Analyzing temporal patterns in frequent emergency department visits among oncology patients using semantic similarity measures

Hyojung Kang, Lloyd Fernandes, John P. Riordan

Research output: Contribution to journalArticlepeer-review

Abstract

Purpose: Patients with cancer face complex challenges that often lead them to seek care in emergency departments (EDs) for acute symptoms and complications. While previous studies have examined preventable ED visits by patients with cancer, less attention has focused on frequent ED use in this population. This study aimed to explore the temporal visit patterns of frequent ED users with cancer. Methods: This retrospective study analyzed ED visit records from 10 large hospitals (five urban, five rural) in Illinois from 2018 to 2019. Adult frequent ED users, defined as those with four or more visits in a year, were classified as cancer or non-cancer users based on cancer-related diagnoses in their first four visits. The Wu-Palmer method was used to calculate semantic similarity between diagnoses. Results: Of the 98,246 frequent ED users, 18.2 % had at least one cancer-related visit. Cancer patients made up 21.7 % of frequent users in urban EDs and 7.6 % in rural EDs. Cancer patients had slightly higher proximity scores than non-cancer patients. Patients with central nervous system malignancies had the highest median proximity score, while breast cancer patients had a lower score. In both urban and rural EDs, cancer patients had significantly higher proximity scores than non-cancer patients, with the difference being more pronounced in urban EDs. Conclusion: Frequent ED visits by oncology patients are more likely to be for similar reasons compared to non-cancer frequent users. These findings highlight the need for targeted care coordination for oncology patients who frequently visit the ED, potentially reducing repeat visits through improved outpatient management. Understanding the distinct patterns of ED utilization among cancer patients could help prioritize resource allocation and care pathways.

Original languageEnglish (US)
Pages (from-to)51-56
Number of pages6
JournalAmerican Journal of Emergency Medicine
Volume89
DOIs
StatePublished - Mar 2025

Keywords

  • Frequent users
  • Natural language processing
  • Oncology patients
  • Visit patterns

ASJC Scopus subject areas

  • Emergency Medicine

Fingerprint

Dive into the research topics of 'Analyzing temporal patterns in frequent emergency department visits among oncology patients using semantic similarity measures'. Together they form a unique fingerprint.

Cite this