SSM - Population Health 23 (2023) 101461 Contents lists available at ScienceDirect SSM - Population Health journal homepage: www.elsevier.com/locate/ssmph Age and cause-of-death contributions to educational inequalities in life expectancy and lifespan variation in a low-mortality country: A cross-sectional study of 1.67 million deaths in Spain (2016–19) Sergi Trias-Llimós a, *, Jeroen J.A. Spijker a, Amand Blanes a, Iñaki Permanyer a, b a Centre d’Estudis Demogràfics, Centres de Recerca de Catalunya (CERCA), Carrer de Ca n’Altayó, Edifici E2, Universitat Autònoma de Barcelona, 08193 Bellaterra, Spain b ICREA, Passeig Lluís Companys 23, 08010, Barcelona, Spain A R T I C L E I N F O A B S T R A C T Keywords: Background: We aim to assess the age- and cause-specific contributions to differences in life expectancy and Causes of death lifespan variation between the high- and low-educated groups in Spain. Low mortality Methods: We use sex-, age-, education- and cause-specific mortality and population data for individuals aged 30 Inequalities and over for 2016-19 in Spain. We estimated life expectancies, and standard deviations of the age-at-death Gender Health inequality distribution (lifespan variation), and we disentangled the contribution of age-causes of death to educational Education differences in both indicators. Findings: Life expectancy at age 30 was higher for high-educated groups compared to low-educated groups, 5.5 years for males and 3.0 years for females. Lifespan variation was higher for low-educated groups compared to high-educated groups, 2.9 years for males and 2.2 years for females. The main contributors to the life expectancy gaps in males were lung cancer (0.58 years) and ischaemic heart diseases (0.42 years), and in females were other cardiovascular causes (0.26 years), and ischaemic heart diseases (0.22 years). The main contributors to the lifespan variation gaps were in males lung cancer (− 0.25 years) and ischaemic heart diseases (− 0.22 years), while in females were other neoplasms and other diseases of the nervous system. Interpretation: Whereas behavioural causes are more important in explaining educational inequalities in mortality among men, ageing-related causes of death seem more important among women. Attempts at narrowing so­ cioeconomic gaps in mortality may benefit from applying gender-specific preventive policy measures. 1. Background outcomes. Such factors thus act across the life course, affecting health and risk of death at any age. This includes decisions regarding education Socioeconomic position is an important predictor of health and and employment during early adulthood which not only shape an in­ mortality (Galobardes et al., 2006). In all populations, low socioeco­ dividual’s life trajectory, but may also be associated with different dis­ nomic groups tend to have worse health and mortality outcomes ease risks as health-damaging exposures or health-enhancing compared to high socioeconomic groups, and these inequalities persist opportunities (including living and working conditions as well as indi­ in all countries, including in countries with low mortality levels and vidual lifestyles) are socially patterned (Commission on Social De­ overall relatively low socioeconomic inequalities in mortality, as is the terminants of Health, 2008; Davey Smith et al., 1998). While most of case of Spain (Kulhánová et al., 2014; Machón et al., 2020; Mackenbach these socioeconomic factors are partly correlated, research focusing on et al., 2018). mortality inequalities has mostly adopted education as the main socio­ Socioeconomic inequalities in health and mortality can be under­ economic variable. Moreover, education is known to be positively stood within the social determinants of health (SDH) framework. These associated with the ability to access health-related knowledge, adapt determinants refer to the conditions in which people are born, grow, behavior and obtain effective treatment (Galobardes et al., 2006). live, work, and age, and how these conditions influence their health Previous studies on cause-specific inequalities in mortality in Spain * Corresponding author. Center for Demographic Studies, Carrer de Ca n’Altayó, Edifici E2, 08193 Bellaterra, Spain. E-mail address: strias@ced.uab.cat (S. Trias-Llimós). https://doi.org/10.1016/j.ssmph.2023.101461 Received 6 March 2023; Received in revised form 22 June 2023; Accepted 23 June 2023 Available online 7 July 2023 2352-8273/© 2023 Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). S. Trias-Llimós et al. SSM - Population Health 23 (2023) 101461 fall into three main categories. First, cross-country studies have used 2011 3) and high (postsecondary vocational and university education, data from the regions of Madrid, Barcelona and/or the Basque Country ISCED-2011 4+) (see https://www.ine.es/metodologia/t20/t and information on the main causes of death (Bartoll-Roca et al., 2022; 2030306_niveduc.pdf). The distribution of the population by age, sex Huisman, 2004; Mackenbach et al., 2018, 2019). Second, other studies and educational group is presented in Fig. S1, and clearly shows the have used 2001 census-linked mortality data to examine socioeconomic rapid educational expansion of the Spanish population. Causes of death inequalities in all-cause and cause-specific mortality (Regidor et al., were grouped using a classification of 45 causes of death, which were 2016; Reques et al., 2014). The most comprehensive study examined then grouped into 10 categories: Neoplasms (ICD-10 codes: C00-D48), absolute and relative educational inequalities in mortality for about 20 endocrine (E00-E90), mental (F00–F99, G31.0, G31.1, G31.8 and causes of death from 2001 to 08. In the 2000s, educational inequalities G31.9), nervous (G00-G98 excluding G31.0, G31.1, G31.8 and G31.9), in mortality in Spain for females were mainly driven by cardiovascular circulatory (I00–I99 excluding I46, I49.0), respiratory (J00-J96), causes, accounting for nearly 50% of the gap. For males the sum of digestive (K00–K93), ill-defined (R00-R94), external causes (chapters S, cardiovascular and respiratory diseases and cancers represented slightly T, V, W, X and Y) and other causes (ICD-10 codes not previously over 50% of the total educational inequality in mortality (Reques et al., mentioned (Table S1). 