Client Work Case Study
Health Care Spending in North Carolina
Health Care Cost Institute (HCCI) is a non-profit research group based in Washington D.C. that studies the U.S. health care market. We’ve worked with HCCI before, most notably on their Healthy Marketplace Index series in 2019 and 2020.
For this collaboration, HCCI teamed up with researchers from Duke University’s Margolis Center for Health Policy and Blue Cross Blue Shield of North Carolina. The goal was to pool data from all three organizations to create a comprehensive view of health care spending in the state of North Carolina. We were brought in to design and develop a data visualization experience to promote the research.
The work was generously supported by the Commonwealth Fund and Arnold Ventures.
March 2020–July 2020
Health care professionals / North Carolina legislators and policymakers / members of the press
We developed a beautiful, visual essay that walked readers through major findings from the research. It featured plenty of interactivity, scrollytelling, and downloadable content. The dashboard at the end allows the reader to dig deeper into spending data for a particular county, making it easy for legislators to understand how their own constituents are impacted.
The output was heralded by the research team and funders alike. It generated buzz on Twitter and calls for research into health care markets in other states.
Publicize the results of a major, multi-org research effort.
Inform North Carolina’s legislators about the levels and variation in health care spending across the state.
Lay out a template that could be used to assess health care markets in other states.
The dataset for this project consisted of de-identified health care claims from individuals with Employer-Sponsored Insurance (ESI), Medicare Advantage, Medicare Fee-For-Service, and Medicaid. Together, the pooled data represents about 12.6 million total member years in 2016 and 2017, or 6.3 million people annually.
HCCI, Blue Cross NC, and Duke handled the number crunching for this project. They focused their analysis on health care spending in North Carolina, both in aggregate and on a per-person basis.
Given the variety of data sources, one of the researchers’ biggest challenges was finding ways to compile and standardize the data. It required making dozens of technical decisions about how to report spending in comparable categories, adjusting for factors expected to influence costs (i.e., age and gender), and addressing other issues to measure costs accurately.
See the methodology document for more information on things like sample selection and claims categorization.
Our work on this project began in earnest in March 2020, before the research was actually complete. In the absence of real data, we had to design visualizations that were flexible enough to accommodate a variety of possible insights and conclusions.
As a first step, we sent a “brainstorming document” around to the partner organizations.
Having the partners answer these questions gave us a better sense of the deliverable and ensured that they were aligned in their expectations for the project.
In early discussions with the HCCI and Duke research teams, it became clear that they wanted both an explanatory and exploratory component to the experience. So we divided the story into two sections: a scrollytelling piece that would introduce the project and highlight key takeaways from the research; and a dashboard that would enable deep exploration into health care spending in each of North Carolina’s 100 counties.
We honed things like colors, page layout, and story flow in our early concepting.
What tangible objectives exist for this project already?
What audience(s) do we anticipate interacting with the tool?
What are the quintessential things that the user needs to be able to do?
Find at least one example of a digital, data-driven experience that you really like. What makes this project so engaging?
We went through three rounds of formal mockups with HCCI and Duke. The mockups outlined the complete experience in higher fidelity, including the functionality for every data viz.
This chart, in particular, took some careful consideration. The goal was to show the distribution of per-person spending across counties for different payer populations.
HCCI also wanted a way of determining which dots mapped to which counties. We achieved this through some clever interactivity, highlighting and linking dots from the same county on hover.
The default view of the dashboard gives readers a sense of the geospatial distribution of per-person health care spending. They can also drill down by payer type or service category.
Then, by zooming in on a particular county, the reader has access to more granular data like:
- The payer type population breakdown
- Per-person spending by payer type
- The percentage of spending derived from inpatient, outpatient, and professional care.
We relied heavily on D3.js to build the data visualizations and make them interactive. To collaborate with researchers on the website copy, we created a lightweight CMS using Google Docs, ArchieML, and Handlebars.
To make the dashboard downloadable, we used Anatolii Saienko’s `dom-to-image` library and some custom code to screenshot the contents of the dashboard container when the user clicked the download button. This meant that the outputted image was dynamic and reflected the selected county and applicable filters.
Our team developed the digital experience in a six-week sprint. We checked in code via GitHub and spun up a password-protected test link using GitHub Pages for HCCI to track progress.
HCCI regularly compiled feedback on the test link via Google Doc, which we triaged and added to our task management system as we went.
We worked with Formations Design Group -- the agency that administers HCCI’s website -- to deploy the project. They helped move our code to HCCI’s servers and troubleshoot small bugs.
The project was released to the public on June 30, 2020.
“Shout out to@TheDataFace for dashboard and visual design elements - we love working with them on projects like this and the Healthy Marketplace Index!!”