Discussion

There were altogether 11 outbreaks in 8 different hospitals and 71 patients acquired a COVID-19 infection during an outbreak. Many of our hospitals have rooms for four patients, which increases the risk of admitting COVID-19 infection from patient-to-patient transmission and resulting in an epidemic. There were no clusters in single rooms. In addition to room size, there must also be other factors contributing to hospital outbreaks, since COVID-19 spread from one room to another only in some of our hospitals while in others, there were no outbreaks at all. Infectious disease specialists and hygiene nurses in all our hospitals meet twice a week to review guidelines and protocols. Our guidance concerning personal protective equipment, including universal masking, and education of healthcare workers in all hospitals in HUH is the same, so difference in infection control practices should not be an explanation for the susceptibility of some hospitals to COVID-19 epidemics. All visitors were also guided to wear a mask during our study period. In all hospitals, SARS-CoV-2 testing criteria were the same. More outbreaks appeared to be in older hospital buildings with smaller spacing between patient beds and older ventilation systems. Hospitals are instructed to ventilate the rooms at minimum 6 times per hour, but unfortunately, this instruction is not monitored. However, a study from USA ^{Reference Trick, Santos and Welbel8} that observed two different acute-care hospitals had an overall incidence of HA COVID-19 of 0.9 per 1,000 patient days in a hospital with only single-occupancy rooms, and 0.39 per 1,000 patient days in another hospital with larger rooms available, implying that implementation of other infection-control methods than single-patient rooms also affects the transmissibility of COVID-19. The incidence in the USA is in the range of ours but one study from Brazil from August 2020 to September 2021 reported an incidence of 5.8 per 1,000 patient days at adult ICUs, which is tenfold higher than ours. ^{Reference Machado de Miranda Costa, Guedes and Nogueira9} We had only three HA COVID-19 infections in ICUs, which may be due to the lower patient load compared to Brazil.

Most of our HA COVID-19 cases occurred in February and March 2022, while at the same time, we had the highest disease burden in the population caused by omicron. ⁷ It is logical that the more virus circulates in the community, the more inpatients have COVID-19 infections in hospital. A recent study ^{Reference Aghdassi, Schwab and Peña Diaz10} presented that both contact with individual COVID-19 patients and being admitted to a ward experiencing an outbreak increased the patient’s risk of maintaining an HA infection. This study also found that most of the HA cases (92%) were a part of an outbreak at a specific ward, while in our study, the outbreaks were a significant etiological factor in HA infections, but their proportion of all HA infections was lower. Many of our patients who contracted COVID-19 from an outbreak were exposed to the infection by a patient that they shared a room with. One study from Switzerland ^{Reference Wolfensberger, Kufner and Zaheri11} observed that prolonged contact time in a shared hospital room is essential in patients acquiring a HA infection.

Higher rates of HA COVID infections have been reported in psychiatric wards even after implementing non-pharmaceutical prevention interventions. ^{Reference Ngandu, Mmotsa, Dassaye, Thabetha, Odendaal, Langdown and Ndwandwe12} In our study, the HA COVID-19 incidence in the psychiatric ward was 1.10 per 1,000 patient days, being higher than in other wards. In addition, the psychiatric patients were younger, were less frequently vaccinated, had a smaller number of comorbidities, and had a lower Charlson Comorbidity Index than the patients in the medicine and operative wards. Higher HA COVID-19 rate in psychiatric patients may be explained by their ability to move around more freely in the hospital settings, eating and spending time together for the purpose of rehabilitation, and having longer treatment periods.

The 30-day mortality of the patients with HA COVID-19 in our study was 11%, and other countries have reported similar mortality rates. For example, a study from Korea that also observed HA COVID-19 infections in the latter part of the year 2021 and early 2022 when omicron was the main variant, reported a mortality rate of 8.4% in 167 patients. ^{Reference Kim, Kim, Choi, Shin and Sim6} Prior to the implementation of COVID-19 vaccinations, obviously higher mortality rates of 32% with HA COVID-19 patients and 34% with community-acquired COVID-19 patients were observed in the United Kingdom. ^{Reference Shiwani, Bilal and Shahzad5} In our study, patients in the medicine wards had a higher mortality rate compared to the operative and psychiatric patients. There is no obvious explanation for this finding because operative patients were about the same age, they had similar rates of comorbidities, and medicine patients were even more often vaccinated than others. On the contrary, medicine patients more frequently had DNR decisions compared to patients in other specialties, indicating that these patients were frailer, and this may partly explain the higher mortality rate among them. While our mortality rate resembled reports from other countries, our study had a relatively low severity of COVID-19 even with high-risk individuals based on low rates of steroid use, number of complications, and ICU admission rate.

There was a slight difference in the vaccination status between specialties, medicine patients being the most vaccinated group (92%), while a slightly smaller fraction of the operative (84%) and psychiatric wards (66%) were vaccinated at least once. It seems, however, that with this sample of patients in various specialties, patients died in association with COVID-19 even though they were vaccinated, probably due to their other comorbidities. In Korea, HA COVID-19 patients had a similar mortality rate and vaccination status, with 78% of the inpatients having at least 1 vaccination, but the inpatients with no vaccinations were reported at a higher risk of mortality. ^{Reference Kim, Kim, Choi, Shin and Sim6} In the Korean study, every deceased patient had received remdesivir for COVID-19, while none was treated with nirmatrelvir combined with ritonavir. In contrast, a study in the United Kingdom reported a 41% mortality rate with HA COVID-19 infections before vaccinations. ^{Reference Duffaydar, Beaumont, Pham, Khamb and Plant13} The nirmatrelvir/ritonavir for prevention of severe COVID-19 was not introduced in Finland until summer of 2022. Remdesivir, with very limited availability, was the only antiviral drug in use during our study period. Remdesivir was available only for defined high-risk patients, which included only a few of HA COVID-19 patients in this study. We cannot conclude if with better availability of antivirals some of the deaths could have been avoided.

The strength of this study is that it is population-based, with patient data that has been collected from all secondary and tertiary care hospitals treating adults from a population area of 1.8 million. Our instructions urge screening asymptomatic exposed inpatients. Vaccination information was widely, although not completely, available, as well as patients’ medical history.

The most relevant limitation of our study is that we were unable to identify all HA COVID-19 infections. In our hospital guidelines, all exposed inpatients are screened for COVID-19 with SARS-CoV-2 NAAT on days 0 and 5 and in addition if symptoms occur. On contrary, discharged asymptomatic patients are not routinely screened. We also did not receive information about majority of patients who tested positive for SARS-CoV-2 after discharge or transfer to primary healthcare facilities. In addition, the clinical sensitivity of SARS-CoV-2 NAAT is estimated to be from 68 to 86%, thus some infected patients remain unrecognized. ^{Reference Kortela, Kirjavainen and Ahava14} These issues imply that there were probably more HA COVID-19 cases than we were able to determine, and our study underestimates the incidence. The extent to which findings of this study can be generalized may also be limited since HA COVID-19 incidence and mortality rates in other countries depend on infection control practices and clinical practices. The transmissibility of new SARS-CoV-2 variants, and morbidity and mortality they cause varies, which also weakens generalizability of our results.