(173) Information Theory and Headache Triggers

Approximately 90% of individuals with headache believe there is at least one trigger, or causal factor, of their headache attacks. The list of headache triggers is extensive, but little is known about how individuals develop such beliefs. The aim of this secondary analysis was to evaluate the degree of information, measured using Shannon entropy, that headache triggers contain and to examine the relationship between joint entropy (i.e., triggers and headache activity) and trigger beliefs. We hypothesized that as information about a headache trigger increased (i.e., an individual was exposed to more states of the trigger), the strength of belief about the trigger also would increase. In a computer-based task, N = 300 participants rated the strength of the relationship between triggers and headaches. For each trigger, participants were asked to rate how often they encountered the trigger and how likely it was to cause a headache on a 0 to 10 scale. Headache triggers contained a wide range of information with binary (i.e., presence or absence) triggers containing between 0.01 and 1 bit of information while continuously scaled triggers (e.g., stress) contained more than 10 bits of information. Each bit of joint information was associated with an 18.4% (95CI: 10.9% to 25.9%) increase in perceived potency of the trigger. However, there was a statistically significant trigger x information interaction (p < 0.001) such that some triggers (e.g., red wine, cinnamon, fruits, and nuts) were associated with decreased potency beliefs when individuals had more trigger-headache pairing information. Certain triggers hold little information for individuals in that they are either largely absent (e.g., MSG) or always present (e.g., air conditioning). Quantifying how much actual information an individual could possess about specific headache triggers appears to be an important aspect of the assessment process.