Adherence to HAART can be measured using a range of methods. This section considers data on their validation, and reviews the strengths and weaknesses of each approach. Particular emphasis is placed on the role of measurement tools as adjuncts to the partnership between patient and health care providers, rather than as tools for catching out ‘disobedience’ on the part of patients. There may be different motivations behind the monitoring of adherence within clinical trials and in routine practice; tools for routine use must be practical.
The lack of both a standard approach to measuring adherence, and the lack of consensus over what constitutes a clinically effective level of adherence, is currently an important challenge for the field. For example, while trials may assess adherence by presenting a measure of the number of doses missed, few include an assessment of whether doses were taken on time or with proper regard to dietary restrictions.
Self-report: Due to the ease with which it may be included in routine clinical practice, patient self-report of adherence is a commonly used measurement. Though it is a good marker, it is not perfect. There is evidence that self-report may over-estimate adherence to antiretroviral therapy compared to other methods102, and the dynamics of provider-patient relationships may clearly impact on the willingness of individuals to disclose problems, particularly in face-to-face settings. Nevertheless, the use of self-complete questionnaires has been validated as a highly accurate means of measuring adherence, and shown to predict virological response when involving both seven day recall103, as well as either two week or one month recall104. Moreover, this method may be important in reinforcing the central role which patients themselves have in managing their own adherence, in comparison to provider-controlled methods which have the potential to undermine this.
Provider estimates: Health care provider estimates of adherence to antiretroviral therapy have been demonstrated to be poor105 106, and their use as a method of measuring adherence is therefore inadvisable.
Drug level monitoring: Low plasma drug levels have been associated with treatment failure107 108, and the results of drug level tests have been proposed as a means of monitoring adherence levels in both HIV-infected adults109 and children110. Their use presents a number of problems however111, including cost, the difficulty in measuring NRTI levels, and the possibility that a one-off sample may not reflect true adherence levels over the longer-term. Other laboratory markers have been employed historically to measure adherence, such as mean corpuscular volume, which is raised by AZT and d4T112.
MEMScap: MEMS (Medication Event Monitoring System) is an electronic device fitted to pill bottle containers which records the removal of the cap, and therefore infers the removal of pills. This method is frequently used to measure adherence in research settings, and has been demonstrated to predict virological response to HAART113. However, MEMS may not be compatible with individual adherence strategies, such as decanting pills on a daily or weekly basis114, and cannot be used with blister packs.
Pill counts: pharmacy records: Pill counts, where patients are asked to attend clinic with their medication which is then counted by a health care provider, have been found to predict response to HAART115. They are time-consuming to perform however, and rely on patients to bring their medication with them. In common with other methods, they are vulnerable to fabrication and may be seen by patients as an unwelcome attempt by health care providers to ‘police’ their adherence. Review of pharmacy dispensing records may be helpful in identifying overdue prescriptions.
Pills identification test: A novel method of detecting low adherence has been reported by Parienti and colleagues116. The Pills Identification Test (PIT) involves inviting patients to distinguish the pills in their regimen from a display of antiretrovirals, including two ‘twin pills’ which are similar but not identical to their own. PIT scores were shown to correlate with a validated self-report adherence measure.
Surrogate markers: The results of viral load and CD4 monitoring tests are commonly used as surrogates for adherence. In the Trilege antiretroviral induction-maintenance study, individuals with virological failure on a PI-containing regimen had low PI blood levels, low adherence level by pill count, and an absence of genotypic resistance to PIs, suggesting their treatment failure had been caused by missed doses117. ACTG 343 similarly found an absence of PI resistance in people with viral load rebound on PI-based HAART118. However, it is important to note that although low adherence is frequently associated with viral rebound, it may also be observed in individuals with suppressed viral load119. Conversely, treatment failure may occur for reasons other than low adherence. Mannheimer and colleagues reported that CD4 count may be a less sensitive marker of adherence than viral load120.