Health Metrics Visualization: Edward Tufte Redux

Edward Tufte

I recently published a post titled Health Metrics Visualization (can be) Beautiful that prominently mentioned the modern pioneer data visualizer Edward Tufte. Today I saw an in-depth article about him in Washington Monthly magazine.

The article in the May/June 2011 issue is The Information Sage: Meet Edward Tufte, the graphics guru to the power elite who is revolutionizing how we see data by Joshua Yaffa, associate editor at Foreign Affairs. The lengthy article is informative and entertaining. Two of my favorite quotes:

“If you display information the right way, anybody can be an analyst,” Tufte once told [the author]. “Anybody can be an investigator.”

“Tufte treats data like good writing,” [Nate Silver] said. “You have a certain thought—how clearly and beautifully are you conveying it?”

The article is an examination of the retired professor and current author, lecturer and sculptor Tufte’s life and work. But Mr. Yaffa places a particular emphasis on the visualizer’s recent work for the Recovery Act Accountability and Transparency Board, the body created by the Obama administration to keep track of the $780 billion in federal stimulus money that has spread out across the country. The program’s aggregate results can be seen on the Recovery.gov website. Tufte’s major contribution to the site, the LIGHTS-ON MAP, which progressively turns on lights across the US illustrating the location of the Recovery Act spending over time, can be seen below. I strongly suggest that you see the visualization in action at the website for the full, informative effect.

The entire article is highly recommended.

The enhanced data capabilities that are are emerging in the healthcare domain will require the leadership and example of data visualization experts like Edward Tufte to achieve their full value and effectiveness. Those who have the responsibility for aggregating and analyzing the emerging mass of health data would do well to observe and emulate the work and teaching of data visualization experts. Those experts and their wisdom will be an essential part of attaining meaningful use of health metrics data in the very near future. The necessity to “display the information in the right way” is central to achieving progress and useful results in healthcare while improving quality and reducing costs.

A data visualizaton guru enhances government data – that’s progress!

Health Metrics Visualization (can be) Beautiful

I believe there is a shared vision in the healthcare community that improved and ubiquitous health metrics will make a significant contribution to the President’s goals to improve the quality of healthcare and reduce its cost in America. The data visualization community, popularized by the work of Edward Tufte, Hans Rosling, David McCandless, and others, would advocate the intensive use of visualization technologies to enhance and improve the presentation of health metrics. There are powerful new tools becoming available through ordinary web browsers that will enable substantially improved visualization of health metrics data. Indeed, the recent PCAST Health Information Technology report [.pdf] suggests that:

“Internet­ based technologies create a platform for “disruptive innovation,” meaning innovations that upset the status quo and can broadly expand markets [...] These types of technologies might allow the 80 percent of physicians who are non­digital [...] [transition] into more modern technologies”

These new health metrics data visualization tools will significantly contribute to the disruptive innovation that the PCAST authors suggest may “help U.S. industry leapfrog to the front of the pack internationally in health IT.”

Traditional health metrics data is typically provided to users as static, absolute information which, in a connected world like healthcare, doesn’t give the complete picture. Relative figures connected to other relevant data enable us to see a fuller picture and perhaps gain unique insights. Connected healthcare data sets presented as insightful visualizations are capable of altering perspectives and changing views. Swedish medical doctor, academic, statistician and public speaker Hans Rosling says “Let the data set change your mindset.” Let’s see how these concepts might be applied to a major issue in the healthcare community: patient compliance.

The Problem of Patient Compliance

Patient compliance, the process of complying with a regimen of treatment, is a universal problem in the healthcare industry. The American Heritage® Stedman’s Medical Dictionary provides a rigorous definition: “The degree of constancy and accuracy with which a patient follows a prescribed regimen, as distinguished from adherence or maintenance.” The importance of patient compliance, and the opportunity for improved healthcare quality, was demonstrated in an often-cited 2004 paper [.pdf] that stated:

“About half of all patients with chronic diseases stop refilling prescriptions by one year.  Several effective interventions are available and adaptations of clinical trials practices offer promise for further improvement.  Poor adherence is a remedial problem in health care quality and its improvement and accountability offer shared opportunities for providers and patients.”

