ABSTRACT
Objective To understand the associations of COVID-19 cases and deaths with policy stringency globally and regionally.
Methods We modeled the marginal effects of new COVID-19 cases and deaths on policy stringency (scored 0–100) in 175 countries and territories, adjusting for gross domestic product (GDP) per capita and health expenditure (% of GDP). Time periods examined were March–August 2020, September 2020– February 2021, and March–August 2021.
Results Policy response to new cases and deaths was faster and more stringent early in the COVID-19 pandemic (March–August 2020) compared to subsequent periods. New deaths were more strongly associated with stringent policies than new cases. In an average week, 1 new death per 100,000 people was associated with a stringency increase of 2.1 units in March–August 2020, 1.3 units in September 2020–February 2021, and 0.7 units in March–August 2021. New deaths in Africa and the Western Pacific were associated with more stringency than in other regions. Higher health expenditure was associated with less stringent policies. GDP per capita did not have consistent patterns of associations with stringency.
Conclusions Our findings demonstrate the need for enhanced mortality surveillance to ensure policy alignment during health emergencies. Countries that invest less of their GDP in health are inclined to enact stringent policies during health emergencies than countries with more significant health expenditure.
INTRODUCTION
Policy responses during the COVID-19 pandemic have a dynamic relationship with epidemiological outcomes.[1,2] The effects of public health policies in slowing the spread of severe acute respiratory syndrome coronavirus 2 (SARS-COV-2) are demonstrated in multiple studies, and new evidence is emerging.[3-5] Public health policies to contain the spread of SARS-COV-2 have included restricting population movements and gatherings, closing schools and businesses, and requiring masks indoors—also known as non-pharmaceutical interventions (NPI). Numerous platforms track the stringency of policies enacted globally since the early stages of the pandemic.[6] However, the drivers of these policies remain poorly understood.
Mathematical modeling suggests that policies for NPI are critical to contain the COVID-19 pandemic alongside improved access to and uptake of COVID-19 vaccines [7]. The emergence of variants of concern with immune evasion capability, such as Omicron, reinforces the need to continually (re)calibrate public health policies to control the pandemic.[8] These modeling findings are layered on the backdrop of persistent inequities in access to and uptake of COVID-19 vaccines in low-income countries compared to middle- and high-income countries.[9]
Member states report new COVID-19 cases and deaths to the World Health Organization (WHO).[10] The reported epidemiological trends create dynamic perceptions about the intensity of the COVID-19 situation in a country [11]. They may trigger policies that are not aligned with the COVID-19 epidemiology. Moreover, seroprevalence assessments of SARS-CoV-2 have uncovered gross underestimation of the population-level burden of infection when compared to officially reported new cases.[12,13] Variations in testing capacity, testing strategies, and people’s willingness to be tested contribute to the underestimation of SARS-CoV-2 infections.[14] COVID-19-related mortality may also be underreported due to weak mortality surveillance systems, inadequate systems for civil registration of deaths, and differences in how causes of death are classified.[15]
How rapidly and aggressively countries respond to the reported new cases and deaths may also be influenced by socioeconomic conditions and political considerations.[16,17] An empirical understanding of how the reported epidemiological data influences policy stringency can help predict policy responses in the current pandemic and other health emergencies. To this end, we aimed to model the associations of reported COVID-19 new cases and deaths with policy stringency.
METHODS AND MATERIALS
Our study examined data for 175 countries and territories representing all regions globally. We used fractional regression to model the marginal effects of new cases and deaths on COVID-19 policy stringency, adjusting for the gross domestic product (GDP) per capita and health expenditure. We adjusted for GDP per capita because prior studies have shown a dynamic relationship between GDP and policy stringency.[19] GDP may influence policy stringency tolerable to a government. Protracted ‘lockdown’ policies, for instance, may not be tenable in low GDP countries. We adjusted for health expenditure (as a percent of GDP) because countries with greater health investments may have greater investments in their COVID-19 response.[20] The GDP and the health expenditure do not change remarkably on an annual basis for the same country. We used the latest reported GDP and health expenditure statistics from the United Nations.[21]
Main outcome
Our main outcome is the stringency of COVID-19 policies as measured daily by the Oxford Tracker [22]; its methods data collection and scoring methods have been described elsewhere.[23] In summary, the stringency index ranges from 0 to 100, with higher scores representing more stringent policies. The score is based on nine component indicators measured on an ordinal scale to assess closing of schools and universities, closing of workplaces, cancelling public events, limits on gatherings, closing of public transport, mandates for shelter-in-place / home-confinement, restrictions on internal movement between cities/regions, restrictions on international travel, and presence of COVID-19 public information campaigns.
