Mortality predictions on admission as a context for organizing care activities

Preoperative Comorbidities Associated With Early Mortality in Hip Fracture Patients: A Multicenter Study

OBJECTIVE

Multiple comorbidities in hip fracture patients are associated with increased mortality and complications. The goal of this study was to characterize the relationship between specific patient factors including comorbidities and outcomes in geriatric hip fractures, including length of stay, unplanned ICU admission, discharge disposition, complications, and mortality.

METHODS

This is a retrospective review of a trauma database from five Level 1 and Level 2 trauma centers of patients with hip fractures of the femoral neck and intertrochanteric region who underwent treatment using hip pinning, hemiarthroplasty, total hip arthroplasty, cephalomedullary nailing, or dynamic hip screw fixation. Mortality was the primary outcome variable (including in-hospital mortality, 30-day mortality, 60-day mortality, and 90-day mortality). Secondary outcome variables included in-hospital adverse events, unplanned transfer to the ICU, postoperative length of stay, and discharge disposition. Regression analyses were used for evaluation of relationships between comorbidities as independent variables and primary and secondary outcomes as dependent variables.

RESULTS

Two thousand three hundred patients were included. The mortality was 1.8%, 7.0%, 10.9%, and 14.1% for in-hospital, 30-day, 60-day, and 90-day mortality, respectively. Diabetes and cognitive impairment present on admission were associated with mortality at all-time intervals. COPD was the only comorbidity that signaled in-hospital adverse event with an odds ratio of 1.67 (P = 0.012). No patient factors, time to surgery, or comorbidities signaled unplanned ICU transfer. Patients with renal failure and COPD had longer hospital stays after surgery.

CONCLUSION

Geriatric hip fractures continue to have high short-term morbidity and mortality. Identifying patients with increased odds of early mortality and adverse events can help teams optimize care and outcomes. Patients with diabetes, cognitive impairment, renal failure, and COPD may benefit from continued and improved medical optimization during the perioperative period as well as being more closely managed by a medicine team without delaying time to the operating room.

Assessment of the disease severity in patients hospitalized for COVID-19 based on the National Early Warning Score (NEWS) using statistical and machine learning methods: An electronic health records database analysis

BACKGROUND

The coronavirus disease 2019 (COVID-19) was a cause of concern in the healthcare system and increased the need for disease severity indicators. However, they still vary in use to evaluate in-hospital outcomes and severity. The National Early Warning Score (NEWS) is routinely used to evaluate patient health status at the hospital. Further research is needed to ensure if NEWS can be a good instrument for an overall health status assessment with or without additional information like laboratory tests, intensive care needs, and history of chronic diseases.

OBJECTIVE

To evaluate if NEWS can be an indicator to measure COVID-19 patient status in-hospital.

METHODS

We used the fully anonymized Electronic Health Records (EHR) characterizing patients admitted to the hospital with COVID-19. Data was obtained from Vilnius University Hospital Santaros Klinikos EHR system (SANTA-HIS) from 01-03-2020 to 31-12-2022. The study sample included 3875 patients. We created several statistical and machine learning models for discrimination between in-hospital death/discharge for evaluation NEWS as a disease severity measure for COVID-19 patients. In these models, two variable sets were considered: median NEWS and its combination with clinical parameters and medians of laboratory test results. Assessment of models' performance was based on the scoring metrics: accuracy, sensitivity, specificity, area under the ROC curve (AUC), and F1-score.

RESULTS

Our analysis revealed that NEWS predictive ability for describing patient health status during the stay in the hospital can be increased by adding the patient's age at hospitalization, gender, clinical and laboratory variables (0.853 sensitivity, 0.992 specificity and F1-score - 0.859) in comparison with single NEWS (0.603, 0.995, 0.719, respectively). A comparison of different models showed that stepwise logistic regression was the best method for in-hospital mortality classification. Our findings suggest employing models like ours for advisory routine usage.

CONCLUSION

Our model demonstrated incremental value for COVID-19 patient's status evaluation.

