Advertisement
Research Article| Volume 32, ISSUE 4, P918-928, April 2022

Download started.

Ok

Education inequalities in cardiovascular and coronary heart disease in Italy and the role of behavioral and biological risk factors

Open AccessPublished:November 24, 2021DOI:https://doi.org/10.1016/j.numecd.2021.10.022
Education inequalities in cardiovascular and coronary heart disease in Italy and the role of behavioral and biological risk factors
Previous ArticleLongitudinal association of egg intake frequency with cardiovascular disease in Chinese adults
Next ArticleIndependent and joint associations of non-exercise cardiorespiratory fitness and obesity with risk of type 2 diabetes mellitus in the Rural Chinese Cohort Study

      Highlights

      • •
        Education inequalities in CVD and CHD were observed In Italy, stronger among women.
      • •
        BBRF mediated education inequalities, but fewer among women.
      • •
        Smoking and psychosocial factors could explain the gender differences.
      • •
        Addressing prevention programmes to less educated women to reduce social inequalities in cardiovascular health.

      Abstract

      Background and aims

      Behavioral and biological risk factors (BBRF) explain part of the variability in socioeconomic differences in health. The present study aimed at evaluating education differences in incidence of cardiovascular disease (CVD) and coronary heart disease (CHD) in Italy and the role of BBRF.

      Methods and results

      All subjects aged 30–74 years (n = 132,686) who participated to the National Health Interview Surveys 2000 and 2005 were included and followed-up for ten years. Exposure to smoking, physical activity, overweight/obesity, diabetes and hypertension at baseline was considered. Education level was used as an indicator of socioeconomic status. The outcomes were incident cases of CVD and CHD. Hazard ratios by education level were estimated, adjusting for sociodemographic covariates and stratifying by sex and geographic area. The contribution of BBRF to education inequalities was estimated by counterfactual mediation analysis, in addition to the assessment of the risk attenuation by comparing the models including BBRF or not.
      22,214 participants had a CVD event and 6173 a CHD event. After controlling for sociodemographic factors, the least educated men showed a 21% higher risk of CVD and a 17% higher risk of CHD compared to the most educated (41% and 61% among women). The mediating effect (natural indirect effect) of BBRF between extreme education levels was 52% for CVD and 84% for CHD among men (16% among women for CVD).

      Conclusions

      More effective strategies aiming at reducing socioeconomic disparities in CVD and CHD are needed, through programs targeting less educated people in combination with community-wide initiatives.

      Keywords

      Introduction

      Cardiovascular diseases (CVD) are the leading health problem in most countries, accounting for 31% of deaths worldwide in 2016 [
      • WHO
      Cardiovascular diseases (CVDs).
      2017
      ]. Social inequalities in health persist in most developed countries for mortality and incidence of CVD and coronary heart disease (CHD) [
      • Manrique-Garcia E.
      • Sidorchuk A.
      • Hallqvist J.
      • Moradi T.
      Socioeconomic position and incidence of acute myocardial infarction: a meta-analysis.
      J Epidemiol Community Health. 2011; 65: 301-309
      ,
      • Mackenbach J.P.
      • Stirbu I.
      • Roskam A.J.
      • Schaap M.M.
      • Menvielle G.
      • Leinsalu M.
      • et al.
      Socioeconomic inequalities in health in 22 European countries.
      N Engl J Med. 2008; 358: 2468-2481
      ,
      • Stringhini S.
      • Carmeli C.
      • Jokela M.
      • Avendaño M.
      • Muennig P.
      • Guida F.
      • et al.
      Socioeconomic status and the 25×25 risk factors as determinants of premature mortality: a multi cohort study and meta -analysis of 1·7 million men and women.
      Lancet. 2017; : 389
      ]. Two recent meta-analysis estimated a 50% higher risk of CVD incidence, 36% of CHD incidence and 39% of CVD mortality for least educated people compared with most educated [
      • Khaing W.
      • Vallibhakara S.A.
      • Attia J.
      • McEvoy M.
      • Thakkinstian A.
      Effects of education and income on cardiovascular outcomes: a systematic review and meta-analysis.
      Eur J Prev Cardiol. 2017; 24: 1032-1042
      ] and stronger socioeconomic inequalities among women than among men for both CVD and CHD incidence [
      • Backholer K.
      • Peters S.A.E.
      • Bots S.H.
      • Peeters A.
      • Huxley R.R.
      • Woodward M.
      Sex differences in the relationship between socioeconomic status and cardiovascular disease: a systematic review and meta-analysis.
      J Epidemiol Community Health. 2017; 71: 550-557
      ].
      Many factors may be involved in determining the occurrence of CVD and CHD, including mainly biological, psychosocial and behavioral ones. The first group is composed of traditional biological CVD risk factors, such as high HLD cholesterol, high blood pressure or diabetes, already identified in large epidemiologic prospective studies [
      • Kannel W.B.
      • McGee D.L.
      Diabetes and cardiovascular disease. The Framingham study.
      JAMA. 1979; 241: 2035-2038
      ,
      • Keil U.
      • Kuulasmaa K.
      WHO MONICA Project: risk factors.
      Int J Epidemiol. 1989; 18: S46-S55
      ].
      Furthermore, there is strong evidence that smoking, physical inactivity, obesity, unhealthy diet, and poor adherence to medications increase the risk of CVD and contribute to an increase in cardiovascular risk factors such as hypertension, lipid abnormalities, insulin resistance, and diabetes mellitus [
      • Havranek E.P.
      • Mujahid M.S.
      • Barr D.A.
      • Blair I.V.
      • Cohen M.S.
      • Cruz-Flores S.
      • et al.
      Social determinants of risk and outcomes for cardiovascular disease: a scientific statement from the American heart association.
      Circulation. 2015; 132: 873-898
      ]. Up to 80% of mortality from cardiovascular diseases could be prevented by eliminating smoking, poor diet, alcohol intake, and inactivity [
      • WHO
      2008-2013 action plan for the global strategy for the prevention and control of noncommunicable diseases.
      2009
      ].
      Unhealthy behaviors are more frequent among individuals in more disadvantaged social positions, and they are one of the main acknowledged mechanisms linking lower socioeconomic position to worse health. Taken together, smoking, alcohol consumption, dietary patterns, physical activity and body mass index, have been shown to explain a substantial part of the variability in socioeconomic differences in incidence and mortality, although with estimates quite heterogeneous across studies and, in general, with a lower contribution in Southern European countries [
      • Petrovic D.
      • de Mestral C.
      • Bochud M.
      • Bartley M.
      • Kivimaki M.
      • Vineis P.
      • et al.
      The contribution of health behaviors to socioeconomic inequalities in health: a systematic review.
      Prev Med. 2018; 113: 15-31
      ]. A recent Italian study on the mediating role of behavioral risk factors on mortality [
      • Piccinelli C.
      • Carna P.
      • Stringhini S.
      • Sebastiani G.
      • Demaria M.
      • Marra M.
      • et al.
      The contribution of behavioural and metabolic risk factors to socioeconomic inequalities in mortality: the Italian Longitudinal Study.
      Int J Publ Health. 2018; 63: 325-335
      ] found that about 20% of cardiovascular mortality was explained by exposure to these risk factors.
      In Italy, a nationwide longitudinal population-based study found among men with primary education a 35% higher risk of death compared to men with a university degree both for all causes and for CVD, and a 25% higher risk of CHD (for women, 40%, and 45%, respectively) [
      • Petrelli A.
      • Zengarini N.
      • Demuru E.
      • Giorgi Rossi P.
      • Sebastiani G.
      • Gaudio R.
      • et al.
      Differences in mortality by educational level in Italy (2012-2014).
      Epidemiol Prev. 2018; 42: 288-300
      ]. Moreover, strong geographic variability was observed, with mortality from CVD higher in the southern regions of Italy, independently of social status [
      • Petrelli A.
      • Di Napoli A.
      • Sebastiani G.
      • Rossi A.
      • Giorgi Rossi P.
      • Demuru E.
      • et al.
      Italian Atlas of mortality inequalities by education level.
      Epidemiol Prev. 2019; 43: 1-120
      ]. Socioeconomic inequalities in the incidence of CVD have also been demonstrated in Italy, mainly through studies conducted in some metropolitan areas [
      • Petrelli A.
      • Gnavi R.
      • Marinacci C.
      • Costa G.
      Socioeconomic inequalities in coronary heart disease in Italy: a multilevel population-based study.
      Soc Sci Med. 2006; 63: 446-456
      ,
      • Ricceri F.
      • Sacerdote C.
      • Giraudo M.T.
      • Fasanelli F.
      • Lenzo G.
      • Galli M.
      • et al.
      The association between educational level and cardiovascular and cerebrovascular diseases within the EPICOR study: new evidence for an old inequality problem.
      PLoS One. 2016; 11e0164130
      ], while no studies are available regarding geographic heterogeneity of socioeconomic inequalities.
      The present study aimed at evaluating education differences in incidence of CVD and of CHD in Italy, and at assessing the role of behavioral (smoking, physical inactivity, overweight) and biological risk factors (diabetes and hypertension) in explaining the gradient of education inequalities in a large nationwide cohort of residents. We also investigated geographic heterogeneity of education differences in CVD and CHD incidence, and of the mediating role of behavioral and biological risk factors. For the purpose of the study we used education level as a proxy of socioeconomic status.

