Do You Know What Messages Your Patient Stories Are Really Sending?

by Brian Zikmund-Fisher on January 9, 2013

Including patient stories in health education materials and patient decision aids might seem like an obvious choice, for many reasons:

  • They are vivid, engaging readers or viewers with the material.
  • They have inherent credibility when told by people who have “been there.”
  • They can provide useful information about health decisions in less boring ways than traditional formats.
  • They can show diversity, either in terms of patient demographics (i.e., “somebody who looks like me”) and/or values (“somebody who thinks like I do”).

Seems like a win-win, right?

Well, sometimes.

On the one hand, it is definitely true that narrative forms of communication are very powerful. Chip and Dan Heath’s bestselling book, Made to Stick: Why Some Ideas Survive and Others Die, makes a convincing argument that good stories embody all of the elements that make information memorable and hence impactful. Good stories are simple in message, concrete in details, often unexpected and emotional and delivered by a credible speaker.

On the other hand, past work by Ubel, Jepson and Baron as well as others has shown that patient stories have the potential to bias decisions by changing how people perceive risk. Specifically, individual narratives have the potential to make rare outcomes appear equally likely as common ones. If there is one story that turns out well and another that turns out poorly, people tend to see those outcomes as equally likely even when numerical risk information is provided.

A review by Winterbottom and colleagues of the use of patient narratives in decision aids drew similarly mixed conclusions, with narratives altering health decisions in some contexts but not in others. To cite the paper:

Narrative information influenced decision making more than the provision of no additional information and/or statistically based information in approximately a third of the studies (5/17); studies employing first person narratives were twice as likely to find an effect.

More recent work (e.g., here and here) has demonstrated that narratives can alter people’s preferences for end-of-life care and increase risk perception for colorectal cancer and interest in screening.

So are patient narratives too dangerous to use in health education and risk communication contexts?

Not necessarily.

You see, all of this prior work has treated all patient narratives as if they were identical. Which, of course, is completely untrue.

A story of how one patient went about looking for information about her breast cancer treatment decision is hardly the same as a story about another patient’s experience with chemotherapy-induced fatigue.

A heartfelt account from a prostate cancer patient of how difficult it is to live with treatment induced impotence is never going to be equivalent to another man’s (or perhaps even the same man’s) upbeat description of the fact that he remains cancer-free 3 years after having radical surgery.

In order to advance the science of if, when, and how to incorporate patient narratives into health communications, we need to accept that narratives are highly heterogenous and to develop common classifications, definitions, and terms to use when developing narratives for use and when reporting them in scientific publications.

In the January 2013 issue of Medical Decision Making, Victoria Shaffer and I have published what we describe as “a purpose-, content-, and valence-based taxonomy of patient narratives in decision aids.”

We argue that all patient stories can be classified on at least three distinct dimensions:

  1. Purpose: Stories can have one or more of the following goals:
    • To provide information
    • To make materials more engaging
    • To model targeted behaviors
    • To persuade people to undertake healthy behaviors
    • To provide comfort
  2. Content: Stories can include information about decision making process (e.g., what to consider, where to look for information, what values might be important), the physical or emotional experience of different conditions, the psychological or physical outcomes of a decision, or any combination of the above.
  3. Evaluative Valence: Stories can have a positive emotional tone, a negative tone, or be be mixed.

Our taxonomy is derived from the research literature, but it is not yet empirically validated. We expect that it will evolve over time, as we and others conduct experimental studies, identify categories or dimensions we omitted, and generally build the evidence base surrounding the use of patient narratives. In fact, we expect that some of our hypotheses may prove to be incorrect. So be it.

But, our main message is simple: Patient narratives are NOT one size fits all.

Consider, for example, the impact of our content types on risk communications. If you go back and look at the Ubel, Jepson, and Baron paper, they tested testimonials that a particular procedure went well vs. poorly. In other words, outcome narratives. We agree that outcome testimonials are likely to bias risk perceptions.

Yet, what about our other content types? Experience narratives that describe what it feels like to, say, have severe shortness of breath might improve risk communications by providing rich detail about the potential negative outcome in question. Process narratives could also be helpful by reminding people to pay attention to certain types of risks that they might otherwise forget about.

In other words, “stories” can be both helpful OR harmful, depending on type.

We wouldn’t dream of giving a patient a drug without knowing whether it was an antibiotic or an anti-inflammatory agent. (But they’re both drugs!) We would never have someone start a physical therapy exercise plan without determining which muscle groups and joints will be strengthened or strained by each exercise. (But they’re all exercises!)

By the same logic, we should strive never to use patient narratives in our communications without systematically identifying what concepts they provide, what purposes they serve, and what emotions they evoke.

Patient narratives are a powerful tool, one that we all can and should use to accomplish a variety of health communications goals. But, the persuasiveness of narratives means that they must be used intentionally and carefully.

Do you know what messages your patient narratives really sending?

Brian J. Zikmund-Fisher is an Assistant Professor of Health Behavior & Health Education at the University of Michigan School of Public Health and a member of the University of Michigan Risk Science Center and the Center for Bioethics and Social Sciences in Medicine. He specializes in risk communication to inform health and medical decision making.

{ 3 comments… read them below or add one }

Hilary Sutcliffe January 11, 2013 at 10:51 am

Very interesting and useful work, thanks Brian.


Elizabeth Fryer January 22, 2013 at 4:20 pm

Interesting comparison, prescribing PT only after knowing what muscles and joints the exercises will improve using narrative only after determining “what concepts they provide, what purposes they serve, and what emotions they evoke.”

Which muscles are affected by exercise is concrete and definite.
Which emotions are evoked by a story is not. I would think a single story may even evoke different emotions from the same individual at different times.

What do you mean by “systematically”? Would you tell the story to a focus group? How do you go about collecting the information?


Brian Zikmund-Fisher January 22, 2013 at 5:32 pm

I actually disagree with your perception of unpredictability or vagueness in people’s responses to stories. That’s actually our major point. Stories aren’t these vague things that are interpreted by different people and/or different times in different and seemingly random ways. Stories have specific (and often formulaic) forms, they include potentially well-defined types of material, and serve concrete purposes. A story that models behavior (good or bad) is different than a story that provides information, which itself is different than one that persuades. One that shows me how you thought about something is different than one that shows me what happened after you chose. By deconstructing what a story is made of, what purpose it is serving, and what tone it conveys, we believe that we can do pretty good job of predicting how someone will react to it.

This isn’t exploratory observational research. It’s a theoretically-driven approach drawing from both specific communication theories (e.g., transportation theory, elaboration likelihood, etc.) and decision-making research (e.g., framing effects, information search, confirmation bias) with the goal of building a science of narrative-form scientific communication. And I believe that such a theory will enable us to predict responses to patient narratives with the psychological equivalent of the knowledge that a sit-up (when done well) will strengthen certain muscles and not others.


Leave a Comment

{ 1 trackback }

Previous post:

Next post: