Analysis of governmental responses

Timeline of COVID-19 control strategies was build for visualisation of strategic interventions put in place as a response of every chosen country to first COVID-19 infection case, first death caused by the virus and the COVID-19 outbreak announcement in China on the 20 January 2020, where the human-to-human transmission was confirmed (figure 1). Based on the dates of responses, Taiwan and Singapore reacted in advance of the outbreak announcement and first deaths in the countries. South Korea reacted on the day of the outbreak announcement with the national alert and contact tracing. USA, Taiwan and Singapore were also ahead of other countries with airport screening to tackle modes of international spread, with the USA being the first country to close its borders. The UK and Sweden had the slowest responses across developed countries that could have lead to the widespread community transmission occurring before government strategies were implemented. African countries generally reacted following the first reported case rather than implementing strategies shortly after the outbreak announcement. This might be a reflection of the lack of healthcare resources. At the beginning of the pandemic, there was a lack of comprehensive resources outlining the strategies put in place to tackle COVID-19. For this reason, figure 1 was constructed from a number of local, country specific and news articles, which relies on the accuracy of the information in non-peer reviewed resources.

The variation in strategies and their speed of implementation is represented in figure 2A,B, analysed as of 24 June 2020. From the portfolio of all OxCGRT indices, those related to early governmental strategic actions were chosen. Since all three ‘containment and health’, ‘government response’ and ‘stringency’ indices are aggregates from the same response indicators considered in OxCGRT, variation patterns are virtually identical for all of them. The data suggest that within 25 days of the outbreak being announced, Singapore, South Korea and Taiwan had the strictest and fastest responses, followed by Ethiopia, USA and UK, while Nigeria responded only after 50 days of the outbreak announcement. The speed of the response is associated with the occurrence of the first COVID-19 case. Sweden is the only country whose indices differ at that stage—the Stringency Index started increasing only on the 45th day—a result that is expected for a ‘herd immunity’ strategy. Countries with high indices, at the day of pandemic announcement, for example, Taiwan and South Korea, had lower indices 3–4 months into the pandemic suggesting that less stricter strategies with higher testing availability could be enough to contain the spread, while countries with zero or low OxCGRT indices had a rapid increase to high levels of indices 1–2 months into the pandemic. ‘Contact tracing’, ‘information campaigning’ and ‘testing policies’ were three indices that were hypothesised to represent the most effective strategies. From the median values of indices and plot densities, it can be inferred that Taiwan had the most coordinated information campaigning; strict contact tracing measures were applied the longest in Singapore and Taiwan followed by Nigeria, Ethiopia and South Korea, while UK, USA and Sweden had minimal or no contact tracing; testing measures were strongest in the USA and South Korea, followed by Taiwan and Singapore. Even though OxCGRT is a beneficial index for the purpose of analysis, there are numerous factors that might have been overlooked in this analysis. OxCGRT was available early in the pandemic and covered multiple factors. Quantifying the spectrum of responses into a numerical scale required subjective decision making, especially where a narrow range was used. Indices were also limited to the availability of data on governmental responses, which was scarce at the beginning of the pandemic.

The mathematical modelling undertaken demonstrates that a combination of self-isolation, household quarantine and manual contact tracing of all contacts was found to be most effective, reducing transmission by 64%, while random mass testing had only a 5% reduction effect.58 Rapid identification and isolation of cases, quarantining of some contacts and monitoring of others was proven to be effective in supressing expansion of the outbreak.59 Some research points to contact tracing and case isolation being sufficient to control a COVID-19 outbreak.60 However, since the feasibility of this method decreases with a delay between first symptom occurrence and isolation, it has to be implemented as fast as possible. Weekly screening of healthcare workers and at-risk groups reduced the COVID-19 transmission by 25%–33%, while widespread testing of the general population was stated to unlikely have a better effect on the transmission than contact tracing and isolation.61 The data from South Korea, however, potentially disputes this, as their early implementation of widespread screening significantly lowered the COVID-19 cases, without need of going into a lockdown.62 It can be argued that widespread testing coupled with contact tracing and isolation would be effective in capturing both asymptomatic and symptomatic cases. A significant concern is the willingness to engage with track and trace programmes and adherence to isolation if identified as at risk. This may point to the impact of cultural norms, as Asian countries were more accepting, while in Western countries, people viewed it as personal choice. For example, in December 2020, many months after the launch, only one-third of residents in England and Wales had installed the phone app enabling contact tracing.63