Comparison of high flow oxygen therapy versus noninvasive mechanical ventilation for successful weaning from invasive ventilation in children: An observational study

Post-extubation respiratory failure is associated with a poor prognosis due to increased ventilator-associated pneumonia, and longer length of stay in the ICU and hospital. In this study, we aimed to evaluate the efficacy of high-flow nasal cannula (HFNC) and noninvasive mechanical ventilation (NIMV) on extubation success in children. A total of 48 patients, aged between 1 month and 18 years, who were weaned to either NIMV or HFNC were included. Patients who had tracheostomy or were not weaned and underwent unplanned extubation were excluded. Age, gender, anthropometric parameters, Pediatric Risk of Mortality and Pediatric Logistic Organ Dysfunction scores, oxygenation index, mechanical ventilation length of stay (LOS), HFNC/NIMV LOS, Modified Downes-Silverman score (MDS), and venous blood gas parameters, pediatric intensive care unit (PICU) LOS were recorded. 24 patients were extubated to NIMV, and 24 patients to HFNC. HFNC LOS and NIMV LOS were similar (P = .621). The failure rates at the 48th hour of HFNC and NIMV were 33% (n = 8), and 33% respectively (n = 8) (P = 1.0). PICU LOS and mortality rate was also similar (P = .06, P = .312 respectively). MDS decreased significantly in both groups (P < .001, P = .02 respectively). Changes in blood gas parameters and MDS within the first 48-hour of device application were similar between the 2 groups. HFNC is not inferior to NIMV in patients with extubation difficulty or those expected to have such difficulty in terms of treatment success, PICU LOS, and mortality. Therefore, HFNC appears to be a weaning technique alternative to NIMV after extubation.

Antibiotic Use in Patients With β-Lactam Allergies and Pneumonia: Impact of an Antibiotic Side Chain-Based Cross-Reactivity Chart Combined With Enhanced Allergy Assessment

BACKGROUND

β-lactam antibiotics with dissimilar R-group side chains are associated with low cross-reactivity. Despite this, patients with β-lactam allergies are often treated with non-β-lactam alternative antibiotics. An institutional β-lactam side chain-based cross-reactivity chart was developed and implemented to guide in antibiotic selection for patients with β-lactam allergies.

METHODS

This single-center, retrospective cohort study analyzed the impact of the implementation of the cross-reactivity chart for patients with pneumonia. Study time periods were defined as January 2013 to October 2014 prior to implementation of the chart (historical cohort) and January 2017 to October 2018 (intervention cohort) following implementation. The primary outcome was the incidence of β-lactam utilization between time periods. Propensity-weighted scoring and interrupted time-series analyses compared outcomes.

RESULTS

A total of 341 and 623 patient encounters were included in the historical and intervention cohorts, respectively. There was a significantly greater use of β-lactams in the intervention cohort (70.4% vs 89.3%; P < .001) and decreased use of alternative therapy (58.1% vs 36%; P < .001). There was no difference in overall allergic reactions between cohorts (2.4% vs 1.6%; P = .738) or in reactions caused by β-lactams (1.3% vs 0.9%; P = .703). Inpatient mortality increased (0% vs 6.4%; P < .001); however, no deaths were due to allergic reactions. Healthcare facility-onset Clostridioides difficile infections decreased between cohorts (1.2% vs 0.2%; P = .032).

CONCLUSIONS

Implementation of a β-lactam side chain-based cross-reactivity chart and enhanced allergy assessment was associated with increased use of β-lactams in patients with pneumonia without increasing allergic reactions.

