An Automated Model to Identify Heart Failure Patients at Risk for 30-Day Readmission or Death Using Electronic Medical Record Data

Factors associated with readmission in chronic kidney disease: Systematic review and meta-analysis

BACKGROUND

Risk factors associated with all-cause hospital readmission are poorly characterised in patients with chronic kidney disease.

OBJECTIVE

A systematic review and meta-analysis were conducted to identify risk factors and protectors of hospital readmission in chronic kidney disease.

DESIGN, PARTICIPANTS & MEASUREMENTS

Studies involving adult patients were identified from four databases from inception to 31/03/2020. Random-effects meta-analyses were conducted to determine factors associated with all-cause 30-day hospital readmission in general chronic kidney disease, in dialysis and in kidney transplant recipient groups.

RESULTS

Eighty relevant studies (chronic kidney disease, n = 14 studies; dialysis, n = 34 studies; and transplant, n = 32 studies) were identified. Meta-analysis revealed that in both chronic kidney disease and transplant groups, increasing age in years and days spent at the hospital during the initial stay were associated with a higher risk of 30-day readmission. Other risk factors identified included increasing body mass index (kg/m2 ) in the transplant group, and functional impairment and discharge destination in the dialysis group. Within the chronic kidney disease ​​​​group, having an outpatient follow-up appointment with a nephrologist within 14 days of discharge was protective against readmission but this was not protective if provided by a primary care provider or a cardiologist.

CONCLUSION

Risk-reduction interventions that can be implemented include a nephrologist appointment within 14 days of hospital discharge, rehabilitation programme for functional improvement in the dialysis group and meal plans in the transplant group. Future risk analysis should focus on modifiable factors to ensure that strategies can be tested and implemented in those who are more at risk.

Predictors of 30-Day Readmission in Patients Hospitalized With Heart Failure as a Primary Versus Secondary Diagnosis

Short-term rehospitalizations are common, costly, and detrimental to patients with heart failure (HF). Current research and policy have focused primarily on 30-day readmissions for patients with HF as a primary diagnosis at index hospitalization, whereas a much larger population of patients are admitted with HF as a secondary diagnosis. This study aims to compare patients initially hospitalized for HF as either a primary or a secondary diagnosis, and to identify the most important factors in predicting 30-day readmission. Patients admitted with HF between 2014 and 2016 in the Nationwide Readmissions Database were included and divided into 2 cohorts: those admitted with a primary and secondary diagnosis of HF. Multivariable logistic regression was performed to predict 30-day readmission. Statistically significant predictors in multivariable logistic regression were used for dominance analysis to rank these factors by relative importance. Co-morbidities were the major driver of increased risk of 30-day readmission in both groups. Individual Elixhauser co-morbidities and the Elixhauser co-morbidity indexes were significantly associated with an increase in 30-day readmission. The 5 most important predictors of 30-day readmission according to dominance analysis were age, Elixhauser co-morbidity indexes of co-morbidity complications and readmission, number of diagnoses, and renal failure. These 5 factors accounted for 68% of the 30-day readmission risk. Measures of patient co-morbidities were among the strongest predictors of readmission risk. This study highlights the importance of expanding predictive models to include a broader set of clinical measures to create better-performing models of readmission risk for HF patients.

Predicting 30-day readmission following total knee arthroplasty using machine learning and clinical expertise applied to clinical administrative and research registry data in an Australian cohort

BACKGROUND

Thirty-day readmission is an increasingly important problem for total knee arthroplasty (TKA) patients. The aim of this study was to develop a risk prediction model using machine learning and clinical insight for 30-day readmission in primary TKA patients.

METHOD

Data used to train and internally validate a multivariable predictive model were obtained from a single tertiary referral centre for TKA located in Victoria, Australia. Hospital administrative data and clinical registry data were utilised, and predictors were selected through systematic review and subsequent consultation with clinicians caring for TKA patients. Logistic regression and random forest models were compared to one another. Calibration was evaluated by visual inspection of calibration curves and calculation of the integrated calibration index (ICI). Discriminative performance was evaluated using the area under the receiver operating characteristic curve (AUC-ROC).

RESULTS

The models developed in this study demonstrated adequate calibration for use in the clinical setting, despite having poor discriminative performance. The best-calibrated readmission prediction model was a logistic regression model trained on administrative data using risk factors identified from systematic review and meta-analysis, which are available at the initial consultation (ICI = 0.012, AUC-ROC = 0.589). Models developed to predict complications associated with readmission also had reasonable calibration (ICI = 0.012, AUC-ROC = 0.658).

CONCLUSION

Discriminative performance of the prediction models was poor, although machine learning provided a slight improvement. The models were reasonably well calibrated, meaning they provide accurate patient-specific probabilities of these outcomes. This information can be used in shared clinical decision-making for discharge planning and post-discharge follow up.

Young Women With Acute Myocardial Infarction: Risk Prediction Model for 1-Year Hospital Readmission

Background

Although young women ( aged ≤ 55 years) are at higher risk than similarly aged men for hospital readmission within 1 year after an acute myocardial infarction (AMI), no risk prediction models have been developed for them. The present study developed and internally validated a risk prediction model of 1-year post-AMI hospital readmission among young women that considered demographic, clinical, and gender-related variables.

Methods

We used data from the US Variation in Recovery: Role of Gender on Outcomes of Young AMI Patients (VIRGO) study (n = 2007 women), a prospective observational study of young patients hospitalized with AMI. Bayesian model averaging was used for model selection and bootstrapping for internal validation. Model calibration and discrimination were respectively assessed with calibration plots and area under the curve.

Results

Within 1-year post-AMI, 684 women (34.1%) were readmitted to the hospital at least once. The final model predictors included: any in-hospital complication, baseline perceived physical health, obstructive coronary artery disease, diabetes, history of congestive heart failure, low income ( < $30,000 US), depressive symptoms, length of hospital stay, and race (White vs Black). Of the 9 retained predictors, 3 were gender-related. The model was well calibrated and exhibited modest discrimination (area under the curve = 0.66).

Conclusions

Our female-specific risk model was developed and internally validated in a cohort of young female patients hospitalized with AMI and can be used to predict risk of readmission. Whereas clinical factors were the strongest predictors, the model included several gender-related variables (ie, perceived physical health, depression, income level). However, discrimination was modest, indicating that other unmeasured factors contribute to variability in hospital readmission risk among younger women.

Pulmonary Congestion on Lung Ultrasound Predicts Increased Risk of 30-Day Readmission in Heart Failure Patients

OBJECTIVES

Heart failure exacerbations are a common cause of hospitalizations with a high readmission rate. There are few validated predictors of readmission after treatment for acute decompensated heart failure (ADHF). Lung ultrasound (LUS) is sensitive and specific in the assessment of pulmonary congestion; however, it is not frequently utilized to assess for congestion before discharge. This study assessed the association between number of B-lines, on LUS, at patient discharge and risk of 30-day readmission in patients hospitalized for acute decompensated heart failure (ADHF).

METHODS

This was a single-center prospective study of adults admitted to a quaternary care center with a diagnosis of ADHF. At the time of discharge, the patient received an 8-zone LUS exam to evaluate for the presence of B-lines. A zone was considered positive if ≥3 B-lines was present. We assessed the risk of 30-day readmission associated with the number of lung zones positive for B-lines using a log-binomial regression model.

RESULTS

Based on data from 200 patients, the risk of 30-day readmission in patients with 2-3 positive lung zones was 1.25 times higher (95% CI: 1.08-1.45), and in patients with 4-8 positive lung zones was 1.50 times higher (95% CI: 1.23-1.82, compared with patients with 0-1 positive zones, after adjusting for discharge blood urea nitrogen, creatinine, and hemoglobin.

CONCLUSION

Among patients admitted with ADHF, the presence of B-lines at discharge was associated with a significantly increased risk of 30-day readmission, with greater number of lung zones positive for B-lines corresponding to higher risk.

Intensive care management for high-risk veterans in a patient-centered medical home - do some veterans benefit more than others?

BACKGROUND

Primary care intensive management programs utilize interdisciplinary care teams to comprehensively meet the complex care needs of patients at high risk for hospitalization. The mixed evidence on the effectiveness of these programs focuses on average treatment effects that may mask heterogeneous treatment effects (HTEs) among subgroups of patients. We test for HTEs by patients' demographic, economic, and social characteristics.

METHODS

Retrospective analysis of a VA randomized quality improvement trial. 3995 primary care patients at high risk for hospitalization were randomized to primary care intensive management (n = 1761) or usual primary care (n = 1731). We estimated HTEs on ED and hospital utilization one year after randomization using model-based recursive partitioning and a pre-versus post-with control group framework. Splitting variables included administratively collected demographic characteristics, travel distance, copay exemption, risk score for future hospitalizations, history of hospital discharge against medical advice, homelessness, and multiple residence ZIP codes.

RESULTS

There were no average or heterogeneous treatment effects of intensive management one year after enrollment. The recursive partitioning algorithm identified variation in effects by risk score, homelessness, and whether the patient had multiple residences in a year. Within each distinct subgroup, the effect of intensive management was not statistically significant.

CONCLUSIONS

Primary care intensive management did not affect acute care use of high-risk patients on average or differentially for patients defined by various demographic, economic, and social characteristics.

