I am occasionally asked for advice on keeping up with the tsunami of medical information while separating facts from spin. Here are my tips.
Timing is everything. Take a look at a few journals the moment they come out. JAMA Internal Medicine puts new articles online on Monday; JAMA Oncology on Thursday; and the New England Journal of Medicine at 5 PM ET on Wednesdays. Pick some journals that you enjoy and check them online the moment new articles drop. Why does timing matter? Keep reading.
Don't read every article; triage them. Reading every article is boring and unsustainable. Instead, I triage. First, I skim the titles. There are some topics and authors that I am a sucker for. I am also a fan of correspondence. I read all of the back-and-forth for papers that previously made my cut.
Next, I focus on the original articles. Do any concern my practice? If so, I glance at the methods: observational or randomized? Sample size greater or less than 100? Multicenter or single center? Control arm or none? Phase 1, 2, or 3? I don't draw a firm line, but larger, multicenter, late-phase randomized trials stay on my screen longer.
Then I start asking questions: What was the effect? Was it for a clinical endpoint or a biomarker? Was the magnitude of the change statistically significant and clinically meaningful? Will it matter to patients? Was the control arm appropriate? Was the use of, or the ban on, crossover correct? (If you're curious about this, read more here.)
If I see anything I don't like, I move on. Unproven surrogate endpoint—skip. Statistical sliver of difference—skip. Straw man control arm—skip. Misuse of crossover—skip.
The word decimate originally meant to eliminate 1 in 10. Here, you will be lucky if 1 in 10 remain. Whatever articles survive get a deep read.
The deep read. The first level of depth is reading the paper and rolling your eyes when you hear the voice of the medical writer. I hear their voice when I see negative findings being spun as positive, when I see subgroup or secondary endpoints emphasized over primary ones, when everything is "well tolerated" no matter how toxic, and when a lousy drug or device "still has a role" and is "just a matter of patient selection." If a new drug fails to show a survival benefit (without crossover) in a highly lethal setting, but that is only because "long post-progression survival and salvage agents" prevented the drug from doing so, I cringe. For the record, if your new, costly drug can't show a survival benefit in a lethal condition against a control arm of old drugs, that isn't a good thing.
The second level is reading the supplement. Look at the figures and tables. Count things in your own mind: the dose of drugs given; the rates of and size of dose reductions and discontinuations. For Kaplan-Meier progression-free survival (PFS) curves, I check how many patients were censored at every time interval. The Lancet makes it easy. For others, I eyeball them. Imbalances, particularly early on, can be problematic. I put my right hand over the Kaplan-Meier curve when less than 10% or 5% of the people who started in that arm remain at risk. That part of the curve is more uncertain.
The third level is the full audit. Compare the paper to Clinicaltrials.gov. Compare it to press releases and abstracts. Look up the drug on Drugs@FDA.gov. Read the package insert. Compare Section 14 of the drug label to the paper, and finally, read the medical and statistical reviews from the US Food & Drug Administration (FDA), which are free online. What does the paper claim that the FDA is not allowing the sponsor to claim? That is where things get interesting.
You may be worried now. Who has time for this? First, remember: This is for papers that made your cut. These will be very few. Second, the more you do it, the quicker it gets. What starts off taking 1 hour can take 10 minutes with practice. And third, you have to do this. CME run by experts taking tens of thousands of dollars from the sponsoring company isn't going to do it for you. The FDA stamp of approval doesn't mean what it used to. This is the job of the 21st century physician.
Participate in the discussion. Timing is everything because of what happens in the hours and the days after the results drop. Doctors will be talking about this paper. The first volley will be on Twitter. Follow doctors, statisticians, patients, and others in cancer medicine, and see what they think of the study in real time. Days later, blogs and podcasts will discuss the topic. My podcast Plenary Session attempts to take a deep dive on at least one oncology paper each week. Some of these posts will be more insightful, fresh, interesting, and thoughtful than the editorial that accompanies the original study. I didn't mention editorials until now, and I didn't forget about them.
At first, it's okay to simply listen to others on social media, but over time, muster the courage to chime in. If you think a trial is good, give your 2 cents and say why. What's stopping you? If you get corrected, so much the better—it's how we learn.
These are just some of my tips. They are not based on randomized trials. If you reflect upon your life, you will see that many decisions are not based on such studies—for instance, how often you brush your teeth, eat food, use the restroom. At the same time, this recognition does not diminish the importance of randomized controlled trials for adjudicating which costly, invasive, harmful medical practices actually work and are worth it. But it takes a good reader of randomized controlled trials to make the final call.
In the 21st century, we have to become lifelong learners who check for ourselves whether new "game-changers," "miracles," and "cures" are as good as advertised.
Medscape Oncology © 2018 WebMD, LLC
Any views expressed above are the author's own and do not necessarily reflect the views of WebMD or Medscape.
Cite this: Vinay Prasad. 21st Century Physician: Triaging the Tsunami of Medical Information - Medscape - Sep 25, 2018.
Comments