      Methods

      Data collection

      The study was conducted on subjects aged 30–74 years participating in the 1999–2000 and 2004–2005 editions of the Italian National Health Interview Survey (NHIS), carried out by the Italian National Institute of Statistics (Istat). The selected households were informed by letter about the purposes and the modalities of the conduction of the survey and were also reassured about confidentiality and protection of personal data. With the exception of some sensitive information, as the response to the survey was mandatory by law, formal consent to participate was not required. For the purpose of the study, subjects participating in either of the two editions of the NHIS were pooled and followed-up for mortality and morbidity by means of deterministic record-linkage with the Istat National Registry of Mortality and with the Italian Ministry of Health's National Hospital Discharge Database, respectively [
      • Sebastiani G.
      • Di Filippo P.
      • Demaria M.
      • Caranci N.
      • Di Minco L.
      • Tamburini C.
      • et al.
      Lo studio longitudinale italiano: integrazione delle indagini sulla salute con dati di mortalità e ospedalizzazione.
      Metodologia e potenzialità di utilizzo, Istat2019
      ]. The follow-up of the two pooled cohorts, called the Italian Longitudinal Study, has been authorized within the National Statistical Programme. All data were fully anonymized before we accessed them.
      The NHIS was conducted on representative samples of the resident population in Italy of 52,332 families and 140,011 individuals and of 50,474 families and 128,040 individuals in the 1999–2000 and 2004–2005 editions, respectively. Data were collected by means of paper-and-pencil interviewing (PAPI) carried out in four distinct phases on a quarterly basis to eliminate the seasonal effect on health. Participation was 87% (1999–2000) and 83% (2004–2005). The survey provides detailed information on health conditions, including self-perceived health, long-term chronic diseases, disability, lifestyles and use of health services, as well as information on individual and household socioeconomic characteristics [
      • Istat
      Condizioni di salute e ricorso ai servizi sanitari. Nota metodologica.
      2008
      https://www.istat.it/it/archivio/10836
      Date accessed: October 19, 2021
      ]. The Istat National Registry of Mortality registers all death events occurring in Italy and their causes. The Ministry of Health's National Hospital Discharge Database contains information concerning all hospital admissions). The linkage key or personal information necessary to reconstruct the database was available for 92% (1999–2000) and 98.3% (2004–2005) of the sample.
      The proportion between observed and expected deaths and hospital discharges for 1999–2000 and 2004–2005 follow up were 90.5% and 89.9% and 85.3% and 88.9%, respectively. All subjects were followed up for 10 years after the date of the interview, unless the outcome or death occurred earlier.
      Our study was conducted on 132,686 individuals (64,329 men and 68,357 women) aged 35–74 years at the time of the interview, a lower age limit motivated on the one hand by the need for participants to have reached a sufficiently stable education level, and on the other hand, to exclude age groups with very low CVD and CHD incidence. Given the long follow-up of the study, the upper limit of 74 years was chosen to avoid the inclusion of health events in very old subjects, as the quality of information on causes of mortality and morbidity decreases among older people, especially the extreme elderly, and because in education differences in morbidity this group are generally attenuated by the lower survival of the less educated.

      Measurements

      The outcomes investigated were incident cases of CVD and CHD occurred after the interview, between 2000 and 2009 for the 1999–2000 cohort and between 2005 and 2014 for the 2004–2005 cohort. Two dichotomous variables were created based on ICD codes in death and hospital admissions registries: one for CVD (ICD-9:390–459; ICD-10:I00–I99) and one for CHD (ICD-9:410-414; ICD-10:I20–I25). Subjects who reported at baseline to have been diagnosed with myocardial infarction, angina pectoris, other cardiac diseases, or stroke were excluded as prevalent cases from the analysis on CVD, while those with a diagnosis of myocardial infarction or angina pectoris at baseline were excluded as prevalent cases from the analysis on CHD. Overall, 1,162,639 person-years of observation were available for the analysis on CVD and 1,232,548 person-years for that on CHD.
      Information on sociodemographics (age, gender, cohort, education level, household type, macro area of residence) and on exposure to behavioral (smoking, physical activity, and overweight/obesity) and biological (diabetes, hypertension) risk factors (BBRF) at baseline was drawn from the NHIS.
      Education level was used as a proxy indicator of socioeconomic status, classified into four categories: primary school or less (up to 5 years schooling), low secondary school (8 years), high school diploma (13 years), and university degree (≥17 years) [
      • Cardano M.
      • Marinacci C.
      La rilevazione della posizione sociale.
      Epidemiol Prev. 2004; 28: 124-142
      ].
      Regarding individual behaviors, smoking was classified into five categories of lifetime smoking history, based on pack-years smoked (20 cigarettes per pack): never smoker (0 py), 0.1–10 py, 10.1–20 py, 20.1–30 py, and >30 py.
      Physical activity practiced with continuity was categorized into four levels: intense activity (sports: competitive or not), regular activity (gym, biking, etc.), light activity (walking, sweet gym, etc.), or no physical activity.
      Overweight and obesity were derived from the body mass index (BMI), calculated on the self-reported height and weight in the survey, according to a standard procedure. Based on the WHO classification, BMI was categorized as normal weight (18.5 ≤ BMI<25), underweight (BMI <18.5), overweight (25≤BMI ≤ 30), and obese (BMI > 30).
      Information on biological risk factors was collected through self-reports, asking subjects to indicate whether, at the time of the interview, they were affected by any of the chronic diseases on a list that included diabetes and hypertension, and confirmed by a diagnosis [
      • Istat
      Condizioni di salute e ricorso ai servizi sanitari. Nota metodologica.
      2008
      https://www.istat.it/it/archivio/10836
      Date accessed: October 19, 2021
      ].

      Statistical analysis

      Baseline characteristics of the study population were expressed as frequencies and percentages and compared across education levels using Pearson's Chi Square.
      Mediation analyses were performed through the approach for multiple mediators [
      • VanderWeele T.J.
      • Vansteelandt S.
      Mediation analysis with multiple mediators.
      Epidemiol Methods. 2013; 2: 95-115
      ], an extension of the counterfactual method to survival outcomes, based on the computation of appropriate weights [
      • Fasanelli F.
      • Giraudo M.T.
      • Ricceri F.
      • Valeri L.
      • Zugna D.
      Marginal time-dependent causal effects in mediation analysis with survival data.
      Am J Epidemiol. 2019; 188: 967-974
      ].
      This technique does not necessitate any specific statistical assumption for mediators and allows partitioning the total effect (TE) of an exposure on an outcome into pure direct effect (PDE) and natural indirect effect (NIE).
      NIE refers to the portion of the total effect of education level on the incidence of cardiovascular diseases (or coronary heart disease) explained by mediators, while PDE describes the remaining part, which would be observed if the path from education level to the mediators were absent. TE is thus obtained by multiplying NIE by PDE. Confidence Intervals for the effects were built considering a 95% percentile bootstrap method. We used CVD and CHD as the outcomes, education level as the exposure, behavioral (smoking, physical activity, and BMI) and biological (diabetes and hypertension) risk factors as the mediators, and age (10-year classes), cohort, household type (single, couple with children, couple without children, single parent), and geographic macro area (North-West, North-East, Center, South) as adjustment covariates. Proportions mediated by BBRF were computed using the formula: ((HR NDE ∗ (HR NIE -1))/(HR NDE ∗ HR NIE -1), following the method proposed by VanderWeele and Vansteelandt [
      • Vanderweele T.J.
      • Vansteelandt S.
      Odds ratios for mediation analysis for a dichotomous outcome.
      Am J Epidemiol. 2010; 172: 1339-1348
      ,
      • Valeri L.
      • VanderWeele T.J.
      SAS macro for causal mediation analysis with survival data.
      Epidemiology. 2015; 26: e23-e24
      ,
      • Vanderweele T.J.
      Explanation in causal inference. Methods for mediation and interaction.
      Oxford University Press, 2015
      ].
      The contribution of BBRF risk factors to education inequalities in CVD or CHD risk was also assessed through risk attenuation analysis conducted through Cox proportional hazards models (the results are shown in Supplementary Table 1). In the first step (Model A), we estimated the HR by education level, taking into account age, cohort, household type, and, only for the models covering all of Italy, the geographic macro area as well. In the second step (Model B), BBRF risk factors were then added to the models. The explained fraction (EF), assessing the risk attenuation by education comparing the two models, was calculated by means of the following formula [
      • Laaksonen M.
      • Talala K.
      • Martelin T.
      • Rahkonen O.
      • Roos E.
      • Helakorpi S.
      • et al.
      Health behaviours as explanations for educational level differences in cardiovascular and all-cause mortality: a follow-up of 60 000 men and women over 23 years.
      Eur J Publ Health. 2008; 18: 38-43
      ]:
      ((Model A HR-Model B HR)/(Model A HR-1))∗100


      All analyses were stratified by sex and performed both at the national level and by macro area.
      Mediation analyses were conducted in R (version 4.0.5), while all other analyses were performed using the STATA V.13 software.