IMPLICATIONS

Reducing acute care use for high-risk patients is complex, and more work is required to identify patients positioned to benefit from intensive management programs.

Associations between continuity of care, perceived control and self-care and their impact on health-related quality of life and hospital readmission-A structural equation model

AIM

The aim of this study is to examine whether a conceptual model including the associations between continuity of care, perceived control and self-care could explain variations in health-related quality of life and hospital readmissions in people with chronic cardiac conditions after hospital discharge.

DESIGN

Correlational design based on cross-sectional data from a multicentre survey study.

METHODS

People hospitalized due to angina, atrial fibrillation, heart failure or myocardial infarction were included at four hospitals using consecutive sampling procedures during 2017-2019. Eligible people received questionnaires by regular mail 4-6 weeks after discharge. A tentative conceptual model describing the relationship between continuity of care, self-care, perceived control, health-related quality of life and readmission was developed and evaluated using structural equation modelling.

RESULTS

In total, 542 people (mean age 75 years, 37% females) were included in the analyses. According to the structural equation model, continuity of care predicted self-care, which in turn predicted health-related quality of life and hospital readmission. The association between continuity of care and self-care was partly mediated by perceived control. The model had an excellent model fit: RMSEA = 0.06, 90% CI, 0.05-0.06; CFI = 0.90; TLI = 0.90.

CONCLUSION

Interventions aiming to improve health-related quality of life and reduce hospital readmission rates should focus on enhancing continuity of care, perceived control and self-care.

IMPACT

This study reduces the knowledge gap on how central factors after hospitalization, such as continuity of care, self-care and perceived control, are associated with improved health-related quality of life and hospital readmission in people with cardiac conditions. The results suggest that these factors together predicted the quality of life and readmissions in this sample. This knowledge is relevant to researchers when designing interventions or predicting health-related quality of life and hospital readmission. For clinicians, it emphasizes that enhancing continuity of care, perceived control and self-care positively impacts clinical outcomes.

PATIENT OR PUBLIC CONTRIBUTION

People and healthcare personnel evaluated content validity and were included in selecting items for the short version.

Influence of Social Determinants of Health on Heart Failure Outcomes: A Systematic Review

Background Prior research suggests an association between clinical outcomes in heart failure (HF) and social determinants of health (SDoH). Because providers should identify and address SDoH in care delivery, we evaluated how SDoH have been defined, measured, and evaluated in studies that examine HF outcomes. Methods and Results Following Preferred Reporting Items for Systematic Reviews and Meta-Analysis guidelines, databases were searched for observational or interventional studies published between 2009 and 2021 that assessed the influence of SDoH on outcomes. Selected articles were assessed for quality using a validated rating scheme. We identified 1373 unique articles for screening; 104 were selected for full-text review, and 59 met the inclusion criteria, including retrospective and prospective cohort, cross-sectional, and intervention studies. The majority examined readmissions and hospitalizations (k=33), mortality or survival (k=29), and success of medical devices and transplantation (k=8). SDoH examined most commonly included race, ethnicity, age, sex, socioeconomic status, and education or health literacy. Studies used a range of 1 to 9 SDoH as primary independent variables and 0 to 7 SDoH as controls. Multiple data sources were employed and frequently were electronic medical records linked with national surveys and disease registries. The effects of SDoH on HF outcomes were inconsistent because of the heterogeneity of data sources and SDoH constructs. Conclusions Our systematic review reveals shortcomings in measurement and deployment of SDoH variables in HF care. Validated measures need to be prospectively and intentionally collected to facilitate appropriate analysis, reporting, and replication of data across studies and inform the design of appropriate, evidence-based interventions that can ameliorate significant HF morbidity and societal costs.

The effects of department of Veterans Affairs medical centers on socio-economic outcomes: Evidence from the Paycheck Protection Program

Do medical facilities also help advance improvements in socio-economic outcomes? We focus on Veterans, a vulnerable group over the COVID-19 pandemic who have access to a comprehensive healthcare network, and the receipt of funds from the Paycheck Protection Program (PPP) between April and June as a source of variation. First, we find that Veterans received 3.5% more loans and 6.8% larger loans than their counterparts (p < 0.01), controlling for a wide array of zipcode characteristics. Second, we develop models to predict the number of PPP loans awarded to Veterans, finding that the inclusion of local VA medical center characteristics adds almost as much explanatory power as the industry and occupational composition in an area and even more than the education, race, and age distribution combined. Our results suggest that VA medical centers can play an important role in helping Veterans thrive even beyond addressing their direct medical needs.

Identifying patients at risk of unplanned re-hospitalisation using statewide electronic health records

Preventing unplanned hospitalisations, including readmissions and re-presentations to the emergency department, is an important strategy for addressing the growing demand for hospital care. Significant successes have been reported from interventions put in place by hospitals to reduce their incidence. However, there is limited use of data-driven algorithms in hospital services to identify patients for enrolment into these intervention programs. Here we present the results of a study aiming to develop algorithms deployable at scale as part of a state government’s initiative to address rehospitalizations and which fills several gaps identified in the state-of-the-art literature. To the best of our knowledge, our study involves the largest-ever sample size for developing risk models. Logistic regression, random forests and gradient boosted techniques were explored as model candidates and validated retrospectively on five years of data from 27 hospitals in Queensland, Australia. The models used a range of predictor variables sourced from state-wide Emergency Department(ED), inpatient, hospital-dispensed medications and hospital-requested pathology databases. The investigation leads to several findings: (i) the advantage of looking at a longer patient data history, (ii) ED and inpatient datasets alone can provide useful information for predicting hospitalisation risk and the addition of medications and pathology test results leads to trivial performance improvements, (iii) predicting readmissions to the hospital was slightly easier than predicting re-presentations to ED after an inpatient stay, which was slightly easier again than predicting re-presentations to ED after an EDstay, (iv) a gradient boosted approach (XGBoost) was systematically the most powerful modelling approach across various tests.

Postpartum Hospital Readmissions and Emergency Department Visits Among High-Risk, Medicaid-Insured Women in New York City

Objectives: To describe the incidence of and characteristics associated with postpartum emergency department (ED) visits and hospital readmissions among high-risk, low-income, predominantly Black and Latina women in New York City (NYC). Methods: We conducted a secondary analysis of detailed survey and medical chart data from an intervention to improve timely postpartum visits among Medicaid-insured, high-risk women in NYC from 2015 to 2016. Among 380 women who completed surveys at baseline (bedside postpartum) and 3 weeks after delivery, we examined the incidence of having an ED visit or readmission within 3 weeks postpartum. We used logistic regression to examine unadjusted and adjusted associations between patient demographic, clinical, and psychosocial characteristics and the odds of postpartum hospital use. Results: In total, 12.8% (n = 48) of women reported an ED visit or readmission within 3 weeks postpartum. Unadjusted odds of postpartum hospital use were higher among women who self-identified as Black versus Latina, U.S. born versus foreign born, and English versus Spanish speaking. Clinical and psychosocial characteristics associated with increased unadjusted odds of postpartum hospital use included cesarean delivery, hypertensive disorders of pregnancy, and positive depression or anxiety screen, and we found preliminary evidence of decreased hospital use among women breastfeeding at three weeks postpartum. The odds of seeking postpartum hospital care remained roughly 2.5 times higher among women with hypertension or depression/anxiety in adjusted analyses. Conclusions: We identified characteristics associated with ED visits and hospital readmissions among a high-risk subset of postpartum women in NYC. These characteristics, including depressive symptoms and hypertension, suggest women who may benefit from additional postpartum support to prevent maternal complications and reduce health disparities.

Machine Learning Patient-Specific Prediction of Heart Failure Hospitalization Using Cardiac MRI-Based Phenotype and Electronic Health Information

Background

Heart failure (HF) hospitalization is a dominant contributor of morbidity and healthcare expenditures in patients with systolic HF. Cardiovascular magnetic resonance (CMR) imaging is increasingly employed for the evaluation of HF given capacity to provide highly reproducible phenotypic markers of disease. The combined value of CMR phenotypic markers and patient health information to deliver predictions of future HF events has not been explored. We sought to develop and validate a novel risk model for the patient-specific prediction of time to HF hospitalization using routinely reported CMR variables, patient-reported health status, and electronic health information.

Methods

Standardized data capture was performed for 1,775 consecutive patients with chronic systolic HF referred for CMR imaging. Patient demographics, symptoms, Health-related Quality of Life, pharmacy, and routinely reported CMR features were provided to both machine learning (ML) and competing risk Fine-Gray-based models (FGM) for the prediction of time to HF hospitalization.

Results

The mean age was 59 years with a mean LVEF of 36 ± 11%. The population was evenly distributed between ischemic (52%) and idiopathic non-ischemic cardiomyopathy (48%). Over a median follow-up of 2.79 years (IQR: 1.59–4.04) 333 patients (19%) experienced HF related hospitalization. Both ML and competing risk FGM based models achieved robust performance for the prediction of time to HF hospitalization. Respective 90-day, 1 and 2-year AUC values were 0.87, 0.83, and 0.80 for the ML model, and 0.89, 0.84, and 0.80 for the competing risk FGM-based model in a holdout validation cohort. Patients classified as high-risk by the ML model experienced a 34-fold higher occurrence of HF hospitalization at 90 days vs. the low-risk group.