Early in the pandemic, it was not clear whether social distancing and lockdown on their own could reduce transmission and delay spread.64 However, the impact of lockdown on transmission in Europe was found to be significant, with 81% reduction rates on average.65 Chinese models have shown that, if implemented early enough, that is, as of 1 January 2020, it could have significantly limited the epidemic, reducing infections by up to 98.9%.66 It was projected that the immediate introduction of different extreme physical distancing measures could push the transmission to households, resulting in a shorter epidemic and a lowered and flattened peak of COVID-19 daily cases.67 Later studies concluded that earlier implementation of lockdown is a better option for reducing COVID-19 spread rather than delayed lockdown after introducing other physical distancing interventions.68 Since almost every country’s response was too slow for this to be achieved, social distancing and lockdown methods were implemented more as a ‘delay than contain’ strategy. Several modelling studies have shown that early implementation of border closures would have limited viral transmission in the early phase of the pandemic; however, for countries that act as global transport hubs, this is more challenging.69 Based on the US early borders closure, and the outcome of their COVID-19 spread, this strategy cannot be effective on its own. While it is important to promote transparency and follow the guidance of international authorities, in critical situations, an immediate response might be necessary. Having extreme measures implemented rapidly for a short period of time is arguably a more effective option long term than delaying and softening the restrictions.

Analysis of deaths associated with COVID-19

Based on publicly available data, as of 13 June 2020, deaths per capita were highest for the UK (611.04) followed by Sweden (480.54) and the USA (343.74). Deaths per capita in South Korea (5.4), Singapore (4.27), Taiwan (0.29), Ethiopia (0.0034) and Nigeria (1.88) were very low. Although difficult to quantify, it is essential to acknowledge also the indirect impact of COVID-19 on death rates, for example, an increase in the rate of suicides. Additionally, data show that people with potentially life-threatening conditions, such as strokes and heart attacks, were less likely to seek urgent hospital care.1 38 39 In the UK, there was a 48% fall in major Emergency department (A&E) visits and a 50% decrease of heart attack admissions in the first half of 2020.70 71 In some countries, the reported number of stroke admissions decreased by 50%–80%.72 Changes to cancer treatment regimes and suspension of clinical trials could have also contributed to death rates.

Country demographics have a significant impact on mortality rates with the number of intensive care unit (ICU) admissions highest in those aged 65 years and above. The percentage of 65+ years old people as a proportion of the population in the selected countries is: USA: 16.2%, UK: 18.7%, Sweden: 20.1%, Taiwan: 15.1%, South Korea: 15.1%, Singapore: 12.3%, Ethiopia: 3.5% and Nigeria: 2.7%.73 74 Given these statistics, it is probable that those countries with predominantly younger populations will have fared better. Death rates are also affected by variation in data collection and should be considered with caution.75 It is acknowledged that this approach relies on countries providing accurate information on numbers of deaths, something that is not possible to determine. It is difficult to account for comorbidities, but the WHO encouraged accurate estimates by normalising a definition of COVID-19 death based on the probable or confirmed COVID-19 infection, unless there was a clear alternative cause of death.

Given higher mortality in the elderly, age-specific mortality would be a more appropriate metric, with age standardised infected population rates being a useful alternative. However, at the beginning of pandemic, there was no detailed data on statistics within different age populations.

This analysis used mortality rates as a metric of effectiveness to governmental response due to the relatively straightforward comparability of values. The numbers of deaths attributable to COVID-19 were chosen because measuring the number of reported cases may not be accurate due to the presence of asymptomatic patients, levels of testing of the general population and a lack of accurate or available diagnostics. Numbers of COVID-19 cases or COVID-19 hospital admissions could also have been used a a benchmark of effectiveness; however, it can be argued that the lack of testing would not have provided an accurate picture of the spread of the pandemic and therefore the effectiveness of any measures implemented. For example, in the USA, a country with the highest number of cases and good availability of tests, the COVID-19 infections count was estimated to be 3–20 times higher than the numbers of confirmed cases.76 Population groups tested, selection criteria and COVID-19 case definition also varied between different countries at the beginning of the pandemic.

