Cumulative social risk and risk of death from cardiovascular diseases and all-causes

Abstract

Mortality rates for cardiovascular diseases (CVD) have declinedsteadily over the past few decades in high-income countries. Thisdeclinehasbyfardisproportionatelyfavoredthosewithhigherincome,educational attainment, and social support or those who are membersof ethnic majority groups [1–7]. Few studies have examined the cumu-lative effects of multiple social risk factors on CVD mortality rates [8].Disparate exposure to multiple social risk factors may contribute tosocial inequalities in CVD mortality rates.We used data on 10,035 adults (age ≥ 30 years) with no history ofCVD, from the NHANES III Mortality Study (1988–1994 survey datalinked to2006mortalitydata), toassesstheprospectiveassociationbe-tween cumulative social risk and CVD deaths, b65-year-old mortality,and all-cause mortality. Linkage with the National Death Index alloweddefinition of CVD deaths as ICD-9 codes 390–459orICD-10codesI00-I99. Income was assessed using the poverty income ratio (ratio offamily income to the federal poverty level) dichotomized into below1.00 (below the official definition of poverty) vs. 1.00 or greater (incomeabove the poverty level). Education level was dichotomized into low(b12 years, representing b high school diploma) vs. high (≥12 years,representinghighschooldiploma,somecollege,orcollegedegree)levels.Self-reported race/ethnicity was classified into a minority group (non-Hispanic Black, Mexican-American and Other) vs. non-Hispanic White.Single-livingstatus (proxyfor social isolation/lowlevel of socialsupport)was classified into two groups, married/living as married vs. never mar-ried, widowed, divorced, or separated. Each of the four social risk factorswere assigned a score of 1 for their presence or 0 for absence and weresummed to create a cumulative social risk score (range 0 to 4). Cox pro-portional models were used to estimate hazard ratios (HR) and 95%confidence intervals (CI) for the association between cumulative socialrisk and mortality. We evaluated the contribution of biological risk fac-tors(bodymassindex[BMI], HbA1c,systolic blood pressure[SBP],cho-lesterol, triglycerides, C-reactive protein [CRP] and estimatedglomerular filtration rate [eGFR]) to the association between cumula-tive social risk and CVD deaths. We hypothesized that these biologicalfactorsareonthepathwayintheassociationbetweenexposuretosocialrisk factors and occurrence of CVD deaths, and thus are mediators of thisassociation.HbA1cwasmeasured usingaBio-RadDiamantionexchangehigh-performanceliquid chromatographysystem. Serumtotalcholester-ol and triglycerides were measured enzymatically by a Hitachi 704 Ana-lyzer. eGFR was based on the Modification of Diet in Renal Diseasestudy equation. Serum CRP was measured using the Behring latex-enhanced CRP assay.Atotalof 31.7%of adultsreportedatleastonesocialriskfactor;7.1%reported3 or more. Over a median14-yearfollow-up, there were 2604deaths (1386 in males and 1218 in females) including 924 deaths relatedto cardiovascular diseases. Table 1 shows the age-and sex-adjusted asso-ciationsofeachsocialriskfactorwithCVDdeaths, b65-year-oldmortality,and all-cause mortality. Ha zard ratios for CVD deaths, b65-year-old mor-tality and all-cause mortality signi ficantly increased with an increasingnumber of social risk factors and were greatest in those exposed to 3 ormore social risk factors compared with those with 0 (Table 1). Table 2shows the association between exposure to 3 or more social risk factorsand CVD deaths, as well as the contribution of biological risk factors tothis association. Biological risk factors accounted for 12% (95% CI: 4% to18%) of the association between exposure to 3 or more social risk factorsand CVD deaths.PreviousstudiesonsocialinequalitiesinCVDmortalityhavetypicallyoperationalized social disadvantage using single measures of socioeco-nomicstatus(e.g.manualoccupationalclass,loweducationlevel,lowin-come) or a composite of socioeconomic measures in several periodsthrough the life course [8]. However, summing the number of times anindividualhadbeeninalowersocioeconomiccategoryasaproxyforcu-mulative social disadvantage, may erroneously attribute CVD mortalityto risks associated with the accumulation of only socioeconomic