Global Excess Mortality during COVID-19 Pandemic: A Systematic Review and Meta-Analysis

Background: Currently, reported COVID-19 deaths are inadequate to assess the impact of the pandemic on global excess mortality. All-cause excess mortality is a WHO-recommended index for assessing the death burden of COVID-19. However, the global excess mortality assessed by this index remains unclear. We aimed to assess the global excess mortality during the COVID-19 pandemic. Methods: We searched PubMed, EMBASE, and Web of Science for studies published in English between 1 January 2020, and 21 May 2022. Cross-sectional and cohort studies that reported data about excess mortality during the pandemic were included. Two researchers independently searched the published studies, extracted data, and assessed quality. The Mantel–Haenszel random-effects method was adopted to estimate pooled risk difference (RD) and their 95% confidence intervals (CIs). Results: A total of 79 countries from twenty studies were included. During the COVID-19 pandemic, of 2,228,109,318 individuals, 17,974,051 all-cause deaths were reported, and 15,498,145 deaths were expected. The pooled global excess mortality was 104.84 (95% CI 85.56–124.13) per 100,000. South America had the highest pooled excess mortality [134.02 (95% CI: 68.24–199.80) per 100,000], while Oceania had the lowest [−32.15 (95% CI: −60.53–−3.77) per 100,000]. Developing countries had higher excess mortality [135.80 (95% CI: 107.83–163.76) per 100,000] than developed countries [68.08 (95% CI: 42.61–93.55) per 100,000]. Lower middle-income countries [133.45 (95% CI: 75.10–191.81) per 100,000] and upper-middle-income countries [149.88 (110.35–189.38) per 100,000] had higher excess mortality than high-income countries [75.54 (95% CI: 53.44–97.64) per 100,000]. Males had higher excess mortality [130.10 (95% CI: 94.15–166.05) per 100,000] than females [102.16 (95% CI: 85.76–118.56) per 100,000]. The population aged ≥ 60 years had the highest excess mortality [781.74 (95% CI: 626.24–937.24) per 100,000]. Conclusions: The pooled global excess mortality was 104.84 deaths per 100,000, and the number of reported all-cause deaths was higher than expected deaths during the global COVID-19 pandemic. In South America, developing and middle-income countries, male populations, and individuals aged ≥ 60 years had a heavier excess mortality burden.


Introduction
As of 21 May 2022, more than 6.29 million people worldwide died from COVID-19 infection [1]. Global countries have adopted a series of public health measures to curb the coronavirus disease 2019 (COVID-19) pandemic, such as strict lockdown policies and wearing masks [2,3]. These measures cut off the main transmission routes of respiratory infectious diseases such as COVID-19 and influenza and lower their prevalence and mortality [2,3]. Notably, despite the reduction of accidental traffic deaths due to strict lockdown policies [4][5][6][7], these policies may increase the deaths of patients with chronic diseases because they have difficulty receiving timely health care [8][9][10]. Additionally, limited medical resources increase the risk of death among patients with chronic diseases [11]. The deaths from mental depression, suicide, and violence also increased during the pandemic [4,12,13]. Therefore, the COVID-19 pandemic is threatening global health resources and economic and political development.
For more than one study in a country, we selected the study that covered the largest population, spanned the longest time period, and performed analyses in different age and sex groups.

Data Extraction
The authors WS and WY independently screened the titles and abstracts, and excluded studies that did not meet the inclusion criteria. Discrepancies were resolved by discussion with the main author (JY). The following data were extracted independently by two authors (WS and WY) from the included studies: first author, publication year, country, study design, the number of reported all-cause deaths, the number of expected all-cause deaths, the number of population, pandemic time and time used to estimate expected deaths. If available, we also extracted the data on the sex and age of reported all-cause deaths, expected all-cause deaths, and the population.

Risk of Bias Assessment
WS and YW independently assessed the risk of bias for each study, which was cross-checked by ZG and JY. Cross-sectional studies were assessed by the Agency for Healthcare Research and Quality (AHRQ) [30] and cohort studies were assessed by the Newcastle-Ottawa scale (Table S1) [31]. Reviewers rated each domain for overall risk of bias as low, moderate, high, or serious/critical (Table S2).