COVID-19 deaths to population proportions were the highest for Sweden, UK and USA (figure 3A,B and C). There was a steep acceleration in numbers of deaths in every country apart from Taiwan. Further statistical analysis was performed to quantify the degree of the exponential growth of COVID-19 cases in the selected countries.

Those countries with most effective strategies would presumably have lower goodness of fit to the exponential growth equation (measured in R²), smaller growth rate constant (measured in k) and longer time it took for cases to double (measured in doubling-time).

The smallest rate constants and the largest doubling-time values were in Ethiopia, Singapore and South Korea. Considering the estimates, all countries apart from Ethiopia and Taiwan had exponential growth in COVID-19 death cases.

It is likely that Ethiopia and Nigeria results are significantly downplayed owing to the lack of postmortem diagnostic facilities.

The importance of the individual characteristics of countries should not be underestimated. It is likely that countries such as the USA, which act as a ‘travel hub’, had widespread community transmission prior to index cases having been identified, hence challenging and limiting contract tracing and isolation efforts.

Cultural aspect in the effectiveness of governmental strategies

Hofstede’s cultural dimensions were analysed to estimate the influence of cultural values on the perception of governmental strategies, decision making and adherence to regulations (figure 4).40 Three out of six values were likely have the strongest impact IDV, LTO and IVR. The USA, UK and Sweden, countries with individualistic norms, fared much worse than collectivism-oriented countries like Singapore, Taiwan or South Korea. Concurrently, the three selected Asian countries are also the most long-term oriented ones and have the highest restraint indications explaining their preparedness and compliance to regulations imposed during the COVID-19 pandemic. High collectivism values might have had a role to play in Nigeria and Ethiopia too. It is acknowledged that it is difficult to measure a complex concept such as culture, comprehensively. Hofstede uses only six dimensions to quantify cultural characteristics of citizens within countries. The data are limited to the set of questions asked in surveys and could thus exclude factors such as migration, diversity and residents living in difficult to reach geographical locations.

Cultural norms of collectivism and obedience to restraints in Asia lead to the acceptance of country-wide shutdowns and contact tracing, whereas the UK, USA and Sweden were more concerned about their citizens privacy rights and often value their liberty more than the societal need of, for example, lockdown. As a result of high levels of environmental pollution, and previous disease outbreaks like SARS, Asian people are more used to wearing masks in everyday life, whereas in other countries, it could be judged and criticised as less acceptable. In Asia or Africa, it could be enough to inform the public of the necessary measures, while in UK or USA, the governments need to appeal to their citizens in other ways, such as via adverts and media. It can be argued that Asian countries might have been better prepared, given their experience of dealing with the SARS and MERS pandemics. Having dealt previously with disease outbreaks, governments are probably more willing to implement stricter measures such as travel restrictions and full lockdown.

There are many other confounders and risk factors, such as obesity and the prevalence of chronic diseases, now known to have a significant impact on death rates; however, these factors were not clearly associated with COVID-19 related deaths at the beginning of pandemic. Apart from cultural impact, there could also be other factors, for example, clear and consistent guidance early on in the pandemic that influenced adherence and hence reduced the number of deaths. The impact of political environments and attempts to minimise economic disruption is acknowledged as a factor that may impact on the effectiveness of any strategy; however, analysis of such data is beyond the scope of this research. In the early phases of the pandemic, death rates were assumed to be the key metric of effectiveness measurement; however, governments could have prioritised minimisation of impact on health services or economic disruption instead, hence influencing their approach to strategy design and implementation.

Healthcare resources availability: implications for the impact of COVID-19 on individual countries

Estimates of healthcare resources were considerably lower for Ethiopia and Nigeria, apart from the absolute value spent on healthcare from GDP in Nigeria, which could perhaps be attributed to the country having the seventh’s highest population in the world (figure 5).39 41–57 In South Korea, Singapore and Taiwan, numbers of ICU beds and ventilators were sufficient to deal with the incoming patients with COVID-19. Estimates for the USA are particularly noteworthy as the number of deaths are very high considering both the healthcare budget and the numbers of ICU beds and ventilators available. Based on the estimates for Sweden and UK, it is most likely that both countries were struggling with high numbers of patients with COVID-19, and this could have contributed to higher death rates in both countries. Healthcare resource analysis was heavily impacted by the lack of data availability at the beginning of the pandemic and is limited to the accuracy of information shared. In particular, the number of hospital beds, ventilators and physicians was obtained from non-peer-reviewed journals, and it is acknowledged that countries may have inflated these figures to improve their global image.