2014). Finally, other studies have made used of the recent availability of mortality data by educational attainment to study socioeconomic in­ 2.2. Methods equalities in mortality (Blanes & Trias-Llimós, 2021; Permanyer et al., 2018; Requena, 2017). We estimated 5-year age group all-cause mortality rates for sex and Recent research on mortality dynamics has stressed the importance educational attainment using conventional approaches. We use two of going beyond average-based indicators, like life expectancy, to mortality indicators to assess educational differences in mortality: life incorporate metrics that measure the variability in the ages at which expectancy at age 30, which is estimated using life table methods individuals die, i.e., lifespan variation (Aburto et al., 2020; van Raalte (Preston et al., 2001); and lifespan variation at age 30, which is esti­ et al., 2018). For example, a population in which everyone dies at age 80 mated using the standard deviation. Using life table notation, it is is more equal to a population in which half of the people die at age 60 defined as: and the other half die at age 100, even if both have a life expectancy of √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ 80 years. It is well-known that, as longevity increases, the variability in √ 95 √∑ the ages at which individuals die tends to decline (Aburto et al., 2020; sd30 = √ dx (x + ax − e30 − 30)2 (1) Smits & Monden, 2009). However, the causes of death that have led to x=30 lower variability in lifespans are not necessarily the same causes that have led to higher life expectancy (Aburto et al., 2018; Aburto & van where x represents age, dx the proportion of deaths at age x, and ax the Raalte, 2018; Seligman et al., 2016). Such information is fundamental to average person-years lived in the age interval by those dying in the better understand whether, and to what extent, it is possible to attain the interval. normatively desirable goals of reducing differences in life expectancy We estimated educational gaps in the two selected mortality mea­ and lifespan variation across education groups simultaneously. sures in two different ways: i) Comparing the outcomes of those with At the national level, individual level mortality data by education high versus low education; which implies that positive life expectancy was only available for Spain from 2012 onwards (Requena, 2017). gaps represent higher life expectancy among the highest educated Recent estimates for 2017-19 in Spain suggest that high-educated groups group, while negative lifespan variation gaps imply lower lifespan have an advantage in life expectancy of over 3 years for females and over variation among higher educated groups; ii) Comparing the outcomes of 5 years for males compared to their low-educated counterparts (Blanes each educational group against the overall population for each sex. Age- & Trias-Llimós, 2021). In terms of lifespan variation, lower-educated and cause-specific contributions to these differences were estimated groups have approximately 25% higher lifespan variation than using the continuous decomposition method derived by Horiuchi high-educated groups for both males and females (Blanes & Tri­ (Horiuchi et al., 2008) implemented in the DemoDecomp package in R as-Llimós, 2021). Trends over the last few decades are scarce and sug­ (Riffe, 2018). gest that educational inequalities in both life expectancy and lifespan variation persisted or even increased (Permanyer et al., 2018). In this 3. Results context, examining the role of a granular set of causes of death on educational inequalities in both life expectancy and lifespan variation Age-specific mortality rates by educational attainment showed a will allow us to better understand mortality inequalities. clear educational gradient in mortality: high-educated groups had the We aim to assess the cause-of-death contribution to educational in­ lowest mortality and low-educated groups had the highest mortality, equalities in life expectancy and lifespan variation in Spain for the most particularly among working age groups (Fig. 1). The differences be­ recent pre-covid period, 2016–2019 and for a group of 45 causes of tween high- and medium-educated groups persisted up to around ages death. 65–69 in men and up to around ages 60–64 in women. The mortality disadvantage observed among the low-educated group persisted up until 2. Methods about the age of 80. For older age groups, educational differences in mortality rates were very small. 