Visualizing Bloodtests

Tuesday, December 21st, 2010

We were happy to win a design competition in Wired US this month.

It was around re-envisioning medical data. Specifically, blood test results. Which suck, design-wise. They still look like secret missives from the CIA circa 1965. Yet their contents are vitally – perhaps mortally – important.

Our challenge was to approach a cholesterol level test. First it looked like this.

Then we designed it thus:

Our goal wasn’t just a polish job. We worked hard on the information too. So there was context for all the facts and figures. Ideally, anyone, of any educational background, could get the gist and plan their next move.

See a hi-res version here. Or download a PDF.

(The image is creative commons. All non-commercial use is cool.)

Great work too from Mucca Design and Jung Und Wenig

Credits —

DESIGN & RESEARCH: David McCandless & Stefanie Posavec
ADDITIONAL DESIGN: Stefanie Posavec, Joe Swainson
CREATIVE DIRECTOR: Tim Leong @ Wired

http://clubneko.net/ nick

This is amazing. I’ll go you one better tho, base it on a living document that will track changes over time, so you can have health history interactivity. That would be swell.
http://www.steigmancommunications.com Daria Steigman

I didn’t realize this was YOUR redesign. I’ve been making everyone I know look at it since I first saw it in the printed magazine. Kudos for taking those tiny, oft-misunderstood numbers and putting them into a format where someone can actually both understand what they say and take action.
Kreso

HDL – should be printed “desirable” instead of “high” or somehow indicate that more is better
Dan

I saw this in Wired last month, didn’t even realize you were behind it. Awesome.
http://cassiewallace.net Cassie

Absolutely fabulous. This would save me some mega headaches at the doctor’s office, and it’s pretty too.
http://www.SortingThingsOut.com Noreen

Fantastic job! Very marketable. Ka-ching, ka-ching.
Matt

Only issue I see is what if the level was low enough to hit the text. Ie: If CRP = 1 it wouldn’t fit without overlapping the title text. Overall great work though. I would love to get something like this for prescriptions. Info about the pills are either so small on the bottle or in huge blocks of text with a standalone sheet / folded up in the bottle. Hopefully this can push a move to something like that! The only thing holding it back would be cost, I worked at a pharmacy for years and know for a fact that ink costs kept prescription information sheets 99% black and white.
- Peter Starokadomskyy
  
  Same question! maybe it worth place dots inside of the bars?
Italo

great work, well designed, congrats!
http://www.mathgon.com/ Mathgon

Nice job, but I won’t work with a matricial printer… try ACSII art ;)
Richard T

Though this is an excellent use of infographics it is based on the assumption that the patient is the consumer of these ‘bloods’. In the real world this data sheet is aimed at the clinician. Would have been nice to see the (presumably reduced & concise) graphic that that would better portray the data to that audience.
Matt Hammond

A massive improvement in so many way – well done. If you were to iterate this design again, can I suggest you might consider substantially increasing the smaller font sizes? Sizable proportions of the population have poor visual acuity, due to, for example, age.
Andrew Scott

Nice work, but I think there are a few issues:

The use of the orange circles ties the design together aesthetically but also creates an unwanted link between the 1, 2, 3, 4 steps, the readings and the “Bacta Medical Centre”. Perhaps this isn’t a good thing.

The cholesterol bars strongly suggest a zero reading is “desirable” when I’m not sure that’s true. Surely very low readings would be a condition called hypocholesterolemia, which Wikipedia tells me is linked with with depression, cancer, and cerebral haemorrhage. The bars should suggest a realistic and healthy target for cholesterol not a dangerous extreme. It might also be useful to include terms for high and low readings (hypercholesterolemia, hypocholesterolemia) to help patients read up on their condition after a consultation with their doctor.

It would be good to see a black and white solution output solution and also one that utilised monochrome output on coloured stock (so you still get the nice coloured bars, etc.) however I realise this is just a mock-up to stimulate ideas and not a complete solution.
Matthew E Martin

Really great except for one thing: the evidence of cholesterol levels (even HDL vs LDV/vLDL) being predictive of heart disease is very poor. Problem is the science is being perverted by big money, for a change. E.g. http://high-fat-nutrition.blogspot.com/search/label/Cholesterol%20heart%20attacks%20and%20JUPITER

But that part’s not your fault, the graphs look pretty!
Brian

Class! I want to get my cholesterol test done now, but only if it looks as cool as this. This would really help people understand what these tests mean and how to act on it, we are visual creatures after all. When you retest you can have the old values faded to show you , hopefully, your improvement over time. Make it interactive and hey presto a very useful tool to help patients.
It reinforces the fact that a lot of information is poorly presented and therefore not very accessible.
http://kaizer.se ulrik

