel opportunities. While the published evidence is limited and presents methodological issues, certain areas show promise, such as decision support systems, detection of adverse events, and prescription error identification. The application of AI in safety may pursue predictive or diagnostic objectives. Implementing AI-based systems necessitates procedures to ensure secure assistance, addressing challenges including trust in such systems, biases, data quality, scalability, and ethical and confidentiality considerations. The development and application of AI demand thorough testing, encompassing retrospective data assessments, real-time validation with prospective cohorts, and efficacy demonstration in clinical trials. Algorithmic transparency and explainability are essential, with active involvement of clinical professionals being crucial in the implementation process.

Development of an enhanced scoring system to predict ICU readmission or in-hospital death within 24 hours using routine patient data from two NHS Foundation Trusts

RATIONALE

Intensive care units (ICUs) admit the most severely ill patients. Once these patients are discharged from the ICU to a step-down ward, they continue to have their vital signs monitored by nursing staff, with Early Warning Score (EWS) systems being used to identify those at risk of deterioration.

OBJECTIVES

We report the development and validation of an enhanced continuous scoring system for predicting adverse events, which combines vital signs measured routinely on acute care wards (as used by most EWS systems) with a risk score of a future adverse event calculated on discharge from the ICU.

DESIGN

A modified Delphi process identified candidate variables commonly available in electronic records as the basis for a 'static' score of the patient's condition immediately after discharge from the ICU. L1-regularised logistic regression was used to estimate the in-hospital risk of future adverse event. We then constructed a model of physiological normality using vital sign data from the day of hospital discharge. This is combined with the static score and used continuously to quantify and update the patient's risk of deterioration throughout their hospital stay.

SETTING

Data from two National Health Service Foundation Trusts (UK) were used to develop and (externally) validate the model.

PARTICIPANTS

A total of 12 394 vital sign measurements were acquired from 273 patients after ICU discharge for the development set, and 4831 from 136 patients in the validation cohort.

RESULTS

Outcome validation of our model yielded an area under the receiver operating characteristic curve of 0.724 for predicting ICU readmission or in-hospital death within 24 hours. It showed an improved performance with respect to other competitive risk scoring systems, including the National EWS (0.653).

CONCLUSIONS

We showed that a scoring system incorporating data from a patient's stay in the ICU has better performance than commonly used EWS systems based on vital signs alone.

TRIAL REGISTRATION NUMBER

ISRCTN32008295.

Prognostic factors for critically ill surgical patients with unplanned intensive care unit readmission: Developing a novel predictive scoring model for predicting readmission

BACKGROUND

Unplanned readmission to the surgical intensive care unit has been demonstrated to worsen patient outcomes. Our objective was to identify risk factors and outcomes associated with unplanned surgical intensive care unit readmission and to develop a predictive scoring model to identify patients at high risk of readmission.

METHODS

We retrospectively analyzed patients admitted to the surgical intensive care unit (2020-2021) and categorized them as either with or without unplanned readmission.

RESULTS

Of 1,112 patients in the derivation cohort, 76 (6.8%) experienced unplanned surgical intensive care unit readmission, with sepsis being the leading cause of readmission (35.5%). Patients who were readmitted had significantly higher in-hospital mortality rates than those who were not. Multivariate analysis identified congestive heart failure, high Sequential Organ Failure Assessment-Hepatic score, use of carbapenem during surgical intensive care unit stay, as well as factors before surgical intensive care unit discharge such as inadequate glycemic control, positive fluid balance, low partial pressure of oxygen in arterial blood/fraction of inspired oxygen ratio, and receipt of total parenteral nutrition as independent predictors for unplanned readmission. The scoring model developed using these predictors exhibited good discrimination between readmitted and non-readmitted patients, with an area under the curve of 0.74. The observed rates of unplanned readmission for scores of <4 points and ≥4 points were 4% and 20.2% (P < .001), respectively. The model also demonstrated good performance in the validation cohort, with an area under the curve of 0.74 and 19% observed unplanned readmission rate for scores ≥4 points.

CONCLUSION

Besides congestive heart failure, clinicians should meticulously re-evaluate critical variables such as the Sequential Organ Failure Assessment-Hepatic score, partial pressure of oxygen in arterial blood/fraction of inspired oxygen ratio, glycemic control, and fluid status before releasing the patient from the surgical intensive care unit. It is crucial to determine the reasons for using carbapenems during surgical intensive care unit stay and the causes for the inability to discontinue total parenteral nutrition before discharging the patient from the surgical intensive care unit.

Perioperative risk factors associated with unplanned neurological intensive care unit readmission following elective supratentorial brain tumor resection

OBJECTIVE

The aim of this study was to describe the clinical and procedural risk factors associated with the unplanned neurosurgical intensive care unit (NICU) readmission of patients after elective supratentorial brain tumor resection and serves as an exploratory analysis toward the development of a risk stratification tool that may be prospectively applied to this patient population.

METHODS

This was a retrospective observational cohort study. The electronic medical records of patients admitted to an institutional NICU between September 2018 and November 2021 after elective supratentorial brain tumor resection were reviewed. Demographic and perioperative clinical factors were recorded. A prognostic model was derived from the data of 4892 patients recruited between September 2018 and May 2021 (development cohort). A nomogram was created to display these predictor variables and their corresponding points and risks of readmission. External validation was evaluated using a series of 1118 patients recruited between June 2021 and November 2021 (validation cohort). Finally, a decision curve analysis was performed to determine the clinical usefulness of the prognostic model.

RESULTS

Of the 4892 patients in the development cohort, 220 (4.5%) had an unplanned NICU readmission. Older age, lesion type, Karnofsky Performance Status (KPS) < 70 at admission, longer duration of surgery, retention of endotracheal intubation on NICU entry, and longer NICU length of stay (LOS) after surgery were independently associated with an unplanned NICU readmission. A total of 1118 patients recruited between June 2021 and November 2021 were included for external validation, and the model's discrimination remained acceptable (C-statistic = 0.744, 95% CI 0.675-0.814). The decision curve analysis for the prognostic model in the development and validation cohorts showed that at a threshold probability between 0.05 and 0.8, the prognostic model showed a positive net benefit.

CONCLUSIONS

A predictive model that included age, lesion type, KPS < 70 at admission, duration of surgery, retention of endotracheal intubation on NICU entry, and NICU LOS after surgery had an acceptable ability to identify elective supratentorial brain tumor resection patients at high risk for an unplanned NICU readmission. These risk factors and this prediction model may facilitate better resource allocation in the NICU and improve patient outcomes.

The risk assessment tool for intensive care unit readmission: A systematic review and meta-analysis

OBJECTIVE

To review and evaluate existing risk assessment tools for intensive care unitreadmission.

METHODS

Nine electronic databases (Medline, CINAHL, Web of Science, Cochrane Library, Embase, Sino Med, CNKI, VIP, and Wan fang) were systematically searched from their inception to September 2022. Two authors independently extracted data from the literature included. Meta-analysis was performed under the bivariate modeling and summary receiver operating characteristic curve method.

RESULTS

A total of 29 studies were included in this review, among which 11 were quantitatively Meta-analyzed. The results showed Stability and Workload Index for Transfer: Sensitivity = 0.55, Specificity = 0.65, Area under curve = 0.63. And Early warning score: Sensitivity = 0.78, Specificity = 0.83, Area under curve = 0.88. The remaining tools included scores, nomograms, machine learning models, and deep learning models. These studies, with varying reports on thresholds, case selection, data preprocessing, and model performance, have a high risk of bias.

CONCLUSION

We cannot identify a tool that can be used directly in intensive care unit readmission risk assessment. Scores based on early warning score are moderately accurate in predicting readmission, but there is heterogeneity and publication bias that requires model adjustment for local factors such as resources, demographics, and case mix. Machine learning models present a promising modeling technique but have a high methodological bias and require further validation.

