Spin Doctoring

Many drugs are dangerous, AIDS drugs in particular. So when a scientific paper's abstract claims great benefits, I am suspicious that the data has been interpreted to give negative research a pro-pharmaceutical spin.

One paper that caught my eye, published in 2003 in AIDS, claimed that the use of protease inhibitors, the AIDS wonder drugs of 1996, caused significantly slower progression in children.

What bothered me is that the researchers claimed the therapy caused slower progression, rather than just being associated with it. This is always difficult to prove, especially now that most clinical trials are more marketing than science. Investigators often have a conflict of interest and placebos are rarely allowed because they would be unethical (but using unproven drugs on trial participants is never unethical).

The most obvious problem with this trial was its inclusion of mothers who injected drugs, a known health risk for both mother and baby. The abstract did not mention this factor; only the text showed babies of drug-using mothers progressed to AIDS significantly faster. The omission of a major factor from the abstract was clearly an attempt to advance the interests of pharmaceuticals and downplay other explanations.

Protease inhibitors are never used alone but usually in combination with two other drugs, most commonly nucleoside analogs, the first AIDS drugs approved for use. Consequently, using them with children creates a burden for the parent. Drugs have to be given before or after eating, at different times of the day, and often cause the child to vomit or spit out the pills or syrup, requiring they be given again to adhere precisely to the schedule.

It is plausible that the complexity of protease inhibitor therapy makes it less likely drug-using mothers would adhere to this regimen, or that doctors would believe it pointless to prescribe them.

I asked the lead author, Dr. Elaine Abrams of Harlem Hospital Center of Columbia University, whether this possibility had been investigated. She replied, "This analysis was not done."

If true, this hypothesis means that only non drug-using mothers administered the protease inhibitors and the outcome of the trial really was that babies of IV drug-using mothers are more likely to be sickly, something everyone already knows.

By spinning the abstract and neglecting to ask a critical question, the authors of this study have spun a haystack of data into a spool of gold for the manufacturers of protease inhibitors.

They're not alone. A 1999 study of papers from top medical journals in 1996 and 1997 found the proportion of inaccurate abstracts ranged from 18 to 68 percent, depending on the journal. Overall 24 percent omitted important information from the abstract and had inconsistencies between the abstract and the body of the paper. Detecting spin doctoring demands a critical eye from the editor of the journals in which these papers are published, as well as the general reader.