Global COVID-19 Pandemic Outcomes: A Cross-Country Comparison Study of Policy Strategies

Abstract

This paper explores global COVID-19 treatment and containment strategies in 108 countries worldwide, specifically the correlation between COVID-19 deaths and the countries’ vaccination rates. Comparison of data across states, provinces, territories, and countries relied upon a common method to evaluate data regarding the impact of COVID-19 policies in the last three years. Data from nine different databases were analyzed to determine if there were correlations between the percentage of countrywide COVID-19 deaths/population and countries’ percent vaccinated. Secondary outcome measures include the effect of other variables on COVID-19 death rates per country population, including health expenditures and annual income per capita, COVID-19 tests per 1000 people, stringency index (a measure of each country’s containment strategies), hydroxychloroquine/ivermectin scores (measure country use), hypertension, obesity, diabetes, and geographic locations. COVID-19 vaccination rates ranged from 0-99% in 108 countries. Bivariate analysis demonstrates the following independent variables to correlate with COVID-19 deaths/population (Spearman correlation coefficient, p value): countrywide COVID-19 vaccination rates (moderate relationship, r=0.39, P < .001); healthcare expenditures per capita per annum (US dollars) (moderate relationship, r=0.46, P < .001), net annual income per capita (moderate relationship, r=0.50, P < .001), COVID-19 tests per 1000 country population (moderate relationship, r=0.36, P < .003); stringency index per country (moderate relationship, r=0.28, P < .003); hydroxychloroquine index (negative relationship, r= 0.15, P = .125); and ivermectin index (negative relationship, r=0.23 P = .018). The authors found that the higher the percentage of a country’s vaccination rate, stringent containment strategies, mass testing, etc., moderately correlated with higher COVID-19 death rates/population. Future studies are required to explore the findings of this study fully.

Introduction

With global infection rates declining to the lowest levels in over a year and many countries having discontinued social containment policies, which, in many cases, had been in place since 2020, it would appear as though the COVID pandemic is winding down. Since its beginning, nearly every facet of the pandemic has been chronicled and subjected to deep analysis in one way or another. Undoubtedly, this will be the most thoroughly documented pandemic in history.

Despite such intense scrutiny, outcomes analysis will ultimately be hampered by a lack of valid historical comparisons regarding the unparalleled modern data-gathering capacity and the unique nature of the SARS-CoV-2 causal agent. Given such limitations, how can one draw meaningful conclusions regarding the effectiveness of the global response? What are the take-home lessons? How will the present COVID-19 response be used to guide future pandemic interventions?

Mmeasures lacked uniform application across the globe. Broad disparities were observed in how individual countries responded to the pandemic based not only upon the availability of resources, per capita wealth, and population composition but also on the ability or inclination of sovereign nations to impose and enforce containment strategies. To complicate matters further, there were wide variations from country to country in the percentage of individuals that ultimately received the vaccines. Finally, there were confounding variables independent of vaccine status, particularly in less developed, socioeconomically disadvantaged nations such as those in Africa: the broad use of prophylactic agents such as hydroxychloroquine and ivermectin.

Using country-level data is a reliable method to study various associations between parameters such as the above.^1-5 To address such striking contradictions in the pandemic response, the authors identified variables to compare global outcomes country-by-country and subjected obtained data to statistical analysis.

Materials and Methods

To assess outcome variability, the authors assembled pandemic-related data on a country-by-country basis using population size, the number of diagnosed COVID-19 cases, and the number of registered deaths, from which the authors calculated COVID-19 mortality/population. These, in turn, were compared to the percentage of individuals in the population who had been vaccinated. All data were obtained from the Johns Hopkins University Coronavirus Resource Center.⁶ Data were collected between August 21 and September 21, 2022.

Descriptive statistics of 108 countries were studied for the final variables in our model, as noted in Table 1. The specific geographic locations were compared to all others by assigning a binary variable (0,1) with 1 assigned to the specific geographic area of interest (country or country groupings) and 0 to all other locations. None of the variables were uniformly distributed as per Shapiro Wilk test (P < .01), and logarithmic data transformation and non-parametric analysis were performed. There were missing data in only one variable (5 missing, n=103) in the number of COVID-19 tests per 1000 population as these data were unavailable for five countries.

Table 1.