      Results

      During a mean follow-up of 8.77 ± 2.38 years, 22,214 participants had a cardiovascular event. For coronary heart disease, mean follow-up duration was 9.30 ± 1.71 years, during which 6173 subjects had a CHD event.
      As shown in Table 1, Table 2, the proportion of women with primary school or less was 40.9%, while among men was 31.6%. The proportion of graduates was 8.8% among men, and 7.4% among women.
      Table 1Baseline characteristics of men aged 35–74 years included in the study, by education level.
      University degreeHigh school diplomaLow secondary schoolPrimary school or lessTotalp-value
      N%N%N%N%N%
      54698.814,12922.024,20137.620,35031.664,329100.0
      Cohort
       2000251644.5653346.211,52347.611,69857.532,27050.2<0.001
       2005313355.5759653.812,67852.4865242.532,05949.8
      Age
       35-442016357591641.9983040.615217.519,28330.0<0.001
       45-54186333.0455332.2732130.3361217.817,34927.0
       55-64118120.9244617.3466119.3696434.215,25223.7
       65-7458910.412148.623899.9825340.612,44519.4
      Geographic area
       North-West116820.7310822.0542322.4402019.813,71921.3<0.001
       North-East117620.8283120.0540922.4402219.813,43820.9
       Center112619.9284920.2415117.2355317.511,67918.2
       South217938.65,34138.637.8921838.1875543.025,49339.6
      Pack-years (py)
       Never smoker (0 py)249644.2538938.1796432.9659732.422,44634.9<0.001
       0.1–10 py77113.7191213.5311612.918809.2767911.9
       10.1–20 py76013.5224215.9407816.9238111.7946114.7
       20.1–30 py5309.4167711.9361414.9276013.6858114.7
       >30 py71112.6221815.7457118.9613830.213,63821.2
       missing3816.76914.98583.65942.925243.9
      Body Mass Index
       Underweight190.3400.3780.31210.62580.4<0.001
       Normal weight267147.2601242.6912937.7654332.224,35537.9
       Overweight253744.9667747.311,84448.910,22250.231,28048.6
       Obesity4227.514009.9315013.0346417.0843613.1
      Physical activity status
       Intense96017.0203814.421208.85102.556288.8<0.001
       Regular168429.8389527.6542422.4402919.815,03223.4
       Light or never300553.2819658.016,65768.815,81177.743,66967.9
      Household type
       Single85615.2164711.724069.9243512.0734411.4<0.001
       Couple with children367365.0965468.316,68268.9981248.239,82161.9
       Couple without children86715.4216115.3394716.3730135.914,27622.2
       Single parent2534.56674.711664.88023.928884.5
      Hypertension (prevalence)86915.4223815.8370215.3507024.911,87918.5<0.001
      Diabetes (prevalence)1703.05303.810924.5208610.338786.0<0.001
      CVD cases (cumulative incidence)82914.7211715.0392216.2577428.412,64219.7<0.001
      CHD cases (cumulative incidence)2724.87595.413645.619669.743616.8<0.001
      Table 2Baseline characteristics of women aged 35–74 years included in the study, by education level.
      University degreeHigh school diplomaLow secondary schoolPrimary school or lessTotalp-value
      N%N%N%N%N%
      50657.413,29719.522,06932.327,92640.968,357100.0
      Cohort
       2000211241.7593144.610,13345.915,87256.834,04849.8<0.001
       2005295358.3736655.411,93654.112,05443.234,30950.2
      Age
       35-44228445.1651149.0928442.119276.920,00629.3<0.001
       45-54166332.8392829.5673730.5541319.417,74126.0788
       55-6478815.6190614.3385517.5951034.116,05923.5
       65-743306.59527.221939.911,07639.714,55121.3
      Geographic area
       North-West101820.1290221.8528323.9545919.614,66221.5<0.001
       North-East99019.6250418.8529124.0533619.114,12120.7
       Center110121.7267120.1368816.7494617.712,40618.2
       South195638.6522039.3780735.412,18543.627,16839.7
      Pack-years (py)
       Never smoker (0 py)290757.4749456.413,15959.621,85178.345,41166.4<0.001
       0.1–10 py89817.7233017.5329114.920927.5861112.6
       10.1–20 py4929.7158611.9246111.215685.661078.9
       20.1–30 py3156.27966.013846.39663.534615.1
       >30 py1933.85604.29734.49803.527064.0
       missing2605.15314.08013.64691.720613.0
      Body Mass Index
       Underweight2525.05674.37063.24861.720112.9<0.001
       Normal weight370773.2922969.413,35060.511,85042.438,13655.8
       Overweight89217.6277320.9598627.110,81738.720,46829.9
       Obesity2144.27285.520279.2477317.1774211.3
      Physical activity status
       Intense56311.110137.610484.84461.630704.5<0.001
       Regular126224.9295622.2401818.2355212.711,78817.2
       Light or never324064.0932870.217,00377.023,92885.753,49978.3
      Household type
       Single75614.9155811.7219810.0532819.1984014.4<0.001
       Couple with children301659.6860064.713,89563.010,58837.936,09952.8
       Couple without children69513.7172012.9368316.7953634.215,63422.9
       Single parent59811.8141910.7229310.424748.967849.9
      Hypertension (prevalence)53910.6148511.2334515.2903132.314,40021.1<0.001
      Diabetes (prevalence)751.52732.16593.027259.837325.5<0.001
      CVD cases (cumulative incidence)3977.811498.6234710.6567920.4957214.0<0.001
      CHD cases (cumulative incidence)591.21601.23841.712094.318122.7<0.001
      Among men, the prevalence of never smokers was highest among subjects with a university degree (44.2%) and lowest among the least educated (32.4%), whereas that of smokers >30 pack-years strongly increased with decreasing education, from 12.6% for university degree to 30.2% for primary school or less. Among women, prevalence of never smokers was instead highest in the primary education group (78.3%) and lowest in the high school diploma category (56.4%), while only slight education differences were observed for smokers >30 pack-years.
      Moreover, the prevalence of physical activity, overweight, obesity, hypertension, diabetes, as well as cumulative incidence of CVD and CHD, all increased as education level decreased among both men and women.

      Cardiovascular diseases

      Among men (Table 3), an inverse gradient between education level and CVD risk was found for the whole country and for all the macro areas. Men with primary school or less had a 21% excess risk of CVD compared to those having a university degree, with the strongest differences observed in Central Italy (HR:1.32) and the lowest in the South (HR:1.13). For men with low secondary school, statistically significant HRs were found at the national level (HR = 1.12) and in Central Italy (HR = 1.28), while for those with a high school diploma, statistically significant HRs were found only in Central Italy (HR = 1.29).
      Table 3Hazard ratios (HR) of cardiovascular diseases by education level, overall and by geographic area, for total effect (TE), pure direct effect (PDE) and natural indirect effect (NIE) and % of TE mediated by behavioral/biological factors (NIE). Age 35–74 years at baseline.
      Models adjusted for age, cohort, household type. Models for Italy are also adjusted for geographic area. University degree was used as reference category.
      MalesFemales
      HR95%CI%TEHR95%CI%TE
      ITALYTotal effect (TE)
      High school diploma1.060.97–1.151.100.97–1.26
      Low secondary school1.121.04–1.231.231.09–1.40
      Primary school or less1.211.12–1.331.411.27–1.63
      Pure direct effect (PDE)
      High school diploma1.010.93–1.101.120.98–1.26
      Low secondary school1.040.96–1.141.211.06–1.38
      Primary school or less1.101.01–1.201.351.19–1.55
      Natural indirect effect (NIE)
      High school diploma1.051.02–1.070.770.990.94–1.02n.s.
      Low secondary school1.081.05–1.110.671.010.97–1.05n.s.
      Primary school or less1.101.07–1.130.521.051.01–1.080.16
      NORTH-WESTTotal effect (TE)
      High school diploma0.980.81–1.231.190.83–1.75
      Low secondary school1.090.93–1.351.361.05–1.83
      Primary school or less1.221.02–1.521.371.07–1.87
      Pure direct effect (PDE)
      High school diploma0.930.78–1.141.170.89–1.58
      Low secondary school1.010.86–1.221.321.03–1.75
      Primary school or less1.100.92–1.371.301.01–1.73
      Natural indirect effect (NIE)
      High school diploma1.051.00–1.11n.a.1.020.90–1.11n.s.
      Low secondary school1.071.02–1.130.851.030.93–1.13n.s.
      Primary school or less1.101.05–1.170.531.050.95–1.15n.s.
      NORTH-EASTTotal effect (TE)
      High school diploma0.930.74–1.141.340.99–1.91
      Low secondary school1.110.88–1.361.511.16–2.18
      Primary school or less1.271.01–1.522.001.48–2.77
      Pure direct effect (PDE)
      High school diploma0.880.71–1.071.371.02–1.92
      Low secondary school1.000.81–1.231.471.17–2.08
      Primary school or less1.120.92–1.331.891.45–2.62
      Natural indirect effect (NIE)
      High school diploma1.060.99–1.10n.s.0.980.93–1.05n.s.
      Low secondary school1.101.04–1.150.961.030.97–1.11n.s.
      Primary school or less1.131.07–1.190.541.061.00–1.130.11
      CENTERTotal effect (TE)
      High school diploma1.291.05–1.650.990.78–1.37
      Low secondary school1.281.08–1.641.260.99–1.76
      Primary school or less1.321.11–1.631.441.10–1.83
      Pure direct effect (PDE)
      High school diploma1.180.96–1.510.960.73–1.30
      Low secondary school1.130.95–1.451.190.94–1.60
      Primary school or less1.150.95–1.421.361.06–1.71
      Natural indirect effect (NIE)
      High school diploma1.091.03–1.160.381.040.97–1.11n.a.
      Low secondary school1.131.07–1.200.531.061.00–1.140.27
      Primary school or less1.151.09–1.230.551.060.99–1.15n.s.
      SOUTHTotal effect (TE)
      High school diploma1.060.89–1.231.030.85–1.27
      Low secondary school1.080.95–1.241.080.88–1.37
      Primary school or less1.131.00–1.311.241.03–1.51
      Pure direct effect (PDE)
      High school diploma1.040.90–1.201.090.93–1.31
      Low secondary school1.030.93–1.171.100.97–1.36
      Primary school or less1.070.96–1.251.211.06–1.45
      Natural indirect effect (NIE)
      High school diploma1.020.97–1.07n.s.0.940.86–1.03n.s.
      Low secondary school1.051.00–1.090.600.990.90–1.07n.s.
      Primary school or less1.051.01–1.100.431.020.93–1.11n.s.
      a Models adjusted for age, cohort, household type. Models for Italy are also adjusted for geographic area. University degree was used as reference category.
      In Italy, BBRF were statistically significant mediators of the association between lower education level, compared to university degree, and CVD risk, as indicated by their NIE: high school diploma (HR:1.05), low secondary school (HR:1.08), primary school or less (HR:1.10), with a mediated proportion of 77%, 67%, and 52%, respectively. HRs of NIE were similar by geographic area, although slightly lower in the South. The proportion of HR mediated by BBFR was also quite similar across areas for the lowest education level, while for the low secondary school it was higher in the North-West (85%) and in the North-East (96%).
      Also among women (Table 3), an inverse gradient between education level and CVD risk, which was stronger than that in men, was found for the whole country and for all the macro areas. Women with primary school or less had a 41% excess risk of CVD compared to those with a university degree, with the strongest differences observed in the North-East (HR:2.00) and the lowest in the South (HR:1.24). HRs for women with low secondary school were also higher in the whole country (HR:1.23), in the North-West (HR:1.36) and in the North-East (HR:1.51), and marginally significantly in Central Italy (HR:1.26).
      The contribution of BBRF to education differences in CVD risk was much lower among women. NIE of BBRF were statistically significant only for those with primary school or less at the national level (HR:1.05), corresponding to 16% of the total effect, and in the North-East (HR:1.06; 11% of the total effect), and for women with low secondary school living in Central Italy (HR:1.06; 27% of the total effect).

