Artículo de revista: Spijker, Jeroen; Trias-Llimós, Sergi (2023) Cause-specific mortality in Spain during the pandemic: educational differences and its impact on life expectancy, European Journal of Public Health, ckad036 (ISSN 1101-1262) https://doi.org/10.1093/eurpub/ckad036 European Journal of Public Health, 1–7 ß The Author(s) 2023. Published by Oxford University Press on behalf of the European Public Health Association. This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial License (https://creativecommons.org/ licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com https://doi.org/10.1093/eurpub/ckad036 ......................................................................................................... Cause-specific mortality in Spain during the pandemic: educational differences and its impact on life expectancy Downloaded from https://academic.oup.com/eurpub/advance-article/doi/10.1093/eurpub/ckad036/7082922 by guest on 25 April 2023 Jeroen J.A. Spijker , Sergi Trias-Llimós Centre d’Estudis Demogràfics, Centres de Recerca de Catalunya (CERCA), Carrer de Ca n’Altayó, Edifici E2, Universitat Autònoma de Barcelona, Bellaterra, Spain Correspondence: Jeroen Spijker, Centre d’Estudis Demogràfics, Centres de Recerca de Catalunya (CERCA), Carrer de Ca n’Altayó, Edifici E2, Universitat Autònoma de Barcelona, 08193 Bellaterra, Spain, Tel: þ31 935813060, Fax: þ31 935813061, e-mail: jspijker@ced.uab.es Background: Life expectancy in Spain fell by 1 year between 2018/19 and 2020. Yet, little is known on the impact on cause-of-death (COD) dynamics and educational inequalities therein. Methods: We use individual-level death counts data by age, sex, education and multiple causes of death (MCOD) and the corresponding population exposures from national registers in Spain. Deaths were examined both as underlying cause of death (UCOD) and as contributory cause. We estimated life expectancies and lifespan inequalities by subpopulation groups and decomposed life expectancy differences between 2018/19 and 2020 by age groups and COD to assess the impact of COVID-19 (as MCOD) and major UCOD. Results: COVID-19 contributed to a decline in male and female life expectancy in Spain between 2018/19 and 2020 (respectively, 1.7 and 1.4 years). Conversely, cancer, respiratory and circulatory system diseases and ill-defined causes as UCOD contributed to life expectancy increases. Life expectancy declines equalled 1.4 years among the low-educated in both sexes (population 30þ), 1.0 and 0.7 years among middle-educated and 1.1 and 0.9 years among high-educated men and women. Without COVID-19, educational inequalities in life expectancy would have remained at similar levels, whereas lifespan variation would have been lower (22% for women and 8% for men). Conclusions: Life expectancy declines in Spain in 2020 were mainly driven by COVID-19, with possible substitution effects, especially for respiratory system diseases (fewer deaths compared to 2018/19 when coded as UCOD but more as contributing cause). We therefore advocate analysing MCOD when studying changing COD patterns during the pandemic. ......................................................................................................... Background of mortality dynamics, particularly in exceptional situations such as the COVID-19 pandemic.15 uring 2020–21, Spain was hit hard by COVID-19 pandemic, However, cause-specific mortality research during the pandemic Dcosting the lives of 90 035 persons according to official statistics. 2–4 1 years is still scarce. Exceptions include Fedeli et al.19 who observed Research on the impact of COVID-19 on excess deaths and life not only the emergence of COVID-19 mortality in the Veneto region expectancy5 is already well-covered. In Spain, the first wave of the (Italy) during the first months of the pandemic but also related re- pandemic led to a 0.9-year decline in life expectancy at birth,6 while spiratory conditions, diabetes mellitus, hypertensive heart disease, annual estimates for 2020 suggest a 1.2-year decline compared to the cerebrovascular diseases and ill-defined causes as the underlying previous year7,8 or 1.5 years when compared to the average yearly cause. Concurrently, when MCOD were analysed, sharp increases change between 2015 and 2019.5 were also found for dementia and Alzheimer’s and chronic lower COVID-19 mortality risks are known to be associated with mor- respiratory diseases.19 A Brazilian study assessing COD from an bidity trajectories.9,10 Most studies on cause-of-death (COD) mor- MCOD perspective revealed that hypertension and diabetes were tality dynamics focus on the assessment of the principal or so-called the most common contributory causes associated with COVID- underlying cause of death (UCOD). However, in a context of ageing 19.15 Thus, assessing the COD dynamics in COVID-19 times from societies and mortality shifts towards older ages, there is a growing an MCOD perspective seems particularly insightful. COVID-19 in- concern regarding the assessment of mortality using only the fection and mortality rates are also known to differ according to UCOD.11–13 The COVID-19 pandemic is not an exception as almost individual-level demographic and socioeconomic characteristics, all COVID-19 deaths recorded as UCOD in 2020 had other condi- including age, sex and educational attainment.20–22 Educational at- tions mentioned on the death certificate (e.g. 92% in the USA14 and tainment is a frequently used indicator of socioeconomic status in 98.4% among 30- to 69-year olds living in three Brazilian capitals15). health and mortality research.23–25 This implies a high prevalence of comorbidity among people who In this article, we aim to analyse the impact of COVID-19 deaths died from COVID-19.15–17 Moreover, the UCOD approach under- alongside all other COD reported on the death certificate on the life estimates the mortality burden from diseases that are not commonly expectancy in 2020 in Spain, one of the most affected countries in selected as the UCOD but nevertheless contribute to the sequence of Europe by the pandemic.5 In doing so, we account for deaths that causes in the death process, as is known to occur with diabetes, had COVID-19 mentioned on the death certificate as underlying or influenza and pneumonia, or hypertension.18 The use of multiple contributing cause. In addition, we assess other COD frequently causes of death (MCOD) should therefore aid the understanding mentioned on the death certificate in the presence of COVID-19, 2 of 7 European Journal of Public Health ascertain any differences in the relative risks of having died from Analytical approach other diseases compared to the average for the two preceding years First, we compare the distribution of the main UCOD (traditional (2018–19) and explore the (change in) educational inequalities in view on COD analyses) and any mention of a selected COD in the mortality during this period. death certificate (MCOD) in 2020 with the 2018/19 average. Herewith, we obtain an indication of the magnitude of both COVID-19 mortality and any changes in the patterns of other im- Methods portant, frequently reported conditions, especially during the initial stages of the pandemic, including flu and pneumonia and other re- Data spiratory system diseases (see also19), dementia/Alzheimer, cardio- We used individual-level mortality and aggregated population level vascular diseases and external causes. Downloaded from https://academic.oup.com/eurpub/advance-article/doi/10.1093/eurpub/ckad036/7082922 by guest on 25 April 2023 data for 2018–20 for Spain that was requested from the Spanish Secondly, we define COVID-19-related deaths as those with a National Statistics Institute (INE) (www.ine.es). Both datasets consist COVID-19 cause on the death certificate, irrespective of their pos- of secondary data of public sources, which did not have any indi- ition within the death certificate. For the remaining deaths, the vidual identifiers. As such, no ethical approval for human subject UCOD was used. Using conventional life table methods, we esti- research from the Institutional Review Board of the respective insti- mated life expectancies at birth by sex and according to educational tutions was required. The datasets contain age, sex and education attainment at age 30.26 Life expectancy gaps between 2018/19 and and, regarding the mortality dataset, also all COD listed on the death 2020 were estimated and differences decomposed into (i) age- and cause- and (ii) education- and cause-specific components using the certificate. We used a detailed COD list of 33 COD including decomposition derived by Horiuchi using the DemoDecomp package COVID-19 (suspected or diagnosed) (Supplementary table S1). in R.27,28 Information on the COD data production process, death certificates Lastly, we estimated the contribution of COVID-19 to life expect- and coding methods can be found elsewhere (see Supplementary ancy and lifespan variation at age 30 (measured as the standard references S1 and S2). Regarding the educational attainment of the deviation in age at death) by sex and educational group by compar- deceased and population at risk, INE estimated this through data ing the corresponding values for the whole population and excluding linkage from multiple datasets, including censuses, municipal popu- COVID-19-related deaths, defined as potential gains in life expect- lation registers (Padrón), data on enrolment and university and non- ancy (PGLE).29 This method was recently applied to estimate the university diplomas from the Ministry of Education. For the 0.2% contribution of alcohol to life expectancy and lifespan variation in that this was not possible, INE probabilistically imputed a value Spain.30 based on the 2011 Census and considering the person’s province of residence and age (see Supplementary reference S3 and figure S1 for the distribution of the Spanish population according to age, Results sex and educational attainment). Based on the mortality and popu- In 2020, 74 839 deaths had COVID-19 as the UCOD (15.2% of all lation data sources, we created three educational attainment catego- deaths) and 8275 additional deaths (1.9% of all deaths) had COVID- ries: low (primary education or less, ISCED-2011 0–1), middle (lower 19 mentioned elsewhere on the death certificate (table 1). While and upper secondary education and lower vocational training, deaths from COVID-19 occurred at all ages, there is a clear positive ISCED-2011 2–3) and high (postsecondary vocational training and age gradient: 18.2% of the 85þ who died had COVID-19 mentioned university, ISCED-2011 4þ). on their death certificate, compared to <1% among infants. Table 1 Number of deaths due to COVID-19 as underlying cause of death (UCOD), contributing cause of death and all-cause mortality in Spain in 2020 according to the age, sex and educational level of the deceased and period of death Number of deaths Ratio (%) All causes COVID-19 COVID-19 UCOD/all Any mention of causes COVID-19/all causes As UCOD As contributing Total 493 776 74 839 8275 15.2 16.8 Age 0 890 4 3 0.4 0.8 1–4 188 1 1 0.5 1.1 5–29 2139 81 17 3.6 4.3 30–44 6715 331 96 4.9 6.4 45–64 54 248 4819 857 8.9 10.5 65–84 192 772 30 416 3374 15.8 17.5 85þ 236 824 39 192 3927 16.5 18.2 Gender Male 249 664 38 917 4264 15.6 17.3 Female 244 112 35 922 4011 14.7 16.4 Education (30þ) Low 282 731 44 286 4673 15.7 17.3 Medium 149 986 21 666 2575 14.4 16.2 High 50 055 7745 867 15.5 17.2 Missing 11 004 1142 160 10.4 11.8 Month of death January–February 79 716 0 4 0.0 0.0 March–May (wave 1) 154 302 45 988 4270 29.8 32.6 June–August (summer) 101 561 2726 1069 2.7 3.7 September–December 158 197 26 125 2932 16.5 18.4 Cause-specific mortality in Spain in 2020 3 of 7 Proportions were relatively similar between men (17.3%) and women 0.10 years; medium educated: 0.01 years; no change among (16.4%) and educational categories (low 17.3%; medium 16.2%; high higher educated). 