Abstract Book Vol. 2 [International Conference on AIDS (14th: 2002: Barcelona, Spain)]

42 Abstracts WePeB5843-WePeB5846

XIV International AIDS Conference

users. Patients were followed for 3 years post-index. Adherence was defined as
cumulative days covered by filled NRTI prescriptions as a proportion of total observed days. Patients were stratified by baseline adherence scores (in year 1)
and assumed to require NRTI therapy over the observation period.
Results: A total of 2,369 patients were studied. Mean adherence rates for individuals stratified by their adherence scores in year 1 and followed through to year 3
are shown. Controlling for age and gender, differences in adherence rates across
time by adherence category were tested through repeated measures ANOVA and
found to be significant (p <0.0001).
Mean Adherence Rates
Adherence Category n Yr 1 Yr 2 Yr 3
0%-19% 431 12% 15% 22%
20%-39% 392 30% 30% 36%
40%-59% 438 50% 48% 50%
60%-79% 459 70% 62% 59%
80%-100% 649 91% 81% 74%
Overall 2369 55% 51% 51%
Conclusions: Overall adherence findings may be misleading as a consequence
of year 1 poor adherents improving and better adherents decreasing their adherence in later years. These patterns could raise critical resistance issues. Our results suggest that adherence programs and regimen simplification targeting treatment experienced patients as well as treatment naive individuals may provide for
improved virologic success rates and assist in minimizing resistant strains of HIV.
Presenting author: Christopher Dezii, 777 Scudders Mill Road, Plainsboro, NJ
08536, United States, Tel.: [phone], Fax: [phone], E-mail: [email protected]
WePeB58431 The impact of depression and substance abuse on
antiretroviral treatment adherence in newly-treated
individuals with hiv infection
C.M. Dezii1, S. Grandy1, B. Burtcel1, S. Hodder', H. Kawabata2. 1IBristo/-Myers
Squibb Virology, 777 Scudders Mill Road, Plainsboro, NJ 08536, United States;
2Bristol-Myers Squibb Outcomes Research, Plainsboro, NJ, United States
Background: Depression and substance abuse are thought to adversely impact
adherence to antiretroviral therapy.
Methods: Retrospective analysis of California Medicaid (Medi-Cal) data using
pharmacy claims from 3Q '94 to 2Q '00 was performed. As most treated patients
during this period were on nucleoside reverse transcriptase inhibitors (NRTI), the
first NRTI claim served as the index. Patients were followed for 1 year post-index.
Adherence was defined as cumulative days covered by filled NRTI prescriptions
as a proportion of total observed days. To identify individuals with substance
abuse, the following ICD-9 codes were selected: all 304 (drug dependencies),
all 305 (nondependent drug use), 969.6 (hallucinogen poisoning) and 969.7 (amphetamine poisoning). ICD-9 codes identifying depression were: 296.2, 296.3,
300.4, 301.12 (single-occurrence, recurrent, neurotic, and chronic depression,
respectively.)
Results: A total of 4,760 patients were studied. Adherence rates by category are
shown.
Category (n) NRTI Adherence Rate
No Substance Abuse Dx/No Depression Dx (3332) 54.0%
No Substance Abuse Dx/Depression Dx (412) 56.6%
Substance Abuse Dx/No Depression Dx (766) 47.8%
Substance Abuse Dx/Depression Dx (250) 47.8%
The independent effect of substance abuse on adherence was statistically significant (GLM, p <0.0001) controlling for age and gender while depression was
not.
Conclusions: Individuals diagnosed with substance abuse exhibited lower NRTI
adherence compared to those without diagnosed substance abuse regardless of
accompanying depression. Similar adherence rates in individuals with or without
diagnosed depression may suggest that recognition and treatment may overcome
the adherence barrier of depression.
Presenting author: Christopher Dezii, 777 Scudders Mill Road, Plainsboro, NJ
08536, United States, Tel.: [phone], Fax: [phone], E-mail: [email protected]
WePeB5844 A single, untimed determination of drug levels
after starting therapy is a significant predictor of
mortality after starting antiretroviral therapy
C.S.A. Alexander, J. Asselin, R.S. Hogg, J.S.G. Montaner,
M.V.O. O'Shaughnessy, L. Ting, P.R. Harrigan. BC Centre for Excellence in
HI V/AIDS, 603-1081 Burrard St, Vancouver, BC, Canada

