Discussion

Universal admission screening for COVID-19 has been adopted as infection control management to prevent in-hospital transmission of SARS-CoV-2. ^{Reference Krüger, Leskien and Schuller3,Reference Saidel-Odes, Shafat, Nativ, Borer and Nesher4,Reference Sutton, Fuchs, D’Alton and Goffman9,Reference Hanson, Caliendo and Arias11,Reference Sastry, Pryor and Raybould13,Reference Alsuhaibani, Kobayashi and Trannel16} To the best of our knowledge, this was the first nationwide population-based cohort study to investigate the performance of universal PCR-based screening for SARS-CoV-2 in newly admitted patients and their companions during periods with varied community incidence rates.

The positivity rate of universal admission screening in newly admitted patients was 0.76% during a period with an average weekly community incidence of 270 per 100,000 population. Previous studies reported that the positivity rate of universal admission screening for SARS-CoV-2 ranged from 0.005% to 13.7%. ^{Reference Saidel-Odes, Shafat, Nativ, Borer and Nesher4,Reference Sutton, Fuchs, D’Alton and Goffman9,Reference Jung, Kim, Lim, Kim, Kim and Kim10,Reference Alsuhaibani, Kobayashi and Trannel16} The wide range of positivity rates in different studies was likely due to the different definitions of universal admission screening and community incidence. The positivity of universal admission screening for SARS-CoV-2 has been positively associated with the contemporary community incidence. ^{Reference Krüger, Leskien and Schuller3,Reference Nakamura and Itoi5} We also observed a similar correlation; the positivity rate of universal admission screening in period 3 with greater community transmission was significantly higher than the positivity rates in periods 1 and 2.

The advantage of universal admission screening is that it provides information for hospitals to implement appropriate IPC measures in advance. ^{Reference Hanson, Caliendo and Arias11,Reference Sastry, Pryor and Raybould13,Reference Alsuhaibani, Kobayashi and Trannel16} Nevertheless, concerns about the potential drawbacks of universal admission screening remain, such as unnecessary isolation of individuals with false-positive test results and delay of necessary treatment for cases. ^{Reference Carrara, Ong and Hussein7,Reference Hanson, Caliendo and Arias11,Reference Hashim, Mahdy, Abdul Rahman, Kalok and Sutan14–Reference Penney, Shah and Doron17} Therefore, in the era of vaccination, it is debatable whether universal admission screening is still necessary, especially in areas with a low community prevalence of SARS-CoV-2. ^{Reference Nakamura and Itoi5,Reference Carrara, Ong and Hussein7,Reference Jung, Kim, Lim, Kim, Kim and Kim10,Reference Hanson, Caliendo and Arias11,Reference Penney, Shah and Doron17,Reference Tenforde, Self and Gaglani22} All positivity rates of the periods observed during this study were less than the prevalence threshold of 10% recommended by international academic associations for the implementation of universal testing for asymptomatic hospitalized patients. ^{Reference Carrara, Ong and Hussein7,Reference Hanson, Caliendo and Arias11}

Although several studies supported universal admission screening, none of these were based on the intervention with cost and effectiveness assessments. ^{Reference Krüger, Leskien and Schuller3,Reference Saidel-Odes, Shafat, Nativ, Borer and Nesher4,Reference Sastry, Pryor and Raybould13,Reference Hashim, Mahdy, Abdul Rahman, Kalok and Sutan14} Krüger et al ^{Reference Krüger, Leskien and Schuller3} reported that the NNTs and average costs of universal admission screening to detect an asymptomatic SARS-CoV-2 patient in a Germany tertiary-care hospital were 425 patients and ∼€25,075 ($28,247) in the high-incidence phase, with a positivity rate of 1.41% and 1,218 patients and ∼€71,862 ($78,969) in the low-incidence phase, with a positivity rate of 0.25%. Nevertheless, no exact beneficiary benefits of the screening strategy were observed in this study. Many studies have examined the cost-effectiveness of various test strategies for SARS-CoV-2, but studies focusing on universal admission screening are limited. ^{Reference Dolatshahi, Nargesi and Sadeghifar23–Reference Diel and Nienhaus25} Stevenson et al ^{Reference Stevenson, Metry and Messenger24} modeled 30 SARS-CoV-2 test strategies for patients admitted from the emergency department, with a community prevalence of 5.3%. These researchers found that the least costly test strategy was testing on hospital admission with retesting 6 days after admission, and they recommended that using tests with shorter turnaround times would be more cost-effective. ^{Reference Stevenson, Metry and Messenger24} Another modeling study in a simulation with a prevalence of 15.6% in Germany showed that testing patients in emergency departments before hospitalization reduced the average cost of hospitalized patients by €213 ($234) per tested patient. ^{Reference Diel and Nienhaus25} Instead of conducting a modeling cost–benefit analysis, we used real-world operational data to approximate costs and potential savings from averted healthcare-associated SARS-CoV-2 infections. Our results demonstrated that implementing universal admission screening for patients and their companions would achieve monetary benefits in the phase with weekly positivity rates of 0.47% and 0.78%, respectively. Thus, the costs of universal admission screening would be outweighed by the benefits only in period 3 of our study from a monetary perspective.