2.1. Data 3.1. Life expectancy and lifespan variation by education groups We use sex-, age-, education- and cause-specific mortality data for individuals aged 30 and over for 2016-19 in Spain from the Spanish In 2016-19, life expectancy at age 30 in Spain was 50.9 years for National Statistics Institute (INE). Educational attainment was retrieved males and 56.2 years for females (Table 1). For both males and females, from multiple data linkages, including municipal population registers the high-educated groups had higher life expectancy compared to their (“Padrón”) and the 2011 census, all carried out by INE. Population es­ low-educated counterparts. For males, life expectancy at age 30 was timates from 2018 by age, sex, and educational attainment, also 53.2 years for the high-educated and 47.7 years for the low-educated retrieved from INE, were used as denominators to produce our mortality (5.5 years difference), whereas the corresponding numbers for females indicators. Based on both data sources, three educational attainment were 57.6 years and 54.6 years, respectively (3.0 years difference). categories could be distinguished: low (primary education or less, Lifespan variation at age 30 was 12.3 years for males and 10.9 years for ISCED-2011 0–2) middle (lower and upper secondary education, ISCED- females. Differences in this indicator across educational groups were 2 S. Trias-Llimós et al. SSM - Population Health 23 (2023) 101461 Fig. 1. Age-specific all-cause mortality rates (95% CI) by sex and educational group, Spain 2016-19. contribution of age groups to educational gaps in lifespan variation was Table 1 largest at ages 50–54 and particularly important across working age Life expectancy and lifespan variation at age 30 in Spain by sex and educational groups. attainment, 2016-19. Life expectancy diff Lifespan variation diff 3.3. Cause-of-death contributions to educational gaps in life expectancy Men All 50.95 12.34 ¡2.93 and lifespan variation High 53.24 − 2.29 11.25 1.10 Medium 51.11 − 0.16 12.49 − 0.14 For both males and females, the causes that contributed the most to Low 47.72 3.24 14.18 − 1.83 educational gaps in life expectancy were cardiovascular causes (1.5 High-Low gap 5.52 − 2.93 years for males, 0.9 years for females) and cancers (1.5 years for males, Women 0.3 years for females). These two major groups of causes accounted for All 56.24 10.95 nearly 50% of the gap for males (48.3%) and females (40.7%). For High 57.62 − 1.38 10.55 0.39 Medium 56.53 − 0.29 10.75 0.20 males, respiratory diseases were the third cause of death that contrib­ Low 54.60 1.64 12.73 − 1.78 uted the most (0.75 years), and they were particularly important at old High-Low gap 3.02 − 2.17 ages. External causes were also important (0.49 years), particularly at young ages. For females, cause-specific patterns were less clear and a larger group of causes contributed in a similar magnitude, for instance, also in favour of the high-educated group, as they were found to have respiratory diseases (0.32 years), nervous system diseases (0.25 years), lower lifespan variation compared to their lower-educated counterparts. digestive diseases (0.25 years) and endocrine diseases (0.21 years). Yet, For males, lifespan variation was 11.2 years for high-educated and 14.2 the contribution of these groups of causes differed across age groups. For years for low-educated, while for females it was 10.6 years and 12.7 example, the contribution of diseases of the nervous system to educa­ years, respectively. tional differences in life expectancy was particularly relevant at working ages, while the contribution of endocrine diseases was mainly attribut­ 3.2. Age contributions to educational gaps in life expectancy and lifespan able to mortality from older age groups. variation The causes that contributed the most to the lifespan variation be­ tween high- and low-educated groups were cancers and cardiovascular The contribution of age groups to the educational gaps in life ex­ causes in both men (− 0.75 years and − 0.53 years, respectively) and pectancy peaked for males at ages 60–64 with almost 0.8 years, and at women (− 0.52 years and − 0.33 years, respectively). External causes ages 50–54 with around 0.35 years of contribution for females (Fig. 2). A accounted for a relatively large part of the gap in men (− 0.47 years), but visual inspection of the graphs suggests higher relative concentration not in women (− 0.16 years), and were particularly important at young across ages for males compared to females. For example, over half (53%) ages. The contributions of digestive, nervous and respiratory diseases of the contribution to the educational life expectancy gap for males was were at similar levels between men and women, ranging from − 0.16 found in the age group 50–69; while for females, a broader age range years in digestive diseases in women to − 0.26 years in nervous diseases was needed to account for at least 50% of the educational life expec­ in women. In contrast with the age contributions to life expectancy gaps, tancy gap, as the age groups 45–74 represented 54.5% of this gap. The the age distribution accounting the most to the educational gaps in 3 S. Trias-Llimós et al. SSM - Population Health 23 (2023) 101461 Fig. 2. Age- and cause-specific decomposition of life expectancy and lifespan variation gaps between high and low educational groups by sex, Spain 2016–19 (list of 10 major causes). lifespan variation was shifted towards the left, and therefore to younger cases, these contributions were particularly important at ages 50–84. age groups. 