Simplified Machine Learning Models Can Accurately Identify High-Need High-Cost Patients With Inflammatory Bowel Disease

A Narrative Review of Financial Burden, Distress, and Toxicity of Inflammatory Bowel Diseases in the United States

INTRODUCTION

Inflammatory bowel diseases (IBD) affect >3 million Americans and are associated with tremendous economic burden. Direct patient-level financial impacts, financial distress, and financial toxicity are less well understood. We aimed to summarize the literature on patient-level financial burden, distress, and toxicity associated with IBD in the United States.

METHODS

We conducted a literature search of US studies from 2002 to 2022 focused on direct/indirect costs, financial distress, and toxicity for patients with IBD. We abstracted study objectives, design, population characteristics, setting, and results.

RESULTS

Of 2,586 abstracts screened, 18 articles were included. The studies comprised 638,664 patients with IBD from ages 9 to 93 years. Estimates for direct annual costs incurred by patients ranged from $7,824 to $41,829. Outpatient costs ranged from 19% to 45% of direct costs, inpatient costs ranged from 27% to 36%, and pharmacy costs ranged from 7% to 51% of costs. Crohn's disease was associated with higher costs than ulcerative colitis. Estimates for indirect costs varied widely; presenteeism accounted for most indirect costs. Severe and active disease was associated with greater direct and indirect costs. Financial distress was highly prevalent; associated factors included lower education level, lower household income, public insurance, comorbid illnesses, severity of IBD, and food insecurity. Higher degrees of financial distress were associated with greater delays in medical care, cost-related medication nonadherence, and lower health-related quality of life.

DISCUSSION

Financial distress is prevalent among patients with IBD; financial toxicity is not well characterized. Definitions and measures varied widely. Better quantification of patient-level costs and associated impacts is needed to determine avenues for intervention.

The cost of inflammatory bowel disease in high-income settings: a Lancet Gastroenterology & Hepatology Commission

The cost of caring for patients with inflammatory bowel disease (IBD) continues to increase worldwide. The cause is not only a steady increase in the prevalence of Crohn's disease and ulcerative colitis in both developed and newly industrialised countries, but also the chronic nature of the diseases, the need for long-term, often expensive treatments, the use of more intensive disease monitoring strategies, and the effect of the diseases on economic productivity. This Commission draws together a wide range of expertise to discuss the current costs of IBD care, the drivers of increasing costs, and how to deliver affordable care for IBD in the future. The key conclusions are that (1) increases in health-care costs must be evaluated against improved disease management and reductions in indirect costs, and (2) that overarching systems for data interoperability, registries, and big data approaches must be established for continuous assessment of effectiveness, costs, and the cost-effectiveness of care. International collaborations should be sought out to evaluate novel models of care (eg, value-based health care, including integrated health care, and participatory health-care models), as well as to improve the education and training of clinicians, patients, and policy makers.

Novel Machine Learning Approach to Predict and Personalize Length of Stay for Patients Admitted with Syncope from the Emergency Department

Background: Syncope, a common problem encountered in the emergency department (ED), has a multitude of causes ranging from benign to life-threatening. Hospitalization may be required, but the management can vary substantially depending on specific clinical characteristics. Models predicting admission and hospitalization length of stay (LoS) are lacking. The purpose of this study was to design an effective, exploratory model using machine learning (ML) technology to predict LoS for patients presenting with syncope. Methods: This was a retrospective analysis using over 4 million patients from the National Emergency Department Sample (NEDS) database presenting to the ED with syncope between 2016−2019. A multilayer perceptron neural network with one hidden layer was trained and validated on this data set. Results: Receiver Operator Characteristics (ROC) were determined for each of the five ANN models with varying cutoffs for LoS. A fair area under the curve (AUC of 0.78) to good (AUC of 0.88) prediction performance was achieved based on sequential analysis at different cutoff points, starting from the same day discharge and ending at the longest analyzed cutoff LoS ≤7 days versus >7 days, accordingly. The ML algorithm showed significant sensitivity and specificity in predicting short (≤48 h) versus long (>48 h) LoS, with an AUC of 0.81. Conclusions: Using variables available to triaging ED clinicians, ML shows promise in predicting hospital LoS with fair to good performance for patients presenting with syncope.