What doctors don't know about the drugs they prescribe | Ben Goldacre

Summary

The speaker discusses the issue of publication bias in scientific research, particularly in medicine, where only positive results are often published while negative or neutral findings are ignored. This skewed representation of data can have serious consequences, such as misleading doctors and patients, wasting resources, and even costing lives. The speaker highlights several examples, including failed cancer studies, dangerous heart drugs, and misleading antidepressant trials. He argues that to fix this problem, all studies, including negative ones, must be published to ensure transparency and true evidence-based medicine.

Takeaways

• 🔮 Precognition claims, like Nostradamus' predictions, are often flukes, and the same bias exists in academia.
• 📊 A study by Daryl Bem on precognition in undergraduates was published, but replications with negative results were rejected, showing publication bias.
• 🧬 Cancer research studies are often unreplicable. A study showed only 6 out of 53 cancer research trials could be replicated, indicating biased results.
• 💔 A drug trial for lorcainide, an anti-arrhythmic drug, was not published, leading to over 100,000 unnecessary deaths due to similar drugs being prescribed.
• 🧑‍⚕️ Publication bias exists in medicine, too, leading to dangerous outcomes. Studies that yield negative results often go unpublished, misleading doctors and patients.
• 💊 Reboxetine, an antidepressant, showed positive results in published studies, but many negative studies were hidden, misleading doctors and patients.
• 🔎 Studies show that half of medical trials, especially on antidepressants, go unpublished, leading to a misleading picture of drug efficacy.
• 📉 The systematic problem of publication bias affects evidence-based medicine, distorting the true effects of treatments and interventions.
• 🦠 The stockpiling of Tamiflu was based on incomplete trial data, with many trials unpublished, leading to false conclusions about its effectiveness.
• 🚨 Current efforts to fix the problem, such as trial registries and FDA regulations, have not been fully effective, with many trials still unpublished or improperly registered.

Q & A

Who is Nostradamus, and why is he mentioned at the beginning of the speech?

Nostradamus is a famous historical figure known for his prophecies. He is mentioned to illustrate the concept of predicting the future, which the speaker uses as a metaphor for unreliable predictions, even in serious fields like academia and medicine.

What is the central issue the speaker is addressing regarding academic research?

The speaker highlights the problem of publication bias, where positive results are more likely to be published than negative or null results, leading to a distorted view of research outcomes and misleading conclusions.

What was the significance of the study conducted by Daryl Bem, and how does it relate to publication bias?

Daryl Bem's study found evidence of precognitive powers in students, but the speaker suggests it was likely a fluke. The study was published, while replication attempts showing negative results were not, illustrating how positive results are favored in academic publishing.

How did publication bias affect cancer research, according to the 2012 study in *Nature*?

Researchers tried to replicate 53 cancer studies and found that only six were replicable. The inability to replicate the majority of studies suggests that only successful or 'freak' results were published, while unsuccessful ones were left out.

What is the problem with not publishing negative results in medical research?

Not publishing negative results leads to the development of treatments or drugs based on incomplete or misleading data. This can result in harmful consequences, as seen with heart drugs like lorcainide and reboxetine, where incomplete data led to unnecessary deaths.

How did the failure to publish early negative results on lorcainide impact patient outcomes?

The failure to publish early results showing that lorcainide increased the death rate contributed to the later development of similar drugs, which ultimately caused over 100,000 unnecessary deaths due to increased heart attack complications.

What role does reboxetine play in the speaker’s argument about publication bias?

The speaker uses reboxetine as a personal example of how incomplete data can mislead doctors. While published studies showed it was effective, many unpublished studies revealed it was less effective than other antidepressants, misleading the speaker into prescribing it.

How does publication bias distort the understanding of medical evidence, according to the FDA study?

The FDA study found that while trials for antidepressants showed a 50-50 split between positive and negative results, the published literature only included a small fraction of negative trials, giving a false impression that these drugs were more effective than they really were.

Why does the speaker refer to publication bias as a 'cancer at the core of evidence-based medicine'?

The speaker calls it a 'cancer' because it systematically undermines the integrity of medical research by creating a skewed view of drug efficacy and safety, which can lead to ineffective or harmful treatments being widely adopted.

What steps does the speaker suggest to fix the problem of missing negative results in research?

The speaker advocates for mandatory publication of all trials conducted in humans, including older trials, to ensure a complete and accurate understanding of the effectiveness of treatments and medications.