IMPLICATIONS FOR CLINICAL PRACTICE

Using reliable risk assessment tools is essential for the early identification of unplanned intensive care unit readmission risk in critically ill patients.  A reliable risk assessment tool must be developed, which is the focus of further research.

Risk factors for hospital mortality in intensive care unit survivors: a retrospective cohort study

BACKGROUND

Deaths can occur after a patient has survived treatment for a serious illness in an intensive care unit (ICU). Mortality rates after leaving the ICU can be considered indicators of health care quality. This study aims to describe risk factors and mortality of surviving patients discharged from an ICU in a university hospital.

METHODS

Retrospective cohort study carried out from January 2017 to December 2018. Data on age, sex, length of hospital stay, diagnosis on admission to the ICU, hospital discharge outcome, presence of infection, and Simplified Acute Physiology Score (SAPS) III prognostic score were collected. Infected patients were considered as those being treated for an infection on discharge from the ICU. Patients were divided into survivors and non-survivors on leaving the hospital. The association between the studied variables was performed using the logistic regression model.

RESULTS

A total of 1,025 patients who survived hospitalization in the ICU were analyzed, of which 212 (20.7%) died after leaving the ICU. When separating the groups of survivors and non-survivors according to hospital outcome, the median age was higher among non-survivors. Longer hospital stays and higher SAPS III values were observed among non-survivors. In the logistic regression, the variables age, length of hospital stay, SAPS III, presence of infection, and readmission to the ICU were associated with hospital mortality.

CONCLUSIONS

Infection on ICU discharge, ICU readmission, age, length of hospital stay, and SAPS III increased risk of death in ICU survivors.

Development and validation of an interpretable 3 day intensive care unit readmission prediction model using explainable boosting machines

Background

Intensive care unit (ICU) readmissions are associated with mortality and poor outcomes. To improve discharge decisions, machine learning (ML) could help to identify patients at risk of ICU readmission. However, as many models are black boxes, dangerous properties may remain unnoticed. Widely used post hoc explanation methods also have inherent limitations. Few studies are evaluating inherently interpretable ML models for health care and involve clinicians in inspecting the trained model.

Methods

An inherently interpretable model for the prediction of 3 day ICU readmission was developed. We used explainable boosting machines that learn modular risk functions and which have already been shown to be suitable for the health care domain. We created a retrospective cohort of 15,589 ICU stays and 169 variables collected between 2006 and 2019 from the University Hospital Münster. A team of physicians inspected the model, checked the plausibility of each risk function, and removed problematic ones. We collected qualitative feedback during this process and analyzed the reasons for removing risk functions. The performance of the final explainable boosting machine was compared with a validated clinical score and three commonly used ML models. External validation was performed on the widely used Medical Information Mart for Intensive Care version IV database.

Results

The developed explainable boosting machine used 67 features and showed an area under the precision-recall curve of 0.119 ± 0.020 and an area under the receiver operating characteristic curve of 0.680 ± 0.025. It performed on par with state-of-the-art gradient boosting machines (0.123 ± 0.016, 0.665 ± 0.036) and outperformed the Simplified Acute Physiology Score II (0.084 ± 0.025, 0.607 ± 0.019), logistic regression (0.092 ± 0.026, 0.587 ± 0.016), and recurrent neural networks (0.095 ± 0.008, 0.594 ± 0.027). External validation confirmed that explainable boosting machines (0.221 ± 0.023, 0.760 ± 0.010) performed similarly to gradient boosting machines (0.232 ± 0.029, 0.772 ± 0.018). Evaluation of the model inspection showed that explainable boosting machines can be useful to detect and remove problematic risk functions.

Conclusions

We developed an inherently interpretable ML model for 3 day ICU readmission prediction that reached the state-of-the-art performance of black box models. Our results suggest that for low- to medium-dimensional datasets that are common in health care, it is feasible to develop ML models that allow a high level of human control without sacrificing performance.

Prognostic value of National Early Warning Score and Modified Early Warning Score on intensive care unit readmission and mortality: A prospective observational study

Background

Modified Early Warning Score (MEWS) and National Early Warning Score (NEWS) are widely used in predicting the mortality and intensive care unit (ICU) admission of critically ill patients. This study was conducted to evaluate and compare the prognostic value of NEWS and MEWS for predicting ICU readmission, mortality, and related outcomes in critically ill patients at the time of ICU discharge.

Methods

This multicenter, prospective, observational study was conducted over a year, from April 2019 to March 2020, in the general ICUs of two university-affiliated hospitals in Northwest Iran. MEWS and NEWS were compared based on the patients’ outcomes (including mortality, ICU readmission, time to readmission, discharge type, mechanical ventilation (MV), MV duration, and multiple organ failure after readmission) using the univariable and multivariable binary logistic regression. The receiver operating characteristic (ROC) curve was used to determine the outcome predictability of MEWS and NEWS.

Results

A total of 410 ICU patients were enrolled in this study. According to multivariable logistic regression analysis, both MEWS and NEWS were predictors of ICU readmission, time to readmission, MV status after readmission, MV duration, and multiple organ failure after readmission. The area under the ROC curve (AUC) for predicting mortality was 0.91 (95% CI = 0.88–0.94, P < 0.0001) for the NEWS and 0.88 (95% CI = 0.84–0.91, P < 0.0001) for the MEWS. There was no significant difference between the AUC of the NEWS and the MEWS for predicting mortality (P = 0.082). However, for ICU readmission (0.84 vs. 0.71), time to readmission (0.82 vs. 0.67), MV after readmission (0.83 vs. 0.72), MV duration (0.81 vs. 0.67), and multiple organ failure (0.833 vs. 0.710), the AUCs of MEWS were significantly greater (P < 0.001).

Conclusion

National Early Warning Score and MEWS values of >4 demonstrated high sensitivity and specificity in identifying the risk of mortality for the patients’ discharge from ICU. However, we found that the MEWS showed superiority over the NEWS score in predicting other outcomes. Eventually, MEWS could be considered an efficient prediction score for morbidity and mortality of critically ill patients.

A Machine Learning Based Discharge Prediction of Cardiovascular Diseases Patients in Intensive Care Units

This paper targets a major challenge of how to effectively allocate medical resources in intensive care units (ICUs). We trained multiple regression models using the Medical Information Mart for Intensive Care III (MIMIC III) database recorded in the period between 2001 and 2012. The training and validation dataset included pneumonia, sepsis, congestive heart failure, hypotension, chest pain, coronary artery disease, fever, respiratory failure, acute coronary syndrome, shortness of breath, seizure and transient ischemic attack, and aortic stenosis patients’ recorded data. Then we tested the models on the unseen data of patients diagnosed with coronary artery disease, congestive heart failure or acute coronary syndrome. We included the admission characteristics, clinical prescriptions, physiological measurements, and discharge characteristics of those patients. We assessed the models’ performance using mean residuals and running times as metrics. We ran multiple experiments to study the data partition’s impact on the learning phase. The total running time of our best-evaluated model is 123,450.9 mS. The best model gives an average accuracy of 98%, highlighting the location of discharge, initial diagnosis, location of admission, drug therapy, length of stay and internal transfers as the most influencing patterns to decide a patient’s readiness for discharge.

