Intragenerational Social Mobility and Cause-Specific Premature Mortality
Sunnee Billingsley, Stockholm University
Anna Bryngelson, Stockholm University
Intragenerationalsocial mobility and cause-specific premature mortality
Wefocus specifically on the nature of the relationship between social mobilityand mortality. First, we observe the relationship net of accumulated advantagesand disadvantages that are related to the origin and destination status and notto the experience of mobility. Second, we explore whether the generalrelationship observed with all-cause mortality is being driven by one or twocauses of death or whether this relationship persists across a wide range ofcauses. To reflect the mortality profile of working-age individuals, we assessmain causes of death for premature mortality. Cause-specific mortality patternsmay offer insight into the conditions under which mobility and mortality arelinked. For example, a divide might be made between mortality due to chronicdiseases (e.g. heart failure); and acute death (e.g. accidents or poisoning)(8). Individuals suffering from chronic diseases are likely to experiencelimited capacity to work for a long time period preceding death; in contrast,the timing of symptoms and mortality should be more similar for those whoexperience acute causes of death. This implies that illness may be less presentand likely to influence careers before individuals experience acute causes ofdeath. To explore these potential differences, we distinguish between deathsdue to physical illness and disease such as cardiovascular diseases, cancer,ischaemic heart disease, and stroke; deaths due to poor health behavior such asalcohol consumption and smoking; and deaths due to accidents, poisoning andsuicide.
Data used for the studycome from Swedish population-based registers in the STAR (Sweden in Time:Activities and Relations) collection, which was provided by Statistics Sweden.This data includes The Longitudinal Integration Database for Health Insuranceand Labor Market Studies (LISA), as well as The Structure of Earnings Survey,and the Cause of Death Register, which covers annual information on basicdemographic information, education, occupation and mortality. Occupationalinformation is only available from 1996 until 2012 and the analysis isrestricted to this time frame. When two occupations were provided for the sameyear, the occupation representing the highest social class was selected. Theoccupational information includes all individuals working in public firms. Italso includes all individuals working in private firms with 500 or moreemployees. Smaller private firms are randomly sampled on the basis of size,where employees of larger firms have a higher chance of being included in theregister. We imputed the previous occupation when it was missing for a certainyear and income remained similar (<10% difference). Nevertheless, samplingsmaller private firms generated missing data.
Menwere at a lower risk of death due to all-cause cancer when they were upwardlymobile and a higher risk when they were downwardly mobile. Only the associationfor downward mobility with all-cause cancer remained statistically significantfor women. Death due to sex-specific cancer types such as prostate and breastcancer are also inconsistently linked to social mobility, where breastcancer-related deaths are unrelated but men were at a lower risk of prostatecancer-related death when upwardly mobile.
CVD,stroke and IHD related deaths showed very similar relationships to upwardmobility for men, where the hazard-odds of death were reduced by a little morethan 30%. We found the estimates for downward mobility with death due to CVDand IHD were also similar for men, but the confidence intervals slightlyoverlapped with 1. A link between social mobility and heart-related mortalityemerged for women as well, where downward mobility was associated with anincreased risk of CVD-related mortality and upward mobility was associated witha lower risk. The estimates for IHD and stroke operated similarly but were lessreliable because of wider confidence intervals.
Whenlooking at health-behavior related mortality (Table 3), we found that when menare upwardly mobile they have lower mortality due to smoking-related cancerthan when they were not mobile. Excess smoking-related cancer mortality wasobserved for women and men when they were downwardly mobile, compared withnon-mobile states. The results further show that alcohol-related mortality wasnot significantly associated with social mobility in either direction in men orwomen.
Table3 shows that upward mobility was inversely associated with suicide mortality inmen and women, respectively. Upward mobility in men was also found to lower therisk of mortality due to accidents and poisoning, compared to being non-mobile.
Table 3. Odds ratios (and 95% confidence intervals) fromdiscrete time hazard analysis for cause-specific death among men and women from1996 to 2012. Adjusted for age, education, country of origin, marital status, residence,origin social class and destination class
Presented in Session 1194: Mortality and Longevity