What Caused Dire Clusters ? ~Analysis of Nosocomial COVID-19 Outbreaks in Kyoto City, Japan From the 1st Wave Through 3rd Wave in 2020/21

Fifteen COVID-19 nosocomial clusters occurred in the rst year of the pandemic in Kyoto City, involving 232 in-hospital patients and 157 healthcare workers (HCWs). Among them, 40 percent of index cases were HCWs. The in-hospital mortality rate among patients was 18.1%, whereas the mortality rate in the Kyoto City area during the study period was 1.79%. The time needed to contain outbreaks was shorter in hospitals with full-time infection control personnel assigned.


Background
Since the initial outbreak, Coronavirus disease 2019 (COVID-19) has spread worldwide, claiming > 3,000,000 lives (https://covid19.who.int/. Accessed May 11, 2021). In addition to its global impact, COVID-19 has alerted the healthcare community to the dangers of nosocomial infection. Nosocomial outbreaks are problematic because of the poor associated prognosis in the presence of underlying diseases in patients and crippling of the entire healthcare system due to the discontinuation of services by medical institutions.
To date, nosocomial infection by COVID-19 has been discovered and reported in numerous healthcare facilities on a global scale; however, concise analysis of nosocomial clusters is limited. [1][2][3] In this study, we clari ed the characteristics and impact of nosocomial clusters that occurred in Kyoto City in the rst year of the pandemic in order to examine preventive measures.

Results
Kyoto City is the capital of Kyoto Prefecture, with 1.4 million inhabitants, and more than 70% of COVID-19 cases in Kyoto Prefecture are from Kyoto City. We analyzed nosocomial clusters that included 5 or more COVID-19 cases of hospitalized patients and healthcare workers (HCWs) in Kyoto City from March 2020 to February 2021. There were three epidemic waves of COVID-19 infections from March 2020 to February 2021 in Japan, among which the 3rd wave led to the highest number of patients. [4] There were 15 clusters in 14 hospitals, which included a total of 459 COVID-19 cases: 145 HCWs, and 282 hospitalized patients. COVID-19 patients discharged from the cluster hospitals caused secondary clusters in subsequent facilities. There were 32 secondarily infected cases outside of the hospital who became infected via transmission from the primary COVID-19 cases. (Table) Suspected index cases of clusters were HCWs in 6 (40%) of the 15 clusters. In all 6 cases, a symptomatic staff member continued to work, which may have led to the spread of the infection. Of the 9 cases in which patients were considered to be the index cases, 4 cases were transferred from other facilities, and 2 cases were suspected to have been infected by family members visiting the hospitals. As many as 61.9% of HCWs and 65.3% of patients with COVID-19 developed symptoms during the course of clusters. The in-hospital mortality rate among patients was 18.1%, and the mean time to death was 11.4 days. The mortality rate in the Kyoto City area during the study period was 1.79%.
The total number of contact tracing PCR tests for 14 clusters was 10,104, with 721 tests per hospital and 22.0 tests per COVID-19 case on average. The average times taken from detection of the rst COVID-19 case in the hospital to expanded testing for contact tracing were 7.3 days in the 1st and 2nd waves, and 2.9 days in the 3rd wave, indicating that expanded testing was done in a shorter period of time in the 3rd wave. The average time taken for cluster containment was 33 days. Regarding the impact on the health care delivery system, all hospitals where clusters occurred restricted hospitalization, and outpatient clinics were closed in 8 of the 15 cases. In addition, 22 cases were secondarily infected by family members or friends after being discharged from the cluster facility, and infection via 2 patients transferred to other hospitals subsequently led to a new cluster of 31 people. Five of the 14 facilities had full-time infection control personnel (ICP).
Although attack rates, time to expanded testing, and in-hospital mortality of COVID-19 patients did not differ by the status of ICP assignment, the average time to contain clusters was signi cantly shorter in facilities with full-time ICP. (Table, Figure) Discussions This is the rst report of the concise analysis of nosocomial clusters in Japan. As indicated in this report, nosocomial clustering has devastating consequences in terms of the prognosis of patients and crippling of the healthcare system both for COVID-19 and non-COVID-19 patients.
When a COVID-19 cluster occurs, close contacts are identi ed, isolated, and tested and infection control measures are strengthened while waiting for the cluster to end. Previous reports focused on the transmission risk and prognosis of COVID-19 patients, and did not examine the details of index cases that contribute to countermeasures against the introduction of COVID-19 to hospitals. In this study, a number of cases were due to transmission via health care workers. The risk of infection among medical personnel is considered to be higher than that among the general public, and the appropriate use of protective equipment is required. [5] On the other hand, as HCWs live in the community (not quarantined), they are at risk of becoming infected depending on the infection status in the community. As such, HCWs need to behave as if they are asymptomatic carriers of COVID-19 and adopt measures to prevent themselves from spreading the infection.
Testing is one of the key strategies to tackle with COVID-19. A low threshold for testing facilitates the prompt identi cation of cases that can be managed by isolation or quarantine, as it has been demonstrated that there are high proportions of asymptomatic and, more importantly, pre-symptomatic individuals. In Japan, it was reported that the PCR testing system was lagging behind those of other countries (accessed as of 11 May 2021 https://ourworldindata.org/coronavirus-testing#how-many-tests-are-performed-each-day), but, in the 3rd wave, the time between index case identi cation and expanded testing was improved compared with the 1st and 2nd waves. In addition, in Kyoto City, the time needed to contain clusters was shorter in facilities with full-time IPCs. Furthermore, facilities with full-time IPCs tended to have lower staff attack rates. (Table) The establishment of a solid chain of command by full-time IPCs and hospital executives may be necessary to contain the situation.
Since huge amounts of medical resources such as personnel involved in contact tracing and PCR testing are required to contain clusters, it is important to adopt measures to prevent introducing the disease to a hospital population. The PCR screening test at the time of admission is not a panacea, and some cases are inevitably missed. As Abbas et al. stated, it may be suitable to perform broad-scale screening of both patients and HCWs, including asymptomatic individuals, in the event that COVID-19 cases are identi ed on "non-COVID" wards to help identify a potential outbreak situation, and to be able to control it. [3] Conclusions According to our analysis, it is extremely important to expand testing to identify index cases and impose standard infection prevention measures, as well as implement sick-leave policies. Those countermeasures are "classic" and simple, and may be implemented much easier if a chain of command is established. We believe that nosocomial transmission can be controlled to some extent by analyzing and sharing the experience of nosocomial clusters indicated in this report in order to ll any gaps in knowledge.

Declarations
Availability of data and materials The datasets used and analysed during the current study are available from the corresponding author on reasonable request.  Figure 1 Time to contain nosocomial clusters by the presence of full-time infection preventionists. The average time to contain clusters was signi cantly shorter in facilities with full-time ICP indicating every-day intervention is needed to contain clusters as quick as possible.