Evaluating the impact of testing strategies for the detection of nosocomial COVID-19 in English hospitals through data-driven modeling

Abstract

Introduction: During the first wave of the COVID-19 pandemic 293,204 inpatients in England tested positive for SARS-CoV-2. It is estimated that 1% of these cases were hospital-associated using European centre for disease prevention and control (ECDC) and Public Health England (PHE) definitions. Guidelines for preventing the spread of SARS-CoV-2 in hospitals have developed over time but the effectiveness and efficiency of testing strategies for preventing nosocomial transmission has not been explored.

Methods: Using an individual-based model, parameterised using multiple datasets, we simulated the transmission of SARS-CoV-2 to patients and healthcare workers between March and August 2020 and evaluated the efficacy of different testing strategies. These strategies were: 0) Testing only symptomatic patients on admission; 1) Testing all patients on admission; 2) Testing all patients on admission and again between days 5 and 7, and 3) Testing all patients on admission, and again at days 3, and 5-7. In addition to admissions testing, patients that develop a symptomatic infection while in hospital were tested under all strategies. We evaluated the impact of testing strategy, test characteristics and hospital-related factors on the number of nosocomial patient infections.

Results: Modelling suggests that 84.6% (95% CI: 84.3, 84.7) of community-acquired and 40.8% (40.3, 41.3) of hospital-associated SARS-CoV-2 infections are detectable before a patient is discharged from hospital. Testing all patients on admission and retesting after 3 or 5 days increases the proportion of nosocomial cases detected by 9.2%. Adding discharge testing increases detection by a further 1.5% (relative increase). Increasing occupancy rates, number of beds per bay, or the proportion of admissions wrongly suspected of having COVID-19 on admission and therefore incorrectly cohorted with COVID-19 patients, increases the rate of nosocomial transmission. Over 30,000 patients in England could have been discharged while incubating a non-detected SARS-CoV-2 infection during the first wave of the COVID-19 pandemic, of which 3.3% could have been identified by discharge screening. There was no significant difference in the rates of nosocomial transmission between testing strategies or when the turnaround time of the test was increased.

Discussion: This study provides insight into the efficacy of testing strategies in a period unbiased by vaccines and variants. The findings are relevant as testing programs for SARS-CoV-2 are scaled back, and possibly if a new vaccine escaping variant emerges.

Keywords: COVID-19; SARS-CoV-2; hospital-associated (or hospital-acquired) infection; modeling; nosocomial transmission; testing.

Figure 1

Classification and infection status of SARS-CoV-2-infected hospital inpatients. (A) Number of community-acquired SARS-CoV-2 admissions (blue) and nosocomial cases (red) per day. (B) Cumulative count of community-acquired admissions and nosocomial COVID-19 cases. (C) Infection status of infected nosocomial cases on the day of discharge. (D) Proportion of nosocomial cases discharged each day by status.

Figure 2

Infection status of community and nosocomially infected inpatients. (A) Number of community and nosocomial cases that are PCR detectable or not before discharge (bars), and proportion of detectable cases that are detected and confirmed before a patient is discharged (lines). (B) Average proportion of detected cases that have a confirmed result before discharge under different turnaround times (TATs) for testing strategy 0. under all scenarios. (C) Number of cases detected per test under each scenario. (D) Average number of tests performed per day under each testing strategy. (E) Additional cases detected per additional discharge test under each scenario, and reduction in nosocomial cases per additional test performed.

Figure 3

Proportion of inpatients that developed a nosocomial infection between 03-Mar-2020 and 01-Sept-2020 under different scenarios. The baseline parameter set (see Methods) was modified to explore the effect of testing strategy (A), turnaround time (TAT, B), test sensitivity (C), hospital occupancy (D), bay size (E), and proportion of non-COVID-19 patients displaying COVID-19-like symptoms on admission (F).