Googlediagnosing
Running was, apparently, Sam’s thing. I don’t know exactly why he runs–he’s a a friend from the neighborhood where I grew up, and we haven’t kept in touch–but he runs, and he loves it. So it was a huge, huge loss when he developed pretty severe pelvic pains every time he tried to run. Specialists one through four recommended rest, but Sam knew there had to be something better. So he turned to Google.
It turns out Sam had bilaternal sports hernias, also known as athletic pubalgia. (I had never heard of it, and most doctors haven’t, either.) He typed in his symptoms, queried away, and found news articles, papers, and PDFs on the condition and its treatment–surgery, performed by one lone sports medicine surgeon in Pennsylvania. (Apparently the intake form has a box, “Check here if you’re a professional athlete,” reassuring Sam that the condition exists in the world outside of textbooks and medical journals.) Sam underwent the surgery, the surgeon repaired the hernias, and he’s now recovering and excited to hit the ground running. Pun intended.
That’s the good–no, the great–side of Google Medicine. Enter symptoms in one end, spit out a diagnosis on the other. In no way is this a new idea: Meg Hourihan’s mom starting Googlecooking in 2002 –you just enter what ingredients you have at home into Google, and the search engine spits out a list of recipes. If I’m trying to find the name of a song on the radio, and I only know bits and pieces of the lyrics, I’ll type them in, add the word “lyrics,” and I’m almost certain to find my match. Googlediagnosing isn’t that far behind. The BMJ recently had an article indicating that more medical journals are getting hits from Google Scholar than they are from PubMed.
In some cases–like Sam’s–this is great. Fantastic. Revolutionary. If I have a very specific set of symptoms–tooth itch, purple-hazed vision, and new-car-smell breath (the classic triad for the Grahamazon syndrome, of course), I can probably figure out a likely diagnosis, see my doctor, and get a treatment (dihydrogen oxide as needed). Hooray. In the most common of syndromes, this also works well. Runny or stuffy nose, fatigue, muscle aches can be a pretty common presentation of the common cold, and we see that this works pretty well –most of the top links have to do with the common cold or the flu. Common things being common, there are probably more webpages out there about common colds than say, dengue fever, and since Google results are essentially a web popularity contest, the common things rise to the top.
Again, this is great. Patients can get information about their possible diagnosis, read about over-the-counter treatments, when to seek medical attention, etc. But there are chasms of caveats to Googlediagnosing. In a world wide web of blogs and forums and comments, anyone can be anyone, and everyone can have an answer. And if Google ranks its information by popularity and not by accuracy , you can end up with some pretty mixed results. Let’s say I have fevers, chills, and muscle aches –holy shit! I have babesiosis? Or Rocky Mountain Spotted Fever? Or leptospirosis? No, I probably have the flu–but not according to Google.
When everyone’s an expert, no one is. I used to frequently respond to Ask Metafilter’s health section (it’s a website where members can post health-related questions), but I’ve stopped outright because there’s just so much bad advice from everyone and their mother. I’ve seen so many users give anecdotal evidence (“my great aunt had the same symptoms you’re describing and ended up having a pheochromocytoma–you should get an MRI before it spreads!”) that’s so scary and harmful and wrong that they must not even know what they’re doing. In other categories of questions–science, travel, computers, etc–I give my anecdotal advice or opinions as well, but I’m not potentially screwing with their mental or physical health.
And there’s potential for great anxiety when Googlediagnoses are incorrect–enlarged lymph nodes and a fever may net a person a diagnosis of lymphoma or HIV, instead of sore throat. The two diagnoses are pretty damn different. (HIV is the latest in a long line of “great imitators” like syphilis and TB that can present in many, many ways.)
Most diseases also present with a specific time course and other symptoms–that physicians are trained to recognize because of their experience seeing diseases so many times–and so often a patient may come in with a diagnosis that, while symptomatically correct, doesn’t fit the way the disease works, the age group, the time course, etc.
Add to this the fact that most patients don’t understand the concept of a differential diagnosis (I didn’t until I entered medical school), and you can get people posting that someone with syptom X definitely has disease Y, or you can get a patient who has done some self-Googlediagnosing, and is adamant that the disease they read about is the one they have. (Or even the disease Y they think they have is treated with drug Z, so they want a prescription for it.)
Another barrier exists between doctor and patient: medicalese. This is the ever-popular “diaphoresis” instead of “sweating,” “erythema” instead of “redness,” “edema” instead of “swelling.” If a patient doesn’t know the lingo, he or she may have the signs and symptoms right, but may lack the ability to learn about his or her condition. How many patients of yours would call a low platelet count a thrombocytopenia? Or how many refer to their diuretics as water pills? If you’re using the wrong keyword, you’ll get the wrong diagnosis–or miss the right one–every time.
Now, there are websites and companies and organizations that try to provide some sort of standard for health information on the web. And there are symptom to disease tools like Dxplain and Gideon that allow people to enter symptoms (both medicalese and non-medical) and, using algorithms and data, generate a list of possible diagnoses. But all of these are still far from perfect.
If we continue to move toward more and more Googlediagnosing (and it certainly looks that way) these tools and standards will be vital. Like the safer-sex mantra, if you’re going to do it (uh, Googlediagnose), do it safely.
Update: My wish at their command. Healthline is a search engine that tries to filter out the bad health info on the net.