Drowning In Risk Factors…

As I discussed previously, we don’t have a clear idea of the causes of many of the degenerative diseases we face today, like heart disease and cancer; all we have are risk factors. In some cases, we have a lot of risk factors. For example did you know there are now over a thousand known risk factors for heart disease? But which of these thousand risk factors actually cause heart disease? All of them? None of them? Why don’t we know what the causes of these diseases are yet?

Unfortunately, these diseases are extremely difficult to study because their causes are non-trivial. The biggest difficulty in studying them is they don’t develop overnight. Some scientists believe that diseases like heart disease and cancer can take 20+ years to develop. So how can we possibly know which combination of factors during all that time caused or contributed to the disease? Hmm…

Observational Studies

The field of epidemiology has been around for hundreds of years. Originally, it was used to study outbreaks of infectious diseases (i.e. epidemics). Today it’s a staple of scientific research, and there are thousands of studies published every year. Unfortunately, as Dr. George Davey Smith and Dr. Shah Ebrahim discuss in their editorial “Epidemiology — Is It Time to Call It a Day?“, the track-record for observational studies correctly identifying the causes of degenerative diseases is less than stellar. In fact, not only do they often find spurious associations, but sometimes they find the opposite of the correct answer. For example, as detailed in the New York Times article “Do We Really Know What Makes Us Healthy?“, doctors for many years prescribed hormone replacement therapy to older women to reduce their risk of heart disease, only to find out recently that it actually increases their risk.

The problem is observational studies are inherently limited. Observational studies involve monitoring a large group of people for some period of time, asking them questions or giving them medical exams at regular intervals, and recording mortality and disease rates. But how often do they monitor the people in these long-term studies? Certainly not every week. In the famous on-going Nurse’s Health Study, for example, they send out questionnaires every two years. Even if you’re monitoring something simple like birth control usage, how do you boil all the variations or permutations of potential usage down into a multiple choice question?

And after the researchers have collected their data, they plug it all into a computer and hope to find correlations. The better studies (prospective studies) as least decide what they’re testing before running the study so they can attempt to minimize confounding factors in the design of their study. The more dubious ones (retrospective studies) go looking for associations in data from studies designed for completely different purposes. Sandy Szwarc from the Junkfood Science blog has a long list of examples of this type of weak “data-dredge” study.

By definition, observational studies are largely uncontrolled (i.e no intervention and no randomization) and thus will have many confounding factors. This is where the judgement and biases of the researchers come in. It’s possible to get different results depending on how the researchers manipulate or “interpret” the data. If the associations being studied are small, the adjustments made to the data can easily skew the data to reveal associations where there are none.

And even if the studies find compelling associations, they’re still just associations. We don’t know why the association exists. They may be causative or could simply be correlated.

In the end, observational studies are like excavators. If you’re looking for big obvious things, like rocks, you’ll find them. But if you’re looking small subtle things, like fossils, you’re going to miss them. It’s the wrong tool for the job.

Clinical Trials

So if observational studies aren’t good at discovering the causes of diseases, what about clinical trials? Clinical trials avoid a number of the limitations of observational studies by design. The best clinical studies are double-blind so neither the doctor nor the patient know whether they’re in the study group or the control group. Clinical studies also are typically short and small in order to control for as many confounding factors as possible.

But if it can take 20+ years to develop these diseases, how much is it going to cost to run clinical studies for that length of time? This, of course, ignores the problem that people won’t like strict controls on how they live their lives for long periods of time either. In the old days, they experimented on patients in mental hospitals since they were a captive audience and not likely to object; but we can’t do that anymore…

Also, some things, like diet, aren’t suitable for double-blind studies because people know whether they’re eating low-fat or low-carb foods, for example, which introduces an additional set of confounding factors.

Clinical trials are useful for testing new drugs because they can use placebos to make the study double-blind, and the experimenters can focus on one specific variable. But for monitoring lifestyle or other human behaviors, clinical studies don’t seem particularly suited…

Biological Research

Our best hope for really understanding these degenerative diseases is to understand them at the molecular level – by understanding the biological mechanisms at work in our bodies. Observational studies and clinical trials can help to at least point us in the right direction since biological research is very expensive and time-consuming. I’ll explore some of the findings of current biological research in future posts.