Descriptive Statistics for 108 countries

Variable	Range	Mean ± SD
Africa (0,1)	N/A	N/A
Australia (0,1)	N/A	N/A
Canada (0,1)	N/A	N/A
Central & South America (%)	N/A	N/A
East Asia (0,1)	N/A	N/A
Europe (0,1)	N/A	N/A
India (0,1)	N/A	N/A
Japan (0,1)	N/A	N/A
Mexico (0,1)	N/A	N/A
Middle East (0,1)	N/A	N/A
Russia (0,1))	N/A	N/A
USA (0,1)	N/A	N/A
Population	586634 – 1.412B	50.85M ±144.1M
COVID-19 tests per 1000 population	5 – 21272	1710 ± 3214
COVID-19 Vaccinated (%)	0 - 99	52.2 ± 0.3
COVID-19 cases	7571 – 93.75M	5.280M±11.92M
COVID-19 deaths	38 – 1.04M	55257±139,091
Pop > 65 years (%)	1 - 29	10.1 ± 7.3
Pop < 14 years (%)	12 - 46	26.9 ± 0.1
Net Annual Income per Capita ($)	174 – 64,140	4272 ±15,763
Health Expense per Capita ($)	20 – 10,921	337 ± 2143
Diabetes (%)	1 - 31	8 ± 4.5
Hypertension (%)	1 - 75	16 ± 17
Obesity (%)	2 – 37	19 ± 9
Hydroxychloroquine Score (0 – 5)	0 - 5	2.2 ± 1.9
Ivermectin Score (0 – 5)	0 - 5	1.6 ± 1.9
Stringency Index (0 – 11)	1 - 11	8.9 ± 2.2
COVID-19 deaths / Population (%)	0.007 –0.47	0.07551 ± 0.128
COVID-19 deaths / COVID-19 cases	0.000763 – 0.0784	0.01484 ± 0.01281

To evaluate potential confounding factors, we constructed a list of country-specific parameters that could potentially influence or modify outcomes in a particular locale: hypertension rates, obesity rates; diabetes rates, percent of the population > 65 years of age, percent of the population < 14 years of age, per capita health expenditures per capita, and net annual per capita income. These were obtained from official sources such as the World Health Organization, World Bank, etc.^7-15

To evaluate the effectiveness of social containment measures, we developed a stringency index based on a binary (0,1) weighting of 11 parameters: vaccine mandates, masking, social distancing, curfews, quarantine, business/school closings, banning or limiting public gatherings, lockdowns, travel bans, contact tracing, and PCR testing. The highest possible stringency score was thus 11, while countries with more lax policies were correspondingly lower. Data were obtained via internet search from the 10 websites.^6-15

To evaluate the potential influence of background hydroxychloroquine and/or ivermectin use, the researchers developed an HCQ score and an IVM score based upon a five-tiered ranking scale: no use (0), sporadic-to-limited use (1), limited-to-moderate use (2), moderate use (3), moderate-to-widespread use (4), and country-wide use (5). Data were obtained via internet searches.

Finally, as a means of cross-correlating various outcomes with global testing initiatives, the researchers assayed the number of COVID-19 tests per 1000 people country-by-country.¹⁴

Statistical analysis was performed using MedCalc^® (MedCalc statistical software, v. 20.115, MedCalc Software Ltd, Ostend, Belgium; https://www.medcalc.org.) statistical software. Bivariate analytics were performed using Spearman correlation coefficients.¹⁶ The Shapiro-Wilk test was used to determine normal distributions and log transformation was performed when necessary. A forward regression modeling approach used P values for entry if < .05 and removal if P > .10.

Results

Countrywide COVID-19 vaccination rates ranged from 0 to 99% (n=108) with a mean ± 1 standard deviation [SD] of 52.2 ± 0.3. The COVID-19 deaths per population ranged from 0.00001 to 0.653 (n=108) with a mean (± 1 standard deviation [SD]) of 0.1309 ± 0.1413 and was not normally distributed (P < .01). Only one country reported a vaccination rate of zero, and that was Burundi, with a low COVID-19 death rate per population at 0.07, the fifth percentile of all the other countries.

For comparative purposes, the 12 countries with the lowest vaccination rates are depicted in Table 2, and the 12 countries with the highest rates are presented in Table 3.

Table 2.

Depiction of the 12 HIGHEST COVID-19 vaccination rates by country, including the COVID-19 death rate per population, COVID-19 testing rate per 1000 population, and Healthcare Expenses per annum per capita.

Country	COVID-19 Vaccine Rate	COVID-19 Death Rate per Population	COVID-19 Testing Rate per 1000 Population	Healthcare Expenses per Annum per Capita
United Arab Emirates	99%	0.02	17884	$1843
Qatar	96%	0.023	2818	$1807
Chile	92%	0.315	2040	$1376
New Zealand	91%	0.035	1416	$4211
Portugal	87%	0.24	4160	$2221
Spain	87%	0.24	1962	$2721
Singapore	86%	0.27	1800	$2632
South Korea	86%	0.05	1935	$2625
Australia	85%	0.052	2831	$5427
Peru	85%	0.653	859	$370
Argentina	84%	0.273	810	$946
Vietnam	84%	0.014	881	$181

Table 3.