      Coronary heart disease

      Compared to men with a university degree (Table 4), the risk of CHD in Italy was higher in all education levels. HRs were similar in the lowest education level across macro areas, while more variable in intermediate education level, with a high risk observed for men with low secondary school living in Central Italy (HR:1.52), compared to those with a university degree.
      Table 4Hazard ratios (HR) of coronary heart diseases by education level, overall and by geographic area, for total effect (TE), pure direct effect (PDE) and natural indirect effect (NIE) and % of TE mediated by behavioral/biological factors (NIE). Age 35–74 years at baseline.
      Models adjusted for age, cohort, household type. Models for Italy are adjusted also for geographic area. University degree was used as reference category.
      MalesFemales
      HR95%CI%TEHR95%CI%TE
      ITALYTotal effect (TE)
      High school diploma1.201.05–1.451.230.83–1.82
      Low secondary school1.191.06–1.401.421.09–2.00
      Primary school or less1.171.03–1.361.611.16–2.37
      Pure direct effect (PDE)
      High school diploma1.110.98–1.331.230.85–1.80
      Low secondary school1.060.93–1.241.371.05–1.98
      Primary school or less1.030.89–1.201.521.11–2.19
      Natural indirect effect (NIE)
      High school diploma1.081.03–1.120.431.000.92–1.10n.s.
      Low secondary school1.121.09–1.180.681.030.95–1.12n.s.
      Primary school or less1.141.11–1.190.841.060.97–1.16n.s.
      NORTH-WESTTotal effect (TE)
      High school diploma1.170.80–1.771.280.55–3.79
      Low secondary school1.260.91–1.811.510.75–4.83
      Primary school or less1.120.77–1.651.770.98–5.23
      Pure direct effect (PDE)
      High school diploma1.080.72–1.671.300.50–4.27
      Low secondary school1.130.81–1.611.520.75–5.46
      Primary school or less0.970.68–1.401.710.88–5.46
      Natural indirect effect (NIE)
      High school diploma1.081.00–1.180.540.980.80–1.20n.s.
      Low secondary school1.121.03–1.230.520.990.79–1.20n.s.
      Primary school or less1.161.07–1.281.261.030.81–1.26n.s.
      NORTH-EASTTotal effect (TE)
      High school diploma0.900.63–1.291.280.69–6.92
      Low secondary school0.950.68–1.451.831.00–10.31
      Primary school or less1.170.84–1.641.851.06–9.12
      Pure direct effect (PDE)
      High school diploma0.830.58–1.201.290.68–6.71
      Low secondary school0.820.60–1.241.690.93–9.14
      Primary school or less0.990.72–1.391.670.95–8.37
      Natural indirect effect (NIE)
      High school diploma1.081.00–1.16n.a.0.990.89–1.14n.s.
      Low secondary school1.161.07–1.27n.a.1.080.97–1.21n.s.
      Primary school or less1.181.11–1.281.061.110.99–1.31n.s.
      CENTERTotal effect (TE)
      High school diploma1.480.86–1.771.180.49–3.22
      Low secondary school1.521.08–1.721.260.72–3.15
      Primary school or less1.200.72–1.521.570.89–3.48
      Pure direct effect (PDE)
      High school diploma1.310.72–1.561.120.45–3.03
      Low secondary school1.310.88–1.511.180.67–3.06
      Primary school or less1.020.59–1.351.470.85–3.37
      Natural indirect effect (NIE)
      High school diploma1.131.06–1.200.361.050.87–1.23n.s.
      Low secondary school1.171.09–1.260.411.060.88–1.22n.s.
      Primary school or less1.181.10–1.280.891.060.90–1.24n.s.
      SOUTHTotal effect (TE)
      High school diploma1.250.98–1.641.190.76–1.82
      Low secondary school1.140.9–1.481.250.81–2.16
      Primary school or less1.180.95–1.51.411.01–2.16
      Pure direct effect (PDE)
      High school diploma1.180.93–1.571.270.81–1.79
      Low secondary school1.040.80–1.361.290.79–2.10
      Primary school or less1.070.86–1.381.451.00–2.13
      Natural indirect effect (NIE)
      High school diploma1.060.98–1.12n.s.0.940.79–1.09n.s.
      Low secondary school1.101.02–1.160.730.970.85–1.15n.s.
      Primary school or less1.101.03–1.160.600.970.85–1.15n.s.
      a Models adjusted for age, cohort, household type. Models for Italy are adjusted also for geographic area. University degree was used as reference category.
      Statistically significant NIE of BBRF were found at the national level for men with any education level below a university degree. For high school diploma, the HR was 1.08, corresponding to 43% of the total effect, for low secondary school, HR was 1.12 (68% of the total effect), and for primary school or less, HR was 1.14 (84% of the total effect). Proportions mediated by BBRF showed high variability across geographic areas for all education levels.
      Among women (Table 4), a stronger inverse education gradient in CHD risk was found for the whole country and for all the macro areas, compared to that among men. Women with primary school or less had a 61% excess risk of CHD compared to those having a university degree, with the strongest differences observed in the North-East (HR:1.85) and the lowest in the South (HR:1.41). Women with low secondary education were also at significantly higher risk of CHD than those with a university degree at the national level (HR = 1.42) and in the North-East (HR = 1.83).
      NIE of BBRF were never statistically significant among women, neither at the national level nor in any geographic area, indicating no significant mediating effects of BBRF on the CHD education gradient.

