30-day Readmission Prediction Research Articles

A Global Data Approach to Prediction of 30-Day Readmission among Patients with Heart Failure and AKI

A Global Data Approach to Prediction of 30-Day Readmission among Patients with Heart Failure and AKI

Predictive Factors for 30-Day Readmission and Increased Healthcare Utilization in Sickle Cell Disease Patients: A Single-Center Comparative Retrospective Study.

SCD poses a significant healthcare burden. Understanding the factors contributing to high healthcare utilization and readmissions is crucial for improving the quality of care provided. This retrospective comparative observational study was conducted at King Saud University Medical City and included 160 SCD patients. A comparison was made between patients with no readmission and patients with at least one 30-day readmission. Another comparison was done between high healthcare-utilizing patients and low healthcare-utilizing patients. A regression model for 30-day readmission prediction was created. Readmission was significantly higher in patients using opioids, following up with pain clinics, and having a history of AVN (p= 0.002, p=0.028 and p=0.025 respectively). Higher healthcare utilization was associated with older age, smoking, use of opioids and GABA analogs, and psychiatric illnesses, including depression, substance use disorder, and anxiety. Predictors of 30-day readmission were hydroxyurea use (odds ratio, 2.819 [95% CI,1.082 to 7.34], p = 0.034), follow-up with pain management clinics (odds ratio, 2.248 [95% CI,1.547 to 3.266], p = 0.001), and SCD genotype (SS genotype) (odds ratio, 1.754 [95% CI,1.012 to 3.042], p = 0.045). Using the Paracetamol/Codeine combination significantly reduced the likelihood of readmission within 30 days of discharge. This study identified factors associated with 30-day readmission rate and high healthcare utilization among SCD patients. Strategies to reduce readmissions may include specialized SCD clinics, educational programs for patients, improved physician awareness of mental health screening, and further research on the impact of opioid use. Limitations include retrospective nature, single-center design, reliance on self-reported data, and exclusion of critically ill patients. However, despite the limitations, this study could lay a foundation for future projects aiming to optimize care and outcomes for patients living with SCD.

Machine learning-based 30-day readmission prediction models for patients with heart failure: a systematic review.

Heart failure (HF) is one of the most frequent diagnoses for 30-day readmission after hospital discharge. Nurses have a role in reducing unplanned readmission and providing quality of care during HF trajectories. This systematic review assessed the quality and significant factors of machine learning (ML)-based 30-day HF readmission prediction models. Eight academic and electronic databases were searched to identify all relevant articles published between 2013 and 2023. Thirteen studies met our inclusion criteria. The sample sizes of the selected studies ranged from 1778 to 272 778 patients, and the patients' average age ranged from 70 to 81 years. Quality appraisal was performed. The most commonly used ML approaches were random forest and extreme gradient boosting. The 30-day HF readmission rates ranged from 1.2 to 39.4%. The area under the receiver operating characteristic curve for models predicting 30-day HF readmission was between 0.51 and 0.93. Significant predictors included 60 variables with 9 categories (socio-demographics, vital signs, medical history, therapy, echocardiographic findings, prescribed medications, laboratory results, comorbidities, and hospital performance index). Future studies using ML algorithms should evaluate the predictive quality of the factors associated with 30-day HF readmission presented in this review, considering different healthcare systems and types of HF. More prospective cohort studies by combining structured and unstructured data are required to improve the quality of ML-based prediction model, which may help nurses and other healthcare professionals assess early and accurate 30-day HF readmission predictions and plan individualized care after hospital discharge. PROSPERO: CRD 42023455584.

Clinical outcome prediction using observational supervision with electronic health records and audit logs

Objective:Audit logs in electronic health record (EHR) systems capture interactions of providers with clinical data. We determine if machine learning (ML) models trained using audit logs in conjunction with clinical data (“observational supervision”) outperform ML models trained using clinical data alone in clinical outcome prediction tasks, and whether they are more robust to temporal distribution shifts in the data. Materials and Methods:Using clinical and audit log data from Stanford Healthcare, we trained and evaluated various ML models including logistic regression, support vector machine (SVM) classifiers, neural networks, random forests, and gradient boosted machines (GBMs) on clinical EHR data, with and without audit logs for two clinical outcome prediction tasks: major adverse kidney events within 120 days of ICU admission (MAKE-120) in acute kidney injury (AKI) patients and 30-day readmission in acute stroke patients. We further tested the best performing models using patient data acquired during different time-intervals to evaluate the impact of temporal distribution shifts on model performance. Results:Performance generally improved for all models when trained with clinical EHR data and audit log data compared with those trained with only clinical EHR data, with GBMs tending to have the overall best performance. GBMs trained with clinical EHR data and audit logs outperformed GBMs trained without audit logs in both clinical outcome prediction tasks: AUROC 0.88 (95% CI: 0.85–0.91) vs. 0.79 (95% CI: 0.77–0.81), respectively, for MAKE-120 prediction in AKI patients, and AUROC 0.74 (95% CI: 0.71–0.77) vs. 0.63 (95% CI: 0.62–0.64), respectively, for 30-day readmission prediction in acute stroke patients. The performance of GBM models trained using audit log and clinical data degraded less in later time-intervals than models trained using only clinical data. Conclusion:Observational supervision with audit logs improved the performance of ML models trained to predict important clinical outcomes in patients with AKI and acute stroke, and improved robustness to temporal distribution shifts.

