2001), a proportion that could conceivably vary by race and ethnicity in the United States, given variation in sexually transmitted diseases by socioeconomic status. In addition, behavior at younger ages could have long-term health effects or determine which people survive to older ages. Other behaviors, such as suicide, are more directly relevant to older adults, though suicide may be better addressed as a consequence of such factors as depression (Myers and Hwang, 2004). (Preventive health behaviors are considered below in relation to health care.)
As an illustration of the potential effect on differences of a range of other behavior risk factors, consider such a simple behavior—with complex causes—as staying in one’s home versus going out regularly. One case shows how this can produce racial and ethnic differences in mortality. A heat wave in Chicago in July 1995 killed at least 521 and possibly as many as 739 people, mostly older adults (Klinenberg, 2002; Whitman et al., 1997). The age-adjusted black death rate was 50 percent higher than the white rate. Hispanic fatalities were so low that a rate could not be calculated. Those most vulnerable proved to be people who lived alone, did not leave home daily, had medical problems, were confined to bed, and had no air conditioning (Semenza et al., 1996). The ethnic difference had much to do with the quality of neighborhood life. Equally poor black and Hispanic neighborhoods were characterized by relatively empty and unsafe versus bustling streets, and few commercial attractions versus animated public spaces. Overall, a pervasive fear of violence in black neighborhoods kept seniors especially indoors, with disastrous results (Klinenberg, 2002). Other behavioral factors can therefore be highly consequential for health differences, and are tied in to neighborhood features, which may in turn have roots in discrimination and lack of social support, factors that we consider further below.
Much is uncertain about the role of behavior risk factors in racial and ethnic differences, leading to a number of important research issues.
Research Need 8: Study how behavior risk factors act over the life course in different racial and ethnic groups.
The effects of behavior among older adults, in contrast to the effects at younger ages, sometimes appear weaker or nonexistent, and carefully designed studies are needed to avoid such pitfalls as inadequate samples and unrecognized selection effects. Behavior does show some consistency over time, but it can also change. The intricacies of attending to cohort and period changes in such factors as smoking, diet, and exercise, their delayed effects on health outcomes in later years, and the differences in these rela-