      Discussion

      Our study shows the presence of education inequalities in CVD and CHD incidence for both men and women, although stronger among women. After controlling for sociodemographic factors, the least educated men showed a 21% higher risk of CVD and a 17% higher risk of CHD compared to men with the highest education, while the least educated women had a 41% higher risk of CVD and a 61% higher risk of CHD compared to the most educated women. High mediated proportions of education differences for both CVD and CHD were estimated for men, while for women they were much lower for CVD and not significant for CHD.
      Some heterogeneity by macro area was observed in the contribution of BBRF to the educational gradient in both CVD and CHD, which however appears in part due to the large uncertainty of the risk estimates after stratification by geographical area, due to the relatively small number of observed cases in each area, especially for CHD.
      This is the first study which assessed education inequalities in the incidence of CVD and CHD in Italy using a representative sample of the entire resident population. Our results are consistent with previous population-based studies conducted on Italian metropolitan cohorts [
      • Petrelli A.
      • Gnavi R.
      • Marinacci C.
      • Costa G.
      Socioeconomic inequalities in coronary heart disease in Italy: a multilevel population-based study.
      Soc Sci Med. 2006; 63: 446-456
      ,
      • Ricceri F.
      • Sacerdote C.
      • Giraudo M.T.
      • Fasanelli F.
      • Lenzo G.
      • Galli M.
      • et al.
      The association between educational level and cardiovascular and cerebrovascular diseases within the EPICOR study: new evidence for an old inequality problem.
      PLoS One. 2016; 11e0164130
      ], as well as with the ample international literature on socioeconomic inequalities in cardiovascular diseases, which in general shows weaker gradients among men in Southern European countries [
      • Khaing W.
      • Vallibhakara S.A.
      • Attia J.
      • McEvoy M.
      • Thakkinstian A.
      Effects of education and income on cardiovascular outcomes: a systematic review and meta-analysis.
      Eur J Prev Cardiol. 2017; 24: 1032-1042
      ,
      • Veronesi G.
      • Ferrario M.M.
      • Kuulasmaa K.
      • Bobak M.
      • Chambless L.E.
      • Salomaa V.
      • et al.
      Educational class inequalities in the incidence of coronary heart disease in Europe.
      Heart. 2016; 102: 958-965
      ].
      The stronger inequalities among women observed in our study appear consistent with the results of a recent systematic review, which estimated that education differences in CVD incidence were about 20–25% higher among women, compared to men [
      • Backholer K.
      • Peters S.A.E.
      • Bots S.H.
      • Peeters A.
      • Huxley R.R.
      • Woodward M.
      Sex differences in the relationship between socioeconomic status and cardiovascular disease: a systematic review and meta-analysis.
      J Epidemiol Community Health. 2017; 71: 550-557
      ].
      Regarding the mediating role of biological and behavioral risk factors, it is not easy to compare our results with the ample international literature of the topic, because of the heterogeneity in the methods of estimation and in the measurement of risk factors, as well as because of frequent lack of stratification by sex. Such heterogeneity in methods has likely contributed to the great variability among studies in the estimates of the contribution of BBRF factors to education differences in CVD/CHD outcomes, with results in a range from less than 20% to more than 70% [
      • Veronesi G.
      • Ferrario M.M.
      • Kuulasmaa K.
      • Bobak M.
      • Chambless L.E.
      • Salomaa V.
      • et al.
      Educational class inequalities in the incidence of coronary heart disease in Europe.
      Heart. 2016; 102: 958-965
      ,
      • Beauchamp A.
      • Peeters A.
      • Wolfe R.
      • Turrell G.
      • Harriss L.R.
      • Giles G.G.
      • et al.
      Inequalities in cardiovascular disease mortality: the role of behavioural, physiological and social risk factors.
      J Epidemiol Community Health. 2010; 64: 542-548
      ,
      • Kershaw K.N.
      • Droomers M.
      • Robinson W.R.
      • Carnethon M.R.
      • Daviglus M.L.
      • Monique Verschuren W.M.
      Quantifying the contributions of behavioral and biological risk factors to socioeconomic disparities in coronary heart disease incidence: the MORGEN study.
      Eur J Epidemiol. 2013; 28: 807-814
      ,
      • Khang Y.H.
      • Lynch J.W.
      • Yang S.
      • Harper S.
      • Yun S.C.
      • Jung-Choi K.
      • et al.
      The contribution of material, psychosocial, and behavioral factors in explaining educational and occupational mortality inequalities in a nationally representative sample of South Koreans: relative and absolute perspectives.
      Soc Sci Med. 2009; 68: 858-866
      ,
      • Kivimaki M.
      • Lawlor D.A.
      • Davey Smith G.
      • Kouvonen A.
      • Virtanen M.
      • Elovainio M.
      • et al.
      Socioeconomic position, co-occurrence of behavior-related risk factors, and coronary heart disease: the Finnish Public Sector study.
      Am J Publ Health. 2007; 97: 874-879
      ,
      • Nordahl H.
      • Rod N.H.
      • Frederiksen B.L.
      • Andersen I.
      • Lange T.
      • Diderichsen F.
      • et al.
      Education and risk of coronary heart disease: assessment of mediation by behavioral risk factors using the additive hazards model.
      Eur J Epidemiol. 2013; 28: 149-157
      ,
      • Panagiotakos D.
      • Georgousopoulou E.
      • Notara V.
      • Pitaraki E.
      • Kokkou E.
      • Chrysohoou C.
      • et al.
      Education status determines 10-year (2002-2012) survival from cardiovascular disease in Athens metropolitan area: the ATTICA study, Greece.
      Health Soc Care Community. 2016; 24: 334-344
      ,
      • Silhol R.
      • Zins M.
      • Chauvin P.
      • Chaix B.
      Investigating the spatial variability in incidence of coronary heart disease in the Gazel cohort: the impact of area socioeconomic position and mediating role of risk factors.
      J Epidemiol Community Health. 2011; 65: 137-143
      ], although cultural and demographic differences between countries may have also concurred [
      • Kershaw K.N.
      • Droomers M.
      • Robinson W.R.
      • Carnethon M.R.
      • Daviglus M.L.
      • Monique Verschuren W.M.
      Quantifying the contributions of behavioral and biological risk factors to socioeconomic disparities in coronary heart disease incidence: the MORGEN study.
      Eur J Epidemiol. 2013; 28: 807-814
      ,
      • Mejean C.
      • Droomers M.
      • van der Schouw Y.T.
      • Sluijs I.
      • Czernichow S.
      • Grobbee D.E.
      • et al.
      The contribution of diet and lifestyle to socioeconomic inequalities in cardiovascular morbidity and mortality.
      Int J Cardiol. 2013; 168: 5190-5195
      ]. A recent review estimated that 18% of the differences in CVD between extreme education classes were attributable to smoking, alcohol consumption, physical activity, and diet, although with a greater contribution found in Northern Europe and North America [
      • Petrovic D.
      • de Mestral C.
      • Bochud M.
      • Bartley M.
      • Kivimaki M.
      • Vineis P.
      • et al.
      The contribution of health behaviors to socioeconomic inequalities in health: a systematic review.
      Prev Med. 2018; 113: 15-31
      ].
      Regarding the low mediating effect of CVD/CHD education differences by BBRF among women in our study, such a finding seems at odds with most results in the literature, although only a few studies have investigated this aspect, mostly conducted in UK and Scandinavia. Three large prospective studies from UK and Sweden on women found that smoking, alcohol consumption, physical activity, and BMI accounted for about 70% of the association between age at leaving full-time education and CHD risk [
      • Floud S.
      • Balkwill A.
      • Moser K.
      • Reeves G.K.
      • Green J.
      • Beral V.
      • et al.
      The role of health-related behavioural factors in accounting for inequalities in coronary heart disease risk by education and area deprivation: prospective study of 1.2 million UK women.
      BMC Med. 2016; 14: 145
      ,
      • Kuper H.
      • Adami H.O.
      • Theorell T.
      • Weiderpass E.
      Psychosocial determinants of coronary heart disease in middle-aged women: a prospective study in Sweden.
      Am J Epidemiol. 2006; 164: 349-357
      ,
      • Lawlor D.A.
      • Ebrahim S.
      • Davey Smith G.
      Adverse socioeconomic position across the lifecourse increases coronary heart disease risk cumulatively: findings from the British women's heart and health study.
      J Epidemiol Community Health. 2005; 59: 785-793
      ]. Three other Scandinavian studies observed lower, although substantial, risk attenuation of education differences by behavioral risk factors in mortality from CHD among women [
      • Laaksonen M.
      • Talala K.
      • Martelin T.
      • Rahkonen O.
      • Roos E.
      • Helakorpi S.
      • et al.
      Health behaviours as explanations for educational level differences in cardiovascular and all-cause mortality: a follow-up of 60 000 men and women over 23 years.
      Eur J Publ Health. 2008; 18: 38-43
      ,
      • Ernstsen L.
      • Bjerkeset O.
      • Krokstad S.
      Educational inequalities in ischaemic heart disease mortality in 44,000 Norwegian women and men: the influence of psychosocial and behavioural factors. The HUNT Study.
      Scand J Publ Health. 2010; 38: 678-685
      ,
      • Hardarson T.
      • Gardarsdóttir M.
      • Gudmundsson K.T.
      • Thorgeirsson G.
      • Sigvaldason H.
      • Sigfússon N.
      The relationship between educational level and mortality. The Reykjavík Study.
      J Intern Med. 2001; 249: 495-502
      ]. In contrast, a Danish study found a low contribution of smoking, physical activity and BMI to education differences in CHD incidence among both women and men [
      • Nordahl H.
      • Rod N.H.
      • Frederiksen B.L.
      • Andersen I.
      • Lange T.
      • Diderichsen F.
      • et al.
      Education and risk of coronary heart disease: assessment of mediation by behavioral risk factors using the additive hazards model.
      Eur J Epidemiol. 2013; 28: 149-157
      ].
      Different hypotheses may explain the low contribution of lifestyle factors to differences in CVD and CHD risk among women in Italy. First of all, smoking has still a direct social gradient among women in Italy, with high proportions of smokers among those with high education, especially among older women, due to a delay in the change of the epidemic curve of tobacco consumption. Lack of control for some relevant behavioral factors, which were not available in our dataset, could partly explain our findings, for example HDL cholesterol, which was found to explain the largest proportion of education inequalities in CHD incidence among women [
      • Veronesi G.
      • Ferrario M.M.
      • Kuulasmaa K.
      • Bobak M.
      • Chambless L.E.
      • Salomaa V.
      • et al.
      Educational class inequalities in the incidence of coronary heart disease in Europe.
      Heart. 2016; 102: 958-965
      ]. Other BBRF factors not available in our study, such as unhealthy diet or alcohol consumption, could play a role, but small education differences in nutrient intake have been observed among Italian women [
      • D'Avanzo B.
      • La Vecchia C.
      • Braga C.
      • Franceschi S.
      • Negri E.
      • Parpinel M.
      Nutrient intake according to education, smoking, and alcohol in Italian women.
      Nutr Cancer. 1997; 28: 46-51
      ], while the social patterns of alcohol consumption is direct, as for smoking [
      • Marmot M.G.
      • Shipley M.J.
      • Hemingway H.
      • Head J.
      • Brunner E.J.
      Biological and behavioural explanations of social inequalities in coronary heart disease: the Whitehall II study.
      Diabetologia. 2008; 51: 1980-1988
      ].
      Several other factors have been hypothesized to explain the different education gradient in CVD incidence between sexes, including differential gender exposure to psychosocial factors, differential identification of subjects at high CHD and CVD risk, differential access and adherence to preventative treatment, and risk factor management across levels of socioeconomic disadvantage [
      • Backholer K.
      • Peters S.A.E.
      • Bots S.H.
      • Peeters A.
      • Huxley R.R.
      • Woodward M.
      Sex differences in the relationship between socioeconomic status and cardiovascular disease: a systematic review and meta-analysis.
      J Epidemiol Community Health. 2017; 71: 550-557
      ].
      Psychosocial factors, such as career paths, work and life conditions, and less control over life and circumstances, could in particular contribute. It is noteworthy that Italy has the strongest gender disparities among the 28 EU countries in the performance of domestic work and care for children [
      European Commission
      Women and unpaid work: recognise, reduce, redistribute!.
      https://ec.europa.eu/social/main.jsp?langId=en&catId=1196&newsId=2492&furtherNews=yes
      Date: 2016
      Date accessed: October 19, 2021
      ], which are mostly responsibility of women [
      • Mencarini L.
      Soddisfazione e uso del tempo nelle coppie italiane.
      in: Romano M.C. Mencarini L. Tanturri M.L. Uso del tempo e ruoli di genere Tra lavoro e famiglia nel ciclo di vita. Istat, Rome2012: 41-64
      ]. The burden posed by domestic activities, especially when combined with paid work, may overload women, activating stress pathways that lead to alterations in the vascular system and to an increase in CVD risk. A few studies have actually observed a higher CHD risk among women who take care of children or grandchildren and also engage in paid work [
      • D'Ovidio F.
      • d'Errico A.
      • Scarinzi C.
      • Costa G.
      Increased incidence of coronary heart disease associated with "double burden" in a cohort of Italian women.
      Soc Sci Med. 2015; 135: 40-46
      ,
      • Haynes S.G.
      • Feinleib M.
      Women, work and coronary heart disease: prospective findings from the Framingham heart study.
      Am J Publ Health. 1980; 70: 133-141
      ,
      • Lee S.
      • Colditz G.
      • Berkman L.
      • Kawachi I.
      Caregiving to children and grandchildren and risk of coronary heart disease in women.
      Am J Publ Health. 2003; 93: 1939-1944
      ], which would indicate that work-family conflicts may also play a role in the development of CVD. Furthermore, various studies have found that women are more exposed than men to the main work stressors associated with CVD risk, especially those in low-grade occupations, such as low job control [
      • Hooftman W.E.
      • van der Beek A.J.
      • Bongers P.M.
      • van Mechelen W.
      Gender differences in self-reported physical and psychosocial exposures in jobs with both female and male workers.
      J Occup Environ Med. 2005; 47: 244-252
      ,
      • Messing K.
      • Stock S.R.
      • Tissot F.
      Should studies of risk factors for musculoskeletal disorders be stratified by gender? Lessons from the 1998 Québec Health and Social Survey.
      Scand J Work Environ Health. 2009; 35: 96-112
      ,
      • d'Errico A.
      • Cardano M.
      • Landriscina T.
      • Marinacci C.
      • Pasian S.
      • Petrelli A.
      • et al.
      Workplace stress and prescription of antidepressant medications: a prospective study on a sample of Italian workers.
      Int Arch Occup Environ Health. 2011; 84: 413-424
      ,
      • Josephson M.
      • Pernold G.
      • Ahlberg-Hultén G.
      • Härenstam A.
      • Theorell T.
      • Vingård E.
      • et al.
      Differences in the association between psychosocial work conditions and physical work load in female- and male-dominated occupations. MUSIC-Norrtälje Study Group.
      Am Ind Hyg Assoc J. 1999; 60: 673-678
      ], high psychological demand [
      • Sterud T.
      Work-related gender differences in physician-certified sick leave: a prospective study of the general working population in Norway.
      Scand J Work Environ Health. 2014; 40: 361-369
      ], and high job strain [
      • d'Errico A.
      • Cardano M.
      • Landriscina T.
      • Marinacci C.
      • Pasian S.
      • Petrelli A.
      • et al.
      Workplace stress and prescription of antidepressant medications: a prospective study on a sample of Italian workers.
      Int Arch Occup Environ Health. 2011; 84: 413-424
      ,
      • Ibrahim S.A.
      • Scott F.E.
      • Cole D.C.
      • Shannon H.S.
      • Eyles J.
      Job strain and self-reported health among working women and men: an analysis of the 1994/5 Canadian National Population Health Survey.
      Women Health. 2001; 33: 105-124
      ,
      • Karlqvist L.
      • Tornqvist E.W.
      • Hagberg M.
      • Hagman M.
      • Toomingas A.
      Self-reported working conditions of VDU operators and associations with musculoskeletal symptoms: a cross-sectional study focussing on gender differences.
      Int J Ind Ergon. 2002; 30: 277-294
      ].