Even the best all-encompassing strategies might not be suitable for every country. Singapore, USA, Taiwan, Sweden, UK and South Korea have GDP based on Purchasing Power Parity (PPP) per capita ranging from $44 700 to $103 200, while in Nigeria, it is $6000 and Ethiopia $2500.77 Despite countries with low GDPs that are highly vulnerable to economic stress, with many lives dependent on having to go to work, many still imposed costly lockdowns.78 These countries faced additional challenges; with the economies of other countries also struggling, the funding of welfare and health programmes were cut.

GDP, provision of healthcare resources and cultural norms influence the death rates in every country. While deaths in the USA, UK, Sweden, South Korea, Taiwan, Ethiopia and Nigeria could largely be attributed to the rates of transmission and demographics in a country, numbers in Singapore require a different explanation: 80% of the cases there originated from foreign workers’ dormitories, with community cases accounting for just 10%.79 All diagnosed patients there were kept in isolation facilities with available medical care if required.80 At the beginning of the pandemic, there were more than 18 000-bed spaces. Due to the younger age of the workers and the mild nature of the symptoms, the majority of them did not require hospitalisation, thereby preventing the healthcare system from being overwhelmed.81 In Europe, half of the COVID-19 fatalities occurred in nursing homes; in Singapore, a limited number of nursing homes had cases. As a result of the SARS epidemic in 2003, Singapore and many other Asian countries improved their preparedness and capacity for future outbreaks. Hospital facilities, availability of ICU beds, ventilators medical staff and PPE influenced government strategies to prevent healthcare systems from becoming overwhelmed. PPE numbers of, for example, face masks or gloves were not analysed due to difficulties in finding comprehensive data, the variability of standards across countries and debates about their effectiveness early on in the pandemic. In addition, there was lack of N95 and other high filter efficient respirators availability to general populations, with prioritisation to medical professionals and infected patients. Early in the pandemic, medical masks filter efficiency was shown to be 2%–38%.82 As the pandemic progressed, evidence of the effectiveness of medical masks in reducing COVID-19 transmisibility was demonstrated, however still not beyond a ‘low to moderate’ level. Countries with low GDP do not have the same level of technologies and healthcare capacities, and access to treatment for rural communities is particularly challenging. In spring 2020, it was estimated that at least 121 000 critical care beds were needed for the whole of Africa, while only 9800 ICU beds were available.83 In support of the analysis performed, it was shown that increased mortality rates, lack of hospital beds and ICU bed capacity is strongly correlated with a country’s income status.84 A large UK study showed that one-third of patients admitted to hospital with COVID-19 died.85 In both studies, the vast majority of deaths were due to comorbidities, with cardiac disease 29%, diabetes 19%, chronic pulmonary disease 19% and asthma 14% prevalence in the UK. In addition to these comorbidities, diseases like HIV, tuberculosis and malnutrition are common in low-income countries. The household setting—key for COVID-19 transmission—is different in lower income countries with a substantially higher proportion of multigenerational households having contact with a wide range of age groups.

Geographical factors such as climate and population density, ethnicity and demographic data are factors that were not heavily considered in the analysis as no conclusive data on their impact on COVID-19 related deaths was available early on in the pandemic. It is likely that COVID-19 spread is higher in countries with larger number of highly populated areas and higher population density. However, in 2020, out of the selected countries Singapore, Taiwan and South Korea had 8358, 672.6 and 527.3 people per square metre, respectively, while for the USA and Sweden, the numbers were 36.19 and 24.61, respectively.86 This suggests that highly populated countries do not necessarily have higher COVID-19 spread, highlighting that strict adhesion to constraint measures with quick identification of infected cases should significantly limit the impact of the spread. Furthermore, it suggests that countries with smaller territories may have advantage as a result of more manageable geographies.