As might be evident to close readers, when it comes to Makeover Monday designs, I often prefer to go in a different direction with my revizzing than what I expect people to choose to pursue. Likewise, if I get off to a late start that week, I see what the earlybirds and jackrabbits (including, but not limited to, @thedataduo) are doing and recalibrate my plans accordingly. I don't want to create "just another version" of someone else's vision.
Why do I approach Makeover Monday this way? To answer that, let me take you back to a simpler time in history: 2005.
Why Zig When Others Zag?
For a few years in the mid-aughts, I used to compete in the Washington Post's weekly "Style Invitational" contest. I'm too lazy to explain it so here's the first paragraph from its Wikipedia article:
The Style Invitational, or Invite, is a long-running humor contest that ran first in the Style section of the Sunday Washington Post before moving to Saturday's Style and later returning to the Sunday paper. Started in 1993, it has run weekly, except for a hiatus in late 1999. In that time, it has had two head judges who select winning entries: "The Czar" and "The Empress." The Czar, who was anonymous but later revealed to be editor Gene Weingarten, abdicated in late 2003, leaving the contest in the hands of his former associate, The Empress, copy editor Patricia (Pat) Myers. The humor ranges from an intellectual vein to a less mature style, and frequently touches on sophisticated political or historical allusions. While the contest theme changes every week, some popular contests are periodically repeated. The S.I. has a loyal following of self-proclaimed "Losers," who refer to having a contest entry published as "getting ink".
I am proud to say that my entries appeared in the paper 32 times, including one victory (for which I received "the Inker;" shown below), two runners-up, and three other top-five finishes. This contest has been going on for almost 25 years, and even though I only competed for maybe two years, total, I'm still among the 200 most published contestants (of the 1900 or so to have ever appeared).
I learned quickly that in a weekly comedy writing contest, your first thought is usually going to be a LOT of people's first thought, and that comedy requires the audience to be surprised. Your first (and second and third) thoughts should usually be discarded, since it's those fourth and fifth thoughts that will be unique to your point of view, and therefore potentially successful in a humor contest with hundreds of other funny people entering. I also learned never to take weeks off, because your comedy writing muscles develop with repetition and with forcing yourself to attempt challenges that don't interest you. And lastly, I learned not to pander to my perception of the Empress's tastes, since that forced me to write in a voice that wasn't mine; that lack of authenticity usually rendered my entries unfunny and unsuccessful.
Many of the lessons I learned from participating in the Style Invitational carry over to Makeover Monday.
Discarding my first (second, third) thoughts often leads me to a more sophisticated, distinctive, or creative design or analytic approach than I would otherwise have settled on.
Participating every week develops my viz muscles, even (especially) when I'm not particularly taken by the topic or the dataset.
You can't predict what an audience is going to fall in love with or what they're going to find uninteresting...but you can make sure that YOU are happy with what you submit, and that only comes when you feel you've put in sufficient mental energy and effort to do the challenge justice.
Finding the Right Story
This week we Monday-Made-Over a crazy simple chart that did not remotely take advantage of the size of the underlying NHS dataset. With nearly 800 million rows of prescription data covering seven-plus years, this EXASOL-ized data from the UK deserved to be delved more deeply than two aggregated line charts would suggest. But where to start?
I was away for the Easter holiday so I didn't jump directly into the project. Before I even had time to look at the row-level data, jackrabbits like Adam Crahen had already built multiple vizzes full of beauty and creativity. These, and others like them, were focusing on the big stories in the data: the big outliers; the combined totals; the geographic distributions; the growth and changes over time; and so on. I knew that I should follow my instinct and look for another path to follow--the path of the small data within the big dataset.
Secondly, my wife is a biostatistician with many years of experience in the pharma industry. I knew that I could rely on her expertise to fill in the gaps of my knowledge regarding pharmaceutical data. But when I thought about it more, I realized that I shouldn't be thinking of her as a handy subject matter expert for my own benefit; but rather, I should be telling a story that might be educational to everyone who ultimately sees the viz, and should help to illuminate some of the more complicated and obtuse aspects of the pharma industry. Basically, I felt obliged to tell a story that would let me bring her specialized knowledge to a wider audience.