Statistical analysis
We analyzed the data in Stata version 17 SE (StataCorp LLC, College Station, TX) by examining three six-month periods separately for March–August 2020, September 2020–February 2021, and March– August 2021. These three periods were selected a priori, and the start period (March 2020) was the onset of when most countries reported their first COVID-19 case.
We first described each of the nine ordinally measured stringency indicators for each period and combined periods (Table 1).We then merged the daily COVID-19 epidemiological data countries report to the WHO [24] with their policy stringency data.[22] New cases and deaths were calculated to reflect cases and deaths per 100,000 people. The daily merged data for cases, deaths, and stringency were aggregated weekly. We added each country’s/territory’s time-invariant data on GDP per capita [25] and health expenditure to the dataset. Finally, we used fixed-effects fractional regression with a quasi-likelihood method to model the marginal effects [18] of the various covariates on policy stringency.
Details of the statistical model-fitting methods we used are in the technical appendix. In summary, We scaled the Oxford Stringency Index to the range [0, 1] for the use in fractional regression modeling.[26] We estimated marginal effects associated with 1 new case per 100,000 people, 1 new death per 100,000 people, 1% larger GDP per capita, and 1% additional share of GDP spent on health. We report marginal effects instead of regression coefficients to enable more meaningful interpretations of the predicted change in the stringency score.
Our models used temporal lags to understand how rapidly governments enacted policies following new cases and deaths We fit separate models for the marginal effects based on 1- and 4-week lags between reporting a new case or death per 100,000 people and the associated average change in the stringency. To examine regional differences, we conducted separate analyses for each of the six WHO regions: Africa, Americas, Eastern Mediterranean, Europe, South-East Asia, and Western Pacific. A p-value less than 0.05 was considered statistically significant in all models.
RESULTS
Policy stringency waned over time. For instance, all levels of schools and universities were required to close on 61.2% of the country-days in March–August 2020 compared to 20.8% in March–August 2021. However, some policies remained relatively unchanged. Restrictions on gatherings between 11 and 100 people were in place for 28.3% of the country-days in March–August 2020 compared to 31.7% in March– August 2021. The global median stringency score was higher for March–August 2020 (68.5) compared to September 2020–February 2021 (56.4) and March–August 2021 (54.6) (Table 1). The unadjusted median stringency score increased as new cases and deaths per 100,000 people increased (Figure 1).
Global findings
Within 1 week on average, 1 new case per 100,000 people was associated with a stringency increase of 0.11 units (95% Confidence Interval [CI]: 0.08 to 0.13) in March–August 2020, 0.02 units (0.01 to 0.02) in September 2020–February 2021, and 0.02 units (0.01 to 0.03) in March–August 2021. In the early period of the pandemic (March–August 2020), increased new cases were associated with more rapidly enacted stringent policies. (Table 2)
Within 1 week on average, 1 new death per 100,000 people was associated with a stringency increase of 2.1 units (1.3 to 2.9) in March–August 2020, 1.3 units (0.9 to 1.6) in September 2020–February 2021, and 0.7 units (0.4 to 1) in March–August 2021. Like new cases, in the early period of the pandemic, increased new deaths were associated with more rapidly enacted stringent policies (Table 2).
Although 1% larger GDP per capita was associated with less stringency in March–August 2020, it was associated with more stringency in the subsequent periods. A 1% additional share of GDP spent on health was associated with less stringency across all three periods (Table 2).
