Current depressive symptoms but not history of depression predict hospital readmission or death after discharge from medical wards: a multisite prospective cohort study

Predictive Modeling of Hospital Readmission: Challenges and Solutions

Hospital readmission prediction is a study to learn models from historical medical data to predict probability of a patient returning to hospital in a certain period, e.g. 30 or 90 days, after the discharge. The motivation is to help health providers deliver better treatment and post-discharge strategies, lower the hospital readmission rate, and eventually reduce the medical costs. Due to inherent complexity of diseases and healthcare ecosystems, modeling hospital readmission is facing many challenges. By now, a variety of methods have been developed, but existing literature fails to deliver a complete picture to answer some fundamental questions, such as what are the main challenges and solutions in modeling hospital readmission; what are typical features/models used for readmission prediction; how to achieve meaningful and transparent predictions for decision making; and what are possible conflicts when deploying predictive approaches for real-world usages. In this paper, we systematically review computational models for hospital readmission prediction, and propose a taxonomy of challenges featuring four main categories: (1) data variety and complexity; (2) data imbalance, locality and privacy; (3) model interpretability; and (4) model implementation. The review summarizes methods in each category, and highlights technical solutions proposed to address the challenges. In addition, a review of datasets and resources available for hospital readmission modeling also provides firsthand materials to support researchers and practitioners to design new approaches for effective and efficient hospital readmission prediction.

Depressive symptoms and malnutrition are associated with other geriatric syndromes and increase risk for 30-Day readmission in hospitalized older adults: a prospective cohort study

BACKGROUND

Readmission in older adults is typically complex with multiple contributing factors. We aim to examine how two prevalent and potentially modifiable geriatric conditions - depressive symptoms and malnutrition - relate to other geriatric syndromes and 30-day readmission in hospitalized older adults.

METHODS

Consecutive admissions of patients ≥ 65 years to a general medical department were recruited over 16 months. Patients were screened for depression, malnutrition, delirium, cognitive impairment, and frailty at admission. Medical records were reviewed for poor oral intake and functional decline during hospitalization. Unplanned readmission within 30-days of discharge was tracked through the hospital's electronic health records and follow-up telephone interviews. We use directed acyclic graphs (DAGs) to depict the relationship of depressive symptoms and malnutrition with geriatric syndromes that constitute covariates of interest and 30-day readmission outcome. Multiple logistic regression was performed for the independent associations of depressive symptoms and malnutrition with 30-day readmission, adjusting for variables based on DAG-identified minimal adjustment set.

RESULTS

We recruited 1619 consecutive admissions, with mean age 76.4 (7.9) years and 51.3% females. 30-day readmission occurred in 331 (22.0%) of 1,507 patients with follow-up data. Depressive symptoms, malnutrition, higher comorbidity burden, hospitalization in the one-year preceding index admission, frailty, delirium, as well as functional decline and poor oral intake during the index admission, were more commonly observed among patients who were readmitted within 30 days of discharge (P < 0.05). Patients with active depressive symptoms were significantly more likely to be frail (OR = 1.62, 95% CI 1.22-2.16), had poor oral intake (OR = 1.35, 95% CI 1.02-1.79) and functional decline during admission (OR = 1.58, 95% CI 1.11-2.23). Malnutrition at admission was significantly associated with frailty (OR = 1.53, 95% CI 1.07-2.19), delirium (OR = 2.33, 95% CI 1.60-3.39) cognitive impairment (OR = 1.88, 95% CI 1.39-2.54) and poor oral intake during hospitalization (OR = 2.70, 95% CI 2.01-3.64). In minimal adjustment set identified by DAG, depressive symptoms (OR = 1.38, 95% CI 1.02-1.86) remained significantly associated with 30-day readmission. The association of malnutrition with 30-day readmission was no longer statistically significant after adjusting for age, ethnicity and depressive symptoms in the minimal adjustment set (OR = 1.40, 95% CI 0.99-1.98).

CONCLUSION

The observed causal associations support screening and targeted interventions for depressive symptoms and malnutrition during admission and in the post-acute period.