Predictors of Repeat Medical Emergency Team Activation in Deteriorating Ward Patients: A Retrospective Cohort Study

Recurrent clinical deterioration and repeat medical emergency team (MET) activation are common and associated with high in-hospital mortality. This study assessed the predictors for repeat MET activation in deteriorating patients admitted to a general ward. We retrospectively analyzed the data of 5512 consecutive deteriorating hospitalized adult patients who required MET activation in the general ward. The patients were divided into two groups according to repeat MET activation. Multivariate logistic regression analyses were used to identify the predictors for repeat MET activation. Hematological malignancies (odds ratio, 2.07; 95% confidence interval, 1.54–2.79) and chronic lung disease (1.49; 1.07–2.06) were associated with a high risk of repeat MET activation. Among the causes for MET activation, respiratory distress (1.76; 1.19–2.60) increased the risk of repeat MET activation. A low oxygen saturation-to-fraction of inspired oxygen ratio (0.97; 0.95–0.98), high-flow nasal cannula oxygenation (4.52; 3.56–5.74), airway suctioning (4.63; 3.59–5.98), noninvasive mechanical ventilation (1.52; 1.07–2.68), and vasopressor support (1.76; 1.22–2.54) at first MET activation increased the risk of repeat MET activation. The risk factors identified in this study may be useful to identify patients at risk of repeat MET activation at the first MET activation. This would allow the classification of high-risk patients and the application of aggressive interventions to improve outcomes.

Patient discharge from intensive care: an updated scoping review to identify tools and practices to inform high-quality care

BACKGROUND

Critically ill patients require complex care and experience unique needs during and after their stay in the intensive care unit (ICU). Discharging or transferring a patient from the ICU to a hospital ward or back to community care (under the care of a general practitioner) includes several elements that may shape patient outcomes and overall experiences. The aim of this study was to answer the question: what elements facilitate a successful, high-quality discharge from the ICU?

METHODS

This scoping review is an update to a review published in 2015. We searched MEDLINE, EMBASE, CINAHL, and Cochrane databases from 2013-December 3, 2020 including adult, pediatric, and neonatal populations without language restrictions. Data were abstracted using different phases of care framework models, themes, facilitators, and barriers to the ICU discharge process.

RESULTS

We included 314 articles from 11,461 unique citations. Two-hundred and fifty-eight (82.2%) articles were primary research articles, mostly cohort (118/314, 37.6%) or qualitative (51/314, 16.2%) studies. Common discharge themes across all articles included adverse events, readmission, and mortality after discharge (116/314, 36.9%) and patient and family needs and experiences during discharge (112/314, 35.7%). Common discharge facilitators were discharge education for patients and families (82, 26.1%), successful provider-provider communication (77/314, 24.5%), and organizational tools to facilitate discharge (50/314, 15.9%). Barriers to a successful discharge included patient demographic and clinical characteristics (89/314, 22.3%), healthcare provider workload (21/314, 6.7%), and the impact of current discharge practices on flow and performance (49/314, 15.6%). We identified 47 discharge tools that could be used or adapted to facilitate an ICU discharge.

CONCLUSIONS

Several factors contribute to a successful ICU discharge, with facilitators and barriers present at the patient and family, health care provider, and organizational level. Successful provider-patient and provider-provider communication, and educating and engaging patients and families about the discharge process were important factors in a successful ICU discharge.

Transitions of Care After Critical Illness-Challenges to Recovery and Adaptive Problem Solving

OBJECTIVES

Investigate the challenges experienced by survivors of critical illness and their caregivers across the transitions of care from intensive care to community, and the potential problem-solving strategies used to navigate these challenges.

DESIGN

Qualitative design-data generation via interviews and data analysis via the framework analysis method.

SETTING

Patients and caregivers from three continents, identified through the Society of Critical Care Medicine's THRIVE international collaborative sites (follow-up clinics and peer support groups).

SUBJECTS

Patients and caregivers following critical illness.

INTERVENTIONS

Nil.

MEASUREMENTS AND MAIN RESULTS

From 86 interviews (66 patients, 20 caregivers), we identified the following major themes: 1) Challenges for patients-interacting with the health system and gaps in care; managing others' expectations of illness and recovery. 2) Challenges for caregivers-health system shortfalls and inadequate communication; lack of support for caregivers. 3) Patient and caregiver-driven problem solving across the transitions of care-personal attributes, resources, and initiative; receiving support and helping others; and acceptance.

CONCLUSIONS

Survivors and caregivers experienced a range of challenges across the transitions of care. There were distinct and contrasting themes related to the caregiver experience. Survivors and caregivers used comparable problem-solving strategies to navigate the challenges encountered across the transitions of care.

Explainable Machine Learning on AmsterdamUMCdb for ICU Discharge Decision Support: Uniting Intensivists and Data Scientists

Unexpected ICU readmission is associated with longer length of stay and increased mortality. To prevent ICU readmission and death after ICU discharge, our team of intensivists and data scientists aimed to use AmsterdamUMCdb to develop an explainable machine learning–based real-time bedside decision support tool.

Factors and Outcomes of Intensive Care Unit Readmission in Elderly Patients

INTRODUCTION

An increase in age has been observed among patients admitted to the intensive care unit (ICU). Age is a well-known risk factor for ICU readmission and mortality. However, clinical characteristics and risk factors of ICU readmission of elderly patients (≥65 years) have not been studied.

METHODS

This retrospective single-center cohort study was conducted in a total of 122-bed ICU of a tertiary care hospital in Seoul, Korea. A total of 85,413 patients were enrolled in this hospital between January 1, 2007, and December 31, 2017. The odds ratio of readmission and in-hospital mortality was calculated by logistic regression analysis.

RESULTS

Totally, 29,503 patients were included in the study group, of which 2,711 (9.2%) had ICU readmissions. Of the 2,711 readmitted patients, 472 patients were readmitted more than once (readmitted 2 or more times to the ICU, 17.4%). In the readmitted patient group, there were more males, higher sequential organ failure assessment (SOFA) scores, and hospitalized for medical reasons. Length of stay (LOS) in ICU and in-hospital were longer, and 28-day and in-hospital mortality was higher in readmitted patients than in nonreadmitted patients. Risk factors of ICU readmission included the ICU admission due to medical reason, SOFA score, presence of chronic heart disease, diabetes mellitus, chronic kidney disease, transplantation, use of mechanical ventilation, and initial ICU LOS. ICU readmission and age (over 85 years) were independent predictors of in-hospital mortality on multivariable analysis. The delayed ICU readmission group (>72 h) had higher in-hospital mortality than the early readmission group (≤72 h) (20.6 vs. 16.2%, p = 0.005).

CONCLUSIONS

ICU readmissions occurred in 9.2% of elderly patients and were associated with poor prognosis and higher mortality.

Adverse Events After Transition From ICU to Hospital Ward: A Multicenter Cohort Study

OBJECTIVES

To examine adverse events and associated factors and outcomes during transition from ICU to hospital ward (after ICU discharge).

DESIGN

Multicenter cohort study.

SETTING

Ten adult medical-surgical Canadian ICUs.

PATIENTS

Patients were those admitted to one of the 10 ICUs from July 2014 to January 2016.

INTERVENTIONS

None.

MEASUREMENTS AND MAIN RESULTS

Two ICU physicians independently reviewed progress and consultation notes documented in the medical record within 7 days of patient's ICU discharge date to identify and classify adverse events. The adverse event data were linked to patient characteristics and ICU and ward physician surveys collected during the larger prospective cohort study. Analyses were conducted using multivariable logistic regression. Of the 451 patients included in the study, 84 (19%) experienced an adverse event, the majority (62%) within 3 days of transfer from ICU to hospital ward. Most adverse events resulted only in symptoms (77%) and 36% were judged to be preventable. Patients with adverse events were more likely to be readmitted to the ICU (odds ratio, 5.5; 95% CI, 2.4-13.0), have a longer hospital stay (mean difference, 16.1 d; 95% CI, 8.4-23.7) or die in hospital (odds ratio, 4.6; 95% CI, 1.8-11.8) than those without an adverse event. ICU and ward physician predictions at the time of ICU discharge had low sensitivity and specificity for predicting adverse events, ICU readmissions, and hospital death.