But what can we do in the meantime? Who should we believe? That’s the hard part…

[Content © 2008 SorryToConfuseYou.com, All Rights Reserved.]

Causality Versus Correlation

We hear regularly about the “causes” and “preventions” for various diseases from the media. For example, high-fat diets cause this, and high cholesterol causes that, or as I discussed last week, coffee helps “prevent” heart disease. But do we actually know whether these things cause or prevent these diseases? If you read any of the major health sites you might think so. But if you read them again more carefully, you’ll notice they don’t actually talk about causes, they talk about risk factors. Take the American Heart Association web site for example:

Disease	Causes	Major Risk Factors
Heart Disease	?	Smoking, high cholesterol, high blood pressure, lack of exercise, obesity, poor diet
Type 2 Diabetes	?	Smoking, high cholesterol, high blood pressure, lack of exercise, obesity, poor diet
Stroke	?	Smoking, high cholesterol, high blood pressure, lack of exercise, obesity, poor diet

So what’s the difference between a “cause” and a “risk factor”? Risk factors sound like risks, which are bad, right?

Let’s look at an example. What’s the leading cause of car accident mortality?

• Reckless driving? Nope.
• Speeding? Nope.
• Getting distracted? Nope.
• Bad weather? Nope.

None of these things actually cause death. Car accident mortality is usually caused by bodily harm incurred when we’re introduced to Newton’s First Law of Motion, i.e. “A body in motion will remain in motion unless acted upon by an outside force”, like a tree or another car!

Now you might argue that this is splitting hairs, but my point is that we need to be careful when talking about cause and effect. In order for something to be causative, we need to be able to link the series of events together with a clear relationship between them. For example:

• Serious bodily harm usually causes death because we lose too much blood and our internal organs fail, or our internal organs become too damaged to function

Straightforward, right? So back to our secondary factors:

• Speeding can increase the probability of an accident, and thus death, because you and the other drivers have less time to react to avoid a collision
• Bad weather can increase the probability of accident, and thus death, because reduced visibility / traction can prevent you from avoiding a collision

Still straightforward, right? If we analyze car accident rates, we’d expect that drivers who speed or drive recklessly would have higher mortality rates. We start with a plausible cause, and we can verify effect. If necessary, we can test our hypothesis in a controlled environment.

But say we notice that drivers with red cars also have higher mortality rates. Now what? We have effect, but what’s the cause? Working backwards is dangerous – just because something is correlated, doesn’t make it causative. For example, what happens if we ban red cars? Will we have fewer car accidents? Not likely. In this case, red cars may be correlated to personality types that are more likely to drive recklessly, rather than being causative in any way. Some people are going to drive recklessly regardless of car color. So red cars may not cause increased mortality, but they are a risk factor for increased mortality.

All of this is intuitive for car accidents, but what about diseases? If we look back at the American Heart Association web site information from above, how many causative factors are listed? None. How many correlated factors are listed? All of them. Oops…

Digging around a little, I couldn’t find any major health sites that said XXX causes YYY, besides smoking causing lung cancer. But some certainly imply causation:

“Extensive clinical and statistical studies have identified several factors that increase the risk of coronary heart disease and heart attack.”

Risk factors increase the risk of heart disease? I don’t think so… What they should have said is “they’ve identified several risk factors that are associated with an increased risk of heart disease”. It’s a subtle but important difference.

“People with diabetes are two to four times more likely to develop cardiovascular disease due to a variety of risk factors”

“Due to” implies “because of” – which certainly implies cause and effect. Hmm…

The definition of a risk factor is:

“A risk factor is a variable associated with an increased risk of disease or infection. Risk factors are correlational and not necessarily causal, because correlation does not imply causation“

So by definition, risk factors don’t cause anything. If diet and obesity were causative, they’d say so – but they don’t. So why are they trying so hard to get us to control risk factors that may or may not be causes? That’s a topic for a future post…

Risk factor != cause

Also, don’t assume that major risk factors are more likely causative than minor risk factors. “Major” simply means that they’re more strongly correlated. For example, yellow teeth are a major risk factor for lung cancer. So yellow teeth cause lung cancer?

