Results and discussion

To observe the co-movement of policy indices and rescaled daily confirmed cases, we provide the time series plots of these two variables together. We observe that the statuses of policy indices are associated with daily new cases in countries and region. We provide illustrative descriptions that appear in Figure 2.

Hong Kong: To stop the transmission of COVID-19 from overseas to Hong Kong, a mandatory quarantine was imposed for travellers from mainland China and South Korea in February 2020 and for travellers from Iran, Italy and other regions in March 2020 [Reference Adam15]. There were also restrictions on public gatherings implemented in April 2020. Therefore, it is not surprising that SI, GRI and CHI increased rapidly in the first 3 months of 2020. Since then, various hygiene measures for controlling the transmission of the disease have been introduced, and there was a slight upward trend in GRI and CHI after mid-2020. Early 2022 presented a sharp increase in confirmed cases, because of the transmission of the Omicron variant in Hong Kong. A few months before the spreading of Omicron, the three indices had increased slowly, probably indicating that there was a tightening of hygiene control measures in late-2021 after recording a few imported Omicron cases.

Japan: The number of confirmed cases surged, exceeding 200 across Japan as of 27 February 2020, excluding the over 700 confirmed cases from the Diamond Princess cruise ship. The Japan government adopted various measures to limit or prevent the outbreak in the first half of 2020, which included inaugurating a Joint Research Coronavirus Task Force to supervise its response to the pandemic and the suspension of all Japanese schooling from elementary to high schools from 27 February until early April 2020 [16]. On 7 April 2020, the Japan government proclaimed a 1-month state of emergency for Tokyo and several prefectures. The state of emergency was lifted for an increasing number of prefectures during May 2020, extending to the whole country by 25 May 2020 [Reference Yamamoto17]. That is why we observe growth and a drop-down in Japan for SI, GRI and CHI in the first 6 months.

South Korea: After 18 February 2020, when the virus was initially confirmed in Daegu, the number of patients with COVID-19 increased rapidly. To respond to the rapid rise in the number of COVID-19 cases, the South Korea government puts the COVID-19 alert level at the highest (level 4) on 23 February 2020, enhanced social distancing measures and infection control measures in hospitals, imposed travel restrictions, cancelled social events and delayed the school start time [Reference Shim18]. We observe that SI, GRI and CHI shot up rapidly in the first 3 months of 2020. In February 2022, new confirmed cases surged to six-digit figures in South Korea.

Taiwan: COVID-19 infections cropped up only sporadically before May 2021. However, daily new confirmed case numbers soared into the triple digits on 15 May 2021. The Central Epidemic Command Center (CECC) raised level 3 COVID-19 alert for Taipei City and New Taipei City, strengthened national restrictions and measures, effective from 15 May, and further raised the epidemic warning to level 3 nationwide from 19 May 2021. Heightened measures and regulations were introduced across Taiwan to reduce community transmission [19, 20]. The government has implemented a mandatory wearing-mask policy, stringent restrictions on all kinds of gatherings, closure of entertainment venues and limiting restaurants to take-out service. CECC partially lifted certain restrictions for level 3 (starting on 13 July 2021 and ending on 26 July 2021). All these related policies reveal the rising and then plummeting SI, GRI and CHI during this period.

The Omicron variant was the predominant variant in Hong Kong, Japan and South Korea from late January 2022. We observe steep increase curves for daily new confirmed cases, while the Omicron outbreak in Taiwan is still in its early stages. The daily count of new cases surged as high as a few multiples of 10 000 outside this study's period (January 2020 to March 2022).

As the 7-day MA removes the fluctuation and derives a more stable trend, we provide Spearman's rank correlation coefficient to measure the association between the 7-day MA of new cases (new deaths) and each index (Tables 1–4). The results indicate the effectiveness of government policies through the specified range when the negative correlation is significant. Findings show significantly negative association measurements during different waves for Hong Kong, Japan, South Korea and Taiwan, but most effects are on confirmed cases with only a few situations on deaths. The exceptions appear on the fourth wave (1 July 2021 to 26 March 2022) in South Korea. Both confirmed and death cases in South Korea have significantly negative associations with the three indices (SI, GRI and CHI) during this period.

Table 1.

Spearman's correlations between 7-day MA of new cases (new deaths) and policy indices for Hong Kong

*, **, and *** denote significance at the 5%, 1%, and 0.1% levels, respectively.

Governments did enforce stringent measures in the second and third quarters of 2020 after the COVID-19 virus broke out and quickly responded to the outbreak, encompassing restrictions on business activities, suspension of schooling, travel restrictions and social distancing rules (Fig. 3a). Interventions, such as enforced isolation, in Hong Kong did work truly well in the early pandemic to contain the virus over the second and third waves in terms of reducing confirmed cases (Table 1), as we find significantly negative associations between the three indices (SI, GRI and CHI) and daily confirmed cases. Since early 2022, Hong Kong has been experiencing huge and fatal virus surges via the highest total confirmed deaths of COVID-19 per million people in our study ([Reference Taylor21] and Fig. 3b).