Development of Multivariable Prediction Models for the Identification of Patients Admitted to Hospital with an Exacerbation of COPD and the Prediction of Risk of Readmission: A Retrospective Cohort Study using Electronic Medical Record Data

Background Approximately 20% of patients who are discharged from hospital for an acute exacerbation of COPD (AECOPD) are readmitted within 30 days. To reduce this, it is important both to identify all individuals admitted with AECOPD and to predict those who are at higher risk for readmission. Objectives To develop two clinical prediction models using data available in electronic medical records: 1) identifying patients admitted with AECOPD and 2) predicting 30-day readmission in patients discharged after AECOPD. Methods Two datasets were created using all admissions to General Internal Medicine from 2012 to 2018 at two hospitals: one cohort to identify AECOPD and a second cohort to predict 30-day readmissions. We fit and internally validated models with four algorithms. Results Of the 64,609 admissions, 3,620 (5.6%) were diagnosed with an AECOPD. Of those discharged, 518 (15.4%) had a readmission to hospital within 30 days. For identification of patients with a diagnosis of an AECOPD, the top-performing models were LASSO and a four-variable regression model that consisted of specific medications ordered within the first 72 hours of admission. For 30-day readmission prediction, a two-variable regression model was the top performing model consisting of number of COPD admissions in the previous year and the number of non-COPD admissions in the previous year. Conclusion We generated clinical prediction models to identify AECOPDs during hospitalization and to predict 30-day readmissions after an acute exacerbation from a dataset derived from available EMR data. Further work is needed to improve and externally validate these models.

Cardiac Troponin T and NT-proBNP for Prediction of 30-Day Readmission or Death in Patients with Acute Dyspnea: Data from the Akershus Cardiac Examination 2 Study

Introduction: N-terminal pro-B-type natriuretic peptide (NT-proBNP) and cardiac troponin T (cTnT) measurements are recommended in patients with acute dyspnea. We aimed to assess the prognostic merit of cTnT compared to NT-proBNP for 30-day readmission or death in patients hospitalized with acute dyspnea. Methods: We measured cTnT and NT-proBNP within 24 h in 314 patients hospitalized with acute dyspnea and adjudicated the cause of the index admission. Time to first event of readmission or death ≤30 days after hospital discharge was recorded, and cTnT and NT-proBNP measurements were compared head-to-head. Results: Patients who died (12/314) or were readmitted (71/314) within 30 days had higher cTnT concentrations (median: 32.6, Q1–Q3: 18.4–74.2 ng/L vs. median: 19.4, Q1–Q3: 8.4–36.1 ng/L; p for comparison <0.001) and NT-proBNP concentrations (median: 1,753.6, Q1–Q3: 464.2–6,862.0 ng/L vs. median 984, Q1–Q3 201–3,600 ng/L; for comparison p = 0.027) compared to patients who survived and were not readmitted. cTnT concentrations were associated with readmission or death within 30 days after discharge both in the total cohort (adjusted hazard ratio [aHR]: 1.64, 95% confidence interval [CI]: 1.30–2.05) and in patients with heart failure (HF) (aHR: 1.58, 95% CI: 1.14–2.18). In contrast, NT-proBNP concentrations were not associated with short-term events, neither in the total cohort (aHR: 1.10, 95% CI: 0.94–1.30) nor in patients with adjudicated HF (aHR: 1.06, 95% CI: 0.80–1.40). Conclusion: cTnT concentrations are associated with 30-day readmission or death in patients hospitalized with acute dyspnea, as well as in patients adjudicated HF.