CONCLUSIONS

Adverse events are common after ICU discharge to hospital ward and are associated with ICU readmission, increased hospital length of stay and death and are not predicted by ICU or ward physicians.

Are there opportunities to improve care as patients transition through the cancer care continuum? A scoping review protocol

INTRODUCTION

Transitions in Care (TiC) are vulnerable periods in care delivery associated with adverse events, increased cost and decreased patient satisfaction. Patients with cancer encounter many transitions during their care journey due to improved survival rates and the complexity of treatment. Collectively, improving TiC is particularly important among patients with cancer. The objective of this scoping review is to synthesise and map the existing literature regarding TiC among patients with cancer in order to explore opportunities to improve TiC among patients with cancer.

METHODS AND ANALYSIS

This scoping review will follow the Preferred Reporting Items for Systematic Reviews and Meta-Analysis-Scoping Review Extension and the Joanna Briggs Institute methodology. The PubMed cancer filter and underlying search strategy will be tailored to each database (Embase, Cochrane, CINAHL and PsycINFO) and combined with search terms for TiC. Grey literature and references of included studies will be searched. The search will include studies published from database inception until 9 February 2020. Quantitative and qualitative studies will be included if they describe transitions between any type of healthcare provider or institution among patients with cancer. Descriptive statistics will summarise study characteristics and quantitative data of included studies. Qualitative data will be synthesised using thematic analysis.

ETHICS AND DISSEMINATION

Our objective is to synthesise and map the existing evidence; therefore, ethical approval is not required. Evidence gaps around TiC will inform a programme of research aimed to improve high-risk transitions among patients with cancer. The findings of this scoping review will be published in a peer-reviewed journal and widely presented at academic conferences. More importantly, decision makers and patients will be provided a summary of the findings, along with data from a companion study, to prioritise TiC in need of interventions to improve continuity of care for patients with cancer.

Routine critical care step-down programmes: Systematic review and meta-analysis

BACKGROUND

Patients discharged from critical care to general hospital wards are vulnerable to clinical deterioration, critical care readmission, and death. In response, routine critical care stepdown programmes (CCSDPs) have been widely developed, which involve the review of all patients on general wards following discharge from critical care by multidisciplinary Outreach teams with critical care skills.

AIMS AND OBJECTIVES

This review aims to answer the question: do routine CCSDPs reduce readmission and/or mortality among patients discharged from critical care?

DESIGN

Systematic review of quantitative studies and meta-analysis.

METHODS

Six databases were comprehensively searched from inception (CENTRAL, Cochrane Reviews, MEDLINE, Embase, CINAHL and web of Science), alongside grey literature and trial registers. Studies investigating the effect of routine CCSDPs delivered by Outreach nurses on readmission and/or mortality following discharge from adult critical care to general hospital wards were included. Study quality was assessed using the Cochrane ROBINS-I tool.

RESULTS

Eight studies met the inclusion criteria, with data from 6 studies pooled in 3 meta-analyses. Among patients exposed to routine CCSDPs, pooled data estimated a statistically nonsignificant reduction in the risk of readmission to critical care (risk ratio [RR] 0.85; 95% confidence interval [CI] 0.66-1.09; P = .19), a statistically significant increase in the risk of readmission to critical care within 72 hours (RR 1.49; 95% CI 1.05-2.12; P = .03), a statistically non-significant reduction in risk of mortality following critical care discharge (RR 0.90; 95% CI 0.75-1.07; P = .22), and no association with mortality within 14 days of discharge.

CONCLUSION

This review is unable to definitively conclude whether routine CCSDPs reduce critical care readmission or mortality following critical care discharge.

RELEVANCE TO CLINICAL PRACTICE

While the synthesized evidence does not suggest a change in policy and practice are warranted, neither does it support routine CCSDPs in the absence of high-quality evidence.

A new analytical framework for missing data imputation and classification with uncertainty: Missing data imputation and heart failure readmission prediction

Background

The wide adoption of electronic health records (EHR) system has provided vast opportunities to advance health care services. However, the prevalence of missing values in EHR system poses a great challenge on data analysis to support clinical decision-making. The objective of this study is to develop a new methodological framework that can address the missing data challenge and provide a reliable tool to predict the hospital readmission among Heart Failure patients.

Methods

We used Gaussian Process Latent Variable Model (GPLVM) to impute the missing values. Specifically, a lower dimensional embedding was learned from a small complete dataset and then used to impute the missing values in the incomplete dataset. The GPLVM-based missing data imputation can provide both the mean estimate and the uncertainty associated with the mean estimate. To incorporate the uncertainty in prediction, a constrained support vector machine (cSVM) was developed to obtain robust predictions. We first sampled multiple datasets from the distributions of input uncertainty and trained a support vector machine for each dataset. Then an optimal classifier was identified by selecting the support vectors that maximize the separation margin of a newly sampled dataset and minimize the similarity with the pre-trained support vectors.

Results

The proposed model was derived and validated using Physionet MIMIC-III clinical database. The GPLVM imputation provided normalized mean absolute errors of 0.11 and 0.12 respectively when 20% and 30% of instances contained missing values, and the confidence bounds of the estimations captures 97% of the true values. The cSVM model provided an average Area Under Curve of 0.68, which improves the prediction accuracy by 7% as compared to some existing classifiers.

Conclusions

The proposed method provides accurate imputation of missing values and has a better prediction performance as compared to existing models that can only deal with deterministic inputs.

Modified Early Warning Score as a predictor of intensive care unit readmission within 48 hours: a retrospective observational study

OBJECTIVE

To evaluate the hypothesis that the Modified Early Warning Score (MEWS) at the time of intensive care unit discharge is associated with readmission and to identify the MEWS that most reliably predicts intensive care unit readmission within 48 hours of discharge.

METHODS

This was a retrospective observational study of the MEWSs of discharged patients from the intensive care unit. We compared the demographics, severity scores, critical illness characteristics, and MEWSs of readmitted and non-readmitted patients, identified factors associated with readmission in a logistic regression model, constructed a Receiver Operating Characteristic (ROC) curve of the MEWS in predicting the probability of readmission, and presented the optimum criterion with the highest sensitivity and specificity.

RESULTS

The readmission rate was 2.6%, and the MEWS was a significant predictor of readmission, along with intensive care unit length of stay > 10 days and tracheostomy. The ROC curve of the MEWS in predicting the readmission probability had an AUC of 0.82, and a MEWS > 6 carried a sensitivity of 0.78 (95%CI 0.66 - 0.9) and specificity of 0.9 (95%CI 0.87 - 0.93).

CONCLUSION

The MEWS is associated with intensive care unit readmission, and a score > 6 has excellent accuracy as a prognostic predictor.

Clinician perception of long-term survival at the point of critical care discharge: a prospective cohort study

Critical care survivors suffer persistent morbidity and increased risk of mortality as compared with the general population. Nevertheless, there are no standardised tools to identify at-risk patients. Our aim was to establish whether the Sabadell score, a simple tool applied by the treating clinician upon critical care discharge, was independently associated with 5-year mortality through a prospective observational cohort study of adults admitted to a general critical care unit. The Sabadell score, which is a measure of clinician-assigned survival perception, was applied to all patients from September 2011 to December 2017. The primary outcome was 5-year mortality, assessed using a multivariable flexible parametric survival analysis adjusted for baseline characteristics and clinically relevant covariates. We studied 5954 patients with a minimum of 18 months follow-up. Mean (SD) age was 59.5 (17.0) years and 3397 (57.1%) patients were men. We categorised 2287 (38.4%) patients as Sadabell 0; 2854 (47.9%) as Sadabell 1; 629 (10.5%) as Sadabell 2; and 183 (3.1%) as Sadabell 3. Adjusted hazard ratios for mortality were 2.1 (95%CI 1.9-2.4); 4.0 (95%CI 3.4-4.6); and 21.0 (95%CI 17.2-25.7), respectively. Sabadell 3 patients had 99.9%, 99.5%, 98.5% and 87.4% mortality at 5 years for patients in the age brackets ≥ 80, 60-79, 40-59 and 16-39 years, respectively. Sabadell 2 patients had 71.0%, 52.7%, 44.8% and 23.7% 5-year mortality for these same age categories. The Sabadell score was independently associated with 5-year survival after critical care discharge. These findings can be used to guide provision of increased support for patients after critical care discharge and/or informed discussions with patients and relatives about dying to ascertain their future wishes.

