Thursday, June 7, 2012

Is Exercise Bad For You?

It’s well-established at this point that news outlets rarely report positive medical findings with the same enthusiasm as negative medical findings, much in the same way that a young man helping an elderly woman across the street doesn’t garner as much attention as a burglary or fire. Unfortunately, when people get their medical information from local or national news, it often means that doctors end up having to do a lot of explaining.

An example of this phenomenon happened this past week when the New York Times reported that for some people, exercise worsens cardiac risk factors like blood pressure and triglycerides. Of course, this flies in the face of our current understanding of lifestyle and its effects on improving health, and it has some docs worried that their patients may let up on exercise. There are two points that this article has brought up for me, and I hope you find them of help in interpreting this result, a result the lead author termed, ‘bizarre’. (If you're interested, here's the original article)

The first is that, in research, we are constantly bombarded by conflicting information, and that this information needs to be analyzed and understood by clinicians in order to make sense. There’s a somewhat dated expression from the world of computer science that I like to use in this situation: ‘Garbage in, garbage out’. In computing, this means that erroneous entries produce erroneous results, but I mean it to mean that disorganized data results in disorganized opinions. Like the blind men describing the elephant, scientists using separate samples of data will reach differing conclusions, and the data doesn’t organize itself – it’s up to us to gather all of this information and synthesize it into a coherent whole.

When we have data that conflicts with prior information, it’s usually telling us something that we ought to know, and were previously unaware of, but rarely does it supersede the other information or somehow overturn it. Thus, the elephant is in some circumstances is like a rope, in others like a wall, and in others like a tree trunk, but it’s always a large grey mammal. Similarly, this information about exercise is telling us something that we didn’t previously know – what exactly it is hasn’t yet been determined, but in no case does it invalidate the information we already have that in most cases, exercise is linked to decreased risk of hypertension, heart disease, diabetes, depression, cancer, and overall mortality.

The second point to bring up in this discussion is that of outcomes versus surrogate markers. In practicing medicine, our goal is always the preservation of the patient’s health and the alleviation of suffering, because these are the things that happen in real life and that matter to patients. In research terms, they’re outcomes – they are the measurable real life effects we aim to achieve. However, research based strictly on outcomes is expensive, time-consuming, and in certain cases, frought with ethical concerns (think about who would willingly take placebo compared to a potentially life-saving substance). To overcome these obstacles, we instead shoot for shorter goals – these are surrogate markers. Surrogate markers include things like cholesterol, blood sugar, blood pressure, and other similar lab markers or physical exam findings. We use these markers to measure someone’s likelihood of reaching one of the real-life outcomes we’re trying to prevent.

Unfortunately, though these are very useful tools, the link isn’t always perfect. A classic example is that fish oil has a somewhat paradoxical effect of slightly raising LDL cholesterol, but significantly lowers one’s likelihood of having a coronary event. Similarly, many people have escaped having a stroke despite ragingly high blood pressure. In analyzing this data on exercise, then, we have to ask ourselves about markers versus outcomes – this study shows that cardiovascular risk markers are adversely affected in a minority of people who start exercising, but on the other hand, many other studies show that exercise is associated with significantly lower risk of having a cardiovascular event. In my book, outcomes trump markers.

In the end, this information is probably going to tell us something useful. Perhaps a minority of people have a transient elevation of certain markers when they start exercise, that is followed by a long-term lowering of risk factors. Perhaps there is a small subset of people who need specially-tailored exercise programs that won’t cause this elevation. As the authors point out in the paper, we really don’t know what this information means. In the interim, I’m going to continue plugging for the benefits of exercise, because while I believe that important information will come from this surprising finding, I don’t think it’s going to overturn the consensus that exercise is beneficial for the vast majority of people.