BMJ Open: first published as 10.1136/bmjopen-2021-050551 on 24 August 2021. Downloaded from http://bmjopen.bmj.com/ on August 24, 2021 at Sectoral & IT Resource CTR World Bank Open access Original research How many infants may have died in low-­income and middle-­income countries in 2020 due to the economic contraction accompanying the COVID-19 pandemic? Mortality projections based on forecasted declines in economic growth Gil Shapira  ‍ ‍, Damien de Walque  ‍ ‍, Jed Friedman  ‍ ‍ To cite: Shapira G, de Walque D, ABSTRACT Friedman J. How many infants Strengths and limitations of this study Objectives  While COVID-19 has a relatively small direct may have died in low-­ income impact on infant mortality, the pandemic is expected to MC-C3-220. Protected by copyright. and middle-­ income countries ►► Our study links gross domestic product (GDP) per indirectly increase mortality of this vulnerable group in in 2020 due to the economic capita data to an especially large dataset of 5.2 low-­income and middle-­ income countries through its contraction accompanying million retrospective birth histories reported in the COVID-19 pandemic? effects on the economy and health system performance. Demographic and Health Surveys conducted in Mortality projections based Previous studies projected indirect mortality by modelling many low-­ income and middle-­ income countries be- on forecasted declines in how hypothesised disruptions in health services will affect tween 1985 and 2018. economic growth. BMJ Open health outcomes. We provide alternative projections, ►► While previous projections of indirect COVID-19 2021;11:e050551. doi:10.1136/ relying on modelling the relationship between aggregate bmjopen-2021-050551 mortality have been based on assumptions regard- income shocks and mortality. ing the magnitude of health service disruptions, ►► Prepublication history and Design  We construct a sample of 5.2 million births by our estimates account for additional mechanisms, additional supplemental material pooling retrospective birth histories reported by women mainly increased household poverty. for this paper are available in Demographic and Health Surveys conducted in 83 ►► Our estimates may represent a lower bound of the online. To view these files, low-­income and middle-­ income countries between 1985 actual excess mortality if the current economic please visit the journal online. and 2018. We employ regression models with country-­ (http://​dx.​doi.o ​ rg/​10.​1136/​ downturn is accompanied by larger disruptions to specific fixed-­effects and flexible time trends to estimate bmjopen-​2021-​050551). the provision of essential health services relative to the impact of gross domestic product per capita on infant previous downturns. mortality rate. We then use growth projections by the Received 22 February 2021 ►► We estimate the short-­ term impact of GDP fluctu- International Monetary Fund to predict the effect of the Accepted 02 August 2021 ations on mortality while long-­ term implications economic downturn in 2020 on infant mortality. for mortality and other adverse outcomes may also Results  We estimate 267 208 (95% CI 112 000 to 422 arise. 415) excess infant deaths in 128 countries, corresponding ►► The analysis ascribes the difference between to a 6.8% (95% CI 2.8% to 10.7%) increase in the total October 2019 and October 2020 economic growth number of infant deaths expected in 2020. projections for 2020 solely to the pandemic, even Conclusions  The findings underscore the vulnerability though some countries have also experienced other of infants to the negative income shocks such as those shocks, such as natural disasters or political crises imposed by the COVID-19 pandemic. While efforts towards that may affect national income levels. prevention and treatment of COVID-19 remain paramount, © Author(s) (or their the global community should also strengthen social safety employer(s)) 2021. Re-­use permitted under CC BY-­ NC. No nets and assure continuity of essential health services. may be substantial. Health and social poli- commercial re-­ use. See rights and permissions. Published by cies should not lose sight of excess indirect BMJ. INTRODUCTION mortality caused by such factors as the inter- Development Research Group, Reducing morality risk in the wake of the ruption of essential health services and the World Bank, Washington, COVID-19 pandemic is a paramount public general economic downturn brought on by Columbia, USA concern. While direct mortality risk as a the pandemic. This study attempts to quan- Correspondence to result of COVID-19 infection has garnered tify the expected indirect mortality over the Dr Gil Shapira; the majority of attention in global media pandemic period for one especially vulner- ​ rg ​gshapira@​worldbank.o and policy discussions, indirect mortality able subpopulation—infants—by modelling Shapira G, et al. BMJ Open 2021;11:e050551. doi:10.1136/bmjopen-2021-050551 1 BMJ Open: first published as 10.1136/bmjopen-2021-050551 on 24 August 2021. Downloaded from http://bmjopen.bmj.com/ on August 24, 2021 at Sectoral & IT Resource CTR World Bank Open access the impact of projected economic decline on the likeli- middle-­income countries between 1985 and 2018. The hood of infant survival. surveys used in the analysis are listed in online supple- Studies from diverse settings find negligible direct mental appendix table A1. The combined sample totals mortality rates for children and infants due to COVID- 5.2 million births, of which 27% and 55% are from low-­ 19.1 However, stringent containment measures and income and lower-­ income countries. Over the full middle-­ the ensuing economic downturn, as well as the need to period of analysis, the sample’s infant mortality rate per reallocate health system resources towards pandemic income, lower-­ 1000 births is 85, 61 and 37 for low-­ middle-­ response, have influenced many social determinants of income and upper-­ middle-­ income countries, respectively. mortality such as the ability to afford nutritious foods and We estimate the relation between aggregate income to access essential healthcare. Unlike economic crises in change and infant mortality with the following framework: high-­income countries, which appear to lower mortality,2 ‍ Dict = αc + β logGDPct + γ1c tct + γ2c t2 3 ct + γ3c tct + εict ‍ economic crises in low-­ income countries generally increase mortality among vulnerable groups, namely is a binary indicator taking the value 1 if child ‍ ‍Dict ‍ i ‍ in young children and the elderly. Earlier studies have docu- mented a robust relationship between short-­ term fluctua- country ‍ c‍died in the first 12 months of life during year ‍ t‍. tion in aggregate income and all-­ cause infant mortality in logGDP ‍ ‍is the natural logarithm of per capita GDP and ‍ ε iact ‍ low-­income and middle-­ income countries.3–8 is the error term. The ‍α‍ and γ coefficients identify country-­ At the very start of the COVID-19 pandemic, model- specific fixed effects and a country-­ specific cubic time ling exercises predicted that the interruption of essen- trend, respectively. SEs are clustered at the country level; β ‍is the coefficient of interest, describing the relationship ‍ tial health services will be severe9–1412 and perhaps the world will experience 250  000 to 1.15  million young between aggregate income shocks and infant mortality. child deaths15 in the first 6  months of the pandemic. We estimate this semi-­ elasticity of infant mortality to Recent studies indicate that barriers to access essential aggregate income decline separately by country income healthcare in low-­ income and middle-­ income countries level, as classified by the World Bank 2020 income groups, MC-C3-220. Protected by copyright. are not just a theoretical concern documenting, for as well as overall. Low-­ income economies are defined by example, disruptions in immunisation services in Paki- a gross national income (GNI) per capita of