Cost-effectiveness is one of the dimensions of the decision-making process when evaluating interventions. Factors such as dominant virus variants in circulation, lack of appropriate vaccinations and pharmaceuticals, and the surge in the number of patients requiring care from the healthcare system should also be considered to justify universal admission screening during the early period of the COVID-19 pandemic. ^{Reference Krüger, Leskien and Schuller3,Reference Kim, Jo and Lee6,Reference Alsuhaibani, Kobayashi and Trannel16,Reference Penney, Shah and Doron17,Reference Dolatshahi, Nargesi and Sadeghifar23} As the pandemic evolved, the virulence of the variants changed, effective vaccinations and antiviral agents were developed, and the immunity of the population increased. ^{Reference Alsuhaibani, Kobayashi and Trannel16,Reference Jian, Cheng, Huang and Liu21,Reference Hammond, Leister-Tebbe and Gardner26,Reference Wolter, Jassat and Walaza27} Therefore, the extra benefits of universal admission screening added to the existing hierarchy of IPC measures (eg, the appropriate use of PPEs, active health surveillance of healthcare workers, optimal unit layouts with enhanced ventilation) might need to be re-evaluated. ^{Reference Alsuhaibani, Kobayashi and Trannel16,Reference Pan, Hsu and Chang28–Reference Malundo, Abad and Salamat30} The Society for Healthcare Epidemiology of America and the American Society of Anesthesiologists did not recommend routine universal testing for SARS-CoV-2 in asymptomatic patients at the time of hospital admission and before hospital procedures. ^{Reference Pan, Hsu and Chang28,29} Healthcare facilities should implement universal admission screening based on individualized risk and benefit assessments and should consider metrics such as the population at risk for severe infection (eg, residents of long-term care facilities, immunocompromised individuals), vaccination status, physical layout of the unit, and community transmission of SARS-CoV-2. ^{Reference Alsuhaibani, Kobayashi and Trannel16,29,Reference Bigelow, Tang and Barshick31,32}

Despite the infectivity of asymptomatic COVID-19 cases, interviews with all patients regarding symptoms through clinical triage and targeted testing would still be an important measure to detect SARS-CoV-2 infections. ^{Reference Krüger, Leskien and Schuller3,Reference Saidel-Odes, Shafat, Nativ, Borer and Nesher4,Reference Hanson, Caliendo and Arias11,Reference Alsuhaibani, Kobayashi and Trannel16} Follow-up symptom-based testing detected ∼9.86% of the confirmed cases in our study within 7 days after negative universal admission screening test results. These patients might have been in the incubation period when they received the admission screening, or they were infected during hospitalization, considering that the mean incubation period of SARS-CoV-2 is ∼5 days. ^{Reference Kim, Jo and Lee6,Reference Hanson, Caliendo and Arias11,Reference Alsuhaibani, Kobayashi and Trannel16} This factor highlights the importance of clinical judgment for inpatients who develop symptoms compatible with COVID-19 during hospitalization. IPC teams in hospitals who implement universal admission screening should be aware of the false sense of security provided by the screening strategy and remind firstline healthcare workers of the significance of clinical assessment.

In our study, patients identified by symptom-based testing had a higher proportion of 30-day mortality than patient cases identified by universal admission screening. Other studies noted similar findings, but no specific explanations were provided in the discussions. ^{Reference Malundo, Abad and Salamat30,Reference Bigelow, Tang and Barshick31} Although we did not evaluate the underlying medical conditions of the cases, older age and a lower proportion of fully vaccinated individuals in patient cases identified by symptom-based testing may partially explain the higher mortality. ^{Reference Wu, Su, Chien, Tseng and Chang1,Reference Tenforde, Self and Gaglani22} A certain proportion of cases detected by universal admission screening might only be identified due to prolonged virus shedding, representing an unknown past infection instead of an acute infection. The majority of blood tests in asymptomatic cases in Malundo’s study were within normal limits. ^{Reference Malundo, Abad and Salamat30} Some asymptomatic cases identified by universal admission screening possibly were caused by persistent shedding of nucleic acid fragments from past infections and this might also have contributed to the lower mortality of the group.

This study had several limitations. First, our data were retrieved from the NHI claims database, and the billing codes used for the tests were not validated. There was a possibility of patient or provider bias when stating or collecting data about the presence or absence of symptoms, thus potentially influencing the type of billing code selected. Second, we did not conduct a formal cost–benefit analysis, and many parameters that should have been included in such modeling studies were not considered, such as the protective effectiveness of vaccination, performance of the test kits, indirect costs incurred by ward closure, and productivity loss incurred by quarantine or isolation. Consequently, potential savings might be underestimated or overestimated; therefore, further studies are warranted. Nevertheless, we used observation data from the operations of Taiwan healthcare systems in the real world, which may still provide useful information for the decision-making process of relevant policies in the future. Third, we did not have the RT-PCR cycle-threshold value results, which is a commonly used proxy for COVID-19 infectiousness. Therefore, we were not able to discriminate between cases identified by universal admission screening as acute or past infections, which could have resulted in the overestimation of the benefits of universal admission screening. Finally, because of the well-operated national laboratory networks for SARS-CoV-2 testing in Taiwan, each hospital could perform universal admission screening with RT-PCR either alone or in collaboration with designated laboratories. Therefore, our results might not be generalizable to countries or regions with different diagnostic capacities.

Controlling SARS-CoV-2 transmission in hospitals is critical for reducing deaths and severe illnesses from COVID-19. Universal admission screening of patients and their companions can control the spread of asymptomatic and presymptomatic SARS-CoV-2 infection in healthcare settings. By determining the appropriate timing and targets for the implementation of this strategy, the benefits can be balanced with the potential unintended harms, and resources can be allocated more efficiently.