[Content © 2008 SorryToConfuseYou.com, All Rights Reserved.]

Science + Media = Bad News

The media seems to love scientific studies. Every week there seems to be a story on the local news about a new study telling us about how to reduce the risk of heart disease or cancer, or some other disease. Unfortunately, the news story invariably distorts the study’s results, doles out some dubious advice, and does a general disservice to their audience…

For example, recently I happened across a WebMD article entitled “Drinking coffee may extend life”. The first paragraph of the article reads:

“Coffee drinkers, rejoice. While you might be using it for a “pick-me-up,” coffee may also be extending your life.”

From this headline, your initial reaction might be:

Coffee’s good for you; it would probably benefit everyone to go out and drink coffee.

Unfortunately this assumption is pretty far from the truth. For example, further down in the article you get a quote from one of the study’s authors:

“We can’t say from this one study that coffee extends your life, but it does appear that it doesn’t increase the risk for death for people who are healthy,”

Wow, coffee doesn’t increase the risk of death for people who are healthy! That’s encouraging…

If you look up the actual study in the Annals of Internal Medicine – “The Relationship of Coffee Consumption with Mortality”, their official conclusion is:

“Regular coffee consumption was not associated with an increased mortality rate in either men or women. The possibility of a modest benefit of coffee consumption on all-cause and CVD mortality needs to be further investigated.”

Hmm… We seem to be getting further and further away from our initial assumption; maybe the devil really is in the details. Looking at the details of the study, I think an appropriate summary of their findings would be:

For those with no history of heart disease or cancer, over 20 years, your risk of dying of heart disease may drop by up to 20% in men and 25% in women, if you drink coffee.

The study doesn’t tell us anything about:

• Benefits to those that already have heart disease or cancer
• Benefits to those who aren’t at risk of heart disease
• Whether coffee increases the risk of dying of something else
• Whether coffee may negatively affect your health in other ways

So, now that we’ve correctly interpreted the study’s results, do we actually believe them? I don’t…

Why does coffee reduce the risk of heart disease? We can guess, but in reality, we have no idea whether coffee actually lowers the risk of heart disease or not, or whether these numbers are due to something else. The authors of this study don’t either. They weren’t working in a lab looking at samples through microscopes, they were crunching numbers in a computer. As I’ll discuss further next week, this is not an effective way to study diseases.

Bottom line, take your coffee (and the results of studies such as these), with a grain of salt…

[Image: www.stockvault.net]

[Content © 2008 SorryToConfuseYou.com, All Rights Reserved.]

What Does Life Expectancy Represent?

Every year, the new life expectancy figures come out to grand fanfare. “Life expectancy is now XXX, up YYY years from last year”, but what does this number actually represent? Is it meaningful? You might think you know, but you’re probably wrong. I know I was…

I became curious about life expectancy statistics because my grandfather died a few years ago at the age of 94. It got me thinking, how long were people expected to live when he was born? Given all the scientific advances we’ve made in the last hundred years, how long did people of his generation actually live? Did he outlive his expected life span?

Life Expectancy vs. Expected Lifetime

Life expectancy for 2005 in the U.S. was 77.8. So, the assumption is that, on average, anyone born in 2005 can expect, on average, to live to the age of 77.8, right? Wrong.

In mathematics, expectancy is a term used to describe the expected future value of something, based on current knowledge. For example, if you flip a coin 10 times and you win $1 for heads and lose a $1 for tails, the expectancy for this game is $0 because you expect, on average, an equal number of heads and tails.

So life expectancy then is the expected average lifetime of someone born today, based on what we know today – i.e. the death rates for today. The way they calculate this (and have since the 17th century) is by using life tables. Life tables give you, based on your current age, the probability of living one more year. For example, they tell you that if you were 39 in 2004, that you had a 99.82% chance of making it to 40.

