Measuring HIV Treatment Adherence in Clinical Practice

Measuring HIV Treatment Adherence in Clinical Practice

Adherence to HIV treatment is a key factor in determining whether our patients develop undetectable HIV viremia or drug-resistant HIV. But how do we determine whether patients are taking their treatment as prescribed, and how can we use this information to improve adherence? There are several possible methods, but only few with documented effectiveness.

Drug Levels

As clinicians, we like to rely on laboratory tests to get accurate information about our patients' medical conditions, and so it seems reasonable to suppose that measuring plasma drug levels will help us monitor patients' adherence to their drug regimens. Unfortunately, though, drug level measurements are of little use in assessing adherence to HIV treatments.The protease inhibitors have such short plasma half-lives -- four hours or less -- that measuring their plasma levels only identifies whether patients took their most recent dose. The serum half-lives of nucleoside reverse transcriptase inhibitors such as AZT and 3TC are similarly short. Efavirenz, a new nonnucleoside reverse transcriptase inhibitor likely to be approved by the FDA soon, has a longer half-life: 40 to 55 hours. Although this is long enough for serum concentration measurements to provide useful information about adherence, this has not been tested yet in clinical practice. Furthermore, low serum concentrations would not necessarily indicate poor adherence, as problems with drug absorption, medication interactions, and other drug metabolism issues may also cause low values.

Physician Judgment

Physician judgment is commonly assumed to be an adequate means to determine which patients are adhering to their prescribed drug regimen. But the track record of physicians in making such determinations is poor, even with patients who are well known to them. For example, Gilbert and colleagues studied the ability of family practitioners to distinguish which of their patients were adhering to digoxin treatment, defined as taking more than 80% of prescribed medication doses (Gilbert et al., 1980). Only patients whom physicians felt they knew well were included in the study. The physician assessment of adherence was compared with a gold-standard measure: unannounced pill counts performed by trained research nurses. The physicians' assessments did predict adherence, but in a negative fashion: Patients whom physicians judged to be adherent were actually less likely to be taking their medicines regularly than those judged to be nonadherent. Thirty percent of patients judged to be adherent were taking less than 80% of their medication by pill count, compared to 22% of patients whom physicians judged to be nonadherent. This suggests that without gathering further information, physicians tend to rely on preconceived notions of which patients are adherent -- notions that are frequently misleading.

Patient Self-Report

Clinicians are often skeptical of relying on patients' self-report of missed doses. In part this is with good reason: Patients can be reluctant to admit that they are not taking medication as prescribed. However, when done properly, asking patients about their pill taking is a key way to assess adherence in clinical practice.

How accurate is patient self-report in detecting adherence problems? Sackett and others assessed this in a study that compared patient self-report to unannounced research pill counts (Sackett et al., 1975). Among patients who reported that they were taking less than 80% of their doses, 21 of 25 (95%) were also found to be nonadherent by pill count. Among patients who reported that they were taking over 80% of their medication, only 36 of 105 (34%) were nonadherent by pill count.

The Gilbert and Sackett studies suggest several useful lessons in the clinical assessment of patient adherence. First, patient self-report seems to be considerably more accurate than physician assessment. If we want to know whether our patients are taking their antiretroviral treatments as we prescribe, we should ask them, not simply rely on our judgment. Second, when patients tell us they are missing doses of medication, we should believe them; they are almost always telling the truth. Third, patient self-report tends to overestimate adherence. Nonadherence may still be a significant problem for some patients who report that they are missing doses infrequently.

How We Ask Matters

Exactly how we ask patients about missed doses makes a significant difference in the quality of the information we get back. Questions should be specific, nonjudgmental, and give permission to disclose nonadherence. Physicians frequently frame questions like this: "Are you taking your medicines the way you are supposed to?" This type of question typically fails to elicit accurate information because of the implied message that patients are bad if they aren't taking their medicines the way they "are supposed to." When faced with a general question about missing doses that does not specify a time period, patients also tend to round up to the best-case scenario.

A better way to phrase the question would be this: "I know how challenging it is to remember to take these medicines, so I want to check in about how taking your HIV medicines is going. Yesterday, how many times did you miss taking a dose of any of your HIV medicines? How about the day before that?" This approach includes a preface acknowledging that it is common to miss doses, which can set the patient at ease about admitting this. It also asks about a specific time period, the last two days. This is an effective time interval to query since the accuracy of patient recall of missed doses falls off rapidly after one to two days. For patients who have not missed any doses in the past one or two days, it may be useful to ask how frequently they have missed doses in the past week, or when they last missed a dose.