Antimicrobial Stewardship by Transitions of Care Pharmacists at Hospital Discharge

Objectives: This study assessed the impact transitions of care (TOC) pharmacists have on optimizing antimicrobial use for patients at high risk for mortality at hospital discharge. In addition, this study aimed to summarize and categorize the types of interventions made. Methods: This was a retrospective descriptive study that included adult patients 18 years of age or older who were at high risk for readmission and mortality. Participants were selected if they had a hospital discharge date between January 2017 and June 2018, but were excluded if they were discharged to a facility where medications were managed by healthcare employees or if they were hospice eligible. TOC pharmacists identified eligible participants and reviewed their discharge medication lists to optimize pharmacological therapy, contacting the discharging prescriber if therapy changes were identified. The therapy recommendations made by TOC pharmacists were documented in an internal database for further analysis. Results: A total of 1100 patients were analyzed by TOC pharmacists during the studied timeframe and a total of 2066 interventions were made. With respect to study objectives, 298 (14.4%) of the interventions made by TOC pharmacists involved antimicrobial recommendations, affecting 255 (23.2%) patients. Recommendations involving dosing (89, 29.9%), treatment duration (74, 24.8%), and drug interactions (41, 13.8%) were the most frequent types of interventions made. Sixty-six (25.9%) patients received multiple interventions and 240 (80.5%) recommendations were accepted by the provider. Conclusion: An opportunity exists to optimize antimicrobial therapy surrounding the time of hospital discharge.

Development and Validation of a 30-Day In-hospital Mortality Model Among Seriously Ill Transferred Patients: a Retrospective Cohort Study

BACKGROUND

Predicting the risk of in-hospital mortality on admission is challenging but essential for risk stratification of patient outcomes and designing an appropriate plan-of-care, especially among transferred patients.

OBJECTIVE

Develop a model that uses administrative and clinical data within 24 h of transfer to predict 30-day in-hospital mortality at an Academic Health Center (AHC).

DESIGN

Retrospective cohort study. We used 30 putative variables in a multiple logistic regression model in the full data set (n = 10,389) to identify 20 candidate variables obtained from the electronic medical record (EMR) within 24 h of admission that were associated with 30-day in-hospital mortality (p < 0.05). These 20 variables were tested using multiple logistic regression and area under the curve (AUC)-receiver operating characteristics (ROC) analysis to identify an optimal risk threshold score in a randomly split derivation sample (n = 5194) which was then examined in the validation sample (n = 5195).

PARTICIPANTS

Ten thousand three hundred eighty-nine patients greater than 18 years transferred to the Indiana University (IU)-Adult Academic Health Center (AHC) between 1/1/2016 and 12/31/2017.

MAIN MEASURES

Sensitivity, specificity, positive predictive value, C-statistic, and risk threshold score of the model.

KEY RESULTS

The final model was strongly discriminative (C-statistic = 0.90) and had a good fit (Hosmer-Lemeshow goodness-of-fit test [X² (8) =6.26, p = 0.62]). The positive predictive value for 30-day in-hospital death was 68%; AUC-ROC was 0.90 (95% confidence interval 0.89-0.92, p < 0.0001). We identified a risk threshold score of -2.19 that had a maximum sensitivity (79.87%) and specificity (85.24%) in the derivation and validation sample (sensitivity: 75.00%, specificity: 85.71%). In the validation sample, 34.40% (354/1029) of the patients above this threshold died compared to only 2.83% (118/4166) deaths below this threshold.

CONCLUSION

This model can use EMR and administrative data within 24 h of transfer to predict the risk of 30-day in-hospital mortality with reasonable accuracy among seriously ill transferred patients.

Impact of an Antibiotic Side-Chain-Based Cross-reactivity Chart Combined With Enhanced Allergy Assessment Processes for Surgical Prophylaxis Antimicrobials in Patients With β-Lactam Allergies

BACKGROUND

β-Lactam antibiotics are first-line therapy for perioperative prophylaxis; however, patient-reported allergies often lead to increased prescribing of alternative antibiotics that may increase the incidence of surgical site infections. The R-group side chain of the β-lactam ring is responsible for allergic cross-reactivity and experts recommend the use of β-lactams that are structurally dissimilar.

METHODS

An internally developed, antibiotic side-chain-based cross-reactivity chart was developed and implemented alongside enhanced allergy assessment processes. This single-center, quasi-experimental study analyzed antibiotic prescribing in all adult patients with a documented β-lactam allergy undergoing an inpatient surgical procedure between quartile (Q) 1 (2012)-Q3 (2014) (historical group) and Q3 (2016)-Q3 (2018) (intervention group). Propensity-weighted scoring analyses compared categorical and continuous outcomes. Interrupted time-series analysis further analyzed key outcomes.