      Strengths and limitations

      The major strength of the study is the use of the Italian Longitudinal Study, which allowed us to evaluate education inequalities in CVD and CHD adjusted for several potential confounders available at baseline in a large representative sample of the Italian population. Another strength is the use of an advanced statistical method, based on the counterfactual approach, to assess the mediation effect of BBRF on such inequalities. In fact, this methodology is considered more reliable than risk attenuation, whose results can be biased by interaction between the exposure and the mediators and whose results provide a more uncertain interpretation, as this method does not allow an evaluation of the statistical significance of the explained fraction [
      • Richiardi L.
      • Bellocco R.
      • Zugna D.
      Mediation analysis in epidemiology: methods, interpretation and bias.
      Int J Epidemiol. 2013; 42: 1511-1519
      ]. Among the limitations, the use of self-reported information on BBRF may have led to a certain degree of misclassification of the exposure to BBRF [
      • Patrick D.L.
      • Cheadle A.
      • Thompson D.C.
      • Diehr P.
      • Koepsell T.
      • Kinne S.
      The validity of self-reported smoking: a review and meta-analysis.
      Am J Publ Health. 1994; 84: 1086-1093
      ,
      • Vartiainen E.
      • Seppälä T.
      • Lillsunde P.
      • Puska P.
      Validation of self reported smoking by serum cotinine measurement in a community-based study.
      J Epidemiol Community Health. 2002; 56: 167-170
      ], with the consequence of an imperfect adjustment for these factors and to an inaccurate estimation of the explained fraction [
      • Stringhini S.
      • Dugravot A.
      • Shipley M.
      • Goldberg M.
      • Zins M.
      • Kivimäki M.
      • et al.
      Health behaviours, socioeconomic status, and mortality: further analyses of the British Whitehall II and the French GAZEL prospective cohorts.
      PLoS Med. 2011; 8
      ]. In addition, information on BBRF was collected only at baseline, without any knowledge on duration of exposure, except for smoking, feature which may have concurred to underestimate their association with the outcomes investigated, and consequently, their mediating role. Further, self-reported information on diabetes and hypertension may have led to an underestimation of their prevalence, especially among men, although apparently very low for diabetes and slightly higher for hypertension [
      • Tolonen H.
      • Koponen P.
      • Mindell J.S.
      • Männistö S.
      • Giampaoli S.
      • Dias C.M.
      • et al.
      Under-estimation of obesity, hypertension and high cholesterol by self-reported data: comparison of self-reported information and objective measures from health examination surveys.
      Eur J Publ Health. 2014; : 941-948
      ].
      Another limitation is that we could not take into account material and psychosocial factors, such as negative life events, family conflicts, or financial difficulties [
      • van Oort F.V.
      • van Lenthe F.J.
      • Mackenbach J.P.
      Material, psychosocial, and behavioural factors in the explanation of educational inequalities in mortality in The Netherlands.
      J Epidemiol Community Health. 2005; 59: 214-220
      ], which are possible confounders of the association between BBRF and occurrence of CVD or CHD. Furthermore, we could not consider in the analysis other behavioral factors, such as dietary habit and alcohol consumption, which could have led to an underestimation of the real fraction explained by lifestyle risk factors.
      Regarding the use of education level as a single proxy for socioeconomic status, both its strengths and limitations should be mentioned. The strength of education level is that it rarely changes over time and is, at least partially, correlated with income and employment conditions. Education level, as an indicator of an individual's cultural and social capital, influences health conditions through multiple mechanisms, which influence behaviors and access to health services throughout life. However, some authors have underlined that education level may lead to an underestimation of the true social gradient compared to occupational social class [
      • D'Errico A.
      • Ricceri F.
      • Stringhini S.
      • Carmeli C.
      • Kivimaki M.
      • Bartley M.
      • et al.
      Consortium Lifepath. Socioeconomic indicators in epidemiologic research: a practical example from the LIFEPATH study.
      PLoS One. 2017 May 30; 12 (2017)e0178071
      ].

      Conclusions

      This study showed education differences in the incidence of CVD and of CHD, which were stronger among women. Adjustment for smoking habit, physical inactivity, BMI, diabetes, and hypertension strongly reduced the risk of CVD and CHD by education level in men, while it had only a small effect in women. However, the variability by gender and geographical area observed in our results regarding both the education differences in CVD/CHD and the contribution of BBRF to these differences, would indicate that prevention programs targeting vulnerable groups may not be able to reduce social inequalities in cardiovascular health without taking into account such a variability in designing and addressing these programs. A recent systematic review pointed out that community-based primary prevention in combination with targeting high-risk groups could be an effective way to reduce social inequalities in CVD [
      • Beauchamp A.
      • Peeters A.
      • Tonkin A.
      • Turrell G.
      Best practice for prevention and treatment of cardiovascular disease through an equity lens: a review.
      Eur J Cardiovasc Prev Rehabil. 2010; 17: 599-606
      ]. Only 5% of clinical trials or observational interventions have been found to incorporate measures of socioeconomic status in their assessment [
      • Havranek E.P.
      • Mujahid M.S.
      • Barr D.A.
      • Blair I.V.
      • Cohen M.S.
      • Cruz-Flores S.
      • et al.
      Social determinants of risk and outcomes for cardiovascular disease: a scientific statement from the American heart association.
      Circulation. 2015; 132: 873-898
      ,
      • Callander E.J.
      • McDermott R.
      Measuring the effects of CVD interventions and studies across socioeconomic groups: a brief review.
      Int J Cardiol. 2017; 227: 635-643
      ], so interventions that address a potential reduction in socioeconomic inequalities in CVD and CHD must be implemented and assessed.

