Development of a COVID-19 vaccine effectiveness and safety assessment system in Japan: The VENUS study

Background There are currently no COVID-19 vaccine assessment systems in Japan that allow for the active surveillance of both vaccinated and unvaccinated persons. Herein, we describe the development of Japan’s first COVID-19 vaccine effectiveness and safety assessment system with active surveillance capabilities. Methods The Vaccine Effectiveness, Networking, and Universal Safety (VENUS) Study was developed as a multi-source database that links four data types at the individual resident level: Basic Resident Register (base population information), Vaccination Record System (vaccination-related information), Health Center Real-time Information-sharing System on COVID-19 (HER-SYS; information on COVID-19 occurrence), and health care claims data (information on diagnoses, hospitalizations, diagnostic tests, and treatments). These data were obtained from four municipalities. Individual residents were linked across the data types using five matching algorithms based on names, birth dates, and sex; the data were anonymized after linkage. To ascertain the viability of the VENUS Study’s database for COVID-19 vaccine assessments, we examined the trends in COVID-19 vaccinations, COVID-19 cases, and polymerase chain reaction (PCR) test numbers. We also evaluated the linkage rates across the data types. Results Our multi-source database was able to monitor COVID-19 vaccinations, COVID-19 cases, and PCR test numbers throughout the pandemic. Using the five algorithms, the data linkage rates between the COVID-19 occurrence information in the HER-SYS and the Basic Resident Register ranged from 85·4% to 91·7%. Conclusion If used judiciously with an understanding of each data source’s characteristics, the VENUS Study can provide a viable data platform that facilitates active surveillance and comparative analyses for population-based research on COVID-19 vaccine effectiveness and safety in Japan.


Introduction
The COVID-19 pandemic is one of the greatest public health crises in modern history. As of July 2022, the estimated number of cumulative cases and deaths worldwide has exceeded 545 million and 6.3 million, respectively [1]. Several COVID-19 vaccines have been authorized for use in many countries, and other vaccine candidates are being tested in phase III clinical trials. Pfizer-BioN-Tech's BNT162b2 mRNA vaccine and Moderna's mRNA-1273 vaccine showed efficacies of over 90% against symptomatic COVID-19 in their respective trials [2,3]. The Japanese government approved the BNT162b2 vaccine, mRNA-1273 vaccine, and ChAdOx1-S recombinant vaccine (Oxford-AstraZeneca) in the first half of 2021, with vaccinations beginning a few days after approval. Although vaccination numbers rose quickly in Japan, the proportion of second-dose vaccinated persons has stalled at 76% in July 2022 [4]. In order to prevent and reduce COVID-19 transmissions within Japan, there is a need to increase vaccination coverage, especially among those who are hesitant to vaccinate.
The provision of accurate scientific evidence may help to overcome vaccine hesitancy and promote vaccinations. In addition to the evidence provided by clinical trials, there is also a need for information on vaccine effectiveness and safety after rollout. With the aim of providing real-world data, many countries have conducted large-scale database research on COVID-19 vaccines in their populations, including the US [5], the UK [6][7][8][9], Israel [10][11][12], Canada [13,14], Sweden [15], Spain [16,17], Italy [18], and Qatar [19][20][21]. In stark contrast to these countries, Japan has yet to establish a data-driven system to assess the post-authorization effectiveness and safety of COVID-19 vaccines. Even though Japan's national strategy for vaccine safety monitoring relies solely on the passive surveillance of vaccinated individuals through spontaneous reports, the Ministry of Health, Labour and Welfare (MHLW) has declared that this strategy ''is considered to work well" [22]. However, it may be imprudent to assert that a vaccine safety monitoring system without active surveillance capabilities is working ''well". As of July 2022, Japan has the eighth highest number of vaccinated persons worldwide with over 103 million people having received at least one COVID-19 vaccine dose [23]. Despite this large population of vaccinated persons, Japan is incapable of producing population-based evidence on COVID-19 vaccines due to the lack of an active surveillance system. This represents an overlooked capacity for research that could substantially contribute to the global understanding of COVID-19 vaccines.
In lieu of any government-led programs, we have developed Japan's first COVID-19 vaccine effectiveness and safety assessment system with active surveillance capabilities. This system builds upon the foundation of the Vaccine Effectiveness, Networking, and Universal Safety (VENUS) Study, which is a data platform that links health care claims data and vaccination records [24]. At its inception, the VENUS Study was designed to evaluate routine vaccinations such as those for influenza and pneumococcal disease [25]. We have since expanded on this study to also assess COVID-19 vaccines. Herein, we describe the development and characteristics of the VENUS Study as a platform for the assessment of COVID-19 vaccines in Japan.