Intelligible Models for HealthCare: Predicting the Probability of 6-Month Unfavorable Outcome in Patients with Ischemic Stroke

Early prediction of unfavorable outcome after ischemic stroke is significant for clinical management. Machine learning as a novel computational modeling technique could help clinicians to address the challenge. We aim to investigate the applicability of machine learning models for individualized prediction in ischemic stroke patients and demonstrate the utility of various model-agnostic explanation techniques for machine learning predictions. A total of 499 consecutive patients with Unfavorable [modified Rankin Scale (mRS) score 3-6, n = 140] and favorable (mRS score 0-2, n = 359) outcome after 6-month from ischemic stroke were enrolled in this study. Four machine learning models, including Random Forest [RF], eXtreme Gradient Boosting [XGBoost], Adaptive Boosting [Adaboost] and Support Vector Machine [SVM] were performed with the area-under-the-curve (AUC): (90.20 ± 0.22)%, (86.91 ± 1.05)%, (86.49 ± 2.35)%, (81.89 ± 2.40)%, respectively. Three global interpretability techniques (Feature Importance shows the contribution of selected features, Partial Dependence Plot aims to visualize the average effect of a feature on the predicted probability of unfavorable outcome, Feature Interaction detects the change in the prediction that occurs by varying the features after considering the individual feature effects) and one local interpretability technique (Shapley Value indicates the probability of unfavorable outcome of different instances) have been applied to present the interpretability techniques via visualization. Thereby, the current study is important for better understanding intelligible healthcare analytics via explanations for the prediction of local and global levels, and potentially reduction of the mortality of patients with ischemic stroke by assisting clinicians in the decision-making process.

Reducing Readmission of Hospitalized Patients With Depressive Symptoms: A Randomized Trial

PURPOSE

To determine if hospitalized patients with depressive symptoms will benefit from post-discharge depression treatment with care transition support.

METHODS

This is a randomized controlled trial of hospitalized patients with patient health questionnaire-9 score of 10 or more. We delivered the Re-Engineered Discharge (RED) and randomized participants to groups receiving RED-only or RED for Depression (RED-D), a 12-week post-discharge telehealth intervention including cognitive behavioral therapy, self-management support, and patient navigation. Primary outcomes were hospital readmission and reutilization rates at 30 and 90 days post discharge.

RESULTS

We randomized 709 participants (353 RED-D, 356 RED-only). At 90 days, 265 (75%) intervention participants had received at least 1 RED-D session (median 4). At 30 days, the intention-to-treat analysis showed no differences between RED-D vs RED-only in hospital readmission (9% vs 10%, incidence rate ratio [IRR] 0.92 [95% CI, 0.56-1.52]) or reutilization (27% vs 24%, IRR 1.14 [95% CI, 0.85-1.54]). The intention-to-treat analysis also showed no differences at 90 days in readmission (28% vs 21%, IRR 1.30 [95% CI, 0.95-1.78]) or reutilization (70% vs 57%, IRR 1.22 [95% CI, 1.01-1.49]). In the as-treated analysis, each additional RED-D session was associated with a decrease in 30- and 90-day readmissions. At 30 days, among 104 participants receiving 3 or more sessions, there were fewer readmissions (3% vs 10%, IRR 0.30 [95% CI, 0.07-0.84]) compared with the control group. At 90 days, among 109 participants receiving 6 or more sessions, there were fewer readmissions (11% vs 21%, IRR 0.52 [95% CI, 0.27-0.92]). Intention-to-treat analysis showed no differences between study groups on secondary outcomes.

CONCLUSIONS

Care transition support and post-discharge depression treatment can reduce unplanned hospital use with sufficient uptake of the RED-D intervention.

Comorbidity of depression and anxiety leads to a poor prognosis following angina pectoris patients: a prospective study

BACKGROUND

Depression and anxiety are two common mood problems among patients with cardiovascular disease (CVD) and are associated with poor cardiac prognoses. The comorbidity of depression and anxiety is considered to be a more severe psychological status than non-comorbid mood disorders. However, little is known about the relationship between depression or anxiety and noncardiac readmission. We conducted a prospective study on the prognostic impact of depression, anxiety, and the comorbidity of the two among angina pectoris (AP) patients.

