Abstract
Background Japan started vaccinations for COVID-19 in late February, 2021, mainly using BNT162b2 mRNA COVID-19 Vaccine.
Object We evaluated COVID-19 vaccination effects on infectiousness in Japan. Method: The effective reproduction number R(t) was regressed on vaccine coverage along with data of temperature, humidity, mobility, and countermeasures. We presumed two definitions for vaccine coverage: at least a single dose with a 12-day delay and a second dose. The study period was February, 2020 through May 16, 2021.
Results Estimation results indicate that vaccine coverage significantly and drastically reduced R(t) for both definitions. Results imply that increased first shot coverage by 1% point decreased R(t) by 0.5 with a 12 day delay. Moreover, increased second shot coverage by 1% point, decreased R(t) by 1 immediately.
Discussion and Conclusion Vaccination effects might be stronger because the periods when vaccination was started and rapid spread coincide with a decreasing phase of R(t).
Introduction
Wide coverage of COVID-19 vaccination has changed the outbreak situation in European countries and US. Unfortunately, vaccination started only in February, 2021 in Japan, among the latest of starting dates in economically developed countries. As of June 6, 2021, the rate of completion for two shots was only about 3% in Japan (Figure 1) [1,2].
Effective reproduction number and vaccine coverage by two definitions.
Note: Line shows the effective reproduction number measured at the right-hand side. Red * denotes vaccine coverage defined by the first dose with 12 days delay. Green + denotes vaccine coverage defined by the second dose. Two vaccine coverages were measured at the left-hand side. Because daily vaccine coverage were not reported in weekend and national holidays, data of vaccine coverages were missing in these days.
Fortunately, prevalence of COVID-19 has been much lower in Japan than in European countries and the US. For instance, prevalence in the US was approximately 10% as of June 6, 2021, but only 0.6% in Japan, which had used no lock down. Although reasons for such differences remain unknown, vaccine efficacy in Japan might be different from efficacy in US and European countries.
Moreover, vaccination has provided not only personal protection, but also herd immunity to unvaccinated persons. The former was investigated in clinical trials [3], but the latter might depend on the surrounding society. Therefore, the present study examined vaccine effectiveness, including herd immunity in Japan. To measure the vaccine effectiveness, we estimated the effects of vaccine coverage on SARS-CoV-2 infectiousness in Japan.
As countermeasures against the COVID-19 outbreak in Japan, school closure and voluntary event cancellation were adopted from February 27, 2020 through the end of March. Large commercial events were cancelled. Subsequently, a state of emergency was declared for April 7 through 25 May, stipulating voluntary restrictions against leaving home. Consumer businesses such as retail shops and restaurants were shut down. During this period, the first peak of infection was reached on April 3. Infections subsequently decreased until July 29. The so-called “ Go To Travel Campaign” (GTTC) started on July 22 as a 50% subsidized travel program aimed at supporting sightseeing businesses with government-issued coupons for use in shopping at tourist destinations. It was expected that the campaign might expand the outbreak. Thereafter, GTTC continued to the end of December, by which time a third wave of infection had emerged. The third wave in December, which was larger than either of the preceding two waves, reached its highest peak at the end of December. Therefore, GTTC was inferred as the main reason underlying the third wave [4].
To force the third wave lower, the second emergency status was declared on January 8, 2021 to March 15, 2021. However, the fourth wave emerged, probably because of the spread of variant strains at the end of February. Moreover, to support the hosting of the Olympics and Paralympics games in Tokyo in July, a third emergency state was declared on April 25, 2021.
Nevertheless, although results have been mixed, some results of earlier studies suggest that COVID-19 might be associated with climate conditions [5–8]. If that were true for Japan, then GTTC might not be the main reason for the third wave.
Moreover, mobility was inferred as the main cause of the outbreak dynamics, at least in the first wave in Japan [9] and throughout the world [10–13]. This study was conducted to estimate vaccine coverage effects on SARS-CoV-2 infectiousness for the outbreak in Japan.
Methods
The numbers of symptomatic patients reported by the Ministry of Health, Labour and Welfare (MHLW) for January 14 – May 16, published [14] as of June 6, 2021 were used. Some patients were excluded from data for Japan: those presumed to be persons infected abroad or infected as Diamond Princess passengers. Those patients were presumed not to represent community-acquired infection in Japan. For some symptomatic patients with unknown onset dates, we estimated onset dates from an empirical distribution with duration extending from onset to the report date among patients for whom the onset date had been reported.