17.2%). Finally, the proportion of deaths directly due to COVID-19 Regarding cause-specific contributions, COVID-19 was actually was highest during wave 1 (29.8%), followed by the last four months responsible for more than the total decline in life expectancy at birth of the year (16.5%). Just 2.7% of deaths during the summer months (1.67 years for men and 1.34 years for women). This implies that had COVID-19 as UCOD. mortality from other COD declined during the pandemic, especially Compared to 2018/19, the mean number of causes per death cer- respiratory system diseases (men: contributing to þ0.21 years in life tificate hardly changed in 2020 (3.8 vs. 3.7) (table 2). When specified expectancy; women: þ0.15 years), male lung cancer (þ0.08 years), according to UCOD, increases can be observed for external causes remaining neoplasms (men: þ0.14 years; women: þ0.07 years), heart (accidental falls, from 3.6 in 2018/19 to 4.2 in 2020, other accidental and other diseases of the circulatory system (men: þ0.04 years; threats to breathing, from 3.6 to 4.0, and suicide and self-inflicted women: þ0.08 years) and ill-defined causes (men: þ0.09 years; Downloaded from https://academic.oup.com/eurpub/advance-article/doi/10.1093/eurpub/ckad036/7082922 by guest on 25 April 2023 injury, from 1.4 to 1.8 causes per death certificate). Regarding the women: þ0.02 years). Concurrently, there were also several COD, frequency of the different UCOD, deaths from diseases related to the which contributed to minor declines in life expectancy in 2020 com- respiratory system declined when they were reported as the under- pared to 2018/19 (diabetes 0.02 years men, 0.01 years women; lying cause (flu and pneumonia 16.4%, chronic lower respiratory hypertensive diseases 0.02 years both sexes). diseases 10.5% and remaining respiratory system diseases 20.6%). Finally, most of the life expectancy gains due to non-COVID-19 Death from sepsis (blood poisoning) also declined (7.3%), as well were made in the 65-84 age group in the case of men (0.28 years). as ill-defined causes (7.0%), transport accidents (18.9%) and Regarding women, the 85þ was the main age group responsible for other accidental threats to breathing (7.0%). Mortality from differ- the gains in life expectancy due to lower non-COVID-19 mortality ent cancer sites remained relatively stable. Conversely, important (0.19 years). increases were observed for diabetes (þ15.5%), other endocrine, nu- tritional and metabolic diseases (þ11.5%), hypertensive disease Counterfactual scenario by eliminating COVID-19 (þ17.2%), diseases of the genitourinary system (þ12.7%), accidental mortality falls (þ12.0%) and suicide and self-inflicted injury (þ9.3%). However, the impact of the pandemic on other diseases becomes Life expectancy at age 30 was estimated to be 1.67 years higher for clearer when analysing MCOD as each COD category analysed men and 1.38 years higher for women when COVID-19 mortality appeared more often on the death certificate in 2020 than in 2018/ was eliminated (table 4). The standard deviation in age at death 19, except transport accidents (17.7%). This was particularly the (lifespan inequality) would be larger as well when eliminating case with flu and pneumonia (þ63.0%), but also the endocrine, nu- COVID-19. Results by educational level suggest similar values in tritional and metabolic diseases (diabetes þ25.8%; obesity þ51.9%; terms of PGLE. The exception is for low-educated females, leading other þ21.7%), dementia and Alzheimer (þ18.9%), hypertensive therefore to lower educational inequalities in lifespan inequality diseases (þ32.6%), diseases of the genitourinary system (þ16.4%) without COVID-19 compared to the observed estimates that include and ill-defined causes (þ19.5%). Although less important numeric- COVID-19 (22% for women and 8% for men). ally, accidental falls also saw an increase of 17.2% during 2020. Hence, the MCOD/UCOD ratio increased for most COD, suggesting Discussion an increase in co- or multimorbidity during 2020. Finally, COVID- 19 co-occurred in all COD categories analysed, ranging from 0.1% We analysed the COD dynamics on life expectancy differences be- when an ill-defined cause was the UCOD to 3.1% in the case of flu, tween 2018/19 and 2020 in Spain. We made use of both the UCOD pneumonia and COPD and 3.2% for malignant neoplasm of lymph- and MCOD data that is available on the death certificate, as well as oid, haematopoietic and related tissue. the information on the educational attainment of the deceased that the INE obtained through data linkage. Life expectancy dropped by 0.99 years among men and 0.95 years among women between 2018– Age, education and cause-specific contributions to life 19 and 2020. Contributions