Defining the best tool to assess adherence to
antiretroviral therapy using sensitivity and specificity analysis of simple
questions and pharmacy records
C. Pereira de Souza1,2, L.A. Alves de Lima1, G. Santoro Lopes1
1Federal University of Rio de Janeiro, Preventive Medicine - Infectious Diseases Service, Rio de Janeiro, Brazil, 2Hospital Escola São Francisco de Assis - Federal University of Rio de Janeiro, AIDS Care Unit, Rio de Janeiro, Brazil
Background: Adherence is essential for a successful antiretroviral therapy.
Incomplete suppression of viral replication has been used as marker of poor
adherence. Self-report questionnaires and pharmacy records are considered strong
predictors of suboptimal adherence, but to establish which questions should be
asked to obtain maximal sensitivity and specificity, is not well established.
A self-reported questionnaire with four simple questions, including a visual
analogue grades scale attributed to self adherence, from zero to ten, was
applied a single time among HIV infected individuals on ART. Additionally,
pharmacy records were computed from 2 months before to 2 months after the
interviews were performed. The answers and registers were confronted with
plasma viral load measurements closest to the date of interview. Incomplete
suppression of viral replication was considered evidence of poor adherence. ROC
curves were constructed, using the parameters found to be statistically
associated with detectable viral load, to determine which combination of them
would result in maximal AUC (area under curve). Statistical analysis was done using
Results: 441 patients, 35.2% female, were interviewed.
356 (80.4%) patients had undetectable viral load. Self attributed grades smaller
than 9 in the past month, multiple delays in drug intake and pharmacy records
showing at least one interval between refills greater than 40 days were independently
associated with detectable viral load. Maximal accuracy for non adherence was
obtained when at least two of those three parameters are present.
Maximal accuracy for suboptimal adherence can be obtained by using a score
based on a simple set of questions and pharmacy records.
Back to the Programme-at-a-Glance