Methods
We sought to develop a system that can identify vaccinated and unvaccinated persons within a population, and actively monitor them to assess outcomes such as COVID-19 occurrence, hospitalizations, and vaccine-related adverse events. However, this requires an array of data items that are not available within any one database in Japan. To compile the necessary data items, we adopted the approach of linking individual persons across several existing databases at the municipal level.

Data sources
The data sources used to develop this system were the (i) Basic Resident Register, (ii) Vaccination Record System (VRS), (iii) Health Center Real-time Information-sharing System on COVID-19 (HER-SYS), and (iv) health care claims data.
To collect these data, we used the system infrastructure of the Longevity Improvement & Fair Evidence (LIFE) Study, which is a multi-region database project managed by Kyushu University, Japan [26]. To analyze COVID-19 vaccines, we invited municipalities that are at or above the administrative division of ''core city" (i.e., population 200,000-500,000) to join the VENUS Study.

Basic Resident Register
We used each municipality's Basic Resident Register to identify the base population. In Japan, the Basic Resident Register is the national registry of citizens and long-term residents, and includes each person's resident registration number, name in kanji (Japanese logographic characters), name in kana (Japanese syllabic characters), birth date, and sex. Resident registration numbers are unique identifiers for individual residents, but are not commonly used in other databases. We requested participating municipalities to provide the Basic Resident Register data for all registered residents from January 1, 2019 onward.

Vaccination Record System: VRS
The Japanese government developed the VRS in 2021 to record the vaccination statuses of residents, including vaccination dates, locations, and types of vaccines. The VRS is a cloud-based system that municipal governments use to manage their vaccinationrelated data. The VRS includes resident registration numbers, which allow for direct data linkage with the Basic Resident Register.
Each resident in Japan is sent a vaccination voucher by their municipal government. The residents bring these vouchers to a participating medical institution or vaccination site in order to receive the vaccine. Each voucher contains a barcode or QR code that, when scanned by a health worker, automatically registers the vaccine recipient and vaccination-related data into the VRS. The health worker is also able to verify that the scanned data matches the data on the vaccine voucher.
The data collection period for VRS data in each municipality began from the initial date of COVID-19 vaccination rollout (February 2021) until the date on which data were collected. The dates of data collection varied among the municipalities, and ranged from February 15, 2022 (City C) to May 2, 2022 (City D).

Health Center Real-time Information-sharing System on COVID-19: HER-SYS
In the HER-SYS, digital information on COVID-19 cases (e.g., symptoms and activity history) are recorded by health care providers. These records are centrally managed by the MHLW, and can be shared among health care institutions, public health centers, prefectural governments, and the national government. Upon diagnosing a person with COVID-19, physicians are required to submit a Notification of Occurrence to public health centers under the Infectious Diseases Act. At the time of this study, it was mandatory to record all positive COVID-19 cases in Japan. However, records have only become required for high-risk groups (e.g., persons aged ! 65 years, hospitalized patients, pregnant women, and persons with underlying conditions) since September 2022. With the exception of a few prefectures, most prefectures have chosen to adopt these new target groups for recording.
Public health centers are responsible for storing and managing these data under the HER-SYS. The HER-SYS does not include resident registration numbers, but contains the following identifying information: name in kanji, name in kana, birth date, and sex. In addition to the number of COVID-19 cases, the HER-SYS also contains patient-reported COVID-19 vaccination statuses and underlying conditions.
The data collection period for HER-SYS data in each municipality began from the date of the first COVID-19 case until the date on which data were collected. The dates of data collection varied among the municipalities, and ranged from March 11, 2022 (City C) to April 28, 2022 (City D).

