Epidemiology and Infection Lockdown measures and relative changes in cambridge.org/hyg the age-specific incidence of SARS-CoV-2 in Spain P.M. De Salazar1 , D. Gómez-Barroso2,3, D. Pampaka2, J.M. Gil4, B. Peñalver2, Original Paper C. Fernández-Escobar2, M. Lipsitch1,5 , A. Larrauri2,3, E. Goldstein1, * *Co-senior authors and M.A. Hernán6, * Cite this article: De Salazar PM et al (2020). Lockdown measures and relative changes in 1 Center for Communicable Disease Dynamics, Department of Epidemiology, Harvard T.H. Chan School of Public the age-specific incidence of SARS-CoV-2 in Health, Boston, MA, USA; 2Centro Nacional de Epidemiología, Instituto de Salud Carlos III, Madrid, Spain; Spain. Epidemiology and Infection 148, e268, 3 Consorcio de Investigación Biomédica en Red de Epidemiología y Salud Pública (CIBERESP), Madrid, Spain; 1–5. https://doi.org/10.1017/ 4 S0950268820002551 Department of Anesthesiology, Santa Creu i Sant Pau Hospital, Barcelona, Spain; 5Department of Immunology and Infectious Diseases, Harvard T.H. Chan School of Public Health, Boston, MA, USA and 6Department of Received: 3 August 2020 Epidemiology and Department of Biostatistics, Harvard T.H. Chan School of Public Health; Harvard-MIT Division of Revised: 8 October 2020 Health Sciences and Technology, Boston, MA, USA Accepted: 13 October 2020 Key words: Abstract Age groups; lockdown; SARS-CoV-2; Spain During the first months of the severe acute respiratory syndrome-coronavirus-2 (SARS-CoV-2) epidemic in 2020, Spain implemented an initial lockdown period on 15 March followed by a Author for correspondence: P.M. De Salazar, E-mail: pablom@hsph. strengthened lockdown period on 30 March when only essential workers continued to com- harvard.edu mute to work. However, little is known about the epidemic dynamics in different age groups during these periods. We used the daily number of coronavirus 2019 cases (by date of symptom onset) reported to the National Epidemiological Surveillance Network among individuals aged 15–19 years through 65–69 years. For each age group g, we computed the proportion PrE(g) of individuals in age group g among all reported cases aged 15–69 years during the pre-lockdown period (1 −10 March 2020) and the corresponding proportion PrL(g) during two lockdown periods (initial: 25 March−3 April; strengthened: 8–17 April 2020). For each lockdown period, we computed the proportion ratios PR(g) = PrL(g)/PrE(g). For each pair of age groups g1, g2, PR(g1)>PR(g2) implies a relative increase in the incidence of detected SARS-CoV-2 infection in the age group g1 compared with g2 for the lockdown period vs. the pre-lockdown period. For the initial lockdown period, the highest PR values were in age groups 50–54 years (PR = 1.21; 95% CI: 1.12,1.30) and 55–59 years (PR = 1.19; 1.11,1.27). For the second lockdown period, the highest PR values were in age groups 15–19 years (PR = 1.26; 0.95,1.68) and 50–54 years (PR = 1.20; 1.09,1.31). Our results suggest that different outbreak control measures led to different changes in the relative incidence by age group. During the initial lockdown period, when non-essential work was allowed, individuals aged 40–64 years, particularly those aged 50–59 years, had a higher relative incidence compared with the pre-lockdown period. Younger adults/older adolescents had an increased relative incidence during the later, strengthened lockdown. The role of dif- ferent age groups during the epidemic should be considered when implementing future miti- gation efforts. Introduction The ongoing severe acute respiratory syndrome-coronavirus-2 (SARS-CoV-2) epidemic led to the implementation of mitigation strategies worldwide. To understand SARS-CoV-2 dynamics under various mitigation strategies, it is important to study the role of age because variations in transmission by age suggest differential impact of various measures such as physical distan- © The Author(s), 2020. Published by cing and workforce-related policies, which in turn has implications for epidemic control. Cambridge University Press. This is an Open The rates of SARS-CoV-2 infection seem to vary by age. Serological studies found the high- Access article, distributed under the terms of est seroprevalence in younger adults and older adolescents in England [1], Switzerland [2], the Creative Commons Attribution licence (http://creativecommons.org/licenses/by/4.0/), Germany [3] and New York State outside of the greater New York City area [4]. A serological which permits unrestricted re-use, study in Spain found higher seroprevalence in older individuals than in younger ones [5]. distribution, and reproduction in any medium, Control measures may have differential effectiveness in different age groups [6], and the provided the original work is properly cited. groups for which the measures are more effective may vary across populations. Social distancing seems to have been less effective