RESULTS

A total of 1119 and 1089 patients were included in the historical and intervention cohorts, respectively. There was a significant difference in patients receiving a β-lactam alternative antibiotic between cohorts (84.9% vs 15.1%; P < .001). There was a decrease in 30-day readmissions in the intervention cohort (7.9% vs 6.3%; P = .035); however, there was no difference in the incidence of SSIs in patients readmitted (14.8% vs 13%; P = .765). No significant differences were observed in allergic reactions (0.5% vs 0.3%; P = .323), surgical site infections, in-hospital and 30-day mortality, healthcare facility-onset Clostridiodes difficile infection, acute kidney injury, or hospital costs.

CONCLUSIONS

Implementation of an antibiotic cross-reactivity chart combined with enhanced allergy assessment processes significantly improved the prescribing of β-lactam antibiotics for surgical prophylaxis.

Impact of a Mortality Prediction Rule for Organizing and Guiding Antimicrobial Stewardship Program Activities

BACKGROUND

Antimicrobial stewardship program (ASP) surveillance at our hospital is supplemented by an internally developed surveillance database. In 2013, the database incorporated a validated, internally developed, prediction rule for patient mortality within 30 days of hospital admission. This study describes the impact of an expanded ASP review in patients at the highest risk for mortality.

METHODS

This retrospective, quasi-experimental study analyzed adults who received antimicrobials with the highest mortality risk score. Study periods were defined as 2011-Q3 2013 (historical group) and Q4 2013-2018 (intervention group). Primary and secondary outcomes were assessed for confounders and analyzed using both unadjusted and propensity score weighted analyses. Interrupted time-series analyses also analyzed key outcomes.

RESULTS

A total of 3282 and 5456 patients were included in the historical and intervention groups, respectively. There were significant reductions in median antimicrobial duration (5 vs 4 days; P < .001), antimicrobial days of therapy (8 vs 7; P < .001), antimicrobial cost ($96 vs $85; P = .003), length of stay (LOS) (6 vs 5 days; P < .001), intensive care unit (ICU) LOS (3 vs 2 days; P < .001), total hospital cost ($10 946 vs $9119; P < .001), healthcare facility-onset vancomycin-resistant Enterococcus (HO-VRE) incidence (1.3% vs 0.3%; P ≤ .001), and HO-VRE infections (0.6% vs 0.2%; P = .018) in the intervention cohort.

CONCLUSIONS

Reductions in antimicrobial use, hospital and ICU LOS, HO-VRE, HO-VRE infections, and costs were associated with incorporation of a novel mortality prediction rule to guide ASP surveillance and intervention.

Validation of a prospective mortality prediction score for hip fracture patients

PURPOSE

Although mortality prediction tools are the subject of significant interest as components of comprehensive hip fracture protocols, few have been applied or validated to prospectively inform ongoing patient management. Five regional hospitals are currently generating real-time mortality risk scores for all adults at the time of admission using available laboratory and comorbidity data (Cowen et al. J Hosp Med 9(11):720-726, 2014). Although results for aggregated conditions have been published, the primary aim of this study is to determine how well prospectively calculated scores predict mortality for hip fracture patients specifically.

METHODS

Using a five-hospital database, 1376 patients who were prospectively scored on admission were identified from January 2013 to April 2017, cross-referencing ICD9/10 diagnosis and procedure codes for AO/OTA 31A1 through 31B3 fractures. Prospective mortality scores have been previously divided into 5 risk categories to facilitate ease of clinical use. Vital status was determined from hospital data, Social Security and Michigan Death Indices.

RESULTS

Prospective scores demonstrated good mortality prediction, with AUCs of 0.80, 0.73, 0.74 and 0.74 for in hospital, 30-, 60- and 90-day mortality, respectively. Patients in the top 2 mortality risk categories represented 30% (410/1376) of the cohort and accounted for 78% (25/32) of the inpatient and 59% (57/97) of the 30 day deaths.