Regional findings
We show the regional-specific marginal effects on policy stringency for a 1-week lag (Table 3) versus a 4-week lag (Table 4) after reporting new cases and deaths. An average increase of 1 new case per 100,000 people was associated with more stringent policies within 1 week in the Americas, Europe, and Western Pacific across all three periods. There was a slow policy response to new cases in Africa early in the pandemic. However, the association between new deaths and stringency was strongest in Africa early in the pandemic. Eastern Mediterranean was the only region where new cases were ever associated with less stringency (March 2021–August 2021).
GDP per capita did not have consistent associations with stringency across all periods within any region, and the pattern of associations was not the same for any two regions. Higher health expenditure was associated with less stringent policies within each region in March 2020–August 2020 and September 2020–February 2021, except for South-East Asia. Higher health expenditure was associated with less stringent policies within each region in March 2021–August 2021, except for Europe and the Eastern Mediterranean.
DISCUSSION
To our knowledge, this is the first study to show the epidemiological drivers of COVID-19 policy stringency globally and within regions, adjusting for GDP per capita and health expenditure. We have empirically demonstrated that countries and territories responded more rapidly and strongly to the rise in new cases and deaths early in the COVID-19 pandemic than in subsequent periods. New deaths were more strongly associated with stringent policies than new cases across all periods examined. Despite African countries having weaker mortality surveillance systems [27], the association between new deaths and policy stringency was more potent in Africa early in the pandemic than other regions (except for Western Pacific). Health expenditure was negatively associated with policy stringency, suggesting that countries with less health expenditures may be more inclined to enact more stringent policies. GDP per capita did not have consistent patterns of associations globally or within regions.
Prior assessments have focused on understanding the effects of public health policies on COVID-19 epidemiological and clinical outcomes. The current evidence shows that public health policies are essential for controlling SARS-COV-2 transmission and will likely remain vital despite the increased accessibility of safe and effective COVID-19 vaccines.[7] In our study, we found that the overall stringency of policies decreased globally, coinciding with an increasing vaccine coverage worldwide, especially in high- and (upper-) middle-income countries. As an example, COVID-19 cases surged in Seychelles after the sudden relaxation of public health policies for non-pharmaceutical interventions in April 2021 before the detection of Omicron, despite rapidly scaling up vaccination coverage to 60%.[28] It has also been documented that maintaining stringent policies for prolonged periods also comes at a cost. School closures showed associations with adverse psychosocial outcomes among children during the pandemic.[29] More broadly, the COVID-19 pandemic has contributed to adverse mental health,[30] food insecurity,[31] and decreased utilization of primary healthcare services [32].
A critical gap in the current literature is the lack of an empirical understanding of how the reported new COVID-19 cases and deaths may influence the stringency of policies. Our study provides new evidence in the global landscape and within geographic regions. An empirical understanding of policy drivers can help public health officials anticipate policy responses in the COVID-19 pandemic and future health emergencies. Thus far, mathematical modelers have not calibrated their models to account for governments’ dynamic drivers of policy responses. Our results can help improve the parametrization of mathematical models attempting to forecast the course of a health emergency.
A prior study showed that countries with weaker health systems, such as those in low- and middle-income countries, were quicker to implement more stringent policies early in the pandemic.[3] High-income countries with more significant investments in their healthcare systems were better equipped to absorb the initial shocks of the COVID-19 burden. Globally, however, healthcare systems have been overwhelmed at peak times of hospitalizations during each COVID-19 wave.[33] We found that places with less health expenditure were more inclined to enact stringent policies. Countries with higher health expenditure may rely more on their capacity to provide medical care to patients instead of instituting stringent policies. Moreover, additional research is needed to understand to what extent countries with weak mortality surveillance may have opted for less stringent policies because of their under-detection of COVID-19 deaths. Unpredictable political and economic considerations make it difficult to anticipate how governments react to this pandemic’s policymaking decisions or future health emergencies.
Our study has several limitations. Firstly, we cannot discern the enforcement of or adherence to government policies. In addition, we did not account for the potential influence of geographic proximity between countries. Although hospitalizations and hospital bed capacities likely mediated policy stringency, standardized global data on these possible mediating variables are lacking and therefore not included in our models.
