PROJECTION BIAS AMONG PERSONS RECEIVING SPINAL INJECTION AS TREATMENT FOR LUMBAR PAIN

Monday, October 24, 2011
Grand Ballroom AB (Hyatt Regency Chicago)
Poster Board # 31
(BEC) Behavioral Economics

Richard Wilson, MD, MS, MetroHealth Medical Center/Case Western Reserve University, Cleveland, OH, Justin Sydnor, PhD, University of Wisconsin, Madison, WI, Brian Snitily, Case Western Reserve University, Cleveland, OH and Adam T. Perzynski, PhD, Case Western Reserve University at MetroHealth, Cleveland, OH

Purpose: Work at the intersection of psychology and economics has documented that people are often unable to accurately predict how their preferences and feelings will change when they are in a different state.  Behavioral economists have labeled this phenomenon “projection bias” in predicting one’s future choice behavior.  A person not in pain may have trouble understanding the value of pain-reduction if she were in fact in pain.  Similarly, a person in extreme pain may overestimate the value placed on reducing modest amounts of pain.  Our study seeks to understand whether these types of projection bias exist.  We examined willingness-to-pay for pain relief and its association with pain level before and after spinal injection treatment among persons with lumbar pain. 

Method: We conducted face-to-face interviews with persons completing treatment in a pain clinic for lumbar stenosis (N=17).  Subjects were 53% male, 63% white, 37% black and the median age was 46.  Prior to treatment, subjects were asked their willingness to pay (WTP) to be free of imagined pain rated a 3, 5, 7 and 10.  After treatment, subjects were asked their willingness to pay for relief of their remaining post-treatment pain.  Based on the imagined pre-treatment pain levels we computed linear extrapolations for pain levels 1-10 and compared responses to WTP at the imagined/extrapolated pain levels to the WTP post-treatment in order to quantify projection bias.  We then examined projection bias by pre-treatment pain level, and pre-post change in pain.

Result: Five subjects (29%) had perfect predictions based upon their imagined pain states.  Three subjects (18%) had higher than predicted WTP (one of whom had a pain increase).  The remaining 9 subjects (53%) all gave answers in line with projection bias (lower than predicted post-treatment WTP).  Subjects with pre-treatment pain > 7 had higher projection bias (p=.02) and projection bias was correlated with magnitude of pain reduction (r=.47, p=.05). 

Conclusion: We found evidence of projection bias in this pilot.  Our findings suggest a need to account for projection bias when using contingent valuation methods to establish the potential benefits of pain-reduction therapies.  Projection bias may also be creating an empathy disconnect for doctors and nurses (who are generally in a non-pain state) trying to weigh costs and benefits of different pain-reduction strategies for their pain patients.