among younger adults and older adolescents in Germany [7] and among persons aged 50–59 years in the Netherlands [8]. In Spain, a national lockdown was instituted on 15 March and was further strengthened to include work restrictions to non-essential workers between 30 March and 14 April [9]. This 2 P.M. De Salazar et al. Fig. 1. Cases of COVID-19 (reported by the day of symptom onset) by age group between 1 March and 30 April, 2020 in Spain. Vertical dashed lines demarcate the complete lockdown period (15 March−30 April) and dotted lines the strengthened lockdown period (30 March−14 April). intervention had a pronounced effect on transmission that led to Relative change in SARS-CoV-2 infection by age group decreasing case counts in all regions of the country. However, Daily counts (by date of symptom onset) of reported COVID-19 much less information is available on the relative incidence of dif- cases in individuals aged 15 through 69 years are plotted in ferent age groups during the different stages of the lockdown. For Figure 1. We excluded children under 15 years because of poten- example, age groups that were overrepresented among those who tial temporal changes in diagnosis/ascertainment of cases, and continued non-remote work between 15 and 29 March could have individuals 70 years and older because significant non- had an increased relative incidence during that period. community transmission of infection in long-term care facilities Here, we estimate changes in detected incidence of (LTCFs) likely affected their relative share of cases [12–14]. In SARS-CoV-2 infection by age group after the implementation sensitivity analyses, we restricted the calculations to (a) two of the different lockdown measures. We applied previously devel- regional clusters to explore heterogeneity based on seroprevalence oped methodology [7, 10, 11] to assess the changes in the inci- levels and (b) hospitalised cases to evaluate the impact of potential dence of age groups between 15 and 69 years after control changes in ascertainment, under the assumption that the detec- measures were implemented in Spain. tion of severe cases requiring hospitalisation is relatively insensi- tive to changes in diagnostic criteria. We selected three periods: 1–10 March (five days before the Methods national lockdown, as some social distancing started on 10 March), 25 March−3 April (starting ten days after implementa- Data sources tion of the initial lockdown to detect changes in symptoms On 20 May we obtained information on coronavirus disease 2019 onset) and 8–17 April (starting ten days after implementation (COVID-19) cases reported from 1 March through 30 April to the of the strengthened lockdown). During the initial lockdown, but National Epidemiological Surveillance Network (RENAVE) via not during the strengthened lockdown, non-essential workers the platform SiViEs (see Fig. 1, Results). We retrieved data on were allowed to work when remote work was not possible [9]. reported PCR-confirmed cases with available information on Using the methodology in [7], we computed the age-specific the date of symptoms onset. We excluded healthcare workers proportion ratios for each of the lockdown periods (25 March−3 due to significant non-community transmission among them. April or 8–17 April) relative to the pre-lockdown period (1–10 Epidemiology and Infection 3 Table 1. Number of reported COVID-19 cases with available information on the date of symptom onset in different age groups for different time periods Period (years) 15–19 20–24 25–29 30–34 35–39 40–44 45–49 50–54 55–59 60–64 65–69 Pre-lockdown 113 250 399 482 646 795 942 970 1065 1200 1342 1−10 March Initial lockdown 139 346 545 674 1004 1409 1783 2032 2200 2177 1882 25 March–3 April Strengthened lockdown 8−17 April 76 135 215 278 330 441 506 618 657 593 522 Case counts increased a great deal with age, likely reflecting higher case ascertainment for older individuals. Fig. 2. Proportion ratio estimates of confirmed COVID-19 cases by age group in Spain for the period 25 March–3 April vs. 1−10 March. March) as follows: for each age group g, let E(g) be the number of younger ones does not necessarily reflect differences in incidence COVID-19 cases in age group g and h E(h) the total number of because infections are more severe in older individuals, and the cases in age groups h = 1 (15–19 years) to h = 11 (65–69 years) likelihood of reporting of infection is higher for severe infections. during  the pre-lock down period. Similarly, let L(g) and Figure 1 plots the epidemic curves of daily COVID-19 cases by h L(h) be the corresponding numbers during the initial lock- age group (15–69 years) between 1 March and 30 April, 2020 down. The proportion ratio (PR) statistic in age group g is (n = 73 650). L(g) E(g) PR(g) =  / After the initial lockdown period h L(h) h E(h) Figure 2 plots the PR estimates for the initial lockdown period