By the end of the first year of the pandemic, the United States had the highest global, cumulative COVID-19 case and death counts [1]. Additional challenges, such as newly emergent lineages of SARS-CoV-2 with increased transmission potential and issues in COVID-19 vaccine distribution and administration, have increased the burden on public health systems worldwide [2]. The implementation of public health and social measures (PHSMs) such as stay-at-home orders, limited gatherings, and the closure of nonessential workplaces is a crucial method for preventing and mitigating the spread of COVID-19. In spring of 2020, the Health Intervention Tracking for COVID-19 (HIT-COVID) database was developed to catalogue the global implementation and relaxation of COVID-19–related PHSMs [3]. Many other PHSM surveillance systems also collect data that can be used to inform pandemic responses [4-6]. These PHSMs have been shown to be effective, both individually and in combination, against COVID-19 globally. For example, a recent study found that from January to May 2020, limiting gatherings, closing businesses, closing schools and universities, and implementing stay-at-home orders were individually effective at reducing the time-varying reproduction number (R_t) of SARS-CoV-2 [7]. An earlier study found that in combination, limited gathering sizes, business closures, educational institution closures, and stay-at-home orders reduced COVID-19 transmission from January to May 2020 in 11 European countries [8]. The implementation of PHSMs is a key marker of how public health systems address the COVID-19 pandemic. Poor COVID-19 outcomes in the United States of America have been attributed to a failure to consistently, quickly, and effectively implement PHSMs [9]. We sought to analyze the epidemiological evidence for the impact of PHSMs on COVID-19 transmission in the United States and compare these data to those for 10 other countries—places that provide sources of learning to the United States.

We analyzed the timing and stringency of PHSMs that were implemented from January 1 to November 25, 2020, compared them against time series for daily case counts of COVID-19, and compared the United States to Canada, China, Ethiopia, Japan, Kazakhstan, New Zealand, Singapore, South Korea, Vietnam, and Zimbabwe.

These 10 countries were chosen for comparison with the United States based on their varying income levels and geographies and were selected as comparators for the US COVID-19 response, which was reported in a previously published, high-profile commentary on global PHSM effectiveness [9]. Canada was chosen for its comparatively lower death rate; China was chosen for being the first country affected by COVID-19 and having a large population; Japan was chosen for its older population; New Zealand was chosen for being geographically isolated; Singapore and South Korea were chosen for their geographic proximity to China; Vietnam was chosen for being a lower-middle–income country; and Ethiopia, Kazakhstan, and Zimbabwe were chosen for having less medical infrastructure and manufacturing capacity.

For ease of analysis, we focused on the following eight categories of PHSMs: border closures, contact tracing, household confinement, mandated face masks, quarantine and isolation, school closures, limited gatherings, and a state of emergency. These categories of PHSMs were chosen based on a brief literature review, data availability in the HIT-COVID database, and an exploratory data analysis. The stringency of PHSM implementation was classified in the HIT-COVID surveillance database as strongly implemented, partially implemented, or not implemented based on the specific details of each PHSM. After being abstracted from government or news websites, data on PHSM timing and stringency in the HIT-COVID database undergo internal auditing by both the person who entered the data and the database management team [3]. PHSM and case count data were derived from the HIT-COVID database, the World Health Organization PHSM global database, and the European Centre for Disease Prevention and Control [5,10]. For each country, the date and stringency of PHSM implementation were plotted against the number of daily national COVID-19 cases from January 1 to November 25, 2020. Additionally, we compared the 11 countries’ cumulative incidence rates to 40 variables that described national measures of social, demographic, economic, and health system characteristics [11-14]. The analysis was performed by using the HIT-COVID R package version 4.0.3 [15]. All data used in this analysis were publicly available, and this study did not constitute human subjects research; therefore, ethical review was not required.

According to the HIT-COVID database, from January 1 to November 25, 2020, 11,999 PHSMs were implemented in 148 countries. Of these PHSMs, 5,695 fell into the eight categories that were defined for analysis in this study. School closure was the most common measure (1592/5695, 27.95%), followed by border closures (1481/5695, 25.98%), quarantine and isolation (705/5695, 12.39%), limited gatherings (629/5695, 11.06%), household confinement (573/5695, 10.05%), face mask mandates (272/5695, 4.78%), a state of emergency (269/5695, 4.73%), and contact tracing (174/5695, 3.04%).

We compared the cross-country timing and stringency of PHSMs to COVID-19 epidemic curves, as shown in Figures 1 and 2. The results show that implementing a specific package of PHSMs—quarantine and isolation, school closures, household confinement, and limited social gatherings—earlier and more stringently coincided with limited case counts and transmission durations in Ethiopia, Kazakhstan, New Zealand, South Korea, Vietnam, and Zimbabwe. Further, these countries’ case counts and transmission durations were lower than those of the United States. Singapore and South Korea implemented similar PHSMs, including less stringent household confinement, but both countries substantially lowered the epidemic curve further than the United States. China implemented fewer PHSMs for shorter durations via more targeted subnational implementation. Japan implemented strong quarantine and isolation measures along with early school closures but did not maintain the same level of stringency as that of the package of 4 PHSMs. Canada initially implemented strict PHSMs but did not maintain this level of stringency across the country. Canada is now facing a resurgent epidemic and has a relatively high cumulative COVID-19 incidence rate, though it is still much lower than that of the United States.

In contrast, the United States did not implement the package of 4 PHSMs that were shown to be effective in the 6 countries mentioned above. The United States implemented few PHSMs fully or early during the pandemic. Although the US government implemented some PHSMs early in the pandemic, these PHSMs were only partially implemented; they were not strongly implemented like the PHSMs of most of the other countries in this analysis.

Border closures, state-of-emergency declarations, and mandated mask wearing did not show strong temporal overlap with epidemic growth across the 11 countries, though some of these PHSMs may have been important in individual countries (eg, border closures in New Zealand). The adherence to mask wearing seemed to vary widely across countries.

The 11 countries differed significantly in terms of many demographic, economic, social, and health system characteristics. These differences may contribute to their varying success. After comparing the countries’ cumulative COVID-19 incidence rates with national measures of social, demographic, economic, and health systems factors, we found only 1 positive correlation (r=0.624; r²=0.389) between national income (measured as gross national income per capita in US dollars) and cumulative COVID-19 incidence (Figure 3). Lower-income countries performed better than higher-income countries; the three countries with the lowest incomes had the lowest incidence rates, and the two countries with the highest incomes had the highest incidence rates.