Socioeconomic status and the 25 × 25 risk factors as determinants of premature mortality: a multicohort study and meta-analysis of 1·7 million men and women

Summary Background In 2011, WHO member states signed up to the 25 × 25 initiative, a plan to cut mortality due to non-communicable diseases by 25% by 2025. However, socioeconomic factors influencing non-communicable diseases have not been included in the plan. In this study, we aimed to compare the contribution of socioeconomic status to mortality and years-of-life-lost with that of the 25 × 25 conventional risk factors. Methods We did a multicohort study and meta-analysis with individual-level data from 48 independent prospective cohort studies with information about socioeconomic status, indexed by occupational position, 25 × 25 risk factors (high alcohol intake, physical inactivity, current smoking, hypertension, diabetes, and obesity), and mortality, for a total population of 1 751 479 (54% women) from seven high-income WHO member countries. We estimated the association of socioeconomic status and the 25 × 25 risk factors with all-cause mortality and cause-specific mortality by calculating minimally adjusted and mutually adjusted hazard ratios [HR] and 95% CIs. We also estimated the population attributable fraction and the years of life lost due to suboptimal risk factors. Findings During 26·6 million person-years at risk (mean follow-up 13·3 years [SD 6·4 years]), 310 277 participants died. HR for the 25 × 25 risk factors and mortality varied between 1·04 (95% CI 0·98–1·11) for obesity in men and 2 ·17 (2·06–2·29) for current smoking in men. Participants with low socioeconomic status had greater mortality compared with those with high socioeconomic status (HR 1·42, 95% CI 1·38–1·45 for men; 1·34, 1·28–1·39 for women); this association remained significant in mutually adjusted models that included the 25 × 25 factors (HR 1·26, 1·21–1·32, men and women combined). The population attributable fraction was highest for smoking, followed by physical inactivity then socioeconomic status. Low socioeconomic status was associated with a 2·1-year reduction in life expectancy between ages 40 and 85 years, the corresponding years-of-life-lost were 0·5 years for high alcohol intake, 0·7 years for obesity, 3·9 years for diabetes, 1·6 years for hypertension, 2·4 years for physical inactivity, and 4·8 years for current smoking. Interpretation Socioeconomic circumstances, in addition to the 25 × 25 factors, should be targeted by local and global health strategies and health risk surveillance to reduce mortality. Funding European Commission, Swiss State Secretariat for Education, Swiss National Science Foundation, the Medical Research Council, NordForsk, Portuguese Foundation for Science and Technology.

included a questionnaire on social, demographic, behavioural and clinical data, and a physical examination including blood collection.
-level data from the Social Security Administration archives in Italy, covering almost 8% of all Italian workers employed in the private sector in 1985information on income, pensions, unemployment benefits, disability indemnities, workplace and job contracts, linked to hospital and mortality follow-up.
NHANES. designed to assess the health and nutritional status of adults and children in the United States. The survey is enters for Disease Control and Prevention (CDC) and the early 1960s and has been conducted as a series of surveys focusing on different population groups or health topics. In 1999, the survey became a continuous program that has a changing focus on a variety of health and nutrition measurements to meet emerging needs. The survey examines a nationally representative sample of about 5,000 persons each year. These persons are located in counties across the country, 15 of health-related questions. The examination component consists of medical, dental, and physiological measurements, as well as laboratory tests administered by highly trained medical personnel. http://www.cdc.gov/nchs/nhanes/ NHIS. of the civilian non-institutionalized population of the United States and is one of the major data collection studies to secure accurate and current statistical information on the amount, distribution, and effects of illness and disability in the United States and the services rendered for or because of such conditions. The survey een conducted continuously since 1957, the content of the survey has been updated about every 10http://www.cdc.gov/nchs/nhis/ Alameda County Study. The Alameda County Study was designed to investigate normal daily routines and social-support factors to determine which might be risk factors for poor health and mortality in a real community.In 1965, a probability sample of the population of Alameda County, California was invited to participate in a study on health status, social networks, and other personal characteristics. The follow-up group contained 6,928 participants who completed questionnaires and were followed at intervals for up to 20 years after the initial investigation.http://www.epi.umn.edu/cvdepi/study-synopsis/alameda-county-study/