      Declaration of competing interest

      None.

      Acknowledgements

      Jacqueline M. Costa, translation/English language editing.

      Appendix A. Supplementary data

      The following is the Supplementary data to this article:

      References

        • WHO
        Cardiovascular diseases (CVDs).
        2017
        View in Article
        • Manrique-Garcia E.
        • Sidorchuk A.
        • Hallqvist J.
        • Moradi T.
        Socioeconomic position and incidence of acute myocardial infarction: a meta-analysis.
        J Epidemiol Community Health. 2011; 65: 301-309
        View in Article
        • Mackenbach J.P.
        • Stirbu I.
        • Roskam A.J.
        • Schaap M.M.
        • Menvielle G.
        • Leinsalu M.
        • et al.
        Socioeconomic inequalities in health in 22 European countries.
        N Engl J Med. 2008; 358: 2468-2481
        View in Article
        • Stringhini S.
        • Carmeli C.
        • Jokela M.
        • Avendaño M.
        • Muennig P.
        • Guida F.
        • et al.
        Socioeconomic status and the 25×25 risk factors as determinants of premature mortality: a multi cohort study and meta -analysis of 1·7 million men and women.
        Lancet. 2017; : 389
        View in Article
        • Khaing W.
        • Vallibhakara S.A.
        • Attia J.
        • McEvoy M.
        • Thakkinstian A.
        Effects of education and income on cardiovascular outcomes: a systematic review and meta-analysis.
        Eur J Prev Cardiol. 2017; 24: 1032-1042
        View in Article
        • Backholer K.
        • Peters S.A.E.
        • Bots S.H.
        • Peeters A.
        • Huxley R.R.
        • Woodward M.
        Sex differences in the relationship between socioeconomic status and cardiovascular disease: a systematic review and meta-analysis.
        J Epidemiol Community Health. 2017; 71: 550-557
        View in Article
        • Kannel W.B.
        • McGee D.L.
        Diabetes and cardiovascular disease. The Framingham study.
        JAMA. 1979; 241: 2035-2038
        View in Article
        • Keil U.
        • Kuulasmaa K.
        WHO MONICA Project: risk factors.
        Int J Epidemiol. 1989; 18: S46-S55
        View in Article
        • Havranek E.P.
        • Mujahid M.S.
        • Barr D.A.
        • Blair I.V.
        • Cohen M.S.
        • Cruz-Flores S.
        • et al.
        Social determinants of risk and outcomes for cardiovascular disease: a scientific statement from the American heart association.
        Circulation. 2015; 132: 873-898
        View in Article
        • WHO
        2008-2013 action plan for the global strategy for the prevention and control of noncommunicable diseases.
        2009
        View in Article
        • Petrovic D.
        • de Mestral C.
        • Bochud M.
        • Bartley M.
        • Kivimaki M.
        • Vineis P.
        • et al.
        The contribution of health behaviors to socioeconomic inequalities in health: a systematic review.
        Prev Med. 2018; 113: 15-31
        View in Article
        • Piccinelli C.
        • Carna P.
        • Stringhini S.
        • Sebastiani G.
        • Demaria M.
        • Marra M.
        • et al.
        The contribution of behavioural and metabolic risk factors to socioeconomic inequalities in mortality: the Italian Longitudinal Study.
        Int J Publ Health. 2018; 63: 325-335
        View in Article
        • Petrelli A.
        • Zengarini N.
        • Demuru E.
        • Giorgi Rossi P.
        • Sebastiani G.
        • Gaudio R.
        • et al.
        Differences in mortality by educational level in Italy (2012-2014).
        Epidemiol Prev. 2018; 42: 288-300
        View in Article
        • Petrelli A.
        • Di Napoli A.
        • Sebastiani G.
        • Rossi A.
        • Giorgi Rossi P.
        • Demuru E.
        • et al.
        Italian Atlas of mortality inequalities by education level.
        Epidemiol Prev. 2019; 43: 1-120
        View in Article
        • Petrelli A.
        • Gnavi R.
        • Marinacci C.
        • Costa G.
        Socioeconomic inequalities in coronary heart disease in Italy: a multilevel population-based study.
        Soc Sci Med. 2006; 63: 446-456
        View in Article
        • Ricceri F.
        • Sacerdote C.
        • Giraudo M.T.
        • Fasanelli F.
        • Lenzo G.
        • Galli M.
        • et al.
        The association between educational level and cardiovascular and cerebrovascular diseases within the EPICOR study: new evidence for an old inequality problem.
        PLoS One. 2016; 11e0164130
        View in Article
        • Sebastiani G.
        • Di Filippo P.
        • Demaria M.
        • Caranci N.
        • Di Minco L.
        • Tamburini C.
        • et al.
        Lo studio longitudinale italiano: integrazione delle indagini sulla salute con dati di mortalità e ospedalizzazione.
        Metodologia e potenzialità di utilizzo, Istat2019
        View in Article
        • Istat
        Condizioni di salute e ricorso ai servizi sanitari. Nota metodologica.
        2008 (Available from:)
        https://www.istat.it/it/archivio/10836
        Date accessed: October 19, 2021
        View in Article
        • Cardano M.
        • Marinacci C.
        La rilevazione della posizione sociale.
        Epidemiol Prev. 2004; 28: 124-142
        View in Article
        • VanderWeele T.J.
        • Vansteelandt S.
        Mediation analysis with multiple mediators.
        Epidemiol Methods. 2013; 2: 95-115
        View in Article
        • Fasanelli F.
        • Giraudo M.T.
        • Ricceri F.
        • Valeri L.
        • Zugna D.
        Marginal time-dependent causal effects in mediation analysis with survival data.
        Am J Epidemiol. 2019; 188: 967-974
        View in Article
        • Vanderweele T.J.
        • Vansteelandt S.
        Odds ratios for mediation analysis for a dichotomous outcome.
        Am J Epidemiol. 2010; 172: 1339-1348
        View in Article
        • Valeri L.
        • VanderWeele T.J.
        SAS macro for causal mediation analysis with survival data.
        Epidemiology. 2015; 26: e23-e24
        View in Article
        • Vanderweele T.J.
        Explanation in causal inference. Methods for mediation and interaction.
        Oxford University Press, 2015
        View in Article
        • Laaksonen M.
        • Talala K.
        • Martelin T.
        • Rahkonen O.
        • Roos E.
        • Helakorpi S.
        • et al.
        Health behaviours as explanations for educational level differences in cardiovascular and all-cause mortality: a follow-up of 60 000 men and women over 23 years.
        Eur J Publ Health. 2008; 18: 38-43
        View in Article
        • Veronesi G.
        • Ferrario M.M.
        • Kuulasmaa K.
        • Bobak M.
        • Chambless L.E.
        • Salomaa V.
        • et al.
        Educational class inequalities in the incidence of coronary heart disease in Europe.
        Heart. 2016; 102: 958-965
        View in Article
        • Beauchamp A.
        • Peeters A.
        • Wolfe R.
        • Turrell G.
        • Harriss L.R.
        • Giles G.G.
        • et al.
        Inequalities in cardiovascular disease mortality: the role of behavioural, physiological and social risk factors.
        J Epidemiol Community Health. 2010; 64: 542-548
        View in Article
        • Kershaw K.N.
        • Droomers M.
        • Robinson W.R.
        • Carnethon M.R.
        • Daviglus M.L.
        • Monique Verschuren W.M.
        Quantifying the contributions of behavioral and biological risk factors to socioeconomic disparities in coronary heart disease incidence: the MORGEN study.
        Eur J Epidemiol. 2013; 28: 807-814
        View in Article
        • Khang Y.H.
        • Lynch J.W.
        • Yang S.
        • Harper S.
        • Yun S.C.
        • Jung-Choi K.
        • et al.
        The contribution of material, psychosocial, and behavioral factors in explaining educational and occupational mortality inequalities in a nationally representative sample of South Koreans: relative and absolute perspectives.
        Soc Sci Med. 2009; 68: 858-866
        View in Article
        • Kivimaki M.
        • Lawlor D.A.
        • Davey Smith G.
        • Kouvonen A.
        • Virtanen M.
        • Elovainio M.
        • et al.
        Socioeconomic position, co-occurrence of behavior-related risk factors, and coronary heart disease: the Finnish Public Sector study.
        Am J Publ Health. 2007; 97: 874-879
        View in Article
        • Nordahl H.
        • Rod N.H.
        • Frederiksen B.L.
        • Andersen I.
        • Lange T.
        • Diderichsen F.
        • et al.
        Education and risk of coronary heart disease: assessment of mediation by behavioral risk factors using the additive hazards model.
        Eur J Epidemiol. 2013; 28: 149-157
        View in Article
        • Panagiotakos D.
        • Georgousopoulou E.
        • Notara V.
        • Pitaraki E.
        • Kokkou E.
        • Chrysohoou C.
        • et al.
        Education status determines 10-year (2002-2012) survival from cardiovascular disease in Athens metropolitan area: the ATTICA study, Greece.
        Health Soc Care Community. 2016; 24: 334-344
        View in Article
        • Silhol R.
        • Zins M.
        • Chauvin P.
        • Chaix B.
        Investigating the spatial variability in incidence of coronary heart disease in the Gazel cohort: the impact of area socioeconomic position and mediating role of risk factors.
        J Epidemiol Community Health. 2011; 65: 137-143
        View in Article
        • Mejean C.
        • Droomers M.
        • van der Schouw Y.T.
        • Sluijs I.
        • Czernichow S.
        • Grobbee D.E.
        • et al.
        The contribution of diet and lifestyle to socioeconomic inequalities in cardiovascular morbidity and mortality.
        Int J Cardiol. 2013; 168: 5190-5195
        View in Article
        • Floud S.
        • Balkwill A.
        • Moser K.
        • Reeves G.K.
        • Green J.
        • Beral V.
        • et al.
        The role of health-related behavioural factors in accounting for inequalities in coronary heart disease risk by education and area deprivation: prospective study of 1.2 million UK women.
        BMC Med. 2016; 14: 145
        View in Article
        • Kuper H.
        • Adami H.O.
        • Theorell T.
        • Weiderpass E.
        Psychosocial determinants of coronary heart disease in middle-aged women: a prospective study in Sweden.
        Am J Epidemiol. 2006; 164: 349-357
        View in Article
        • Lawlor D.A.
        • Ebrahim S.
        • Davey Smith G.
        Adverse socioeconomic position across the lifecourse increases coronary heart disease risk cumulatively: findings from the British women's heart and health study.
        J Epidemiol Community Health. 2005; 59: 785-793
        View in Article
        • Ernstsen L.
        • Bjerkeset O.
        • Krokstad S.
        Educational inequalities in ischaemic heart disease mortality in 44,000 Norwegian women and men: the influence of psychosocial and behavioural factors. The HUNT Study.
        Scand J Publ Health. 2010; 38: 678-685
        View in Article
        • Hardarson T.
        • Gardarsdóttir M.
        • Gudmundsson K.T.
        • Thorgeirsson G.
        • Sigvaldason H.
        • Sigfússon N.
        The relationship between educational level and mortality. The Reykjavík Study.
        J Intern Med. 2001; 249: 495-502
        View in Article
        • D'Avanzo B.
        • La Vecchia C.
        • Braga C.
        • Franceschi S.
        • Negri E.
        • Parpinel M.
        Nutrient intake according to education, smoking, and alcohol in Italian women.
        Nutr Cancer. 1997; 28: 46-51
        View in Article
        • Marmot M.G.
        • Shipley M.J.
        • Hemingway H.
        • Head J.
        • Brunner E.J.
        Biological and behavioural explanations of social inequalities in coronary heart disease: the Whitehall II study.
        Diabetologia. 2008; 51: 1980-1988
        View in Article
        • European Commission
        Women and unpaid work: recognise, reduce, redistribute!.
        (Available from:)
        https://ec.europa.eu/social/main.jsp?langId=en&catId=1196&newsId=2492&furtherNews=yes
        Date: 2016
        Date accessed: October 19, 2021
        View in Article
        • Mencarini L.
        Soddisfazione e uso del tempo nelle coppie italiane.
        in: Romano M.C. Mencarini L. Tanturri M.L. Uso del tempo e ruoli di genere Tra lavoro e famiglia nel ciclo di vita. Istat, Rome2012: 41-64
        View in Article
        • D'Ovidio F.
        • d'Errico A.
        • Scarinzi C.
        • Costa G.
        Increased incidence of coronary heart disease associated with "double burden" in a cohort of Italian women.
        Soc Sci Med. 2015; 135: 40-46
        View in Article
        • Haynes S.G.
        • Feinleib M.
        Women, work and coronary heart disease: prospective findings from the Framingham heart study.
        Am J Publ Health. 1980; 70: 133-141
        View in Article
        • Lee S.
        • Colditz G.
        • Berkman L.
        • Kawachi I.
        Caregiving to children and grandchildren and risk of coronary heart disease in women.
        Am J Publ Health. 2003; 93: 1939-1944
        View in Article
        • Hooftman W.E.
        • van der Beek A.J.
        • Bongers P.M.
        • van Mechelen W.
        Gender differences in self-reported physical and psychosocial exposures in jobs with both female and male workers.
        J Occup Environ Med. 2005; 47: 244-252
        View in Article
        • Messing K.
        • Stock S.R.
        • Tissot F.
        Should studies of risk factors for musculoskeletal disorders be stratified by gender? Lessons from the 1998 Québec Health and Social Survey.
        Scand J Work Environ Health. 2009; 35: 96-112
        View in Article
        • d'Errico A.
        • Cardano M.
        • Landriscina T.
        • Marinacci C.
        • Pasian S.
        • Petrelli A.
        • et al.
        Workplace stress and prescription of antidepressant medications: a prospective study on a sample of Italian workers.
        Int Arch Occup Environ Health. 2011; 84: 413-424
        View in Article
        • Josephson M.
        • Pernold G.
        • Ahlberg-Hultén G.
        • Härenstam A.
        • Theorell T.
        • Vingård E.
        • et al.
        Differences in the association between psychosocial work conditions and physical work load in female- and male-dominated occupations. MUSIC-Norrtälje Study Group.
        Am Ind Hyg Assoc J. 1999; 60: 673-678
        View in Article
        • Sterud T.
        Work-related gender differences in physician-certified sick leave: a prospective study of the general working population in Norway.
        Scand J Work Environ Health. 2014; 40: 361-369
        View in Article
        • Ibrahim S.A.
        • Scott F.E.
        • Cole D.C.
        • Shannon H.S.
        • Eyles J.
        Job strain and self-reported health among working women and men: an analysis of the 1994/5 Canadian National Population Health Survey.
        Women Health. 2001; 33: 105-124
        View in Article
        • Karlqvist L.
        • Tornqvist E.W.
        • Hagberg M.
        • Hagman M.
        • Toomingas A.
        Self-reported working conditions of VDU operators and associations with musculoskeletal symptoms: a cross-sectional study focussing on gender differences.
        Int J Ind Ergon. 2002; 30: 277-294
        View in Article
        • Richiardi L.
        • Bellocco R.
        • Zugna D.
        Mediation analysis in epidemiology: methods, interpretation and bias.
        Int J Epidemiol. 2013; 42: 1511-1519
        View in Article
        • Patrick D.L.
        • Cheadle A.
        • Thompson D.C.
        • Diehr P.
        • Koepsell T.
        • Kinne S.
        The validity of self-reported smoking: a review and meta-analysis.
        Am J Publ Health. 1994; 84: 1086-1093
        View in Article
        • Vartiainen E.
        • Seppälä T.
        • Lillsunde P.
        • Puska P.
        Validation of self reported smoking by serum cotinine measurement in a community-based study.
        J Epidemiol Community Health. 2002; 56: 167-170
        View in Article
        • Stringhini S.
        • Dugravot A.
        • Shipley M.
        • Goldberg M.
        • Zins M.
        • Kivimäki M.
        • et al.
        Health behaviours, socioeconomic status, and mortality: further analyses of the British Whitehall II and the French GAZEL prospective cohorts.
        PLoS Med. 2011; 8
        View in Article
        • Tolonen H.
        • Koponen P.
        • Mindell J.S.
        • Männistö S.
        • Giampaoli S.
        • Dias C.M.
        • et al.
        Under-estimation of obesity, hypertension and high cholesterol by self-reported data: comparison of self-reported information and objective measures from health examination surveys.
        Eur J Publ Health. 2014; : 941-948
        View in Article
        • van Oort F.V.
        • van Lenthe F.J.
        • Mackenbach J.P.
        Material, psychosocial, and behavioural factors in the explanation of educational inequalities in mortality in The Netherlands.
        J Epidemiol Community Health. 2005; 59: 214-220
        View in Article
        • D'Errico A.
        • Ricceri F.
        • Stringhini S.
        • Carmeli C.
        • Kivimaki M.
        • Bartley M.
        • et al.
        Consortium Lifepath. Socioeconomic indicators in epidemiologic research: a practical example from the LIFEPATH study.
        PLoS One. 2017 May 30; 12 (2017)e0178071
        View in Article
        • Beauchamp A.
        • Peeters A.
        • Tonkin A.
        • Turrell G.
        Best practice for prevention and treatment of cardiovascular disease through an equity lens: a review.
        Eur J Cardiovasc Prev Rehabil. 2010; 17: 599-606
        View in Article
        • Callander E.J.
        • McDermott R.
        Measuring the effects of CVD interventions and studies across socioeconomic groups: a brief review.
        Int J Cardiol. 2017; 227: 635-643
        View in Article

