Recall bias can occur whenever an attempt is made to collect data retrospectively. Human memory is imperfect. Some events are more likely to be remembered than others; some are remembered incorrectly; some are forgotten altogether. For example, when bragging about my golf game, I find it easier to remember a recent excellent round of golf and tend to forget about the (more numerous) bad rounds.
Recall bias is especially problematic when the outcome of interest is likely to influence memory. For example, a cancer patient may remember more exposures in their life that they consider “toxic”, but forget about the many other life experiences that could actually have contributed to their diagnosis. Similarly, a cancer patient may be more likely to remember their visit to a (boring) nuclear power plant than a person without cancer.
Although the focus is often on the patient as the source of bias, recall bias can also be provoked by the researcher. Unblinded researchers might ask certain patients the same question multiple times, or in multiple different ways, if they expect a specific answer.
Recall bias is especially problematic in an trials that rely on self-reporting, such as some case-control and retrospective cohort studies.
Blinding and prospective collection of data are important tools in avoiding recall bias.
This post is part of a series of posts on bias in medical research. You can find the whole bias catalogue here.
Sackett DL. Bias in analytic research. Journal of chronic diseases. 1979; 32(1-2):51-63. PMID: 447779