HALS.
Kingdom) in order to inform health strategies of the Primary Care Trusts. A stratified random sample of the population was asked to participate and a questionnaire was posted to those people who consented. The compelling message this survey sends out is the importance of deprivation in health inequalities and behavioural risks. Deprivation is a complex interaction of factors, which include environment, income, education, housing, employment and social capacity. Therefore to improve health and reduce inequalities, the o-ordinated action at all levels and across all agencies, no improvement will be seen in this unacceptable variation.

HRS.
a longitudinal panel study that surveys a representative sample of approximately 20,000 people in America over the age of 50 every two years. Supported by the National Institute on Aging (NIA U01AG009740) and the Social Security plores the changes in labor force participation and the health transitions that individuals undergo toward the end of their work lives and in the years that follow. Since its launch in 1992, the study has collected information about income, work, assets, pension plans, health insurance, disability, physical health and functioning, cognitive functioning, and health care expenditures. Through its unique and inresearchers can use to address important questions about the challenges and opportunities of aging. http://hrsonline.isr.umich.edu/index.php?p=start MIDUS. The first national survey of Midlife Development in the U.S. (MIDUS) was conducted in 1995/96 by the MacArthur Foundation Research Network on Successful Midlife Development. The study was conceived by a multidisciplinary team of scholars from fields of psychology, sociology, epidemiology, demography, anthropology, medicine, and health care policy. Their collective aim was to investigate the role of behavioral, psychological, and social factors in accounting for age-related variations in health and wellbeing in a national sample of Americans. In addition to a national probability sample (N = 3,487), the study included over-samples in select metropolitan areas (N = 757), a sample of siblings (N = 950) of the main respondents, and a national sample of twin pairs (N=1,914). http://midus.wisc.edu/scopeofstudy.php WLSG and WLSS. The Wisconsin Longitudinal Study has followed a random sample of 10317 participants (5326 women, 4991 men) who were born between 1937 and 1940 and who graduated from Wisconsin high schools in 1957 (graduate sample, WLSG). After baseline data collection in 1957, survey data have been collected from the participants or their parents in 1964, 1975, 1992/3, and 2003/5. The present study used data from the 1993/1994 follow-up. The WLS sample is broadly representative of white, non-American men and women who have completed at least a high school education (among Americans aged 50 to 54 in 1990 and 1991, approximately 66 percent were non-ted at least 12 years of schooling). It is estimated that about 75 percent of Wisconsin youth graduated from high school in the late 1950s -everyone in the primary WLS sample graduated from high school. In addition to the main sample of the 1957 high school graduates, the WLS has also collected data on a selected sibling of a sample of the graduates (sibling sample, WLSS). The data collection in adulthood has been very similar although not entirely identical for the siblings as for the graduates. For the present purposes, the sibling sample was analyzed separately from the graduate sample, because the sampling frame of the individuals for the graduate cohort and sibling cohort was considered to sufficiently to justify the decision of not combining the samples. Baseline data were from the 1994/1995 follow-up.

Appendix 3. Quality assessment of included studies
To assess study quality, we evaluated potential bias in the following domains: 1. Was selection of exposed and non-exposed cohorts drawn from the same population? 2. Can we be confident in the assessment of exposure? 3. Can we be confident that the outcome of interest was not present at start of the study? 4. Did the statistical analysis adjust for the confounding variables? 5. Can we be confident in the assessment of the presence or absence of confounding factors? 6. Can we be confident in the assessment of outcome? 7. Was the follow up of cohorts adequate?
The studies were evaluated in relation to each question using 4 categories: ++ "definitely yes"," + probably/mostly yes", -"probably/mostly no", and --"definitely no". The quality of the study was considered high if all domains were evaluated favourably (++ or +). In all studies, individuals with high and low SES were from the same cohort and response at baseline was 60% or higher in all studies, except COLAUS, GAZEL and E3N for which it was less than 50%. In MCCS and EPIC Italy, the eligible population was not clearly defined as the participants were recruited from universities, cafeterias etc. Self-reported assessment was considered reliable for SES, but this was missing for MCCS. There were no exclusions to prevalent cases in mortality analyses. The 25 x 25 risk factors (smoking, blood pressure, body mass index, diabetes, alcohol consumption, physical activity) were assessed in all studies -1996 which missed assessment of one or more 25x25 risk factors.
- III, 1999III, , 2001III, , 2003III, , 2005III, and 2007, whereas other studies measured baseline variables using self-report assessment. Mortality ascertainment was reliable in all studies. The follow-up for mortality was optimal (>10 years) in most studies and adequate (>3 years) in all studies. Of the 48 studies, 30 were of high quality with all 7 domains evaluated favorably.

