Patient education as an adherence intervention

A pharmaceutical company ran a twelve-week coaching program for patients starting one of its two GLP-1 medications for type 2 diabetes, delivered by a digital health coaching company. Each patient worked with a health coach on a weekly call, alongside a digital experience of articles, videos, and check-in questions through a web and mobile app.

The program was unbranded by design. That gave the client room to use it with more than one medication population, and it kept the program outside the stricter medical-legal-regulatory review that branded patient-facing content has to clear, which allowed a wider range of topics in the education. The tradeoff was that the content could not speak to a specific medication, so it could not answer the questions patients had in their first weeks, which were about what was happening in their bodies and what to expect next.

About 45% of enrolled patients left during the first week, and another 25% left during the second. The program was losing most of its patients before they reached the content built to support them, though the patients who stayed rated it well; its Net Promoter Score sat around 70.

Coaching observations, patient interviews run by the quality assurance team, and a review of the clinical literature pointed to two causes, both of which sat upstream of the program. Dose escalation had not been explained to these patients, so they did not know their dose was meant to change over time, or that a different dose might arrive in a different delivery format; when a change came, it read as an error rather than as the treatment working as designed. They also did not know that gastrointestinal side effects are common in the early weeks of GLP-1 therapy and during escalation, or that those effects usually ease as the body adjusts, so many stopped the medication on their own without talking to their care team.

Two branded education modules, replacing the program’s first and second weeks, which is where patients were leaving. I built a separate set for each of the two medications, and the two were never shown together, so a patient only saw what applied to their own treatment.

Everything had to work inside pharmaceutical content rules, which meant no direct medical recommendations, nothing that could read as telling patients to change what they were doing with their medication, and adverse events flagged and routed rather than softened or left out. Beyond the standard rules, branded content had not been placed inside an unbranded program this way before, so the review board had no template for what review should look like and no settled view of where the line sat.

The client’s own patient-facing website had already cleared review, which set the boundary: content that stayed close to what the client had published was safe ground, and anything past it drew scrutiny. The productive approach was to restructure and reformat guidance that already existed rather than write new clinical material. Some of the inclusions proposed for the modules were unlikely to clear review in this format, and part of the work was steering the scope, respectfully, back to what could realistically be approved.

Approval took sixteen months, and most of that time was neither writing nor a matter of my team’s pace. It was the back-and-forth between what the client wanted to include and what the review board would accept in a structure none of us had a precedent for. The scope that cleared in the end was close to what I had recommended at the start.

The modules cleared review and launched, which set a precedent for this kind of structure. Coaches reported more engaged conversations in the first weeks, patient feedback was positive, and the client was discussing updates to the modules for upcoming product changes. There is no post-launch quantitative data, because both companies went through organizational changes shortly after launch.

The attrition came from a knowledge gap that existed before anyone enrolled, and patient education was the intervention built to close it; the early signals were positive. These patients were prescribed a medication with a well-documented escalation process and well-documented early side effects, and they arrived prepared for neither, which creates avoidable attrition for any program trying to support adherence downstream.