to this drop were highest among the 65- expectancy changes to 84-year-old age group (men: 0.65 years; women: 0.52 years). Life expectancy at birth declined by 0.99 years for men and 0.95 years Infant, child and young-adult mortality did not increase. for women between 2018/19 and 2020 (table 3). Most of the decline COVID-19, measured in death certificates as either underlying or occurred among the 65- to 84-year age group (0.65 years among contributory COD, exceeded the drop in total (all-cause) life expect- men and 0.53 years among women), followed by the 85þ age group ancy. In contrast, improvements were observed for other COD, par- (respectively, 0.22 and 0.34 years). Life expectancy among 45- to ticular respiratory system diseases, but also circulatory system diseases 64-year-old men and women declined by 0.12 and 0.09 years, (except hypertensive diseases) and cancer (except female lung cancer) respectively, and among 30- to 44-year olds merely 0.03 and showed improvements. With regard to the latter two COD categories, 0.01 years. One- to twenty-nine-year olds even gained 0.02 years, this is likely due to a continuing declining trend, although from just while no change took place in infant mortality (both sexes). one year of data this is difficult to confirm. For example, we know that Among all three educational categories, declines in life expectancy cancer patients have been at greater risk to manifest complications and at age 30 were observed between 2018/19 and 2020, whereby declines develop severe events from COVID-19, including intensive care unit were highest among the lowest educated, with minor sex differences admission and death, compared with patients without cancer, which (men: 1.36 years; women: 1.43 years), followed by the highest- would contribute to fewer cancer deaths because of a substitution educated group (men: 1.06 years; women: 0.86 years). The effect. Conversely, care, diagnostic tests and therapeutic intervention medium-educated group observed the lowest drop (men: for cancer patients were often delayed during the pandemic as the 0.96 years; women 0.66 years). Again, COVID-19 explained health care system distracted towards COVID-19. As a result, cancer most of these declines. Yet, differences in the dynamics of other survival rates may have experienced periodic declines.31,32 causes across educational groups are to be noted. While the medium Similarly, we observed notable declines in mortality from several and high-educated recorded life expectancy improvements in most COD when the underlying cause was analysed. For example, 16% cardiovascular disease categories (þ0.21 years for the total COD fewer deaths from pneumonia and influenza were recorded in 2020 chapter in the case of middle-educated women), the lower educated compared to 2018/19, there was a 11% decline in chronic lower did not, and even observed minor losses in hypertensive diseases respiratory diseases, 21% decline in remaining lower respiratory sys- (0.03 years among both sexes). Educational differences are also tem diseases and 7% fewer deaths from ill-defined causes. Mortality observed for external causes, but only among men (lower educated: declines from these causes were also found using UCOD in England 4 of 7 European Journal of Public Health Table 2 Distribution of causes of death registered in 2020 in Spain compared to 2018/19: underlying cause of death (UCOD) and multiple causes of death (MCOD) Cause of death Mean n of causes per Number of deaths death certificate by UCOD 2020 UCOD 2020 MCOD 2020 Ratio MCOD/UCOD % COVID-19 deaths on certificate 2020 N D 2020 vs. N D 2020 vs. N D 2020 vs. 2020 2018/19 2018/19 (%) 2018/19 (%) 2018/19 (%) All deaths 3.8 3.7 493 776 16.7 – – – – 16.5 1 COVID-19 3.8 – 74 839 NAb 83 114 – 1.1 NAb 97.9a 2 Flu, pneumonia 3.4 3.4 9 662 16.4 48 889 63.0 5.1 2.6 3.1 Downloaded from https://academic.oup.com/eurpub/advance-article/doi/10.1093/eurpub/ckad036/7082922 by guest on 25 April 2023 3 Chronic lower respiratory diseases 4.2 4.1 12 351 10.5 27 088 7.3 2.2 1.8 3.1 4 Remaining respiratory system diseases 3.4 3.4 18 837 20.6 82 235 4.5 4.4 3.3 3.0 5 Sepsis 3.6 3.4 2 745 7.3 14 587 6.3 5.3 4.6 1.5 6 Remaining infectious and parasitic 4.5 4.4 3 211 2.6 14 917 38.2 4.6 3.3 2.8 diseases 7 Malignant neoplasm of lymphoid, 3.7 3.7 8 374 0.7 10 453 7.5 1.2 1.2 3.2 haematopoietic and related tissue 8 Lung cancer 3.7 3.6 23 099 1.0 25 590 1.7 1.1 1.1 2.2 9 Neoplasms of reproductive organs 3.9 3.8 16 803 1.8 21 397 6.0 1.3 1.2 2.0 10 Remaining neoplasms 3.8 3.7 64 465 0.2 126 676 2.3 2.0 1.9 1.8 11 Diseases of the blood (-forming organs), 4.4 4.3 2 062 4.2 16 028 10.4 7.8 7.3 2.8 immunological disorders 12 Diabetes mellitus 4.6 4.6 11 297 15.5 44 156 25.8 3.9 3.6 2.1 13 Obesity 4.6 4.6 503 1.7 3 952 51.9 7.9 5.3 2.2 14 Other endocrine, nutritional and meta- 4.3 4.3 3 490 11.5 33 772 21.7 9.7 8.9 2.2 bolic diseases 15 Dementia, Alzheimer 4.1 4.1 37 793 0.3 74 349 18.9 2.0 1.6 2.2 16 Other mental/behavioural disorders and 4.1 4.0 11 386 4.2 44 316 4.0 3.9 3.9 2.6 diseases of nervous system 17 Hypertensive diseases 4.3 4.3 14 271 17.2 73 194 32.6 5.1 4.5 1.4 18 Ischaemic heart diseases 3.6 3.6 29 654 1.8 44 210 5.0 1.5 1.4 1.7 19 Other heart diseases 3.9 3.8 14 357 0.6 36 885 8.9 2.6 2.4 1.8 20 Cerebrovascular diseases 3.7 3.6 25 817 1.0 43 659 5.1 1.7 1.6 1.8 21 Other diseases of the circulatory system 3.4 3.4 35 640 0.4 114 677 9.2 3.2 2.9 1.3 22 Chronic liver diseases 4.3 4.1 3 976 0.9 6 749 3.1 1.7 1.6 2.3 23 Other diseases of the digestive system 4.0 3.9 17 589 1.5 46 973 1.1 