icon-folder.gif   Conference Reports for NATAP  
  5th IAS Conference on HIV Pathogenesis, Treatment and Prevention
July 19th-22nd 2009
Capetown, South Africa
Back grey_arrow_rt.gif
Influence of patient baseline clinical and demographic characteristics on choice of initial antiretroviral therapy regimen:
evidence of channeling bias in HIV clinical care

  Reported by Jules Levin
5th IAS Conference on HIV Pathogenesis Treatment and Prevention Capetown July 19-22 2009
ES Brouwer1,2, S Napravnik1,2, JJ Eron2 1University of North Carolina Gillings School of Global Public Health, Epidemiology, Chapel Hill, United States, 2University of North Carolina, Department of Medicine, Chapel Hill, United States
Clinical and socio-demographic characteristics influence initial antiretroviral regimen choice
Some "channeling" is expected due to published known effects of certain antiretrovirals such as teratogenicity of efavirenz and known renal toxicity of tenofovir
The channeling of patients with hypertension and LDL cholesterol > 100 mg/dl to an initial abacavir use may result in channeling bias in studies assessing cardiovascular outcomes related to abacavir.
Advanced epidemiologic analytic methods are necessary to account for channeling bias in clinical cohort studies.
Study population included all UNC HIV Clinical Cohort participants who initiated cART after 09/2004 (FDA TDF/FTC combination approval) (N=150). We evaluated use of the most commonly prescribed nucleoside reverse transcriptase inhibitors (NRTIs): lamivudine or emtricitabine (3TC/FTC), abacavir (ABC), tenofovir (TDF), and zidovudine (AZT); nonnucleoside reverse transcriptase inhibitor (NNRTI): efavirenz (EFV); and protease inhibitors (PIs): atazanavir (ATV[RTV]), and lopinavir (LPV/RTV). Multivariable logistic regression models were fit to identify independent predictors of ARV use, including: demographic characteristics; comorbidities (diabetes, hypertension, and coronary artery disease); and laboratory measures (cholesterol, liver function tests, calculated glomerular filtration rate, CD4cell count and HIVRNA level).
One-third of patients were women (31%), 51% African American, with median age of 41 years (IQR: 32, 50).
Patients received: 3TC/FTC (100%), TDF (75%), ABC (11%), AZT (11%), EFV (45%), LPV/RTV (30%), ATV (3%) and ATV/RTV (11%).
Elevated cholesterol level (LDL-C>100 mg/dL) and hypertension predicted greater ABC versus TDF use (all P<0.05).
Elevated liver function tests (ALT>35 units/L) predicted greater TDF versus ABC use (P=0.002).
MSM predicted AZT versus TDF use (P=0.02).
Being a woman and having a lower CD4 predicted LPV/RTV versus EFV use (P=0.03 and 0.01).
Higher CD4 and elevated ALT predicted EFV versus ATV/RTV use (P=0.04 and 0.004). Elevated ALT predicted LPV/RTV versus ATV/RTV use (P=0.02).
Conclusions: Patient clinical and demographic characteristics predict choice of initial cART regimen, including NNRTI versus PI, and NRTI backbone use. Advanced analytic techniques are available and needed to account for ARV channeling bias in observational clinical studies.
Observational HIV clinical cohort studies are often used to study long-term outcomes of combination antiretroviral therapy (cART) not easily addressed through randomized controlled trials.
Due to known side effects of antiretrovirals, clinical and socio-demographic patient characteristics may influence initial antiretroviral treatment choices resulting in the channeling of antiretroviral exposure to patients with different characteristics.
Channeling bias may interfere with the causal evaluation of medication effects due to the differential allocation of medications to patient groups with varying risk factors for disease outcomes, especially if factors that influence medication selection are not captured in observation data or accounted for in cohort analyses (StatMed.1991;10:577-581).


Table 1: Clinical and Demographic Characteristics of the Study Population (N=150)


*GFR calculated using the modified diet and renal disease (MDRD) equation: 186 x (Scr)-1.154 x (Age)-0.203 x (0.742 if patient is female) x (1.210 if patient is black)
Figure 1: Initial NNRTI or PI and NRTI backbone (N=150)
Yellow: EFV; light blue: Kaletra; it looks like beige 17 is Reyataz/r & dark blue 16 is unboosted Reyataz; purple other NNRTI/PI. On the right panel- orange 113 FTC/3TC + TDF; green FTC/3TC + ABC; orange FTC/3TC + AZT; purple 5 FTC/3TC + other NRTI.


Table 2. Final Predictive Models of Initial Antiretroviral Choice


This project was supported by The University of North Carolina at Chapel Hill, The Center for AIDS Research, National Institutesof Health funded program P30 AI50410; National Center of Research Resources, National Institutes of Health funded program UL1RR025747, and the North Carolina Translational and Clinical Sciences Institute Large Pilot Grant 3-42036. Emily S. Brouwer is a recipient of an unrestricted fellowship from the UNC-GlaxoSmithKline Center of Excellence in Pharmacoepidemiology and Public Health at the University of North Carolina, Gillings School of Global Public Health.