Medicine can never be like Facebook, despite what Matt Herper argues over at Forbes. Perhaps he was just trolling for hits on a day when everyone is thinking about the Facebook IPO, but Herper proposed, with apparently seriousness, that medicine needs to model itself on the tech world in order to match the kind of progress– and profits– of a Facebook. But the medical news this week provided ample evidence why this will never happen. Biology is much more complex and resistant than the digital world.
For a medical journalist like myself this was a frustrating week. There were a whole bunch of large, major studies on important subjects published in top journals. But the take-away message from these studies, both individually and combined, is that achieving any kind of real progress in medicine is incredibly hard.
Let’s take a quick look at these studies:
1. Coffee in the New England Journal of Medicine: Despite some of the breathless news reports, some of which erroneously claimed that the study proved that drinking coffee can extend your life, this large study added little or nothing new to our knowledge about coffee. Even the editor of the journal, Jeff Drazen, acknowledged the limitations of this sort of study. The simple truth is this: although coffee is ubiquitous and has been the subject of hundreds of different studies of all different types and designs, we will almost certainly never learn to any degree of certainty whether coffee is good or bad for us. An enormous, decades-long randomized controlled clinical trial, which is the only possible way to ascertain the truth about coffee with any degree of certainty, would be nearly impossible to perform, for multiple reasons.
I don’t want to overstate my pessimism here. I think there is a much more limited lesson that can be derived from this NEJM study and the rest of the coffee literature. From the totality of the evidence it seems highly unlikely that coffee has any large effect, either positive or negative, on important outcomes like mortality or cancer. But we’ll never know for sure about small effects, and we will certainly never know if there are small populations or individuals who are particularly likely to derive benefit or harm from coffee.
2. HDL Cholesterol in the Lancet. In some respects the HDL cholesterol story is exactly the opposite of the coffee story. Unlike coffee, the epidemiology of HDL is clear-cut, and therefore the reverse association of HDL with cardiovascular disease is among the best established facts in all of medicine. But association is not causation, and despite more than a generation of intense research we still don’t know how– or even if– HDL works. In fact, as its name implies, high density lipoprotein is not so much a biological entity as an artificial construct of something that we can measure easily.
I don’t want to get bogged down in the details of the Mendellian randomization studies reported in the Lancet, but there are a few conclusions that we can take from the paper. First, a simple measurement of HDL may be meaningless by itself, and a therapy that raises the HDL number will not therefore produce a beneficial effect. This doesn’t mean that one day HDL won’t produce a billionaire on the scale of Mark Zuckerberg. It’s still entirely possible that research into HDL will identify a pathway or a compound that will have effects comparable to or perhaps even greater than the statins. But, unlike Facebook, this won’t happen overnight. There will be a long, hard slog through the mud. There will be no 20-year-old HDL multibillionaires, but there may be a whole lot of middle-aged researchers, businessmen, and investors. And you will “like” these guys for better reasons than a cute cat picture.
3. Azithromycin in the New England Journal of Medicine. This is another demonstration of the limitations of our knowledge. Azithromycin is one of the most popular antibiotics of all time: as heavily promoted as a sugar-coated cereal and routinely given for no good medical reason whatsoever. Now we learn that it may have induced lethal arrhythmias in a few people, especially those already at high risk for cardiovascular disease. Further, if you read the NEJM article carefully, you’ll learn that we don’t have a very good idea about the cardiovascular effects of most other antibiotics.
In his Forbes article Herper writes about the efforts to encourage innovation in developing new antibiotics. I can’t argue with him about this, as we desperately need new antibiotics. But we will still be left with this azithromycin-type problem. There will be no easy shortcuts to find and identify rare but serious adverse effects. The biology is just too complex.
4. Statins For Primary Prevention in the Lancet. Our knowledge about statins is unmatched in the medical literature. This latest report from the the Cholesterol Treatment Trialists’ (CTT) Collaborators includes data from more than 175,000 patients, and quantifies to a very precise level the relative and absolute benefit of statins in most populations. So this paper represents one of the great triumphs of contemporary medicine.
And yet, despite the massive evidence base, there is a battle raging within the medical community over who should get statins. The main part of the debate centers on whether statins should be used for primary prevention, in people who don’t already have, or who don’t know they have, cardiovascular disease, in order to prevent future cardiovascular events. In this case, the problem isn’t about there being enough data. The problem is about how to interpret the data or, in other words, the values and perspectives we bring to the data. Should we treat people for diseases they don’t have but might develop? How do we assess the risks and benefits of such treatments? Why should someone who is not sick be treated like a patient? And here, I’m afraid, we’ve reached the frontier of biology and medicine, since these are questions that science and medicine will never be able to answer. As I’ve tried to show, the limitations of our medical knowledge are already severe, but as the statin story demonstrates, even when we have an overwhelming amount of data it’s not always easy to achieve consensus and certainty.