Make the Serious Illness Conversation Guide easy for clinicians to use during a tough conversation


  • Design strategy

  • Analysis and research

  • Information architecture

  • Visual design

  • Production coordination and oversight


  • Established a framework for considering the types of information required for guiding conversations

  • Over a period of years, the core design has been refined and revised to reflect lessons learned from extensive use and testing

  • Over 6,500 clinicians worldwide have been trained on using the Conversation Guide (Kaiser Health News, Feb 2018)


A team of palliative care experts at Dana-Farber Cancer Institute and the Harvard School of Public Health (which later became part of Ariadne Labs) were working on a project to improve communication between clinicians and people who are seriously ill. 

The team had developed and validated a set of carefully worded questions that clinicians can use to uncover and understand the goals and wishes of people with serious illness, and they had developed a checklist-like framework to help clinicians have these important but difficult conversations.

We were originally asked to help optimize the Serious Illness Checklist in 2012. One year later, we refined the design of what had become known as the Serious Illness Conversation Guide, along with related program materials, based on the results of testing and feedback from clinician training. In 2016 we again worked with the Ariadne Labs team to help them evaluate lessons learned and revise the Conversation Guide to better meet clinician needs

How we helped

Our work on the project began with an analysis and mapping of the proposed program components and a detailed analysis of the content on the draft checklist.

We also helped the team evaluate user needs (in this case, clinicians were the target users of the tool) and document the times and places in which end-of-life conversations occur and which contexts we would consider when developing the tool. The team had a number of ideas about how to support the conversation, so we used a priority mapping exercise to help them consider which possible components were most important and easiest to implement

Key to the challenge

Making a conversation guide that is easy to use is challenging, in part, because of the underlying tension between different kinds of information:

  • specific steps and a sequence for the clinician to follow (first do this, then do that)

  • themes for the clinician to address during the conversation (e.g., assess, share, explore, etc.)

  • and suggested words for the clinician to say to the seriously ill person ("What is your understanding now of where you are with your illness?)."

It's hard for people to navigate and process those different kinds of information in real time, especially in an emotionally charged situation. To explore what might work best, we considered a variety of options (small cards, a large shared visual like a placemat, a simple hand-held page, tablet, etc.) and created a variety of prototypes that varied in level of detail, labeling, order of elements, and visual presentation. 

Elements of the solution

Although we experimented with several approaches and the tool evolved over time, the solutions we delivered:

  • Teased apart different types of information and labeled them clearly

  • Addressed the distinct needs of people just learning the approach and the needs of those who were experienced but still wanted a simple reference

  • Allowed individual clinicians to choose the format (from among several choices) that would best work for them

  • Included a Reference Guide for Clinicians that described key ideas of the approach, described the family of Serious Illness Conversation tools, explained the structure of the tool, and offered strategies for common scenarios

  • Included a resource for patients and their families to help people talk to loved ones and caregivers about their illness and the future

  • Included a user-modifiable version in Word so that facilities can make changes locally when needed

Links and resources

Learn more about this tool and related efforts to improve end-of-life care:


Facilitation and design strategy

Part of our role is to help teams reach and articulate a common understanding of their project. Here are a few excerpts from the 2016 redesign work in which we helped the team crisply define their efforts to create a revised version of the conversation guide that could be deployed widely as a stand-alone tool. (Note that it is helpful to be explicit about what the tool can't do, in addition to what you hope it will do.)


What is the Serious Illness Conversation Guide?
The conversation guide is a learning tool to share good practices and proven language. It is intended to inform clinician practice, rather than serve as a procedural task list.

The desired end state is for clinicians to enhance their conversations with patients.

  • clinicians will use the guide only a short while and may never use the language verbatim
  • the guide should be a useful tool in a clinician’s toolbox, but will not by itself transform individual or system behavior

Who is it for?
Primary audience
Busy clinicians (MDs, PAs, NPs) who see seriously ill patients daily and want to improve their own practice with patients

  • they will not receive training
  • they are not likely to have system support; e.g., EHR prompts, triggers for the conversation, etc.

Secondary audience
Busy clinicians (MDs, PAs, NPs) and administrators who work in an environment in which seriously ill patients are treated daily and who want to improve the system in which they work

  • they may be interested in other components of the overall intervention
  • they will still need the Conversation Guide as one element

Information inventory

This "simple" tool to shape a conversation is actually quite complicated. Our analysis of one draft showed that it included:

  • 6 major categories of information

  • 3 kinds of framing content (title, logo, footer)

  • 8 labels used to identify types of content for the reader

  • 6 parts to the conversation

  • 9 actions and 3 instructions (across the 6-part conversation)

  • 2 kinds of branching logic

  • Nearly 200 words of suggested wording made up of 2 kinds of statements and 8 questions


Testing with users

The team understood that some people struggle to use the Conversation Guide, even in the context of a program that included training and support. As they prepared to deploy the Conversation Guide as an unsupported stand-alone tool, the team had to wrestle with:

  • Why do some people "leave" the tool at certain times?

  • How much information do clinicians need, based on the team’s experience and observation?

  • How much information do clinicians think they need vs. how much do they actually need?

  • Which information makes tasks easier and which makes them more challenging?

We developed approaches to test the level of complexity and different use cases and wrote an interview protocol to help the team learn from a group of clinicians testing the new approach with patients.