August 21, 2010

In Practice: Dr. Google has mixed results

Using the search engine can be a boon for physicians — if they do it right.

By Rahul K. Parikh, Special to the Los Angeles Times
August 23, 2010

During an otherwise unremarkable afternoon at the office, I was sitting at my desk plowing through paperwork when a colleague came knocking — and forever changed how I practice medicine.

"Hey, Rahul, what's McMurray's sign?" he asked.

It is said that a physician needs to carry some 2 million facts to practice medicine. Though I knew that McMurray's sign had something to do with examining a patient's knee, it was one of the 2 million facts that, like some tattered scraps of paper pushed to the back of a file cabinet, I had long since forgotten.

I shrugged and turned toward my bookshelf to pull down a decade-old volume of Nelson's Textbook of Pediatrics. Somewhere in that tome was the definition of McMurray's sign and how to elicit it on exam.

Just as I was reaching for the book, I had another idea. I turned back to my workstation, clicked on that little "e" to bring up my Internet browser and went to Google. I typed "McMurray's sign" in the search box and hit "enter," and up popped several links to YouTube videos. We clicked on one and, like two medical students learning the basics of our craft, watched the short mini-lecture that showed an instructor moving the knee joint to check for a positive McMurray's sign. And with that, my colleague was off to try it on his own patient.

Eureka. No more pencils, no more books. With a computer and Google (sorry, Bing, statistics show that Google is where we go first and frequently) in your office and your exam rooms, it's a whole new world — for better or worse.

I've used Google regularly in my practice for the last four to five years, and statistics suggest I'm not alone (though it's Google's data, so it's perhaps best taken with a grain of salt): Eighty-six percent of doctors say they now regularly use the Internet on the job. Of that group, the majority start at Google, which they use as a springboard to look for general information about diseases and drugs.

And why not? As writer and physician Atul Gawande likes to point out, in medicine today there are 13,000 diseases, 6,000 drugs and 4,000 medical and surgical procedures. Even for the best and most motivated of physicians, it's a Sisyphean battle to keep up. Having the Internet at my fingertips makes me a better doctor, though I'll admit that sometimes it feels a bit like cheating on an exam.

Use it judiciously

I was sitting with the parent of a new patient a while back. As she was filling me in about her son's medical history, she told me his father has a history of retinitis pigmentosa. Like McMurray's sign, it was a distant but familiar phrase. As I was nodding my head listening, I stopped typing into his chart, brought up Google and typed in "retinitis pigmentosa."

A mere 0.36 seconds later, I got links 1-10 of 866,000. I clicked on a Medline link and glanced over some bullet points to see if I needed to send this child for any screening tests. Because, as I read, the "signs and symptoms often first appear in childhood, but severe vision problems do not usually develop until early adulthood," I knew I just had to make sure I did his vision test annually. Didn't miss a beat.

Most doctors seem to agree that using Google to educate yourself or patients makes sense (for a pediatrician like me, that means showing parents pictures of different diaper rashes and kids pictures of poison oak so they don't walk into a plant for the ninth time this summer). But things get trickier — and more controversial — when doctors start using Google to make diagnoses.

There's a now famous story first told in the New England Journal of Medicine about a trainee in allergy and immunology who was presenting a case to some colleagues. The audience included a visiting distinguished professor in the same field. The case was of an infant with diarrhea, an unusual rash and multiple immunologic abnormalities. The attending physicians and house staff discussed several diagnostic possibilities, but no consensus was reached. They were stumped. Finally, the visiting professor asked the trainee if she had made a diagnosis. She reported that she had: The child had a rare syndrome known as IPEX.

The distinguished physician asked the young trainee how she'd figured this out.

"Well, I had the skin biopsy report, and I had a chart of the immunologic tests. So I entered the salient features into Google, and it popped right up," she was quoted as saying. The author of that letter to the New England Journal, who was present at the meeting, wondered: "Are we physicians no longer needed? Is an observer who can accurately select the findings to be entered in a Google search all we need for a diagnosis to appear, as if by magic?"

He needn't worry. It turns out that we're not all as good at Googling as that young doctor was.

In a 2006 study published in the British Medical Journal, researchers had physicians read the histories of 26 tricky cases published in the New England Journal of Medicine and enter search terms into Google to see if they could make the diagnosis. They nailed it 58% of the time. Not bad, but not much better than the flip of a coin, either.

I tried this experiment myself with a handful of those New England Journal cases — I didn't get a single one right. So my advice is: Unless you're Internet savvy enough and have time to scan pages and pages of websites, skip Googling for diagnoses and just call your nearest specialist for help.

How scholarly?

Another of Google's potential pitfalls has to do with Google Scholar — a product the company launched in 2004 that searches academic publications and websites.

A colleague of mine recently declared that this was now his go-to site for medical research. And, in fact, more visitors were led to biomedical journal websites by Google Scholar within a year of its release than by PubMed, the free Internet database published by the National Library of Medicine and the standard-bearer for doctors and others looking for academic publications.

But at least one study, in the Journal of the Medical Library Assn. in 2007, suggests that Google Scholar may not be the best choice. In a head-to-head comparison of Google Scholar and PubMed, Scholar searches yielded significantly more total citations of primary research articles than did PubMed, suggesting it's less specific than PubMed.

Researchers also found that Google Scholar brings older articles to the top of the results. This is fine if you want a history lesson but problematic if you're looking for cutting-edge clinical data. (It turns out Google Scholar places more weight on studies that are cited more, regardless of their validity.)

Plus, the researchers concluded, users often don't search more than five pages to answer a question, leaving knowledge susceptible to the most popular articles, not necessarily the best.

This study raises an important point about Google, and really all search engines. These are businesses whose success largely rests on proprietary math and technology — algorithms, which in Google's case, are trade secrets, no different than the formula for Coca-Cola. As such, what you find on the top of the page depends on an often mysterious process.

That doesn't mean Google is doing something bad (the company's motto is "Don't be evil"). But it does mean users lack the kind of transparency some of us need when we place our professional faith in a search engine.

The solution for doctors (and everyone), I suspect, is best answered by taking note of that lack of transparency and by making it a professional priority to become search savvy. A growing number of medical professionals have suggested this as well.

Searchers also can use Google — and other sites — more narrowly. That little "advanced search" button next to the big search box, for example, can narrow the results, as can using quotation marks for exact phrases and excluding words by putting a minus sign in front of them. There are lots of tips about how to do this floating around on the Web (just Google "tips to using Google").

This little bit of expertise, especially in a world in which medical knowledge exponentially outpaces our ability to know it all, is a valuable skill set for young doctors to gain.

Parikh, a Walnut Creek, Calif., physician, writes the Vital Signs medical column for Salon.com. http://www.rahulkparikh.com/.

Source

No comments:

Post a Comment