Global minimum estimates of children affected by COVID-19-associated orphanhood and deaths of caregivers: a modelling study

Background: The COVID-19 pandemic priorities have focused on prevention, detection, and response. Beyond morbidity and mortality, pandemics carry secondary impacts, such as children orphaned or bereft of their caregivers. Such children often face adverse consequences, including poverty, abuse, and institutionalisation. We provide estimates for the magnitude of this problem resulting from COVID-19 and describe the need for resource allocation. Methods: We used mortality and fertility data to model minimum estimates and rates of COVID-19-associated deaths of primary or secondary caregivers for children younger than 18 years in 21 countries. We considered parents and custodial grandparents as primary caregivers, and co-residing grandparents or older kin (aged 60–84 years) as secondary caregivers. To avoid overcounting, we adjusted for possible clustering of deaths using an estimated secondary attack rate and age-specific infection–fatality ratios for SARS-CoV-2. We used these estimates to model global extrapolations for the number of children who have experienced COVID-19-associated deaths of primary and secondary caregivers. Findings: Globally, from March 1, 2020, to April 30, 2021, we estimate 1 134 000 children (95% credible interval 884 000–1 185 000) experienced the death of primary caregivers, including at least one parent or custodial grandparent. 1 562 000 children (1 299 000–1 683 000) experienced the death of at least one primary or secondary caregiver. Countries in our study set with primary caregiver death rates of at least one per 1000 children included Peru (10·2 per 1000 children), South Africa (5·1), Mexico (3·5), Brazil (2·4), Colombia (2·3), Iran (1·7), the USA (1·5), Argentina (1·1), and Russia (1·0). Numbers of children orphaned exceeded numbers of deaths among those aged 15–50 years. Between two and five times more children had deceased fathers than deceased mothers. Interpretation: Orphanhood and caregiver deaths are a hidden pandemic resulting from COVID-19-associated deaths. Accelerating equitable vaccine delivery is key to prevention. Psychosocial and economic support can help families to nurture children bereft of caregivers and help to ensure that institutionalisation is avoided. These data show the need for an additional pillar of our response: prevent, detect, respond, and care for children. Funding: UK Research and Innovation (Global Challenges Research Fund, Engineering and Physical Sciences Research Council, Medical Research Council), UK National Institute for Health Research, US National Institutes of Health, and Imperial College London.

