Simply put, positive results are good business. There is money to be made every time a new drug or device is brought to market. It
draws public interest and makes for good news copy. And it is something that most consumers and patients are hungry for. Wouldn't it be great if we could
lower our blood pressure, narrow our waistlines, increase our energy levels, elevate our moods, and prolong our lives simply by taking some new pill or
making use of some new medical device?
Similar incentives apply to researchers. The careers of physicians and scientists depend in part on how much interest we generate in our discoveries and
how much research funding we can attract. Generally speaking, positive results are far more likely than negative ones to result in a presentation at a
scientific meeting or a publication in a scientific journal. Major awards are virtually never presented to researchers for negative results, including even
results that contradict previous positive reports. Egaz Moniz received the Nobel Prize in Medicine for developing a form of frontal lobotomy, but the
researchers who later showed its poor benefit/risk ratio were not similarly recognized.
In some cases, teams of researchers vie with one another in pursuit of statistically significant results. When this happens, the probability that at least
one team will come up with positive results is increased, largely because unlikely results become more likely as we increase the number of trials. If one
person flips a coin, the probability of coming up with four heads in a row is very low. On the other hand, if 100 people flip a coin four times, there is a
good chance that at least one person will get four heads in a row. If only the people who get positive results report their findings, it will appear as
though the probability of getting four heads a in a row is far higher than it really is.
Likewise, a drug may have little or no beneficial effect. But if a sufficient number of researchers conduct trials to assess its efficacy, and if these
studies are designed in such a way that the probability of a positive result is high (for example, by looking for only a very small effect), then there is
a good chance that some researchers will be able to report positive results. In such situations, which are relatively common, the probability that positive
results will be found may be not only high but greater than that of negative results. As Ioannidis has demonstrated, the probability of reporting false
results often exceeds that of true results.
In most cases, there is probably little harm in eating more foods that are high in antioxidants, making use of nutritional supplements such as fish oil, or
taking an extra vitamin tablet each day. We may be wasting money, but the sums involved are generally not high, and it is quite possible that we are
achieving some benefit from the placebo effect associated with believing that we are taking good care of ourselves. But when such drugs and devices cost
large sums of money and carry substantial risks, we should think twice and then think again before jumping on the parade of fads that often characterizes
the health beat.