Impact of Intensive Care Unit Readmissions on Patient Outcomes and the Evaluation of the National Early Warning Score to Prevent Readmissions: Literature Review

Background

Intensive care unit (ICU) readmissions have been shown to increase a patient’s in-hospital mortality and length of stay (LOS). Despite this, no methods have been set in place to prevent readmissions from occurring.

Objective

The aim of this literature review was to evaluate the impact of ICU readmission on patient outcomes and to evaluate the effect of using a risk stratification tool, the National Early Warning Score (NEWS), on ICU readmissions.

Methods

A database search was performed on PubMed, Cumulative Index of Nursing and Allied Health Literature, Google Scholar, and ProQuest. In the initial search, 2028 articles were retrieved; after inclusion and exclusion criteria were applied, 12 articles were ultimately used in this literature review.

Results

This literature review found that patients readmitted to the ICU have an increased mortality rate and LOS at the hospital. The sample sizes in the reviewed studies ranged from 158 to 745,187 patients. Readmissions were most commonly associated with respiratory issues about 18% to 59% of the time. The NEWS has been shown to detect early clinical deterioration in a patient within 24 hours of transfer, with a 95% CI of 0.89 to 0.94 (P<.001), a sensitivity of 93.6% , and a specificity of 82.2%.

Conclusions

ICU readmissions are associated with worse patient outcomes, including hospital mortality and increased LOS. Without the use of an objective screening tool, the provider has been solely responsible for the decision of patient transfer. Assessment with the NEWS could be helpful in decreasing the frequency of inappropriate transfers and ultimately ICU readmission.

Is my patient ready for a safe transfer to a lower-intensity care setting? Nursing complexity as an independent predictor of adverse events risk after ICU discharge

BACKGROUND

Patients discharged from intensive care units (ICUs) are at risk for adverse events (AEs). Establishing safe discharge criteria is challenging. No available criteria consider nursing complexity among risk factors.

OBJECTIVES

To investigate whether nursing complexity upon ICU discharge is an independent predictor for AEs.

METHODS

Prospective observational study. The Patient Acuity and Complexity Score (PACS) was developed to measure nursing complexity. Its predictive power for AEs was tested using multivariate regression analysis.

RESULTS

The final regression model showed a very-good discrimination power (AUC 0.881; p<0.001) for identifying patients who experienced AEs. Age, ICU admission reason, PACS, cough strength, PaCO₂, serum creatinine and sodium, and transfer to Internal Medicine showed to be predictive of AEs. Exceeding the identified PACS threshold increased by 3.3 times the AEs risk.

CONCLUSIONS

The level of nursing complexity independently predicts AEs risk and should be considered in establishing patient's eligibility for a safe ICU discharge.

Readmissions to General ICUs in a Geographic Area of Poland Are Seemingly Associated with Better Outcomes

BACKGROUND

Various factors can contribute to high mortality rates in intensive care units (ICUs). Here, we intended to define a population of patients readmitted to general ICUs in Poland and to identify independent predictors of ICU readmission.

METHODS

Data derived from adult ICU admissions from the Silesian region of Poland were analyzed. First-time ICU readmissions (≤30 days from ICU discharge after index admissions) were compared with first-time ICU admissions. Pre-admission and admission variables that independently influenced the need for ICU readmission were identified.

RESULTS

Among the 21,495 ICU admissions, 839 were first-time readmissions (3.9%). Patients readmitted to the ICU had lower mean APACHE II (21.2 ± 8.0 vs. 23.2 ± 8.8, p < 0.001) and TISS-28 scores (33.7 ± 7.4 vs. 35.2 ± 7.8, p < 0.001) in the initial 24 h following ICU admission, compared to first-time admissions. ICU readmissions were associated with lower mortality vs. first-time admissions (39.2% vs. 44.3%, p = 0.004). Independent predictors for ICU readmission included the admission from a surgical ward (among admission sources), chronic respiratory failure, cachexia, previous stroke, chronic neurological diseases (among co-morbidities), and multiple trauma or infection (among primary reasons for ICU admission).

CONCLUSIONS

High mortality associated with first-time ICU admissions is associated with a lower mortality rate during ICU readmissions.

Developing a framework for evidence-based grading and assessment of predictive tools for clinical decision support

BACKGROUND

Clinical predictive tools quantify contributions of relevant patient characteristics to derive likelihood of diseases or predict clinical outcomes. When selecting predictive tools for implementation at clinical practice or for recommendation in clinical guidelines, clinicians are challenged with an overwhelming and ever-growing number of tools, most of which have never been implemented or assessed for comparative effectiveness. To overcome this challenge, we have developed a conceptual framework to Grade and Assess Predictive tools (GRASP) that can provide clinicians with a standardised, evidence-based system to support their search for and selection of efficient tools.

METHODS

A focused review of the literature was conducted to extract criteria along which tools should be evaluated. An initial framework was designed and applied to assess and grade five tools: LACE Index, Centor Score, Well's Criteria, Modified Early Warning Score, and Ottawa knee rule. After peer review, by six expert clinicians and healthcare researchers, the framework and the grading of the tools were updated.

RESULTS

GRASP framework grades predictive tools based on published evidence across three dimensions: 1) Phase of evaluation; 2) Level of evidence; and 3) Direction of evidence. The final grade of a tool is based on the highest phase of evaluation, supported by the highest level of positive evidence, or mixed evidence that supports a positive conclusion. Ottawa knee rule had the highest grade since it has demonstrated positive post-implementation impact on healthcare. LACE Index had the lowest grade, having demonstrated only pre-implementation positive predictive performance.

CONCLUSION

GRASP framework builds on widely accepted concepts to provide standardised assessment and evidence-based grading of predictive tools. Unlike other methods, GRASP is based on the critical appraisal of published evidence reporting the tools' predictive performance before implementation, potential effect and usability during implementation, and their post-implementation impact. Implementing the GRASP framework as an online platform can enable clinicians and guideline developers to access standardised and structured reported evidence of existing predictive tools. However, keeping GRASP reports up-to-date would require updating tools' assessments and grades when new evidence becomes available, which can only be done efficiently by employing semi-automated methods for searching and processing the incoming information.

Prediction of ICU Readmissions Using Data at Patient Discharge

Unplanned readmissions to ICU contribute to high health care costs and poor patient outcomes. 6-7% of all ICU cases see a readmission within 72 hours. Machine learning models on electronic health record data can help identify these cases, providing more information about short and long-term risks to clinicians at the time of ICU discharge. While time-toevent techniques have been used in clinical care, models that identify risks over time using higher-dimensional, non-linear machine learning models need to be developed to present changes in risk with non-linear techniques. This work identifies risks of ICU readmissions at 24 hours, 72 hours, 7 days, 30 days, and bounceback readmissions in the same hospital admission with an AUROC for 72 hours of 0.76 and for bounceback of 0.84.

Risk prediction models for intensive care unit readmission: A systematic review of methodology and applicability

OBJECTIVE

We conducted a systematic review of primary models to predict intensive care unit (ICU) readmission.

