Fighting the Number-One Killer In the US With Data 121
mattydread23 writes "Often, the signs of eventual heart failure are there, but they consist of a lot of weak signals over a long period of time, and doctors are not trained to look for these patterns. IBM and a couple heathcare providers, Sutter Health and Geisinger Health System, just got a $2 million grant from NIH to figure out how better data analysis can help prevent heart attack. But the trick is that doctors will have to use electronic records — it also means a lot more tests. Andy Patrizio writes, 'What this means is doctors are going to have to expand the tests they do and the amount of data they keep. Otherwise, the data isn't so Big.'"
Don't believe the salesman's hype (Score:5, Informative)
It is a hypothesis that collecting more data will find a pattern that will predict heart failure earlier, and that will lead to earlier interventions.
They haven't demonstrated that it works.
In order to demonstrate it, they have to do a controlled trial. They have to use these data collection systems in a group of 5,000 patients, and use the usual methods in another 5,000 similar patients, and see if there's any difference in a meaningful outcome. Do the patients live any longer? Are they any less likely to get strokes?
Sometimes it works, sometimes it doesn't. The New England Journal of Medicine just published a report on the use of a high-tech surgical intervention -- implanting cardiac resynchronizing devices in a new subset of heart failure patients. http://www.nejm.org/doi/full/10.1056/NEJMoa1306687 [nejm.org] It turned out the resynchronization patients had more deaths than the control group, and they stopped the study early. You don't know until you've done the randomized, controlled trial. That's the method of science, the experimental method. You take your brilliant ideas and put them to a test.
That's science. Everything else is bullshit.
There was a study of using an electronic medical record in a pediatric intensive care unit. The patients with the EMR had a higher death rate than the control patients. The doctors said that when they needed to write a prescription in a hurry, they would just take out their Rx pad and write it. When they needed to write it with the EMR, they had to sign in, go through screens, and find what they were looking for.
EMR replaced a simple, effective system -- paper and pen -- with a more difficult system. What's the point?
Read what doctors are actually saying about electronic medical records, http://www.kevinmd.com/blog/ [kevinmd.com] http://www.nejm.org/ [nejm.org]
There are systems that actually make it easier to treat patients. As I understand it, the Veterans Affairs and Kaiser Permanente have systems that actually collect useful data. The Scandinavians have great useful databases. http://www.bmj.com/content/347/bmj.f5906 [bmj.com] But a lot of the new systems, particularly the ones that are merely being installed because they're required and subsidized under new federal regulations, are driving doctors crazy. They complain that they have to log in, go through screens, fill out checklist after checklist, and wind up with records that go on for hundreds of pages that nobody ever looks at again. Traditionally, on paper, they were forced to write a concise narrative for their colleagues and themselves, of useful information that got to the point and helped them make a decision about what to do next. These poorly-designed EMRs stopped forcing doctors to think. It simply forced them to collect a lot of data. Data isn't information. Useless data is noise.
And maybe most of all, they complain that instead of looking at their patients, they're looking at a computer screen. If you have to tell somebody that he's going to die in 6 months if he doesn't stop smoking, you shouldn't be looking at your computer screen. Maybe there's an element of human communication that computer nerds don't appreciate.
In any computerized records, there's a tradeoff between how much data you collect, and how much time you have to spend entering data. You can spend an extra hour a day just entering more data. Is this pill a tablet or a capsule?
And more important than time, when you write a medical record, you should be filtering information for just the important information. Otherwise you're just adding noise to the record, and making it harder for the humans to spot patterns.
If you want to prevent heart failure, the basic job is to stop smoking, lose weight, and exercise. When patients get outside of certain well-understood parameters, you can give the
Re:Waste of money (Score:4, Informative)
Utter rubbish. The French eat meat and have a high fat diet, but have a very low incidence of heart problems
Your link is to a fad-diet site.
I think they consume far less sugar and soda and more fiber, which may explain things. You might find these two (long) talks interesting. The first is research/study based, the second is more anecdotal, but with some research. I thought they were both excellent and interesting.