The patient compliance issue will take on new prominence as recent healthcare legislation is implemented. The incentives changed by healthcare reform will require economic participants in the healthcare market to transition from placing primary emphasis on reducing costs as a method of enhancing bottom lines to one where physicians, pharmacists, and hospitals will make money by putting emphasis on healthier people and improved patient health outcomes. Also, the legislation provides for a patient-centered medical home program that will integrate patients as active participants of their own health management with their primary caregiver serving as the point person in the medical home program. A goal of this movement is to improve patient health in a cost-effective manner. Patient compliance will be one available path for cost improvements. Better visualization could be an effective tool to improve compliance.

Visualizing the CRP Test

Bloodwork VisualizationThe potential ability of health metrics data visualization to enable improved patient compliance is demonstrated by a recent prize-winning visualization of the humble and routine CRP (c-reactive protein) laboratory test. The Wired Magazine article,The Blood Test Gets a Makeover,” shows the original traditional computer output CRP report and the suggested visualization of the data by the independent data journalist and information designer David McCandless.

Although the McCandless visualization was harshly and justifiably criticized on medical grounds by a number of commentators on the Wired article, as well as commentators on his own site, it is clear to me that the visualized CRP report is more likely to result in increased patient compliance. Mr. McCandless defends his visualization by reminding his critics that his work was done without any input from the medical community while his most strident critics remind the visualizer that the CRP test is intended for physicians to understand and interpret for the patient considering his or her complete health profile.

Both the visualization author and his critics have valid points. Nonetheless, the dramatic visual contrast of the reports demonstrates the potential for visualization to improve medical professionals’ communication with patients and caregivers. A real world implementation of enhanced visualization of medical test results would be thoroughly vetted by qualified medical personnel for technical content and desired outcomes.

I thought the most provocative and insightful reader comment regarding the blood test report visualization was by commenter Nick on the McCandless website who said:

“This is amazing. I’ll go you one better tho (sic), base it on a living document that will track changes over time, so you can have health history interactivity. That would be swell.”

I agree, Nick, that would be swell! And useful too. Let’s meet the man that says we can have exactly the living document that Nick wants–some now, some soon.

New and Improved Tools

Let me introduce a new data visualizer, a young Frenchman, Paul Rouget, an engineer at the Mozilla Foundation, maker of the Firefox web browser, who says:

“Most of the infographics we see are beautiful, but sooooo (sic) static. You can make them much more alive if you use the new web technologies.”

Paul adds, “It’s a great time to be a data visualizer. After a long period of hibernation, the standards bodies and browser vendors have been extremely busy over the past few years, generating a torrent of exciting technology.” He has written a blog post, Why you should build your infographics in HTML5 and CSS3, providing an overview of a few of these emerging technologies that will bring life to data visualizations.

Lay readers may have the impression that HTML5 and CSS3 are two distinct bodies of browser-based technologies for the presentation of information on the web. In fact, HTML5 is a collection of many separate technical modules, including CSS3. As Mr. Rouget explains:

“the persistent use of blanket terms, especially HTML5, as a sort of brand shorthand for “emerging web technology” is a useful shortcut. It allows nontechnical people to grasp—in a generalized way—the exciting work being done in the standards space right now.”

Those who are interested in the full, detailed description of these new technologies should look at Rob Larsen’s blog post primer HTML5, CSS3, and related technologies: A rapid-fire guide to new and emerging web standards.

Features and Examples

Paul Rouget presents a list of some of the new features of these technologies that are now beginning to enhance every web user’s experience.

Interactive content: You can change the style/content of an element if the user interacts with it: :hover effects, fold/unfold on click, buttons to decide what to draw and select options (select a country, choose how to sort, etc.

Live data: With an image, you’re just showing a “snapshot” of your data at a certain time. The web allows you to fetch data. Your infographic could look completely different depending on when it is viewed. Build your graph with SVG or Canvas.

Make it move!: With SVG, SMIL, CSS Transitions, Canvas (see ProcessingJS), CSS Animations, you can build beautiful animations.

Make it sexy: Well, it’s obvious, your infographic must be beautiful. With CSS3, you have infinite ways to make things beautiful: gradients, font-faces, multiple background, transforms, SVG background, …

Side effects: Your infographic will be accessible (copy/paste!) People will be able to enhance it (view source! Host it on github.) You can have multiple layouts, like a mobile layout (mediaQueries).

Well, Nick, it seems that with these new technologies you could soon have exactly the “living document that will track changes over time, so you can have health history interactivity” that you were asking for. Moreover, since these new technologies will be delivered through the web browser, they will be available throughout the entire range of devices that support web browsers, including mobile, tablet, notebook, laptop, desktop, and many other form factors as they exist today and as new ones evolve.