REVIEW METHODS

We searched MEDLINE, PubMed, Scopus, and Embase for studies on the development of ICU readmission prediction models that are published until January 2017. Data were extracted on the source of data, participants, outcomes, candidate predictors, sample size, missing data, methods for model development, and measures of model performance and model evaluation. The quality and applicability of the included studies were assessed using the CHecklist for critical Appraisal and data extraction for systematic Reviews of prediction Modelling Studies.

RESULTS

We identified five studies describing the development of the primary prediction models of ICU readmission. Studies ranged in size from 343 to 704,963 patients with the mean age of 58.0-68.9 years. The proportion of readmission ranged from 2.5% to 9.6%. The discriminative ability of prediction models measured by area under the receiver operating characteristic curve was 0.66-0.81. None of the studies performed external validations. The quality scores ranged from 42 to 54 out of 62, and the applicability scores from 24 to 32 out of 38.

CONCLUSION

We identified five prediction models for ICU readmission. However, owing to the numerous methodological and reporting deficiencies in the included studies, physicians using these models should interpret the predictions with precautions until an external validation study shows the acceptable level of calibration and accuracy of these models.

Subjective score and outcomes after discharge from the intensive care unit: a prospective observational study

Objective

Intensive care unit (ICU) discharge is a decision process that is usually performed subjectively. We evaluated whether a subjective score (Sabadell score) is associated with hospital outcomes.

Methods

We conducted a prospective cohort study from August 2014 to May 2015 at a tertiary-care private hospital in Brazil. We analyzed 425 patients who were discharged alive from the ICU to the wards. We used univariate and multivariate analysis to identify risk factors associated with a composite endpoint of worse outcomes (later ICU readmission or ward death) during the same hospitalization.

Results

Forty-three patients (10.1%) were readmitted after ICU discharge, and 19 died in the ward. Compared with patients with successful outcomes, those with the composite endpoint were older and more severely ill, had a nonsurgical reason for hospitalization, more frequently came from the ward, were less frequently independent during daily activities, had sepsis, had higher C-reactive protein concentrations at ICU admission, and had higher Sabadell scores at discharge. The multivariate analysis showed that sepsis and the Sabadell score were independently and significantly associated with worse outcomes.

Conclusion

Sepsis at admission and the Sabadell score were predictors of worse hospital outcomes. The Sabadell score might be a promising predictive tool.

Analysis and prediction of unplanned intensive care unit readmission using recurrent neural networks with long short-term memory

BACKGROUND

Unplanned readmission of a hospitalized patient is an indicator of patients' exposure to risk and an avoidable waste of medical resources. In addition to hospital readmission, intensive care unit (ICU) readmission brings further financial risk, along with morbidity and mortality risks. Identification of high-risk patients who are likely to be readmitted can provide significant benefits for both patients and medical providers. The emergence of machine learning solutions to detect hidden patterns in complex, multi-dimensional datasets provides unparalleled opportunities for developing an efficient discharge decision-making support system for physicians and ICU specialists.

METHODS AND FINDINGS

We used supervised machine learning approaches for ICU readmission prediction. We used machine learning methods on comprehensive, longitudinal clinical data from the MIMIC-III to predict the ICU readmission of patients within 30 days of their discharge. We incorporate multiple types of features including chart events, demographic, and ICD-9 embeddings. We have utilized recent machine learning techniques such as Recurrent Neural Networks (RNN) with Long Short-Term Memory (LSTM), by this we have been able to incorporate the multivariate features of EHRs and capture sudden fluctuations in chart event features (e.g. glucose and heart rate). We show that our LSTM-based solution can better capture high volatility and unstable status in ICU patients, an important factor in ICU readmission. Our machine learning models identify ICU readmissions at a higher sensitivity rate of 0.742 (95% CI, 0.718-0.766) and an improved Area Under the Curve of 0.791 (95% CI, 0.782-0.800) compared with traditional methods. We perform in-depth deep learning performance analysis, as well as the analysis of each feature contribution to the predictive model.

CONCLUSION

Our manuscript highlights the ability of machine learning models to improve our ICU decision-making accuracy and is a real-world example of precision medicine in hospitals. These data-driven solutions hold the potential for substantial clinical impact by augmenting clinical decision-making for physicians and ICU specialists. We anticipate that machine learning models will improve patient counseling, hospital administration, allocation of healthcare resources and ultimately individualized clinical care.

Towards a decision support tool for intensive care discharge: machine learning algorithm development using electronic healthcare data from MIMIC-III and Bristol, UK

OBJECTIVE

The primary objective is to develop an automated method for detecting patients that are ready for discharge from intensive care.

DESIGN

We used two datasets of routinely collected patient data to test and improve on a set of previously proposed discharge criteria.

SETTING

Bristol Royal Infirmary general intensive care unit (GICU).

PATIENTS

Two cohorts derived from historical datasets: 1870 intensive care patients from GICU in Bristol, and 7592 from Medical Information Mart for Intensive Care (MIMIC)-III.

RESULTS

In both cohorts few successfully discharged patients met all of the discharge criteria. Both a random forest and a logistic classifier, trained using multiple-source cross-validation, demonstrated improved performance over the original criteria and generalised well between the cohorts. The classifiers showed good agreement on which features were most predictive of readiness-for-discharge, and these were generally consistent with clinical experience. By weighting the discharge criteria according to feature importance from the logistic model we showed improved performance over the original criteria, while retaining good interpretability.

CONCLUSIONS

Our findings indicate the feasibility of the proposed approach to ready-for-discharge classification, which could complement other risk models of specific adverse outcomes in a future decision support system. Avenues for improvement to produce a clinically useful tool are identified.

Diffusing an Innovation: Clinician Perceptions of Continuous Predictive Analytics Monitoring in Intensive Care

BACKGROUND

The purpose of this article is to describe neonatal intensive care unit clinician perceptions of a continuous predictive analytics technology and how those perceptions influenced clinician adoption. Adopting and integrating new technology into care is notoriously slow and difficult; realizing expected gains remain a challenge.

METHODS

Semistructured interviews from a cross-section of neonatal physicians (n = 14) and nurses (n = 8) from a single U.S. medical center were collected 18 months following the conclusion of the predictive monitoring technology randomized control trial. Following qualitative descriptive analysis, innovation attributes from Diffusion of Innovation Theory-guided thematic development.

RESULTS

Results suggest that the combination of physical location as well as lack of integration into work flow or methods of using data in care decisionmaking may have delayed clinicians from routinely paying attention to the data. Once data were routinely collected, documented, and reported during patient rounds and patient handoffs, clinicians came to view data as another vital sign. Through clinicians' observation of senior physicians and nurses, and ongoing dialogue about data trends and patient status, clinicians learned how to integrate these data in care decision making (e.g., differential diagnosis) and came to value the technology as beneficial to care delivery.

DISCUSSION

The use of newly created predictive technologies that provide early warning of illness may require implementation strategies that acknowledge the risk-benefit of treatment clinicians must balance and take advantage of existing clinician training methods.

Nighttime intensive care unit discharge and outcomes: A propensity matched retrospective cohort study

Background

Nighttime ICU discharge, i.e., discharge from the ICU during the night hours, has been associated with increased readmission rates, hospital length of stay (LOS) and in-hospital mortality. We sought to determine the frequency of nighttime ICU discharge and identify whether nighttime ICU discharge is associated with worse outcomes in a private adult ICU located in Brazil.

Methods

Post hoc analysis of a cohort study addressing the effect of ICU readmissions on outcomes. This retrospective, single center, propensity matched cohort study was conducted in a medical-surgical ICU located in a private tertiary care hospital in São Paulo, Brazil. Based on time of transfer, patients were categorized into nighttime (7:00 pm to 6:59 am) and daytime (7:00 am to 6:59 pm) ICU discharge and were propensity-score matched at a 1:2 ratio. The primary outcome of interest was in–hospital mortality.

