Self-rated health, gender, and mortality in older persons: introduction to a special section.

Self-rated health (SRH) is widely recognized as a comprehensive indicator of health. Its comprehensiveness is often referred to as surprising because the indicator is based on a single question. Yet, numerous studies have demonstrated that this question may include more relevant aspects of health than can be measured in a survey, however broad the range of health measures used. The most widely quoted proof of SRH’s comprehensiveness is its predictive ability for mortality, evidence for which has been found in many studies from different countries using various designs (Benyamini & Idler, 1999; Idler & Benyamini, 1997). However, little evidence exists for the reasons why SRH is such a good predictor of mortality. Possible explanations include (a) SRH, in addition to comprehensive health, also reflects preclinical, and so far unmeasurable, aspects of health; (b) SRH, in addition to current health, dynamically reflects recent changes in health; (c) the way of responding to the SRH question is correlated with health behavior, for example, optimistic responses are linked with positive health behavior, which in turn are associated with better survival; (d) self-perceptions of health are related to the psychoneuroimmunologic responses that determine the susceptibility to diseases (see Idler & Benyamini, 1997). Moreover, upon closer inspection, differences and inconsistencies in predictive ability can be observed among subgroups according to gender and age and also according to the duration of the mortality follow-up period. This special section explicitly addresses gender differences in the ability of self-rated health to predict mortality, as well as possible genderspecific differences with respect to age, cohort, and duration of mortality follow-up. Idler and Kasl’s (1991) study of 2,812 older people in New Haven, Connecticut, provide an example of gender differences in the predictive ability for mortality. Their study showed that, in comparison with the reference category of ‘‘excellent’’ self-rated health, among men the adjusted odds ratio (OR) for poor (OR 1⁄4 6:7), fair (OR 1⁄4 4:1) and good (OR 1⁄4 3:2) self-rated health was noticeably greater than among women (OR 1⁄4 3:1 [poor], OR 1⁄4 2:8 [fair], and OR 1⁄4 2:4 [good]). Four other studies showed similar differences. However, McCallum, Shadbolt, and Wang’s study (1994) of 1,050 older people in Sydney, Australia, showed that whereas fair (OR 1⁄4 2:4) and good (OR 1⁄4 3:2) compared with excellent self-rated health remained predictive of mortality among women with sociodemographics, morale, and social support as covariates, self-rated health was no longer predictive among men. In more recent studies, lower self-rated health remained predictive of mortality in adjusted models among men but not among women (Hays, Schoenfeld, Blazer, & Gold, 1996; Franks, Gold, & Clancy, 1996; van Doorn, 1998; van Doorn & Kasl, 1998). In one study, lower self-rated health remained predictive of mortality among women but not among men (Simons, McCallum, Friedlander, & Simons, 1996). The observed differences in predictive ability among subgroups might shed light on the ‘‘working elements’’ in SRH that contribute to the prediction of mortality. For possible explanations of the association between SRH and mortality when SRH predicts mortality in one, but not in another subgroup, researchers should examine the characteristics of the subgroups for factors that might further explain the association between SRH and mortality. However, the reported differences among subgroups are not Address correspondence to Prof. dr. D. J. H. Deeg, Vrije Universiteit Medical Center/LASA, Van der Boechorststraat 7, 1081 BT Amsterdam, The Netherlands. E-mail: [email protected] Vrije Universiteit Medical Center/LASA, Amsterdam, The Netherlands. Health Informatics Research Group, Centre for Health Information Management Research (CHIMR), Department of Information Studies, University of Sheffield, Western Bank, Sheffield, United Kingdom.