We used methods similar to Lotka and colleagues22 and those used by the UNAIDS Reference Group for estimating AIDS orphans5, 23 to estimate COVID-19-associated orphanhood among children younger than 18 years (appendix pp 2–5). We extended these to incorporate deaths of grandparents aged 60–84 years who lived with their grandchildren. We assumed co-residing grandparents helped to provide some type of relational, practical, or financial caregiving for grandchildren. Key aspects of such care include face-to-face contact or psychosocial support, caregiving behaviours (eg, feeding, teaching, or supervision), and financial support for household and educational expenses.16 We developed estimates of pandemic-associated orphanhood and caregiver deaths using excess mortality and COVID-19 deaths for 21 countries that accounted for 76·4% of global COVID-19 deaths up to April 30, 2021 (Argentina, Brazil, Colombia, England and Wales, France, Germany, India, Iran, Italy, Kenya, Malawi, Mexico, Nigeria, Peru, Philippines, Poland, Russia, South Africa, Spain, the USA, and Zimbabwe). We based calculations of orphanhood and caregiver deaths on age-and-sex-stratified excess death data when available, because reported counts of confirmed COVID-19 deaths underestimate pandemic-associated deaths.3 We further examined effects of age and sex variations in mortality on orphanhood and loss of caregivers. For countries without disaggregated excess mortality data, we adjusted the total number of children experiencing COVID-19-associated deaths of parents or caregivers using data on differences between COVID-19 deaths and excess deaths (see appendix pp 7–60 for methodological details). As we used aggregate country-level information and estimates from statistical models, no individuals are studied. All analyses were done using R (version 4.0.2). We extracted available excess deaths and COVID-19 deaths from March 1, 2020, to April 30, 2021, using 5-year age bands or the level of disaggregation provided. For countries reporting COVID-19 and excess deaths, we used the larger of these two values in each age band to calculate the number of orphans, because we are interested in orphans associated with the pandemic as a whole. In this Article, we use the term COVID-19-associated deaths to refer to the combination of deaths caused directly by COVID-19 and those caused indirectly by other associated causes, such as lockdowns, restrictions on gatherings and movement, and decreased access or acceptability of health care and of treatment for chronic diseases, which are reported in the excess deaths. If excess deaths were not routinely reported for a given country, we calculated them by subtracting the monthly deaths in 2020–21 from the monthly average between 2015 and 2019. For Russia, where age-stratified and sex-stratified COVID-19 deaths and excess deaths were unavailable, we disaggregated excess deaths using published age-specific COVID-19 infection–fatality ratios estimates (appendix pp 2, 48–50). Where appropriate, we adjusted our deaths by the excess-to-COVID-19-deaths ratio (appendix pp 7–60). In light of the rise in COVID-19-associated deaths in India since February, 2021, we further used estimates of COVID-19-associated deaths to illustrate the impact of such a crisis on increases in orphanhood and death of caregivers. To estimate numbers of children orphaned as a result of the pandemic, we needed female and male fertility rates at the same disaggregation level as deaths (5-year age bands) for the years in which children younger than 18 years were born (2003–20). We assumed fertility in 2021 was the same as in 2020. In England and Wales, we used country-specific data available for both male and female fertility. For countries with Demographic and Health Survey data, we used the own-child method to calculate male and female fertility using the same source (appendix pp 2–3). For all other countries, we used the UN World Prospects female fertility rates and calculated male fertility rates using the UN Statistics Division data on men’s fertility and fatherhood, alongside population estimates (appendix pp 2–3). We calculated the average number of children that each adult of a given age would have in 2020 by summing the average number of children born to a man or woman over each of the past 17 years at the age the adult would have been in each year, and adjusting for child mortality where necessary (appendix p 4). We assumed a fertility rate of zero for women older than 50 years but used data for men up to age 80 years. We then multiplied the average number of children for each 5-year age band by the number of male and female deaths in corresponding parental age bands to calculate the number of children losing a mother (maternal orphans) or father (paternal orphan). We adjusted for possible clustering of deaths between parents using an estimate of secondary attack rates and age-specific infection–fatality ratios to provide unduplicated counts for children losing one parent (single orphan) or both parents (double orphan; appendix p 4).24 It was not possible to consider families with two parents of the same gender because of lack of available data for the proportion of same gender parents in every country. Due to the lack of globally consistent data on orphanhood, estimates of pre-existing single orphans were not available, so our estimates of double orphans only capture those whose parents both died during the pandemic. We report the ratio of orphanhood to age-specific COVID-19-associated deaths. We calculated this ratio by dividing our estimate of the number of children orphaned by parental age group in broad age categories (which varied sightly due to country-specific differences in reporting) by the number of COVID-19-associated deaths in the age category. A ratio larger than 1 suggests a larger family size, such that one parental death can lead to multiple children orphaned. While the ratios are similar to age-specific fertility rates, they take into account the age pattern of COVID-19 mortality. We further calculated rates of orphanhood and caregiver loss per 1000 people using International Census Data. Rates are provided for combined categories, not individual categories, so as to provide estimates of minimum total numbers of children affected. To estimate COVID-19-associated deaths in co-residing grandparents, we used two UN Population Division measures of household composition: custodial grandparents and other co-residing grandparents. Custodial grandparents are skip-generation grandparents, defined as grandparents aged 60–84 years who live with their grandchildren in absence of parents. Other co-residing grandparents (or kin) are grandparents aged 60–84 years (or other co-residing kin aged 60–84 years, such as aunts or uncles) who live in multigenerational households with at least one family member younger than 18 years, along with at least one of their parents (appendix pp 4–5).20 We truncated deaths of grandparents at 85 years, since we were aiming for a conservative estimate, and a large proportion of deaths in people older than 85 years in Europe and the USA were in care homes, which are excluded from the household composition data. We again adjusted these numbers using the estimated secondary attack rate and age-specific infection–fatality ratios to avoid overcounting the estimated 1·03% of children who lost parents and grandparents (appendix pp 4–5). We limited loss of grandparents or older kin to a maximum of two per child (one male and one female), owing to limitations in global household composition data. However, low secondary household attack rates means the number of children experiencing further co-residing caregiver losses is likely to be negligible. We assumed that extended family members aged 60 years or older who lived with family members younger than 18 years most likely represented grandparents and grandchildren, although the older adult might be an aunt, uncle, or cousin. A systemic review addressing grandparents co-residing with grandchildren reports that in multigenerational families, grandparents and parents provide care for the children together, through involvement or resources.16 Because the death of a parent or abrupt death of a close family member is the most frequently cited type of trauma exposure experienced by children, we considered the abrupt death of either co-residing grandparents or other co-residing extended family aged 60 years or older to represent a substantial loss for the affected child.25 From a public health perspective, the inclusion of these older adult family members is crucially important, as their prioritisation for vaccines renders their premature deaths highly preventable. We used data from the 21 countries to develop global extrapolations for the impact of COVID-19-associated deaths on the numbers of children orphaned due to deaths of parents, losing primary caregivers (parents or custodial grandparents), and losing primary or secondary caregivers (parents, custodial grandparents, or co-residing grandparents or kin; figure 1). We based our estimates on the larger of either excess deaths or COVID-19 deaths for 12 countries with data available, and on COVID-19 deaths for nine countries with unavailable excess death data. To extrapolate beyond these 21 countries, we relied on the high correlation between total fertility rate (TFR) and the ratio of orphans to deaths (Pearson r2=0·93) and fit a logistic model using least squares to estimate the two logistic parameters and gamma, a scaling parameter. We obtained COVID-19 deaths from each country from Johns Hopkins University and TFRs from the UN Population Division World Prospects data (appendix p 5). We considered uncertainty from the TFR in our global estimates by assuming our TFR was normally distributed with the medium fertility variant estimate for 2020–25 as our mean, and estimating the SD from the low and high variants given. We then calculated the global numbers of orphans by sampling the TFR for each country 1000 times and using our previously fitted logistic model. Our central estimates include country-specific estimates from our study, but 95% credible intervals (CrIs) are based solely on the samples. Classification of deaths of parents, custodial (skip-generation) grandparents, and other co-residing grandparents or older kin *Grandparents or other older kin (≥60 years of age) co-residing with family members younger than 18 years. We completed a leave-one-out sensitivity analysis to show how our central estimates of total number of children experiencing death of parents and caregivers varied if we fit the models to our data leaving out one country each time (appendix p 5). A second sensitivity analysis also considered whether our findings differed when we used TFRs from the Institute for Health Metrics and Evaluation (IHME) instead of UN Population Division TFRs. The funders of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report.

• hillis, susan d.
• unwin, h. juliette t.
• chen, yu
• cluver, lucie
• sherr, lorraine
• goldman, philip s.
• ratmann, oliver
• donnelly, christl a.
• bhatt, samir
• villaveces, andrés
• butchart, alexander
• bachman, gretchen
• rawlings, laura b.
• green, phil
• nelson, charles a.
• flaxman, seth r.