Petersen ML, Wang Y, van der Laan MJ, et al. Assessing the effectiveness of antiretroviral adherence interventions: using marginal structural models to replicate the findings of randomized controlled trials

Division of Biostatistics, University of California at Berkeley, Berkeley, CA 94720-7360, USA.
JAIDS Journal of Acquired Immune Deficiency Syndromes (Impact Factor: 4.56). 01/2007; 43 Suppl 1:S96-S103. DOI: 10.1097/01.qai.0000248344.95135.8d
Source: PubMed


Randomized controlled trials of interventions to improve adherence to antiretroviral medications are not always feasible. Marginal structural models (MSM) are a statistical methodology that aims to replicate the findings of randomized controlled trials using observational data. Under the assumption of no unmeasured confounders, 3 MSM estimators are available to estimate the causal effect of an intervention. Two of these estimators, G-computation and inverse probability of treatment weighted (IPTW), can be implemented using standard software. G-computation relies on fitting a multivariable regression of adherence on the intervention and confounders. Thus, it is related to the standard multivariable regression approach to estimating causal effects. In contrast, IPTW relies on fitting a multivariable logistic regression of the intervention on confounders. This article reviews the implementation of these methods, the assumptions underlying them, and interpretation of results. Findings are illustrated with a theoretic data example in which MSM are used to estimate the effect of a behavioral intervention on adherence to antiretroviral therapy.

Download full-text


Available from: Maya Liv Petersen,
  • Source
    • "This method has a close association with concurrent viral load [2], electronic pill cap adherence assessment [20], development of resistance [21], [22], and progression to AIDS [23]. We used marginal structural models [24]–[26] and targeted maximum likelihood estimation [27] to adjust for potential confounders of adherence and viral suppression. "
    [Show abstract] [Hide abstract]
    ABSTRACT: We hypothesized that the percent adherence to antiretroviral therapy necessary to maintain HIV suppression would decrease with longer duration of viral suppression. Eligible participants were identified from the REACH cohort of marginally housed HIV infected adults in San Francisco. Adherence to antiretroviral therapy was measured through pill counts obtained at unannounced visits by research staff to each participant's usual place of residence. Marginal structural models and targeted maximum likelihood estimation methodologies were used to determine the effect of adherence to antiretroviral therapy on the probability of virologic failure during early and late viral suppression. A total of 221 subjects were studied (median age 44.1 years; median CD4+ T cell nadir 206 cells/mm(3)). Most subjects were taking the following types of antiretroviral regimens: non-nucleoside reverse transcriptase inhibitor based (37%), ritonavir boosted protease inhibitor based (28%), or unboosted protease inhibitor based (25%). Comparing the probability of failure just after achieving suppression vs. after 12 consecutive months of suppression, there was a statistically significant decrease in the probability of virologic failure for each range of adherence proportions we considered, as long as adherence was greater than 50%. The estimated risk difference, comparing the probability of virologic failure after 1 month vs. after 12 months of continuous viral suppression was 0.47 (95% CI 0.23-0.63) at 50-74% adherence, 0.29 (CI 0.03-0.50) at 75-89% adherence, and 0.36 (CI 0.23-0.48) at 90-100% adherence. The risk of virologic failure for adherence greater than 50% declines with longer duration of continuous suppression. While high adherence is required to maximize the probability of durable viral suppression, the range of adherence capable of sustaining viral suppression is wider after prolonged periods of viral suppression.
    PLoS ONE 09/2009; 4(9):e7196. DOI:10.1371/journal.pone.0007196 · 3.23 Impact Factor
  • Source
    • "This was consistent with our previous analysis of the same and other data sets [20]. Actually, AEs were a rather weak predictor that would have been eliminated had we used aggressive search algorithms with cross-validation or other model selection tools that target prediction error (as suggested in [21,22]). However, we left it in the model as we would rather overfit than underfit the model. "
    [Show abstract] [Hide abstract]
    ABSTRACT: The true dose effect in flexible-dose clinical trials may be obscured and even reversed because dose and outcome are related. To evaluate dose effect in response on primary efficacy scales from 2 randomized, double-blind, flexible-dose trials of patients with bipolar mania who received olanzapine (N = 234, 5-20 mg/day), or patients with schizophrenia who received olanzapine (N = 172, 10-20 mg/day), we used marginal structural models, inverse probability of treatment weighting (MSM, IPTW) methodology. Dose profiles for mean changes from baseline were evaluated using weighted MSM with a repeated measures model. To adjust for selection bias due to non-random dose assignment and dropouts, patient-specific time-dependent weights were determined as products of (i) stable weights based on inverse probability of receiving the sequence of dose assignments that was actually received by a patient up to given time multiplied by (ii) stable weights based on inverse probability of patient remaining on treatment by that time. Results were compared with those by unweighted analyses. While the observed difference in efficacy scores for dose groups for the unweighted analysis strongly favored lower doses, the weighted analyses showed no strong dose effects and, in some cases, reversed the apparent "negative dose effect." While naïve comparison of groups by last or modal dose in a flexible-dose trial may result in severely biased efficacy analyses, the MSM with IPTW estimators approach may be a valuable method of removing these biases and evaluating potential dose effect, which may prove useful for planning confirmatory trials.
    BMC Psychiatry 02/2008; 8(1):3. DOI:10.1186/1471-244X-8-3 · 2.21 Impact Factor
  • Source
    • "For example , randomized controlled trials have provided evidence that behavioral interventions improve adherence to HAART . Such interventions are increasingly considered the standard of care , making additional randomized trials less likely ( Petersen , Wang , van der Laan , & Bangsberg , 2006 ) . "
    [Show abstract] [Hide abstract]
    ABSTRACT: Background: Adherence to antiretroviral (ARV) medications in excess of 90-95% is necessary for optimal response to suppress HIV replication and to maintain and/or restore immune function. A number of interventions have been shown to improve ARV adherence, but no research has been conducted which evaluates proactive monitoring of pharmacy refill adherence and subsequent intervention when inadequate adherence is identified. Purpose: The purpose of this project was to compare treatment response, pharmacy refill adherence and self-reported medication adherence between two groups of patients: those participating in an AIDS Drug Assistance Program (ADAP) and those participating in a Medicaid-funded medication access program. The ADAP served as a structured adherence intervention (SAI) based on procedural and administrative processes required by the state-managed program Additionally, covariates that can impact adherence were studied including utilization of adherence services and interventions and factors related to HIV disease, antiretroviral agents and sociodemographic factors. Method: This retrospective comparative study examined secondary data to assess 424 patients who received clinical and pharmacy services at one treatment site in 2005. Analysis: Logistic regression was performed to test the effects of the SAI on treatment response (CD4 and HIV RNA response), self-reported adherence, and pharmacy refill adherence while controlling for the covariates. Results: Patients participating in the SAI demonstrated higher levels of both self-reported and pharmacy refill adherence compared to patients receiving usual care. Although patients participating in the SAI program demonstrated better virologic (HIV RNA) responses to HAART compared to patients receiving usual care, immunologic (CD4 lymphocyte) responses to HAART were not significantly different compared to subjects in the usual care program. Conclusion/Discussion: This study provides information on the effects of a structured programmatic intervention on medication adherence and response to treatment and will be used to inform policy decision making at the local and State level.
Show more