Machine learning vs. conventional methods for prediction of 30-day readmission following percutaneous mitral edge-to-edge repair

BackgroundIdentifying predictors of readmissions after mitral valve transcatheter edge-to-edge repair (MV-TEER) is essential for risk stratification and optimization of clinical outcomes. AimsWe investigated the performance of machine learning [ML] algorithms vs. logistic regression in predicting readmissions after MV-TEER. MethodsWe utilized the National-Readmission-Database to identify patients who underwent MV-TEER between 2015 and 2018. The database was randomly split into training (70 %) and testing (30 %) sets. Lasso regression was used to remove non-informative variables and rank informative ones. The top 50 informative predictors were tested using 4 ML models: ML-logistic regression [LR], Naive Bayes [NB], random forest [RF], and artificial neural network [ANN]/For comparison, we used a traditional statistical method (principal component analysis logistic regression PCA-LR). ResultsA total of 9425 index hospitalizations for MV-TEER were included. Overall, the 30-day readmission rate was 14.6 %, and heart failure was the most common cause of readmission (32 %). The readmission cohort had a higher burden of comorbidities (median Elixhauser score 5 vs. 3) and frailty score (3.7 vs. 2.9), longer hospital stays (3 vs. 2 days), and higher rates of non-home discharges (17.4 % vs. 8.5 %). The traditional PCA-LR model yielded a modest predictive value (area under the curve [AUC] 0.615 [0.587–0.644]). Two ML algorithms demonstrated superior performance than the traditional PCA-LR model; ML-LR (AUC 0.692 [0.667–0.717]), and NB (AUC 0.724 [0.700–0.748]). RF (AUC 0.62 [0.592–0.677]) and ANN (0.65 [0.623–0.677]) had modest performance. ConclusionMachine learning algorithms may provide a useful tool for predicting readmissions after MV-TEER using administrative databases.

Modelling 30-day hospital readmission after discharge for COPD patients based on electronic health records

Chronic Obstructive Pulmonary Disease (COPD) is the third most common chronic disease in China with frequent exacerbations, resulting in increased hospitalization and readmission rate. COPD readmission within 30 days after discharge is an important indicator of care transitions, patient’s quality of life and disease management. Identifying risk factors and improving 30-day readmission prediction help inform appropriate interventions, reducing readmissions and financial burden. This study aimed to develop a 30-day readmission prediction model using decision tree by learning from the data extracted from the electronic health record of COPD patients in Macao. Health records data of COPD inpatients from Kiang Wu Hospital, Macao, from January 1, 2018, to December 31, 2019 were reviewed and analyzed. A total of 782 hospitalizations for AECOPD were enrolled, where the 30-day readmission rate was 26.5% (207). A balanced dataset was randomly generated, where male accounted for 69.1% and mean age was 80.73 years old. Age, length of stay, history of tobacco smoking, hemoglobin, systemic steroids use, antibiotics use and number of hospital admission due to COPD in last 12 months were found to be significant risk factors for 30-day readmission of CODP patients (P < 0.01). A data-driven decision tree-based modelling approach with Bayesian hyperparameter optimization was developed. The mean precision-recall and AUC value for the classifier were 73.85, 73.7 and 0.7506, showing a satisfying prediction performance. The number of hospital admission due to AECOPD in last 12 months, smoke status and patients’ age were the top factors for 30-day readmission in Macao population.

Identification of Factors Associated With 30-Day Readmissions After Posterior Lumbar Fusion Using Machine Learning and Traditional Models: A National Longitudinal Database Study.

Retrospective cohort study. To identify factors associated with readmissions after PLF using machine learning and logistic regression (LR) models. Readmissions following posterior lumbar fusion (PLF) place significant health and financial burden on the patient and overall healthcare system. The Optum Clinformatics® Data Mart database was used to identify patients who underwent posterior lumbar laminectomy, fusion, and instrumentation between 2004 and 2017. Four machine learning models and a multivariable LR model were used to assess factors most closely associated with 30-day readmission. These models were also evaluated in terms of ability to predict unplanned 30-day readmissions. The top performing model (Gradient Boosting Machine; GBM) was then compared to the validated LACE index in terms of potential cost savings associated with implementation of the model. A total of 18,981 patients were included, of which 3,080 (16.2%) were readmitted within 30 days of initial admission. Discharge status, prior admission, and geographic division were most influential for the LR model, while discharge status, length of stay, and prior admissions had greatest relevance for the GBM model. GBM outperformed LR in predicting unplanned 30-day readmission (mean AUC 0.865 vs. 0.850, P<0.0001). Use of GBM also achieved a projected 80% decrease in readmission-associated costs relative to those achieved by the LACE index model. Factors associated with readmission vary in terms of predictive influence based on standard logistic regression and machine learning models used, highlighting the complementary roles these models have in identifying relevant factors for prediction of 30-day readmissions. For posterior lumbar fusion procedures, Gradient Boosting Machine yielded greatest predictive ability and associated cost savings for readmission. 3.

