Session number: 1093
Abstract title: Use of Pharmacy Benefits Management Data to Assess Medication Adherence
Author(s):
Wagner JH - VA Pittsburgh Healthcare System
Justice AC - VA Pittsburgh Healthcare System
Good CB - VA Pittsburgh Healthcare System
Cunningham F - PBM/SHG and Univ. of Illinois at Chicago
Rodriguez-Barradas M - VAMC Houston, TX
Objectives: Pharmacy Benefits Management database version 3.0 (PBM) provides an electronic data warehouse of prescriptions filled within the VA system. We hypothesized that PBM: 1) can be used to measure both current antiretroviral medications and adherence to these medications 2) will be independently associated with HIV-1 viral load after adjustment for patient and provider reported adherence.
Methods: VACS 3 is a multi-site cohort of 881 HIV+ adults from Cleveland, Manhattan, and Houston VA HIV clinics assembled between 6/99-7/00, representing 85% of the patients seen during this interval. Patients filled out a survey that included questions regarding adherence to their HIV medication (PtADH). Providers were also queried about their patients' adherence to HIV therapy (PrADH) and were asked to list specific antiretroviral medications prescribed (follow up survey only). HIV-1 RNA viral load was obtained from VA computerized laboratory data and was analyzed as a continuous measure (log 10). Electronic pharmacy data available from PBM included the specific medications, their prescription date, refills, and dosages. Using an algorithm developed by J. F. Steiner that requires at least one refill, this data was converted to an electronic measure of adherence (PBMadh).
Results: Validation Results: Provider report of current antiretroviral medications was available for 604 patients and of these, 515 (85.3%) had PBM data available. Provider and PBM agreement for antiretrovirals was good for all commonly prescribed medications (Agreement 90-73%; kappa 0.5-0.4).
Adherence Results: At baseline 793 patients were on antiretroviral therapy. Of these 505 (75%) had at least one refill and a recent viral load and were included in the multivariate model. Using multivariate analyses, each measure of adherence (patient, provider, and PBM) was independently associated with viral load after adjustment for the other measures (in all cases p <.001, R sq=0.25).
Conclusions: PBM v3.0 data represents an accurate record of current antiretroviral medications prescribed. Further, a measure of adherence based upon PBM refill data was independently associated with HIV-1 viral load after adjustment for patient and provider report of adherence.
Impact statement: PBM is a valid source of information on current antiretroviral medications and adherence for veterans in care.