A2B-COVID: A Tool for Rapidly Evaluating Potential SARS-CoV-2 Transmission Events

Abstract

Identifying linked cases of infection is a critical component of the public health response to viral infectious diseases. In a clinical context, there is a need to make rapid assessments of whether cases of infection have arrived independently onto a ward, or are potentially linked via direct transmission. Viral genome sequence data are of great value in making these assessments, but are often not the only form of data available. Here, we describe A2B-COVID, a method for the rapid identification of potentially linked cases of COVID-19 infection designed for clinical settings. Our method combines knowledge about infection dynamics, data describing the movements of individuals, and evolutionary analysis of genome sequences to assess whether data collected from cases of infection are consistent or inconsistent with linkage via direct transmission. A retrospective analysis of data from two wards at Cambridge University Hospitals NHS Foundation Trust during the first wave of the pandemic showed qualitatively different patterns of linkage between cases on designated COVID-19 and non-COVID-19 wards. The subsequent real-time application of our method to data from the second epidemic wave highlights its value for monitoring cases of infection in a clinical context.

Keywords: SARS-CoV-2; evolution; hospital; transmission.

Fig. 1.

Overview of our method. Our approach estimates the likelihood that transmission could have occurred between pairs of individuals. The model takes as input dates on which individuals became symptomatic for COVID-19 infection. Further data which can be considered includes viral genome sequence data, and time-resolved location data for each individual. Our model combines details of COVID-19 infection dynamics with a model of viral evolution, information about potential contacts between individuals, and measurement error in the sequence data. Increasing amounts of data provide increasing amounts of resolution about the potential for viral transmission.

Fig. 2.

Analysis of simulated data. Simulations were performed describing (A) direct and (B) indirect transmission events. (C) Results of analyses using A2B-COVID. 95% of data sets from direct transmission events were identified as consistent with direct transmission, as designed. Data from increasingly separated pairs of individuals showed decreasingly fewer events identified as consistent with direct transmission. (D) Days between symptom onsets for selected simulated data sets. The low mean and high variance in the time between symptom dates leads to a tradeoff between the recall and the precision of our method.

Fig. 3.

Analysis of the full data sets collected from wards X and Y. (A) Output from the A2B-COVID package given data from ward X. The plot shows potential links between cases, assessed in a pairwise fashion between potential donors (rows) and recipients (columns). Identifiers of individuals are colored in either black (patients) or red (HCWs). Squares in the grid indicate that transmission from one individual to another is consistent with our model (red), borderline (yellow), or unlikely (blue). (B) Locations of individuals linked to the ward X outbreak. Black lines indicate presence on ward X. Red lines indicate known household contacts between three individuals. Dots show times at which individuals first reported symptoms. (C) Output from the A2B-COVID package given data from ward Y. (D) Locations of individuals linked to the ward Y outbreak. Black lines indicate presence on ward Y. Red and blue lines show presence in locations other than ward Y.

Fig. 4.

Output analysis from the real-time application to clinical wards. Output from the A2B-COVID app applied to data from a COVID-19 ward during the second wave of infection in the UK. Data from the patients 1, 2, and 3 is consistent with the direct infection of the health care workers HCW_1 and HCW_2.

Fig. 5.

Notation used in our method. An overview of our model for transmission events is shown in figure 3. We divide time into discrete days. For the individual A, we denote by S_A the date at which that individual became symptomatic, and by D_A the date at which a sample of viruses were collected for genome sequencing. For each pair of individuals A and B, we denote by w_AB(t) the probability that A and B were colocated on day t. Within our model, we assume that dates of sample collection are known, whereas times of symptom onset are known or estimated. Using these data, in combination with viral sequence data, we calculate a statistic describing the consistency of the data with individual A having infected individual B on any given day T. Summing this statistic across T, we obtain an estimate of the consistency of our data with transmission having occurred between the two individuals.