CONCLUSION
The findings from this study reinforce the need for enhanced mortality surveillance to ensure policy alignment during health emergencies. Timely detection of excess deaths may help policymakers communicate their policy decisions more effectively to the public. Countries with lesser health expenditure may be more inclined to enact stringent policies during health emergencies than countries with more significant health expenditure as a percent of GDP. Our findings call for public health authorities to tailor their support to policymakers to ensure alignment between the stringency of containment policies and the public health requirements of the health emergency.
Data Availability
All data produced in the present study are available upon reasonable request to the authors.
Contributors
MFJ, ZZ, SAN, DP, AAN, MW, and HN conceptualized the study. MFJ, ZZ, HN, DP, AH, SBO, NTB made methodological contributions. MFJ conducted the literature review with additional literature identified by ZZ, AH, HN, LS, and WG. ZZ and MFJ analyzed, curated, and visualized the data with substantial contributions from MW, HN, AAN, and LS. MFJ, ZZ, SAN, DP, AAN, AH, MW, LS, WG, SBO, NTB, and HN contributed to the interpretation of the results. MFJ and ZZ drafted the initial manuscript with contributions from SAN, DP, AAN, AH, MW, LS, WG, SBO, NTB, and HN. All authors contributed to reviewing and editing the manuscript. The final manuscript was read and approved by all authors.
Competing interests
Noel T. Brewer is a paid consultant of Merck, World Health Organization, and U.S. Centers for Disease Control and Prevention (CDC) for advising on vaccination strategy, vaccination beliefs, and vaccination uptake. Several authors are employees of the CDC. The findings and conclusions in this paper are those of the authors and do not necessarily represent the official position of the CDC.
Funding
No specific funding was provided for this study. Aasli A Nur is supported by grants from the U.S. National Institutes of Health (NIH T32 HD101442-01 and P2C HD042828) to the Center for Studies in Demography & Ecology at the University of Washington.
SUPPLEMENTARY MATERIAL
We fit several fixed-effects fractional regression models using a quasi-likelihood method with a logit link to examine associations between weekly new cases/deaths and policy stringency, adjusting for GDP per capita and health expenditure in 175 countries and territories. Our fractional regression models accounted for a country’s/territory’s specific and time-varying policy stringency on a nonlinear-trend basis by including interaction terms for (1) country × time and (2) country × time-squared. The time-varying effect is measured weekly over 6-month periods. The time is mean-centered for the avoidance of multicollinearity between time and time-squared. The mean centering is done by subtracting the middle week number of the studied period from each week number. The country-specific intercept in the fixed-effects model, was replaced by other time-invariant GDP per capita and health expenditure on country level to avoid the dummy trap problem (perfect multicollinearity).
The stringency index, scaled in the interval [0,1], for country i at time t was modeled as
where Eit is the vector of weekly COVID-19 cases and deaths, Zi the vector of observed and time-invariant country-specific GDP per capita and health expenditure, and β and α are unknown regression parameters to be estimated. Also, νit is the unobservable time-varying country’s specific performance at time t, modeled as:
with unknown parameters γ0, γ1i and γ2i, to be estimated for country i. Alternatively, the above country-specific performance at time t is presented as in the fixed-effects notation. It can be equivalently represented as:
where Di is the ith country dummy variable.
For sensitivity analysis, the time-invariant predictors are used in a random-effects beta regression model, using glmmTMB library in R statistical software package version 1.1.2.3 (R Foundation for Statistical Computing, Vienna, Austria). The beta regression model was fitted as: , with γji = γj + uji, j = 0,1,2, and uji a random variable.
The results of the beta regression and the fractional regression were consistent whenever the beta regression is applicable. The beta regression was not appropriate for extreme situations with 0 or 100 stringency scores (0,1 on a fractional scale). Such extreme situations were rare in the dataset, especially after March—August 2020 period. We only report here the results from the fractional regression.
Acknowledgements
We acknowledge various technical support from Dr. Barbara Marston from the U.S. Centers for Disease Control and Prevention and Dr. Oliver Morgan and Dr. Olivier Le Polain from the World Health Organization.