Multimodal Data Matters: Language Model Pre-Training Over Structured and Unstructured Electronic Health Records.

As two important textual modalities in electronic health records (EHR), both structured data (clinical codes) and unstructured data (clinical narratives) have recently been increasingly applied to the healthcare domain. Most existing EHR-oriented studies, however, either focus on a particular modality or integrate data from different modalities in a straightforward manner, which usually treats structured and unstructured data as two independent sources of information about patient admission and ignore the intrinsic interactions between them. In fact, the two modalities are documented during the same encounter where structured data inform the documentation of unstructured data and vice versa. In this paper, we proposed a Medical Multimodal Pre-trained Language Model, named MedM-PLM, to learn enhanced EHR representations over structured and unstructured data and explore the interaction of two modalities. In MedM-PLM, two Transformer-based neural network components are firstly adopted to learn representative characteristics from each modality. A cross-modal module is then introduced to model their interactions. We pre-trained MedM-PLM on the MIMIC-III dataset and verified the effectiveness of the model on three downstream clinical tasks, i.e., medication recommendation, 30-day readmission prediction and ICD coding. Extensive experiments demonstrate the power of MedM-PLM compared with state-of-the-art methods. Further analyses and visualizations show the robustness of our model, which could potentially provide more comprehensive interpretations for clinical decision-making.

Prediction of 30-Day Readmission in Hospitalized Older Adults Using Comprehensive Geriatric Assessment and LACE Index and HOSPITAL Score.

(1) Background: Elders have higher rates of rehospitalization, especially those with functional decline. We aimed to investigate potential predictors of 30-day readmission risk by comprehensive geriatric assessment (CGA) in hospitalized patients aged 65 years or older and to examine the predictive ability of the LACE index and HOSPITAL score in older patients with a combination of malnutrition and physical dysfunction. (2) Methods: We included patients admitted to a geriatric ward in a tertiary hospital from July 2012 to August 2018. CGA components including cognitive, functional, nutritional, and social parameters were assessed at admission and recorded, as well as clinical information. The association factors with 30-day hospital readmission were analyzed by multivariate logistic regression analysis. The predictive ability of the LACE and HOSPITAL score was assessed using receiver operator characteristic curve analysis. (3) Results: During the study period, 1509 patients admitted to a ward were recorded. Of these patients, 233 (15.4%) were readmitted within 30 days. Those who were readmitted presented with higher comorbidity numbers and poorer performance of CGA, including gait ability, activities of daily living (ADL), and nutritional status. Multivariate regression analysis showed that male gender and moderately impaired gait ability were independently correlated with 30-day hospital readmissions, while other components such as functional impairment (as ADL) and nutritional status were not associated with 30-day rehospitalization. The receiver operating characteristics for the LACE index and HOSPITAL score showed that both predicting scores performed poorly at predicting 30-day hospital readmission (C-statistic = 0.59) and did not perform better in any of the subgroups. (4) Conclusions: Our study showed that only some components of CGA, mobile disability, and gender were independently associated with increased risk of readmission. However, the LACE index and HOSPITAL score had a poor discriminating ability for predicting 30-day hospitalization in all and subgroup patients. Further identifiers are required to better estimate the 30-day readmission rates in this patient population.

Updated Perspectives on the Role of Biomechanics in COPD: Considerations for the Clinician.

