IMPACT OF PERSONALIZED AND COMPARATIVE RISK INFORMATION ON DECISIONS ABOUT COLORECTAL CANCER SCREENING

Purpose: Personalized medicine could allow individuals to make decisions about prevention based on their specific levels of health risk, but little is known about how personalized data about risk and benefit informs patient decisions. We conducted an online study of responses to a hypothetical scenario in which individuals were assigned differing information about their personal risk of colorectal cancer (CRC), with or without comparative data on the average person’s CRC risk.

Methods: 622 participants aged 50-75 were recruited from a national on-line survey panel. All participants viewed information about CRC and two approved tests. Participants were then randomized to view additional information following a 2x3 between-subjects experimental design (Table 1):
(a) one of three personalized-risk conditions: no personalized risk information, or a hypothetical scenario in which lifetime risk of CRC mortality was either 1.5% (low personalized risk) or 6% (high personalized risk).
(b) one of two average-risk conditions: viewing vs. not viewing average lifetime risk of CRC mortality (3%). Viewing average risk allowed individuals to determine whether their personalized risk was above or below average (“comparative risk”).

The no-risk-information condition served as the control group. Participants then completed a questionnaire assessing perceived CRC risk, screening intention and test choice.

Results: There was a significant interactive effect of average and personalized risk information on perceived likelihood of dying of CRC  (F_2,622 = 4.40, p=.013, eta²=.014).

Participants who viewed only average-person risk information, and participants assigned to the low personalized risk condition, both had significantly lower perceived CRC mortality risk than participants in the no-risk-information control group. However, participants assigned to the high personalized-risk condition did not have significantly different perceived CRC risk than the control group. (Table 1.)

There was no significant difference between experimental groups in intent to be screened or in test choice. Numeracy did not moderate the effects of the experimental conditions on perceived risk, intent, or test choice.

Conclusions: Low personalized risk information decreased participants’ perceived CRC mortality risk; however, high personalized risk information had no impact on individuals’ perceived CRC risk. Neither personalized nor average risk information influenced participants’ screening intentions. These findings suggest that personalized and comparative risk data may only influence behavior if that data is supplemented with description of potential implications for screening.