Conclusion

In this study we demonstrated capacity for routinely reported CMR phenotypic markers and patient health information to be combined for the delivery of patient-specific predictions of time to HF hospitalization. This work supports an evolving migration toward multi-domain data collection for the delivery of personalized risk prediction at time of diagnostic imaging.

Predicting robotic-assisted total knee arthroplasty operating time

Aims

No predictive model has been published to forecast operating time for total knee arthroplasty (TKA). The aims of this study were to design and validate a predictive model to estimate operating time for robotic-assisted TKA based on demographic data, and evaluate the added predictive power of CT scan-based predictors and their impact on the accuracy of the predictive model.

Methods

A retrospective study was conducted on 1,061 TKAs performed from January 2016 to December 2019 with an image-based robotic-assisted system. Demographic data included age, sex, height, and weight. The femoral and tibial mechanical axis and the osteophyte volume were calculated from CT scans. These inputs were used to develop a predictive model aimed to predict operating time based on demographic data only, and demographic and 3D patient anatomy data.

Results

The key factors for predicting operating time were the surgeon and patient weight, followed by 12 anatomical parameters derived from CT scans. The predictive model based only on demographic data showed that 90% of predictions were within 15 minutes of actual operating time, with 73% within ten minutes. The predictive model including demographic data and CT scans showed that 94% of predictions were within 15 minutes of actual operating time and 88% within ten minutes.

Conclusion

The primary factors for predicting robotic-assisted TKA operating time were surgeon, patient weight, and osteophyte volume. This study demonstrates that incorporating 3D patient-specific data can improve operating time predictions models, which may lead to improved operating room planning and efficiency.

Cite this article: Bone Jt Open 2022;3(5):383–389.

Utilizing Artificial Intelligence to Enhance Health Equity Among Patients with Heart Failure

Patients with heart failure (HF) are heterogeneous with various intrapersonal and interpersonal characteristics contributing to clinical outcomes. Bias, structural racism, and social determinants of health have been implicated in unequal treatment of patients with HF. Through several methodologies, artificial intelligence (AI) can provide models in HF prediction, prognostication, and provision of care, which may help prevent unequal outcomes. This review highlights AI as a strategy to address racial inequalities in HF; discusses key AI definitions within a health equity context; describes the current uses of AI in HF, strengths and harms in using AI; and offers recommendations for future directions.

Electronic Health Records and Heart Failure

Increasing the global adoption of electronic health records (EHRs) is transforming the delivery of clinical care. EHRs offer tools that are useful in the care of heart failure ranging from individualized risk stratification and decision support to population management. EHR tools can be combined to target specific areas of need such as the standardization of care, improved quality of care, and resource management. Leveraging EHR functionality has been shown to improve select outcomes including guideline-based therapies, reduction in adverse clinical outcomes, and improved cost-efficiency. Central to success is participation by clinicians and patients in the design and feedback of EHR tools.

Information Extraction From Electronic Health Records to Predict Readmission Following Acute Myocardial Infarction: Does Natural Language Processing Using Clinical Notes Improve Prediction of Readmission?

Background

Social risk factors influence rehospitalization rates yet are challenging to incorporate into prediction models. Integration of social risk factors using natural language processing (NLP) and machine learning could improve risk prediction of 30‐day readmission following an acute myocardial infarction.

Methods and Results

Patients were enrolled into derivation and validation cohorts. The derivation cohort included inpatient discharges from Vanderbilt University Medical Center between January 1, 2007, and December 31, 2016, with a primary diagnosis of acute myocardial infarction, who were discharged alive, and not transferred from another facility. The validation cohort included patients from Dartmouth‐Hitchcock Health Center between April 2, 2011, and December 31, 2016, meeting the same eligibility criteria described above. Data from both sites were linked to Centers for Medicare & Medicaid Services administrative data to supplement 30‐day hospital readmissions. Clinical notes from each cohort were extracted, and an NLP model was deployed, counting mentions of 7 social risk factors. Five machine learning models were run using clinical and NLP‐derived variables. Model discrimination and calibration were assessed, and receiver operating characteristic comparison analyses were performed. The 30‐day rehospitalization rates among the derivation (n=6165) and validation (n=4024) cohorts were 15.1% (n=934) and 10.2% (n=412), respectively. The derivation models demonstrated no statistical improvement in model performance with the addition of the selected NLP‐derived social risk factors.

Conclusions

Social risk factors extracted using NLP did not significantly improve 30‐day readmission prediction among hospitalized patients with acute myocardial infarction. Alternative methods are needed to capture social risk factors.

Readmissions performance and penalty experience of safety-net hospitals under Medicare's Hospital Readmissions Reduction Program

BACKGROUND

The Hospital Readmissions Reduction Program (HRRP), established by the Centers for Medicare and Medicaid Services (CMS) in March 2010, introduced payment-reduction penalties on acute care hospitals with higher-than-expected readmission rates for acute myocardial infarction (AMI), heart failure, and pneumonia. There is concern that hospitals serving large numbers of low-income and uninsured patients (safety-net hospitals) are at greater risk of higher readmissions and penalties, often due to factors that are likely outside the hospital's control. Using publicly reported data, we compared the readmissions performance and penalty experience among safety-net and non-safety-net hospitals.

METHODS

We used nationwide hospital level data for 2009-2016 from the Centers for Medicare and Medicaid Services (CMS) Hospital Compare program, CMS Final Impact Rule, and the American Hospital Association Annual Survey. We identified as safety-net hospitals the top quartile of hospitals in terms of the proportion of patients receiving income-based public benefits. Using a quasi-experimental difference-in-differences approach based on the comparison of pre- vs. post-HRRP changes in (risk-adjusted) 30-day readmission rate in safety-net and non-safety-net hospitals, we estimated the change in readmissions rate associated with HRRP. We also compared the penalty frequency among safety-net and non-safety-net hospitals.

RESULTS

Our study cohort included 1915 hospitals, of which 479 were safety-net hospitals. At baseline (2009), safety-net hospitals had a slightly higher readmission rate compared to non-safety net hospitals for all three conditions: AMI, 20.3% vs. 19.8% (p value< 0.001); heart failure, 25.2% vs. 24.2% (p-value< 0.001); pneumonia, 18.7% vs. 18.1% (p-value< 0.001). Beginning in 2012, readmission rates declined similarly in both hospital groups for all three cohorts. Based on difference-in-differences analysis, HRRP was associated with similar change in the readmissions rate in safety-net and non-safety-net hospitals for AMI and heart failure. For the pneumonia cohort, we found a larger reduction (0.23%; p < 0.001) in safety-net hospitals. The frequency of readmissions penalty was higher among safety-net hospitals. The proportion of hospitals penalized during all four post-HRRP years was 72% among safety-net and 59% among non-safety-net hospitals.

CONCLUSIONS

Our results lend support to the concerns of disproportionately higher risk of performance-based penalty on safety-net hospitals.

Mortality and Readmission Rates After Heart Failure: A Systematic Review and Meta-Analysis

Objective

The current work aimed to examine the rates of and risk factors for mortality and readmission after heart failure (HF).

Setting

A systematic search was carried out in PubMed, the Cochrane Library, and EMBASE to identify eligible reports. The random-effects model was utilized to evaluate the pooled results.

Participants

A total of 27 studies with 515,238 participants were finally meta-analysed. The HF patients had an average age of 76.3 years, with 51% of the sample being male, in the pooled analysis.

Primary and Secondary Outcome Measures

The outcome measures were 30-day and 1-year readmission rates, mortality, and risk factors for readmission and mortality.

Results

The effect sizes for readmission and mortality were estimated as the mean and 95% confidence interval (CI). The estimated 30-day and 1-year all-cause readmission rates were 0.19 (95% CI 0.14-0.23) and 0.53 (95% CI 0.46-0.59), respectively, while the all-cause mortality rates were 0.14 (95% CI 0.10-0.18) and 0.29 (95% CI 0.25-0.33), respectively. Comorbidities were highly prevalent in individuals with HF.

Conclusion

Heart failure hospitalization is followed by high readmission and mortality rates.

Predicting hospital readmission risk: A prospective observational study to compare primary care providers' assessments with the LACE readmission risk index

Purpose

This study aims to determine if the primary care provider (PCP) assessment of readmission risk is comparable to the validated LACE tool at predicting readmission to hospital.

Methods

A prospective observational study of recently discharged adult patients clustered by PCPs in the primary care setting. Physician readmission risk assessment was determined via a questionnaire after the PCP reviewed the hospital discharge summary. LACE scores were calculated using administrative data and the discharge summary. The sensitivity and specificity of the physician assessment and the LACE tool in predicting readmission risk, agreement between the 2 assessments and the area under receiver operating characteristic (AUROC) curves were calculated.

Results

217 patient readmission encounters were included in this study from September 2017 till June 2018. The rate of readmission within 30 days was 14.7%, and 217 discharge summaries were used for analysis. The weighted kappa coefficient was 0.41 (95% CI: 0.30–0.51) demonstrating a moderate level of agreement. Sensitivity of physician assessment was 0.31 (95% CI: 0.22–0.40) and specificity was 0.80 (95% CI: 0.77–0.83). The sensitivity of the LACE assessment was 0.42 (95% CI: 0.25–0.59) and specificity was 0.79 (95% CI: 0.73–0.85). The AUROC for the LACE readmission risk was 0.65 (95% C.I. 0.55–0.76) demonstrating modest predictive power and was 0.57 (95% C.I. 0.46–0.68) for physician assessment, demonstrating low predictive power.