Patients with chronic obstructive pulmonary disease (COPD) demonstrate extra-pulmonary functional decline such as an increased prevalence of falls. Biomechanics offers insight into functional decline by examining mechanics of abnormal movement patterns. This review discusses biomechanics of functional outcomes, muscle mechanics, and breathing mechanics in patients with COPD as well as future directions and clinical perspectives. Patients with COPD demonstrate changes in their postural sway during quiet standing compared to controls, and these deficits are exacerbated when sensory information (eg, eyes closed) is manipulated. If standing balance is disrupted with a perturbation, patients with COPD are slower to return to baseline and their muscle activity is differential from controls. When walking, patients with COPD appear to adopt a gait pattern that may increase stability (eg, shorter and wider steps, decreased gait speed) in addition to altered gait variability. Biomechanical muscle mechanics (ie, tension, extensibility, elasticity, and irritability) alterations with COPD are not well documented, with relatively few articles investigating these properties. On the other hand, dyssynchronous motion of the abdomen and rib cage while breathing is well documented in patients with COPD. Newer biomechanical technologies have allowed for estimation of regional, compartmental, lung volumes during activity such as exercise, as well as respiratory muscle activation during breathing. Future directions of biomechanical analyses in COPD are trending toward wearable sensors, big data, and cloud computing. Each of these offers unique opportunities as well as challenges. Advanced analytics of sensor data can offer insight into the health of a system by quantifying complexity or fluctuations in patterns of movement, as healthy systems demonstrate flexibility and are thus adaptable to changing conditions. Biomechanics may offer clinical utility in prediction of 30-day readmissions, identifying disease severity, and patient monitoring. Biomechanics is complementary to other assessments, capturing what patients do, as well as their capability.

Nationwide hospital admission data statistics and disease-specific 30-day readmission prediction.

PurposeHospital readmission prediction uses historical patient visit data to train machine learning models to predict risk of patients being readmitted after the discharge. Data used to train models, such as patient demographics, disease types, localized distributions etc., play significant roles in the model performance. To date, many methods exist for hospital readmission prediction, but answers to some important questions still remain open. For example, how will demographics, such as gender, age, geographic, impact on readmission prediction? Do patients suffering from different diseases vary significantly in their readmission rates? What are the nationwide hospital admission data characteristics? and how do hospital speciality, ownership, and locations impact on their readmission rates? In this study, we carry systematic investigations to answer the above questions, and propose a predictive modeling framework to predict disease-specific 30-day hospital readmission.MethodsWe first implement statistics analysis by using National Readmission Databases (NRD) with over 15 million hospital visits. After that, we create features and disease-specific readmission datasets. An ensemble learning framework is proposed to conduct hospital readmission prediction and Friedman test and Nemenyi post-hoc test is used to validate our proposed method.ResultsUsing National Readmission Databases (NRD), with over 15 million hospital visits, as our testbed, we summarize nationwide patient admission data statistics, in related to demographic, disease types, and hospital factors. We use feature engineering to design 526 representative features to model each patient visit. Our studies found that readmission rates vary significantly from diseases to diseases. For six diseases studied in our research, their readmission rates vary from 1.832 (Pneumonia) to 8.761% (Diabetes). Using random sampling and voting approaches, our study shows that soft voting outperforms hard voting on majority results, especially for AUC and balanced accuracy which are the main measures for imbalanced data. Random under sampling using 1.1:1 for negative:positive ratio achieves the best performance for AUC, balanced accuracy, and F1-score.ConclusionThis paper carries out systematic studies to understand US nationwide hospital readmission data statistics, and further designs a machine learning framework for disease-specific 30-day hospital readmission prediction. Our study shows that hospital readmission rates vary significantly with respect to different disease types, gender, age groups, any other factors. Gradient boosting achieves the best performance for disease specific hospital readmission prediction.

Ability of the LACE Index to Predict 30-Day Readmissions in Patients with Acute Myocardial Infarction.

Aims: This study aimed to utilize the existing LACE index (length of stay, acuity of admission, comorbidity index and emergency room visit in the past six months) to predict the risk of 30-day readmission and to find the associated factors in patients with AMI. Methods: This was a retrospective study and LACE index scores were calculated for patients admitted with AMI between 2015 and 2019. Data were utilized from the hospital’s electronic medical record. Multivariate logistic regression was performed to find the association between covariates and 30-day readmission. The risk prediction ability of the LACE index for 30-day readmission was analyzed by receiver operating characteristic curves with the C statistic. Results: A total of 205 (5.7%) patients were readmitted within 30 days. The odds ratio of older age group (OR = 1.78, 95% CI: 1.54–2.05), admission via emergency ward (OR = 1.45; 95% CI: 1.42–1.54) and LACE score ≥10 (OR = 2.71; 95% CI: 1.03–4.37) were highly associated with 30-day readmissions and statistically significant. The receiver operating characteristic curve C statistic of the LACE index for AMI patients was 0.78 (95% CI: 0.75–0.80) and showed favorable discrimination in the prediction of 30-day readmission. Conclusion: The LACE index showed a good discrimination to predict the risk of 30-day readmission for hospitalized patients with AMI. Further study would be recommended to focus on additional factors that can be used to predict the risk of 30-day readmission; this should be considered to improve the model performance of the LACE index for other acute conditions by using the national-based administrative data.