Not only patients and caregivers will benefit from the new visualization technologies, but all data users in the healthcare field thanks to an improved and enhanced presentation of health metrics data through continuing progress in visualization, analytics and data science. The opportunities for practitioners’ clinical decision support and hospital operations should equal or exceed the benefits to patients and caregivers. The combined advantages to the entire healthcare community will significantly contribute to the goals to improve the quality of healthcare and reduce its cost in America.

New tools for data visualization – that’s Progress!

Opportunities for Reader Action

Visualization Resources

Infographics Collections on the Web

PCAST to ONC: Eat the dog food

While perusing some very old blog posts, I saw Brian Ahier‘s PCAST ~ Where it all began… article with it’s link to the original President’s Council of Advisors on Science and Technology (PCAST) Health IT Report announcement in July, 2010. This led me to the PCAST Report itself and it’s collateral documents and impacts. If anyone wants the full PCAST Report journey, here are the bread crumbs that I followed:

Much has already been written about the details of the PCAST Report. Here are a few summary points:

First, President Obama asked PCAST how health IT could improve the quality of healthcare and reduce its cost, and whether existing Federal efforts in health IT are optimized for these goals. The response to the President’s request comprise the broad landscape and deep analysis of the PCAST Report.

The Executive Summary of the Report lists the potential benefits of Health IT:

Information technology (IT) has the potential to transform healthcare as it has transformed many parts of our economy and society in recent decades. Properly implemented, health IT can:

  • Integrate technology into the flow of clinical practice as an asset, while minimizing unproduc­tive data entry work.
  • Give clinicians real­time access to complete patient data, and provide them with information support to make the best decisions.
  • Help patients become more involved in their own care.
  • Enable a range of population ­level public health monitoring and real­time research.
  • Improve clinical trials, leading to more rapid advances in personalized medicine.
  • Streamline processes, increase their transparency, and reduce administrative overhead, as it has in other industries.
  • Lead to the creation of new high ­technology markets and jobs.
  • Help support a range of economic reforms in the healthcare system that will be needed to address our Nation’s long­term fiscal challenges.

The Report reached six major conclusions:

1. HHS’s vigorous efforts have laid a foundation for progress in the adoption of electronic health records, including through projects launched by ONC, and through the issuance of the 2011 “meaningful use” rules under HITECH.
2. In analyzing the path forward, we conclude that achievement of the President’s goals requires significantly accelerated progress toward the robust exchange of health information.
3. National decisions can and should be made soon to establish a “universal exchange language” that enables health IT data to be shared across institutions; and also to create the infrastructure that allows physicians and patients to assemble a patient’s data across institutional boundaries, subject to strong, persistent, privacy safeguards and consistent with applicable patient privacy preferences. Federal leadership is needed to create this infrastructure.
4. Creating the required capabilities is technically feasible, as demonstrated by technology frameworks with demonstrated success in other sectors of the economy.
5. ONC should move rapidly to ensure the development of these capabilities; and ONC and CMS should focus meaningful use guidelines for 2013 and 2015 on the more comprehensive ability to exchange healthcare information.
6. Finally, as CMS leadership already understands, CMS will require major modernization and restructuring of its IT platforms and staff expertise to be able to engage in sophisticated exchange of health information and to drive major progress in health IT.

The Report particularly emphasized the advantages of focusing on a universal exchange language:

Briefly, the approach described in this report, focused on the technical ability to exchange data in uniform ways, has multiple advantages:

  • It will improve healthcare quality, by making it possible for a physician to integrate accurately all of a patient’s medical information.
  • It will improve healthcare quality and decrease costs, by making it possible for third ­party innovators to compete to create widely applicable services and tools serving patients, providers, payers, public health officials, and researchers.
  • It will provide much stronger privacy protection than available under current approaches, allowing persistent privacy assurances (including applicable patient preferences) to be attached to different kinds of information and using data ­level encryption to prevent access of data by unauthorized persons.
  • It will not require universal patient identifiers, nor will it require the creation of Federal databases of patients’ health information.
  • It will simplify the regulatory burden on providers, by decreasing the focus of meaningful use regulations on ad hoc list of data items.
  • It will help U.S. industry leapfrog to the front of the pack internationally in health IT, by providing exchange standards that can be more broadly adopted by others.
  • It will facilitate public health and medical research, by providing a secure way to de­identify data.
  • It will not require that existing systems be replaced, but only be modestly upgraded or augmented by “middleware.”