2.7 2.6 1.9 24 Diseases of the skin and subcutaneous 3.9 3.8 1 902 5.6 9 415 12.3 5.0 4.7 1.7 tissue 25 Diseases of the musculoskeletal system/ 4.4 4.3 5 225 0.8 14 768 13.3 2.8 2.5 2.5 connective tissue 26 Diseases of the genitourinary system 3.7 3.6 15 810 12.7 84 741 16.4 5.4 5.2 2.4 27 Other diseases 4.0 3.6 1 433 4.8 3 680 23.2 2.6 2.0 2.2 28 Ill-defined causes 1.7 1.6 11 107 7.0 496 992 19.5 44.7 34.8 0.1 29 Transport accidents 1.7 1.6 1 686 18.9 1 719 17.7 1.0 1.0 0.3 30 Accidental falls 4.2 3.6 3 605 12.0 4 211 17.2 1.2 1.1 2.1 31 Other accidental threats to breathing 4.0 3.6 2 511 7.0 17 115 1.7 6.8 6.2 1.1 32 Suicide and self-inflicted injury 1.8 1.4 3 941 9.3 3 955 9.6 1.0 1.0 0.3 33 Other external causes 3.3 3.1 4 335 0.4 30 568 13.7 7.1 6.2 1.3 Notes: See Supplementary table S1 for the ICD codes of each cause of death category. a: Does not equal to 100% because about 1600 COVID-19 cases did not have COVID-19 in the death certificate. Following the WHO recommendations, in March 2020, the optical character recognition (OCR) dictionary (used to scan the death certificate) was updated to include terms related to COVID-19 and was deployed by all the regional delegations of the INE. In the first version of the OCR dictionary, the terms included were mainly theoretical or based on the actual terms used by doctors in the first certificates that the INE had access to and where COVID-19 was reported (but whereby COVID-19 was not necessarily mentioned on the death certificate). Subsequently, a second version of the OCR dictionary was deployed in which more real terms used by doctors to name this disease were included. The discrepancy therefore pertains to cases that were posteriori coded by IRIS as COVID-19, as well as more complex cases that had been transferred to the WHO for recoding (see Supplementary reference S2). b: There were no COVID-19 deaths in Spain prior to 2020. and Wales.33 Given that COVID-19 is caused by the severe acute accidents (19%), which is line with to the reduction in mobility respiratory syndrome coronavirus 2 (SARS-CoV-2) a substitution early on during the pandemic and ensuing lockdowns.34 effect is likely behind the declines in these UCOD as we observed On the other hand, minor increases in mortality were observed for notable increases in the deaths with respiratory diseases as contribu- hypertensive diseases (þ17%), accidental falls (þ12%) and suicide tory cause. For example, flu and pneumonia appeared 63% more (þ9%). Taking external causes as a whole, we found it had a strong times on the death certificate in 2020 (table 2). This was particularly educational component in terms of years of life expectancy lost, as a evident when COVID-19 was identified as the UCOD, with flu and decline was only observed among low-educated men. There was not pneumonia being identified as contributing causes on 30% of these an external COD that stood out, but most of the contribution to the death certificates. This suggests that flu and pneumonia were, at least change in life expectancy occurred at working age, which could partially, consequences of COVID-19 (Supplementary table S2). imply that lower-skilled men were more likely to be physically pre- Considerable declines in deaths were also observed for transport sent at work. Cause-specific mortality in Spain in 2020 5 of 7 Table 3 Change in life expectancy between 2018/19 and 2020 according to several main cause of death groupings: COVID-19 as underlying cause of death (UCOD) or contributing cause of death and other causes of death as UCODa Total Age Edu (age 301) 0 1–29 30–44 45–64 65–84 851 Low Medium High Men Total mortality 0.99 0.00 0.02 0.03 0.12 0.65 0.22 1.36 0.96 1.06 COVID19 1.67 0.00 0.00 0.03 0.35 0.92 0.37 1.78 1.60 1.74 Other causes 0.67 0.00 0.02 0.00 0.23 0.28 0.14 0.42 0.64 0.68 Lung cancer 0.08 0.00 0.00 0.00 0.04 0.03 0.00 0.07 0.08 0.11 Remaining Neoplasms 0.16 0.00 0.00 0.01 0.07 0.07 0.01 0.17 0.15 0.16 Downloaded from https://academic.oup.com/eurpub/advance-article/doi/10.1093/eurpub/ckad036/7082922 by guest on 25 April 2023 Diabetes mellitus 0.01 0.00 0.00 0.00 0.00 0.01 0.00 0.02 0.02 0.02 Dementia, Alzheimer 0.03 0.00 0.00 0.00 0.00 0.01 0.02 0.04 0.04 0.03 Hypertensive diseases 0.02 0.00 0.00 0.00 0.01 0.01 0.00 0.03 0.01 0.01 Heart and other diseases of the circulatory system 0.04 0.00 0.00 0.01 0.02 0.05 0.02 0.00 0.05 0.11 Cerebrovascular diseases 0.02 0.00 0.01 0.00 0.01 0.01 0.01 0.01 0.03 0.04 Respiratory system diseases 0.21 0.00 0.01 0.00 0.04 0.09 0.06 0.17 0.20 0.16 Chronic liver diseases 0.02 0.00 0.00 0.00 0.01 0.00 0.00 0.02 0.00 0.00 Other diseases of the digestive syst. 0.03 0.00 0.00 0.00 0.01 0.01 0.01 0.01 0.02 0.03 Other diseases 0.04 0.00 0.01 0.00 0.02 0.01 0.01 0.00 0.05 0.03 Ill-defined causes 0.09 0.00 0.01 0.02 0.05 0.01 0.00 0.08 0.06 0.03 External causes 0.01 0.00 0.01 0.01 0.00 0.01 0.00 0.10 0.01 0.00 Women Total mortality 0.95 0.00 0.02 0.01 0.09 0.52 0.34 1.43 0.66 0.86 COVID-19 1.34 0.00 0.01 0.01 0.17 0.63 0.53 1.60 1.29 1.41 Other causes 0.39 0.00 0.02 0.00 0.08 0.11 0.19 0.17 0.64 0.55 Lung cancer 0.01 0.00 0.00 0.00 0.02 0.01 0.00 0.01 0.01 0.04 Remaining neoplasms 0.07 0.00 0.01 0.00 0.04 0.03 0.01 0.01 0.07 0.12 Diabetes mellitus 0.01 0.00 0.00 0.00 0.00 0.01 0.01 0.03 0.00 0.01 Dementia, Alzheimer 0.02 0.00 0.00 0.00 0.00 0.01 0.03 0.01 0.07 0.05 Hypertensive diseases 0.02 0.00 0.00 0.00 0.00 0.01 0.01 0.03 0.01 0.00 Heart and other diseases of the circulatory system 0.08 0.00 0.00 0.00 0.00 0.04 0.04 0.04 0.14 0.10 Cerebrovascular diseases 0.03 0.00 0.00 0.00 