Timing Is Everything: Outpatient Follow-Up Between One And Two Weeks Post-Admission For Heart Failure Is Associated With The Lowest Rate Of Readmission Before 30 Days.

Introduction Hospital readmission following hospitalization for heart failure (CHF) is common and challenging to prevent. Get With The Guidelines performance measures include follow-up (F/U) within 7 days, which is logistically challenging. Optimized timing of F/U may both reduce rehospitalizations and burden on providers/systems. Hypothesis There is no difference in 30-day hospital readmission rates between patients who F/U within 1-7 days vs. F/U within 8-14 days. Both intervals will have lower readmission rates than those who had no F/U within the first 30 days. Methods We retrospectively reviewed a cohort of patients discharged alive after admission for CHF between January 1, 2010 and March 13, 2021 and who had available F/U data. We examined the association between all-cause 30-day readmission type (cardiology vs. general) and timing of F/U appointment [0-7 (1W); 8-14 (2W); 15-30 days (1MO); 30-day or longer F/U (LATE)]. We used generalized estimating equations survival-time analysis models to account for clustering. We performed a subgroup analysis of admissions since 2015 controlling for risk of 30-day readmission using a prediction score previously developed using machine learning of demographic and clinical data. Results Among 15,766 admissions of 10,523 patients, the 30-day readmission rates were 19.2%, 18.8%, 20.7%, and 23.7% for 1W, 2W, 1MO and LATE respectively, with an overall rate of 21.1%. There was no difference in odds of readmission between 1W and 2W (OR 0.96; 95%CI 0.85-1.09; p=0.530). F/U at either 1W or 2W had lower odds of readmission compared to LATE (OR 0.79; 95%CI 0.73-0.87; p<0.001; Figure 1) and there was a nonsignificant trend toward lower odds of readmission among those with cardiologist F/U as compared to general medicine (OR 0.91; 95%CI 0.82-1.00; P=0.053). In the subgroup analysis comprising 5,853 admissions with 30-day readmission prediction score, the odds of readmission were lower for 2W compared to 1W (OR 0.81; 95%CI 0.66-0.99; p= 0.048). F/U at either 1W or 2W had lower odds of readmission compared to LATE (OR 0.84; 95%CI 0.73-0.98; p=0.026). Conclusions After discharge for CHF hospitalization, outpatient follow-up between 8 and 14 days may be the optimal timeframe for reducing the rate of readmission at 30 days within at least one health system. Hospital readmission following hospitalization for heart failure (CHF) is common and challenging to prevent. Get With The Guidelines performance measures include follow-up (F/U) within 7 days, which is logistically challenging. Optimized timing of F/U may both reduce rehospitalizations and burden on providers/systems. There is no difference in 30-day hospital readmission rates between patients who F/U within 1-7 days vs. F/U within 8-14 days. Both intervals will have lower readmission rates than those who had no F/U within the first 30 days. We retrospectively reviewed a cohort of patients discharged alive after admission for CHF between January 1, 2010 and March 13, 2021 and who had available F/U data. We examined the association between all-cause 30-day readmission type (cardiology vs. general) and timing of F/U appointment [0-7 (1W); 8-14 (2W); 15-30 days (1MO); 30-day or longer F/U (LATE)]. We used generalized estimating equations survival-time analysis models to account for clustering. We performed a subgroup analysis of admissions since 2015 controlling for risk of 30-day readmission using a prediction score previously developed using machine learning of demographic and clinical data. Among 15,766 admissions of 10,523 patients, the 30-day readmission rates were 19.2%, 18.8%, 20.7%, and 23.7% for 1W, 2W, 1MO and LATE respectively, with an overall rate of 21.1%. There was no difference in odds of readmission between 1W and 2W (OR 0.96; 95%CI 0.85-1.09; p=0.530). F/U at either 1W or 2W had lower odds of readmission compared to LATE (OR 0.79; 95%CI 0.73-0.87; p<0.001; Figure 1) and there was a nonsignificant trend toward lower odds of readmission among those with cardiologist F/U as compared to general medicine (OR 0.91; 95%CI 0.82-1.00; P=0.053). In the subgroup analysis comprising 5,853 admissions with 30-day readmission prediction score, the odds of readmission were lower for 2W compared to 1W (OR 0.81; 95%CI 0.66-0.99; p= 0.048). F/U at either 1W or 2W had lower odds of readmission compared to LATE (OR 0.84; 95%CI 0.73-0.98; p=0.026). After discharge for CHF hospitalization, outpatient follow-up between 8 and 14 days may be the optimal timeframe for reducing the rate of readmission at 30 days within at least one health system.