Results

Among 4,313 eligible patients admitted to the ICU between June 2013 and May 2015, 1,934 patients were matched at 1:2 ratio [649 (33.6%) nighttime and 1,285 (66.4%) daytime discharged patients]. The median (IQR) cohort age was 66 (51–79) years and SAPS III score was 43 (33–55). In-hospital mortality was 6.5% (42/649) in nighttime compared to 5.6% (72/1,285) in daytime discharged patients (OR, 1.17; 95% CI, 0.79 to 1.73; p = 0.444). While frequency of ICU readmission (OR, 0.95; 95% CI, 0.78 to 1.29; p = 0.741) and length of hospital stay did not differ between the groups, length of ICU stay was lower in nighttime compared to daytime ICU discharged patients [1 (1–3) days vs. 2 (1–3) days, respectively, p = 0.047].

Conclusion

In this propensity-matched retrospective cohort study, time of ICU discharge did not affect in-hospital mortality.

Predicting ICU readmission among surgical ICU patients: Development and validation of a clinical nomogram

BACKGROUND

Unplanned intensive care unit readmission within 72 hours is an established metric of hospital care quality. However, it is unclear what factors commonly increase the risk of intensive care unit readmission in surgical patients. The objective of this study was to evaluate predictors of readmission among a diverse sample of surgical patients and develop an accurate and clinically applicable nomogram for prospective risk prediction.

METHODS

We retrospectively evaluated patient demographic characteristics, comorbidities, and physiologic variables collected within 48 hours before discharge from a surgical intensive care unit at an academic center between April 2010 and July 2015. Multivariable regression models were used to assess the association between risk factors and unplanned readmission back to the intensive care unit within 72 hours. Model selection was performed using lasso methods and validated using an independent data set by receiver operating characteristic area under the curve analysis. The derived nomogram was then prospectively assessed between June and August 2017 to evaluate the correlation between perceived and calculated risk for intensive care unit readmission.

RESULTS

Among 3,109 patients admitted to the intensive care unit by general surgery (34%), transplant (9%), trauma (43%), and vascular surgery (14%) services, there were 141 (5%) unplanned readmissions within 72 hours. Among 179 candidate predictor variables, a reduced model was derived that included age, blood urea nitrogen, serum chloride, serum glucose, atrial fibrillation, renal insufficiency, and respiratory rate. These variables were used to develop a clinical nomogram, which was validated using 617 independent admissions, and indicated moderate performance (area under the curve: 0.71). When prospectively assessed, intensive care unit providers' perception of respiratory risk was moderately correlated with calculated risk using the nomogram (ρ: 0.44; P < .001), although perception of electrolyte abnormalities, hyperglycemia, renal insufficiency, and risk for arrhythmias were not correlated with measured values.

CONCLUSION

Intensive care unit readmission risk for surgical patients can be predicted using a simple clinical nomogram based on 7 common demographic and physiologic variables. These data underscore the potential of risk calculators to combine multiple risk factors and enable a more accurate risk assessment beyond perception alone.

Predicting Intensive Care Unit Readmission with Machine Learning Using Electronic Health Record Data

RATIONALE

Patients transferred from the intensive care unit to the wards who are later readmitted to the intensive care unit have increased length of stay, healthcare expenditure, and mortality compared with those who are never readmitted. Improving risk stratification for patients transferred to the wards could have important benefits for critically ill hospitalized patients.

OBJECTIVES

We aimed to use a machine-learning technique to derive and validate an intensive care unit readmission prediction model with variables available in the electronic health record in real time and compare it to previously published algorithms.

METHODS

This observational cohort study was conducted at an academic hospital in the United States with approximately 600 inpatient beds. A total of 24,885 intensive care unit transfers to the wards were included, with 14,962 transfers (60%) in the training cohort and 9,923 transfers (40%) in the internal validation cohort. Patient characteristics, nursing assessments, International Classification of Diseases, Ninth Revision codes from prior admissions, medications, intensive care unit interventions, diagnostic tests, vital signs, and laboratory results were extracted from the electronic health record and used as predictor variables in a gradient-boosted machine model. Accuracy for predicting intensive care unit readmission was compared with the Stability and Workload Index for Transfer score and Modified Early Warning Score in the internal validation cohort and also externally using the Medical Information Mart for Intensive Care database (n = 42,303 intensive care unit transfers).

RESULTS

Eleven percent (2,834) of discharges to the wards were later readmitted to the intensive care unit. The machine-learning-derived model had significantly better performance (area under the receiver operating curve, 0.76) than either the Stability and Workload Index for Transfer score (area under the receiver operating curve, 0.65), or Modified Early Warning Score (area under the receiver operating curve, 0.58; P value < 0.0001 for all comparisons). At a specificity of 95%, the derived model had a sensitivity of 28% compared with 15% for Stability and Workload Index for Transfer score and 7% for the Modified Early Warning Score. Accuracy improvements with the derived model over Modified Early Warning Score and Stability and Workload Index for Transfer were similar in the Medical Information Mart for Intensive Care-III cohort.

CONCLUSIONS

A machine learning approach to predicting intensive care unit readmission was significantly more accurate than previously published algorithms in both our internal validation and the Medical Information Mart for Intensive Care-III cohort. Implementation of this approach could target patients who may benefit from additional time in the intensive care unit or more frequent monitoring after transfer to the hospital ward.

Elevated red blood cell distribution width at ICU discharge is associated with readmission to the intensive care unit

BACKGROUND

Red blood cell distribution width (RDW) is a predictor of mortality in critically ill patients. Our objective was to investigate the association between the RDW at ICU discharge and the risk of ICU readmission or unexpected death in the ward.

METHODS

A secondary analysis of prospectively collected data study was conducted including patients discharged alive from the ICU to the ward. The target variable was the RDW collected at ICU discharge. Elevated RDW was defined as an RDW > 16%. Outcomes of interest included readmission to the ICU, unexpected death in the ward and in-hospital death. Variables with a p-value <0.1 in the univariate analysis or with biological plausibility for the occurrence of the outcome were included in the Cox proportional hazards model for adjustment.

RESULTS

We included 813 patients. A total of 138 readmissions to the ICU and 44 unexpected deaths in the ward occurred. Elevated RDW at ICU discharge was independently associated with readmission to the ICU or unexpected death in the ward after multivariable adjustment (HR: 1.901; 95% CI 1.357-2.662). Other variables associated with this outcome included age, tracheostomy and mean corpuscular volume (MCV) at ICU discharge. Similar results were obtained after the exclusion of unexpected deaths in the ward (HR 1.940; CI 1.312-2.871) and for in-hospital deaths (HR 1.716; 95% CI 1.141-2.580).

CONCLUSIONS

Elevated RDW at ICU discharge is independently associated with ICU readmission and in-hospital death.

Development and implementation of a risk identification tool to facilitate critical care transitions for high-risk surgical patients

Quality problem

Patients recently discharged from the intensive care unit (ICU) are at high risk for clinical deterioration.

Initial assessment

Unreliable and incomplete handoffs of complex patients contributed to preventable ICU readmissions. Respiratory decompensation was responsible for four times as many readmissions as other causes.

Choice of solution

Form a multidisciplinary team to address care coordination surrounding the transfer of patients from the ICU to the surgical ward.

Implementation

A quality improvement intervention incorporating verbal handoffs, time-sensitive patient evaluations and visual cues was piloted over a 1-year period in consecutive high-risk surgical patients discharged from the ICU. Process metrics and clinical outcomes were compared to historical controls.

Evaluation

The intervention brought the primary team and respiratory therapists to the bedside for a baseline examination within 60 min of ward arrival. Stakeholders viewed the intervention as such a valuable adjunct to patient care that the intervention has become a standard of care. While not significant, in a comparatively older and sicker intervention population, the rate of readmissions due to respiratory decompensation was 12.5%, while 35.0% in the control group (P = 0.28).

