On analyzing frequent readmissions using a group-based trajectory model

Ofir Ben-Assuli, Rema Padman, Martha Bowman, Moshe Leshno, Itamar Shabtai

Research output: Contribution to conferencePaperpeer-review

Abstract

The problem of readmission, wherein some patients return shortly after they are discharged and are readmitted for the same or related condition, has become a challenge worldwide due to care quality and financial concerns. Predicting who is likely to be readmitted and understanding the factors contributing to preventable readmissions is being widely researched, but few studies have examined the role of clinical disease markers in predicting frequent readmissions related to disease progression. In this study, we explore 30-day readmission risk prediction using the developmental trajectories of patients' creatinine levels, a key laboratory marker of serious illness, as a potential predictor of future readmission for patients with a large number of repeat hospital visits. Using Electronic Health Record data on 7,722 patients over seven to twelve separate visits to seven major Israeli EDs, we explore risk stratification and prediction using a semi-parametric, group-based, statistical model. Preliminary results suggest four distinct creatinine-based trajectories over time, with significantly differing readmission rates by age, gender, creatinine levels and length of stay that may enable readmission risk stratification of the patient population for targeted interventions. Further research will incorporate other disease markers using multi-trajectory models, combined with time-invariant and time-varying covariates, to identify more insightful set of longitudinal factors for readmission risk reduction in patients with frequent hospital visits.

Original languageEnglish
StatePublished - 2015
Event25th Annual Workshop on Information Technologies and Systems, WITS 2015 - Dallas, United States
Duration: 12 Dec 201513 Dec 2015

Conference

Conference25th Annual Workshop on Information Technologies and Systems, WITS 2015
Country/TerritoryUnited States
CityDallas
Period12/12/1513/12/15

Fingerprint

Dive into the research topics of 'On analyzing frequent readmissions using a group-based trajectory model'. Together they form a unique fingerprint.

Cite this