that is, the ratio between the proportion of cases in age group g in compared with the pre-lockdown period. The highest estimates the initial lockdown period and the proportion of cases in age correspond to individuals aged 50–54 years (PR = 1.21; 95% CI group g in the pre-lockdown period. We repeated these calcula- 1.12,1.30) and 55–59 years (PR = 1.19; 1.11,1.27). PR estimates tions after replacing the cases in the initial lockdown period by for individuals aged 15–44 years and 60–69 years were signifi- the cases in the strengthened lockdown period. cantly lower than the estimates for those aged 50–59 years In sensitivity analyses, we computed the age-group PR statistic (Supplementary Material, online Supplementary Table S2). (a) disaggregating the cases in two groups depending on regional seroprevalence estimates [15] and (b) including only hospitalised After the strengthened lockdown period cases, which would be less affected by potential changes in ascer- tainment given higher severity. The Supplementary Material Figure 3 plots the PR estimates for the strengthened lockdown describes the methods to obtain confidence bounds for the PR period vs. the pre-lockdown period. The highest estimates corres- statistic [16]. pond to individuals aged 15–19 years (PR = 1.26; 0.95,1.68), 50–54 years (PR = 1.20; 1.09,1.31) and 55–59 years (PR = 1.16; 1.06,1.27). PR estimates for individuals aged 50–54 years were Results significantly higher than the estimates for those aged 35–49 years Table 1 classifies the confirmed cases by age group and period. and 60–69 years (Supplementary Material, online Supplementary The higher number of cases in older individuals compared with Table S3). 4 P.M. De Salazar et al. Fig. 3. Proportion ratio estimates of confirmed COVID-19 cases by age group in Spain for the period 8−17 April vs. 1−10 March. A comparison of Figures 2 and 3 suggests an increase in the [8], which is in agreement with the elevated relative incidence proportion of COVID-19 cases for the strengthened lockdown in persons (excluding healthcare workers) aged 50–59 years period compared with the initial lockdown period in younger throughout the lockdown period in Spain. Our findings for age groups (up to 34 years) relative to the middle ones (35– the strengthened lockdown period in Spain echo those from 64 years). other countries: a higher relative incidence of infection in In sensitivity analyses, PR estimates did not materially change younger persons was found in England [1], Switzerland [2] across regions with different seroprevalence or when we restricted and Germany (Figure 6A in [3]), and the highest proportion the analysis to hospitalised cases (Supplementary Material, ratio estimates for the post lockdown period in Germany are Sections S1–S2). in younger persons [7]. Our findings could be affected by age-differential changes in case ascertainment, over time or across regions. However, there Discussion is no evidence of fundamental diagnostic changes for the adult We applied the methodology described in [7, 10, 11] to show that population during the lockdown period in Spain, where the the relative incidence of detected SARS-CoV-2 infections in differ- focus was on testing the more severe cases. Moreover, our sensi- ent age groups changed during the lockdown periods in Spain. tivity analysis restricted to hospitalised cases, which are less likely Compared with the pre-lockdown period, individuals aged 40– affected by changes in ascertainment, as well as region-specific 64 years, and particularly those aged 50–59 years, had an elevated analyses (with splitting into regions reflecting places with higher relative incidence during the first lockdown period. The corre- and lower-medium seroprevalence) yielded estimates that were sponding relative incidence of younger adults/older adolescents, consistent with those of the main analysis. as well as persons aged 50–59 years, increased during the strength- In summary, our paper provides evidence for an elevated rela- ened lockdown period compared with the pre-lockdown period. tive incidence of individuals aged 40–64 years during the initial These differences by age group might be explained by a num- lockdown period, when non-essential work was allowed, and for ber of factors. First, the elevated relative incidence in middle-aged an elevated relative incidence of younger adults/older adolescents adults during the first lockdown period, when non-essential when only essential workers continued to work. These results sug- workers were allowed to work, is consistent with the higher gest that age structure is an important factor in the effect of lock- employment rates in Spain in middle-aged adults compared down interventions. with younger adults [17]. Second, adherence to social distancing Supplementary material. The supplementary material for this article can may vary with age: relative increases for younger