Lessons learned

The implementation of this ICU transition protocol is feasible and internationally applicable, and results in improved care coordination and communication for a high-risk group of patients.

Does the implementation of a novel intensive care discharge risk score and nurse-led inpatient review tool improve outcome? A prospective cohort study in two intensive care units in the UK

OBJECTIVE

To develop a clinical prediction model for poor outcome after intensive care unit (ICU) discharge in a large observational data set and couple this to an acute post-ICU ward-based review tool (PIRT) to identify high-risk patients at the time of ICU discharge and improve their acute ward-based review and outcome.

DESIGN

Retrospective patient cohort of index ICU admissions between June 2006 and October 2011 receiving routine inpatient review. Prospective cohort between March 2012 and March 2013 underwent risk scoring (PIRT) which subsequently guided inpatient ward-based review.

SETTING

Two UK adult ICUs.

PARTICIPANTS

4212 eligible discharges from ICU in the retrospective development cohort and 1028 patients included in the prospective intervention cohort.

INTERVENTIONS

Multivariate analysis was performed to determine factors associated with poor outcome in the retrospective cohort and used to generate a discharge risk score. A discharge and daily ward-based review tool incorporating an adjusted risk score was introduced. The prospective cohort underwent risk scoring at ICU discharge and inpatient review using the PIRT.

OUTCOMES

The primary outcome was the composite of death or readmission to ICU within 14 days of ICU discharge following the index ICU admission.

RESULTS

PIRT review was achieved for 67.3% of all eligible discharges and improved the targeting of acute post-ICU review to high-risk patients. The presence of ward-based PIRT review in the prospective cohort did not correlate with a reduction in poor outcome overall (P=0.876) or overall readmission but did reduce early readmission (within the first 48 hours) from 4.5% to 3.6% (P=0.039), while increasing the rate of late readmission (48 hours to 14 days) from 2.7% to 5.8% (P=0.046).

CONCLUSION

PIRT facilitates the appropriate targeting of nurse-led inpatient review acutely after ICU discharge but does not reduce hospital mortality or overall readmission rates to ICU.

Patient transfer from the intensive care unit to a general ward

The transfer of patients from the intensive care unit (ICU) to a general ward can present several challenges for nurses. Such patients are at high risk of adverse outcomes, including readmission to the ICU, and increased nosocomial infections and mortality, with a resultant increase in hospital costs. This article explores the challenges of transferring patients from the ICU and uses evidence to examine ways to address them to ensure optimal care for a complex patient group. Transfer time, factors affecting general ward care, handover processes, recognition of deterioration and education, intensive care outreach, and the psychological factors affecting these patients are examined.

Prediction of early unplanned intensive care unit readmission in a UK tertiary care hospital: a cross-sectional machine learning approach

OBJECTIVES

Unplanned readmissions to the intensive care unit (ICU) are highly undesirable, increasing variance in care, making resource planning difficult and potentially increasing length of stay and mortality in some settings. Identifying patients who are likely to suffer unplanned ICU readmission could reduce the frequency of this adverse event.

SETTING

A single academic, tertiary care hospital in the UK.

PARTICIPANTS

A set of 3326 ICU episodes collected between October 2014 and August 2016. All records were of patients who visited an ICU at some point during their stay. We excluded patients who were ≤16 years of age; visited ICUs other than the general and neurosciences ICU; were missing crucial electronic patient record measurements; or had indeterminate ICU discharge outcomes or very early or extremely late discharge times. After exclusion, 2018 outcome-labelled episodes remained.

PRIMARY AND SECONDARY OUTCOME MEASURES

Area under the receiver operating characteristic curve (AUROC) for prediction of unplanned ICU readmission or in-hospital death within 48 hours of first ICU discharge.

RESULTS

In 10-fold cross-validation, an ensemble predictor was trained on data from both the target hospital and the Medical Information Mart for Intensive Care (MIMIC-III) database and tested on the target hospital's data. This predictor discriminated between patients with the unplanned ICU readmission or death outcome and those without this outcome, attaining mean AUROC of 0.7095 (SE 0.0260), superior to the purpose-built Stability and Workload Index for Transfer (SWIFT) score (AUROC=0.6082, SE 0.0249; p=0.014, pairwise t-test).

CONCLUSIONS

Despite the inherent difficulties, we demonstrate that a novel machine learning algorithm based on transfer learning could achieve good discrimination, over and above that of the treating clinicians or the value added by the SWIFT score. Accurate prediction of unplanned readmission could be used to target resources more efficiently.

Incidence and Etiology of Potentially Preventable ICU Readmissions

OBJECTIVES

The rate of unplanned ICU readmissions is often considered a measure of hospital performance. However, the degree to which these readmissions are preventable and the causes of preventable readmissions are unknown, creating uncertainty about the feasibility and value of reducing ICU readmission rates. To inform this issue, we sought to determine the frequency and underlying causes of potentially preventable ICU readmissions.

DESIGN

Retrospective cohort study.

SETTING

Urban, academic medical center in the mid-Atlantic United States.

PATIENTS

Adult patients discharged alive from their first ICU admission with an unplanned readmission within 48 hours of discharge.

MEASUREMENTS AND MAIN RESULTS

Each patient's medical chart was reviewed by two independent investigators who rated each readmission's preventability according to standardized scale and assessed the etiology of both preventable and nonpreventable readmissions. We assessed concordance between raters using the κ statistic and resolved disagreements through iterative discussion. Of 136 readmissions in the final analysis, 16 (11.8%; 95% CI, 6.9-18.4) were considered preventable and 120 (88.2%; 95% CI, 81.5-93.1) were considered nonpreventable. Of nonpreventable readmissions, 67 were due to a new clinical problem and 53 were due to an existing clinical problem. Among preventable readmissions, six were attributable to system errors, six were attributable to management errors, two were attributable to procedural events, one was attributable to a diagnostic error, and one was attributable to a medication error. Compared to nonpreventable readmissions, preventable readmissions tended to have shorter index ICU lengths of stay (2 vs 3 d; p = 0.05) and a shorter duration of time on the ward prior to readmission (16.6 vs 23.6 hr; p = 0.05).

CONCLUSIONS

The majority of early ICU readmissions are nonpreventable, raising important concerns about ICU readmission rates as a measure of hospital performance.

Readmissions to Intensive Care: A Prospective Multicenter Study in Australia and New Zealand

OBJECTIVES

To determine factors independently associated with readmission to ICU and the independent association of readmission with subsequent mortality.

DESIGN

Prospective multicenter observational study.

SETTING

Forty ICUs in Australia and New Zealand.

PATIENTS

Consecutive adult patients discharged alive from ICU to hospital wards between September 2009 and February 2010.

INTERVENTIONS

Measurement of hospital mortality.

MEASUREMENTS AND MAIN RESULTS

We studied 10,210 patients and 674 readmissions. The median age was 63 years (interquartile range, 49-74), and 6,224 (61%) were male. The majority of readmissions were unplanned (84.1%) but only deemed preventable in a minority (8.9%) of cases. Time to first readmission was shorter for unplanned than planned readmission (3.2 vs 6.9 d; p < 0.001). Primary diagnosis changed between admission and readmission in the majority of patients (60.2%) irrespective of planned (58.2%) or unplanned (60.6%) status. Using recurrent event analysis incorporating patient frailty, we found no association between readmissions and hospital survival (hazard ratios: first readmission 0.88, second readmission 0.90, third readmission 0.44; p > 0.05). In contrast, age (hazard ratio, 1.03), a medical diagnosis (hazard ratio, 1.43), inotrope use (hazard ratio, 3.47), and treatment limitation order (hazard ratio, 17.8) were all independently associated with outcome.

CONCLUSIONS

In this large prospective study, readmission to ICU was not an independent risk factor for mortality.