Health care claims data
Under Japan's universal health insurance system, insurance is provided to all residents through three schemes: employer-based insurance (for salaried employees of businesses), the National Health Insurance System, and the Latter-Stage Older Persons Health Care System. The region-based National Health Insurance System, in which municipal governments fulfill the role of insurer, provides coverage to the self-employed, unemployed, retired persons aged 65-74 years, and their dependents. The Latter-Stage Older Persons Health Care System, in which prefectural governments fulfill the role of insurer, provides coverage to all persons aged ! 75 years. Although municipal governments do not serve as insurers for the Latter-Stage Older Persons Health Care System, they are able to acquire health care claims data produced under this system for the purpose of analyzing health issues within their regions. For the VENUS Study, health care claims data are collected from enrollees of the National Health Insurance System and the Latter-Stage Older Persons Health Care System by the participating municipalities. Each individual is assigned a unique research identification code based on their name, birth date, and sex to allow their linkage across the data sources. This identification code also allows the tracking of individuals that switch from the National Health Insurance System to the Latter-Stage Older Persons Health Care System after their 75th birthday. Although a large proportion of enrollees in these two insurance systems are aged ! 65 years, these schemes also cover the self-employed and their dependents. Therefore, the study population encompasses all age groups. Japanese health care providers send electronic claims to insurers to be reimbursed for the provision of insurance-covered care. These claims data are produced using a standardized format for all insurance schemes. In the VENUS Study, we obtained claims data for all enrollees of the National Health Insurance System and the Latter-Stage Older Persons Health Care System regardless of whether their data indicated COVID-19 occurrence. The data contain information on diagnoses, hospitalizations, diagnostic tests, and treatments. The claims data do not include resident registration numbers, but contain the following identifying information: name in kanji, name in kana, birth date, sex, and insurance enrollee number. However, records of names in kana are not compulsory, and are not always included for all individuals.
The data collection period for health care claims data in all participating municipalities was from January 1, 2019 until January 31, 2022. We collected data from the start of 2019 to obtain information on the enrollees' underlying conditions prior to the pandemic.

Data linkage
Ideally, the resident registration numbers in the Basic Resident Register should form the basis for linking the various data types as this would allow for the easy and accurate identification of all residents within a municipality. However, the lack of resident registration numbers in the HER-SYS and health care claims data precludes the direct linkage of data. Therefore, we used a sequential series of five matching algorithms to link the four data types. The first algorithm (Algorithm 1) involved the linkage of individuals across these data sources if they had complete matching of four criteria: name in kanji, name in kana, birth date, and sex. Each individual who could be linked across the data sources was assigned a unique research identification code to facilitate subsequent analyses. However, a preliminary analysis of the HER-SYS showed that it contains numerous errors such as missing birth dates, missing names in kana, and erroneous names in kanji. Furthermore, an individual's name may have different recorded kanjis in the Basic Resident Register and the health care claims data due to the use of rare kanji characters that are unavailable in standard character code sets. Therefore, individuals who could not be matched in Algorithm 1 were carried over to the second algorithm (Algorithm 2), which matched individuals based on name in kanji, birth date, and sex. Next, individuals who could not be matched in the first two algorithms were carried over to the third algorithm (Algorithm 3), which matched individuals based on name in kana, birth date, and sex. Because of the missing birth dates in the HER-SYS, the fourth algorithm (Algorithm 4) matched individuals based on name in kanji and sex only. Finally, individuals who could not be matched in the first four algorithms were carried over to the fifth algorithm (Algorithm 5), which matched individuals based on name in kana and sex only. If there were ! 2 matching candidates for any algorithm, we excluded all candidates from the matching process.

