Bad Health Reporting Associated With Higher Risk of Blogging

Last month a published systematic review concluded that there is no evidence that treating high blood pressure down to levels lower than 140/90 has any benefit. It is also true that a blood pressure of 115/70 is associated with a lower risk of cardiovascular "events" compared with a blood pressure of 140/90. These are not contradictory.

Another recent study showed that high cholesterol in midlife is associated with a higher risk of dementia later in life. Does this mean that if you are middle-aged and have high cholesterol, taking cholesterol-lowering medication will help prevent dementia? No.

If this is confusing, blame bad health reporting, combined with an American medical tendency to treat symptoms, which in itself is a symptom of a larger malaise of failing to focus on the bigger picture (overly specific focus while engaging in healthcare). More on that in a bit. First I want to share some exciting news from my lab, the Institute for the Advancement of Idiotic Analogies (I am the founder and CEO).

I have just concluded an observational study that shows visiting convenience stores is associated with higher risk of mortality from cancer. The effect is exposure-dependent, showing a positive correlation between frequency of visits and cancer mortality rates. Clearly this means that avoiding convenience stores will lower risk of cancer death, right? Well, I must sadly admit that my study shows nothing of the kind. Since it's an observational study, if something interesting has come of it, I should use that to generate hypotheses, then design studies to test them. Or I could just go to an uncritical health reporting pool and let them breathlessly warn against the dangers of convenience-store proximity. If I'm honest, which, unfortunately, I am, I also must admit that visiting convenience stores correlated positively with purchasing and, more importantly, smoking, cigarettes. So it's likely that the root cause of any increase in mortality is due to smoking and not due to the convenience stores themselves. In other words, frequency of visiting convenience stores is simply a marker that correlates with something else, and not a cause. So banning convenience stores, or following government advice to limit exposure, would actually not affect outcomes at all.

Likewise, is high cholesterol a cause of dementia? (No evidence of that so far. In fact, low HDL correlates with higher risk of dementia as you get older.) Or is whatever is causing high cholesterol also the cause of dementia? (Much more likely I think.) In that case, treating the cholesterol and not the root cause would likely have no effect, or possibly make things worse (side effects of drug treatment, ignoring causal factors because of focusing on a single measure of risk). Likewise, high (or low) blood pressure can be both a direct cause of some bad effects, but also an effect of some other cause. That's why bp-lowering drugs might be of no benefit, or even be harmful, when lowering bp to a level which in a non-treated person shows reduced risk of some poor health outcomes. Really, this should be pretty simple, it's amazing how many reporters don't get this. Suppose you are hypertensive because of stress. Stress has many effects on the body, not just blood pressure. But blood pressure is easy to measure and easy to focus on. So if you focus on one effect, and, say, take ACE-inhibitors to lower blood pressure, will that help? Maybe a bit. What if you focused on lifestyle changes to reduce stress, achieving the same lower blood pressure (without medication)? Seems like a better state, since you've addressed the root cause, and mitigated more effects than just the one you've measured, right?

Too narrow a focus on specific outcomes can be a dangerous medical myopia. If a drug decreases heart attack mortality you may find cardiologists prescribing it, even if it increases cancer mortality at an equal and offsetting rate. If sunlight exposure increases risk of skin cancer you may find dermatologists telling you to avoid it, even if it decreases risk of much more common and much more fatal internal cancers. Cardiologists don't care what you die of, as long as it's not cardiovascular disease. Dermatologists don't care what you die of, as long as it's not a skin disease. (Personally, avoiding a heart attack is not necessarily a big win if I instead die of cancer after developing dementia.)

Gosh, I've really babbled on here. Lemme sum up a bit.

• observational studies can be interesting, are great for generating testable hypotheses, but do not show cause
  
• risk association does not mean cause
• directly treating a risk measure does not necessarily mean you are doing anything preventive nor even lowering your risk
• most journalists, and even some doctors, don't understand those three points