Conclusion

The LACE index shows moderate discriminatory power in identifying high-risk patients for readmission when compared to the PCP’s assessment. If this score can be provided to the PCP, it may help identify patients who requires more intensive follow-up after discharge.

Disease evolution and risk-based disease trajectories in congestive heart failure patients

Congestive Heart Failure (CHF) is among the most prevalent chronic diseases worldwide, and is commonly associated with comorbidities and complex health conditions. Consequently, CHF patients are typically hospitalized frequently, and are at a high risk of premature death. Early detection of an envisaged patient disease trajectory is crucial for precision medicine. However, despite the abundance of patient-level data, cardiologists currently struggle to identify disease trajectories and track the evolution patterns of the disease over time, especially in small groups of patients with specific disease subtypes. The present study proposed a five-step method that allows clustering CHF patients, detecting cluster similarity, and identifying disease trajectories, and promises to overcome the existing difficulties. This work is based on a rich dataset of patients' records spanning ten years of hospital visits. The dataset contains all the health information documented in the hospital during each visit, including diagnoses, lab results, clinical data, and demographics. It utilizes an innovative Cluster Evolution Analysis (CEA) method to analyze the complex CHF population where each subject is potentially associated with numerous variables. We have defined sub-groups for mortality risk levels, which we used to characterize patients' disease evolution by refined data clustering in three points in time over ten years, and generating patients' migration patterns across periods. The results elicited 18, 23, and 25 clusters respective to the first, second, and third visits, uncovering clinically interesting small sub-groups of patients. In the following post-processing stage, we identified meaningful patterns. The analysis yielded fine-grained patient clusters divided into several finite risk levels, including several small-sized groups of high-risk patients. Significantly, the analysis also yielded longitudinal patterns where patients' risk levels changed over time. Four types of disease trajectories were identified: decline, preserved state, improvement, and mixed-progress. This stage is a unique contribution of the work. The resulting fine partitioning and longitudinal insights promise to significantly assist cardiologists in tailoring personalized interventions to improve care quality. Cardiologists could utilize these results to glean previously undetected relationships between symptoms and disease evolution that would allow a more informed clinical decision-making and effective interventions.

Prior Frequent Emergency Department Use as a Predictor of Emergency Department Visits After a New Cancer Diagnosis

PURPOSE

To determine whether emergency department (ED) visit history prior to cancer diagnosis is associated with ED visit volume after cancer diagnosis.

METHODS

This was a retrospective cohort study of adults (≥ 18 years) with an incident cancer diagnosis (excluding nonmelanoma skin cancers or leukemia) at an academic medical center between 2008 and 2018 and a safety-net hospital between 2012 and 2016. Our primary outcome was the number of ED visits in the first 6 months after cancer diagnosis, modeled using a multivariable negative binomial regression accounting for ED visit history in the 6-12 months preceding cancer diagnosis, electronic health record proxy social determinants of health, and clinical cancer-related characteristics.

RESULTS

Among 35,090 patients with cancer (49% female and 50% non-White), 57% had ≥ 1 ED visit in the 6 months immediately following cancer diagnosis and 20% had ≥ 1 ED visit in the 6-12 months prior to cancer diagnosis. The strongest predictor of postdiagnosis ED visits was frequent (≥ 4) prediagnosis ED visits (adjusted incidence rate ratio [aIRR]: 3.68; 95% CI, 3.36 to 4.02). Other covariates associated with greater postdiagnosis ED use included having 1-3 prediagnosis ED visits (aIRR: 1.32; 95% CI, 1.28 to 1.36), Hispanic (aIRR: 1.12; 95% CI, 1.07 to 1.17) and Black (aIRR: 1.21; 95% CI, 1.17 to 1.25) race, homelessness (aIRR: 1.95; 95% CI, 1.73 to 2.20), advanced-stage cancer (aIRR: 1.30; 95% CI, 1.26 to 1.35), and treatment regimens including chemotherapy (aIRR: 1.44; 95% CI, 1.40 to 1.48).

CONCLUSION

The strongest independent predictor for ED use after a new cancer diagnosis was frequent ED visits before cancer diagnosis. Efforts to reduce potentially avoidable ED visits among patients with cancer should consider educational initiatives that target heavy prior ED users and offer them alternative ways to seek urgent medical care.

Congestive Heart Failure 30-Day Readmission: Descriptive Study of Demographics, Co-morbidities, Heart Failure Knowledge, and Self-Care

Background Congestive heart failure (CHF) readmissions are associated with substantial financial and medical implications. We performed a descriptive study to determine demographic, clinical, and behavioral factors associated with 30-day readmission. Materials and methods Patients hospitalized with CHF at William Beaumont Hospital in Royal Oak, MI, from March 2019-May 2019 were studied. Response to heart failure knowledge and self-care questionnaires along with the patients' demographic and clinical factors were collected. Thirty-day readmission to any of the eight hospitals in the Beaumont Health System was documented. Results One-hundred ninety-six (196) patients were included. The all-cause 30-day readmission rate was 23%. A numerical higher rate of readmissions was observed among males (23.7% vs 22.2%), current smokers (27.3% vs 22.9%), and patients with peripheral vascular disease (PVD; 28.9% vs 21.2%), diabetes mellitus (DM; 26.4% vs 18.9%), hypertension (HTN; 26.4% vs 10%), coronary artery disease (CAD; 24.6% vs 19%), and prior history of cerebrovascular accident (CVA; 28.9% vs 21.2%) (p>0.05). Reduced left ventricular ejection fraction (LVEF) was associated with higher readmissions (24.4% vs 20.5%, p=0.801). Patients with the highest reported questionnaire scores corresponding to better heart failure knowledge and self-care behaviors at home were readmitted at a similar rate compared to those scoring in the lowest interval (25%, p=0.681). Conclusion Though statistically insignificant due to the limitations of sample size, a higher percentage of readmissions was observed in male patients, current smokers, reduced LVEF, and higher comorbidity burden. Better reported patient self-care behavior, medication compliance, and heart failure knowledge did not correlate with reduced readmission rates. While the impact of medical comorbidities on 30-day readmissions is better established, the role of socioeconomic factors remains unclear and might suggest a focus for future work.

Enhancing the value to users of machine learning-based clinical decision support tools: A framework for iterative, collaborative development and implementation

BACKGROUND

Health care organizations are integrating a variety of machine learning (ML)-based clinical decision support (CDS) tools into their operations, but practitioners lack clear guidance regarding how to implement these tools so that they assist end users in their work.

PURPOSE

We designed this study to identify how health care organizations can facilitate collaborative development of ML-based CDS tools to enhance their value for health care delivery in real-world settings.

METHODOLOGY/APPROACH

We utilized qualitative methods, including 37 interviews in a large, multispecialty health system that developed and implemented two operational ML-based CDS tools in two of its hospital sites. We performed thematic analyses to inform presentation of an explanatory framework and recommendations.

RESULTS

We found that ML-based CDS tool development and implementation into clinical workflows proceeded in four phases: iterative solution coidentification, iterative coengagement, iterative coapplication, and iterative corefinement. Each phase is characterized by a collaborative back-and-forth process between the technology's developers and users, through which both users' activities and the technology itself are transformed.

CONCLUSION

Health care organizations that anticipate iterative collaboration to be an integral aspect of their ML-based CDS tools' development and implementation process may have more success in deploying ML-based CDS tools that assist end users in their work than organizations that expect a traditional technology innovation process.

PRACTICE IMPLICATIONS

Managers developing and implementing ML-based CDS tools should frame the work as a collaborative learning opportunity for both users and the technology itself and should solicit constructive feedback from users on potential changes to the technology, in addition to potential changes to user workflows, in an ongoing, iterative manner.

Assessment of Automating Safety Surveillance From Electronic Health Records: Analysis for the Quality and Safety Review System

BACKGROUND AND OBJECTIVES

In an effort to improve and standardize the collection of adverse event data, the Agency for Healthcare Research and Quality is developing and testing a patient safety surveillance system called the Quality and Safety Review System (QSRS). Its current abstraction from medical records is through manual human coders, taking an average of 75 minutes to complete the review and abstraction tasks for one patient record. With many healthcare systems across the country adopting electronic health record (EHR) technology, there is tremendous potential for more efficient abstraction by automatically populating QSRS. In the absence of real-world testing data and models, which require a substantial investment, we provide a heuristic assessment of the feasibility of automatically populating QSRS questions from EHR data.

METHODS

To provide an assessment of the automation feasibility for QSRS, we first developed a heuristic framework, the Relative Abstraction Complexity Framework, to assess relative complexity of data abstraction questions. This framework assesses the relative complexity of characteristics or features of abstraction questions that should be considered when determining the feasibility of automating QSRS. Questions are assigned a final relative complexity score (RCS) of low, medium, or high by a team of clinicians, human factors, and natural language processing researchers.