Data evaluation
In order to ascertain the viability of the VENUS Study's database for COVID-19 vaccine assessments, we examined the daily trends in vaccinations using VRS data, daily trends in COVID-19 cases using HER-SYS data, as well as the numbers of individuals who had undergone a polymerase chain reaction (PCR) test and publicly funded hospitalization using health care claims data. COVID-19 cases were identified according to age group (0-19, 20-64, 65-74, and ! 75 years) using the HER-SYS, with the date of a positive PCR test result set as the sample collection date. From the health care claims data, we calculated the number of outpatient and inpatient PCR tests according to age group. Individuals who had undergone multiple PCR tests were counted as single cases. The procedural and administrative codes used to identify PCR tests and use of public funds are provided in Supplementary Tab**le 1.
Next, we evaluated the data linkage rates between the Basic Resident Register and the HER-SYS for Algorithms 1 to 5. For individuals that could be linked using these algorithms, we also assessed their linkage rates with health care claims data according to age group. Individuals who did not have a single clinical encounter at a health care institution (i.e., did not generate any claims data) during their municipality's data collection period were excluded from data linkage.
Finally, we evaluated the validity of the HER-SYS records of COVID-19 vaccination statuses and underlying conditions. As a database specifically designed to record vaccination statuses, the VRS was assumed to have accurate records for these data. Similarly, we assumed that the health care claims data had accurate records of underlying conditions as these would affect patient case-mix and reimbursements. With these assumptions, we analyzed the validity (measured using sensitivity) of COVID-19 vaccination statuses and underlying conditions in the HER-SYS based on their concordance with the VRS and health care claims data, respectively. We assumed that the VRS contains true information on vaccination statuses, and calculated the sensitivity of HER-SYS for vaccination records as the proportion of true positives (individuals recorded as vaccinated in the HER-SYS who are actually vaccinated) to the total number of individuals who are actually vaccinated. This evaluation was limited to individuals who were assigned research identification codes in the HER-SYS. Similarly, we assumed that the health care claims data contain true information on underlying conditions, and calculated the sensitivity of HER-SYS for these conditions as the proportion of true positives (individuals recorded with an underlying condition in the HER-SYS who actually have the condition) to the total number of individuals who actually have the condition. This evaluation was limited to individuals aged ! 75 years who were assigned research identification codes in the HER-SYS. As the residents included in all three data types were those aged ! 75 years, these older residents comprised the study sample for the validity assessments. The identification criteria for the underlying conditions in the health care claims data are presented in Supplementary Tab**le 1. In the identification of underlying conditions from these data, we only included confirmed diagnoses, and excluded those classified as ''suspected diagnoses". The monitoring period for the underlying conditions was the 12-month period preceding the month before the positive COVID-19 test result.
The VENUS Study was approved by the Kyushu University Institutional Review Board for Clinical Research (Approval No. 2021-399). In addition, the Personal Information Protection Review Board of each participating municipality reviewed and authorized the study. Individuals who wish to be withdrawn from the VENUS Study may opt-out by contacting their municipal government or VENUS Study personnel, and their data will be deleted from the database.

