Authors:  Eldar Shafir (Princeton University); Donald A. Redelmeier (Sunnybrook Health Sciences Centre)
Title: “Post Hoc Bias in Treatment Decisions”
Journal: JAMA Network Open

The widespread use of unnecessary or dubious medications, including assorted vitamins and so-called “alternative” treatments, to alleviate diverse vague symptoms has numerous unfavorable effects, says Shafir, the Class of 1987 Professor in Behavioral Science and Public Policy, a professor of psychology and public affairs, and the inaugural director of Princeton SPIA’s Kahneman-Treisman Center for Behavioral Science & Public Policy.

These include significant financial costs, considerable waste, “lots of magical thinking, distraction, and, perhaps most importantly, the occasional medical neglect and missing of potentially serious diagnoses that later become incurable,” he says.

With such dangers as a backdrop, Shafir and Redelmeier set out to test whether patients would wrongly attribute a marginal improvement in symptoms to their scientifically questionable treatments, and continue those treatments as a result.

The researchers presented a variety of scenarios – involving treatments ranging from sugar supplements and horse shampoo to vitamins and antibiotics – that gauged patients’ attitudes towards questionable medications following a brief improvement in conditions. They even went so far as to have actors ask pharmacists for advice on whether to continue a questionable medication following slight improvement.

Shafir and Redelmeier found that patients were more likely to continue the treatment – and pharmacists more likely to recommend that they continue the treatment – when their initial care was followed by minor improvement.

“Although a modicum of enthusiasm is natural,” Shafir and Redelmeier write, “slight improvement after treatment is not necessarily an improvement because of treatment.”

The researchers call this a post hoc bias – a belief among patients that because their symptoms improved slightly after treatment, the improvement was caused by the treatment.

“Deciding to proceed after a slight and brief subjective improvement can lead to the long-term use of ill-advised medication,” they write.

The bias Shafir and Redelmeier describe can contribute to several harmful effects, including complacency; a failure to seek alternative, possibly more effective treatments; and shortfalls in care. In addition, it can lead patients to prematurely halt the use of a potentially appropriate treatment when improvement is not immediately apparent. Finally, it can lead to the under-use of preventative medication when patients fail to note the lack of adverse effects – a lack resulting from that earlier treatment.

“An awareness of the bias will not make it disappear,” Shafir notes. “However, it might help if we all remind ourselves that minor subjective improvements provide limited information and merit considering alternative explanations.”

The simple implication is that patients should refrain from pursuing questionable treatments following momentary feelings of improvement.

“But,” Shafir adds, “‘people’ are not easy to reach. Part of our message is for physicians to be attuned to such proclivities and to discuss these with their patients.”