RESULTS

One hundred thirty-four QSRS questions were coded using this framework by a team of natural language processing and clinical experts. Fifty-five questions (41%) had high RCS and would be more difficult to automate, such as "Was use of a device associated with an adverse outcome(s)?" Forty-two questions (31%) had medium RCS, such as "Were there any injuries as a result of the fall(s)?" and 37 questions (28%) had low RCS, such as "Did the patient deliver during this stay?" These results suggest that Blood and Hospital Acquired Infections-Clostridium Difficile Infection (HAI-CDI) modules would be relatively easier to automate, whereas Surgery and HAI-Surgical Site Infection would be more difficult to automate.

CONCLUSIONS

Although EHRs contain a wealth of information, abstracting information from these records is still very challenging, particularly for complex questions, such as those concerning patient adverse events. In this work, we developed a heuristic framework, which can be applied to help guide conversations around the feasibility of automating QSRS data abstraction. This framework does not aim to replace testing with real data but complement the process by providing initial guidance and direction to subject matter experts to help prioritize, which abstraction questions to test for feasibility using real data.

Man vs. Machine: Comparing Physician vs. Electronic Health Record-Based Model Predictions for 30-Day Hospital Readmissions

BACKGROUND

Electronic health record (EHR)-based readmission risk prediction models can be automated in real-time but have modest discrimination and may be missing important readmission risk factors. Clinician predictions of readmissions may incorporate information unavailable in the EHR, but the comparative usefulness is unknown. We sought to compare clinicians versus a validated EHR-based prediction model in predicting 30-day hospital readmissions.

METHODS

We conducted a prospective survey of internal medicine clinicians in an urban safety-net hospital. Clinicians prospectively predicted patients' 30-day readmission risk on 5-point Likert scales, subsequently dichotomized into low- vs. high-risk. We compared human with machine predictions using discrimination, net reclassification, and diagnostic test characteristics. Observed readmissions were ascertained from a regional hospitalization database. We also developed and assessed a "human-plus-machine" logistic regression model incorporating both human and machine predictions.

RESULTS

We included 1183 hospitalizations from 106 clinicians, with a readmission rate of 20.8%. Both clinicians and the EHR model had similar discrimination (C-statistic 0.66 vs. 0.66, p = 0.91). Clinicians had higher specificity (79.0% vs. 48.9%, p < 0.001) but lower sensitivity (43.9 vs. 75.2%, p < 0.001) than EHR model predictions. Compared with machine, human was better at reclassifying non-readmissions (non-event NRI + 30.1%) but worse at reclassifying readmissions (event NRI - 31.3%). A human-plus-machine approach best optimized discrimination (C-statistic 0.70, 95% CI 0.67-0.74), sensitivity (65.5%), and specificity (66.7%).

CONCLUSION

Clinicians had similar discrimination but higher specificity and lower sensitivity than EHR model predictions. Human-plus-machine was better than either alone. Readmission risk prediction strategies should incorporate clinician assessments to optimize the accuracy of readmission predictions.

Healthcare utilization and mortality outcomes in patients with pre-existing psychiatric disorders after intensive care unit discharge: A population-based retrospective cohort study

PURPOSE

Pre-existing psychiatric disorders may lead to negative outcomes following intensive care unit (ICU) discharge. We evaluated the association of pre-existing psychiatric disorders with subsequent healthcare utilization and mortality in patients discharged from ICU.

MATERIALS AND METHODS

We retrospectively studied adult patients admitted to 14 medical-surgical ICUs (January 2014-June 2016) with ICU length stay ≥24 h who survived to hospital discharge. Pre-existing psychiatric disorders were identified using algorithms for diagnostic codes captured ≤5 years before ICU admission. Outcomes were healthcare utilization (emergency department visit, hospital or ICU readmission) and mortality. We used logistic regression models with propensity scores to estimate associations, converted to risk ratios (RR).

RESULTS

We included 10,598 patients. 37.6% (n = 3982) had a psychiatric history. Patients with pre-existing psychiatric disorders were at higher risk of subsequent emergency department visits (RR 1.49, 95%CI 1.29-1.71), hospital readmission (RR 1.49, 95%CI 1.34-1.66), ICU readmission (RR 2.64, 95%CI 1.55-4.49) one-year post-ICU discharge, compared to patients without pre-existing psychiatric disorders. Patients with pre-existing psychiatric disorders had a higher risk of mortality (RR 1.31, 95%CI 1.00-1.71) six-months post-ICU discharge.

CONCLUSION

Critically ill patients with pre-existing psychiatric disorders have an increased risk of healthcare utilization and mortality outcomes following an ICU stay.

Prediction models for hospital readmissions in patients with heart disease: a systematic review and meta-analysis

OBJECTIVE

To describe the discrimination and calibration of clinical prediction models, identify characteristics that contribute to better predictions and investigate predictors that are associated with unplanned hospital readmissions.

DESIGN

Systematic review and meta-analysis.

DATA SOURCE

Medline, EMBASE, ICTPR (for study protocols) and Web of Science (for conference proceedings) were searched up to 25 August 2020.

ELIGIBILITY CRITERIA FOR SELECTING STUDIES

Studies were eligible if they reported on (1) hospitalised adult patients with acute heart disease; (2) a clinical presentation of prediction models with c-statistic; (3) unplanned hospital readmission within 6 months.

PRIMARY AND SECONDARY OUTCOME MEASURES

Model discrimination for unplanned hospital readmission within 6 months measured using concordance (c) statistics and model calibration. Meta-regression and subgroup analyses were performed to investigate predefined sources of heterogeneity. Outcome measures from models reported in multiple independent cohorts and similarly defined risk predictors were pooled.

RESULTS

Sixty studies describing 81 models were included: 43 models were newly developed, and 38 were externally validated. Included populations were mainly patients with heart failure (HF) (n=29). The average age ranged between 56.5 and 84 years. The incidence of readmission ranged from 3% to 43%. Risk of bias (RoB) was high in almost all studies. The c-statistic was <0.7 in 72 models, between 0.7 and 0.8 in 16 models and >0.8 in 5 models. The study population, data source and number of predictors were significant moderators for the discrimination. Calibration was reported for 27 models. Only the GRACE (Global Registration of Acute Coronary Events) score had adequate discrimination in independent cohorts (0.78, 95% CI 0.63 to 0.86). Eighteen predictors were pooled.

CONCLUSION

Some promising models require updating and validation before use in clinical practice. The lack of independent validation studies, high RoB and low consistency in measured predictors limit their applicability.

PROSPERO REGISTRATION NUMBER

CRD42020159839.

Published models that predict hospital readmission: a critical appraisal

INTRODUCTION

The number of readmission risk prediction models available has increased rapidly, and these models are used extensively for health decision-making. Unfortunately, readmission models can be subject to flaws in their development and validation, as well as limitations in their clinical usefulness.

OBJECTIVE

To critically appraise readmission models in the published literature using Delphi-based recommendations for their development and validation.

METHODS

We used the modified Delphi process to create Critical Appraisal of Models that Predict Readmission (CAMPR), which lists expert recommendations focused on development and validation of readmission models. Guided by CAMPR, two researchers independently appraised published readmission models in two recent systematic reviews and concurrently extracted data to generate reference lists of eligibility criteria and risk factors.

RESULTS

We found that published models (n=81) followed 6.8 recommendations (45%) on average. Many models had weaknesses in their development, including failure to internally validate (12%), failure to account for readmission at other institutions (93%), failure to account for missing data (68%), failure to discuss data preprocessing (67%) and failure to state the model's eligibility criteria (33%).

CONCLUSIONS

The high prevalence of weaknesses in model development identified in the published literature is concerning, as these weaknesses are known to compromise predictive validity. CAMPR may support researchers, clinicians and administrators to identify and prevent future weaknesses in model development.

Feature-based multi-criteria recommendation system using a weighted approach with ranking correlation

With the increase of online businesses, recommendation algorithms are being researched a lot to facilitate the process of using the existing information. Such multi-criteria recommendation (MCRS) helps a lot the end-users to attain the required results of interest having different selective criteria – such as combinations of implicit and explicit interest indicators in the form of ranking or rankings on different matched dimensions. Current approaches typically use label correlation, by assuming that the label correlations are shared by all objects. In real-world tasks, however, different sources of information have different features. Recommendation systems are more effective if being used for making a recommendation using multiple criteria of decisions by using the correlation between the features and items content (content-based approach) or finding a similar user rating to get targeted results (Collaborative filtering). To combine these two filterings in the multicriteria model, we proposed a features-based fb-knn multi-criteria hybrid recommendation algorithm approach for getting the recommendation of the items by using multicriteria features of items and integrating those with the correlated items found in similar datasets. Ranks were assigned to each decision and then weights were computed for each decision by using the standard deviation of items to get the nearest result. For evaluation, we tested the proposed algorithm on different datasets having multiple features of information. The results demonstrate that proposed fb-knn is efficient in different types of datasets.

