2012 HSR&D/QUERI National Conference Abstract
1038 — An Empirically-Based Typology of VA Users Representing Patterns of VA Service Utilization: Latent Class Analysis
Vaughan Sarrazin MS, and Rosenthal GE, Iowa City VAMC;
Understanding patterns of VA utilization is critical to forecasting future utilization of VA services and the impact of healthcare reform on demand for VA services. This study sought to empirically derive a typology for classifying Veterans’ use of VA services.
VA users in VISN 23 during FY2009 were identified (n = 259,567). The use of specific outpatient, inpatient, and pharmacy services was quantified using VA administrative and pharmacy files. Outpatient encounters were classified as Emergency, Mental Health, Specialty, or Primary Care. Prescription medications were collapsed into 23 mutually exclusive categories. Latent cluster analysis was used to explain associations between the manifest (observed) use of specific services and latent (unobserved) variables. A 5-class model provided the best fit; patients were assigned to one of the 5 classes based on the posterior probability of cluster membership.
Low Users (26%) represent the largest group. Only 22% of this group had >1 primary care visit, and only 10% used any specialty or mental health services. Less than 3% received prescriptions for >4 drug categories, and only 1% received psychiatric medications. Pharmacy VA users (23%) were similar to ‘Low Users’, with few primary or specialty care visits. However, nearly 70% received prescriptions for >4 drug categories. Moderate Users (22%) use more primary care, mental health, and drug categories than Low Users. Like ‘Low Users’, they were unlikely to receive psychiatric medications or mental health services. Mental Health Users (14%) are similar to ‘Moderate Users’, except they use more mental health and pharmacy services. Nearly 80% receive 4 or more categories of drugs and 100% receive psychiatric medications. High Users (15%) use all VA services at a high level, and account for 90% of all hospitalizations. Notably, 97% receive prescriptions for >4 drug categories.
Distinct categories of VA users can be defined by levels of use of primary, specialty, mental health, pharmacy, and inpatient services. The different categories likely reflect differences in Veterans’ needs for services and their access to specific VA and non-VA services.
This empirically derived typology representing patients with similar patterns of service utilization is feasible and may provide important information for VA planning purposes.