CONCLUSIONS

Implementation of this real-time mortality risk tool is feasible and valid for the prediction of short- to medium-term mortality risk for hip fracture patients, and potentially offers valuable information to guide ongoing patient management decisions such as admitting service or level of care.

Clinical Analytics Prediction Engine (CAPE): Development, electronic health record integration and prospective validation of hospital mortality, 180-day mortality and 30-day readmission risk prediction models

BACKGROUND

Numerous predictive models in the literature stratify patients by risk of mortality and readmission. Few prediction models have been developed to optimize impact while sustaining sufficient performance.

OBJECTIVE

We aimed to derive models for hospital mortality, 180-day mortality and 30-day readmission, implement these models within our electronic health record and prospectively validate these models for use across an entire health system.

MATERIALS & METHODS

We developed, integrated into our electronic health record and prospectively validated three predictive models using logistic regression from data collected from patients 18 to 99 years old who had an inpatient or observation admission at NorthShore University HealthSystem, a four-hospital integrated system in the United States, from January 2012 to September 2018. We analyzed the area under the receiver operating characteristic curve (AUC) for model performance.

RESULTS

Models were derived and validated at three time points: retrospective, prospective at discharge, and prospective at 4 hours after presentation. AUCs of hospital mortality were 0.91, 0.89 and 0.77, respectively. AUCs for 30-day readmission were 0.71, 0.71 and 0.69, respectively. 180-day mortality models were only retrospectively validated with an AUC of 0.85.

DISCUSSION

We were able to retain good model performance while optimizing potential model impact by also valuing model derivation efficiency, usability, sensitivity, generalizability and ability to prescribe timely interventions to reduce underlying risk. Measuring model impact by tying prediction models to interventions that are then rapidly tested will establish a path for meaningful clinical improvement and implementation.

Comparison of Artificial Neural Networks and Logistic Regression for 30-days Survival Prediction of Cancer Patients

Introduction

A machine learning technique that imitates neural system and brain can provide better than traditional methods like logistic regression for survival prediction and create an algorithm by determining influential factors.

Aim

To determine the influential factors on survival time of palliative care cancer patients and to compare two statistical methods for better prediction of survival.

Methods

One-year data is gathered from the patients that we followed in the palliative care clinic of our hospital (2017-2018) (n = 189). All data were retrospectively evaluated. After descriptive statistics, we used Pearson and Spearman correlations for parametric and non-parametric variables. The Artificial Neural Networks (ANN) and logistic regression model were applied to parameters which have a significant correlation with short survival.

Results

Significantly correlated variables with short survival were Palliative Performance Scale (PPS), Edmonton Symptom Assessment System (ESAS), Karnofsky Performance Scale (KPS), brain, liver, and distant metastasis, hemogram parameters, cero-reactive protein (CRP) and albumin (ALB). ANN model showed 89.3% prediction accuracy while the logistic regression model showed 73.0%. ANN model achieved a better AUC value of 0.86 than logistic regression model (0.76).

Discussion

There are several prognostic evaluation tools such as PPS, KPS, CRP, albumin, leukocytes, neutrophil were reported several studies as survival-related parameters in logistic regression models, also. Many studies compare ANN with logistic regression. When we evaluated these parameters totally, we observed the same relations with survival then we used the same parameters in the ANN model. The effectivity of the survival prediction models can be improved with the use of ANN.

Conclusion

ANN provides a more accurate estimation than logistic regression. ANN model is an important statistical method for survival prediction of cancer patients.

Leveraging Advances in Artificial Intelligence to Improve the Quality and Timing of Palliative Care

In recent years, research on artificial intelligence (AI) in medicine has seen great advances, especially with regards to the detection of diseases [...].

Heart failure mortality prediction using PRISM score and development of a classification and regression tree model to refer patients for palliative care consultation

INTRODUCTION

We sought to assess one-year mortality in heart failure (HF) patients by using (Placement Resource Indicator for Systems Management) PRISM, a disease nonspecific risk stratification score, and use it along with modified Seattle Heart Failure Model (SHFM) to guide patient selection for palliative care consultation.