Appendix 4. Details on measurements of the 25x25 risk factors
Measures. Socioeconomic position (SES) was measured through occupational position, obtained from the European Socio-economic Classification (ESEC)( https://www.iser.essex.ac.uk/archives/esec/user-guide), higher professionals and managers, lower professionals and managers, higher clerical, services and sales workers (Class 1, 2 and 3 ESEC); Intermediate, including small employers and self-employed, farmers, lower supervisors and technicians (Class 4, 5, and 6 ESEC); and low, including lower clerical, services and sales workers, skilled workers, semi-and unskilled workers (Class 7,8,9 ESEC). Occupational position was the last known one (current occupation for working individuals and last know occupation for those not currently working). For E3N, occupational position was current occupation 2-yr after baseline. In the cohorts from the United States, a similar occupational categorization was applied based on occupational groups defined by the Standard Occupational Classification codes.
Smoking was self-reported and was categorized into current smoking for individuals currently smoking, former smoking for individuals not currently smoking but who smoke in the past, and never smoking for individuals who never smoke.
Alcohol consumption was measured in alcohol units. For most cohorts, it was the average number of alcohol units over a week. For ELSA, it was the highest intake over a week. For EPIC-Italy, EPIPORTO and MCCS alcohol intake was measured in grams and they were converted to alcohol units by assuming that 12grams equal one alcohol unit. For E3N and GAZEL, alcohol consumption was available 3-yr after baseline.
women. Abstinence was defined as no alcohol consumption, and moderate intake was defined as consumption less or equal 21 alcohol units for men and 14 for women.
Physical inactivity was considered as less than 20 minutes of physical activity per week in COLAUS, neither moderate nor intense physical activity per day in EPIPORTO, neither moderate nor intense physical activity per week in E3N, none/light activity only for ELSA, as less than 1 hour of moderate and vigorous , no physical activity in EPIC Italy, as not practicing sport in GAZEL, as a physical activity score of zero in MCCS. Physical inactivity was determined as participating in no weekly leisure-; no moderate or vigorous physical Body mass index (BMI) was measured as weight (Kg) divided by the square of height (m^2). Obesity is defined as BMI 30; Overweight is defined as BMI 25 and BMI<30; Normal BMI is defined as BMI 18 5 and BMI<25. For NCDS, the self-reported values of weight and height at wave 6 were used whenever the measured values during the biomedical survey at 2003 were not available. For E3N and GAZEL the values were self--Hypertension was defined as the presence of at least one of the following conditions: systolic blood pressure -hypertensive medication, self-report. For most cohorts, the four conditions were available, for GAZEL only the self-reported information was available and for E3N only self-reported blood pressure values, and for NCDS knowledge of ant-hypertensive medication was not available. For most cohorts blood pressure values were the average of two readings, while EPIC-Italy had only one. Blood pressure was measured as the mean of two or three -Diabetes was defined as the presence of at least one of the following conditions: fasting glucose 7 mmol/L, 2h postload glucose 11 1 mmol/L, glycated hemoglobin 6 5%, self-report. All but E3N have self--II further has both fasting glucose and 2h post load glucose, ELSA and NCDS have glycated hemoglobin, and EPIPORTO and COLAUS fasting glucose. In the -reported) or taking medication, and self-reported as whether the participant had been diagnosed of diabetes by a medical