3.5. The role of detailed causes of death on educational inequalities in 3.4. The role of detailed causes of death on educational inequalities in life lifespan variation expectancy The cause-specific contribution to educational differences in lifespan The contribution of specific causes of death to the educational gaps variation between high and low educated groups was led by lung cancer in mortality presented notable differences across sex. For males, lung (− 0.25 years) and ischaemic diseases (− 0.22 years) in males, and for cancer was the largest contributor to educational differences in life ex­ other neoplasms (− 0.22 years) and other nervous system diseases pectancy with 0.58 years (10.6%), followed by ischaemic heart diseases (− 0.21 years) in females (Table 3). Other cause-specific contributions (0.42 years, 7.6%) and by other chronic lower respiratory diseases (0.33 that contributed more than − 0.15 years to the lifespan variation were years, 6.0%) (Table 2). For females, lung cancer was found to play the other neoplasms, other nervous system diseases, other heart diseases in opposite role, and therefore contributed toward reducing the life ex­ males and other diseases in females. Regarding the age-specific contri­ pectancy gap between the highest and lowest educational groups (− 0.10 butions within these cause-specific contributions an interesting pattern years, − 3.2%), whereas the main drivers of the gap were all cardio­ emerges, namely that contrary to the absolute educational differences in vascular causes: other cardiovascular causes (0.26 years, 8.7%), life expectancy, it is not the older age groups that contribute the most ischaemic diseases (0.22 years, 7.2%), and cerebrovascular diseases but the 30–49 and 50–64 year age groups (− 2.17 years out of − 2.93 (0.20 years, 6.7%); as well as all cancers (0.22 years, 7.3%). In most years in males and − 2.25 years out of − 2.17 years in females). For 4 S. Trias-Llimós et al. SSM - Population Health 23 (2023) 101461 Table 2 Age- and cause-specific contributions to the high-low educational gaps in life expectancy at age 30, Spain, 2016-19. List of 45 causes of death. d Males Females 30–49 50–64 65–84 85+ 30+ 30–49 50–64 65–84 85+ 30+ AIDS 0.03 0.03 0.00 0.00 0.06 0.02 0.01 0.00 0.00 0.02 Other infectious diseases 0.03 0.05 0.03 0.00 0.11 0.02 0.02 0.03 0.01 0.09 Benign neoplasms 0.01 0.01 0.01 0.00 0.03 0.01 0.01 0.01 0.00 0.02 Breast cancer 0.00 0.00 0.00 0.00 0.00 0.01 0.02 − 0.04 − 0.01 ¡0.02 Intestine cancer 0.02 0.06 0.08 0.00 0.16 0.01 0.03 0.01 0.00 0.05 Liver cancer 0.03 0.07 0.03 0.00 0.13 0.01 0.02 0.03 0.00 0.06 Lung cancer 0.05 0.31 0.23 0.00 0.58 0.02 0.00 − 0.11 − 0.01 ¡0.10 Other neoplasms 0.06 0.15 0.10 − 0.02 0.29 0.06 0.12 0.06 − 0.02 0.22 Prostate cancer 0.00 0.01 0.03 0.00 0.04 0.00 0.00 0.00 0.00 0.00 Stomach cancer 0.01 0.03 0.05 0.00 0.10 0.01 0.02 0.03 0.01 0.07 Upper Aero cancers 0.02 0.09 0.05 0.00 0.16 0.00 0.01 0.00 0.00 0.02 Blood diseases 0.00 0.01 0.01 0.00 0.02 0.01 0.01 0.00 0.00 0.02 Diabetes 0.01 0.04 0.06 0.01 0.13 0.01 0.03 0.08 0.05 0.16 Other endocrine diseases 0.01 0.02 0.01 0.00 0.05 0.01 0.02 0.02 0.01 0.06 Alcohol psychosis 0.00 0.01 0.01 0.00 0.02 0.00 0.00 0.00 0.00 0.00 Dementia 0.00 0.01 0.05 0.02 0.08 0.00 0.01 0.06 0.04 0.12 Drug dependence 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 Other mental diseases 0.01 0.01 0.01 0.00 0.03 0.01 0.01 0.01 0.00 0.02 Alzheimer 0.00 0.01 0.03 0.00 0.04 0.00 0.01 0.04 0.02 0.07 Epilepsy 0.03 0.02 0.00 0.00 0.05 0.02 0.01 0.01 0.00 0.04 Other diseases nervous system 0.10 0.06 0.03 − 0.01 0.17 0.07 0.05 0.02 − 0.01 0.14 Cerebrovascular diseases 0.04 0.09 0.11 0.02 0.25 0.02 0.05 0.08 0.05 0.20 Hypertension 0.01 0.02 0.03 0.01 0.06 0.01 0.01 0.05 0.04 0.10 Ischaemic heart diseases 0.06 0.21 0.15 − 0.01 0.42 0.03 0.06 0.10 0.02 0.22 Other circulatory diseases 0.03 0.05 0.06 0.00 0.14 0.02 0.04 0.06 0.01 0.13 Other heart diseases 0.08 0.11 0.11 0.01 0.30 0.03 0.05 0.12 0.07 0.26 Acute lower respiratory diseases 0.03 0.06 0.05 0.01 0.15 0.03 0.03 0.03 0.02 0.10 Asthma 0.00 0.00 0.00 0.00 0.01 0.00 0.01 0.01 0.01 0.03 Chronic lower respiratory diseases 0.01 0.08 0.19 0.05 0.33 0.01 0.02 − 0.02 0.00 0.01 Other diseases of the respiratory system 0.08 0.09 0.08 0.02 0.27 0.04 0.04 0.06 0.02 0.17 Chronic liver disease 0.04 0.12 0.05 0.00 0.21 0.01 0.03 0.02 0.00 0.06 Other digestive diseases 0.05 0.09 0.09 0.02 0.25 0.03 0.05 0.07 0.03 0.18 Skin diseases 0.00 0.00 0.01 0.00 0.01 0.00 0.00 0.01 0.00 0.02 Musculoskeletal diseases 0.00 0.01 0.01 0.00 0.02 0.01 0.01 0.02 0.00 0.03 Genitourinary diseases 0.01 0.03 0.04 0.01 0.09 0.01 0.03 0.06 0.03 0.12 Other diseases 0.04 0.04 0.00 0.00 0.09 0.05 0.05 0.01 0.00 0.10 Mechanisms of the death 0.02 0.02 0.01 0.00 0.05 0.01 0.01 0.00 0.00 0.02 Other ill-defined causes 0.06 0.06 0.02 0.00 0.15 0.03 0.02 0.01 0.00 0.06 Accidental drowning and exposure to substances 0.07 0.02 0.01 0.00 0.10 0.01 0.01 0.00 0.00 0.02 Accidental falls 0.02 0.02 0.01 0.00 0.04 0.01 0.00 0.01 0.00 0.02 Other accidents 0.02 0.02 0.01 0.00 0.05 0.01 0.01 0.01 0.00 0.03 Assault 0.02 0.00 0.00 0.00 0.02 0.01 0.00 0.00 0.00 0.01 Other external causes 0.03 0.02 0.01 0.00 0.06 0.01 0.01 0.01 0.00 0.03 Suicide 0.06 0.04 0.03 0.00 0.13 0.01 0.00 0.00 0.00 0.01 Transport accidents 0.05 0.02 0.01 0.00 0.08 0.01 0.01 0.00 0.00 0.01 All causes 1.30 2.20 1.88 0.14 5.52 0.69 0.97 0.96 0.39 3.02 instance, in men aged 30–49 other nervous disease system diseases ischaemic and other heart diseases, lung cancer and other respiratory contributed − 0.14 years to the lifespan variation, respiratory system (males). diseases − 0.11 years and other heart diseases, accidents and suicide This study used individual-level mortality data for Spain of 1.67 − 0.10 years (each). At ages 50–64 the contributions were highest for million deaths in the pre-Covid-19 