Onset dates among patients who did not report this information and reporting delay were adjusted using the same procedures as those used for our earlier research [15,16]. As described hereinafter, we estimated the onset dates of patients for whom onset dates had not been reported. Letting f(k) represent this empirical distribution of incubation period and letting Nt denote the number of patients for whom onset dates were not published and available at date t, then the number of patients for whom the onset date was known is t-1. The number of patients with onset date t-1 for whom onset dates were not available was estimated as f(1)Nt. Similarly, patients with onset date t-2 and for whom onset dates were not available were estimated as f(2)Nt. Therefore, the total number of patients for whom the onset date was not available, given an onset date of s, was estimated as Σk=1f(k)Ns+k for the long duration extending from s.
Moreover, the reporting delay for published data from MHLW might be considerable. In other words, if s+k is larger than that in the current period t, then s+k represents the future for period t. For that reason, Ns+k is not observable. Such a reporting delay engenders underestimation of the number of patients. For that reason, it must be adjusted as Σk=1t-sf(k)Ns+k /Σk=1t-sf(k). Similarly, patients for whom the onset dates were available are expected to be affected by the reporting delay. Therefore, we have Ms|t /Σk=1t-sf(k), where Ms|t represents the reported number of patients for whom onset dates were period s as of the current period t.
We defined R(t) as the number of infected patients on day t divided by the number of patients who were presumed to be infectious. The number of infected patients was calculated from the epidemic curve by the onset date using an empirical distribution of the incubation period, which is Σk=1f(k)Et+k, where Et denotes the number of patients for whom the onset date was period t. The distribution of infectiousness in symptomatic and asymptomatic cases g(k) was assumed to be 30% on the onset day, 20% on the following day, and 10% for the subsequent five days [17]. Then the number of infectiousness patients was Σk=1g(k)Et-k. Therefore, R(t) was defined as Σk=1f(k)Et+k/Σk=1g(k)Et-k.
We use average temperature and relative humidity data for Tokyo during the day as climate data because national average data were not available. We obtained data from the Japan Meteorological Agency (https://www.data.jma.go.jp/gmd/risk/obsdl/index.php). We identified several remarkable countermeasures in Japan: two emergency status declarations, GTTC, and school closure and voluntary event cancellation (SCVEC). The latter, SCVEC, extended from February 27 through March: this countermeasure required school closure and cancellation of voluntary events, and even of private meetings. Then the first state of emergency was declared April 7. It ceased at the end of May. It required voluntary restriction against going out, school closures, and shutdown of businesses. To subsidize travel and shopping at tourist destinations, GTTC started on July 22. It was halted temporarily at the end of December. The second state of emergency was declared on January 7, 2021 for the 11 most affected prefectures. This countermeasure required restaurant closure at 8:00 p.m., with voluntary restrictions against going out, but did not require school closure. It continued until March 21, 2021. The third state of emergency was declared on April 25, 2021 for four prefectures: Tokyo, Osaka, Hyogo, and Kyoto. Later, the application areas were extended gradually, but never covered the entirety of Japan.
To clarify associations among R(t) and vaccine coverage as well as climate, mobility, and countermeasures, we regressed the daily R(t) on dummy variables for daily vaccine coverage, daily climate, mobility, and countermeasures using ordinary least squares regression. Temperatures were measured in degrees Celsius, humidity, and mobility as percentages in regression, not as standardized. When some variables were found to be not significant, we excluded them and estimated the regression line again. If some variables were not significant in the full specification estimation, then we estimated them again stepwisely without those non-significant variables.
We examine two definitions for vaccine coverage: rate of receiving at least a single dose with a 12 day delay and the completion rate of the second dose. Delay in the former was the response period for increasing antibody [18]. If a vaccine protects the recipient from infection perfectly, the estimated coefficient of vaccine coverage should be 0.01 if one assumes an average of R(t) with no vaccination in the study period. That means that increasing vaccine coverage by 1% point was expected to reduce R(t) by 1% point. If the estimated coefficients of vaccine coverage were smaller than 0.01, then it might reflect imperfect personal prevention. Conversely, if the estimated coefficients of vaccine coverage were smaller than 0.01, it might mean that herd immunity contributed to prevention from infection among non-recipients.
We also examined another delay in vaccine coverage to verify robustness. We found 14 days of delay in the first dose and 7 days of delay in the second dose. We adopted 5% as the level at which to infer significance.
