Online medical symptom checkers examined

I'm a wreck. I think I might have osteomyelitis, or peripheral neuropathy, or even retrocalcaneal bursitis. I just hope it's not tarsal tunnel syndrome.

I'm also overstressed, overweight, losing my hearing and not sleeping enough. Even worse, I have a low IQ and my "real" age (taking into account my bad habits and so on) is 10 years older than my actual age.

Who needs doctors anymore? We've got the Web.

Yup, I learned all this stuff from visiting online health sites. There are dozens of them out there, and they all promise to help figure out what ails you. Of course, they all feature prominent disclaimers emphasizing that they should be used for information purposes only and you should see a medical professional for any real diagnoses.

But why bother? Even if you could schedule an appointment within the next three months (by which time the symptoms have usually disappeared ... or you have), it's much easier to type in few search terms or answer a few questions and get a list of potential causes of your symptoms.

And, people being people, you'll probably focus on the worst-sounding ailment and decide it's what you have. That's called "cyberchondria" by a couple of Microsoft researchers who have studied health-related search behavior. In their research paper, Ryen White and Eric Horvitz declared, "The Web has the potential to increase the anxieties of people who have little or no medical training, especially when Web search is employed as a diagnostic procedure."

No kidding. If I didn't suffer from anxiety before, I surely do now. Later, I'm going to take an online test to be certain.

Web consultation

My own self-diagnosis research wasn't scientific like Microsoft's. I just checked out five of the most popular (according to Google) health sites to see if they could tell me what's causing pain in my foot. (The sites are presented in the order I tested them.)

But I cheated. I know what I have, because a doctor told me. So I confess, I don't really believe I might have all those conditions listed above -- but I might have believed that if Doc hadn't told me I had plantar fasciitis. That's medical-speak for a strain of the fibrous tissue that runs along the bottom of the foot from heel to toe.

So, to gauge the effectiveness of online health sites, I used that example to focus on "symptom checkers" that typically let you indicate what part of the body is causing you problems and answer some questions to narrow down the list of possible causes.

The results weren't very encouraging. Some sites winnowed possibilities down to a manageable number that included plantar fasciitis, as shown in the video below. But some didn't. And the logic underpinning the decision-tree process sometimes seemed absurd.

Being a layperson (with a low IQ, remember), I asked some bona-fide medical professionals to take a look at the sites and provide their assessments. Helping me analyze the symptom checkers were Dr. Laura Beaty, a family practice physician at Alliance Primary Care in Atlanta; Dr. Andy Spooner, a general practitioner with a specialty in pediatrics at Cincinnati General Hospital; and Dr. Viren Bavishi, a pediatrician at Wheaton Franciscan Healthcare in Franklin, Wis.

Dr. Bavishi said symptom checkers can be a good resource, but their usefulness varies on a case-by-case basis depending on what a patient's symptoms are, so he didn't want to compare them. One common problem he found: "There's a spectrum on what information they're giving you. It could be a lack of information depending on the illness, or it could be far too much [information] for what we're concerned about."

He would prefer that patients come in first so he can figure out what's going on, and then these sites could be "a great adjunct as an education resource. That's where I found it to be useful." But when the sites are consulted first, "I think it causes, in some cases, more stress and anxiety because people end up getting fixated on the worst-case scenario."

"Some of these symptom checkers are great," he said, "but it does sometimes leave out some of the questions or things that we may ask. Some variables that I think are important aren't accounted for."

Dr. Beaty and Dr. Spooner took my approach and worked through each of the sites to evaluate them individually. I conducted my own analysis before speaking to them.

Here's what we found.


My take: This is probably the most popular health-related site on the Web. Google ranks it No. 1. But I found some of that aforementioned absurdity here.

Upon landing on the symptom checker page, I was presented with a list of 76 possible symptoms or an interactive map of the human body that lets you select the bothersome part with a mouse hover.

After answering questions about age and gender (with optional questions about ZIP code and e-mail address, just in case you want another newsletter in your in-box), I eventually found the path to plantar fasciitis by clicking on the map's back view, then "leg" and then "sole."

I was then given a list of 21 possible conditions. A series of questions ensued, designed to zero in on the most probable cause.

Question No. 1 was the first of my problems -- how to describe the pain. I had a hard time deciding if it was:

-- sharp or stabbing

-- dull or achy

-- burning or stinging

-- throbbing

Is "sharp" that much different from "stinging"?

It turns out it didn't make any difference what answer I chose -- the end result was the same for every path I took through the process.

Not yet knowing that, I pressed forward through seven questions (although each one offered the option to "finish") and ended up with a list of 15 "conditions associated with the selected symptoms," with plantar fasciitis as No. 8 -- not that there was any indication that the ordering meant anything.

Starting over and choosing the option to "finish" after the first question -- the pain description question mentioned above -- I was immediately rewarded with a list of the same exact 15 possibilities, but in a different order.

I started over yet again and chose "don't know" or "none of the above" for every single question, and yes, I got the same exact 15 possibilities.

So after choosing "sole" and "pain or discomfort," there was apparently no result possible other than the 15.

Except for one. The only way the list changed was if I indicated that I had suffered a catfish sting.

That's right, question No. 6 was:

Pain or discomfort associated with

-- recent injury or trauma

-- catfish sting

-- none of the above

Choosing option No. 2 added -- you guessed it -- "catfish sting" to the list of possibilities.

So if your sole hurts and you recently tangled with a catfish, you're told that the cause of your pain could be a catfish sting -- or 15 other things.