Development and validation of an interpretable machine learning scoring tool for estimating time to emergency readmissions.

SummaryBackgroundEmergency readmission poses an additional burden on both patients and healthcare systems. Risk stratification is the first step of transitional care interventions targeted at reducing readmission. To accurately predict the short- and intermediate-term risks of readmission and provide information for further temporal risk stratification, we developed and validated an interpretable machine learning risk scoring system.MethodsIn this retrospective study, all emergency admission episodes from January 1st 2009 to December 31st 2016 at a tertiary hospital in Singapore were assessed. The primary outcome was time to emergency readmission within 90 days post discharge. The Score for Emergency ReAdmission Prediction (SERAP) tool was derived via an interpretable machine learning-based system for time-to-event outcomes. SERAP is six-variable survival score, and takes the number of emergency admissions last year, age, history of malignancy, history of renal diseases, serum creatinine level, and serum albumin level during index admission into consideration.FindingsA total of 293,589 ED admission episodes were finally included in the whole cohort. Among them, 203,748 episodes were included in the training cohort, 50,937 episodes in the validation cohort, and 38,904 in the testing cohort. Readmission within 90 days was documented in 80,213 (27.3%) episodes, with a median time to emergency readmission of 22 days (Interquartile range: 8-47). For different time points, the readmission rates observed in the whole cohort were 6.7% at 7 days, 10.6% at 14 days, 13.6% at 21 days, 16.4% at 30 days, and 23.0% at 60 days. In the testing cohort, the SERAP achieved an integrated area under the curve of 0.737 (95% confidence interval: 0.730-0.743). For a specific 30-day readmission prediction, SERAP outperformed the LACE index (Length of stay, Acuity of admission, Charlson comorbidity index, and Emergency department visits in past six months) and the HOSPITAL score (Hemoglobin at discharge, discharge from an Oncology service, Sodium level at discharge, Procedure during the index admission, Index Type of admission, number of Admissions during the last 12 months, and Length of stay). Besides 30-day readmission, SERAP can predict readmission rates at any time point during the 90-day period.InterpretationBetter performance in risk prediction was achieved by the SERAP than other existing scores, and accurate information about time to emergency readmission was generated for further temporal risk stratification and clinical decision-making. In the future, external validation studies are needed to evaluate the SERAP at different settings and assess their real-world performance.FundingThis study was supported by the Singapore National Medical Research Council under the PULSES Center Grant, and Duke-NUS Medical School.

Abstract 10834: Novel Biomarkers and Machine Learning Models for the Prediction of 30-Day Readmission Following Pediatric Congenital Heart Surgery

Introduction: Hospital readmissions after congenital heart surgery (CHS) are common in pediatric patients and are linked to higher risks of complications, mortality, and increased treatment costs. Accurate prediction of 30-day readmission after CHS could help clinicians identify high risk patients and enable tailored care. Hypothesis: Novel biomarkers alone enhance prediction of readmission risk using XGBoost model and augmenting clinical features to the Society of Thoracic Surgeons (STS)-risk model improves predictions of 30-day readmissions. Methods: The prospective cohort included children of 18 years or younger who underwent congenital heart surgery at Johns Hopkins Hospital. Of the 162 patients, 159 survived discharge and were therefore included in the study. We used pre- and post-operative biomarkers ST2, Galectin-3, NT-proBNP and GFAP as features for the biomarker model, covariates from the STS CHS Database Mortality Risk Model for the STS-risk model and augmenting clinical features to STS-risk variables for clinical model. We implemented the XGBoost model and compared the biomarker model and clinical model performance to the STS-risk model. Results: Readmissions within 30 days of surgery occurred among 9% of pediatric patients. Using XGBoost, the STS-risk model derived an AUROC of 0.978 (95% CI: 0.955-1.0). The biomarker model performed better than the STS-risk model with AUROC of 0.982 (95% CI: 0.959-1.0, p=0.8). The augmented clinical model outperformed the STS-risk model with an AUROC of 0.997 (95% CI: 0.99-1, p=0.1). Conclusions: Novel biomarkers alone perform better compared to the STS- risk model in predicting 30-day readmission risk. XGBoost model improves risk prediction when adding additional clinical features to the existing STS-risk model which outperformed the STS-risk model. Depending on healthcare systems clinical resources and data availability, an optimized XGBoost model can improve the readmission risk prediction outcome.

Identification of elderly patients at risk for 30-day readmission: Clinical insight beyond big data prediction.