Following a detailed analysis of the Health IT landscape in the body of the Report, the penultimate Section VIII presents a Guidance to Agencies, including 1) A Feasible Roadmap to the Future, 2) Guidance on Necessary Design Choices and 3) Guidance on Meaningful Use Requirements.

Finally, Section IX of the Report presents PCAST’s specific and comprehensive recommendations to various governmental organizations, including the Chief Technology Officer of the United States, Office of the National Coordinator, Centers for Medicare & Medicaid Services (CMS), Department of Health and Human Services and other departments and agencies, such as the Agency for Healthcare Research and Quality, Food and Drug Administration, Center for Disease Control, Department of Defense and Department of Veteran Affairs. The details of these recommendations have been much discussed in the trade press and elsewhere, can be found in the Report itself, and do not bear repetition in a brief overview of the PCAST Report.

However, I was singularly intrigued by one particular recommendation in the Report which, as far as I could discover, has not received a single published commentary. The first PCAST recommendation is to the Chief Technology Officer of the United States:

The Chief Technology Officer of the United States should:

• In coordination with the Office of Management and Budget (OMB) and the Secretary of HHS, and using technical expertise within ONC, develop within 12 months a set of metrics that measure progress toward an operational, universal, national health IT infrastructure. Research, prototype, and pilot efforts should not be included in this metric of operational progress.

• Annually, assess the Nation’s progress in health IT by the metrics developed, and make recom­mendations to OMB and the Secretary of HHS on how to make more rapid progress.

The discussion of Recommendations in the Executive Summary near the beginning of the Report is more forthcoming about PCAST’s intent with this particular recommendation, and relates it to the President’s original charge to the PCAST Advisors:

Among our recommendations, we therefore suggest that the Chief Technology Officer of the United States in coordination with the Office of Management and Budget and HHS, develop within 12 months a set of metrics that measure progress toward an operational, universal, national health IT infrastructure that has the desirable features that we have discussed. Focusing these metrics on operational progress, as distinct from research, prototype, and pilot efforts, will enable a more accurate continuing assessment of whether Federal efforts in health IT, including both executive initiatives and legislative mandates, are in fact supportive of the President’s goal of increasing the quality, and decreasing the cost, of healthcare.

In other words, the ONC will, itself along with others, be measured regarding it’s “progress toward an a operational, universal, national health IT infrastructure that has the desirable features that we [PCAST] have discussed.”

Only in these brief paragraphs was there a direct response to the President’s original query of “how health IT could improve the quality of healthcare and reduce its cost, and whether existing Federal efforts in health IT are optimized for these goals” by recommending an “accurate continuing assessment of whether Federal efforts in health IT, including both executive initiatives and legislative mandates, [that] are in fact supportive of the President’s goal of increasing the quality, and decreasing the cost, of healthcare.” Somehow, the most important recommendation responsive to the President’s direct inquiry has been the least noticed, and uncommented particular, in the entire PCAST Report.

In short, the measurers will be measured. In other domains, particularly the software development world, this is called “dogfooding.” That is, the designers being required to use the tools that their users will use to better design them. The CMS meaningful use regulations were designed primarily by Federal Advisory Committees operating under the ONC. While the measurements of ONC will differ entirely from those required by CMS’s meaningful use regulations, the process of being measured should have a similar significant impact on those measured. I am sure that the irony of this particular recommendation is not lost on the hospitals or among the practitioners that are being subjected to the health metrics involved in the regulations implementing meaningful use.

Annual assessment of ONC efforts to support the President’s Health IT goalsthat’s Progress!

Health Metrics and Common Sense

I read an interesting post today by Jennifer L. Middleton, MD MPH, Whatever happened to common sense? Dr. Middleton is a faculty member at a large family medicine residency program. Her heartfelt story about two cases involving infants clearly illustrate the dangers of the tyranny of numbers and the human tendency to follow them without sufficient thought. Well worth a read as a reminder to all of her fellow practitioners as well as her students and anyone in the health metrics community.  Recommended.

Caution when using health metrics – that’s Progress!

World Health Metrics Data: Free, and lots of it

The Institute of Health Metrics and Evaluation (IHME), located at the University of Washington in Seattle and funded primarily by the Bill and Melinda Gates Foundation, has announced  the formation of the Global Health Data Exchange (GHDx). IHME’s goal is to make public-health data widely and freely available.

The database currently hosts about 1,000 readily available datasets and includes detailed information about health metrics surveys, censuses, administrative data, statistical yearbooks, vital statistics, public health records and hospital data, globally. The database is freely accessible to any member of the public and will provide an invaluable resource for researchers, students, analysts or anyone interested in global health metrics issues.

