E lectronic devices for measuring medication-taking have been available for decades, and have demonstrated substantial variability in adherence behavior. 1 These devices are integrated into medication packaging, and record the date and time when a pill container is opened or an inhaled medication dose is dispensed. Electronic adherence monitors have been used in many research studies 2 but remain uncommon in clinical practice. While the lack of widespread adoption of such devices has in part been due to persistent concerns about cost and technical reliability, a more formidable barrier to clinical implementation has been the lack of answers to several fundamental questions. These questions include: how to select medications for monitoring, how to present adherence information to clinicians, which clinicians should receive the information, when they should receive it, and how the information should be integrated into decision-making.A study in this issue of JGIM sheds light on these questions. Kronish and colleagues report a cluster-randomized trial of an intervention that provided clinicians with data from electronic pillbox monitors. 3 The study took place in two hospital-based primary care practices serving vulnerable populations in New York City. Twenty-four clinicians were randomized either to continue usual care or to receive information about recent adherence with antihypertensive medications. The information was presented at a single clinic visit along with a brief report suggesting responses to that information. Each clinician in the intervention group received information from a median of four patients, all of whom met JNC-7 criteria for uncontrolled hypertension on two consecutive clinic visits. 4 These patients were complex, both socioeconomically and medically. Most were Hispanic (75 %), and 83 % received Medicaid or were uninsured. More than half the participants were taking three or more antihypertensive medications, and the clinician addressed a mean of four other health problems at the same clinic visit. Adherence monitoring demonstrated that 41 % of participants had taken < 80 % of their antihypertensive medications over roughly the preceding 2 months. Although the evidence for using an 80 % threshold to distinguish adherence from non-adherence is scanty, at least one recent paper suggests that this threshold identifies individuals at higher risk for adverse outcomes after myocardial infarction. 5 Identification of non-adherence is an important Bbranch point^in clinician decision making. For individuals whose hypertension is uncontrolled despite high adherence, clinicians should consider intensifying treatment by increasing medication doses or adding new medications. For individuals who are not adherent, clinicians should identify and remove adherence barriers and provide counseling to promote behavior change, since intensification of treatment is likely to be futile or counterproductive. 6 Accordingly, these researchers designated clinician behavior in response to adherence information as their...