period (2016–2019). Pulling the data lung cancer (− 0.19 years) and ischaemic heart disease (− 0.13 years). over 4 years allowed us to focus beyond main groups of causes such as Females presented a similar pattern although values were generally ICD chapters (e.g. cardiovascular diseases, diseases of the respiratory slightly lower: at ages 30–49 other diseases of the nervous system system) as we could analyse a much wider range of specific causes of (− 0.14 years) and other forms of cancer (− 0.10 years) were the most death (e.g. dementia, intestine cancer) as well as educational differ­ important contributors; while at the 50–64 age group were mostly ences. However, some limitations deserve to be discussed. First, we responsible for the lifespan variation due to other forms of cancer relied on educational attainment data estimated from INE. Most of the (− 0.13 years). data were obtained by linking it with either the 2001 Census or “Padrón” (>80%), but also from the official degrees issued by the Ministry of 4. Discussion education or data on enrolment at Spanish universities or non-university education (see https://www.ine.es/metodologia/t20/t2030306_ni This study showed that higher educated groups had a higher life veduc.pdf for more details). When studying health inequalities two expectancy at age 30 as compared to lower educated groups in both critical choices need to be made: the distribution and grouping of in­ males (5.5 years) and females (3.0 years), and lower lifespan variation. dividuals across educational groups; and the pair-wise comparison of These education inequalities were driven by a wide variety of factors groups. Education has been widely used as a socioeconomic indicator in and causes of death, which were led by cardiovascular diseases, diseases several other mortality studies (Mackenbach et al., 2018; Permanyer of the respiratory system, and cancer. Our detailed analysis of 45 causes et al., 2018). In order to ensure a sufficiently large sample size in all of death by age groups pointed towards the important contribution of groups we distinguished three educational groups (Spijker & 5 S. Trias-Llimós et al. SSM - Population Health 23 (2023) 101461 Table 3 Age- and cause-specific contributions to the high-low educational gaps in lifespan variation at age 30, Spain, 2016-19. List of 45 causes of death. Males Females 30–49 50–64 65–84 85+ 30+ 30–49 50–64 65–84 85+ 30+ AIDS − 0.03 − 0.02 0.00 0.00 ¡0.06 − 0.03 − 0.01 0.00 0.00 ¡0.04 Other infectious diseases − 0.04 − 0.03 0.00 0.00 ¡0.07 − 0.03 − 0.03 − 0.01 0.01 ¡0.06 Benign neoplasms − 0.01 − 0.01 0.00 0.00 ¡0.02 − 0.01 − 0.01 0.00 0.00 ¡0.02 Breast cancer 0.00 0.00 0.00 0.00 0.00 − 0.02 − 0.03 0.00 − 0.01 ¡0.05 Intestine cancer − 0.03 − 0.04 0.00 0.00 ¡0.06 − 0.02 − 0.03 0.00 0.00 ¡0.06 Liver cancer − 0.04 − 0.05 0.00 0.00 ¡0.09 − 0.01 − 0.02 − 0.01 0.00 ¡0.04 Lung cancer − 0.06 − 0.19 − 0.01 0.00 ¡0.25 − 0.03 − 0.01 0.03 − 0.01 ¡0.01 Other neoplasms − 0.09 − 0.09 − 0.01 − 0.02 ¡0.21 − 0.10 − 0.13 − 0.02 − 0.01 ¡0.27 Prostate cancer 0.00 − 0.01 0.00 0.00 ¡0.01 0.00 0.00 0.00 0.00 0.00 Stomach cancer − 0.02 − 0.02 0.00 0.00 ¡0.03 − 0.02 − 0.02 − 0.01 0.00 ¡0.05 Upper Aero cancers − 0.03 − 0.05 0.00 0.00 ¡0.09 − 0.01 − 0.01 0.00 0.00 ¡0.02 Blood diseases − 0.01 0.00 0.00 0.00 ¡0.01 − 0.01 − 0.01 0.00 0.00 ¡0.02 Diabetes − 0.02 − 0.03 0.01 0.01 ¡0.02 − 0.01 − 0.03 − 0.01 0.03 ¡0.01 Other endocrine diseases − 0.02 − 0.01 0.00 0.00 ¡0.03 − 0.02 − 0.02 0.00 0.00 ¡0.04 Alcohol psychosis 0.00 − 0.01 0.00 0.00 ¡0.01 0.00 0.00 0.00 0.00 0.00 Dementia 0.00 − 0.01 0.01 0.02 0.02 0.00 − 0.01 0.00 0.03 0.01 Drug dependence 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 Other mental diseases − 0.02 − 0.01 0.00 0.00 ¡0.03 − 0.02 − 0.01 0.00 0.00 ¡0.03 Alzheimer 0.00 0.00 0.01 0.00 0.00 0.00 − 0.01 − 0.01 0.01 0.00 Epilepsy − 0.04 − 0.01 0.00 0.00 ¡0.05 − 0.04 − 0.01 0.00 0.00 ¡0.05 Other diseases nervous system − 0.14 − 0.04 0.00 − 0.01 ¡0.20 − 0.14 − 0.06 − 0.01 − 0.01 ¡0.21 Cerebrovascular diseases − 0.05 − 0.05 0.01 0.01 ¡0.08 − 0.04 − 0.05 − 0.01 0.04 ¡0.06 Hypertension − 0.01 − 0.01 0.01 0.01 ¡0.01 − 0.01 − 0.01 0.00 0.02 0.00 Ischaemic heart diseases − 0.08 − 0.13 0.00 − 0.01 ¡0.22 − 0.04 − 0.06 − 0.02 0.01 ¡0.11 Other circulatory diseases − 0.04 − 0.03 0.00 0.00 ¡0.06 − 0.04 − 0.04 − 0.01 0.00 ¡0.09 Other heart diseases − 0.10 − 0.07 0.01 0.01 ¡0.16 − 0.04 − 0.06 − 0.01 0.04 ¡0.07 Acute lower respiratory diseases − 0.04 − 0.04 0.01 0.01 ¡0.07 − 0.06 − 0.03 0.00 0.01 ¡0.08 Asthma 0.00 0.00 0.00 0.00 0.00 − 0.01 − 0.01 0.00 0.00 ¡0.01 Chronic lower respiratory diseases − 0.02 − 0.05 0.03 0.05 0.02 − 0.01 − 0.03 0.00 0.00 ¡0.04 Other diseases of the respiratory system − 0.11 − 0.06 0.01 0.01 ¡0.14 − 0.08 − 0.04 − 0.01 0.01 ¡0.11 Chronic liver disease − 0.05 − 0.08 0.00 0.00 ¡0.13 − 0.02 − 0.04 − 0.01 0.00 ¡0.06 Other digestive diseases − 0.07 − 0.06 0.01 0.02 ¡0.10 − 0.05 − 0.05 − 0.01 0.02 ¡0.10 Skin diseases 0.00 0.00 0.00 0.00 ¡0.01 0.00 0.00 0.00 0.00 ¡0.01 Musculoskeletal diseases 0.00 − 0.01 0.00 0.00 ¡0.01 − 0.01 − 0.01 0.00 0.00 ¡0.02 Genitourinary diseases − 0.02 − 0.02 0.00 0.01 ¡0.03 − 0.02 − 0.03 − 0.01 0.02 ¡0.04 Other diseases − 0.06 − 0.03 0.00 0.00 ¡0.09 − 0.09 − 0.05 0.00 0.00 ¡0.15 Mechanisms of the death − 0.03 − 0.01 0.00 0.00 ¡0.04 − 0.02 − 0.01 0.00 0.00 ¡0.02 Other ill-defined causes − 0.09 − 0.04 0.00 0.00 ¡0.12 − 0.05 − 0.02 0.00 0.00 ¡0.07 Accidental drowning and exposure to substances − 0.10 − 0.02 0.00 0.00 ¡0.11 − 0.02 − 0.01 0.00 0.00 ¡0.03 Accidental falls − 0.02 − 0.01 0.00 0.00 ¡0.03 − 0.01 0.00 0.00 