Discussion

We surveyed hospital epidemiologists from 23 leading US hospitals to understand COVID-19 infection control policies in the context of the initial wave of COVID-19 caused by the SARS-CoV-2 omicron variant in the United States. The 2 policies that were nearly uniform were the type of respiratory protection for COVID-19 patients (96% of survey hospitals used N95s for routine care) and universal testing of patients on admission (91% of survey hospitals). Consensus was moderately high regarding universal eye protection for non–COVID-19 patients (70% of survey hospitals), preference for nasopharyngeal swabs for symptomatic patients (78% of survey hospitals), and targeted rather than mandated surveillance testing of employees (78% of survey hospitals). Beyond this, however, we observed substantial variation in care models (dedicated COVID-19 units vs mixed units vs a combination of both), use of airborne infection isolation rooms, use of portable HEPA filters, postadmission testing strategies, specimen source sites for asymptomatic patients, use of algorithms to clear precautions from asymptomatic patients with high Ct values on their initial positive SARS-CoV-2 tests, what Ct values were considered to be high, and use of time versus test-based strategies to clear precautions from patients with confirmed COVID-19.

The routine use of N95s for all COVID-19 care rather than only for AGPs corresponds with shifts in the CDC and WHO guidance that reflect the growing evidence of aerosol-based transmission, even in the absence of AGPs. ^{4–Reference Greenhalgh, Jimenez, Prather, Tufekci, Fisman and Schooley6} Although most hospitals used medical masks for universal respiratory protection and source control when caring for patients without suspected or confirmed COVID-19, 5 hospitals reported using N95s for this purpose, and several explicitly tied this to high community rates. This finding may reflect the growing recognition that patients with early unsuspected infection can be highly contagious, even in the absence of symptoms, and that most clusters tend to occur in non–COVID-19 wards where transmissions can be rapidly sparked by patients or healthcare workers with unsuspected acute infections. ^{Reference Braun, Moreno and Buys7,Reference Jinadatha, Jones and Choi8} N95 respirators provide better respiratory protection for healthcare workers than medical masks and also provide better source control to reduce the risk of transmission to patients. ^{Reference Klompas, Rhee and Baker9–Reference Baker, Rhee and Tucker11} The advantages of universal N95s over medical masks, however, must be balanced against higher costs, less comfort, and logistical challenges related to fit testing.

The variability in cohorting strategies and use of AIIRs may reflect variability in the physical infrastructure and engineering capabilities of hospitals. Some hospitals, for example, can convert standard pressure rooms, wards, or even entire buildings to negative pressure, whereas others may only have a small, fixed number of AIIRs. Hospitals also varied in the criteria they used to place COVID-19 patients into AIIRs, ranging from all COVID-19 patients to just those requiring AGPs to not using AIIRs at all. Some hospitals used non-AGP risk-stratification schemes that incorporated low Ct values, severity of symptoms, duration from symptom onset, and immunocompromised status. These criteria may reflect emerging data that AGPs do not generate aerosols, though they can sometimes be proxies for factors that do increase transmission risk (eg, heavy breathing, coughing, and high viral loads). ^{Reference Brown, Gregson and Shrimpton13–Reference Thuresson, Fraenkel and Sasinovich17} Notably, few data are available on the clinical impact of AIIRs on preventing nosocomial spread of SARS-CoV-2 to other patients or healthcare workers. ^{Reference Klompas, Ye and Vaidya19}

Interestingly, 9 hospitals reported not using portable HEPA filters at all as part of their infection control strategy; on the other end of the spectrum, 1 hospital used them for all patient rooms, even those without COVID-19. Many hospitals had a more targeted strategy, including using them when COVID-19 patients met their criteria for negative pressure but AIIRs were unavailable. Some hospitals also deployed portable HEPA filters outside the rooms of COVID-19 patients in selected high-risk areas, or in shared patient rooms. The latter indication makes sense insofar as transmission risk between roommates is very high when one has an occult infection, ^{Reference Trannel, Kobayashi and Dains20,Reference Karan, Klompas and Tucker21} and portable HEPA filters have been shown to help rapidly clear infectious aerosols and airborne virus from the rooms of COVID-19 patients. ^{Reference Conway Morris, Sharrocks and Bousfield22–Reference Thuresson, Fraenkel and Sasinovich26}

