Non-parametric recurrent events analysis with BART and an application to the hospital admissions of patients with diabetes.

Much of survival analysis is concerned with absorbing events, i.e., subjects can only experience a single event such as mortality. This article is focused on non-absorbing or recurrent events, i.e., subjects are capable of experiencing multiple events. Recurrent events have been studied by many; however, most rely on the restrictive assumptions of linearity and proportionality. We propose a new method for analyzing recurrent events with Bayesian Additive Regression Trees (BART) avoiding such restrictive assumptions. We explore this new method via a motivating example of hospital admissions for diabetes patients and simulated data sets.

Use of secondary clinical data for research related to diabetes self-management education.

BACKGROUND: Diabetes self-management education (DSME) is a key component of ensuring optimal diabetes outcomes. Electronic medical record (EMR) systems have transformed diabetes management by providing organized and useful data. However, important gaps remain in the process of how practice settings track referrals and attendance to DSME. PURPOSE: The purpose of this study was to use EMR data to examine patients' demographic, behavioral, and diabetes risk factors by referral pattern to a DSME program in a large midwestern Academic Medical Center. METHODS: A retrospective cross-sectional design using 2006-2013 EMR data from a Clinical Research Data Warehouse (CRDW). Data on 10,000 patients with type 2 diabetes mellitus (T2DM) were randomly extracted from the CRDW for analysis. Multiple logistic regression analysis was employed to explore adjusted associations with referral to DSME. RESULTS: Seven hundred forty patients with T2DM were referred to DSME. Results show that age at diagnosis, insurance status, race/ethnicity, language, alcohol use, use of insulin, HbA1c, LDL, systolic blood pressure, ophthalmology appointment, coronary artery disease, neuropathy, diabetic-retinopathy, and nephropathy were found to be factors significantly associated with a referral to DSME. Language emerged as a significant result; non-English speakers were more likely to receive a referral to DSME. CONCLUSIONS: Patients referred for DSME had appropriate medical complications or social needs that would benefit from intensive education; however, there remains a considerable opportunity for improving the DSME referral process. Aspects of the physician decision-making process to refer or not refer patients to DSME warrant further investigation.