0.00 ¡0.01 Other accidents − 0.03 − 0.01 0.00 0.00 ¡0.04 − 0.03 − 0.01 0.00 0.00 ¡0.04 Assault − 0.03 0.00 0.00 0.00 ¡0.03 − 0.01 0.00 0.00 0.00 ¡0.02 Other external causes − 0.04 − 0.01 0.00 0.00 ¡0.05 − 0.02 − 0.01 0.00 0.00 ¡0.03 Suicide − 0.10 − 0.02 0.00 0.00 ¡0.11 − 0.02 0.00 0.00 0.00 ¡0.02 Transport accidents − 0.07 − 0.01 0.00 0.00 ¡0.09 − 0.01 − 0.01 0.00 0.00 ¡0.02 All causes − 1.76 − 1.41 0.11 0.12 ¡2.93 − 1.22 − 1.03 − 0.15 0.23 ¡2.17 Trias-Llimós, 2023). Second, we chose to compare the highest and the results, even in our case where a rather large number of causes of death lowest educational groups. A sensitivity analysis comparing each were assessed. educational group with the total population suggests consistent results Assessing both life expectancy and lifespan variation inequalities was (Figs. S2 and S3), therefore our choices of educational attainment cat­ particularly insightful for the (groups of) causes in which the differences egories did not have a major impact on our conclusions. Further research between high and low educational groups are concentrated either at should explore how other socioeconomic dimensions, for instance in­ young or old ages (Seligman et al., 2016). For example, among males, come or labour market trajectories, are associated with cause-specific respiratory system diseases explained 13.8% of the life expectancy gap mortality. Third, we used life expectancy and lifespan variation as an but only 6.5% of the lifespan variation gap; while external causes -more indicator of age-at-death variability. We should acknowledge that these prevalent at younger ages-explained 8.7% and 15.7%, respectively. metrics use cross-sectional data to build fictitious cohorts, and therefore Similarly, cardiovascular diseases among females -particularly impor­ that our results do not reflect inequalities from any specific cohort (Luy tant at old ages-had a notable impact on life expectancy gaps (30.1%), et al., 2020). Fourth, due to data limitations we adopted a but a much lower impact on lifespan variation gaps (12.0%). In other cross-sectional perspective. We acknowledge that, particularly in the words, deaths that kill at young ages (e.g. external causes) have a higher case of Spain with a rapid educational expansion and a lag in time be­ impact on population-level mortality indicators (Lazarus et al., 2022). tween genders and social classes in the adoption of unhealthy lifestyles, Whereas this is a straightforward conclusion, the relative difference a longitudinal study could provide further insights to our results. Un­ between these contributions and contributions from causes that tend to fortunately, longitudinal long-term mortality data is not yet available in be fatal at older ages (e.g. cardiovascular diseases) seems not negligible. Spain. Fifth, we used a detailed list of 45 causes of death, including the Therefore, this suggests that efforts to reduce mortality inequalities major diseases. More granular detail and estimates for a list of 72 causes should carefully consider both mean and length of life indicators as they of death can be found in Table S2. Finally, as in any study focusing on could yield different results, and provide essential knowledge when causes of death, the grouping of the causes has a direct impact on the aiming at reducing mortality inequalities. 6 S. Trias-Llimós et al. SSM - Population Health 23 (2023) 101461 The results of this study suggest that educational inequalities per­ mortality in ageing societies. These patterns observed in Spain are sisted up to the pre-Covid-19 times, particularly in premature mortality insightful and could motivate studies on mortality inequalities in other (before age 75). Our cause-specific results on the important contribution low-mortality countries. Yet, monitoring detailed cause-of-death mor­ of lung cancer and other respiratory diseases to overall inequalities in tality inequalities is a priority in light of the potential increases of them mortality for males are in line with previous results from the early 2000s derived from the consequences of the COVID-19 pandemic (Bambra (Reques et al., 2014). Yet, we identified a set of causes of death that were et al., 2020). In conclusion, improving our understanding of inequalities overlooked in previous Spanish studies and that had an important in mortality in low-mortality settings would allow policymakers to draw impact on mortality inequalities. For instance, Alzheimer’s and other better preventive policies to improve health and survival across all dementias contributed to almost three months (6.3%) of the life ex­ educational groups while reducing the differences therein. pectancy gap for females. In Spain, the prevalence of dementia at age 65+ ranged between 4 and 9%, although it was higher for females and Funding increased rapidly with age (Villarejo Galende et al., 2021). Whereas robust studies on dementia prevalence by education in Spain are lack­ This study was supported with funding provided by the European ing, a systematic review of international studies suggests lower de­ Research Council (ERC-2019-COG agreement 864616, HEALIN) and by mentia prevalence among higher-educated groups (Bodryzlova et al., the Spanish Ministry of Science and Innovation (projects PID2020- 2022). Our findings for Spain are also supported by other international 113934RB-I00 and PID2021-128892OB-I00). STL acknowledges studies that suggest poorer Alzheimer’s prognosis among low social research funding from the Ramón y Cajal program of the Spanish Min­ classes (Ono et al., 