top of page

Some Patients Know How to Ask. That Is a Design Problem.

Oct 26
5 min read

What cultural health capital says about why better information does not reach everyone equally, and what we are doing about it.

Oct 26, 2026 · @Dan

A 2010 sociology paper describes the room our videos are preparing people for

We build clinician-approved personalized video for patients getting ready for a procedure. It runs across 11 hospitals in Israel, in a pilot at Mayo Clinic Rochester, and it is now rolling out with GI Alliance. Our working assumption has always been that if a patient understands what is about to happen, the care goes better. We still believe that. But a paper complicates it, and we think it is worth reading in public.

In 2010, Janet Shim, a sociologist at UCSF, published Cultural Health Capital: A Theoretical Approach to Understanding Health Care Interactions and the Dynamics of Unequal Treatment in the Journal of Health and Social Behavior. It has been cited more than 280 times, and it introduced a concept the field still uses.

The argument is narrow and uncomfortable. The clinical encounter is not a neutral channel. It rewards patients who arrive already equipped: with medical vocabulary, with the ability to tell a history in a focused way, with the confidence to ask a follow-up question, with some sense of how the system works, and with an interactional style clinicians read as engaged. Those patients get more time, more explanation and more trust. Shim calls that bundle cultural health capital. Her point is that the conversion happens habitually, on both sides, without anyone intending it.


Why this is a problem for anyone selling patient education

If Shim is right, then information is not the whole job. Information is an input that gets converted, and the conversion rate is not the same for everyone.

Follow that through. Give every patient the same excellent explanation. The patient who already had the vocabulary now has a little more. The patient who did not may still not know which part of it entitles them to ask a question, or that asking is permitted at all. Both patients improved. The distance between them did not close, and on a bad day it widened.

That is a real risk and it applies to us. It is also not a new worry in the literature. It is close to what the knowledge gap hypothesis predicted about mass media in 1970, and to what Link and Phelan argued in 1995 about why health inequalities keep reappearing even as the diseases behind them change. Shim's contribution is to move the mechanism indoors. The gap is not only produced by who gets reached. Some of it is produced in the room, in the ninety seconds where a patient either asks or does not.

What our own numbers do and do not tell us

Two things need saying plainly before we go further.

The first is that Shim's paper is not evidence about GistMD. It is a theoretical argument, and she does not test it in that paper. It is not a trial of patient video, it says nothing about digital interventions, and we would be misreading it badly if we presented it as validation of anything we have built. What it gives us is a mechanism to design against, which is a different and more useful thing.

The second is about our own results. In our deployments we have measured a 37 percent reduction in inadequate bowel preparation, a 68 percent reduction in cancellations, and a 55 percent reduction in pre-anesthesia meetings. We are proud of those numbers. They are also averages, and an average is exactly the shape of result that can hide the problem Shim describes. A program can lift the mean while leaving the spread untouched.

We do not currently report whether those gains are evenly distributed across literacy levels, languages and education. That is a gap in our own evidence, not a gap in the literature. We are saying so here because we would rather raise it ourselves than have a reviewer raise it for us.

What it changes in how we build

Read this way, the job is not only to transmit information. It is to send the patient into the room able to use it. That shifts what the content is for, and it has concrete consequences.

It means the end of a video matters more than we used to think. A patient who has just learned what a bowel preparation is may still not know that the cramping they are worried about is a reasonable thing to raise, or that rescheduling is an option rather than a failure. So the close of a video should hand them the specific questions they are entitled to ask, in their own language, phrased the way a confident patient would phrase them. Not a generic prompt to speak up. The actual sentences.

It means vocabulary is part of the intervention, not a readability score to optimize. If the encounter rewards patients who can name things, then naming things is a deliverable.

And it means the delivery channel is doing equity work that is easy to overlook. Video in more than 20 languages, adapted for literacy, sent by SMS or QR with no app to download, reaches people who were never going to extract this from a leaflet in a second language. Shim's framework is a reason to care about that, and we had been treating it mainly as a coverage feature.


How we would actually test it

A framework that cannot be tested is just a nice way of talking, so here is the version we can measure.

The construct is patient-initiated questions: how many a patient asks, unprompted, in the consultation. Shim's argument predicts that this count tracks social position. Our hypothesis is that a video which hands patients the questions raises the count, and raises it most for the patients who started lowest. The outcome we care about is therefore not the average. It is the spread, reported stratified by language, education, and literacy. If our effect turns out to be concentrated among patients who were already fluent, we will have built a very good product that widens a gap, and we would want to know that early rather than late.

Two honest caveats. The framework has been challenged: Anna Schneider-Kamp argued in Social Theory and Health in 2021 that it is tied too tightly to the formal clinical encounter and to cultural capital alone. And it has a blind spot we should not paper over. It treats the consultation as the decisive arena, which means it is close to silent on transport, childcare, and a lost day of wages. No video solves those. A patient who cannot get to the clinic is not helped by knowing what to ask when they arrive.


The question we are putting to our partners

So here is what we are asking the clinicians and researchers we work with.

If a patient education program improves the average and leaves the distribution alone, has it succeeded? We think the answer is no, and we think the field has been slow to ask the question because the average is the number that sells. We would rather find out that we widen a gap and fix it than keep reporting a mean and never look underneath it.

If you run a screening or procedural program and you would be willing to measure patient-initiated questions with us, stratified rather than averaged, we would like to hear from you.


Sources

 
 
 
bottom of page