The number of confirmed cases increased substantially across Japan in late-February 2020. The Japan government adopted various hygiene measures to limit or prevent the outbreak in the first half of 2020. These could explain why we observe a significantly negative association between new confirmed cases and policy in the first wave in Table 2 for this country.

Table 2.

Spearman's correlations between 7-day MA of new cases (new deaths) and policy indices for Japan

*, **, and *** denote significance at the 5%, 1%, and 0.1% levels, respectively.

We observe evidence of the impact of non-pharmaceutical measures in controlling the spread of the virus in Japan during the whole study period except during the second and sixth waves (Table 2). The ‘Go To Travel’ campaign originally ran in the second half of 2020 to support the travel industry in Japan. We do not observe any significant association between new cases and the index in the second wave (Table 2). The subsidy programme was suspended on 28 December 2020, complicating government efforts to address the economic fallout from the pandemic and extending the State of Emergency during the third wave. This may explain the negative association between new cases and SI, GRI and CHI during the third and fourth waves. Daily new cases in Japan soared following the start of the Olympics in the last week of July and in early August of 2021. Consequently, the policy indices of GRI and CHI fail to show effective prevention of confirmed cases in the fifth wave. Because of a state of emergency declaration and a spectator-less Tokyo 2020 Olympic Games, the number of confirmed cases in Tokyo dropped after the fifth wave. The negative association between new cases and SI also confirms this fact. Currently, in the sixth wave Japan has seen a rapid increase in new confirmed cases by more than three times over the previous wave, fuelled by the surge of Omicron. None of the indices negatively affect new confirmed cases in the sixth wave.

The economic reactivation and public health control of the COVID-19 pandemic have been asymmetric. When social activities were reactivated to boost the economy, the second and third waves in South Korea started, but there were differences in the public health response [Reference Seong12]. Our results of the second and third waves in Table 3 confirm a negative association between new cases and each index in the third wave, while a positive association exists between new cases and each index in the second wave. The results reveal South Korea's trending down of new death cases in the fourth wave and the entire period (Table 3). South Korea presents two seemingly contradictory pandemic indicators. It recorded over 100 000 new confirmed cases every day from 17 February 2022 to March 2022, as seen from the trend in Figure 3a. At the same time, the country has one of the lowest virus death rates globally (https://covid19.who.int/table).

Table 3.

Spearman's correlations between 7-day MA of new cases (new deaths) and policy indices for South Korea

*, **, and *** denote significance at the 5%, 1%, and 0.1% levels, respectively.

During the first wave, Taiwan had nearly unblemished success at keeping the COVID-19 pandemic at bay with the world's longest track of case-free days. Due to the outbreak of infections derived from tea houses in the red-light district of Wanhua in Taipei on 23 April 2021, Taiwan was in the grip of its first major COVID-19 surge. The CECC announced a nationwide level 3 epidemic alert on 19 May 2021, but later eased some restrictions for level 3 from 13 July to 26 July 2021. The intervention results during the second wave seem to have been not so effective, showing insignificant associations between cases (deaths) and each index. The intervention during the third wave by Taiwan was relatively significant compared to the previous waves (Table 4).

Table 4.

Spearman's correlations between 7-day MA of new cases (new deaths) and policy indices for Taiwan

*, **, and *** denote significance at the 5%, 1%, and 0.1% levels, respectively.

A lead–lag effect, especially in economics, describes when one (leading) variable cross-correlates with the values of another (lagging) variable at later times. This study aims to quantify the time-lag effect reflected in new confirmed cases from authorities' response to the COVID-19 pandemic using SI (GRI or CHI) as an indicator of policy effectiveness. From the healthcare perspective, there are COVID-19-related studies that do investigate the time lag between the mobility response and government policy [Reference Bian22] and the lead–lag relationship between physical and mental health [Reference Luo, Chui and Li23]. Our findings show no policy lag in Taiwan (Fig. 5). At the same time, there is no significant relationship between new confirmed cases and each index for Hong Kong, Japan and South Korea.

In May 2020 14% (79 out of 557) of COVID-19-related deaths in Japan took place in long-term care facilities [24]. Long-term care facilities with elderly residents are especially high-risk facilities in the event of COVID-19 viral transmissions. This may explain why the case fatality ratio of Japan is relatively higher than in other East Asia countries during the first wave (Fig. 4). For comparison, we also include the case fatality ratio of the elderly group for Sweden in Table 5. By the beginning of December 2020, more than 7000 people had died of COVID-19 in Sweden. Of these, almost 90% were 70 years or older, half were living in a long-term residential care facility, and just under 30% received home help services [Reference Melin25].