In an announcement by Peter Speyer, director of data development at IHME, published on the Health.Data.gov blog, he said:

GHDx is our user-friendly and searchable data catalog for global health, demographic, and other health-related datasets … [it]  features powerful search tools and multiple ways to browse global health data.

The database provides an intuitive user interface and a rich search engine. The site also offers a selection of visualizations of datasets, data collection instruments used by IHME, computational tools, and a GIS tool for map displays.

More Features

GHDx offers even more sophisticated, useful features for health metrics analysts, researchers and students. In an announcement from IHDE, the database “makes high-quality metadata freely available, reconciling conflicting information and providing the critical information needed to pinpoint the time, source, and content of the datasets that are so important to researchers in the field.” Also, “[the GHDx] contains useful data sources, data providers, and the information needed to easily create citations – some of which are difficult to locate or scattered among different resources.” Director Speyer adds “[while] acquiring and analyzing data about health conditions in communities all over the world … global health researchers like the faculty at the IHME put puzzle pieces together and draw conclusions about the effectiveness of global-health services and systems. This allows researchers to provide valuable insight to lawmakers and foundations when they make decisions impacting the health of global communities.”

While IHME claims over 1,000 datasets, my wildcard query returned 4,593 files on the site. Also, I found 16 series, comprised of many separate datasets.

Peers

The GHDx takes its place among a limited set of large-scale health metrics databases. Among the other large players are the World Health Organization’s Global Health Observatory and CABI’s World Health Database. IHME database will also complement the U.S. health metrics data efforts at the Health Data Community at Health.Data.gov. In his announcement of GHDx to Health.Data.gov Director Speyer said:

We were very excited about the launch of HealthData.gov with information about and direct access to US Government health data. We are using US Government data ranging from surveys like BRFSS and NHANES to HCUP databases and USAID-funded international research, and we will be active participants in HealthData.gov’s community.

Open, Clean Data

Large, open data aggregations are excellent examples of the current trend toward transparency and open data. Governments, NGOs, and other institutions interested in local or world health metrics can utilize these data sources to influence policy and decision-making, directly impacting the health of people and populations everywhere.

An important feature of these large, institutional databases is that they often provide “clean” data for researchers. When presented with raw data, a user can often spend more time preparing it than analyzing it: an inefficient use of scarce research resources.

Stability for a Coral Reef in Seattle

Another important feature of the IHME sponsorship and funding of the GHDx is its stability. The Institute’s University setting, as well as the continuing funding from (by far) the largest grantmaking foundation in the U.S., bodes well for the Institute’s and the database’s durability. The ongoing availability, assurance of dataset growth and maintenance of the database are critical features for prospective analytical, research and student users.

However, I believe that the most significant part of IHME’s plan for the GHDx is the positioning of the database as an epicenter for other data owners to share their data with the public. IHME has said that they would “like to make it possible for researchers to use it as a platform to share information and experiences and work together on research projects.” The Institute has made it easy for others to join in as they “will share all of its research results through the GHDx and invites countries and organizations to make their data available to the public on the GHDx at high transfer speeds with no fees.” Such a generous, open offer is likely to be appealing to researchers from smaller institutions, underfunded grant-supported projects and students everywhere.

Director Speyer illuminates the platform concept in an article in The Daily of the University of Washington by Ryan Dunn:

We feel that this would be a very good place to start a global community of global-health researchers, market researchers, nongovernmental organization folks and people that fund global health, like foundations. I think this would be a good place for all of those people that work with information to share their experiences and document it for other people and basically turn it into a global-health community.

The organic growth of the GHDx database through external contributions, as well as from the Institute’s own research, is like a coral reef–a little at a time–with many collaborating and cooperating organisims involved in the process. It demonstrates the capability of the health metrics community to grow and prosper through community, cooperation, collaboration and sharing.

GHDx represents a significant increase in marketplace capabilities of world health metrics data. It will be widely welcomed by its users and extensively utilized. A final word from Director Speyer: “Health-related data, used as evidence for policy and decision-making, can have a tremendous impact on the health of people and populations, if they are made broadly available.”

Bon voyage, GHDx!

Free World Health Metrics Data Marketplace. That’s Progress!

Opportunities for Reader Action

IHME Website

GHDx Database

Subscribe via RSS feed to GHDx Blog

Subscribe via RSS feed for New and Updated GHDx Database Records