Depiction of the 12 LOWEST COVID-19 vaccination rates by country and includes the COVID-19 death rate per population, COVID-19 testing rate per 1000 population, and Healthcare Expenses per annum per capita.

Country	COVID-19 Vaccine Rate	COVID-19 Death Rate per Population	COVID-19 Testing Rate per 1000 Population	Healthcare Expenses per Annum per Capita
Burundi	0	0.0700	128	$21
Haiti	2%	0.0300	18	$57
Papua New Guinea	3%	0.0100	25	$65
Madagascar	5%	0.0200	16	$20
Cameroon	5%	0.0200	100	$54
Senegal	7%	0.0200	66	$59
Sudan	10%	0.086	12	$23
Gabon	12%	0.0800	12	$215
Malawi	12%	0.0100	683	$30
Somalia	13%	0.0300	30	$22
Nigeria	14%	0.0500	29	$71
South Sudan	14%	0.0100	25	$23

Bivariate Analysis

A bivariate analysis was conducted on the following attributes as displayed in Figures 1-8 below: vaccination rate per country, healthcare expenditure per capita per annum (US dollars), annual income per capita (US dollars), COVID-19 testing per country, stringency index by country, and the hydroxychloroquine (HCQ) and ivermectin (IVM) indexes.

Figure 1.

depicts a scatter plot of COVID-19 deaths per population by the country’s vaccination rates. The correlation coefficient (r) is 0.39 (n=108, P < .001. The higher the countrywide COVID-19 vaccination rates, the higher the COVID-19 deaths per country population (P < .001).

Figure 2.

depicts a scatter plot of COVID-19 deaths per country population versus healthcare expenditure per capita per annum (US dollars). The correlation coefficient is +0.46 (n=108, P < .001). The higher the country’s healthcare costs per capita per annum, the higher the COVID-19 deaths per country population (P < .001).

Figure 3.

depicts a scatter plot and regression line with 95% confidence interval of COVID-19 deaths per country population versus annual income per capita (US dollars). The correlation coefficient was 0.50 (n=108, P < .001). The higher the annual income per capita, the higher the COVID-19 deaths per country population (P < .001).

Figure 4.

depicts a scatter plot and regression line with 95% confidence interval of COVID-19 deaths per country population versus the COVID-19 tests per 1000 country population. A log transformation was performed because the data was not uniform (P < .01). The correlation coefficient (r) is +0.36 (n=103, P < .001). The higher the countrywide COVID-19 testing, the higher the COVID-19 deaths per country population (P < .001).

Figure 5.

depicts a scatter plot and regression line with 95% confidence interval of COVID-19 deaths per population versus stringency index (1-11). A log transformation was performed because the data was not uniform (P < .01). The correlation coefficient (r) is +0.28 (n=103, P < .003). The higher the country’s stringency index, the higher the COVID-19 deaths per country population (P < .003). Stringency index is based on a binary (0,1) weighting of 11 parameters: mandates, masking, social distancing, curfews, quarantine, business/school closings, banning or limiting public gatherings, lockdowns, travel ban, contact tracing, and PCR testing. The highest possible stringency score was thus 11 while countries with more lax policies were correspondingly lower. Data were obtained via internet search from the 9 websites.^1-10

Figure 6.

depicts a scatter plot with a regression line and 95% confidence interval of COVID-19 deaths per population versus the hydroxychloroquine (HCQ) index. The correlation coefficient is 0.15 (n=108, P = .125). A higher countrywide HCQ index is associated with lower COVID-19 deaths per country population (P = .125).

Figure 7.

depicts a scatter plot and regression line with 95% confidence interval of COVID-19 deaths per population versus the ivermectin (IVM) index. The correlation coefficient (r) is 0.23 (n=108, P < .018). A higher Ivermectin (IVM) index is associated with lower COVID-19 deaths per country population (P < .018).

Figure 8.

depicts global monthly cases of COVID-19 and the monthly COVID-19 deaths from the World Health Organization.

Multivariate Analysis

A multivariate analysis was also conducted, as presented in Table 4. A forward regression model using COVID-19 deaths per country population as the dependent variable was performed using P values for entry < .05 and removal if P > .10.

