Is Too Skinny More Dangerous Than Too Fat?

Obesity is a top cause of health problems. But is being thin a free ticket to health or could it kill you sooner than being fat?

Can Skinny Be More Lethal?

A while ago I wrote about how lazy people of normal weight are healthier than active overweight people. That invites a conclusion that looks logical on the surface: thinner always is better.

Researchers from Toronto came to a different one. One that says skinny can be more lethal than obese.

But is it really? Or is this another case of bad reporting?

The Risk Of Dying Is 1/3 Higher

Let’s start at the beginning. The researchers compared the results of 51 studies that each tracked people for five years or longer. Then they looked at how their Body Mass Index (BMI) influenced their health. The outcome looked like this:

• People who were underweight (BMI equal to or lower than 18.4), had a 1.8 times higher risk of dying than those within the normal range of 18.5 to 24.9
• Obese persons (BMI of 30 to 34.9) had a 1.2 times higher risk than those of normal weight
• Severely obese people (35+ BMI) had a risk 1.3 times higher

The Study’s Shortcomings

That’s the gist media outlets report about this. They missed that the study’s main aim was showing other researchers how important it is to use statistical tools correctly (and, by Jupiter, a lot need to learn that).

Hence the title “J-shapedness: an often missed, often miscalculated relation: the example of weight and mortality” and an abstract that reads like this:

We present three considerations in analysing the association between weight and mortality, as well as other relations that might be non-linear in nature. First, authors must graphically plot their independent and dependent variables in a continuous manner. Second, authors should assess the shape of that relation, and note its shape. If it is non-linear, and specifically, J-shaped or U-shaped, careful consideration should be given to using the ‘best’ statistical model, of which multivariate fractional polynomial regression is a reasonable choice. Authors should also refrain from truncating their data to avoid dealing with non-linear relations.

That’s Greek to most and has awfully little to do with body weight. In simple words it means they wanted to show how to properly calculate correlations.

Causation Is The Important Thing

“Correlation” means “when A changes, B does as well. ” What it doesn’t mean is “any change in B is caused by A.” That would be “causation.”

The Toronto guys didn’t find causation, because they never looked for it. When they examined mortality, they only vaguely accounted for mentally unstable people or alcohol and drug abusers. Which is fine when all you want to show is the proper use of statistical analysis.

But within those groups you find underweight persons aplenty, and their health conditions or deaths are not caused by their underweight per se. Rather, the low body weight and mortality is a result of the drug abuse.

Dr. J over at CalorieLab also wrote about this paper and had the chance to ask lead researcher Joel Ray about how these factors were accounted for. Dr. Ray replied:

Smoking was well controlled for, but alcohol intake to the degree that it should be (was not), nor was mental illness. As they are prevalent, the (added) consideration of alcohol overuse and mental illness would likely attenuate the underweight effect.

There Could Be A Relation

I’m not doubting that we are on to something when investigating a possible sour relation between underweight and health. A very skinny person’s body has fewer reserves to cope with stress and illness.

But this paper is bad evidence for that. It does a splendid job showing how important it is to do your statistics right – which is what it was meant to. It would be really bad if someone took it as an excuse to stay fat.

Picture courtesy of Tony Alter.