Numerous studies have indicated that lower socioeconomic status (SES) is related to poor health, in terms of both morbidity and mortality. immortal,"''4(0"237) resulting in an artificially low Latino mortality rate.Although the salmon bias hypothesis has not been tested, some evidence suggests that it is plausible. One study'5 estimated return migration rates ofvarious foreign-born groups based on data from a program requiring immigrants to submit yearly address reports to the Immigration and Naturalization Service. Lower-and upper-bound return migration estimates (assuming a 50% and 100% response rate for filing address reports) ranged from 15.6% to 56.2% for Mexicans, 52.4% to 72.5% for South Americans, and 49.6% to 69.5% for Central Americans and Caribbean persons (excluding Cubans).Although return migration (both permanent and temporary) depends on specific community, economic, and social network factors, 1617 it can be substantial. As many as 75% of households in Mexican migrant towns engage in return migration from the United States.'8 Despite the methodologic shortcomings and the specificity of communities surveyed, these studies suggest that the salmon effect and healthy migrant hypothe-

The study tests the thesis of pathologic adaptation for youth exposed to community violence, where high levels of exposure to community violence lead to increased aggressive behavior but decreased psychological distress. Four hundred seventy-one 6th graders and 1 of their parents were interviewed. The results showed, for a small but important subgroup of youth, that high levels of exposure to community violence were associated with more child- and parent-reported aggressive behavior and less child-reported psychological distress. Targeted prevention strategies for these high-risk youth are especially needed.

A causal model is formulated for the thesis that in inner-city youth exposed to high levels of violence, cognitions that normalize violence mitigate affective effects of exposure while increasing risk for violent behavior, thus perpetuating violence in the very process of adapting to it psychologically. Gender differences in the cognitive normalization of violence may explain gender differences in affective and behavioral effects of exposure. Empirical studies are needed to directly test this model.

Opioid use for migraine is associated with more severe headache-related disability, symptomology, comorbidities (depression, anxiety, and cardiovascular disease and events), and greater HRU for headache. Longitudinal studies are needed to further assess the directionality and causality between opioid use and the outcomes we examined.