Almost all hospitals reported universal testing of patients on admission. This is an important infection control measure, given that a large fraction of infectious individuals are asymptomatic, presymptomatic, or paucisymptomatic. ^{Reference Johansson, Quandelacy and Kada27} The strategies for repeated testing after admission, however, varied greatly, with some hospitals conducting repeated surveillance testing as frequently as every 4 days, while others only did so for new symptoms or known exposures. Prior studies have suggested that surveillance testing can identify occult nosocomial infections, but the overall yield may be low depending on community transmission rates. ^{Reference Smith, Pau and Fallon28}

Most hospitals used nasopharyngeal swabs for testing symptomatic patients, but many used anterior nasal swabs for asymptomatic screening. Although anterior nasal swabs may have slightly lower sensitivity than nasopharyngeal swabs, they have the advantage of greater patient comfort. ^{Reference Tsang, So, Ng, Cowling, Leung and Ip29} Interestingly, no hospitals used saliva testing, although this also appears to perform well and is noninvasive. ^{Reference Bastos, Perlman-Arrow, Menzies and Campbell30}

More than half of the hospitals used algorithms to clear precautions from patients who incidentally tested positive with high Ct values. Some of these algorithms have been shown to safely facilitate rapid discontinuation of isolation in a large fraction of SARS-CoV-2–positive patients. ^{Reference Rhee, Baker and Kanjilal31,Reference Kobayashi, Trannel and Holley32} However, the precise algorithm and Ct value varied between hospitals (with 30 being the lowest threshold used). Some hospitals may be reluctant to consider Ct values for infection control decisions, however, given potential variability between assays and sample quality, given the lack of a uniform standard for inferring absence of infectiousness, and because some hospitals’ assays or laboratories may not generate or report Ct values. ^{Reference Rhoads, Peaper and She33}

Only 22% of hospitals in this survey required regular surveillance testing of employees. Conducting routine surveillance testing for large numbers of healthcare workers poses logistical challenges and is costly. False-positive results can exacerbate staff shortages, and the benefit on top of universal masking is unclear. ^{Reference Shenoy and Weber34} Nonetheless, pre-emptive testing when community incidence rates are high (as was the case with the initial pandemic surge due to the SARS-CoV-2 omicron variant) may be a reasonable additional strategy to minimize infections in the workplace. ^{Reference Klompas, Baker and Rhee35–Reference Klompas, Baker and Griesbach37}

Our study had several limitations. First, the survey sample was drawn from 30 large academic hospitals, many with considerable resources relative to other hospitals, and with an incomplete response rate. Hence, these results cannot be considered generalizable to all US hospitals. However, our goal was to identify practices at leading institutions that many hospitals strive to emulate. Second, our study was conducted at a specific point in time, but COVID-19 infection control policies continue to evolve rapidly in response to new data and changes in local incidence rates. Third, our survey was designed only to describe the infection control policies being used across these hospitals and did not attempt to assess their effectiveness. Lastly, we attempted to focus on the most pressing questions regarding COVID-19 policies, but numerous other nuanced infection control issues likely vary across hospitals, such as the quarantine period and testing strategy for COVID-exposed inpatients, repeated testing practices prior to discharge to facilities, whether and how employee or patient vaccine and booster status affect testing strategies, and policies for visitations and visitor masking.

In conclusion, in this survey of 23 leading US hospitals, we documented some infection control policies that were common across institutions, but many more that were highly variable. This variability underscores the need for ongoing research into the effectiveness of different infection control policies. Clearer public health guidance and transparency around hospitals’ policies may also facilitate more consistent national standards.