Results
We acquired the Basic Resident Register, VRS, HER-SYS, and health care claims data from four municipalities (designated City A, City B, City C, and City D), and constructed a multi-source database that linked these data types at the individual resident level. Table 1 presents the numbers of records in the VRS, HER-SYS, and health care claims data for each of the four municipalities collected until July 1, 2022. The HER-SYS included data until March or April 2022, and the number of COVID-19 cases ranged from 6,732 to 21,040 across the municipalities. Fig. 1A presents the proportion of vaccinated persons in the Japanese population according to age group. Immediately following the approval of the COVID-19 vaccines in Japan, the initial vaccination efforts focused on practicing health care workers. Therefore, the early period had higher proportions of persons aged 20-64 years. Thereafter, older persons were prioritized for vaccinations due to their increased risk of serious COVID-19 symptoms, which resulted in a rapid increase in vaccinations for persons aged ! 65 years. This was followed by vaccinations for persons aged 20-64 years and persons aged 19 years. Fig. 1B shows the number of days from the start of COVID-19 vaccinations until 80% coverage of each municipality's population (84 days for City A, 110 days for City B, 92 days for City C, and 84 days for City D). Fig. 1C shows the distribution of the vaccination interval between the first and second doses of the BNT162b2 vaccine. The figure is centered around the recommended three-week interval (designated ''0 00 days) for this vaccine. The vast majority of vaccinated residents (91Á48-94Á78%) strictly followed this recommended interval in all municipalities. Fig. 1D illustrates the daily trends in COVID-19 cases from the HER-SYS data. The numbers of individuals who had undergone PCR tests (outpatient and inpatient) and publicly funded hospitalization numbers according to age group are presented in Table 2. Approximately 6Á9% of the entire study population within the four municipalities had undergone outpatient PCR tests, and there were no substantial differences among the age groups. In contrast, the proportions of individuals who had undergone inpatient PCR tests increased with increasing age, culminating at 9Á6% in persons aged ! 75 years. Table 3 shows the numbers of linked residents and linkage rates between the HER-SYS and the Basic Resident Register for each municipality according to matching algorithm. The proportion of residents that could not be linked between these two data types ranged from 8Á3% (City B) to 14Á6% (City D). In most cases, residents could be matched using Algorithm 1 (name in kanji, name in kana, birth date, and sex). Each of the other three algorithms had a linkage rate of approximately 18% or less. After identifying the individuals that could be linked between the HER-SYS and the Basic Resident Register, we assessed their linkage rates with health care claims data according to age group (Table 4). Among residents aged ! 75 years, the linkage rates with health care claims data ranged from 86Á7% to 95Á6%. However, the linkage rates were lower in the younger age groups. The linkage rates between the VRS and Basic Resident Registers were 100% as both systems use the same resident registration numbers to identify individuals.
We evaluated the validity of the HER-SYS records based on their concordance with these data types ( Table 5). The HER-SYS had low sensitivities for first-dose vaccinations (46Á8%) and second-dose vaccinations (45Á7%). Among the underlying conditions, hypertension showed the highest sensitivity at 42Á6%. Sensitivity was lower than 10% for several other conditions.

