“Painful Memories” – How Memory Biases Affect Patient-reported Outcome Measures

Memory is a funny thing.  People are surprisingly bad at remembering past events, both the factual details and how they were feeling at the time. You may think you remember exactly what happened at your 10th birthday party, and how excited you were to get that new bike… but do you really? And do you really remember all of the details about your back pain last week and how bad it was? In a TED Radio Hour, Daniel Kahneman - a Nobel laureate for his work in behavioral economics - describes “two selves” for how we perceive events:

“There is an experiencing self, who lives in the present and knows the present. It's capable of reliving the past but basically it has only the present. And then there is a remembering self, and the remembering self is the one that keeps score and maintains the story of our life, and it's the one that the doctor approaches in asking the question, how have you been feeling lately?”

The question is, how well does the “remembering self” agree with the “experiencing self”?

Kahneman and fellow researchers conducted a colonoscopy study comparing the pain levels and memories of the procedure between two randomized groups of patients - one group selected to undergo the normal procedure and the other to experience an extended procedure with an additional several minutes before the tip of the colonoscope was removed. They theorized that even though the second group experienced a longer painful procedure, because the procedure ended with less severe pain they would remember it as being less painful and unpleasant (due to a memory effect called the peak-end rule) - and subsequently would be more likely to return for a repeat colonoscopy in the future.

The study successfully proved this theory, that the patients experiencing a longer, objectively more painful procedure actually remembered it as being less painful than those having the shorter colonoscopy.

While this is good news for the researchers trying to increase patient return rates for preventative procedures, it sheds light on the differences and biases between real-time and retrospective pain assessments. This same study found that the patients’ real-time evaluations of their pain showed no difference between the two groups in their pain levels at the beginning, middle and worst part of the colonoscopy, and that the patients in the extended-procedure group had objectively more total pain. This implies memory biases in recalling pain. In trying to assess pain levels in clinical trials, it would seem that real-time evaluations (or at least closer to real time) are more accurate than retrospective pain questionnaires.

A study published by the Journal of Allergy and Clinical Immunology in 2011 did in fact find that daily symptom diaries are a better assessment tool than a retrospectively administered questionnaire (given up to every four weeks) at assessing treatment responses for asthma control.

So it would seem that having patients complete daily assessments - or even better, real-time assessments - of pain, quality of life measures, treatment response and side effects would reveal a more complete, accurate and objective measure of how patients respond to treatments.

From an analysis of 2013-2014 clinical protocols in Medidata’s PICASⓇ database, we’ve found that 60% of Phase II and 85% of Phase III trials use patient-reported outcome (PRO) questionnaires, and that this trend is continuing to increase. Most of these questionnaires are administered at site visits that often occur weeks apart. For example, a commonly used PRO form, according to our analysis, asks patients about their pain, overall health, and activity levels over the past four weeks. Four weeks! Most people can barely remember accurately where they were last Tuesday, let alone how they felt physically one month prior.

E-diaries and electronic versions of questionnaires are quickly becoming more popular in clinical trials, according to a recent industry report. Apps on patients’ own smartphones (that include reminders) are gaining more and more traction and gets closer to assessing how the patient is feeling at a given point in time throughout their daily life. This eliminates recall biases and provides a much more complete and accurate set of patient-reported data.

Even further, mHealth promises to supplement e-diaries in eliminating recall bias.Today, mHealth devices can track patients’ activity levels, sleep patterns, heart rate and other vitals.

And what if there was a way to objectively measure pain, without even asking the patient?  Stanford Medicine researchers are investigating ways to objectively measure pain using brain imaging and non-invasive devices. Soon, these devices and methods may be used to capture continuous, real-time pain data that can be used to objectively compare treatments.

As clinical trials and regulatory approvals are founded on objective, data-driven research, these approaches will advance measurement of pain, quality of life, and patient treatment outcomes in studies.

But as humans who don’t like experiencing pain or memories of pain, sometimes we have to wonder which is more important: current and remembered perception or objective reality?