Results
Figure 2 presents an empirical distribution of the duration of onset to reporting in Japan. The maximum delay was 31 days. Figure 3 presents an empirical distribution of incubation periods among 91 cases for which the exposed date and onset date were published by MHLW in Japan. The mode was six days. The average was 6.6 days. The calculated R(t) is shown in Figure 1.
Empirical distribution of duration from onset to report by MHLW, Japan.
Note: Bars represent the probability of duration from onset to report based on 657 patients for whom the onset date was available in Japan. Data were obtained from MHLW, Japan.
Empirical distribution of the incubation period published by MHLW, Japan.
Notes: Bars show the distribution of incubation periods for 91 cases for which the exposure date and onset date were published by MHLW, Japan. Patients for whom incubation was longer than 14 days are included in the bar shown for day 14.
Table 1 presents estimation results of two definitions of vaccine coverage to R(t). In both definitions, climate conditions and the third emergency status were not significant in specifications (1) and (3). Therefore, we deleted these variables as explanatory variables. All remaining variables in the specification (2) and (4) were significant.
Estimation results of R(t) on vaccine coverage with the climate condition, mobility, and countermeasures
Mobility was positive and significant. The first two emergency status periods and GTTC were found to be negative and significant. However, SCVEC was found to be significant and positive. Vaccine coverage was significantly negative, as expected. However, its estimated coefficients were approximately 0.5 in the specification (2) and one in the specification (4). These were 50–100 times larger than expectations from personal prevention.
Another definition adopted for the delay period until sufficiently increasing antibodies to protect from infection yielded similar results to those presented above. The estimated coefficients for vaccine coverage were −0.44365 (p-value was 0.000) for the first doze with 14 days delay and −0.97141 (0.000) for the second doze with 7 days delay.
Discussion
Estimation results showed that vaccine coverage was significantly negative and that the estimated coefficients were far larger than expected from personal prevention. That finding might mean that herd immunity strongly reduced R(t). However, it might be biased by a shorter period with vaccine in Japan because the period at which vaccination was started and spread rapidly coincided with the decreasing phase of R(t). Vaccination effects might more strongly incorporate another effect to reduce R(t), but might not be included as explanatory variables.
Because we presumed a linear function of vaccine coverage and presumed that the estimated coefficients were larger than expected, the predicted R(t) is expected to be negative. That result was inconsistent with definition of R(t). Therefore, such a linear association should disappear with increased vaccine coverage. This nonlinearity might also diminish herd immunity effects.
The third emergency status was not significant. That finding is probably attributable to overlap of the period of the third emergency and initiation of the vaccine campaign. Consequently, multicollinearity might occur among these two variables. Actually, if vaccine coverage is dropped in specification (1) or (3), then the estimated coefficient of the third emergency status was −0.89263; its p value was 0.000. The value was comparable to the first two emergency status periods.
The present study has some limitations. First, as described above, effects of vaccine coverage and estimated herd immunity might be overestimated. Accumulation of the data will probably resolve this problem.
Secondly, readers must be reminded when interpreting the obtained results that our obtained results do not indicate causality. This study demonstrated that some association exists among vaccine coverage and infectiousness. That finding does not necessarily mean that vaccine coverage reduces infectiousness: it might imply that lower infectiousness, for instance, an effective reproduction number being smaller than one, induces people to approach vaccination more enthusiastically.
Conclusion
Because the period when vaccination was started and spread rapidly coincides closely with the decreasing phase of R(t), vaccination effects might be estimated as stronger than they actually were. Accumulation of data and longer periods of data might decrease that misleading effect.
The present study is based on the authors’ opinions: it does not reflect any stance or policy of their professionally affiliated bodies.
Data Availability
Japan Ministry of Health, Labour and Welfare. Press Releases.(in Japanese)
Ethical considerations
All information used for this study was published data on the web were used. There is therefore no ethical issue related to this study.
Competing Interest
No author has any conflict of interest, financial or otherwise, to declare in relation to this study.
Acknowledgments
We acknowledge the great efforts of all staff at public health centers, medical institutions, and other facilities who are fighting the spread and destruction associated with COVID-19.
Footnotes
ICMJE Statement Contributors JK was responsible for the coordination of the study and responsible for the data setting. YO developed the model and TS illustrated the results. All authors contributed to the writing of the final manuscript.
