Why you shouldn't blindly believe 'medical research'

Research in very young birds and animals has shown that, when light falls on the retina in the eye, it makes the eyeball grow longer. If the eyeball grows too long, rays of light are focused in front of the retina instead of upon the retina. The result, as every student knows, is the common vision disorder known as short sight, or myopia.

Does this effect of light on the eye apply to human infants, too? Twenty years ago, a team of researchers in the USA asked the parents of 479 children about the exposure of these children to night lighting before the age of two years. They found that myopia was present in 9 per cent of children who had slept in the dark, in 31 per cent of those who had slept with a night light, and in 48 per cent of those who had slept with full room lighting.

There was no relationship between myopia and night lighting at later ages.

The study findings suggested the startling conclusion that, because eyelids are not wholly opaque to light, leaving night lights on during the first two years of life may stimulate the growth of the eyeball and result in myopia.

Myopia commonly develops during childhood and adolescence. If myopia can be prevented by switching night lights off, why isn't the whole world aware of this important news?

Within a year of the first study, two more studies were published, again by teams from the USA. Both studies found no association between myopia and night lighting during infancy. Interestingly, both studies found that parents with myopia were more likely to leave night lights on!

So there is the answer. Myopia is a hereditary condition, and the connection between night lighting in infancy and myopia in childhood is heredity because short-sighted parents leave lights on at night so as to see better when checking on or attending to infants. In other words, night-lighting is a red herring.

Expressed in scientific terms, night-lighting was a marker for the higher risk of myopia; it was not the cause. Parental myopia was the cause.

Reading research right

Research is conducted in two important ways. One involves experimentation; the scientist does something and looks at the effects. The other involves observation; the scientist merely examines relationships between subjects of interest in the study.

Cause and effect may reasonably be inferred from certain kinds of experimental studies. For example, if patients who are given a drug become better whereas those who are given a dummy pill do not improve, we might justifiably conclude that the drug is effective. Inferring cause and effect, however, is tricky in observational studies, as the myopia study showed.

Similar considerations apply to all studies of cause-effect relationships in observational research. For example, many studies have shown that elderly depressed patients who receive antidepressant drugs are more likely to suffer a hip fracture. So should antidepressants be avoided in the elderly?

No, because there are many biological mechanisms through which depression can itself increase the risk of hip fracture. So antidepressant use is merely a marker for the risk and not the cause of the risk. In fact, a very recent study from Sweden showed that the risk of hip fracture was actually higher in the months before starting antidepressants than in the months after starting antidepressants. Clearly, antidepressants were not to blame.

Another instance has to with anti-depressants and autism.

More than a dozen studies have found that children of women who take antidepressant drugs during pregnancy are at higher risk of developing autism. This can be interpreted in two ways. One is that antidepressants directly increase the risk of autism. The other is that it is depression and behaviours associated with depression increases the risk of autism, and the use of antidepressant is merely a marker for the depression.

The latter interpretation has merit. This is because studies show that when fathers (and not the mothers) take antidepressants during pregnancy, the child is still at higher risk of autism. This suggests that paternal genes are involved. Studies also show that the child is at increased risk of autism even if women stop the antidepressant long before pregnancy. This suggests that maternal genes, or depression-related changes, are involved.

Studies also show that sedentary behaviour, including TV watching, is associated with poorer health outcomes. Are these studies flawed?

Watching TV and sedentary lifestyles can certainly be markers of health that is already poor. That is, people with symptoms of early, perhaps undiscovered disease may exercise less and watch more TV; so if they do develop a disease, later, their lifestyle may not have been responsible. However, importantly, experimental studies show that changing lifestyle behaviour improves health outcomes. This definitely discourages sedentary lifestyle behaviours.

The bottom line is that research findings should be interpreted with caution. The onus lies on scientists to communicate research findings responsibly, and on the mass media to not sensationalise study results or present research findings out of context.

(Chittaranjan Andrade is Dean (Basic Sciences) and Professor, Psychopharmacology, at NIMHANS, Bangalore)

The views expressed above are the author’s own. They do not necessarily reflect the views of DH.