METHOD

In this prospective study, 443 patients with AP were included in the analysis. Follow-up assessments were performed 1 year, and 2 years after patient discharges. Clinical outcomes of interest included noncardiac readmission, major adverse cardiovascular events (MACEs), and composite events. Depression and anxiety symptom scores derived from the patient health questionnaire-9 (PHQ-9) and generalised anxiety disorder-7 (GAD-7) questionnaire were used to assess mood symptoms at baseline. Participants with symptom scores of ≥10 on both the depression and anxiety questionnaires formed the clinical comorbidity subgroup. We used multivariable Cox proportional hazards models to evaluate the impact of individual mood symptom and comorbidity on clinical outcomes.

RESULTS

Among all the AP patients, 172 (38. 9%) were determined to have depression symptoms, 127 (28.7%) patients had anxiety symptoms and 71 (16.0%) patients suffered from their comorbidity. After controlling covariates, we found that patients who endured clinical depression (hazard ratio [HR] = 2.38, 95% confidence interval [CI] 1.06-5.33, p = 0.035) and anxiety ([HR] 2.85, 95% [CI] 1.10-7.45, p = 0.032) had a high risk of noncardiac readmission. Compared to participants with no mood symptoms, those with clinical comorbidity of depression and anxiety presented a greater risk of noncardiac readmission ([HR] 2.91, 95% [CI] 1.03-8.18, p = 0.043) MACEs ([HR] 2.38, 95% [CI] 1.11-5.10, p = 0.025) and composite event ([HR] 2.52, 95% [CI] 1.35-4.69, p = 0.004).

CONCLUSION

Depression and anxiety were found to have predictive value for noncardiac readmission among patients with AP. Furthermore, prognoses were found to be worse for patients with comorbidity of depression and anxiety than those with single mood symptom. Additional attention needs to be focused on the initial identification and long-term monitoring of mood symptom comorbidity.

Association between depression and increased risk of readmission in patients with heart failure: a systematic review and meta-analysis

INTRODUCTION

Heart failure (HF) is one of the world leading causes of admission and readmission. Recent studies have shown that the presence of depression is associated with hospital readmission in patients after an index admission for heart failure (HF). However, there is disagreement between published studies regarding this finding. We performed a systematic review and meta-analysis to evaluate the effect of depression on readmission rates in HF patients.

EVIDENCE ACQUISITION

We searched the databases of MEDLINE and EMBASE from inception to March 2020. Included studies were published study evaluating readmission rate of HF patients, with and without depression. Data from each study were combined using a random-effects model, generic inverse variance method of DerSimonian and Laird to calculate risk ratios and 95% confidence intervals.

EVIDENCE SYNTHESIS

Ten studies were included in the meta-analysis with a total of 53,165 patients (6194 patients with depression). The presence of depression was associated with an increased risk of readmission in patients with HF (pooled HR=1.54, 95% CI: 1.22-1.94, P<0.001, I²=55.4%). In a subgroup analysis, depression was associated with an increased risk of readmission in patients with HF in both short-term (≤90 days) follow-up (pooled HR=1.75, 95% CI: 1.07-2.85, P=0.025, I²=76.0%) and long-term (>90 days) follow-up (pooled HR=1.58, 95% CI: 1.32-1.90, P<0.001, I²=0.0%).

CONCLUSIONS

Our meta-analysis demonstrated that depression is associated with an increased risk of hospital readmission in patients with HF.

Post-discharge adjustment of medication in geriatric patients : A prospective cohort study

BACKGROUND

Little is known to what extent general practitioners (GP) change hospital discharge medications in older patients.

OBJECTIVE

This prospective cohort study aimed to analyze medication changes at the interface between hospital and community in terms of quality, quantity and type of drugs.