And finally, I'm fascinated with how plaque psoriasis drugs are marketed in the United States. If you aren't in the U.S., you maybe can't appreciate how commonplace television advertising is for certain classes of prescription medicine. Mostly it's for impotence, anxiety, asthma or allergies, and the like; but plaque psoriasis, by its nature as an autoimmune disorder, also has a fairly large overlap with people who suffer from psoriatic arthritis, ulcerative colitis, and Crohn's disease. What's interesting to me is:
how the same drug can be marketed in two different ways to two different audiences (since these drugs are approved for multiple "indications," or accepted uses, by the FDA); and
how commonplace the ads are, given how few people (compared to the entire population) have this level of severe psoriasis.
But when you consider how expensive these treatments are (multiple thousands of dollars monthly, in the U.S.), you can see where there's value to be had in marketing these medicines directly to the consumer (who of course has to get a doctor to sign off on prescribing it; but doesn't necessarily have to first go through the same step-by-step treatment protocol that people in a nationally-managed-care system have to).
Even in the time I was working on this viz, both the Otezla and the Taltz ads happened to come up on TV. Here are sample ads from the drugs in the plaque psoriasis category, focusing on that particular indication. You can see just from the screenshots that the goal is to make life seem fun again, and to show patients feeling confident in showing their skin. (Except for Remicade, which goes exactly the other direction.) I've seen all of these at least once, with the exception of Remicade; that one is a print ad only.
Now, in my viz, which focused on NHS adopting the oral tablet Otezla as part of its approved treatment protocol, I show the NICE flowchart of treatment steps. In it, we see that a patient must go through assessment, topical therapy, specialist referral, phototherapy, steroids, and oral tablets (Otezla) before proceeding to the "injection-delivered systemic biological therapy" stage of treatment.
But look back at the ads above! The disclaimers (the ones you can see) all say some version of "for adults for whom systemic therapy OR PHOTOTHERAPY is appropriate." The ads imply that in the U.S., you don't have to go through steps 4 and 5 before jumping to 6; just ask your doctor! Advocate for yourself!
So anyway. That's what got me interested in looking at plaque psoriasis in the U.K. system. And the discovery was not surprising to me, and was covered in the viz itself: once NHS added a less-invasive, more cost-effective level of plaque psoriasis treatment to its protocol, overall costs of care went down and adoption of the oral tablet treatments went up.
The Truth Isn't Always Beautiful
But in setting out to tell this complicated story, I realized something:
It wasn't going to be a very beautiful viz.
I knew this because the story, and the details of the story more specifically, was complicated. I was going to have to lead people through the story step by step and explain some nitpicky elements along the way. That meant I was going to have to ask people to do what they hate to do: read. Even worse, I was going to have to put the main visual element at the bottom of a long chart that wasn't going to be especially eye-catching.
But these were the tradeoffs I was willing to make in order to tell the story of Otezla as authentically and as accurately as I could. I'd post the screenshot to Twitter with the proper hashtag; I'd post the workbook on Public as always; but I knew all along that I'd get, at best, half or a third as much engagement as I could get with just a pretty picture.
In exchange for that smaller engagement, I earned the right to tell the story properly. In my mind, that required several non-negotiable elements:
Explain the nature of plaque psoriasis and how it relates to other diseases.
Explain why "psoriasis," which some people might dismiss as just some skin irritations, is actually debilitating--even life-threatening--to many sufferers. Suicidal thoughts among patients with the most severe cases are not unheard of.
Explain that many plaque psoriasis drugs are "biologics," derived from genetic material, and that biologics do not have generic versions; they have bio-similars that work similarly to their parent biologic, but are not identical.