Discussion
The first COVID-19 vaccine was approved for use in Japan on February 14, 2021. While a considerable amount of time has since passed, Japan has yet to establish a national database that can actively monitor COVID-19 occurrence and adverse events in both vaccinated and unvaccinated persons. In this study, we described the development of a researcher-led database project that can facilitate active surveillance and comparative analyses at the municipal level. We also assessed the ability of the VENUS Study's database to analyze vaccination statuses, vaccination coverage, COVID-19 cases, PCR tests, and COVID-19 hospitalizations. Our results indicate that this database, if used with prudence and a thorough understanding of each data source's strengths and limitations, can provide a viable data platform for population-based research on COVID-19 vaccine effectiveness and safety in Japan. The analysis also found that although the HER-SYS contains information on positive COVID-19 cases, its records on vaccination statuses and underlying conditions have comparatively low levels of accuracy and completeness.
Despite the potential advantages conferred by including COVID-19 occurrences from the HER-SYS, our analysis raised doubts about the accuracy of its data. In addition to COVID-19 cases, the HER-SYS collects data on vaccination statuses and underlying condi- tions. However, the latter two items are self-reported by patients upon receiving a COVID-19 diagnosis. By comparing the HER-SYS vaccination statuses and underlying conditions with the corresponding records in the VRS and health care claims data, we found that the sensitivities for the HER-SYS records of first-dose and second-dose vaccinations were only 46Á8% and 45Á7%, respectively. The sensitivities for the HER-SYS records of underlying conditions were even lower, with the highest at only 42Á6% for hypertension.   In our identification of underlying conditions from the health care claims data, we excluded suspected diagnoses, which are frequently given to justify diagnostic tests. However, as accurate diagnoses cannot always be ascertained from the recorded use of drugs or medical procedures, it is possible that the recorded diagnoses in the claims data may have overestimated the actual disease conditions, which could lead to a degree of underestimation of the HER-SYS's sensitivities. These findings suggest that it is difficult to accurately ascertain vaccination status and underlying conditions in clinical settings. In this way, the HER-SYS can be limited to necessary data items, which could increase its overall accuracy. We compared the VENUS Study with other countries' databases that enable COVID-19 vaccine assessments. First, the quality of health care information in the VENUS Study does not appear to be markedly inferior to the databases of other countries. As Japan uses a universal health insurance system, it is possible to obtain almost all health care records of insurance enrollees. Therefore, comprehensive claims data on all health care encounters can be acquired for both vaccinated and unvaccinated persons. While the relatively lower proportions of younger insurance enrollees and the lack of employer-based insurance enrollees are notable gaps in the coverage of the VENUS Study, our study population included persons of all age groups and information on all insurance-covered care. Moreover, there is an increasing number of validation studies on the use of Japan's claims data in identifying diseases such as respiratory diseases [27], cardiovascular diseases [28,29], and cancer [30]. Such studies contribute to the growing evidence that algorithms combining diagnoses, treatments, and prescriptions recorded in these claims data are able to accurately identify target diseases. Second, the study populations are slightly smaller than those used in other databases. As the VENUS Study is not a government-led initiative and does not have easy access to national-level data, it is limited to municipalities that voluntarily agree to participate. At present, the VENUS Study database covers approximately 500,000 persons residing in four municipalities. Until more municipalities are recruited, the current numbers may be insufficient to produce high-quality evidence on vaccine safety. Nevertheless, there are numerous studies that have assessed COVID-19 vaccine effectiveness in smaller study populations [6][7][8][11][12][13]14,[16][17][18][19][20][21], suggesting that research based on the VENUS Study database would not be inferior to the existing literature. The VENUS Study aims to increase the number of participating municipalities for a target study population of 5 million. Third, the VENUS Study still has the capacity to be further developed. In Japan, region-based medical care, long-term care, and health activities are often developed and implemented at the municipal level. By continuing to build upon the VENUS Study together with cooperation from municipalities, it may be possible   to also link health checkup data (e.g., prenatal checkups and infant health checkups) and long-term care claims data (e.g., home-based and facility-based long-term care) 26 . This would widen the range of studies that can be conducted using this database. The VENUS Study has the following limitations. First, the age structure of our study population is not representative of the overall population in Japan. As the VENUS Study focuses on enrollees of the National Health Insurance System and Latter-Stage Older Persons Health Care System, the study population is overrepresented by older persons, and under-represented by children and adults who are covered under social insurance. The proportions of residents in each municipality who are enrolled in the National Health Insurance System are 12.0% for persons aged 0-19 years, 19.2% for persons aged 20-64 years, and 70.7% for persons aged 65-74 years [31,32]. In contrast, the HER-SYS covers residents of all ages within each municipality. This age-dependent difference in coverage between the HER-SYS and health care claims data reduced their linkage rates, particularly in younger persons. Also, the types and prevalences of underlying conditions may vary between enrollees of the different insurance schemes. Therefore, the VENUS Study is vulnerable to systematic bias as it only covers enrollees of the National Health Insurance System and Latter-Stage Older Persons Health Care System. Second, the four municipalities participating in the VENUS Study may not be representative of Japan as a whole. Third, there are limitations to the immediacy of data updates. Currently, VENUS Study researchers periodically visit each participating municipality to receive updated data, and the numbers of these visits were heavily restricted during the pandemic. Furthermore, each municipal government is generally only able to receive health care claims data approximately two months after care is provided. Therefore, even if VENUS Study researchers rushed to collect data, there will always be an unavoidable time lag of at least two months between the provision of care and data collection. This lack of real-time data is an inherent limitation of health care claims data. Fourth, the different data types are not easily linked. Due to a strong national disposition toward the protection of personal information, Japan has no mandatory identification system used to identify residents across various data types.
The VENUS Study database described in this study has already been used to generate evidence on the effectiveness of the COVID-19 vaccine [33,34]. The effectiveness of COVID-19 vaccination against infection can be analyzed using only the VRS (containing vaccination records) and the HER-SYS (containing information on COVID-19 cases), which allows for an investigation of all residents within the participating municipalities. In September 2022, however, the Japanese government streamlined the mandatory recording of COVID-19 cases to apply only to high-risk groups comprising persons aged ! 65 years, hospitalized patients, pregnant women, and persons with underlying conditions. Consequently, this diminishes the ability to investigate the effectiveness of vaccination against infection in lower-risk groups and younger persons using HER-SYS data. Research on the effectiveness of vaccination against severe COVID-19 requires information on hospitalization, which can be acquired by linking the HER-SYS with health care claims data. Similarly, research on COVID-19 vaccine safety requires information on adverse events, which can be acquired by linking health care claims data with the VRS. As adverse events following immunization are frequently measured using rates per 100,000 person-years, there is a need to increase the number of participating municipalities. In addition, the accuracy of recorded diagnoses in the health care claims data must be assessed. Specifically, comprehensive validation studies are needed to test the levels of agreement between diagnoses recorded in claims data and medical records in Japan's health care institutions. Furthermore, it is also important to conduct a validation study to determine if COVID-19 occurrence can be ascertained from health care claims data with the same level of accuracy as the HER-SYS. If COVID-19 occurrence can be accurately identified using algorithms that combine various information from health care claims data (e.g., publicly funded health care, drug prescriptions, COVID-19-specific management fees, and recorded diagnoses), future studies could be conducted without the need for HER-SYS data.