0.00 0.01 0.02 0.00 0.05 0.04 Respiratory system diseases 0.15 0.00 0.00 0.00 0.01 0.05 0.09 0.16 0.19 0.14 Chronic liver diseases 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 Other diseases of the digestive syst. 0.01 0.00 0.00 0.00 0.00 0.01 0.01 0.00 0.02 0.00 Other diseases 0.01 0.00 0.01 0.00 0.00 0.01 0.00 0.04 0.03 0.02 Ill-defined causes 0.02 0.00 0.00 0.00 0.01 0.01 0.00 0.03 0.03 0.04 External causes 0.03 0.00 0.02 0.00 0.00 0.01 0.00 0.02 0.02 0.01 Note: Age- and/or cause-specific contributions may not always add up to the total/education-specific change due to rounding. a: Excludes deaths that had COVID-19 mentioned on the death certificate as a contributing cause. Table 4 Contribution of COVID-19 (as underlying and contributing cause of death) to all-cause life expectancy and lifespan inequality at age 30 by sex; Spain, 2020 Educational groups Inequality gap Changes in inequal- ity by eliminating All Low Medium High COVID-19 (%) Men e30 49.99 48.46 49.91 52.20 3.74 e30 eliminating COVID-19 51.67 50.33 51.67 54.10 3.76 PGLE COVID-19 1.68 1.87 1.76 1.89 0.02 0.5 SD30 12.06 13.18 12.27 10.94 2.24 SD30 eliminating COVID-19 12.44 13.39 12.71 11.33 2.05 Difference 0.38 0.21 0.44 0.40 0.19 8.4 Women e30 55.31 54.00 55.76 56.70 2.71 e30 eliminating COVID-19 56.69 55.65 57.24 58.26 2.61 PGLE COVID-19 1.38 1.65 1.48 1.55 0.10 3.6 SD30 10.83 12.30 10.99 10.31 1.99 SD30 eliminating COVID-19 10.99 12.19 11.19 10.65 1.54 Difference 0.16 0.11 0.21 0.33 0.44 22.3 Note: PGLE, potential gains in life expectancy. Strengths and limitations of using MCOD only detected the most severe cases of COVID-19 deaths during the Our methodological approach using multiple COD allowed us to initial stages of the pandemic.35 Moreover, inconsistent coding, es- capture all COVID-19 mentions in death certificates. This was par- pecially early on, may have led to certain COVID-19 deaths being ticularly necessary in the context of changing definitions, inconsist- coded as the contributing rather than UCOD.15 An MCOD approach ent and incomplete reporting and limited testing availability which also provides additional information regarding other COD involved 6 of 7 European Journal of Public Health in the death process. For instance, given the seasonality of COVID- health and crisis management policies should consider the potential 19, we conducted a sensitivity analysis by repeating the comparative unequal impact of a pandemic on social outcomes (e.g. education, analysis presented in table 2 for the four distinguishable COVID-19 employment) and on the health of the population.20 For instance, we periods in 2020 (table 1). Results showed that respiratory system already know from previous research that hospitalized patients, nur- disease mortality was actually lower during all four periods when it sery home residents and health workers were particularly vulner- was the UCOD, while concurrently any mention of these causes able.40 For future pandemics health impact evaluation tools should increased substantially during the COVID-19 waves. This was up therefore be developed as they permit for the management of the to 172.9% in the case of flu and pneumonia during March–May. pandemic equity to be introduced in the design of interventions, (Supplementary table S3). Interestingly, in the case of flu and pneu- positive and negative impacts of said measures on different popula- monia, as a contributing cause of COVID-19 the proportion tion groups to be predicted and recommendations that advocate remained remarkably the same throughout the year, including dur- equity in health to be issued.20 Downloaded from https://academic.oup.com/eurpub/advance-article/doi/10.1093/eurpub/ckad036/7082922 by guest on 25 April 2023 ing the summer months when there were few COVID-19 deaths (Supplementary table S2). While MCOD data allows to capture more COVID-19-related death Supplementary data than when only relying on UCOD this approach also has its limita- Supplementary data are available at EURPUB online. tions. COVID-19-related death may be overstated if health authorities were classifying COVID-19 deaths as those who tested positive for the virus but had died of another pathology.17 However, in 90% of all Funding death where COVID-19 was mentioned on the death certificate, it This work was supported by the research project “The health of people was considered the UCOD, i.e. the disease or injury that began the in advanced ages: the analysis of comorbidity, multiple causes of chain of events that ultimately led to the person’s death. This is similar death and gender and socioeconomic inequalities in health to what has been reported in other countries.14,15 Nevertheless, an (COMORHEALTHSES)” (IP: JS), funded by the Spanish Ministry of argument could still be made for a possible overestimation of Science and Innovation (PID2020-113934RB-I00). STL also acknowl- COVID-19 deaths as the contribution of COVID-19 was actually edges research funding from the same Ministry (FJC-2019-039314-I higher than the overall one-year drop in life expectancy. This is partly and RYC2021-033123-I) and JS and STL from the European attributed to the fact that deaths coded as underlying or contributing Research Council (ERC-2019-COGagreement No 864616, HEALIN) cause were included here. But also our sensitivity analysis showed that and the Catalan Government under the CERCA programme. when COVID-19 was only included as UCOD, there would still have been a higher reduction in life expectancy than overall (1.5 and Conflicts of interest: None declared. 