Table 4.

notes the forward regression model using COVID-19 deaths/population as the dependent variable and only two variables were retained (t value, P value): East Asia (+2.932, .041) and Europe (+4.648, <.0001). The following independent variables were eliminated (enter variable if P < .05 and remove variable if P > .1): Africa, Australia, Canada, Japan, Mexico, Middle East, diabetes rate, hypertension rate, obesity rate, Russia, Sweden, USA, Age > 65, Age < 14, India, and Central & South America.

Independent Variables	Coefficient	Standard Error	t value	P value
Constant	0.03807
East Asia (0,1)	0.1569	0.05352	2.932	.0041
Europe (0,1)	0.1207	0.02597	4.648	<.0001

Discussion

The findings of this study demonstrate COVID-19 death rates per population were moderately correlated with countrywide COVID-19 vaccination rates (r=0.39, P < .001) = .002); the higher the percentage of a country’s vaccination rate, the higher the COVID-19 death rate/population. Other moderate correlations of secondary measures include healthcare expenditure per capita (r = 0.46, P < .001), national average income per capita (r = 0.50, P < .001), rate of COVID-19 testing per 1000 population (r = 0.36, P < .001), and stringency index (r = 0.28, P = .003).

The researchers undertook this analysis to investigate outstanding questions regarding the efficacy of global pandemic management strategies: social containment measures such as masking and lockdowns, widespread testing measures, and the vaccine initiative. In so doing, the authors sought to address disparities concerning the stated efficacies of these measures related to reported global and regional case numbers, which, in the end, seem to tell a different story.

One question assumed primacy: how effective was the highly centralized pandemic management strategy, and was it a viable approach for future pandemics? It should not be overlooked that this is the first pandemic in which widescale orchestrated efforts were implemented. Is the world in better stead because these strategies were employed?

Comparing reported case numbers at three points during the pandemic – early January 2020, 2021, and 2022 – a pattern emerges of unimpeded spread regardless of enacted measures. On January 1, 2020, as the pandemic was emerging (and before data were available), global case numbers were (perhaps) in the thousands. During the first week of January 2021, after nearly 10 months of containment measures, 4 985 723 new cases were reported globally. During the first week of January 2022, after a year of containment measures and vaccine initiatives, 16 138 104 new cases translated to a 3.2-fold increase. By late January, at the peak of the Omicron surge, numbers had skyrocketed even further to 23,205,305, equating to a 4.6 × increase (Figure 8).^6,15,17

Similar trends were seen in the US^6,15,17 in the first week of January 2021, when the US tallied 1 667 173 new cases. During the first week of January 2022, these numbers climbed to 4 682 921, nearly a 2.8-fold increase. By mid-January, the total hit 5650958 new cases, corresponding to a 3.4-fold increment.¹⁵ Based on such data, even a hint of beneficial outcomes related to social containment or mass vaccination was unfounded and may have increased the risk of COVID-19 deaths per population. This is corroborated by the findings of this study as outlined in Figure 5: the higher the countrywide stringency index, the higher the COVID-19 deaths per country population (P = .047).

The reasons behind the uncontrolled spread are well established. At least 50% of viral transmission occurred through asymptomatic or pre-symptomatic carriers.^11-18 This is to say that policies intended to curtail viral spread were, from the beginning, doomed to fail. This, in turn, called into question the value of mass populational testing, which, even under optimal circumstances, has a sensitivity of about 80% and, in the real world, no more than 50-60%. These facts argue against testing measures favorably impacting transmission dynamics.^26-39

A similar case can be made regarding the vaccines. Mass deployment began in early 2021, and by Spring, a handful of countries, Israel, in particular, were nearing the hypothetical threshold for herd immunity. Yet, once the Delta variant emerged, it spread like wildfire independent of a country’s vaccine status.^30,31 It was only later recognized that the vaccines did not confer immunity but induced short-term protection by stimulating an antibody response. Therefore, booster doses became necessary.^32-40

Despite an aggressive booster campaign, the same results occurred when the Omicron variant emerged in Fall 2021. By January 2022, case rates, both globally and in the US, reached the highest pandemic levels, far surpassing those at the beginning of 2021 when the vaccines were being rolled out15. A large percentage of those infected had received both vaccination and booster jabs. It is difficult to explain these results on any basis other than the failure of containment and vaccine strategies.^41-43

The failure to affect pandemic outcomes significantly incriminated top-down centralized management strategies by entities such as WHO and large countries like the US. In March 2020, WHO Director-General Ghebreysus announced, “We have a simple message for all countries: test, test, test.” He claimed that testing was essential to contain the spread of the virus. But how does a bureaucrat in New York City know what is happening “on the ground” in Brazil, Indonesia, or Nigeria? The one-size-fits-all pandemic management strategy was an unqualified disaster. The only viable solution was radical decentralization of authority and policymaking.