Designing COVID-19 mortality predictions to advance clinical outcomes: Evidence from the Department of Veterans Affairs

Using administrative data on all Veterans who enter Department of Veterans Affairs (VA) medical centres throughout the USA, this paper uses artificial intelligence (AI) to predict mortality rates for patients with COVID-19 between March and August 2020. First, using comprehensive data on over 10 000 Veterans’ medical history, demographics and lab results, we estimate five AI models. Our XGBoost model performs the best, producing an area under the receive operator characteristics curve (AUROC) and area under the precision-recall curve of 0.87 and 0.41, respectively. We show how focusing on the performance of the AUROC alone can lead to unreliable models. Second, through a unique collaboration with the Washington D.C. VA medical centre, we develop a dashboard that incorporates these risk factors and the contributing sources of risk, which we deploy across local VA medical centres throughout the country. Our results provide a concrete example of how AI recommendations can be made explainable and practical for clinicians and their interactions with patients.

Effects of Neighborhood-level Data on Performance and Algorithmic Equity of a Model That Predicts 30-day Heart Failure Readmissions at an Urban Academic Medical Center

BACKGROUND

Socioeconomic data may improve predictions of clinical events. However, owing to structural racism, algorithms may not perform equitably across racial subgroups. Therefore, we sought to compare the predictive performance overall, and by racial subgroup, of commonly used predictor variables for heart failure readmission with and without the area deprivation index (ADI), a neighborhood-level socioeconomic measure.

METHODS AND RESULTS

We conducted a retrospective cohort study of 1316 Philadelphia residents discharged with a primary diagnosis of congestive heart failure from the University of Pennsylvania Health System between April 1, 2015, and March 31, 2017. We trained a regression model to predict the probability of a 30-day readmission using clinical and demographic variables. A second model also included the ADI as a predictor variable. We measured predictive performance with the Brier Score (BS) in a held-out test set. The baseline model had moderate performance overall (BS 0.13, 95% CI 0.13-0.14), and among White (BS 0.12, 95% CI 0.12-0.13) and non-White (BS 0.13, 95% CI 0.13-0.14) patients. Neither performance nor algorithmic equity were significantly changed with the addition of the ADI.

CONCLUSIONS

The inclusion of neighborhood-level data may not reliably improve performance or algorithmic equity.

Effect of a Real-Time Risk Score on 30-day Readmission Reduction in Singapore

OBJECTIVE

To develop a risk score for the real-time prediction of readmissions for patients using patient specific information captured in electronic medical records (EMR) in Singapore to enable the prospective identification of high-risk patients for enrolment in timely interventions.

METHODS

Machine-learning models were built to estimate the probability of a patient being readmitted within 30 days of discharge. EMR of 25,472 patients discharged from the medicine department at Ng Teng Fong General Hospital between January 2016 and December 2016 were extracted retrospectively for training and internal validation of the models. We developed and implemented a real-time 30-day readmission risk score generation in the EMR system, which enabled the flagging of high-risk patients to care providers in the hospital. Based on the daily high-risk patient list, the various interfaces and flow sheets in the EMR were configured according to the information needs of the various stakeholders such as the inpatient medical, nursing, case management, emergency department, and postdischarge care teams.

RESULTS

Overall, the machine-learning models achieved good performance with area under the receiver operating characteristic ranging from 0.77 to 0.81. The models were used to proactively identify and attend to patients who are at risk of readmission before an actual readmission occurs. This approach successfully reduced the 30-day readmission rate for patients admitted to the medicine department from 11.7% in 2017 to 10.1% in 2019 (p < 0.01) after risk adjustment.

CONCLUSION

Machine-learning models can be deployed in the EMR system to provide real-time forecasts for a more comprehensive outlook in the aspects of decision-making and care provision.

Leveraging machine learning to characterize the role of socio-economic determinants on physical health and well-being among veterans

INTRODUCTION

We investigate the contribution of demographic, socio-economic, and geographic characteristics as determinants of physical health and well-being to guide public health policies and preventative behavior interventions (e.g., countering coronavirus).

METHODS

We use machine learning to build predictive models of overall well-being and physical health among veterans as a function of these three sets of characteristics. We link Gallup's U.S. Daily Poll between 2014 and 2017 over a range of demographic and socio-economic characteristics with zipcode characteristics from the Census Bureau to build predictive models of overall and physical well-being.

RESULTS

Although the predictive models of overall well-being have weak performance, our classification of low levels of physical well-being performed better. Gradient boosting delivered the best results (80.2% precision, 82.4% recall, and 80.4% AUROC) with perceptions of purpose in the workplace and financial anxiety as the most predictive features. Our results suggest that additional measures of socio-economic characteristics are required to better predict physical well-being, particularly among vulnerable groups, like veterans.

CONCLUSION

Socio-economic characteristics explain large differences in physical and overall well-being. Effective predictive models that incorporate socio-economic data will provide opportunities to create real-time and personalized feedback to help individuals improve their quality of life.

Early Prediction of Unplanned 30-Day Hospital Readmission: Model Development and Retrospective Data Analysis

Background

Existing readmission reduction solutions tend to focus on complementing inpatient care with enhanced care transition and postdischarge interventions. These solutions are initiated near or after discharge, when clinicians’ impact on inpatient care is ending. Preventive intervention during hospitalization is an underexplored area that holds potential for reducing readmission risk. However, it is challenging to predict readmission risk at the early stage of hospitalization because few data are available.

Objective

The objective of this study was to build an early prediction model of unplanned 30-day hospital readmission using a large and diverse sample. We were also interested in identifying novel readmission risk factors and protective factors.

Methods

We extracted the medical records of 96,550 patients in 205 participating Cerner client hospitals across four US census regions in 2016 from the Health Facts database. The model was built with index admission data that can become available within 24 hours and data from previous encounters up to 1 year before the index admission. The candidate models were evaluated for performance, timeliness, and generalizability. Multivariate logistic regression analysis was used to identify readmission risk factors and protective factors.

Results

We developed six candidate readmission models with different machine learning algorithms. The best performing model of extreme gradient boosting (XGBoost) achieved an area under the receiver operating characteristic curve of 0.753 on the development data set and 0.742 on the validation data set. By multivariate logistic regression analysis, we identified 14 risk factors and 2 protective factors of readmission that have never been reported.

Conclusions

The performance of our model is better than that of the most widely used models in US health care settings. This model can help clinicians identify readmission risk at the early stage of hospitalization so that they can pay extra attention during the care process of high-risk patients. The 14 novel risk factors and 2 novel protective factors can aid understanding of the factors associated with readmission.

A novel method for clinical risk prediction with low-quality data

In real-world data, predictive models for clinical risks (such as adverse drug reactions, hospital readmission, and chronic disease onset) are constantly struggling with low-quality issues, namely redundant and highly correlated features, extreme category imbalances, and most importantly, a large number of missing values. In most existing work, each patient is represented as a value vector with the fixed-length from some feature space, and missing values are forced to be imputed, which introduces much noise for prediction if the data set is highly incomplete. Besides, other challenges are either remaining unresolved or only partially solved when modeling, but without a systematic approach. In this paper, we propose a novel framework to address these low-quality problems, that we first treat patients as bags with the various number of feature-value pairs, called instances, and map them to an embedding space through our proposed feature embedding method to learn from it directly. In this way, predictive models can avoid the negative impact of missing data naturally. A novel multi-instance neural network is then connected, using two computational modules to deal with the problems of correlated and redundant features: multi-head attention and attention-based multi-instance pooling. They are capable of capturing the instance correlations and locating valuable information in each instance or bag. The feature embedding and multi-instance neural network are parameterized and optimized jointly in an end-to-end manner. Moreover, the training process is under both main and auxiliary supervision with focal loss functions to avoid the caveat of a highly imbalanced label set. This proposed framework is named AMI-Net3. We evaluate it on three suitable data sets from real-world settings with different clinical risk prediction tasks: adverse drug reaction of risperidone, schizophrenia relapse, and invasive fungi infection, respectively. The comprehensive experimental results demonstrate the effectiveness and superiority of our proposed method over competitive baselines.

Hospital readmissions of patients with heart failure from real world: timing and associated risk factors

AIMS

This study aims to investigate hospital readmissions and timing, as well as risk factors in a real world heart failure (HF) population.

METHODS AND RESULTS

All patients discharged alive in 2016 from Sahlgrenska University Hospital/Östra, Gothenburg, Sweden, with a primary diagnosis of HF were consecutively included. Patient characteristics, type of HF, treatment, and follow-up were registered. Time to first all-cause or HF readmission, as well as number of 1 year readmissions from discharge were recorded. In total, 448 patients were included: 273 patients (mean age 78 ± 11.8 years) were readmitted for any cause within 1 year (readmission rate of 60.9%), and 175 patients (mean age 76.6 ± 13.7) were never readmitted. Among readmissions, 60.1% occurred during the first quarter after index hospitalization, giving a 3 month all-cause readmission rate of 36.6%. HF-related 1 year readmission rate was 38.4%. Patients who were readmitted had significantly more renal dysfunction (52.4% vs. 36.6%, P = 0.001), pulmonary disease (25.6% vs. 15.4%, P = 0.010), and psychiatric illness (24.9% vs. 12.0%, P = 0.001). Number of co-morbidities and readmissions were significantly associated (P < 0.001 for all cause readmission rate and P = 0.012 for 1 year HF readmission rate). Worsening HF constituted 63% of all-cause readmissions. Psychiatric disease was an independent risk factor for 1 month and 1 year all-cause readmissions. Poor compliance to medication was an independent risk factor for 1 month and 1 year HF readmission.