Background: Plasma drug levels, influenced by pharmacokinetic (PK) parameters and medication adherence, may predict response to antiretroviral therapy.
However, using routine therapeutic drug monitoring in the clinic is limited by both
its relatively cumbersome nature and lack of agreement on optimum target values and sampling strategies. We wished to determine whether a simplified approach of examining plasma drug levels in single untimed samples collected for

viral load testing can predict eventual therapeutic outcomes (plasma viral load
and death).
Methods: Plasma drug concentrations of all commercially available NNRTIs and
PIs were determined by a validated assay (HPLC-MS/MS) in the first post-therapy
viral load sample of patients who initiated triple therapy at the BC Centre for Excellence in HIV/AIDS between Aug/96 and Sept/99 with CD4 counts <50 (N=129).
Drug levels were defined as unexpectedly low if below the published Ctrough -
1SD or the assay limit (2610 ng/mL for NVP, 2289 (DLV), 761 (EFZ), 134 (APV),
85 (IDV), 1485(LPV), 648 (NFV), 1110 (RTV), or 50 (SQV)).
Results: In the first post-therapy HIV viral load sample, lower than expected levels
of antiretroviral medication were observed in 50 individuals (38.8%). This single
measure was associated with failure to ever achieve a confirmed plasma viral load
below 500 HIV RNA copies/mi (p=0.001) and with increased mortality (p=0.04).
These observations may be mediated by both pharmacokinetic and adherence
factors, since individuals who had unexpectedly low drug levels were less likely
to renew prescriptions on time in the following year than those who did not (48%
vs 76%).
Conclusion: A single, untimed measure of unexpectedly low drug levels is both
common and a statistically significant predictor of time to death after starting antiretroviral therapy This simple laboratory measure may be a practical method to
spark interventions targeting adherence and/or PK problems.
Presenting author: Paul Richard Harrigan, 603-1081 Burrard St, Vancouver, BC,
Canada, Tel.: [phone], Fax: [phone], E-mail: [email protected].
ca
WePeB5845 Differences between provider-estimated and
self-reported adherence to antiretroviral therapy
S. Mannheimer1, M. Perkins2, J. Ford', B. Diamond3, L. Farrell2, S. Findley4,
A. Hofmann1, W. EI-Sadr.. Columbia University, Harlem Hospital, Division of
Infectious Diseases, Room 3101A, Harlem Hospital, 506 Lenox Avenue, New
York, NY United States; 2Harlem Hospital New York, United States; 3Columbia
University, Irving Center for Clinical Research, New York, United States;
4 Co/umbia University, Mailman School of Public Health, New York, United States
Background: While provider assessment is critical in decisions concerning antiretroviral therapy (ART), providers have been shown to be poor at estimating
patient (pt) adherence (adh).
Methods: Provider-estimated and self-reported adh were compared among pts
enrolled in an adh support study at an inner city hospital-based clinic. Adh was
measured concurrently by 3-day self-report and by provider assessment (on a
scale of 0-100%). Self-report adh scores were calculated as a ratio of doses
taken to doses prescribed; >90% was considered adherent for analysis. Additional baseline data collected through interviews included demographics, psychosocial factors, and substance use. CD4 count and HIV RNA levels were collected by chart abstraction.
Results: Of 169 pts, 56% were woman, 80% African-American, and 15% Latino.
Agreement between provider and pt self report of adh was low (Kappa=0.11,
p=0.15). 61% of pts were adherent by self-report, compared to 44% by provider
estimate. By univariate analysis, Latinos were more likely to be adherent by
provider estimate and by self-report (p=.03). Pts with depression self-reported
poorer adh (p=.03). A higher CD4 count (p<.001), lower HIV RNA level (p=.001)
and having an AIDS diagnosis (p=.021) were associated with provider-estimated
adh, but not self-report. Multivariate analysis confirmed the association between
provider-estimated adh, lower HIV RNA levels (p=.001), and higher CD4 counts
(p=.005). Prior psychiatric hospitalization (p=.02) and current homelessness
(p=.001) were associated with nonadherence by self-report, but not by provider
assessment.
Conclusions: There was low agreement between provider and self-reported adh
to ART. Provider dependence on laboratroy values can be misleading in settings
of HIV resistance to ART or variability of ART absorption. Providers should be
aware of the impact of psychosocial factors on adherence, and that pts with special needs may require additonal adherence support.
Presenting author: Sharon Mannheimer, Division of Infectious Diseases, Room
3101A, Harlem Hospital, 506 Lenox Avenue, New York, NY, United States, Tel.:
[phone], Fax: [phone], E-mail: [email protected]
WePeB5846 Self-efficacy expectation and adherence to
antiretroviral therapy in HIV infected patients: a
cross-sectional study in Southern Brazil
C.A.T. Pinheiro1, J.C.C. Leite2, M.L. Drachler2, V.L. Silveira3, M.O. Centeno2.
Universidade Federal de Pelotas, Pelotas RS, Brazi; 2Universidade Catolica de
Pelotas, Brazil; 3Secretaria Municipal de Saude de Pelotas, Pelotas, Brazil
Background: In the management of HIV-infected patients, adherence to antiretroviral therapy is crucial to maintaining viral suppression. The purpose of this
study was to explore factors that may be associated with adherence.

Methods: A cross-sectional study was conducted with HIV-infected adults on
treatment with antiretroviral drugs at a reference service in Southern Brazil. Participants answered a sociodemographic questionnaire and scales assessing sociocognitive variables. Adherence was assessed by a self-report inventory developed for the study. Clinical information was extracted from the patients' records.
Significance test were conducted using univariate logistic regressions folowed by
multivariate logistic regression analisys.