Ambient audit and feedback: a prototype data display

The recognition of suboptimal performance can act as a cue for action, encouraging those who are both motivated and capable to take action to reduce the discrepancy.

Ivers et al., 2014
A map of relationships and similarities between features of clinical performance improvement interventions

My final studio project in my graduate education focused on answering a research question I had been forming while working as Human-Centred Design Lead for the Alberta Physician Learning Program:

How could healthcare performance data be more approachable in clinical settings?

Project goals

  1. Integrate data into a primary care clinic so that it is can be seen by anyone in the clinic.
  2. Be a low-profile piece of the environment that provides a visual reminder of change.
  3. Be data-driven, reacting to updating datasets and changing form.
  4. Reduce the time required to access and understand data.
  5. Encourage peer-to-peer discussion in response to the changing display.

Challenges with data in healthcare

Not looking at issues of data integration and silo’d healthcare data, the way that performance data is integrated into clinical settings means that providers are forced to go out of their way to view something unrewarding. I was inspired to develop this prototype by my time spent designing individualized practice reports for physicians. Reviewing data that describes clinical performance is a standard method of improvement, but often requires providers to take time out of their already busy day.

The presentation of the data is often traditional – bar charts, line graphs, statistical process controls – and unappealing to look at. So healthcare providers are going out of their way to look at something unappealing; which leaves their motivation to engage with the data up to the individual’s desire to understand and improve.

Healthcare providers would have to access their performance data on the computer, through dashboard or some other report document.

Generating a guiding framework

Because of the four years I spent designing clinical performance reports, I already had a lot of practical knowledge of best practices. Though I lacked understanding in the context of how data has been used to improve clinical performance. An environmental scan helped me identify the activities available, but additional research was needed.

A review of literature in the areas of interprofessional feedback, self-regulated learning and information use in primary care helped me form a framework to describe the types of performance improvement activities available to healthcare providers.

This framework describes these activities in four dimensions:

  1. Goal of the activity: is it to build knowledge, aid in reflection, or build capacity?
  2. Consideration of context: how is the activity delivered with respect to their context?
  3. Delivery method: is it facilitated, digital or physical, or a social activity?
  4. Needs influencing activity design: is the activity meant to meet needs that were provider-expressed, behaviour-indicated, comparison-based, expert-defined, or infrastructure based?
There’s no one right combination of the above framework – every provider has their own needs.

Developing the Ambient Audit & Feedback prototype

With this framework as a guide, I did some ideation and sketched of possible info displays, exploring different variations and thought processes. Situating the prototype in the clinical environment was important, so many of the ideation focused on physical manifestation of data.

A number of sketches looked at modifying EMR interfaces, adapting EMR reminder systems or creating detailed email reports.

Looking at how to visualize impact, i came up with the idea of a trash can that would only open far enough to easily use while under target.

Exploring how to show progression towards a target lead to the idea of water, a plunger and containers to indicate appropriateness.

Abstract visualizations were a big focus of the sketching, some which considered the relationship between the data and a person.

Physical prototyping involved cardboard, paper and a toy figurine to present the final concept to my peers for feedback.

Next steps

Building a more refined prototype using a Raspberry Pi would be the next steps. Though that would involve learning more about Raspberry Pis and physical information displays.