Cumulative biological risk and socio-economic differences in mortality: MacArthur Studies of Successful AgingIntroductionMaterials and methodsAll-cause mortalitySocio-economic statusAllostatic loadCovariatesAnalysesResultsDiscussionAcknowledgementsReferencesSocial Science & Medicine 58 (2004) 1985–1997Cumulative biological risk and socio-economic differences inmortality: MacArthur Studies of Successful AgingTeresa E. Seemana,*, Eileen Crimminsb, Mei-Hua Huanga, Burton Singerc,Alexander Bucura, Tara Gruenewaldd, Lisa F. Berkmane, David B. ReubenaaDivision of Geriatrics, Geffen School of Medicine, University of California, 10945 Le Conte Ave Suite 2339, Los Angeles,CA 90095-1687, USAbAndrus Gerontology Center, University of Southern California, Los Angeles, CA, USAcOffice of Population Research, Princeton University, Princeton NJ, USAdDepartment of Psychology, University of California, Los Angeles, CA, USAeDepartment of Health and Social Behavior, School of Public Health, Harvard University, Cambridge MA, USAAbstractPrevious research has suggested that socio-economic status (SES) differences in mortality are only partially explainedby differences in life-style, psychological and social factors. Seven year mortality data (1988–1995) from the MacArthurStudy of Successful Aging, a longitudinal study of adults, aged 70–79, from New Haven, CT; East Boston, MA; andDurham, NC; were used to test the hypothesis that a cumulative measure of biological dysregulation (‘‘allostatic load’’),reflecting multiple regulatory systems, would serve as a further mediator of SES differences in mortality. Logisticregression analyses revealed that a cumulative index of biological risk explained 35.4% of the difference in mortalityrisk between those with higher versus lower SES (as measured by less than high school education versus high school orgreater educational attainment). Importantly, the cumulative index provided independent explanatory power, over andabove a measure of doctor-diagnosed disease, though the latter also contributed to education-related variation inmortality risks. The summary measure of biological risk also accounted for more variance than individual biologicalparameters, suggesting the potential value of a multi-systems view of biological pathways through which SES ultimatelyaffects morbidity and mortality.r 2003 Elsevier Ltd. All rights reserved.Keywords: Socio-economic status; Mortality; Allostatic load; Education; USAIntroductionAn individual’s socio-economic status (SES) repre-sents one of the most enduring of all risk factors.Extensive evidence documents not only the highermorbidity and mortality associated with lower SES(Adler et al., 1994; Pincus & Callahan, 1995; Macintyre,1997) but also the persistence of these SES gradients intoolder age (House, Kessler, & Herzog, 1990; Crimmins,Hayward, & Saito, 1996; Rogers, 1992). Indeed, recentevidence suggests that SES-related mortality differen-tials at older ages are growing (Manton, Stallard, &Corder, 1997; Crimmins & Saito, 2001; Preston & Elo,1995).Though a variety of possible explanations for thesesocial inequalities in health have been offered, differ-ential exposure to chronic and acute stressors arehypothesized to contribute importantly. Evidence forSES-related differentials in exposure to stressors, parti-cularly chronic ones, is abundant, with lower SESindividuals consistently reporting more of various typesof stressors—e.g., physical, economic and/or social(Dohrenwend, 1973; Kessler & Cleary, 1980; Houseet al., 1992; Turner, Wheaton, & Lloyd, 1995). MoreARTICLE IN P RESS*Corresponding author. Tel.: +1-310-825-8253; fax: +1-310-794-2199.E-mail address: [email protected] (T.E. Seeman).0277-9536/$ - see front matter r 2003 Elsevier Ltd. All rights reserved.doi:10.1016/S0277-9536(03)00402-7recently, there has been growing interest in elucidatingthe biological pathways that mediate between such SES-related exposures and increased disease/mortality (Mar-mot, Bobak, & Davey Smith, 1995; Seeman & Crim-mins, 2001).Gradients in biological risk factors by SES, withhigher risk profiles being seen more commonly at lowerlevels of SES, have been shown for a number oftraditional cardiovascular risk factors such as bloodpressure (BP), cholesterol, relative weight, glucose andfibrinogen in many though not all studies (Bobak,Hertzman, Skodova, & Marmot, 1999; Brunner et al.,1997; Ishizaki, Martikainen, Nakagawa, & Marmot,2000; Dyer et al., 1999; Marmot, et al., 1998; Feldman,Makuc, Kleinman, & Cornoni-Huntley, 1989; Karla-mangla et al., submitted; Rose & Marmot, 1981;Bartley, Fitzpatrick, Firth, & Marmot, 2000; Kristen-son, Kucinskiene, Bergdahl, & Orth-Gomer, 2001;Rosengren, Orth-Gomer, & Wilhelmsen, 1998). How-ever, these gradients in cardiovascular risk factorsappear to explain only a portion of SES differentialsin mortality (Marmot et al., 1991; Marmot, Shipley, &Rose, 1984; Lynch, Kaplan, Cohen, Tuomilehto, &Salonen, 1996; Luoto, Pekkanen, Uutela, & Tuomilehto,1994). Prior research, however, has generally focused ontraditional cardiovascular risk factors and has largelyexamined the role of individual biological parametersand/or sets of these cardiovascular risk factors.The potential value of considering biological media-tion of the SES-health relationships from a broadermulti-systems viewpoint is suggested by several facets ofthe SES-health relationship. First, SES gradients areobserved with respect to a wide range of differentpathophysiological disease processes and with respect tomost major causes of death (Adler et al., 1994). Second,most of these outcomes are known to have multi-factorial etiologies, encompassing multiple sources ofbiological dysregulation.The concept of allostatic load has been proposed as amulti-systems view of the cumulative physiological tollthat may be exacted on the body through attempts atadaptation to life’s demands (McEwen & Stellar, 1993;McEwen, 1998). As such, allostatic load may represent auseful approach to conceptualizing biological mediationof SES effects on health and longevity. The concept ofallostatic load originates from the idea that healthyfunctioning requires on-going adjustments of the inter-nal physiologic milieu, with physiologic systems exhibit-ing fluctuating levels of activity as they respond andadapt to environmental demands—a concept referred toas allostasis (Sterling & Eyer, 1988). Importantly,allostasis emphasizes that while healthy functioningrequires