Regularly asking about adherence will reinforce its importance for patients. Identifying when missed doses have occurred provides an opportunity to help patients develop customized adherence strategies. For example, patients who are missing doses because they are away from home without their medicines may need to develop strategies to help them remember to take scheduled doses with them when they go out. Asking about adherence also can provide opportunities to identify gaps in patient education. For example, patients often believe that they cannot take ritonavir with them when they leave the house because it must be refrigerated at all times. Teaching them that ritonavir can be unrefrigerated for a few hours can make a big difference in preventing missed doses.

Other Methods for Measuring Patient Adherence

Pill bottles can now be equipped with special caps containing an electronic device that records when the bottle is opened (under the assumption that medicine is then taken). Such devices are mainly useful for research studies unless further steps are taken. One example of how these devices are being used in clinical practice is to team an electronic pill bottle cap with a special modem that allows long-distance monitoring of bottle opening. When the electronically capped bottle is placed on the modem device at night, information from the cap device is sent to a central site. The data are reviewed by a trained adherence-support nurse, who can call the patient when doses are missed to troubleshoot adherence problems. A summary of the data is also sent to the physician. A randomized, controlled trial of this system is now being tested with HIV-positive patients. Although the precision of this system is appealing, it will be important to determine whether its expense and complexity are justified by adequate increases in patient adherence. A simpler approach consists of using an electronic pill bottle cap that displays for the patient the last time a dose was taken and sounds an alarm when the next dose is due. An important drawback of even the simplest system is the difficulty of fitting it into a patient's lifestyle. In order to get an accurate record of when doses are taken, patients must take all doses of medication directly from the pill bottle, rather than taking doses out to carry with them. This is an inconvenience for many patients, especially since protease inhibitors typically require large pill bottles.

Pill counts have been used for years to measure patient adherence in research settings. This type of measure is vulnerable to patient manipulation by "pill dumping." In a research setting, this weakness can be at least partially overcome by making unannounced home visits to count pills (with prior patient permission!). This approach is clearly not practical in a clinical setting; instead, patients can be asked to bring in their pill bottles so that the pills remaining since the prescription was filled can be counted. But this approach has several drawbacks too, in addition to the possibility of pill dumping: Patients have to remember to bring their medications to the medical visit, pill counting takes up precious clinical time, and the count can be inaccurate if the patient uses other medication containers to dispense pill doses. Still, in some settings, pill counts can usefully supplement patient self-report.

View this table:
[in a new window]

Table 1. Assessing Adherence: Helpful and Not-So-Helpful Questions

Checking the time interval between prescription refills provides a minimum estimate of nonadherence. Of course, without information about whether the prescription was filled, it cannot be known with certainty whether the medication was actually taken. In addition, making such a check can be cumbersome for a busy clinician. Reviewing prescription refills may be most useful when they are noted in a computerized pharmacy record that is accessible to clinicians, and the patient can use only one pharmacy to fill prescriptions.

The key goal of assessing adherence in a clinical setting is usually to enhance adherence. To the extent that patients are unwilling to acknowledge that they are missing doses, measuring adherence with some of these alternatives to self-report may not attain this goal. Using measures of adherence other than self-report is most useful if they can help a patient accurately identify problems with adherence. Using them to confront patients with the inaccuracy of their self-report can be counterproductive if it turns adherence assessment into an adversarial inquisition rather than a patient-provider partnership aimed at improving patient outcomes.

Summary

Assessing patient adherence to HIV treatments is a key means to enhance patient understanding of the medication regimen, to identify potential obstacles to taking medications as prescribed, and ultimately to prevent virologic breakthrough. Querying patients about their adherence, using well-chosen words, is the most important technique for clinical practice. Asking specific, nonjudgmental questions can maximize our opportunity to get information helpful in enhancing adherence to treatment. Measuring drug levels should be regarded as a supplementary measure for determining whether drug interactions or problems with absorption are contributing to inadequate responses to therapy, since for most antiretroviral medications it provides information only about the most recent dose. Electronic pill monitoring, pill counts, and reviewing prescription refills can be useful adjuncts to patient self-report in specific contexts, but each method has important limitations.

— Frederick M. Hecht, MD

Dr. Hecht is an Associate Editor of ACC.

Published in AIDS Clinical Care August 1, 1998