Latent Class Analysis (LCA) to Identify Medical Device Usage in Real-World Data

Author(s)

Taeko Minegishi, PhD, MS, Geri Cramer, MBA, RN, PhD;
Boston Scientific, Marlborough, MA, USA

Presentation Documents

OBJECTIVES: In real-world data, identifying device usage can be challenging unless specific procedure codes are established. The ability to track both device usage and associated procedures is critical for evaluating real-world clinical outcomes and economic impacts of new technology. The Apollo OverStitch™ Endoscopic Suturing System can be used for procedures such as endoscopic sleeve gastroplasty (ESG), transoral outlet reduction (TORe), and defect closure within the gastrointestinal tract, for which specific procedure coding is still in development. We applied Latent Class Analysis (LCA), a probabilistic modelling to a real-world dataset utilizing the OverStitch device to subset patients into the primary procedures in which this device is used.
METHODS: A retrospective analysis of 2,231 outpatient encounters associated with the OverStitch device were identified using the Premier Healthcare Database between Jan 1, 2019 and Dec 31, 2022. Patient demographics, diagnosis, and procedure codes were included and reported on the most parsimonious model.
RESULTS: A four-class model provided optimal fit and interpretability, of which, two of the subgroups aligned closely with our anticipated characteristics of patients undergoing ESG or TORe. One subgroup accounted for 40% of the cohort, consisted primarily of females younger than 55 years old with commercial insurance or self-pay designation. The dominant codes included obesity, post-gastric surgery status, and unlisted code for the stomach. An additional subset comprised 25% of the population, characterized by Medicare beneficiaries with comorbidities such as hypertension and gastroesophageal reflux disease, and who had previously undergone bariatric procedures. This subgroup was likely, indicative of patients who received the TORe procedure.
CONCLUSIONS: We successfully classified OverStitch encounters into those likely to be ESG and TORe using LCA, enhancing the understanding about the demographics of patients undergoing procedures using OverStitch. This approach may be beneficial to isolate devices and procedures within real-world data when procedure codes are not available.

Conference/Value in Health Info

2025-05, ISPOR 2025, Montréal, Quebec, CA

Value in Health, Volume 28, Issue S1

Code

MT7

Topic

Medical Technologies

Disease

No Additional Disease & Conditions/Specialized Treatment Areas, SDC: Diabetes/Endocrine/Metabolic Disorders (including obesity)

Your browser is out-of-date

ISPOR recommends that you update your browser for more security, speed and the best experience on ispor.org. Update my browser now

×