Recall bias: why memory shifts after a diagnosis

Recall bias is a systematic difference in how accurately or completely people remember and report past events or exposures. It is a particular problem for studies that ask people to look backward, since two groups being compared can remember the same kind of past exposure with different levels of care, not because their memories are equally poor, but because their circumstances give them different reasons to search their memory harder.

The clearest setting is a case-control study, where people who have a condition, the cases, and people who do not, the controls, are both asked about their past. A parent of a child with a birth defect, for instance, may have spent months turning over every choice made during pregnancy, while a parent of an unaffected child has no comparable reason to have rehearsed the same period of their life. The cases end up reporting more of their past exposures simply because they looked harder for them.

Two traps follow. First, recall bias can manufacture an association between an exposure and an outcome that does not really exist, purely because the affected group remembered and reported the exposure more completely. Second, it is easy to confuse with a real effect, because the reported difference in exposure looks exactly like the pattern a true cause would produce, and the two can only be told apart by checking memory against an independent record.

When a study's findings rest on people's memory of past behavior or exposure, especially one comparing people with a condition to people without it, check whether the exposure was confirmed against records made at the time, such as medical charts, pharmacy data, or purchase history, rather than relying on recollection alone. Pair any claim built on retrospective recall with a look at whether a prospective study, one that measured exposure before the outcome occurred, reaches the same conclusion.