      Article info

      Publication history

      Published online: November 24, 2021
      Accepted: October 26, 2021
      Received in revised form: October 20, 2021
      Received: March 16, 2021
      Handling Editor: F. Galletti

      Identification

      DOI: https://doi.org/10.1016/j.numecd.2021.10.022

      Copyright

      © 2021 The Authors. Published by Elsevier B.V. on behalf of The Italian Diabetes Society, the Italian Society for the Study of Atherosclerosis, the Italian Society of Human Nutrition and the Department of Clinical Medicine and Surgery, Federico II University.

      User license

      Creative Commons Attribution – NonCommercial – NoDerivs (CC BY-NC-ND 4.0) |
      How you can reuse Information Icon
      Creative Commons Attribution – NonCommercial – NoDerivs (CC BY-NC-ND 4.0)

      Permitted

      For non-commercial purposes:

      • Read, print & download
      • Redistribute or republish the final article
      • Text & data mine
      • Translate the article (private use only, not for distribution)
      • Reuse portions or extracts from the article in other works

      Not Permitted

      • Sell or re-use for commercial purposes
      • Distribute translations or adaptations of the article

      Elsevier's open access license policy

      ScienceDirect

      Access this article on ScienceDirect

      The content on this site is intended for healthcare professionals.


      We use cookies to help provide and enhance our service and tailor content. To update your cookie settings, please visit the Cookie Preference Center for this site.
      Copyright © 2023 Elsevier Inc. except certain content provided by third parties.


      RELX