Data Synthesis and Statistical Analysis
We performed a meta-analysis of global excess mortality during the pandemic, and we reported the pooled risk difference (RD) as excess mortality. The Mantel-Haenszel random-effects method [32] was adopted to estimate the pooled risk difference and their 95% confidence intervals (CIs). The lower limit of 95% CI > 0 indicated that the number of reported all-cause deaths was higher than that of expected deaths; the upper limit of 95% CI < 0 indicated that the number of reported all-cause deaths was lower than that of expected deaths; that the 95% CI included 0 suggested no significant difference between reported and expected deaths. The heterogeneity among the studies was estimated using I 2 values. Very low, low, moderate, and high degrees of heterogeneity were defined as I 2 ≤ 25%, 25% to ≤50%, 50% to ≤75%, and ≥75%, respectively [33].
We performed subgroup analyses in continents (Asia vs. Africa vs. Europe vs. North America vs. South America vs. Oceania), country development levels (developing country vs. developed country), World Bank income levels (lower middle-income country vs. upper middle-income country vs. high-income country), age groups (<40 years vs. 40-60 years vs. ≥60 years) and sex (male vs. female). We performed sensitivity analyses by excluding countries with populations less than 1 million. All analyses were performed using Stata software (version 12.0; Stata SE Corporation LP, College Station, TX, USA). A two-sided p value < 0.05 was considered statistically significant [34].

Publication Bias and Sensitivity Analysis
We did not analyze publication bias because the study subject in our study was the country, not the originally published studies. After excluding eight countries with a study population of less than 1 million, the pooled excess mortality

Publication Bias and Sensitivity Analysis
We did not analyze publication bias because the study subject in our study was the country, not the originally published studies. After excluding eight countries with a study population of less than 1 million, the pooled excess mortality [106.99 deaths (95% CI: 86.71-127.27) per 100,000] was similar to the original result. (Figure S1)

Discussion
In this systematic review and meta-analysis, we found that the pooled excess mortality was 104.84 deaths (95% CI: 85.56-124.13) per 100,000. We found that the excess mortality was higher in South America, North America, Europe, developing countries, loweror upper-middle-income countries, the male population, and the population aged ≥60 years.
To the best of our knowledge, the current study is the first systematic review to evaluate all-cause excess mortality during the pandemic. In this study, the pooled excess mortality was 104.84 per 100,000 globally. Our results are consistent with previous findings

