Why is WebMD so awful?
And what would a "good" WebMD look like?
WebMD—and its imitators—are terrible. Often the first stop for health questions, WebMD bombards you with vague, unhelpful articles strewn with garish pharmaceutical ads—an ocean of "content" without substance.
And I'm not just some crank with an ax to grind—satisfaction with online health information is incredibly low, around 38%. What's more, this satisfaction has been very stable over time—one study found that users in 2008 were just as unhappy with health information as they were in 2017!
This is pretty odd. In those nine years two things happened: (1) software got vastly more powerful and (2) researchers published eight million new health citations—a 47% increase in all published health knowledge. So why hasn't any of this innovation made online health information even slightly better?
It's baffling. The sum of human health knowledge is large and growing rapidly—but the meager, watered-down portion allotted to consumers hasn't grown with it.
This has always been a problem, but COVID has brought this knowledge gap into sharp relief: consumers have been flocking to health websites in droves, searching in vain for answers to their COVID-related questions. And—thanks to the blistering pace of COVID-related research—those answers are changing frequently, making it almost impossible for a typical person to find up-to-date COVID recommendations.
In other words, online health information has always been bad, but COVID made the situation dire. So why is no one talking about it?
I think we've collectively gotten so used to terrible health information that we can't even imagine a better version of WebMD. So let's toss around some ideas—what would good health information look like?
For starters, we'd change the format. Most online health content is buried in long, wordy articles that manage to say almost nothing in 2,000 words (and 8-10 pharmaceutical ads). This isn't by accident—the goal of these articles isn't to improve your health, it's to rank higher on Google! And every SEO expert knows that long articles rank higher than short ones, so now "treating covid headache" takes you to a thousand-word article when really I just want short summaries of the top three treatments.
We don't tolerate this "editorial style" of information on other websites. Imagine if each Zillow listing was a college-length essay, and instead of saying "1,700 square feet, 3bd/2ba, currently listed for $800k" it said "this house is medium-sized, although some visitors report feeling that this house is smaller or larger than that. It probably has some number of bedrooms, and is listed for a certain number of US dollars." We'd never use Zillow again! And yet for some reason we accept this as the norm with health information.
Instead, it'd be great to have a bit more structure in our health data. What if I want a list of antidepressants, ranked by efficacy? What if I want to filter out the ones with nausea as a side effect? Or maybe I just want to see the ones with the best evidence. And why—dear god why—can't I see the percent of people who experience a particular side effect? I'm obviously not going to take Lexapro if there's a 90% chance I'll experience "brief feelings similar to electric shock"—but if it's a 0.01% chance then I'd consider it. I know that data exists—WebMD just doesn't want to give it to me.
So better health information would be quantitative, sortable, filterable, etc.—in short, it would have all of the features we'd expect from Zillow, Amazon, Expedia, Airbnb, Uber, etc.—all of whom are very good at conveying lots of complicated information to consumers through a 3x5'' screen that they glance at for three seconds.
Next, we'd make sure this data was updated frequently, ideally every day. New research often changes the established medical orthodoxy (that's the point of new research!) but the current editorial-based health references usually update their articles once every few months, and big overhauls happen more like annually, if ever. Even well-studied diseases (e.g. diabetes) have new research every few weeks—so a health reference that's a year old can easily have outdated information.
Finally, we'd put the evidence front and center. Most consumer health websites either (1) bury their evidence in the footnotes or (2) dispense with evidence altogether. You might be saying, "well, I wasn't going to read the evidence anyway, I'm sure WebMD has someone vet their articles." But I promise: the evidence does matter, even for the "non-science" folks.
For example: the WebMD articles on essential oils and aspirin are written in almost the same tone—even though the evidence for aspirin is an order of magnitude stronger than the evidence for essential oils. But because both pages lack a good summary of the evidence, a typical user might think that both treatments are equally likely to be effective. Gah!
Instead, a good health reference would give you a rough idea of the confidence that researchers have about the topic. It might list all the studies for a claim, along with structured info about the study: size, design, affiliation, etc.—as well as a clear summary of what this evidence means.
So to summarize: a good version of WebMD would have
structured, quantitative information
real-time updates
summaries of supporting evidence
So why hasn't anyone done this yet?
The short answer: cost.
As I've already mentioned, there's a lot of medical research out there, with more and more piling up every day. And it costs a lot of money to hire a subject-matter expert to comb through mountains of data and write a 2,000-word treatise on zinc supplements. Imagine how much more it would cost to pay this person to organize this data in a structured way, make daily updates, and provide exhaustive evidence to back up their claims.
And these days, ad-based publishers are seeing their margins decline—most of the big online publishers are moving toward paid models instead. Ad-based publishers like WebMD certainly don't have piles of cash sitting around waiting to be dumped into quality research and product development.
But—but!—there are two trends that might just solve this problem:
Automated evidence synthesis is improving rapidly
More consumers are paying for premium content (Substack, podcasts, online news, etc.)
Each of these trends approaches the cost problem from a different direction—(1) means publishers can now provide higher-quality health info at lower cost, and (2) means publishers can make more money from that same health info, which they can plow back into improving the information. Let's look at each of these trends in turn.
Automated evidence synthesis is a fancy way of saying "turning health studies into useful information, automatically." And it's 2022 so obviously "automatically" is a euphemism for "using a computer." We're a pretty long way from total automation, but several parts of the synthesis pipeline have gotten much easier recently, thanks to ✨technology✨.
And the other trend—a move away from ad-based models toward subscription-based models—also seems like a boon for online health information. Not only does this mean that publishers can spend more money on their data—it also aligns the incentives between the publisher and the user. WebMD's primary goal is to increase ad revenue—which means writing lots of SEO fluff and packing it with garish ads for the latest hair-loss pill. "Providing useful, evidence-based health information" is way down the list of WebMD's priorities. But subscriber-supported websites need to be constantly worried about user happiness—one bad experience and a user will immediately cancel their subscription.
Taken together, these two trends make me optimistic about the future of online health information.
So here's the tl;dr, which I've helpfully buried at the bottom of this article:
Online health information is terrible
We should make it structured, timely, and evidence-based
This might actually be possible, thanks to better automation and subscription-based models
And—wouldn't you know it—I was so convinced by my own argument that I made a prototype of "good WebMD," demonstrating that this isn't a pipe dream. It's called GlacierMD and yes I've tried this before.
And it's live! You can check it out at glaciermd.com or you can click this fancy button:
So far it only has data on long COVID, but hopefully it gives you a flavor of what online health information could be. Thanks for reading! If you have any thoughts or ideas on this topic—or some brutal feedback on my prototype—I'd love to hear about it at tom@tjcx.me. Thanks for reading!