Determine, and explain, that NHS guidance is to prescribe medicines with the "generic" name when at all possible, even when (as in the case of biologics, or patent-protected branded medicines) there is no such thing as a generic. This means that there are British National Formulary (BNF) codes and names for medicines that appear to be generic, but really are just other names for the biologic or the branded medicine that is actually being dispensed.
Explain that there is a standard guideline within the NHS for the treatment of most diseases, including plaque psoriasis.
This standard guideline includes a multi-step course of treatment, and biologic systemic treatment comes at the very end, as a measure of last resort.
Within the last few years, many new drugs have been approved in Europe and in the US for treatment of this condition. Many of these are "bio-similars" to already approved medicines that are in the protocol flowchart.
Show that these approvals happened mostly in the last few years, after a fairly large time gap. (This is the most technical thing in the viz, from a Tableau-dashboard-creation perspective. It's shown below this list.)
However, NICE emphasizes an "evidence-based" model of medicine, which would make these new drugs less likely to be added to the official protocol unless they offered something completely new and beneficial to patients and NHS in terms of treatment and/or cost.
Explain that NICE added Otezla to the treatment protocol both for cost benefits (treatment is much cheaper than treatments with biologics) and for patient care benefits (oral tablets are much less invasive than injections or IVs, and could put the psoriasis into remission without requiring such medicine).
Finally, show that costs have begun to drop as Otezla usage has ramped up.
In the end, I wound up with a viz that looks more like a magazine article than a dashboard or an independent viz. It's text-heavy, for sure; color-wise, it's based on the Otezla palette; there are tooltips in every section but they aren't necessary to tell the whole story; and all of the stray datasources I needed are identified, as best as I was able to do so, in the footer. (I needed to find out the EMA and FDA approval dates of all the drugs; I needed to find the manufacturer and the molecular names of each medicine; I needed to find and create an interactive version of the treatment flowchart from NICE; and I needed to find estimated monthly costs for treatments for each medicine--including the ones not prescribed by NHS.)
I did make one mistake: originally I aggregated the data by quarter, but then realized that the Q1 2017 numbers would be skewed (data only ran through one month of that quarter), so I rebuilt it.
At this point I'd be remiss not to say that Daniel Caroli built an excellent, clean, and informative viz about Salbutamol and asthma that has a similar story to tell and does so with many fewer (although not zero) words. Good on Eva for featuring him in the weekly write-up, as he's certainly one-to-watch.
I Thought It Mattered....But Not Compared to How People Mattered
Believe it or not, I'm quite proud of this text-heavy visualization. I think it functions best as an article that leans on charts to support its points, rather than as a viz that stands on its own. The world of pharma is complex and dense at times, and telling a story like this requires a reader/viewer to understand a fair amount of back story and context. To that end, I was satisfied with the way I was able to lead a reader/viewer from point to point, even if it felt like there was 80% context and 20% story.
Most importantly, I felt a real obligation to produce something my industry-savvy wife would be proud of. She has often said that her role in many situations is that she "speaks for the data," meaning she has to take the most objective view possible, and ensure that no analysis over- or mis-states anything that the data doesn't truly say. In clinical trials, determining the efficacy of a drug is important, but so is ensuring its safety; in the past I've explained her job to people by saying "she uses math to save people's lives."
In a week that dropped the better part of a billion rows of pharmaceutical data into my lap, I wouldn't have been satisfied with putting forth a submission that in any way trivialized, overstated, or misrepresented the facts of the matter, because to do so would have been disrespectful to her. If it's overly textual; if it's laden down with provisos and explainers; if it takes a lot of pixels and real estate to ultimately make a relatively simple point; so be it.
After all, this is data about people at their most vulnerable. It's the medicines they seek out and take, in some cases, to stay alive. It's a million stories of individuals worried about themselves and their families, trying to live their best lives and admitting they need medicine--and each other, via a national managed-care health system--to accomplish that. It's all too easy to forget about the humanity and vulnerability encoded in every line of data this recordset provides.