adults/older ado- be found at https://doi.org/10.1017/S0950268820002551 lescents during the second lockdown period might reflect lower adherence to lockdown measures, whereas perception for risk of Data availability. The data that support the findings of the study are avail- severe disease could have led to stronger individual adherence able as aggregates in Table 1 of the main text (and at the end of the supple- for persons aged 60–69 years. Finally, household transmission mentary material, online Supplementary Table S1 for sensitivity analysis). in the context of the high proportion of multigenerational house- Line list data are available upon request from the National Center of holds in Spain may also have contributed to the observed patterns. Epidemiology, Carlos III Institute of Health, Madrid, Spain. Further work is needed to understand those issues to better Acknowledgements. We thank all the contributors to the surveillance and inform future mitigation efforts. investigation of COVID-19 epidemic in Spain, including colleagues at local Our results are aligned with findings in other European and regional Public Health Services, as well as the Centre of Health Alerts countries. Persons aged 50–59 years were least impacted in and Emergencies at the Ministry of Health, the National Centre of terms of mixing/number of contacts with people following the Epidemiology and the National Centre of Microbiology at the Carlos III introduction of social distancing measures in the Netherlands Institute of Health. Epidemiology and Infection 5 Financial support. P.M.D. was supported by the Fellowship Foundation transmissibles = European communicable disease bulletin 2020; 25Published Ramon Areces. Marc Lipsitch and Edward Goldstein were supported by online: April 2020. doi:10.2807/1560-7917.ES.2020. 25.17.2000596. Award Number U54GM088558 from the National Institute of General 8. Backer JA et al. The impact of physical distancing measures against Medical Sciences and Lipsitch also by the Morris-Singer Fund. The content COVID-19 transmission on contacts and mixing patterns in the is solely the responsibility of the authors. Netherlands: repeated cross-sectional surveys. 9. BOE. BOLETÍN OFICIAL DEL ESTADO Num. 87. 2020(https://www.boe. Conflict of interest. None es/boe/dias/2020/03/29/pdfs/BOE-A-2020-4166.pdf). Accessed 11 June 2020. 10. Goldstein E et al. (2017) Temporally varying relative risks for infectious diseases: implications for infectious disease control. Epidemiology References (Cambridge, Mass.) 28, 136–144. 1. Ward H et al. Antibody prevalence for SARS-CoV-2 in England following 11. Goldstein E et al. (2018) On the relative role of different age groups dur- first peak of the pandemic: REACT2 study in 100,000 adults. medRxiv ing epidemics associated with respiratory syncytial virus. The Journal of Cold Spring Harbor Laboratory Press, 2020: 2020.08.12.20173690. Infectious Diseases 217, 238–244. 2. Stringhini S et al. (2020) Seroprevalence of anti-SARS-CoV-2 IgG anti- 12. Statement – Invest in the overlooked and unsung: build sustainable bodies in Geneva, Switzerland (SEROCoV-POP): a population-based people-centred long-term care in the wake of COVID-19. World Health study. The Lancet 396, 313–319. Organization, 2020; Published online: 11 June 2020. 3. Streeck H et al. Infection fatality rate of SARS-CoV-2 infection in a 13. Resources to support community and institutional long-term care German community with a super-spreading event. responses to COVID-19. Mortality associated with COVID-19 outbreaks 4. Rosenberg ES et al. Cumulative incidence and diagnosis of SARS-CoV-2 in care homes: early international evidence. 2020(https://ltccovid.org/ infection in New York. 2020/04/12/mortality-associated-with-covid-19-outbreaks-in-care-homes- 5. Pollán M et al. (2020) Prevalence of SARS-CoV-2 in Spain early-international-evidence/). Accessed 22 June 2020. (ENE-COVID): a nationwide, population-based seroepidemiological 14. Radiografía del coronavirus en residencias de ancianos en España. study. The Lancet 396, 535–544. RTVE. 2020; Published online: 21 June 2020. 6. Zhang J et al. (2020) Changes in contact patterns shape the dynamics 15. Murtagh F and Legendre P (2014) Ward’s Hierarchical Agglomerative of the COVID-19 outbreak in China. Science (New York, N.Y.) 368, Clustering Method: which Algorithms Implement Ward’s Criterion? 1481–1486. Journal of Classification 31, 274–295. 7. Goldstein E and Lipsitch M. Temporal rise in the proportion of younger 16. Altman DG (1990) Practical Statistics for Medical Research. London: adults and older adolescents among coronavirus disease (COVID-19) cases Chapman and Hall/CRC. following the introduction of physical distancing measures, Germany, 17. INE. INE. Instituto Nacional de Estadística. (https://www.ine.es). Accessed March to April 2020. Euro surveillance: bulletin Europeen sur les maladies 11 June 2020.