CONCLUSIONS

In our real world cohort of HF patients, frequent hospital readmissions occurred in the early post-discharge period and were mainly driven by worsening HF. Co-morbidity was one of the most important factors for readmission.

Insight Into the Posthospital Syndrome: A 3-Month Longitudinal Follow up on Geriatric Syndromes and Their Association With Functional Decline, Readmission, and Mortality

Background

Acute hospitalization may lead to posthospital syndrome, but no studies have investigated how this syndrome manifests and geriatric syndromes are often used as synonym. However, studies on longitudinal associations between syndromes and adverse outcomes are scarce. We aimed to analyze longitudinal associations between geriatric syndromes and functional decline (FD), readmission, and mortality.

Methods

Prospective cohort study, including 401 acutely hospitalized patients (aged ≥ 70). We performed: (i) logistic regression analyses to assess associations between patterns of geriatric syndromes as they develop over time (between admission and 1 month postdischarge), and FD and readmission; (ii) generalized estimating equations to assess longitudinal associations between geriatric syndromes over five time points (admission, discharge, 1, 2, and 3 months postdischarge) and FD, mortality, and readmission at 3 months postdischarge.

Results

After syndrome absent, syndrome present at both admission and 1 month postdischarge was most prevalent. Persistent patterns of apathy (odds ratio [OR] = 4.35, 95% confidence interval [CI] = 1.54–12.30), pain (OR = 3.26, 95% CI = 1.21–8.8), malnutrition (OR = 3.4, 95% CI = 1.35–8.56), mobility impairment (OR = 6.65, 95% CI = 1.98–22.38), and fear of falling (OR = 3.17, 95% CI = 1.25–8.02) were associated with FD. Developing cognitive impairment (OR = 6.40, 95% CI = 1.52–26.84), fatigue (OR = 4.71, 95% CI = 1.03–21.60), and fall risk (OR = 4.30, 95% CI = 1.21–16.57) postdischarge, was associated with readmission; however, only 4%–6% developed these syndromes. Over the course of five time points, mobility impairment, apathy, and incontinence were longitudinally associated with FD; apathy, malnutrition, fatigue, and fall risk with mortality; malnutrition with readmission.

Conclusion

Most geriatric syndromes are present at admission and patients are likely to retain them postdischarge. Several geriatric syndromes are longitudinally associated with mortality and, particularly, persistently present syndromes place persons are at risk of FD. Although few persons develop syndromes postdischarge, those developing cognitive impairment, fatigue, and fall risk were at increased readmission risk.

Dynamic evaluation of the comparative effectiveness of an integrated program for heart failure care

RATIONALE, AIMS AND OBJECTIVES

An integrated care program for heart failure (HF) was developed in the Basque Country in 2013. The objective of this research was to evaluate its effectiveness through the number of hospital admissions in three integrated healthcare organizations (IHOs), taking into account the longitudinal nature of the disease and the intensity of the implementation.

METHODS

A retrospective observational study was carried out, based on data entered in administrative and clinical databases between 2014 and 2018 for a total population of 230 000. In addition to conventional statistical analyses, Andersen-Gill models for recurrent events were used, incorporating dynamic variables that allowed assessment of the intervention's intensity before each hospitalization.

RESULTS

A total of 6768 patients were analysed. Age (hazard ratio [HR] = 1.016; 95% confidence interval [CI] 1.011-1.022), the Charlson index (HR = 1.067, 95% CI 1.047-1.087), and the number of previous hospitalizations (HR = 1.632, 95% CI 1.557-1.712) were risk factors for readmission. Differences between IHOs were also statistically significant. Greater intervention intensity was associated with a lower hospitalization rate (HR = 0.995, 95% CI 0.990-1.000). As indicated by the interaction between intervention intensity and IHO, differences between IHOs disappeared when intensity rose. No inequities in hospitalization were found as a function of deprivation index or sex. Nonetheless, inequity in the implementation of the program by sex was clear, women with HF receiving less intense intervention than men with the same level of comorbidity and age.

CONCLUSIONS

The extent of program implementation measured by intervention intensity is a main driver of the effectiveness of an educational and monitoring program for HF. The evaluation of HF program effectiveness on readmissions must take into account the entire natural history of the disease. Implementation intensity explains differences between IHOs.

Temporal Trends in Racial Differences in 30-Day Readmission and Mortality Rates After Acute Myocardial Infarction Among Medicare Beneficiaries

Importance

The association of the Hospital Readmission Reduction Program (HRRP) with reductions in racial disparities in 30-day outcomes for myocardial infarction (MI), is unknown, including whether this varies by HRRP hospital penalty status.

Objective

To assess temporal trends in 30-day readmission and mortality rates among black and nonblack patients discharged after hospitalization for acute MI at low-performing and high-performing hospitals, as defined by readmission penalty status after HRRP implementation.

Design, Setting, and Participants

This observational cohort analysis used data from the multicenter National Cardiovascular Data Registry Chest Pain-MI Registry centers that were subject to the first cycle of HRRP, between January 1, 2008, and November 30, 2016. All patients hospitalized with MI who were included in National Cardiovascular Data Registry Chest Pain-MI Registry were included in the analysis. Data were analyzed from April 2018 to September 2019.

Exposures

Hospital performance category and race (black compared with nonblack patients). Centers were classified as high performing or low performing based on the excess readmission ratio (predicted to expected 30-day risk adjusted readmission rate) for MI during the first HRRP cycle (in October 2012).

Main Outcomes and Measures

Thirty-day all-cause readmission and mortality rates.

Results

Among 753 hospitals that treated 155 397 patients with acute MI (of whom 11 280 [7.3%] were black), 399 hospitals (53.0%) were high performing. Thirty-day readmission rates declined over time in both black and nonblack patients (annualized 30-day readmission rate: 17.9% vs 20.8%). Black (compared with nonblack) race was associated with higher unadjusted odds of 30-day readmission in both low-performing and high-performing centers (odds ratios: before HRRP: low-performing hospitals, 1.14 [95% CI, 1.03-1.26]; P = .01; high-performing hospitals, 1.17 [95% CI, 1.04-1.32]; P = .01; after HRRP: low-performing hospitals, 1.23 [95% CI, 1.13-1.34]; P < .001; high-performing hospitals, 1.25 [95% CI, 1.12-1.39]; P < .001). However, these racial differences were not significant after adjustment for patient characteristics. The 30-day mortality rates declined significantly over time in nonblack patients, with stable (nonsignificant) temporal trends among black patients. Adjusted associations between race and 30-day mortality showed that 30-day mortality rates were significantly lower among black (compared with nonblack) patients in the low-performing hospitals (odds ratios: pre-HRRP, 0.79 [95% CI, 0.63-0.97]; P = .03; post-HRRP, 0.80 [95% CI, 0.68-0.95]; P = .01) but not in high-performing hospitals. Finally, the association between race and 30-day outcomes did not vary after the HRRP period began in either high-performing or low-performing hospitals.

Conclusions and Relevance

In this analysis, 30-day readmission rates among patients with MI declined over time for both black and nonblack patients. Differences in race-specific 30-day readmission rates persisted but appeared to be attributable to patient-level factors. The 30-day mortality rates have declined for nonblack patients and remained stable among black patients. Implementation of the HRRP was not associated with improvement or worsening of racial disparities in readmission and mortality rates.

Comparison of deep learning with traditional models to predict preventable acute care use and spending among heart failure patients

Recent health reforms have created incentives for cardiologists and accountable care organizations to participate in value-based care models for heart failure (HF). Accurate risk stratification of HF patients is critical to efficiently deploy interventions aimed at reducing preventable utilization. The goal of this paper was to compare deep learning approaches with traditional logistic regression (LR) to predict preventable utilization among HF patients. We conducted a prognostic study using data on 93,260 HF patients continuously enrolled for 2-years in a large U.S. commercial insurer to develop and validate prediction models for three outcomes of interest: preventable hospitalizations, preventable emergency department (ED) visits, and preventable costs. Patients were split into training, validation, and testing samples. Outcomes were modeled using traditional and enhanced LR and compared to gradient boosting model and deep learning models using sequential and non-sequential inputs. Evaluation metrics included precision (positive predictive value) at k, cost capture, and Area Under the Receiver operating characteristic (AUROC). Deep learning models consistently outperformed LR for all three outcomes with respect to the chosen evaluation metrics. Precision at 1% for preventable hospitalizations was 43% for deep learning compared to 30% for enhanced LR. Precision at 1% for preventable ED visits was 39% for deep learning compared to 33% for enhanced LR. For preventable cost, cost capture at 1% was 30% for sequential deep learning, compared to 18% for enhanced LR. The highest AUROCs for deep learning were 0.778, 0.681 and 0.727, respectively. These results offer a promising approach to identify patients for targeted interventions.

Impact of the Iranian Health Sector Evolution Plan on Rehospitalization: An Analysis of 158000 Hospitalizations

Background:

In May 2014, Iran launched the most far-reaching reform for the health sector, so-called Health Sector Evolution Plan (HSEP), since introduction of the primary health care network, with a systematic plan to bring about Universal Health Coverage. We aimed to analyze the time to first all-caused rehospitalization and all-caused 30-day readmission rate in the biggest referral hospital of Northwest of Iran before and after the reform.

