Methods: We conducted a retrospective analysis of patients with ≥ 1 HIV clinic visit at the Duke Adult Infectious Diseases Clinic between 2008 and 2013. Healthcare utilization was characterized by four indicators: clinic attendance in each half of the calendar year (yes/no), number of emergency department (ED) visits (0, 1 or 2+), inpatient admissions per year (0, 1, 2+), and viral suppression (never, intermittent, always). Healthcare engagement patterns were modeled using latent class/latent transition analysis, with model fit assessed using the Bayesian Information Criterion.
Results: The cohort included 2686 patients (median age 42.9 years, 72% male, 56% black). A three-class model best fit the data: “Adherent” “Non-adherent” and “Sick”. “Adherent” patients had high rates of clinic attendance in each half of the year (84%), rarely visited the ED (3.6% with ≥1 ED visit per year), and moderate rates of (54%) viral suppression. “Non-adherent” patients rarely attended clinic visits in both halves of the year (1.5%), used the ED more than “adherent” patients (10.3% with ≥1 ED visit per year), and had low rates of viral suppression (19%). “Sick” patients also had high rates of clinic attendance (75%), were frequent users of the ED (53% with ≥1 ED visit per year), and comparable rates of viral suppression to the “adherent” group (55%) viral suppression. Non-white race (OR 1.9) and age ≤ 40 (OR 3.76) were associated with membership in the “non-adherent” class. Movement between classes was dynamic, especially in the “sick” group (30-40% of whom moved to a different class the following year). Across all years, persons in the “non-adherent” class were more likely to completely disengage from care the following year than “adherent” persons (23.6% v. 0.2%, p <0.001).
Conclusion: A broader definition of healthcare engagement revealed distinct and dynamic patterns among persons with HIV that would have been hidden had only clinic attendance and viral suppression been considered. These patterns may be useful for designing engagement-targeted interventions.
N. L. Okeke,
M. McKellar, None
J. Stout, None