One or Two Weeks of Dosing Data is All You Need for an Optimized Patient Pool

By Rich Christie, MD PhD

Medication adherence can make or break a clinical trial. A lot of effort is spent by site teams engaging with patients to get them to take their medications regularly and correctly. This can be a real burden for sites, bogging them down and making everything go slower. 

But what if you could identify those patients most likely to show high adherence and focus your enrollment on that population? 

You can – and it will only take a week or two.

The Case for a Lead-in Period

Dosing is a behavior, and behaviors are predictable. Taking this thought, AiCure tested to see if their Patient Connect platform could be used in a special lead-in period to accurately predict dosing behavior over the course of an entire study. We wanted to see if remote, real-time measurement of medication dosing, done via a smart device, could tell us which patients would be adherent and which ones wouldn’t. 

To prove the ability of the technology to predict adherence, we first needed to define the target. Clinical standards for research recommend adherence rates of 80% or better in order to adequately demonstrate treatment effects. Based on this, we built classification algorithms for prediction of adherence above or below that 80% threshold.

We selected a pool of patient data, representing 4,182 patients, that ranged widely in terms of demographics, disease state and other factors. We went to work looking at the first week and then the first two weeks of dosing data for these patients. We wanted to compare dosing behaviors in these short timeframes to dosing behaviors in full-length studies over time. By comparing the two, we examined the ability of the data from just one or two weeks of adherence to predict overall dosing behaviors. 

What we found was that the first week of dosing data accurately predicted beginning-to-end adherence at a rate of 72.2%. After two weeks, that accuracy rose to 76.6%. This means that roughly 75% of the time, the one or two-week timeframes were able to accurately identify patients that would – and would not – meet or exceed that 80% rate of medication adherence.

Taking all of this into account, we can apply these findings in new trials by adding a one- or two-week lead-in period to help optimize the patient pool. For a short amount of time, we can ask patients to take part in a pre-trial placebo test period where we ask them to use the remote functionality of the Patient Connect app to record dosing. The adherence shown during this period will help predict which patients will adhere at a high rate, and thus which patients should be enrolled in the full study.

The Power of an Optimized Patient Pool

Optimizing your patient pool for maximum medication adherence, now that you know it’s possible, can have a game-changing impact on your research. Patient pools with 80% or higher adherence mean:

  • Less burden on sites – Higher performing patients require less follow-up and proactive engagement from site-based clinicians, allowing those clinicians to be more efficient
  • Smaller – Patients with typically low rates of adherence are also the patients most likely to drop out of studies. Using data from the lead-in period, you can avoid enrolling these patients to begin with, reducing the need to “over-enroll” as a contingency for dropouts. 
  • Faster trials – Studies that require fewer patients to produce statistically significant, quality data are usually faster studies.
  • Better data – Very simply, the higher rate of medication adherence, the more complete and accurate the data becomes.

By just adding one or two weeks, pre-trial, to monitor and evaluate the medication adherence of prospective patients, you can use AiCure’s technology to accurately predict overall adherence. With this information, you can identify and enroll the patients that will take their medications reliably and as directed. High rates of reliable dosing mean more efficient studies, reduced burden on trial sites and accurate, high-quality data. For more information about AiCure Patient Connect, visit