The problem is, to calculate the 77.8 life expectancy figure for 2005, they’re applying those chances of someone making it from 39 to 40 in 2005, and projecting them 39 years in the future to calculate future survivability. So, 77.8 is really a fictional number that says that for someone born in 2005, if death rates at a given age were to remain as they were in 2005, this is how long they’d live on average.

We can probably all agree that it’s likely a lot of things are going to change in the next hundred years that will impact mortality rates. For example, there may be some big scientific breakthroughs on cancer or heart disease, there could be another world war or infectious disease pandemic, or we could be attacked by aliens…

Either way, life expectancy is not the same as expected lifetime.

Life Expectancy vs. Actual Lifetime

So if life expectancy is a projection of present knowledge on future longevity, we should be able to see how far off these figures were. For example, life expectancy for someone born in 1900 was 49.2; what was their actual average lifetime?

As I was trying to find actual longevity information, I came across a text book called “Development Through Life” by Newman & Newman that said that current life expectancy values are calculated using life tables, whereas life expectancy values for someone born in 1900 are now based on actual rates of death. This completely confused me because it implied that life expectancy was a dynamic number. That somehow life expectancy values from previous years would be recalculated using actual death rates. This would also imply that comparing the life expectancy for someone born in 1900 vs. someone born in 2000 would be completely bogus because we’d be comparing apples and oranges.

Fortunately, this assertion doesn’t seem to be true. Going back to the life tables from the CDC published in 1900, the numbers appear the same as the numbers reported for 1900 today, so I’m not sure what Newman & Newman were talking about…

In the end, I couldn’t find the numbers I was looking for, so I calculated them myself. To estimate the actual average lifetime for someone born in 1900 you can incrementally use all the life tables produced over the last hundred years because those life tables are calculated based on actual death rates for that year. For example, if you managed to survive to the age of 40, then the life tables from 1940 for someone aged 40 will tell you what the chances were of living to the age of 41. The biggest limitation to this approach is we can’t control for things like immigration and migration. Oh well, c’est la vie.

From my calculations, the average lifetime for someone born in 1900 was actually something like 55 years, rather than the original life expectancy of 49.2. Notice that over time, average lifetime seems to consistently be about 10 years longer than life expectancy.

[Note that the lines converge because as we get closer to the present, more and more people are still alive, which means I had to use more and more life table data rather than actual death data. As we progress into the future, I expect the average lifetime to retroactively creep as people continue to live beyond their original expectancy (as least as long as life expectancy continues to rise).]

Life Span

Life span usually refers to the maximum lifetime of a species. So, given that life expectancy is going up, you might ask if human life spans also being extended? It doesn’t appear so. The maximum life span for humans still seems to be about 110 years.

So the fact that my grandfather lived to the age of 94 wasn’t really statistically extraordinary; it was just a lot less likely than me living to the age of 94.

Usefulness?

So now that we understand life expectancy, what can we do with it? You and me personally, probably not much. The fact that overall life expectancy was 77.8 years in 2005 is pretty meaningless because it’s a composite number for everyone alive in 2005. It’s not that useful for a new baby, because the figures don’t attempt to predict the future. And it’s not relevant to someone born 40 years ago because how do you combine it with the life expectancy from 40 years ago? The number might as well be 75, 80, or 85.

But the change in the expectancy figure year to year can tell us something. Namely, if life expectancy is going up, it tells us that we’re all currently, on average, living longer – which is probably a good thing (assuming we have reasonable quality of life).

What about the fact that the life expectancy for someone who was 80 in 2005 was 9 years? Medical science isn’t likely to change significantly in the next 9 years, so the number is at least relevant. That said, your remaining lifetime at that age is probably more heavily affected by your current health rather than these general averages, so maybe that’s not all that useful either…

What about retirement? Can you make financial plans using these life expectancy numbers? If you do, you may be in for an unpleasant surprise… If the current trend continues, life expectancy will continue to under-estimate actual lifetimes. Life expectancy was off by 6 years for people born in 1900. How far off will it be for your generation?

[Content © 2008 SorryToConfuseYou.com, All Rights Reserved.]