What the work has looked like.

Half the patients left in the first week

A twelve-week coaching program supported patients starting one of two GLP-1 medications for type 2 diabetes. Close to half of them left before week two, a critical window for medication adherence. The cause sat upstream of the coaching. Dose escalation had not been explained to these patients, so they did not know their dose could change. They also did not know that side effects could get worse during escalation. Many stopped the medication on their own.

I worked on the teaching modules built to close that gap: a separate set for each medication, placed at the front of the program where the drop-off was happening, so each group saw only what applied to them. Getting them approved meant medical-legal-regulatory review. The modules carried the medication’s branding; the coaching program around them did not. No one had put the two together that way before, so the review had no precedent to work from.

Read the full case study

Dozens of country programs, one curriculum

A global drug company ran a separate patient support program for insulin therapy in every country it worked in. Nothing was shared between them. The task was one core curriculum that each market could adapt instead of rebuild.

I worked on the evidence base it was built from, and on the structure. It was layered, so a market could change what it had to change and leave the rest alone. It went through regulatory review in several countries.

Read the full case study

Why some people get the care they need and others do not

My dissertation research asked the question: what factors separate people who get the mental health care they need from people who do not? The people in the study were adults who had lived through a suicide attempt. This is an under-studied group, and the pathways toward or away from care were complex. I set out to map the combinations of conditions that could lead to met or unmet mental health care needs.

The study used two kinds of data: surveys from 74 people, read with fuzzy-set qualitative comparative analysis, and open-ended survey responses alongside 21 in-depth interviews, analyzed for themes. Fuzzy-set analysis looks for combinations of conditions that lead to an outcome. It does not ask what one variable does on average. That is the right tool when no single factor explains much by itself.

That is what the data showed. Stigma, trust in providers, insurance, and income worked together in a variety of common combinations, not separately. The same person may move in and out of good care as those conditions shift. The study also found a middle group that does not sit squarely at either end of a met-or-unmet needs measure: people whose needs were partly met. Many people sit in that gap.

This type of academic study is far from the style of patient-facing materials I create, but it is a good example of how I work. Look at conditions in combination, and stay open to hearing from patients to find out what they need and where systems may fail them. Take the gray areas seriously. Choose methods and measures matched to the question, which can catch what a simpler design would hide.

Mixed methods · fsQCA · 74 surveys, 21 interviews