2022), and an important role in socioeconomic in­ istry of Science and Innovation (RYC2021-033123-I). equalities in old age (Korhonen et al., 2020). A particularly interesting finding was the observed differences in the Author statement contribution of causes of death to educational inequalities in mortality between males and females. In men, most inequalities were concen­ Conceptualization: STL, JS, IP. trated at younger ages and were due to lung cancer, cardiovascular, Data curation: STL. respiratory and digestive diseases. A relatively large fraction of educa­ Formal analysis: STL. tion gaps in mortality seems to be explained by behavioural-related Funding acquisition: STL, JS, IP. causes of death such as lung, lip, pharynx and oesophagus cancers or Investigation: STL, JS, AB, IP. respiratory diseases, which is well-aligned with previous research sug­ Methodology: STL, IP. gesting that smoking account for 1.5-year educational gap in life ex­ Project administration: STL. pectancy among men (Piñeiro et al., 2022). Chronic liver diseases, which Resources: STL, JS, IP. are partly caused by alcohol consumption, were also important in men, Software: STL. but not in women, and a similar story is observed for suicides and other Supervision: JS, AB, IP. external causes. From a social determinants of health perspective, these Validation: STL. lifestyle-related causes can also be influenced by living and working Visualization: STL. conditions or the neighborhood environment (Commission on Social Roles/Writing - original draft: STL, JS. Determinants of Health, 2008). In contrast, for females, the age contri­ Writing - review & editing: STL, JS, AB, IP. butions were shifted towards older ages, and the causes that contributed the most to mortality inequalities were driven by cardiovascular disease, Data availability diabetes and dementia-related diseases. This is in line with recent research that found higher educational inequalities in females compared The authors do not have permission to share data. to males in cardiovascular mortality (Haeberer et al., 2020), as well as in obesity and hypertension prevalence (Gullón et al., 2021). The fact that Appendix A. Supplementary data the detrimental effects of unhealthy lifestyles impacted a greater num­ ber of men with low educational attainment compared to both other Supplementary data to this article can be found online at https://doi. men and women may be related to the social determinants of health org/10.1016/j.ssmph.2023.101461. framework as education places a crucial role in health literacy, i.e. and individual’s ability to obtain, understand, and apply health information (Kuh & Shlomo, 2004). Individuals with higher education levels are References better able to make informed decisions about their health and engage in Aburto, J. M., & van Raalte, A. (2018). Lifespan Dispersion in times of life expectancy preventive measures. In a similar way, due to gender norms, traditional Fluctuation: The case of Central and Eastern Europe. Demography, 55(6), 2071–2096. masculine ideals often discourage help-seeking behaviours and vulner­ Dec. ability (Evans et al., 2011). As individuals grow older, however, Aburto, J. M., Villavicencio, F., Basellini, U., Kjærgaard, S., & Vaupel, J. W. (2020). Dynamics of life expectancy and life span equality. Proc Natl Acad Sci, 117(10), educational differences in age-related disease mortality decreased 5250–5259. among men, possibly because the male population became more Aburto, J. M., Wensink, M., van Raalte, A., & Lindahl-Jacobsen, R. (2018). Potential selected than the female population. gains in life expectancy by reducing inequality of lifespans in Denmark: An international comparison and cause-of-death analysis. BMC Public Health, 18(1), 831. Dec. 5. Conclusions Bambra, C., Riordan, R., Ford, J., & Matthews, F. (2020). The COVID-19 pandemic and health inequalities. Journal of Epidemiology & Community Health, 74(11), 964–968. Nov 1. This study showed that important differences existed in the cause- Bartoll-Roca, X., Rodríguez-Sanz, M., Sánchez-Ledesma, E., Pérez, K., & Borrell, C. specific contribution to educational differences in both life expectancy (2022). Inequalities in life expectancy by educational level and its decomposition in and lifespan variation. Reducing inequalities in respiratory disease will Barcelona, 2004-2018. Gaceta Sanitaria, 36(6), 520–525. Blanes, A., & Trias-Llimós, S. (2021). Shorter lives with poor health: The Toll on Spain’s mostly contribute to reducing life expectancy gaps, but will have a more less educated population. Perspect Demogràfiques., 24, 1–4. limited impact on lifespan variation gaps. Additionally, an increasing Bodryzlova, Y., Kim, A., Michaud, X., André, C., Bélanger, E., & Moullec, G. (2022). part of the educational gap in both life expectancy and lifespan variation Social class and the risk of dementia: A systematic review and meta-analysis of the is attributable to ageing factors, especially in females. This suggests that, prospective longitudinal studies. Scand J Public Health, Article 14034948221110020. Commission on Social Determinants of Health. (2008). Closing the gap in a generation: as mortality switches towards older ages, there is a growing importance Health equity through action on the social determinants of health. In Final Report of of dementias and other neurodegenerative diseases on all-cause the Commission on social determinants of health. Geneva: World Health Organization. 