Graduation Year


Document Type




Degree Granting Department

Public Health

Major Professor

Barbara Langland Orban


care management, chronic disease, health, managed care, quality


Research supports the clinical and societal benefits of antiretroviral treatment (ART) for people living with HIV or AIDS (PLWHA). However, factors associated with ART adherence and the relationship of ART adherence to total healthcare expenditures are not well understood.

The research data included Florida Medicaid claims for five years (July 2006 through June 2011). All subjects (n=514) were HIV-positive, adult, non-pregnant, and ART naïve for at least 12 months prior to their 24 month measurement period. Each subject was categorized as adherent (>90%#37;) or non-adherent (<90%#37;) based upon medication possession ratios (MPR). Total expenditures were payments Medicaid made to providers and pharmacies.

Objective 1 modeled the logit probability of a subject being non-adherent to ART (versus adherent). Certain factors were expected to have significant negative associations with non-adherence to ART, including females, older age group, AIDS diagnosis, adherence to antidepressants, severe mental illness, meeting the minimum recommended number of outpatient visits, ART regimen type, number of medications in the ART regimen, residing in a county with a high median income, and residing in a county with an urban population density. The variables expected to have significant positive associations included race/ethnicity, substance or alcohol abuse diagnosis, depression or anxiety diagnoses, disease progression from HIV to AIDS, discontinuous Medicaid coverage, Medicaid eligibility type, and co-morbidity count levels.

The Objective 1 results showed that all non-white race/ethnicity categories had at least twice the odds of being classified as non-adherent. Also, increasing numbers of concurrent ART medications significantly predicted non-adherence; the odds ratio for three to five ART medications was 2.04 (95%#37; CI=1.04-4.01, p<.05), and six or more prescriptions category odds ratio was 4.58 (95%#37; CI=1.82-11.56, p<.01), as compared to a single medication. Finally, a chronic diseases diagnosis was protective against non-adherence (OR=.46, 95%#37; CI=.26-.84, p<.01), as was adherence to antidepressants (OR=.28, 95%#37; CI=.14-.54, p<.01).

In Objective 2, it was expected that the ART adherence group, the explanatory variable, would have significantly less monthly mean total healthcare expenditures, the outcome variable. For each of the HIV-positive (n=232) and the AIDS-diagnosed (n=270) groups, a generalized linear model predicted the mean total expenditures for the ART non-adherence group (<90%#37; MPR) versus the ART adherence group, controlling for other factors. For the HIV-positive subjects, the predicted mean total healthcare expenditures for the ART non-adherent group was $1,291 (95%#37; CL $840-$2,004); the predicted mean for the adherent group was $1,926 (95%#37; CL $1,157-$3,231). The difference was statistically significant, but the hypothesis was not supported. The non-adherent group mean was less than the adherent group (-40%#37;, p<.001). However, for the AIDS-diagnosed subjects, there was no statistical difference between the non-adherent and adherent groups. The predicted mean for the non-adherent group was $2,279 (95%#37; CL $1,572-$3,322), and $2,005 (95%#37; CL $1,387-$2,913) for the adherent group.

The findings of this research support the need for translating evidence on racial/ethnic disparities in ART adherence, along with behavioral, social, or cultural barriers and effective interventions, into policy and practice. Also, certain medication management strategies should be implemented to reduce the number of medications in ART regimen. Finally, the results of the present study underscores the necessity for appropriate financial incentives and purposeful risk-adjusted capitation payment structures that would support ART adherence among Medicaid-insured PLWHA.