Conclusion
Although a year behind many countries, we developed and tested the viability of Japan's first system to assess COVID-19 vaccine effectiveness and safety. However, we discovered several fundamental problems in Japan's database environment when developing this system. In its present condition, Japan's vaccine monitoring strategy is inherently limited and cannot be said to be working well. Instead, the HER-SYS in its current form could even be considered a serious misstep in Japan's digitalization strategy against COVID-19. Despite its inclusion of information on vaccination statuses and underlying conditions, the accuracy of these records is extremely low. In contrast, accurate vaccination records and diagnostic records can be obtained from the VRS and health care claims data, respectively. We propose that instead of recording low-quality information on these items in the HER-SYS, it would be more useful to re-design it as a database that includes personally identifiable information (e.g., resident registration numbers or insurance enrollee numbers) that would enable its precise linkage with the VRS, health care claims data, and other databases. In order to prepare for future public health crises, there is a need to rebuild Japan's database environment to support the development of timely and accurate surveillance systems.

Author's contributors
HF conceived and designed the study. HF, MM, MF obtained and managed the data. HF developed the analytical strategy. HF conducted analysis in consultation with MM and MF. HF, and MM interpreted the data and drafted the figures. HF wrote the first draft of the manuscript. Other authors provided input to finalize the paper. HF, MM, and MF had full access to all data used in this study. All authors were responsible for submitting the article for publication.

Data sharing
Data cannot be shared for privacy or ethical reasons.

Funding
This research was supported by AMED under Grant Number JP21nf0101635.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.