Excuse me? I mean, how many people have even seen a catfish? I can't figure out why "catfish sting" was one of the options and "stepped on a nail" wasn't, or maybe "walked over hot coals on a bet."

Conditions rounding out the 15 ranged from multiple sclerosis to poorly fitting shoes. Gee, thanks for narrowing it down. If you decide to click on each of the 15 items to get detailed descriptions and try to pinpoint your problem, I hope yours isn't No. 15 on the list.

WebMD does contain plenty of good information and services and looks like it could be quite useful, as do all the other sites. But for quickly and easily figuring out your specific problem through the symptom checker, not so much.

Dr. Beaty's take: She found WebMD and (below) to be very similar. Specifically, "they were literally too inclusive."

Her first test ailment of choice was a urinary tract infection. Going through the WebMD process, she ended up with 16 possible causes. "And then when you increase the number of symptoms -- which obviously in medicine helps you to better isolate the condition -- for these sites, it actually increases how many conditions they are including. So I went from 16 to 21 just by adding a few more symptoms. I thought that was kind of all over the map."

Dr. Spooner's take: He liked the way WebMD used a "Bayesian logic calculator" that lets users add multiple symptoms and tailored its results so, for example, the symptoms "cough and fever" would have a different result than "cough and wheeze." However, the results "are still very narrow. Clearly, these are small databases."

Physicians use a similar tool, he noted, that might come up with 200 possible causes, compared to the much smaller number offered by WebMD and some of the other sites. That helps him consider diseases that he might not otherwise have thought of. He said the professional tool also makes it easy to add many different symptoms and combine them in various ways to reach this large number of possible conditions.

Revolution Health

My take: Here you're greeted with a standard (if rather ape-like) interactive human body map. But it operates a little differently.

Clicking on "Leg, Hip, Knee and Foot" and then "Toe, Foot and Ankle Injuries" brings up an overview page with all kinds of possibilities, some of which are categorized as "acute injury" or "overuse injuries." There are eight main acute possibilities and five under overuse, with plantar fasciitis No. 4 under overuse injuries.

But scroll down and things get weird.

Under a heading of "Check Your Symptoms" are a number of additional questions. One asks "Are you unable to free a trapped foot from an object, such as a pipe, toy or jar?" Another asks "Is an object, such as a nail, embedded in your toe, foot or ankle?" Answering "yes" brings up a page telling you how soon you should contact a medical professional.

I don't know about you, but if my foot is stuck in a pipe or has a nail sticking out of it, I'm thinking my first impulse isn't to go online to see what the problem might be and what I should do about it. I'm pretty sure I could muster up an accurate self-diagnosis here.

Giving credit where it's due, the site does go on to suggest calling a health professional immediately if -- and this is an actual example the site uses -- a high-pressure nail gun fired a nail into your foot. You know, just in case you logged on because you're confused about what to do: "Hmm, I just fired a nail gun into my foot, should I see a doctor or just try to walk it off?"

When I went back and clicked "noninjury," plantar fasciitis was No. 2 in a list of seven items under the heading of "Pain."

Dr. Beaty's take: "This is a more educational site. When you entered the symptoms, it didn't try to lead you to a diagnosis necessarily. It just opened up an entire page of information about -- for example -- urinary issues, so you went through everything and it was just sort of a dissertation on what all the problems were with little links where you could click and go into to get more information. So it was educational, but it would take a lot of time, and I don't think that most people want to spend that much time. Maybe some do."

Dr. Spooner's take: "They offer a very nice decision-tree-based ability to move through their information while paying attention to possibly serious conditions." He added that the site is "pretty much a database of articles with a symptom-based navigator."

Dr. Spooner noted that although the site offered limited combinations of symptoms, there wasn't a "blank box" where a second symptom could be typed in. He used the symptom "headache" to go through the checker and observed that if a child had a headache, there was no way to add the symptom of diarrhea. "It's quite necessary" to know if those two symptoms are being experienced at the same time, he said.

My take: No interactive map here. It starts out with a list of 28 adult symptoms and 17 child symptoms.

Going through the decision tree, I chose "foot or ankle pain" as the symptom. After several attempts to narrow down the final results, I ended up with:

"Located in: heel"

"Triggered by: injury"

"Accompanied by: inability to bear weight"

That brought up seven probable causes and a list of associated factors for each (with your choices in bold). Only three of those seven were included on WebMD's list.

Plantar fasciitis was No. 5 on the list. Clicking on it brought up pages detailing its definition, symptoms, causes, risk factors and other information, similar to all of the sites.

All combinations after choosing "heel" resulted in 11-item lists with just a few differences.

Dr. Beaty's take: She found to present fewer symptoms, but compared to WebMD "it was very similar in that you couldn't narrow down the scope at all, so I'm just not sure how helpful that would be to someone even with a common problem."

She thought both WebMD and let her navigate through the process well. But she was disappointed in both sites when she entered general leg pain as a symptom and neither site had prompts "to find out whether you had a deep venous thrombosis, which is something very significant with leg pain."

One feature that Dr. Beaty liked on the site was that after offering up a possible diagnosis, such as cystitis, "it listed the symptoms so you could go through at that point to see which one best fit you even if the symptom checker didn't get you to the place you wanted to get."

She was amused when she couldn't figure out how to specify female symptoms, "so it gave me prostitis. I thought that was kind of interesting."

1 2 Page 1
Page 1 of 2