Table 5.

Information on the elderly group and test (per 1000) for all ages

The case fatality ratios by age group in Japan, Hong Kong, South Korea and Taiwan as of March 2022 appear in Figure 6.Footnote ¹ This confirms an earlier report [Reference Melin25] that case fatality ratios exhibit age group-specific disparity and a high percentage of deaths in elderly people. It shows that the case fatality ratios of the elderly group (age ≥70) in Japan, Hong Kong and South Korea are 4.69%, 4.72% and 1.48%, respectively, while the case fatality ratio of the elderly group (age ≥70) in Taiwan is 25.01% (Table 5). The proportion of deaths in the elderly group (age ≥70) of Japan as of March 2022 is 86.53% of confirmed deaths, in which the first wave may dominate. The reason for elderly deaths could be that some countries performed mass testing or initiated substantial prevention measures, which helped avoid a large number of potential deaths; it could also be that insufficient vaccinations led to them being unable to prevent these issues or families abandoning their elderly amid this pandemic [Reference Brooke and Jackson26].

We also provide total tests per thousand (total tests/population × 1000) as of March 2022 in the fifth column of Table 5. Hong Kong and South Korea have undergone mass testing compared to Japan and Taiwan. At the start of the pandemic in 2020, South Korea was hit by some of the worst early COVID-19 outbreaks, but it appeared to have brought them under control through mass testing and aggressive contact tracing. However, the South Korea government announced on 7 February 2022 that it was abandoning the test-and-trace system in the face of a surge of Omicron cases that threatened to overwhelm the country's health system [Reference Smith and Lee27]. Authorities of South Korea prioritise tests for people aged 60 or older.

We add the proportion of elderly in East Asia and Sweden in the last column of Table 5. Unfortunately, we only can provide the proportion of people aged over 65 in 2021 instead of the elderly group aged over 70. We observe that the proportion of elderly people in 2021 ranges from around 15% to 30% and also note that the proportion of the elderly in Japan is the highest in this study. There is clearly no obvious association between this proportion and the case fatality ratio for the elderly group.

It is impossible to examine how the Hong Kong and Taiwan governments are responding to COVID-19 without considering their past experience with the SARS epidemic in 2003. Hong Kong and Taiwan both faced severe consequences from the SARS outbreak [Reference Hsieh, Chen and Hsu28, Reference Leung, Hedley and Ho29]. Both governments have effectively provided non-pharmaceutical interventions in terms of the negative association between COVID-19 confirmed cases and the indices. Table 1 reveals such negative associations between COVID-19 confirmed cases and the indices of Hong Kong during the second, third and fifth waves. However, the proportion of deaths in the elderly group (age ≥70) of Hong Kong is 87.51% of confirmed deaths, which is similar to Sweden's elderly group at 88.75% (Table 5). As of 25 March 2022, the proportion of people aged over 80 in Hong Kong who have received two vaccine doses or more is 53.19% [30]. Around one-half of people aged 80 and over in Hong Kong have been given two doses or more of a vaccine before the Omicron surge, compared to over 90% in Singapore. A recent report (accessed 27 March 2022) by the Ministry of Health, Singapore shows that 16% of confirmed death in those aged 80 and over in Singapore were non-fully vaccinated, while 2.8% and 0.42% were fully vaccinated-without booster and fully vaccinated-with booster, respectively (Fig. 7).

Fig. 7.

Proportion of COVID-19 cases in Singapore who died, by age and vaccination status.

There is a steep case fatality ratio for Taiwan after May 2021. The case fatality ratio of March 2022 in the elderly group (age ≥70) is 25.01%, which is the highest case fatality ratio for the elderly group in our study. Taiwan began Oxford-AstraZeneca COVID-19 vaccinations on 22 March 2021, while the Moderna COVID-19 vaccine rollout started on 9 June 2021. Its CECC announced a COVID-19 vaccine distribution strategy on 20 June 2021. Table 6 presents the 10 approved priority groups.

Table 6.

COVID-19 vaccination rollout in Taiwan announced by the CECC on 20 June 2021

^a Unit is 10 000 people.

The elderly people aged 75 and over and pregnant women rank in the sixth priority group, while those aged 65–74 ranked as the eighth priority, and adults aged 50–64 with the largest number rank at tenth priority. This may explain the highest case fatality ratio of March 2022 in Taiwan in the elderly group, as they are insufficiently vaccinated. Based on the death rate of Hong Kong's largely unvaccinated elderly group [Reference Taylor21] and the lower priority order for Taiwan's elderly group after vaccines rolled out, we draw a conclusion that a fully vaccinated person with a booster is presently the best strategy for reducing deaths from COVID-19 in the elderly group.