METHODS

A retrospective study design was used to examine 1-year mortality in 689 HF patients admitted from 2012 to 2014. One-year mortality was calculated using Pmort30/PRISM and modified SHFM scores, and the predicted scores were validated using the area under the ROC curve. CART was used to develop an algorithm to classify patients based on their mortality risk.

RESULTS

The discriminatory ability of PRISM categorical score (AUC = 0.701) was not significantly different than the discriminatory ability of modified SHFM (AUC = 0.686) (DeLong's test p = 0.56) but improved significantly with the combination of PRISM (categorical) score + modified SHFM (AUC = 0.740) (p = 0.002). The predictive capability of the CART tree model after cross-validation was 72.2% (AUC 0.631).

CONCLUSION

Our study suggests PRISM score performed as well as modified SHFM for one-year mortality prediction. Moreover, the addition of modified SHFM to PRISM score increases discriminatory ability in predicting 1-year mortality in heart failure patients compared to either of the two models alone. Together, when combined in a CART model, they can be used to identify the population subset with the highest mortality risk and hence guide goals of care discussion.

Improving palliative care with deep learning

BACKGROUND

Access to palliative care is a key quality metric which most healthcare organizations strive to improve. The primary challenges to increasing palliative care access are a combination of physicians over-estimating patient prognoses, and a shortage of palliative staff in general. This, in combination with treatment inertia can result in a mismatch between patient wishes, and their actual care towards the end of life.

METHODS

In this work, we address this problem, with Institutional Review Board approval, using machine learning and Electronic Health Record (EHR) data of patients. We train a Deep Neural Network model on the EHR data of patients from previous years, to predict mortality of patients within the next 3-12 month period. This prediction is used as a proxy decision for identifying patients who could benefit from palliative care.

RESULTS

The EHR data of all admitted patients are evaluated every night by this algorithm, and the palliative care team is automatically notified of the list of patients with a positive prediction. In addition, we present a novel technique for decision interpretation, using which we provide explanations for the model's predictions.

CONCLUSION

The automatic screening and notification saves the palliative care team the burden of time consuming chart reviews of all patients, and allows them to take a proactive approach in reaching out to such patients rather then relying on referrals from the treating physicians.

Implementation of a Novel Algorithm to Decrease Unnecessary Hospitalizations in Patients Presenting to a Community Emergency Department With Atrial Fibrillation

OBJECTIVES

Atrial fibrillation (AFib) is the most common dysrhythmia in the United States. Patients seen in the emergency department (ED) in rapid AFib are often started on intravenous rate-controlling agents and admitted for several days. Although underlying and triggering illnesses must be addressed, AFib, intrinsically, is rarely life-threatening and can often be safely managed in an outpatient setting. At our academic community hospital, we implemented an algorithm to decrease hospital admissions for individuals presenting with a primary diagnosis of AFib. We focused on lenient oral rate control and discharge home. Our study evaluates outcomes after implementation of this algorithm.

METHODS

Study design is a retrospective cohort analysis pre- and postimplementation of the algorithm. The primary outcome was hospital admissions. Secondary outcomes were 3- and 30-day ED visits and any associated hospital admissions. These outcomes were compared before (March 2013-February 2014) and after (March 2015-February 2016) implementation. Chi-square tests and logistic regressions were run to test for significant changes in the three outcome variables.

RESULTS

A total of 1,108 individuals met inclusion criteria with 586 patients in the preimplementation group and 522 in the postimplementation group. Cohorts were broadly comparable in terms of demographics and health histories. Admissions for persons presenting with AFib after implementation decreased significantly (80.4% pre vs. 67.4% post, adjusted odds ratio [OR] = 3.4, p < 0.001). Despite this difference there was no change in ED return rates within 3 or 30 days (adjusted ORs = 0.93 and 0.89, p = 0.91 and 0.73, respectively).