This is nice and all, one caveat though: Cholesterol is not dangerous, and the levels used for “dangerous levels” are so low that 90% of 60-year-olds (Sweden) have at least that level; that can hardly be considered to tell you anything about heart conditions, if you point out that 90% of the population might have them. The most profitable medicine worldwide is the cholesterol-lowering medicine Lipitor!
gerrit

Great, yet I suggest to improve readability by color-coding which parts of the document directly derive from the patient’s data and which ones are interpreting the data. E.g. blood readings should not look the same as paragraph numbers (white in orange circle), and the 15% could have a similar shading as blood readings.
NHSer

This is a stunning, my first impression was wow….

…then a horrible feeling dawned on me that this will never be realised. Can you knock something up that will automatically convert the data in various hospital systems into a consistent document?

Something like this is more feasible in the form of a care plan from GP systems.
http://www.alessandromingione.com Alessandro

Wonderful!
The only issue I have with this is that (i think) the previous one was b/w with monospaced text for a reason…
But since some of the latest tests are downloadable from internet (yes, even here in Italy) there shouldn’t be any problem at all.
Clay B

In section 3 (“Your risk would be lowered to”), are the 12%/10%/6% numbers if you do those three things independently, or are they cumulative effects? It’s not clear.
Ravi

The colors of the bars could be green ,amber and red -like traffic signals -indicating severity of the readings.
Aceeptable-green
Borderline -Amber
and
Not acceptable/beyond limit-Red
http://nicolasgallagher.com Nicolas Gallagher

It would be interesting to know what constraints a practical redesign would have to work within. Issues surrounding the equipment used to produce this paperwork, the cost of production, who will really be consuming the information, etc. As far as I can tell, none of the Wired redesigns involved researching those factors.
http://viedevelo.wordpress.com Matthew Perry

I like the visualizations BUT … The scientific evidence to support these cholesterol recommendations are not very solid. See “Good Calories, Bad Calories” for an overview of the science and why looking at cholesterol numbers in this manner is a dangerously misleading.

The charts could be improved dramatically if “desirable” and “optimal” were not placed directly above the 0 mg/DL label – to claim or infer that 0 mg/DL is desirable is completely irresponsible. Cholesterol is an essential part of our bodies and while very elevated levels have been *correlated* with poor health, so have very low levels of cholesterol. That’s not to say that either high or low levels have ever been proven as causative factors.
http://www.andreyshipilov.ru/ Andrey Shipilov

Visually looks good. But the information is given even worser than it was.

Name: John Doe. Gender: male. DOB: 01/01/1950. Age: 40
Should be given much shorter:

John Doe. Male, age 50.
The dob is not important in medicine. But the name and age are.
- Anna
  
  I’m a nurse and we always use DOB in my (large) hospital system. Always.
  Age changes, leading to confusion. DOB does not.
  
  I think the main idea here is the whole concept, I’m sure that details of the document would be tweaked and reviewed by medical professionals in the real world.
  EVERY document and form that we use goes through an approval/review process at the hospital where I work.
Jonathan

So in summary, this is a misleading info graphic (as explained by some commenters here) designed for an audience that the original was not intended to address, and which adds information the original doesn’t have. Not only that, but it’s presented with hardly any rationale for those changes other than it simply looks better than a dot-matrix print out.

Wow – I wonder what other entries to this design competition were like!?
http://snuze.blogspot.com Snuze

Excellent design! The use of colours is fabulous and the explanations given for each data point was done in an informative and creative manner.

♥♥♥
Omar Mustafa

This is a great idea. It helps a lot with our patients education. I’m thinking of re-designing the blood results in order to help non-specialists understand specialist tests. Would like to help the team.
http://eurogene.blogspot.com Keith Grimaldi

Very nice work -

BTW all the above comments about the danger or not of cholesterol are not really relevant, the scope of the project was to turn a standard blood test report into a graphical version, not to reinterpret it. Sadly or otherwise the official interpretation of high cholesterol is “danger”

Another BTW – to make this happen needs some sophisticated and flexible Rules software – we developed that as part of our EU project Eurogene :) Contact me for any more info!
http://www.wunderlich.ca John Wunderlich

Would love to see the same kind of redesign of a web site privacy policy.
ExitToShell

That’s great and all. But the document is meant for clinicians, not patients.

Further you have increased to use of paper for a print out almost 4x. The color inks are more expensive to print than B&W or graytone. Text should be larger overall for older patients (heck even I need reading glasses and am far from 40 yo).