Discussion
In this systematic review and meta-analysis, we found that the pooled excess mortality was 104.84 deaths (95% CI: 85.56-124.13) per 100,000. We found that the excess mortality was higher in South America, North America, Europe, developing countries, lower-or upper-middle-income countries, the male population, and the population aged ≥ 60 years.
To the best of our knowledge, the current study is the first systematic review to evaluate all-cause excess mortality during the pandemic. In this study, the pooled excess mortality was 104.84 per 100,000 globally. Our results are consistent with previous findings (120.30 per 100,000) from COVID-19 Excess Mortality Collaborators, although their findings were derived from a model estimation covering 187 countries/regions [14]. Previous literature reported all-cause excess mortality from January to August 2020 for 22 countries but did not calculate the pooled excess mortality [15]. Similarly, another study calculated all-cause excess mortality during the SARS-CoV-2 pandemic in 67 countries, and no pooled excess mortality data were presented [26]. The coronavirus not only directly kills people but also causes a chain reaction of premature deaths in society. For example, in response to the ongoing epidemic crisis, the Greek public healthcare system ceased most of its regular activities and redirected available resources to COVID-19 treatment and caused excess non-COVID-19 deaths (representing 62% of all-cause excess deaths) during the first 9 months of the epidemic [50]. A similar situation occurred in Italy and England, where 20% and 25% of excess deaths during the first wave of the epidemic could not be directly attributed to COVID-19, respectively [51,52]. Besides, lacking guidelines and personal protective equipment also downsized the clinical activities of primary care centers, which may have increased excess non-COVID-19 deaths during the pandemic [50].
We found that South America had the highest pooled excess mortality, followed by North America, Europe, Asia, Africa, and Oceania, and reported all-cause deaths in Oceania were lower than expected deaths. Our findings are consistent with the COVID-19 Excess Mortality Collaborators' results [14]. In our study, the top three countries in South America for excess mortality were Ecuador, Bolivia, and Peru, which is consistent with the findings of Karlinsky et al. [3]. In the early stages of the pandemic, the number of deaths increased dramatically in Ecuador due to limited detection capacity and inadequate emergency measures, such as social distancing and wearing masks [49]. In Peru, many factors contributed to all-cause deaths, including coronavirus infection, overloaded health systems, lack of medical services, limited number of ICU beds, and inadequate oxygen supply equipment during the pandemic [9].
North America and Europe were two continents with excess mortality that was only lower than that of South America. Lower mask use, more frequent population mobility, and fewer social distancing mandates may cause high all-cause excess mortality in the United States and parts of European countries [14]. In this study, Africa had low excess mortality, but the prevalence of COVID-19 was severe in sub-Saharan Africa. Thus, we consider that underreporting of deaths or lack of mortality-related surveillance or reporting in some countries may lead to a low rate [14,53,54]. The number of reported all-cause deaths in Oceania was lower than the expected deaths, which is consistent with the findings of previous studies [14,15]. This phenomenon may be related to the following reasons: First, Australia and New Zealand have implemented strict entry-exit screening, timely detection, vaccination and mask-wearing requirements, close contact tracking, and vulnerable group attention during the pandemic [55][56][57][58]. Second, the medical information surveillance system plays an important role in the timely response to public health emergencies in Australia [59]. Third, unique meteorological factors and the Australian government's influenza vaccination campaign during the pandemic reduced the number of influenza deaths [15]. All the above actions may potentially reduce the number of all-cause deaths in Oceania during the pandemic.
We found that the pooled excess mortality in developing countries was higher than that in developed countries, and middle-income countries had higher excess mortality than high-income countries. The results of the COVID-19 excess mortality collaborators supported that East Asia, Australia, and the high-income Asia-Pacific region had low excess mortality [14], which is similar to our findings. The pandemic has brought shocks to health systems in countries worldwide. Our analysis suggests that developed countries have better quality and more adequate quantities of health care services (e.g., number of intensive care beds, oxygen ventilators, etc.) [60] compared to developing countries. Thus, these advantages might reduce all-cause excess mortality in developed countries. In addition, vaccination is an important protective factor in reducing excess deaths globally, and studies have shown that the number of new deaths per million people decreases over time as vaccine coverage rises [61]. At the beginning of the outbreak, developed countries had better access to the COVID-19 vaccine and higher public accessibility to vaccination, so their vaccine coverage was higher than that of developing countries [62].
In this systematic review, the male population had higher excess mortality than the female population, which is consistent with previous studies [41,[63][64][65]. Males with COVID-19 infection have longer courses and worse prognoses than females. In addition, androgens, especially testosterone, are considered a possible risk factor [66]. Populations older than 60 years had higher excess mortality, and several country studies also indicate the same results [42,45,67,68]. We believe that the senior population has lower physical function and immunity compared to younger people, and they are more susceptible to the neo-crown virus during the pandemic [69,70]. Meanwhile, elderly people tend to suffer from one or more chronic diseases, and they are at a higher risk of death due to neo-coronavirus or post-infection complications after unfortunate infections. Moreover, the elderly population is more concerned about the effects of adverse vaccine reactions, and therefore, vaccine hesitancy leads to relatively low vaccination rates in this population. Especially developed countries have a much larger proportion of the elderly population and the excess number of deaths from this population is greater [69,70].
Our meta-analysis still has several limitations. First, the number of reported allcause deaths is real-world data from the mortality surveillance system or death survey in 79 countries. Despite quality control of the data, it is possible that all-cause excess mortality in some countries is underestimated due to delayed or omissive. Because robust vital registration systems do not exist in many parts of the world, the WHO estimated that 40% of global deaths that occurred in 2020 were unregistered [71]. Second, only 10.1% of countries reported the number of all-cause reported deaths and expected deaths by sex and Vaccines 2022, 10, 1702 14 of 17 age groups. It is necessary to refine and supplement excess mortality results for sex and age by including more country data in the future. Third, the number of African countries included in this study is small. However, Africa has a severe prevalence of COVID-19 with a potentially high number of excess deaths. Relevant studies in Africa are required to further complement all-cause excess mortality globally.

Conclusions
In this meta-analysis, the pooled global excess mortality was 104.84 deaths per 100,000, and the number of all-cause reported deaths was higher than expected deaths during the COVID-19 pandemic worldwide. Excess mortality was higher in South America, North America, Europe, developing countries, middle-income countries, the male population, and individuals aged ≥60 years. Further research needs to more accurately estimate all-cause excess mortality attributed to the COVID-19 pandemic.
Supplementary Materials: The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/vaccines10101702/s1, Text S1: Search strategy; Table S1: Risk of bias of all included cross-sectional studies using the Agency for Healthcare Research and Quality scale (AHRQ-Tool) (n = 18); Table S2: Risk of bias of all included cohort studies using the Newcastle-Ottawa quality assessment scale (n = 2); Figure S1: Sensitive analysis of excess mortality among 71 countries/regions.