METHODS

A total of 121 out of 248 consecutively enrolled patients admitted to an acute geriatric hospital unit participated in the study. Medication regimens were recorded at admission and discharge and 4 weeks after hospital discharge the general practitioners in charge were contacted to provide the current medication charts. Changes in the extent of polypharmacy, in the type of drugs using anatomical therapeutic chemical classification (ATC) codes and potentially inappropriate medications (PIM) were analyzed.

RESULTS

Medication charts could be obtained for 98 participants in primary care. Only 21% of these patients remained on the original discharge medication. Overall, the average number of medications rose from hospital admission (6.58 SD ± 3.45) to discharge (6.96 SD ± 3.49) and again post-discharge in general practice (7.22 SD ± 3.68). The rates of patients on excessive polypharmacy (≥10 drugs) and on PIM were only temporarily reduced during hospital stay. The GPs stopped anti-infective drugs (ATC-J) and prescribed more antirheumatic drugs (ATC-M). Although no significant net changes occurred in other ATC groups, a substantial number of drugs were interchanged regarding the subgroups.

CONCLUSION

The study found that GPs extensively adjusted geriatric discharge medications. Whereas some changes may be necessary due to alterations in patients' state of health, a thorough communication between hospital doctors and GPs may level off different prescribing cultures and contribute to consistency in medication across sectors.

On the interpretability of machine learning-based model for predicting hypertension

BACKGROUND

Although complex machine learning models are commonly outperforming the traditional simple interpretable models, clinicians find it hard to understand and trust these complex models due to the lack of intuition and explanation of their predictions. The aim of this study to demonstrate the utility of various model-agnostic explanation techniques of machine learning models with a case study for analyzing the outcomes of the machine learning random forest model for predicting the individuals at risk of developing hypertension based on cardiorespiratory fitness data.

METHODS

The dataset used in this study contains information of 23,095 patients who underwent clinician-referred exercise treadmill stress testing at Henry Ford Health Systems between 1991 and 2009 and had a complete 10-year follow-up. Five global interpretability techniques (Feature Importance, Partial Dependence Plot, Individual Conditional Expectation, Feature Interaction, Global Surrogate Models) and two local interpretability techniques (Local Surrogate Models, Shapley Value) have been applied to present the role of the interpretability techniques on assisting the clinical staff to get better understanding and more trust of the outcomes of the machine learning-based predictions.

RESULTS

Several experiments have been conducted and reported. The results show that different interpretability techniques can shed light on different insights on the model behavior where global interpretations can enable clinicians to understand the entire conditional distribution modeled by the trained response function. In contrast, local interpretations promote the understanding of small parts of the conditional distribution for specific instances.

CONCLUSIONS

Various interpretability techniques can vary in their explanations for the behavior of the machine learning model. The global interpretability techniques have the advantage that it can generalize over the entire population while local interpretability techniques focus on giving explanations at the level of instances. Both methods can be equally valid depending on the application need. Both methods are effective methods for assisting clinicians on the medical decision process, however, the clinicians will always remain to hold the final say on accepting or rejecting the outcome of the machine learning models and their explanations based on their domain expertise.

Home Care Program Increases the Engagement in Patients With Heart Failure

Heart failure patients seen in the outpatient home care setting and in cardiology clinics have repeated emergency department visits, hospital admissions, and readmissions despite receiving education about their medications diet and self-care interventions such as daily weights. The objective of this evidence-based practice change was to determine, in home care patients, whether the use of standardized teach-back methodology educational materials would improve their knowledge and confidence in the self-care of their chronic disease. Of the 22 patients enrolled, 15 were not readmitted to the hospital within 9 months of home care admission. The Self-Care of Heart Failure Index revealed an improvement in patient and caregiver contributions to heart failure self-care maintenance (daily adherence and symptom monitoring). The findings suggest that engaging patients by increasing their knowledge of their disease and their self-confidence can reduce hospitalizations and subsequent readmissions.

Association of the Trauma of Hospitalization With 30-Day Readmission or Emergency Department Visit

IMPORTANCE

Trauma of hospitalization refers to the depersonalizing and stressful experience of a hospital admission and is hypothesized to increase the risk of readmission after discharge.