The “Recommendations” at the end should not be there. That is for a doctor to recommend based on the patient, again see senior citizens.

What we really need is to get a set-in-stone data storage format for medical data that can be visualized in any manner and easily transmitted/saved (even localized) by patients and doctors. It could be XML, or even something like JSON.

Once we have that we can use smart cards to carry our info (encrypted of course) on us, get it to a doctor instantly (instead of endlessly filling out forms with the same info time after time) and can append it easily when third-party testing facilities are used.

That is what we need.
http://www.tactix4.com Rob Dyke

@NHSer

“Something like this is more feasible in the form of a care plan from GP systems.”

I’m going to contact InformationIsBeautiful right now to help with that job.
Gregory Goldmaker

Nice work!
Debra

So much better! Thank you for making the world easier to understand.
Anderson Fagundes

Very good! It’s really necessary for a lot of tests results in medicine.
“Users” of these tests will thanks
Alex Floyd

C Reactive Protein, or CRaP test is one of the least high predictive value tests in Medical Technology. Predictive value of a test describes sensitivity and specificity. Sensitivity can be high yet have high false positives. Specificity pertains to false negatives. That one is pretty sad. If you have false negatives, how do you know whether to retest? C Reactive Protein levels are related to inflammation of unspecified origin. A high level could be due to almost a thousand different maladies. Not a very good test.
Furthermore, the result in the first case was 0.4 mg/dL. In the redesigned format the result was revealed to be 3.3 mg/dL. Fat the whuck?
http://www.abhimanyughoshal.com Abhimanyu Ghoshal

Simply brilliant. This is exactly the direction in which future advancements in medical care should go – friendly, simple, clear and useful information that anyone can understand. Great job on this.
Rob

Great piece of work. Congrats. I would like to offer a couple of comments if you do similar work in the future. They both concern the context of the outcome figure, in this case 15%.

1. Instead of showing a percentage figure, I like the absolute numbers approach. Does the 15% figure mean: “Of every 100 people like you, 15 of them will have a heart attack in the next 10 years” ? If so, I think it is more meaningful like that. I have been completely convinced by the book by Gerd Gigerenzer.

2. This is more difficult, and when I asked my GP the same question, he shifted in his seat and looked uncomfortable; nevertheless I think it is appropriate: “Of 100 people of my age how many expect to have a heart attack in the next ten years if we ignore risk factors and how does that help me to understand my figure?” The answer to this question puts my 15% (e.g.) figure in a wider context.
http://therearemonkeys.com/twowords/ Jono

Something missing is a unique identifier. If this were given to a patient and subsequently it was the wrong person that would be a tad upsetting. Also, the gender could be represented as a male / female icon like the ones on loo doors.

I read the other comments and see the ones attacking the use of paper and ink. What this redesign addresses is value. If the money spent on successfully making the patient understand the importance of the result and getting them to take action then the price of paper and ink is money well spent. The original doesn’t even delineate where the result and the guidance information starts / ends – easy to misinterpret. The original was a product of the technology put in place years ago, continuation paper and dot matrix printers. We have moved on somewhat.
http://diagnosticinformationsystem.com Bob Coli, MD

Although I am now living just 15 miles from the Dartmouth Hitchcock Medical Center, I first saw a reference to this WIRED design competition yesterday in the KevinMD blog at: http://www.kevinmd.com/blog/2011/01/blood-test-designed-results.html

The 38 individuals who commented, especially “ExitToShell”, might be interested in the comments I posted today at: http://www.wired.com/magazine/2010/11/ff_bloodwork/comment-page-1/#comment-10874
Anna

That is absolutely awesome!

I am a nurse and I also see this idea from a clinician’s view… since medical charting is becoming paperless, our hospital system uses Cerner… please, for the love of all that’s holy, get with Cerner and collaborate.

ALL DAY, EVERY DAY we have to use what is basically the CIA circa 1965 computer screen. It’s awful and not only is it simply unappealing, I would bet that more charting/medical MISTAKES are probably made on it because of the way it looks/layout, etc.
Tara

:O I am speechless…simply AMAZING! It would be awesome to have this implemented in real medical lab
Ted

Hi, this is a zillion years old in internet time already, but I wanted to jump in here and point out that the “good cholesterol/bad cholesterol” dichotomy is really misleading. LDLs, like HDLs, are extremely important in regulating the body’s metabolic processes. Having a higher HDL:LDL ratio is generally accepted as “more healthy” as the alternative, but most experts agree the raw numbers are more important than the ratios, and there’s also no evidence that decisively links increases in HDL concentrations to improved health.

Visualizing Bloodtests

Tags —

Credits —

Related