The Missing Link

by Howard Fienberg
October 31, 2002

Schizophrenia is a frightening mental disease, afflicting more than 25 million Americans. New drugs can treat the problem with fewer side effects than ever before, but recent research has shown these medications to have their own disastrous side-effects, including diabetes and even death. Does this mean the cure is worse than the disease? Given certain deficiencies in the data, there is no reason to think so.


The research alleging the danger in these medicines has been spearheaded by Dr. Elizabeth Koller, a former Food and Drug Administration (FDA) endocrinologist. Although working outside of her area of specialty, and in a private capacity, Koller's FDA affiliation elevates the stature of her work and gives it increased publicity. She has published numerous letters to the editor in major medical journals since last fall, outlining the results of her work. Combining published case reports with FDA database records, she found many patients taking the newest anti-schizophrenic drugs - such as olanzapine (sold as "Zyprexa") and clozapine (sold as "Clozaril") - were developing diabetes of varying severity or seeing their pre-existing diabetes worsen within months of starting to take the drugs. She also discovered that some of the diabetes cases improved after patients stopped taking the drugs.


Medical science needs this kind of research, but it does not prove the drugs are dangerous, especially considering the dubious nature of Koller's data. Aside from scattered published reports of individual cases, Koller relied upon the FDA's MedWatch database, which compiles reports of "adverse events" tied to pharmaceuticals. These reports are submitted by physicians and companies on a voluntary basis and usually lack detailed background information, like patients' risk factors for diseases. The resulting data can suggest possible problems in new drugs, but there are many idiosyncracies in the database that can lead researchers to false patterns. Small changes in drug-related publicity, user demographics or how manufacturers and physicians record and report "adverse events," can lead to much larger changes in patterns of supposed side effects.


The reported "adverse events," of course, may have absolutely nothing to do with the drugs. The database lists reports of bad or strange things that happened to patients after they started taking a drug. It is all too easy to fall prey to the fallacy post hoc ergo propter hoc - just because one event occurred after another event does not mean that the former caused the latter. There may be no link whatsoever.


In order to make comparisons, which might show if users are more or less likely to develop diabetes, we need a "control" group (a group of similar patients not taking the drug). But it turns out that sufferers of schizophrenia as a whole are two to four times more likely to develop diabetes than the general population. This was observed for decades before the arrival of pharmacological treatments. No one is entirely certain why, but the cause may lie in the tendency for schizophrenics to lead sedentary lifestyles, leaving them prone to risk factors for diabetes, such as poor nutrition, obesity and smoking. Schizophrenics are also generally in poor health and often miss out on regular medical care. That is why Dr. John Buse, a diabetes expert and chief of the Division of General Medicine and Clinical Epidemiology at the University of North Carolina, suggests that "there may be simply a higher risk of those people developing diabetes than the general population, regardless of their medication profile."


To date, we don't have the kind of in-depth research required to say for certain, one way or the other. The National Institutes of Mental Health is currently funding a large research project, looking at most of the new anti-schizophrenic drugs, their interactions with other drugs, their side effects, and so forth, using long-term randomized controlled clinical trials - the scientific gold standard. Unfortunately, good science is slow and careful, and results from the study won't be available for several years.


Koller effectively concedes all these points. She admits that "causality cannot be ascertained" from her investigations, but offers them as a warning and a spur for further research. Her work may be an important part of the scientific research process, identifying possible trends, but it is of minimal use to the public. Supposition and preliminary observations are not particularly enlightening to the average schizophrenic, who would simply like to know if the new drugs will give them diabetes or worsen their current cases of it.


In May, the Medicines Control Agency in the United Kingdom and the Health and Welfare Ministry in Japan both issued warnings about the possible link between anti-schizophrenic drugs and diabetes. Some people here think we should follow their lead, but that would probably be an over-reaction given the paucity of evidence available. The newest medications have proven to be the most efficient way to treat schizophrenia with the fewest associated side effects. The risks involved in taking them appear insignificant compared to the risks of leaving schizophrenics to suffer needlessly.

See the original:

return to Howard Fienberg's page