When going through journals, many busy pediatricians scan the abstracts and look at the conclusion. In the process, however, they may miss some red flags.
Reading a journal article is a lot like examining a patient. You need to take a systematic approach, said Thomas Newman, M.D., M.P.H., FAAP, chief, Division of Clinical Epidemiology, UCSF Department of Epidemiology and Biostatistics, San Francisco.
Dr. Newman will describe how to take a step-by-step approach to reviewing randomized, double-blind trials during a session titled “Don't Believe Everything You Read: the Impact of Industry Funding on the Clinical Research Literature (F1141).” The session will be held from 3-3:45 p.m. Saturday in Room 309 of Moscone South.
The first thing Dr. Newman suggests is to look at who sponsored the study. If it is a pharmaceutical company, a closer look at the research question, subjects, drug doses, outcome variables and results can be illuminating.
He will present statistics showing that most industry-sponsored studies focus on drugs for chronic diseases of the wealthy as opposed to diseases that cause more morbidity and mortality in developing countries.
“Even if someone gets very sick, if they’re only sick for a short time and you treat them and they either are cured or they die, there’s not much money in that,” Dr. Newman said. “People go where the money is.”
Industry-sponsored studies also tend to include subjects who are most likely to benefit from the drug and are least likely to have adverse effects. For example, studies in adults tend to exclude the elderly and those with co-morbidities, even though they will be the ones taking the drug, he said.
The next thing to look at is the comparison group. If you want to make your drug look good, you compare it to something that’s not very good like a placebo or the least effective drug at the lowest possible dose. If you want to make your drug look safe, you compare it to a drug with the most possible side effects, said Dr. Newman, a member of the AAP Section on Epidemiology, Public Health and Evidence and the Council on Quality Improvement and Patient Safety.
For example, in a trial of an expensive new drug for constipation, the comparison group received a placebo.
“But there are a lot of laxatives out there that we know work,” Dr. Newman said. “So the question is if you pay hundreds of dollars (for the new drug), do you get something that actually works better than mineral oil?”
Study authors also can present their results in many different ways. For example, do they focus on statistical significance or clinical significance?
“If you’re a drug company, you can study thousands of people and have a little tiny p value for an effect, which is not really clinically important,” Dr. Newman said.
Or, the study has multiple outcomes and only the ones that turn out good are highlighted and the rest are ignored.
These issues don’t necessarily mean that trials sponsored by industry are worthless.
“The most reliable indicator that you can trust the results is if they come out unfavorable to the funder’s product,” Dr. Newman said. However, such studies are rare, partly because they can be stopped early and buried rather than published.
“Results that favor a sponsor’s product can also be informative, but a much more critical reading of the methods and results is required,” Dr. Newman said. “Especially for industry-sponsored research, you can’t just rely on the last sentence of the abstract.”
For more coverage of the AAP National Conference & Exhibition visit http://www.aappublications.org/collection/cme