7 S. Trias-Llimós et al. SSM - Population Health 23 (2023) 101461 Davey Smith, G., Hart, C., Hole, D., MacKinnon, P., Gillis, C., Watt, G., et al. (1998). based study of 15 European countries between 1990 and 2015. European Journal of Education and occupational social class: Which is the more important indicator of Epidemiology, 34(12), 1131–1142. mortality risk? Journal of Epidemiology & Community Health, 52(3), 153–160. Ono, R., Uchida, K., Nakatsuka, K., Megumi, M., & Fukuda, H. (2022). Economic Status Evans, J., Frank, B., Oliffe, J. L., & Gregory, D. (2011). Health, illness, men and and mortality in Patients with Alzheimer’s disease in Japan: The longevity masculinities (HIMM): A theoretical framework for understanding men and their Improvement and Fair Evidence study. Journal of the American Medical Directors health. Journal of Men’s Health, 8(1), 7–15. Association, 23(1), 161–164. Galobardes, B., Shaw, M., Lawlor, D. A., Lynch, J. W., & Smith, G. D. (2006). Indicators of Permanyer, I., Spijker, J., Blanes, A., & Renteria, E. (2018). Longevity and lifespan socioeconomic position (part 1). Journal of Epidemiology & Community Health, 60(1), variation by educational attainment in Spain: 1960–2015. Demography, 55(6), 7–12. 2045–2070. Dec. Gullón, P., Díez, J., Cainzos-Achirica, M., Franco, M., & Bilal, U. (2021). Social inequities Piñeiro, B., Trias-Llimós, S., Spijker, J. J., Llorens, A. B., & Permanyer, I. (2022). in cardiovascular risk factors in women and men by autonomous regions in Spain. Estimation of smoking-related mortality and its contribution to educational Gaceta Sanitaria, 35(4), 326–332. inequalities in life expectancy in Spain: An observational study, 2016–2019. BMJ Haeberer, M., León-Gómez, I., Pérez-Gómez, B., Tellez-Plaza, M., Rodríguez-Artalejo, F., Open, 12(8), Article e059370. & Galán, I. (2020). Social inequalities in cardiovascular mortality in Spain from an Preston, S., Patrick, H., & Michel, G. (2001). Demography: Measuring and modeling intersectional perspective. Rev Esp Cardiol Engl Ed, 73(4), 282–289. population processes. Oxofrd, UK: Blackwell Publ. Horiuchi, S., Wilmoth, J. R., & Pletcher, S. D. (2008). A decomposition method based on van Raalte, A. A., Sasson, I., & Martikainen, P. (2018). The case for monitoring life-span a model of continuous change. Demography, 45(4), 785–801. inequality. Science, 362(6418), 1002–1004. Nov 30. Huisman, M. (2004). Socioeconomic inequalities in mortality among elderly people in 11 Regidor, E., Vallejo, F., Granados, J. A. T., Viciana-Fernández, F. J., de la Fuente, L., & European populations. Journal of Epidemiology & Community Health, 58(6), 468–475. Barrio, G. (2016). Mortality decrease according to socioeconomic groups during the Jun 1. economic crisis in Spain: A cohort study of 36 million people. The Lancet, 388 Korhonen, K., Einiö, E., Leinonen, T., Tarkiainen, L., & Martikainen, P. (2020). Midlife (10060), 2642–2652. Nov. socioeconomic position and old-age dementia mortality: A large prospective register- Requena, M. (2017). La desigualdad ante la muerte: Educación y esperanza de vida en based study from Finland. BMJ Open, 10(1), Article e033234. Jan. España. 2017; Perspect Demogràfiques, 6, 1–4. Kuh, D., & Shlomo, Y. B. (2004). A life course approach to chronic disease epidemiology. Reques, L., Giráldez-García, C., Miqueleiz, E., Belza, M. J., & Regidor, E. (2014). Oxford university press. Educational differences in mortality and the relative importance of different causes Kulhánová, I., Bacigalupe, A., Eikemo, T. A., Borrell, C., Regidor, E., Esnaola, S., et al. of death: A 7-year follow-up study of Spanish adults. Journal of Epidemiology & (2014). Why does Spain have smaller inequalities in mortality? An exploration of Community Health, 68(12), 1151–1160. Dec. potential explanations. Eur J Public Health, 24(3), 370–377. Riffe, T. (2018). DemoDecomp: Decompose Demographic Functions [R Package Version Lazarus, J. V., Ortiz, A., Tyrovolas, S., Fernández, E., Guy, D., White, T. M., et al. (2022). 1.0.1] https://cran.r-project.org/web/packages/DemoDecomp/index.html. A GBD 2019 study of health and Sustainable Development Goal gains and forecasts Seligman, B., Greenberg, G., & Tuljapurkar, S. (2016). Equity and length of lifespan are to 2030 in Spain. Scientific Reports, 12(1), 1–19. not the same. Proc Natl Acad Sci, 113(30), 8420–8423. Luy, M., Di Giulio, P., Di Lego, V., Lazarevič, P., & Sauerberg, M. (2020). Life expectancy: Smits, J., & Monden, C. (2009). Length of life inequality around the globe. Social Science Frequently used, but hardly understood. Gerontology, 66(1), 95–104. & Medicine, 68(6), 1114–1123. Mar. Machón, M., Mosquera, I., Larrañaga, I., Martín, U., & Vergara, I. (2020). Desigualdades Spijker, J. J. A., & Trias-Llimós, S. (2023). Cause-specific mortality in Spain during the socioeconómicas en la salud de la población mayor en España. Gaceta Sanitaria, 34, pandemic: Educational differences and its impact on life expectancy. Eur J Public 276–288. Sep 11. Health, 33(3), 543–549. Mackenbach, J. P., Valverde, J. R., Artnik, B., Bopp, M., Brønnum-Hansen, H., Villarejo Galende, A., Ortiz, M. E., Velasco, S. L., Luque, M. L., de Miguel, CL. de S., & Deboosere, P., et al. (2018). Trends in health inequalities in 27 European countries. Jurczynska, C. P. (2021). Report by the Spanish Foundation of the Brain on the social Proc Natl Acad Sci, 115(25), 6440–6445. Jun 19. impact of Alzheimer disease and other types of dementia. Neurol Engl Ed, 36(1), Mackenbach, J. P., Valverde, J. R., Bopp, M., Brønnum-Hansen, H., Costa, G., 39–49. Deboosere, P., et al. (2019). Progress against inequalities in mortality: Register- 8