Table 12 The effect of considering exposure gradient when judging quality of evidence
Factor | Example |
|---|---|
Exposure gradient | In our review we did not observe evidence of an exposure-response gradient in any of the cases identified. However, an exposure-response gradient is observed in a cross-sectional study of headache, which was not included in our review because cross-sectional studies should not be used to study causes and prognosis. Anda and colleagues [44] evaluated, in a sample of adults with frequent headaches, whether the number of adverse childhood events (including childhood abuse and household dysfunction) prior to and including the age of 18 was associated with the presence of self-reported headaches in adulthood. A significant exposure-response gradient was identified, with an increase in the size of the odds ratio associated with the likelihood of headache in adulthood from 1.2, to 1.7, to 2.1 as the number of childhood adverse events also increases from 1, to 3, to 5, respectively. |