Case: a paper that was retracted

In 1998, Andrew Wakefield and colleagues published a paper in The Lancet about 12 children with bowel problems and developmental disorders. It suggested a link with the MMR vaccine. The paper caused years of fear and a drop in vaccination. In 2010, The Lancet fully retracted it. Here is what a careful reader could have seen in the paper itself.

Warning signs visible in the paper

• Tiny sample: 12 children. That is far too few to establish a link to anything.
• No control group: there was no comparison with children who were not vaccinated. Without it, you cannot tell whether the problems were more common after the vaccine.
• Timing is not cause: many developmental problems become noticeable at the age when the vaccine is usually given. Close timing is expected by chance alone.
• Claim bigger than design: a case series can describe cases and raise questions. It cannot show cause and effect. Much of the public reading of the paper went far beyond what the design could support.

What came out later

Investigations found that the children had been selected in misleading ways, data had been misreported, and there were undisclosed conflicts of interest. Wakefield was struck off the UK medical register. Large studies of hundreds of thousands of children found no link between MMR and autism.

The lessons

1. Match the claim to the design: case series raise questions, controlled studies test them.
2. Check the sample size and the comparison group before anything else.
3. Read the conflicts of interest statement.
4. Check whether a paper has been retracted. Publishers mark retracted papers, and the Retraction Watch database tracks them.

RememberA paper in a top journal can still be wrong, or worse. The design, the sample and the comparison tell you more than the journal's name.