Adherence data needs to be collected in real time, before the damage is done

Innovative methods to assess a person’s adherence to treatment are being developed which may enable clinicians to intervene and offer support before virological failure occurs, Jessica Haberer told the Fifth International Conference on HIV Treatment Adherence in Miami this week.

The need for real-time data is clear: the current methods for adherence monitoring detect missed doses several weeks to months after they occur, but viral replication is likely to commence within 2 days of a break in treatment, and viral rebound occurs in half the people who miss treatment for 15 days.

While current studies are generally at a pilot or feasibility stage, to assess the validity of new methods, Haberer stressed that if these approaches could be shown to be valid, they would be of practical use outside a research setting. They would enable clinicians to provide adherence support to those who needed it at the time, in order to prevent drug resistance and treatment failure occuring.

Some of the most common currently used methods to assess patients’ adherence are:

• self-report (cheap and simple, but usually over-estimates adherence),
• pill counts of the number left in the container,
• monitoring of the frequency of pharmacy refills,
• devices such as a MEMS cap which record the date and time that a pill container is opened (expensive but quite accurate), and
• monitoring of viral load (the only method which gives an indication that a drug has actually been swallowed, but expensive and viral load may vary for other reasons).

However, for all these methods, missed doses can only be identified retrospectively.

On Tuesday, Haberer presented findings from a pilot study on the use of a device named Wisepill in rural Uganda. This is a somewhat similar to a MEMScap in that it registers the date and time that the pill container was opened; however, it uses wireless technology to sends this information over a mobile phone network to a clinician or researcher. If the network is not available at that time, the message will be stored and sent later.

The pilot study assessed the effectiveness and acceptability of the device with ten HIV-positive people in rural Uganda (a country with 90% mobile phone coverage). Most participants were women, had undetectable viral load and were taking a single fixed-dose pill twice a day.

In the first three months, there were a number of technical problems, largely battery failures and problems with signal transmission. However, these were resolved and in the second three-month period there was only one technical failure.

According to Wisepill, the mean adherence level was 90%. This compared to a level of 100% by self-report and 99% by unannounced pill count. Prior to the study, adherence measured by MEMScaps was 87%. Although there is some discrepancy in these results, the researchers consider the results to be comparable enough to continue testing the device in further studies.

The technology was well-received by the individuals in the study. All described it as easy or convenient to use. Many appreciated the fact that their health was being monitored, rather than seeing it as an intrusion of their privacy.

Participants appreciated the fact that the device did not look like a pill box, which is associated with HIV in this community. In this small study, no participants had their HIV status accidentally disclosed through use of the device.

Wisepill is one of a large number of similar devices which are currently in development, and which Haberer outlined at another talk the previous day. While most take the form of a pill container or pill box, some are in the more portable form of electronic blister packs.

These second-generation electronic monitoring devices have a number of potential advantages, Haberer said. As well as collecting real-time data, they show the time at which a drug is taken (not just the total number of doses). Moreover, the device itself can help remind people to take their pills as well as recording whether they have done so – it may take the form of a pill box, incorporate a flashing display or an alarm, or send a reminder text message to a mobile phone.

However, they are expensive (prototypes are $100 to $200), can only take a few weeks of medication at a time, and many are relatively bulky. A person may choose to remove pills to put them somewhere more convenient, distorting the information collected.

In fact, some researchers worry that the problem with almost all methods is that they don’t indicate whether someone has actually swallowed the drug. In some settings, there is so much social pressure on patients to take their drugs, that some people will apparently open containers and throw drugs away to fool doctors.

As a result, a number of emerging high-tech methods are being developed for clinicians and researchers to see whether a drug has actually been swallowed. These technologies (such as ID-Cap, MagneTrace, ChipSkin and Smart System) all combine a drug with an apparently harmless antenna, magnet or microchip which will send a signal when it is swallowed. Discussing these, Haberer did ask the questions of whether researchers are going too far in their efforts to monitor adherence and if such methods would be acceptable to people taking treatment.

However, Haberer does believe that real-time, accurate, adherence data could be of benefit in clinical practice. While many of these methods are expensive, if any go into production, costs will go down. Moreover, if clinicians in resource-limited settings had accurate adherence data, they could spend less on laboratory tests, while identifying problems before treatment regimes fail and symptoms occur.