1.2 years for men and women, respectively) (Supplementary table S4). As this raises the question of whether many COVID-19 deaths were in fact deaths from other causes, further research should delve Data availability deeper into the specific contributing causes associated with COVID- Multiple causes of death data and population exposures by educational 19. attainment were obtained from the Instituto Nacional de Estadıstica The observed discrepancy could also be related to another limita- (INE). These data are not publicly available but interested researchers tion of the MCOD approach, which concerns the assumption of have the possibility to obtain data access by contacting INE (www.ine.es). independent competing risks between COD. Our approach has im- plicitly assumed that eliminating COVID-19 does not alter mortality rates from other COD, which may not necessarily be the case. Although removing only one COD (COVID-19) from the calcula- tions limits this possible bias, if the underlying processes that lead to Key points COVID-19 are not independent from certain respiratory system dis- • In Spain, 15.2% of all deaths in 2020 had COVID-19 diagnosed eases,19 the true effect of COVID-19 on the loss of life expectancy could be slightly lower. Future research should therefore compare or suspected as underlying cause of death on the death estimates of life years lost due to COVID-19 using different methods, certificates, rising to 16.8% when including the COVID-19 including those that deal with competing risks.36 cases coded as contributing cause. • Besides COVID-19, cause-of-death patterns between 2018/19 and 2020 were heterogeneous with very minor increases from Conclusion and policy recommendations some causes (e.g. diabetes, hypertensive diseases) and more Using both UCOD and MCOD data allows researchers to provide a substantial declines from other causes (e.g. cancer, respiratory more holistic view of COD dynamics in a context of the COVID-19 diseases, heart and other diseases of the circulatory system pandemic. This is especially the case regarding old-age mortality due to (women only) and ill-defined causes). the complexity of morbid processes that lead or might contribute to • Compared to 2018/19, life expectancy at birth and age 30 death.13 Although the pandemic only led to a minor increase in the declined by 1.0 year in both men and women. At age 30, mean number of causes per death certificate (table 2), it is well known these declines were higher among lower educated (1.4 years in that people who recover from COVID-19 may develop persistent or both sexes) compared to higher educated (1.1 years in men new symptoms, also referred to as “long COVID”.37 Although long and 0.9 years in women), whereby COVID-19 was COVID does not itself result in death, the risk of death is increased responsible for more than the total decline in life expectancy. during the following months.38 It is therefore essential that we continue • A hypothetical elimination of COVID-19-related mortality in monitoring cause-specific mortality trends closely from an MCOD per- 2020 would have increased lifespan inequality (age at death spective, even after the COVID-19 pandemic has subsided. As the variation), except for low-educated females. Without COVID- UCOD is not necessarily registered and coded as COVID-19,39 we 19, educational inequalities in lifespan variation would therefore recommend analysing monthly mortality data in order to discern have been lower (22% for women and 8% for men). whether the COVID-19 pandemic has led to trend breaks in mortality • Disentangling the role of COVID-19 and other causes of death from different causes, in particular cancer, as cancer patients need con- using multiple cause-of-death approaches aids our tinuous care, and undergo diagnostic tests or therapeutic interventions. understanding of cause-specific mortality during the Finally, given the increase in educational inequalities in lifespan pandemic. variation at age 30 due to COVID-19, especially among women, Cause-specific mortality in Spain in 2020 7 of 7 References 20 Bacigalupe A, Martın U, Franco M, et al. Socioeconomic inequalities and COVID-19 in Spain. SESPAS report 2022. Gac Sanit 2022;36:S13–S21. 1 National Center of Epidemiology. Incidencias acumuladas e indicadores de trans- 21 Bambra C, Riordan R, Ford J, et al. The COVID-19 pandemic and health inequal- misibilidad [Cumulative incidences and transmissibility indicators]. Health Institute ities. J Epidemiol Community Health 2020;74:964–8. Carlos III, 2022. Available at: https://cnecovid.isciii.es/covid19/resources/casos_ hosp_uci_def_sexo_edad_provres.csv (12 May 2022, date last accessed). 22 Niedzwiedz CL, O’Donnell CA, Jani BD, et al. Ethnic and socioeconomic differences 2 Weinberger DM, Chen J, Cohen T, et al. Estimation of excess deaths associated with in SARS-CoV-2 infection: prospective cohort study using UK Biobank. BMC Med the COVID-19 pandemic in the United States, March to May 2020. JAMA Intern 2020;18:1–14. Medi 2020;180:1336–44. 23 Mackenbach JP, Kulhanova I, Bopp M, et al. Variations in the relation between edu- 3 Basellini U, Alburez-Gutierrez D, Del Fava E, et al. Linking excess mortality to cation and cause-specific mortality in 19 European populations: a test of the mobility data during the first wave of COVID-19 in England and Wales. SSM Popul “fundamental causes” theory of social inequalities in health. Soc Sci Med 2015;127:51–62. Health 2021;14:100799. https://doi.org/10.1016/j.ssmph.2021.100799. 