Methods:

We retrospectively analyzed discharge data for all hospitalization occurred in the six-year period of 2011–2017. The primary endpoints were readmission-free survival, and overall 30-day readmission rate. Using multivariate cox proportional hazards regression and logistic regression, we assessed between-period differences for readmission-free survival time and overall 30-day rehospitalization, respectively.

Results:

Overall, 157969 admissions were included. After adjusting for available confounders including age; sex; ward of admission; length of stay; and admission in first/second half of year, the risk of being readmitted within 30 days after the reform was significantly higher (worse) compared to pre-reform hospitalization (odd ratio 1.22, P<0.001, 95% CI, 1.15–1.30). Adjusting for the same covariates, after-reform period also was slightly significantly associated with decreased (deteriorated) readmission-free time compared with pre-HSEP period (HR 1.06, P=0.005, 95% CI 1.01–1.11).

Conclusion:

HSEP seems insufficient to improve neither readmission rate, nor readmission-free time. It is advisable some complementary strategies to be incorporated in the HSEP, such as continuity of care promotion, self-care enhancement, effective information flow, and post-discharge follow up programs.

CREATE: A New Data Resource to Support Cardiac Precision Health

Background

The initiatives of precision medicine and learning health systems require databases with rich and accurately captured data on patient characteristics. We introduce the Clinical Registry, AdminisTrative Data and Electronic Medical Records (CREATE) database, which includes linked data from 4 population databases: Alberta Provincial Project for Outcome Assessment in Coronary Heart Disease (APPROACH; a national clinical registry), Sunrise Clinical Manager (SCM) electronic medical record (city-wide), the Discharge Abstract Database (DAD), and the National Ambulatory Care Reporting System (NACRS). The intent of this work is to introduce a cardiovascular-specific database for pursuing precision health activities using big data analytics.

Methods

We used deterministic data linkage to link SCM electronic medical record data to APPROACH clinical registry data using patient identifier variables. The APPROACH-SCM data set was subsequently linked to DAD and NACRS to obtain inpatient and outpatient cohort data. We further validated the quality of the linkage, where applicable, in these databases by comparing against the Alberta Health Insurance Care Plan registry database.

Results

We achieved 99.96% linkage across these 4 databases. Currently, there are 30,984 patients with 35,753 catheterizations in the CREATE database. The inpatient cohort contained 65.75% (20,373/30,984) of the patient sample, whereas the outpatient cohort contained 29.78% (9226/30,984). The infrastructure and the process to update and expand the database has been established.

Conclusions

CREATE is intended to serve as a database for supporting big data analytics activities surrounding cardiac precision health. The CREATE database will be managed by the Centre for Health Informatics at the University of Calgary, and housed in a secure high-performance computing environment.

A stacking-based model for predicting 30-day all-cause hospital readmissions of patients with acute myocardial infarction

BACKGROUND

Acute myocardial infarction (AMI) is a serious cardiovascular disease, followed by a high readmission rate within 30-days of discharge. Accurate prediction of AMI readmission is a crucial way to identify the high-risk group and optimize the distribution of medical resources.

METHODS

In this study, we propose a stacking-based model to predict the risk of 30-day unplanned all-cause hospital readmissions for AMI patients based on clinical data. Firstly, we conducted an under-sampling method of neighborhood cleaning rule (NCR) to alleviate the class imbalance and then utilized a feature selection method of SelectFromModel (SFM) to select effective features. Secondly, we adopted a self-adaptive approach to select base classifiers from eight candidate models according to their performances in datasets. Finally, we constructed a three-layer stacking model in which layer 1 and layer 2 were base-layer and level 3 was meta-layer. The predictions of the base-layer were used to train the meta-layer in order to make the final forecast.

RESULTS

The results show that the proposed model exhibits the highest AUC (0.720), which is higher than that of decision tree (0.681), support vector machine (0.707), random forest (0.701), extra trees (0.709), adaBoost (0.702), bootstrap aggregating (0.704), gradient boosting decision tree (0.710) and extreme gradient enhancement (0.713).

CONCLUSION

It is evident that our model could effectively predict the risk of 30-day all cause hospital readmissions for AMI patients and provide decision support for the administration.

Predictors of surgical readmission, unplanned hospitalization and emergency department use in head and neck oncology: A systematic review

OBJECTIVE

To identify predictors of unplanned hospitalization and emergency department (ED) use among head and neck oncology patients.

METHODS

Peer reviewed publications were identified through a systematic search of MEDLINE, Embase and Cochrane CENTRAL. Studies describing a cohort of HNC patients that detailed predictors of unplanned hospitalization or ED use in risk-adjusted models were eligible for inclusion. The methodologic quality of included studies was assessed using the Quality In Prognostic Studies (QUIPS) tool and an adapted version of the GRADE framework.

RESULTS

Of the 932 articles identified, 39 studies met our inclusion criteria with 31/39 describing predictors of surgical readmission and 10/39 describing predictors of ED use or unplanned hospitalization during radiation/chemoradiation treatment. Risk factors were classified into either 'patient-related', 'cancer severity' or 'process' factors. In the subset of studies looking at readmission following surgery wound complications (10/14 studies), presence of comorbidity (16/28 studies), low socioeconomic status (8/17 studies), cancer stage (9/14 studies), and prolonged hospital stay (7/18 studies) were the variables most frequently associated with readmission on multivariable analysis. Presence of comorbidity (6/10) and chemotherapy use (4/10) were more frequently associated with ED use and unplanned hospitalization.

CONCLUSIONS

Several consistent predictors have been identified across a variety of studies. This work is a critical first step towards the development of readmission and ED prediction models. It also enables meaningful comparison of hospital readmission rates with risk adjustment in HNC patients.

Predicting Postoperative Length of Stay for Isolated Coronary Artery Bypass Graft Patients Using Machine Learning

Purpose

Predictive analytics (PA) is a new trending approach in the field of healthcare that uses machine learning to build a prediction model using supervised learning algorithms. Isolated coronary artery bypass grafting (iCABG), an open-heart surgery, is commonly performed in the treatment of coronary heart disease.

Aim

The aim of this study was to develop and evaluate a model to predict postoperative length of stay (PLoS) for iCABG patients using supervised machine learning techniques, and to identify the features with the highest contribution to the model.

Methods

This is a retrospective study that uses historic data of adult patients who underwent isolated CABG (iCABG). After initial data pre-processing, data imputation using the kNN method was applied. The study used five prediction models using Naïve Bayes, Decision Tree, Random Forest, Logistic Regression and k Nearest Neighbor algorithms. Data imbalance was managed using the following widely used methods: oversampling, undersampling, "Both", and random over-sampling examples (ROSE). The features selection process was conducted using the Boruta method. Two techniques were applied to examine the performance of the models, (70%, 30%) split and cross-validation, respectively. Models were evaluated by comparing their performance using AUC and other metrics.

Results

In the final dataset, six distinct features and 621 instances were used to develop the models. A total of 20 models were developed using R statistical software. The model generated using Random Forest with "Both" resampling method and cross-validation technique was deemed the best fit (AUC=0.81; F1 score=0.82; and recall=0.82). Attributes found to be highly predictive of PLoS were pulmonary artery systolic, age, height, EuroScore II, intra-aortic balloon pump used, and complications during operation.

Conclusion

This study demonstrates the significance and effectiveness of building a model that predicts PLoS for iCABG patients using patient specifications and pre-/intra-operative measures.

Assessment of Value of Neighborhood Socioeconomic Status in Models That Use Electronic Health Record Data to Predict Health Care Use Rates and Mortality

Importance

Prediction models are widely used in health care as a way of risk stratifying populations for targeted intervention. Most risk stratification has been done using a small number of predictors from insurance claims. However, the utility of diverse nonclinical predictors, such as neighborhood socioeconomic contexts, remains unknown.

Objective

To assess the value of using neighborhood socioeconomic predictors in the context of 1-year risk prediction for mortality and 6 different health care use outcomes in a large integrated care system.

Design, Setting, and Participants

Diagnostic study using data from all adults age 18 years or older who had Kaiser Foundation Health Plan membership and/or use in the Kaiser Permantente Northern California: a multisite, integrated health care delivery system between January 1, 2013, and June 30, 2014. Data were recorded before the index date for each patient to predict their use and mortality in a 1-year post period using a test-train split for model training and evaluation. Analyses were conducted in fall of 2019.

Main Outcomes and Measures

One-year encounter counts (doctor office, virtual, emergency department, elective hospitalizations, and nonelective), total costs, and mortality.

Results

A total of 2 951 588 patients met inclusion criteria (mean [SD] age, 47.2 [17.4] years; 47.8% were female). The mean (SD) Neighborhood Deprivation Index was -0.32 (0.84). The areas under the receiver operator curve ranged from 0.71 for emergency department use (using the LASSO method and electronic health record predictors) to 0.94 for mortality (using the random forest method and electronic health record predictors). Neighborhood socioeconomic status predictors did not meaningfully increase the predictive performance of the models for any outcome.

Conclusions and Relevance

In this study, neighborhood socioeconomic predictors did not improve risk estimates compared with what is obtainable using standard claims data regardless of model used.