Application of Prevalent New-User Cohort Designs to a Claims Data Study of Incidence of Nutritional Deficiency in People With Diabetes Using GLP-1 Receptor Agonists
Author(s)
Thadchaigeni Panchalingam, PhD1, Andrew Chang, PhD, MPH, MS2, Kirk W. Kerr, PhD3, Dominique R. Williams, MD, MPH4, Suela Sulo, PhD5, Refaat Hegazi, MD, MPH, MBA, PHD4, Steven Heymsfield, MD6, Scott Goates, MBA, PhD1;
1Abbott Medical Devices, Sylmar, CA, USA, 2Abbott Medical Devices, Santa Clara, CA, USA, 3Abbott Nutrition, Senior Manager, Columbus, OH, USA, 4Abbott Nutrition, Columbus, OH, USA, 5Abbott Nutrition, Abbott Park, IL, USA, 6Pennington Biomedical Research Center, Baton Rouge, LA, USA
1Abbott Medical Devices, Sylmar, CA, USA, 2Abbott Medical Devices, Santa Clara, CA, USA, 3Abbott Nutrition, Senior Manager, Columbus, OH, USA, 4Abbott Nutrition, Columbus, OH, USA, 5Abbott Nutrition, Abbott Park, IL, USA, 6Pennington Biomedical Research Center, Baton Rouge, LA, USA
Presentation Documents
OBJECTIVES: Apply prevalent new-user cohort design to real-world data for comparison of incidence of nutrition deficiencies in patients using and not using Glucagon-like peptide-1 receptor agonists (GLP-1RAs).
METHODS: Cohorts of patients with type 2 diabetes prescribed metformin with or without an additional prescription for GLP-1RAs between 07/2017-06/2023 were formed from Inovalon Insights claims data. Patients using insulin, with type 1 diabetes, without continuous 6-months of enrollment at baseline or with prior diagnosis of nutrition deficiency were excluded. Patients were observed longitudinally to identify when GLP-1RAs were prescribed. Monthly hazard for GLP-1RA prescriptions were calculated for all patients using sociodemographic variables, comorbidities, and baseline health care costs. Each GLP-1RA patient was matched to the closest non-GLP-1RA patient using hazard of GLP-1RA prescription with exact match on cohort year and the months of metformin use.
RESULTS: Of 4,557 GLP-1RA users that met the inclusion criteria, 4,505 were matched to a GLP-1RA non-users. Matching reduced the standardized mean difference between GLP-1RA users and non-users to less than 0.1 for almost all variables. Comparison reveals that GLP-1RA users had slightly higher incidence of nutrition deficiencies in the year following GLP-1RA prescription than non-users (18.6% vs 16.5% p<0.01).
CONCLUSIONS: Application of prevalent new-user study design to claims data demonstrates the feasibility of developing comparisons of patients using and not using a medication while mitigating time-zero bias, while showing that GLP-1RA users had higher incidence of nutrition deficiencies than non-users (18.6% vs 16.5%) . This statistical methodology can support future analysis of real-world nutrition data. incidence of nutrition deficiencies in the year following GLP-1RA prescription than non-users (18.6% vs 16.5% p<0.01).
METHODS: Cohorts of patients with type 2 diabetes prescribed metformin with or without an additional prescription for GLP-1RAs between 07/2017-06/2023 were formed from Inovalon Insights claims data. Patients using insulin, with type 1 diabetes, without continuous 6-months of enrollment at baseline or with prior diagnosis of nutrition deficiency were excluded. Patients were observed longitudinally to identify when GLP-1RAs were prescribed. Monthly hazard for GLP-1RA prescriptions were calculated for all patients using sociodemographic variables, comorbidities, and baseline health care costs. Each GLP-1RA patient was matched to the closest non-GLP-1RA patient using hazard of GLP-1RA prescription with exact match on cohort year and the months of metformin use.
RESULTS: Of 4,557 GLP-1RA users that met the inclusion criteria, 4,505 were matched to a GLP-1RA non-users. Matching reduced the standardized mean difference between GLP-1RA users and non-users to less than 0.1 for almost all variables. Comparison reveals that GLP-1RA users had slightly higher incidence of nutrition deficiencies in the year following GLP-1RA prescription than non-users (18.6% vs 16.5% p<0.01).
CONCLUSIONS: Application of prevalent new-user study design to claims data demonstrates the feasibility of developing comparisons of patients using and not using a medication while mitigating time-zero bias, while showing that GLP-1RA users had higher incidence of nutrition deficiencies than non-users (18.6% vs 16.5%) . This statistical methodology can support future analysis of real-world nutrition data. incidence of nutrition deficiencies in the year following GLP-1RA prescription than non-users (18.6% vs 16.5% p<0.01).
Conference/Value in Health Info
2025-05, ISPOR 2025, Montréal, Quebec, CA
Value in Health, Volume 28, Issue S1
Code
EPH59
Topic
Epidemiology & Public Health
Disease
SDC: Cardiovascular Disorders (including MI, Stroke, Circulatory), SDC: Diabetes/Endocrine/Metabolic Disorders (including obesity), STA: Nutrition