AIDS and Behavior

, 10:249

Monitoring Adherence to HIV Antiretroviral Therapy in Routine Clinical Practice: The Past, the Present, and the Future

Authors

Original Paper

DOI: 10.1007/s10461-006-9121-7

Cite this article as:
Bangsberg, D.R. AIDS Behav (2006) 10: 249. doi:10.1007/s10461-006-9121-7

Keywords

HIVadherenceresistanceelectronic medication monitorCASImedisetpill box organizer

Adherence to antiretroviral therapy is the strongest predictor of viral suppression, drug resistance, disease progression and death in HIV infected individuals (Bangsberg et al., 2000, 2003; Bangsberg, Hecht et al., 2001; Bangsberg, Perry et al., 2001; de Olalla et al., 2002; Hogg et al., 2002; Paterson et al., 2000). There is, however, no standard approach to adherence assessment in routine clinical practice. This situation is analogous to the general internist managing hypertension with out a blood pressure cuff or the critical care specialist managing a ventilator without measures of oxygenation. Worse yet, providers in routine clinical practice have rarely predicted adherence better than random (Bangsberg, Hecht et al., 2001; Bangsberg, Perry et al., 2001; Gross, Bilker, Friedman, Coyne, and Strom, 2002) which means that we are leaving the most critical determinant of HIV treatment outcomes to chance.

In this edition of AIDS and Behavior, Simoni and colleagues conducted a systematic review of patient reported measures of adherence (Simoni et al., 2006). Simoni et al. determined that patient reported measures of adherence are associated with objective adherence measures, viral suppression, and CD4 cell response. Questioning patients about adherence over longer periods tended to detect more non-adherence and had a slightly greater association with viral load, although the differences were not significant. Simoni et al. conclude that “physicians and researchers may proceed with the use of self-reported measures with some confidence in their validity at least in terms of their association with other indirect measures of adherence and viral load.”

So is this the end of the story? Do we simply ask our patients about missed doses over the last 30 days in a non-judgmental manner at each clinic visit in order to assess adherence and, hopefully, improve adherence with the goal of optimizing virologic, immunologic, and clinical response? As Simoni et al. indicate, self report has several advantages, including: low cost, minimal participant burden, ease of administration, flexibility in term of mode of administration and timing of assessment, and the potential to yield specific information regarding dosing behavior such as timing with food. In spite of these advantages, there are also many limitations. While self reported adherence is significantly associated with other measures, it remains a coarse measure of adherence and suffers from a ceiling effect; specifically it does not distinguish individuals with moderate, high, and prefect adherence. There are strategies to reduce this ceiling effect. These include asking about adherence over several intervals (Mannheimer, Friedland, Matts, Child, and Chesney, 2002), asking about the last time a does was missed, or using private computer assisted interviewing devices where the patient discloses nonadherence to a computer program rather than an individual which may inhibit disclosure of socially sensitive information (Bangsberg, Bronstone, Chesney, and Hecht, 2002). Pharmacy refill data may also provide important information during clinical visits which can supplement patient interview and improve adherence assessment (Gross, Zhang, and Grossberg, 2005).

The challenge to improve the precision of adherence assessment is daunting. The rapid evolution of HIV antiretroviral regimens is changing adherence goals. While Paterson et al. (2000) found that levels of electronically monitored medication adherence of less than 95% was associated with incomplete viral suppression in the majority of individuals, this study was conducted on partially suppressive regimens that are no longer in use. More potent antiretroviral therapy including nonnucleoside reverse transcriptase therapy and ritonavir boosted protease inhibitor therapy are capable of reliably suppressing viral load in patients at moderate levels of adherence (King, Brun, and Kempf, 2005; Maggiolo et al., 2005; Weiser, Guzman, Riley, Clark, and Bangsberg, 2004). Furthermore, the ability of resistant virus to replicate in difference levels of drug lead to resistance within different adherence windows for each regimen. For single protease inhibitors, the window is 80–95% adherence, for non-nucleoside reverse transcriptase inhibitors, the window is 2–70% adherence, and the window has yet to be defined for ritonavir-boosted protease inhibitors, entry inhibitors, or future regimens (Bangsberg et al., 2005). Patterns of adherence may be more critical than overall level of adherence. Patient reported treatment discontinuation of more than 48 h is an independent risk factor for non-nucleoside reverse transcriptase inhibitor resistance, even controlling for average adherence over time (Parienti et al., 2004; Spacek et al., 2006). As a result, rapid changes in antiretroviral therapy will lead to ever changing adherence goals. The ultimate goal of adherence measurement is to detect specific levels and specific patterns of adherence that put a patient's particular regimen at risk for resistance. Current self-reported measures may not live up to this goal.

How do we detect levels and patterns of adherence prior to the development of resistance? One such possibility is to further develop the work of Simoni and other investigators to identify specific questions that are both sensitive and specific to patterns of adherence that put patients at risk for resistance. Such questions could be regularly administered to patients via web-based systems in order to detect resistance before it happens (Bangsberg et al., 2002). Alternatively, systems that electronically monitor patient adherence with real-time reporting may prove effective. Historically, the electronic medication monitor has not penetrated the routine clinical practice setting because it only monitors one medication at a time and precludes the use of pill box organizers that may improve adherence (Bova et al., 2005). More recent systems, however, have incorporated electronic monitoring within a pill box design. Such systems monitor individuals medication adherence in real time and can calculate the risk of virologic failure which is communicated via an internet connection to a clinical care team that is capable of intervening on adherence before the development of resistance. While such systems have been promising in congestive heart failure and schizophrenia (Artinian et al., 2003; Ruskin et al., 2003), their full utility for HIV antiretroviral therapy remain untested.

While continuous and objective adherence monitoring systems are promising, they will never replace health care providers interviewing patients about their experience with taking medication. Patient understanding how their medication impacts their health, communicating side effects, and establishing trust in both patients and therapeutic strategy are essential parts of medical care. Just as physical exam skills have declined with advances in imaging systems (Jauhar, 2006), our interest in talking with patients about their treatment strategies should not diminish with more accurate technologies to monitor adherence.

Copyright information

© Springer Science+Business Media, Inc. 2006