The African experience was a powerful testament to this approach. For decades throughout Africa, hydroxychloroquine (HCQ) and chloroquine (CQ) have been staples for preventing and treating malaria. Its efficacy and safety were empirically well-established, and it was inexpensive and widely available. It seemed inevitable these two agents would figure into the African pandemic strategy until the publication of one study. The findings in this study were consistent with this strategy.

In May 2020, the medical journal Lancet published a meta-analysis of 96 000 hospitalized COVID-19 patients from 671 hospitals across the globe, claiming that HCQ had no benefit and was associated with an increased risk of cardiac arrhythmias and death.³⁷ The article, however, was fraudulent and subsequently retracted by Lancet. But the damage was far-reaching. Not only were several ongoing clinical trials discontinued, but oversight agencies such as the European Medicines Agency (EMA), WHO, and the U.S. Food and Drug Administration issued warnings against their use in the following months. These recommendations were met with skepticism on the African continent.³⁷

Countries such as Egypt, Zambia, Nigeria, Tunisia, and South Africa chose to continue clinical trials under the support of Africa Centres for Disease Control and Prevention (ACDC). The director of Nigeria’s National Agency for Food and Drug Administration and Control (NAFDAC), Dr. Mojisola Adeyeye, affirmed ongoing support of HCQ and claimed that Nigeria would continue clinical trials despite the WHO warning: “The narrative might change later, but for now, we believe in hydroxychloroquine.”⁴⁵

Other countries chose to ignore the edicts. The Economic Community of West African States (ECOWAS) approved HCQ and CQ to treat COVID-19 infection. Ghana’s health minister, Kweku Agyeman-Manu, also approved HCQ for widespread use and supported its efficacy. Similarly, Uganda continued its use in conjunction with azithromycin and claimed beneficial results. Djibouti continued to treat all COVID-19 infections with CQ/HCQ and azithromycin. Djibouitian health officials claimed that the death rate was only 0.5%, and Dr. Maad Nasser Mohammed, the top official of the COVID-19 response center, claimed that the treatment regimen was the main reason for the low death rate.

Algeria also used HCQ for COVID-19 despite the cessation of the WHO-sponsored trials. Mohamed Bekkat, a member of the COVID-19 treatment committee, claimed that thousands of cases had been successfully treated with the HCQ + azithromycin combination with very few undesirable reactions. Moroccan Minister of Health Khalid Ait Taleb vigorously defended its use in COVID-19 infections and claimed Morrocco would continue to use CQ despite warnings from the WHO. Meanwhile, however, as more data have emerged from countries across the globe, the efficacy of HCQ and CQ in early COVID-19 infection has been widely substantiated.

Dr. Peter McCullough most eloquently communicated this important concept in his now famous “Lesson Learned” Testimony to the Texas State Senate on June 28, 2022.⁴⁶ As McCullough emphasized, the community standard of care should not emanate from a top-down decree but quite the opposite. Community standards of care should be established at the ground level with experienced, local care providers treating their patients with their wisdom, due diligence, and research. Communication of these grassroots experiences, among others, then establishes a regional standard.

For decades throughout Africa, hydroxychloroquine (HCQ) and chloroquine (CQ) have been staples for preventing and treating malaria. Its efficacy and safety were empirically well-established, and treatment was inexpensive and widely available. It seemed inevitable these two agents would figure into the African pandemic strategy until the publication of one study. The greater the availability of HCQ by country, the fewer COVID-19 deaths per country population (-0.1337, P = .0678).

The centralized statistically based management strategies implemented during the COVID-19 pandemic utterly failed to achieve their stated goals. The socioeconomic and individual consequences of the failed containment strategies, such as lockdowns, banning public gatherings, and business closures, far outweighed any possible benefit regarding loss of life or social well-being. This study demonstrates negative associations between containment strategies, widespread populational testing, mass vaccination, and disease outcomes. Indeed, these interventions were associated with more COVID-19 deaths per country population. Evidence suggests that future mass infectious outbreaks should be managed more efficiently and effectively on the ground at regional levels where consequences are most directly felt.^46,47

Limitations of this study’s conclusions are only as reliable as the validity of the data abstracted from the 10 sources used to assemble the database.^6-15 There is wide variation not only in case reporting but in mortality, testing, etc. from country to country, and inconsistencies could potentially limit global outcomes. Additionally, the binary stringency index utilized presented a limitation to this study as each country exhibited varying levels of targeted strategies used in their pandemic response. Non-uniform data, collinearity, and heteroskedasticity may limit Multiple regression analytics. Additional country-comparison model studies are needed to understand fully the association between pandemic management strategy and outcome.