This study explores the potential benefit of combining clinicians' risk assessments and the automated 30-day readmission prediction model. Automated readmission prediction models based on electronic health records are increasingly applied as part of prevention efforts, but their accuracy is moderate. This prospective multisource study was based on self-reported surveys of clinicians and data from electronic health records. The survey was performed at 15 internal medicine wards of three general Clalit hospitals between May 2016 and June 2017. We examined the degree of concordance between the Preadmission Readmission Detection Model, clinicians' readmission risk classification and the likelihood of actual readmission. Decision trees were developed to classify patients by readmission risk. A total of 694 surveys were collected for 371 patients. The disagreement between clinicians' risk assessment and the model was 34.5% for nurses and 33.5% for physicians. The decision tree algorithms identified 22% and 9% (based on nurses and physicians, respectively) of the model's low-medium-risk patients as high risk (accuracy 0.8 and 0.76, respectively). Combining the Readmission Model with clinical insight improves the ability to identify high-risk elderly patients. This study provides algorithms for the decision-making process for selecting high-risk readmission patients based on nurses' evaluations.

Nomogram for the Prediction of 30-Day Readmission in Acute Pancreatitis.

A prediction model for 30-day readmission in patients with acute pancreatitis (AP) was needed. To develop a nomogram to predict 30-day readmission in patients with AP and validate the usefulness of serum indicators after discharge for the prediction of 30-day readmission. This was a retrospective cohort study enrolling patients with the first attack of AP. Baseline characteristics, clinical profiles, and serum indicators after discharge were compared. Multivariate logistic regression analysis and a nomogram were employed to determine the independent risk factors for 30-day readmission. A total of 7.32% (121/1653) of the patients were readmitted within 30days after discharge. Different etiologies (biliary pancreatitis (adjusted odds ratio (AdjOR), 9.63; 95% confidence interval (CI), 1.28-72.52; P = 0.028), other causes (AdjOR, 9.37; 95% CI, 1.15-76.12, P = 0.026), mixed causes (AdjOR, 10.76; 95% CI, 1.27-91.35; P = 0.03) compared with alcoholic pancreatitis)), infected pancreatitis necrosis (IPN) (AdjOR, 2.3; 95% CI, 1.2-4.42; P = 0.013), total bilirubin level ≥ 20.5µmol/L (AdjOR, 2.42; 95% CI, 1.23-4.77; P = 0.01), glucose level ≥ 6.1mmol/L (AdjOR, 1.93; 95% CI, 1.16-3.19; P = 0.011), and albumin level < 40g/L (AdjOR, 4.25; 95% CI, 2.44-7.41; P < 0.001) were independently associated with 30-day readmission. A nomogram incorporating these factors demonstrated good discrimination, calibration, and clinical utility. Serum indicators after discharge added predictive value compared with clinical variables alone (AUC, 0.78 vs. 0.685; P = 0.0001). The nomogram combining etiology, IPN, and serum indicators after discharge has favorable predictive performance for 30-Day readmission. The close monitoring and reexamination of serum indicators are essential for AP patients at high risk.

Effects of routinely collected health information system variables on the readmission of patients with type 2 diabetes

ObjectivesThis research explores the association between variables routinely collected in a health information system and the readmission of patients with type 2 diabetes within 30 days of discharge. MethodsThis retrospective cohort study was conducted at King Fahd Hospital of the University (KFHU) in Al-Khobar, KSA. The study population comprised patients with type 2 diabetes who were admitted to the hospital from January 2016 to November 2016. Data were obtained from the hospital's information system at KFHU. The association between the readmission of patients with type 2 diabetes and routinely collected health information system variables such as demographics, type of diabetes, length of stay, and discharge type were analyzed. ResultsA total of 497 cases met the inclusion criteria. Of these, 31 (6.2%) cases were readmitted within 30 days. Type 2 diabetes was the only variable found to be significantly associated with readmission within 30 days (χ2 (1, N = 497) = 6.116, p = 0.0134). Diabetes type (p = 0.0133) and discharge type (p = 0.0403) were the only variables that displayed significance utilizing a logistic regression model. ConclusionOverall, the routinely collected demographic, diagnostic, and administrative variables were found to be poor predictors of 30-day readmission for type 2 diabetes at the institution studied. Nonetheless, the only significant variables in the prediction of 30-day readmission were diabetes type and discharge type. To determine the predictors of readmission, it is recommended that future studies include height and weight to the routinely collected health information system variables. We also suggest that future studies be based on data collected over several years or on pooled data collected from several hospitals.