24 Krieger N, Williams DR, Moss NE. Measuring social class in US public health Downloaded from https://academic.oup.com/eurpub/advance-article/doi/10.1093/eurpub/ckad036/7082922 by guest on 25 April 2023 4 Modig K, Ahlbom A, Ebeling M. Excess mortality from COVID-19: weekly excess research: concepts, methodologies, and guidelines. Annu Rev Public Health 1997;18: death rates by age and sex for Sweden and its most affected region. Eur J Public 341–78. Health 2021;31:17–22. 25 Permanyer I, Spijker J, Blanes A, et al. Longevity and Lifespan Variation by 5 Aburto JM, Schöley J, Kashnitsky I, et al. Quantifying impacts of the COVID-19 Educational Attainment in Spain: 1960–2015. Demography 2018;55:2045–70. pandemic through life-expectancy losses: a population-level study of 29 countries. 26 Preston SH, Heuveline P, Guillot M. Demography: Measuring and Modeling Int J Epidemiol 2021;51:63–74. Population Processes. Oxford: Wiley-Blackwell, 2000. 6 Trias-Llimós S, Riffe T, Bilal U. Monitoring life expectancy levels during the 27 Horiuchi S, Wilmoth JR, Pletcher SD. A decomposition method based on a model of COVID-19 pandemic: example of the unequal impact of the first wave on Spanish continuous change. Demography 2008;45:785–801. regions. PLoS One 2020;15:e0241952. 28 Riffe T. DemoDecomp: Decompose Demographic Functions [R Package Version 1.0.1], 7 INE. Indicadores de Mortalidad. Resultados nacionales. Esperanza de Vida al 2018. Nacimiento según sexo [Mortality Indicators. National Results. Life Expectancy at 29 Tsai SP, Lee ES, Hardy RJ. The effect of a reduction in leading causes of death: Birth by Sex]. Madrid: INE (National Statistics Institute), 2022. Available at: https:// potential gains in life expectancy. Am J Public Health 1978;68:966–71. www.ine.es/jaxiT3/Datos.htm?t=1414.updated (12 May 2022, date last accessed). 30 Trias-Llimós S, Spijker J. Educational differences in alcohol-related mortality and 8 Islam N, Jdanov DA, Shkolnikov VM, et al. Effects of covid-19 pandemic on life their impact on life expectancy and lifespan variation in Spain (2016–2018): a cross- expectancy and premature mortality in 2020: time series analysis in 37 countries. sectional analysis using multiple causes of death. BMJ Open 2022;0:e053205. BMJ 2021;375:e066768. 31 Cortiula F, Pettke A, Bartoletti M, et al. Managing COVID-19 in the oncology clinic 9 Bonow RO, Fonarow GC, O’Gara PT, et al. Association of coronavirus disease 2019 and avoiding the distraction effect. Ann Oncol 2020;31:553–5. (COVID-19) with myocardial injury and mortality. JAMA Cardiol 2020;5:751–3. 32 Moujaess E, Kourie HR, Ghosn M. Cancer patients and research during COVID-19 10 Onder G, Rezza G, Brusaferro S. Case-fatality rate and characteristics of patients pandemic: a systematic review of current evidence. Crit Rev Oncol Hematol 2020; dying in relation to COVID-19 in Italy. JAMA 2020;323:1775–6. 150:102972. 11 Trias-Llimós S, Permanyer I. A need for better understanding old-age mortality 33 Laliotis I, Stavropoulou C, Ceely G, et al. Excess Deaths in England and Wales during dynamics. Correspondence letter in response to: Bergeron-Boucher M-P, Aburto the First Year of COVID-19. Essen: Global Labor Organization (GLO), 2022. JM, Raalte A van. Diversification in causes of death in low-mortality countries: Available at: http://hdl.handle.net/10419/260493 (12 May 2022, date last accessed). emerging patterns and implications. BMJ Glob Health 2020;5:e002414. 34 Saladié Ò, Bustamante E, Gutiérrez A. COVID-19 lockdown and reduction of traffic 12 Désesquelles A, Demuru E, Egidi V, et al. Cause-specific mortality analysis: is the accidents in Tarragona province, Spain. Transp Res Interdiscip Perspect 2020;8: underlying cause of death sufficient? Revue Quetelet 2014;2:119–35. 100218. 13 Gorina Y, Lentzner HR. Multiple causes of death in old age. Aging Trends 2008;9: 35 Karanikolos M, McKee M. How comparable is COVID-19 mortality across coun- 1–9. tries? Eurohealth 2020;26:45–50. 14 Fineberg HV. The toll of COVID-19. JAMA 2020;324:1502–3. 36 Andersen PK, Canudas-Romo V, Keiding N. Cause-specific measures of life years 15 Nogales Vasconcelos AM, Ishitani L, Abreu DMX, et al. Covid adult mortality in lost. Demogr Res 2013;29:1127–52. Brazil: an analysis of multiple causes of death. Front Public Health 2022;9:788932. 37 Raveendran AV, Jayadevan R, Sashidharan S. Long COVID: an overview. Diabetes 16 Grippo F, Navarra S, Orsi C, et al. The role of COVID-19 in the death of SARS- Metab Syndr Clin Res Rev 2021;15:869–75. CoV-2–positive patients: a study based on death certificates. J Clin Med 2020;9:3459. 38 Shoucri SM, Purpura L, DeLaurentis C, et al. Characterising the long-term clinical 17 Kumar A, Kalra Y, Guan WW, et al. COVID-19 impact on excess deaths of various outcomes of 1190 hospitalised patients with COVID-19 in New York City: a causes in the United States. Ann GIS 2022;28:457–478. retrospective case series. BMJ Open 2021;11:e049488. 18 Redelings MD, Sorvillo F, Simon P. A comparison of underlying cause and multiple 39 Singh B. International comparisons of COVID-19 deaths in the presence of comor- causes of death: US vital statistics, 2000–2001. Epidemiology 2006;17:100–3. bidities require uniform mortality coding guidelines. Int J Epidemiol 2021;50:373–7. 19 Fedeli U, Schievano E, Avossa F, et al. Different approaches to the analysis of causes 40 López-Gay A, Spijker J, Cole HVS, et al. Sociodemographic determinants of intra- of death during the COVID-19 epidemic. Eur Rev Med Pharmacol Sci 2021;25: urban variations in COVID-19 incidence: the case of Barcelona. J Epidemiol 3610–3. Community Health 2021;76:1–7.