CONCLUSIONS

Implementation of a novel algorithm to identify and treat low-risk patients with AFib can significantly decrease the rate of hospital admissions without increased ED returns. This simple algorithm could be adopted by other community hospitals and help lower costs.

Risk Score to Predict Need for Intensive Care in Initially Hemodynamically Stable Adults With Non-ST-Segment-Elevation Myocardial Infarction

Background

Intensive care unit (ICU) use for initially stable patients presenting with non–ST‐segment–elevation myocardial infarction (NSTEMI) varies widely across hospitals and minimally correlates with severity of illness. We aimed to develop a bedside risk score to assist in identifying high‐risk patients with NSTEMI for ICU admission.

Methods and Results

Using the Acute Coronary Treatment and Intervention Outcomes Network (ACTION) Registry linked to Medicare data, we identified patients with NSTEMI aged ≥65 years without cardiogenic shock or cardiac arrest on presentation. Complications requiring ICU care were defined as subsequent development of cardiac arrest, shock, high‐grade atrioventricular block, respiratory failure, stroke, or death during the index hospitalization. We developed and validated a model and integer risk score (Acute Coronary Treatment and Intervention Outcomes Network (ACTION) ICU risk score) that uses variables present at hospital admission to predict requirement for ICU care. Of 29 973 patients with NSTEMI, 4282 (14%) developed a complication requiring ICU‐level care, yet 12 879 (43%) received care in an ICU. Signs or symptoms of heart failure, initial heart rate, initial systolic blood pressure, initial troponin, initial serum creatinine, prior revascularization, chronic lung disease, ST‐segment depression, and age had statistically significant associations with requirement for ICU care after adjusting for other risk factors. The ACTION ICU risk score had a C‐statistic of 0.72. It identified 11% of patients as having very high risk (>30%) of developing complications requiring ICU care and 49% as having low likelihood (<10%) of requiring an ICU.

Conclusions

The ACTION ICU risk score quantifies the risk of initially stable patients with NSTEMI developing a complication requiring ICU care, and could be used to more effectively allocate limited ICU resources.

Data-driven optimization methodology for admission control in critical care units

The decision of whether to admit a patient to a critical care unit is a crucial operational problem that has significant influence on both hospital performance and patient outcomes. Hospitals currently lack a methodology to selectively admit patients to these units in a way that patient health risk metrics can be incorporated while considering the congestion that will occur. The hospital is modeled as a complex loss queueing network with a stochastic model of how long risk-stratified patients spend time in particular units and how they transition between units. A Mixed Integer Programming model approximates an optimal admission control policy for the network of units. While enforcing low levels of patient blocking, we optimize a monotonic dual-threshold admission policy. A hospital network including Intermediate Care Units (IMCs) and Intensive Care Units (ICUs) was considered for validation. The optimized model indicated a reduction in the risk levels required for admission, and weekly average admissions to ICUs and IMCs increased by 37% and 12%, respectively, with minimal blocking. Our methodology captures utilization and accessibility in a network model of care pathways while supporting the personalized allocation of scarce care resources to the neediest patients. The interesting benefits of admission thresholds that vary by day of week are studied.

The risk-outcome-experience triad: Mortality risk and the hospital consumer assessment of healthcare providers and systems survey

BACKGROUND

Studies have shown an association between the Hospital Consumer Assessment of Healthcare Providers and Systems Survey (HCAHPS) scores and clinical quality. The mortality risk on admission predicts adverse events. It is not known if this risk also portends a suboptimal patient experience.

OBJECTIVE

To determine if the admission mortality risk identifies an experience of care risk.

DESIGN

A retrospectively assembled cohort in which individual HCAHPS survey responses were linked to the admission risk of dying.

SETTING

Five community hospitals of various sizes in Michigan.

PATIENTS

There were 17,509 HCAHPS medical and surgical respondents; 2513 (14.4%) were at high risk of dying.

MEASUREMENTS

Odds ratio (OR) (high-risk patients to low-risk patients) for providing a top box score for HCAHPS dimensions, controlling for hospital and the standard HCAHPS patient mix adjustment factors.

