Overview of Electronically Measured Adherence Data
All electronic adherence measurement solutions record the date and time a medication event occurs.
A single medication event or absence of a medication event within a 24 hour period (missed dose) can be important data point, but the real value of adherence data comes from sequencing individual data points into a time series of medication events that can highlight specific drug dosing problems. Independent of the measurement technique, the measured adherence data that is collected has the same properties and can be analyzed and visualized the same way.
Typical drug dosing problems that patients frequently have in clinical trials and in clinical practice can clearly be seen with adherence data. These include:
- Variable or Sporadic timing of doses
- Variable or Sporadic time between doses (too close; too far apart)
- Frequent short duration Drug Holidays (2-3 consecutive days of missed doses)
- Extended Drug Holiday (>3 consecutive days of missed doses)
- Early Discontinuation
- Problem with Evening Dose
- Day of the Week Effect
Adherence Data Plots of Patient Dosing Histories
Most Adherence Measurement solution providers have the ability to produce data plots for their customers. In the examples provided below, the data plots are defined by the following parameters.
- Horizontal or X-Axis is defined by the duration of measurement time (months, years)
- Vertical or Y-Axis is defined by the 24 hour period of measurement each day
- Blue or Green Dots represent a dose taken as defined by the actual measurement technique of the solution used
- Grey Vertical bars or lack of colored Dot signify that there was no dose detected within that 24 hour period
Examples of drug dosing histories that highlight common patient dosing errors
Adherent to prescribed dosing regimen (Data courtesy of CMT Cares)
The patient below is from a Phase IV study where the patient was asked to take a QD dose between 9 and 11 am. The patient misses 8 doses over the measurement period but never has an episode of consecutive days missed.
Variable time of doses and Variable time between doses (Data courtesy of IMC)
The patient history below is from a Phase III clinical trial that required an exacting and highly complicated dosing regimen of 4 different dosing times per day as well as a “4 days on” -“3 days off” dosing schedule. The four dosing windows are depicted on the graph by the four horizontal gray shaded areas. Doses taken within the prescribed time window are green dots and outside the window by red dots. A perfectly adherence patient would be identified as having 4 green dots each day. The patient below had two different dosing problems:
- Variable time of doses can be seen by the inconsistent pattern of doses when viewed horizontally. Ideally, one would want to see 4 distinct horizontal patterns of dots. The morning dose taken around 8 am each day is a pattern you would like to see in the other three doses taken later in the day.
- Variable time between doses can be seen when looking at the data vertically. For example, on Nov 6 (depicted by blue arrows), you see 12 hours between doses 1&2 and only 1 hour between doses 2&3.
Worsening Dosing Regimen (Data courtesy of AARDEX)
The patient history below is from a Phase III study where the patient is asked to take a QD dose for 12 weeks. Over the course of the 12 week period, it is clear that the patient’s dosing behavior worsens as they go from missing 2 doses in the first 6 weeks of treatment to missing 9 doses in the last 6 weeks. Identifying behavioral trends such as worsening regimen has proven to be a key indicator that a patient is getting to ready to discontinue treatment or drip out of a clinical study.
Data of specific dosing times and patterns of dosing behavior provide significant utility for:
- Patient Adherence Interventions Programs
- “Quick Start” Adherence Programs
- Predictive Modeling of dosing patterns for patient discontinuation
- Population Triage for critical dosing errors
- In Clinical Trials and limited distribution Specialty Rx medications
- PK/PD modeling
- Decentralized Clinical Trial protocols requiring remote patient monitoring
- Pharmacovigilance
- Measurement inclusion with Patient Registry participants
- Adjudication of Value Based Risk Sharing Agreements
The downside of using pill counts as a measure of drug exposure in clinical trials – It’s not just about how many doses were taken, it’s about how and when those doses were taken.
Bernard Vrijens, who is the COO of AARDEX Group and a Professor of Biostatistics at the University of Liège (Belgium) is the Leading Researcher and Expert in the analysis of electronically measured adherence data and interpretation of patient drug dosing histories. Over the last few years, he has used the following data to visually highlight the downside of using pill count to estimate patient adherence and drug exposure in clinical trials.
Below are 4 patient histories from a Phase III HIV study. All 4 of the patients took 75% of prescribed doses during a 3-month period. Statistically, these 4 patients would be grouped together if the study used pill-count to measure drug exposure. However, using an electronic measurement method that provides a rich and reliable sample of data, you can detect that each patient took their 75% of the medicine very differently and correspondingly, their exposure to the study drug over the 3-month period would be different. One would expect that the health outcome for each patient would be different as well. Since this was an infectious disease trial, the drug holiday detected in the lower left-hand patient history could have led to drug resistance and possible failed outcome. (Data courtesy of AARDEX Group)
