STANFORD – We are constantly bombarded with information about the purported risks or protective effects of one or another food, dietary supplement, chemical, drug, or activity. In July, for example, an article in The Annals of Internal Medicine reported that people who work at least 11 hours a day have a 67% higher risk of having a heart attack or dying from heart disease than those who work 7-8 hours a day.
However, the study cannot conclude that long work hours cause cardiac disease. In fact, one of the most common misunderstandings about scientific studies is the critical difference between correlation and causation. Working long days might be only a marker for being at higher risk of heart disease. For example, it could be that highly stressed people with Type A personality who think that bacon cheeseburgers and creamy sauces are staple foods, and who already have an increased, long-term risk of heart disease, also tend to work longer hours.
Studies that show an association between a factor and a health effect should be regarded as no more than a preliminary result that points researchers toward further research and analysis. But even professional regulators, who should know better, sometimes get sidetracked by this sort of misapprehension, and overreact.
Last December, the US Food and Drug Administration issued a public Drug Safety Communication about results from a large study conducted in France – the Santé Adulte GH Enfant, or SAGhE, study – which found an increased risk for death with biosynthetic human growth hormone therapy relative to the general French population. There were 93 deaths in the treated group, compared with the 70 expected from a statistical calculation.
One news article reported the study and FDA alert with this headline, “Are children dying for an inch or two?” – language guaranteed to terrify the tens of thousands of patients (and the innumerable body-builders and athletes who use the drug off-label) who have taken the hormone over the past 26 years.
This brings us back to the conundrum of correlation versus causation. The French study compared all patients who took growth hormone to the general population. But the two groups are not at all comparable: kids who receive the drug are far from normal.
Growth hormone is given to children for various kinds of short stature. These include growth-hormone deficiency of unknown cause or as a result of impairment of the pituitary gland’s ability to produce the hormone, owing to traumatic injury or irradiation to treat cancer. It is also administered to children who fail to grow because of chronic renal insufficiency or significant genetic abnormalities, including Turner syndrome, Noonan syndrome, and Prader-Willi syndrome.
These kids are severely impaired. Turner syndrome, a chromosomal abnormality in which all or part of one of the sex chromosomes is absent, manifests itself in several ways. These include characteristic physical abnormalities (most often non-functioning ovaries, which results in the absence of menstrual cycles and sterility) and various other health problems, such as congenital heart disease, low thyroid hormone levels, and autoimmune diseases.
Similarly, Noonan syndrome, a relatively common genetic disorder, is marked by abnormal development in various systems of the body that causes unusual facial characteristics, short stature, and cardiac abnormalities. Prader-Willi syndrome is characterized by poor muscle tone, low levels of sex hormones, and a defect in the part of the brain that controls feelings of satiety or hunger, which leads to overeating and obesity.
Thus, to expect the same death rate in such patients as in the general population would be like blaming insulin for decreased longevity in diabetics (who suffer cardiac and renal disease more frequently than non-diabetics). In fact, it would be big medical news if there were no increase in deaths in patients treated with growth hormone, because that would imply that the drug actually provides a life-prolonging effect in a population expected to die prematurely.
Another finding in the SAGhE study arouses skepticism about a cause-and-effect relationship between the drug and the increased death rate: the mortality came from causes that are mechanistically and physiologically quite different – mainly bone cancers, cardiovascular diseases, and cerebrovascular events. A more narrow, unimodal manifestation of toxicity would be more persuasive in establishing causation.
American regulators jumped the gun. The European Medicines Agency was more measured, saying in a statement, “Based on this observational study alone, the risk cannot be associated with certainty to the growth hormone treatment. The results need to be confirmed and complemented with further analyses.”
There is a reproducible, undeniable association between fires and fire engines, but that doesn’t mean that the latter cause the former. If you or a loved one has taken (or is taking) human growth hormone under a physician’s supervision, you shouldn’t lose sleep over it.
Henry I. Miller, a physician and molecular biologist, is a Fellow in Scientific Philosophy and Public Policy at Stanford University’s Hoover Institution and a fellow at the Competitive Enterprise Institute. He was the founding director of the Office of Biotechnology at the FDA.
Copyright: Project Syndicate, 2011.