RESULTS

High-risk respondents were less likely to provide the most favorable response (unadjusted) for all HCAHPS domains, although the difference was not significant (P = 0.09) for discharge information. Multivariable analyses indicated that high-risk patients were less likely to report a top box experience for doctor communication (OR: 0.85; 95% confidence interval [CI]: 0.77-0.94) and responsiveness of hospital staff (OR: 0.77; 95% CI: 0.69-0.85), but were more likely to have received adequate discharge information (OR: 1.30, 95% CI: 1.14-1.48).

CONCLUSIONS

Patients at high risk of dying who completed surveys were less likely to report favorable physician communication and staff responsiveness. Further understanding of these relationships may help design a care model to improve both outcomes and experience. Journal of Hospital Medicine 2016;11:628-635. © 2016 Society of Hospital Medicine.

Implementation of a mortality prediction rule for real-time decision making: feasibility and validity

BACKGROUND

A previously published, retrospectively derived prediction rule for death within 30 days of hospital admission has the potential to launch parallel interdisciplinary team activities. Whether or not patient care improves will depend on the validity of prospectively generated predictions, and the feasibility of generating them on demand for a critical proportion of inpatients.

OBJECTIVE

To determine the feasibility of generating mortality predictions on admission and to validate their accuracy using the scoring weights of the retrospective rule.

DESIGN

Prospective, sequential cohort.

SETTING

Large, tertiary care, community hospital in the Midwestern United States

PATIENTS

Adult patients admitted from the emergency department or scheduled for elective surgery

RESULTS

Mortality predictions were generated on demand at the beginning of the hospitalization for 9312 (92.9%) out of a possible 10,027 cases. The area under the receiver operating curve for 30-day mortality was 0.850 (95% confidence interval: 0.833-0.866), indicating very good to excellent discrimination. The prospectively generated 30-day mortality risk had a strong association with the receipt of palliative care by hospital discharge, in-hospital mortality, and 180-day mortality, a fair association with the risk for 30-day readmissions and unplanned transfers to intensive care, and weak associations with receipt of intensive unit care ever within the hospitalization or the development of a new diagnosis that was not present on admission (ie, complication).

CONCLUSIONS

Important prognostic information is feasible to obtain in a real-time, single-assessment process for a sizeable proportion of hospitalized patients.

The sentinel hospitalization and the role of palliative care

With current healthcare reform and calls for improving care quality and safety, there is renewed emphasis on high-value care. Moreover, given the significant healthcare resource utilization for patients with chronically progressive illnesses or for patients at the end of life, innovative and efficient care delivery models are urgently needed. We propose here the concept of a sentinel hospitalization, defined as a transitional point in the patient's disease course that heralds a need to reassess prognosis, patient understanding, treatment options and intensities, and goals of care. Hospitalists are well positioned to recognize a patient's sentinel hospitalization and use it as an opportunity for active integration of palliative care that provides high-quality and cost-saving care through its patient- and family-oriented approach, its interdisciplinary nature, and its focus on symptom control and care coordination.

Palliative care in the emergency department

The Emergency Department (ED) is the place where people most frequently seek urgent care. For patients living with chronic disease or malignancy who may be in a crisis, this visit may be pivotal in determining the patients' trajectory. There is a large movement in education of emergency medicine physicians, hospitalists, and intensivists from acute aggressive interventions to patient-goal assessment, recognizing last stages of life and prioritizing symptom management. Although the ED is not considered an ideal place to begin palliative care, hospital-based physicians may assist in eliciting the patient's goals of care and discussing prognosis and disease trajectory. This may help shift to noncurative treatment. This article will summarize the following: identification of patients who may need palliation, discussing prognosis, eliciting goals of care and directives, symptom management in the ED, and making plans for further care. These efforts have been shown to improve outcomes and to decrease length of stay and cost. The focus of this article is relieving "patient" symptoms and family distress, honoring the patient's goals of care, and assisting in transition to a noncurative approach and placement where this may be accomplished.