OBJECTIVES

To characterize the trauma of hospitalization by measuring patient-reported disturbances in sleep, mobility, nutrition, and mood among medical inpatients, and to examine the association between these disturbances and the risk of unplanned return to hospital after discharge.

DESIGN, SETTING, AND PARTICIPANTS

This prospective cohort study enrolled participants between September 1, 2016, and September 1, 2017, at 2 academic hospitals in Toronto, Canada. Participants were adults admitted to the internal medicine ward for more than 48 hours. Participants were interviewed before discharge using a standardized questionnaire to assess sleep, mobility, nutrition, and mood. Responses for each domain were dichotomized as disturbance or no disturbance. Disturbance in 3 or 4 domains (the upper tertile) was considered high trauma of hospitalization, and disturbance in 0 to 2 domains (the lower 2 tertiles) was considered low trauma.

MAIN OUTCOME AND MEASURES

The primary outcome was readmission or emergency department visit within 30 days of discharge. The association between trauma of hospitalization and the primary outcome was examined using logistic regression, adjusted for age; sex; length of stay; Charlson Comorbidity Index Score; Laboratory-Based Acute Physiology Score; and baseline disturbances in sleep, mobility, nutrition, and mood.

RESULTS

A total of 207 patients participated, of whom 82 (39.6%) were women and 125 (60.4%) were men, with a mean (SD) age of 60.3 (16.8) years. Among the 207 participants, 75 (36.2%) reported sleep disturbance, 162 (78.3%) reported mobility disturbance, 114 (55.1%) reported nutrition disturbance, and 48 (23.2%) reported mood disturbance. Nearly all participants (192 [92.8%]) described a disturbance in at least 1 domain, and 61 participants (29.5%) had high trauma exposure. A statistically significant 15.8% greater absolute risk of readmission or emergency department visit was found in participants with high trauma (37.7%; 95% CI, 25.9%-51.1%) compared with those with low trauma (21.9%; 95% CI, 15.7%-29.7%), which remained statistically significant after adjusting for baseline characteristics (adjusted odds ratio, 2.52; 95% CI, 1.24-5.17; P = .01) and propensity score matching (odds ratio, 2.47; 95% CI, 1.11-5.73; P = .03).

CONCLUSIONS AND RELEVANCE

Disturbances in sleep, mobility, nutrition, and mood were common in medical inpatients; such trauma of hospitalization may be associated with a greater risk of 30-day readmission or emergency department visit after hospital discharge.

Symptoms Reported by Frail Elderly Adults Independently Predict 30-Day Hospital Readmission or Emergency Department Care

OBJECTIVES

To assess the degree to which self-reported symptoms predict unplanned readmission or emergency department (ED) care within 30 days of high-risk, elderly adults enrolled in a posthospitalization care transition program (CTP).

DESIGN

Retrospective cohort study.

SETTING

Posthospitalization CTP at Mayo Clinic, Rochester, Minnesota, from January 1, 2013, through March 3, 2015.

PARTICIPANTS

Frail, elderly adults (N = 230; mean age 83.5 ± 8.3, 46.5% male).

MEASUREMENTS

Charlson Comorbidity Index (CCI) and self-reported symptoms, measured using the Edmonton Symptom Assessment System (ESAS), were ascertained upon CTP enrollment.

RESULTS

Mean CCI was 3.9 ± 2.3. Of 51 participants returning to the hospital within 30 days of discharge, 13 had ED visits, and 38 were readmitted. Age, sex, and CCI were not significantly different between returning and nonreturning participants, but returning participants were significantly more likely to report shortness of breath (P = .004), anxiety (P = .02), depression (P = .02), and drowsiness (P = .01). Overall ESAS score was also a significant predictor of hospital return (P = .01).

CONCLUSION

Four self-reported symptoms and overall ESAS score, but not CCI, ascertained after hospital discharge were strong predictors of hospital return within 30 days. Including symptoms in risk stratification of high-risk elderly adults may help target interventions and reduce readmissions.