Reconstruction of transmission chains of SARS-CoV-2 amidst multiple outbreaks in a geriatric acute-care hospital: a combined retrospective epidemiological and genomic study

Mohamed Abbas; Anne Cori; Samuel Cordey; Florian Laubscher; Tomás Robalo Nunes; Ashleigh Myall; Julien Salamun; Philippe Huber; Dina Zekry; Virginie Prendki; Anne Iten; Laure Vieux; Valérie Sauvan; Christophe E Graf; Stephan Harbarth

doi:10.7554/eLife.76854

. 2022 Jul 19;11:e76854. doi: 10.7554/eLife.76854

Reconstruction of transmission chains of SARS-CoV-2 amidst multiple outbreaks in a geriatric acute-care hospital: a combined retrospective epidemiological and genomic study

Mohamed Abbas ^1,^2,^3,^✉, Anne Cori ^2,⁴, Samuel Cordey ^3,⁵, Florian Laubscher ⁵, Tomás Robalo Nunes ^1,⁶, Ashleigh Myall ^7,⁸, Julien Salamun ⁹, Philippe Huber ¹⁰, Dina Zekry ¹⁰, Virginie Prendki ^10,¹¹, Anne Iten ¹, Laure Vieux ¹², Valérie Sauvan ¹, Christophe E Graf ¹⁰, Stephan Harbarth ^1,^3,¹¹

Editors: Joshua T Schiffer¹³, Wendy S Garrett¹⁴

PMCID: PMC9328768 PMID: 35850933

Abstract

Background:

There is ongoing uncertainty regarding transmission chains and the respective roles of healthcare workers (HCWs) and elderly patients in nosocomial outbreaks of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) in geriatric settings.

Methods:

We performed a retrospective cohort study including patients with nosocomial coronavirus disease 2019 (COVID-19) in four outbreak-affected wards, and all SARS-CoV-2 RT-PCR positive HCWs from a Swiss university-affiliated geriatric acute-care hospital that admitted both Covid-19 and non-Covid-19 patients during the first pandemic wave in Spring 2020. We combined epidemiological and genetic sequencing data using a Bayesian modelling framework, and reconstructed transmission dynamics of SARS-CoV-2 involving patients and HCWs, to determine who infected whom. We evaluated general transmission patterns according to case type (HCWs working in dedicated Covid-19 cohorting wards: HCW_covid; HCWs working in non-Covid-19 wards where outbreaks occurred: HCW_outbreak; patients with nosocomial Covid-19: patient_noso) by deriving the proportion of infections attributed to each case type across all posterior trees and comparing them to random expectations.

Results:

During the study period (1 March to 7 May 2020), we included 180 SARS-CoV-2 positive cases: 127 HCWs (91 HCW_covid, 36 HCW_outbreak) and 53 patients. The attack rates ranged from 10% to 19% for patients, and 21% for HCWs. We estimated that 16 importation events occurred with high confidence (4 patients, 12 HCWs) that jointly led to up to 41 secondary cases; in six additional cases (5 HCWs, 1 patient), importation was possible with a posterior probability between 10% and 50%. Most patient-to-patient transmission events involved patients having shared a ward (95.2%, 95% credible interval [CrI] 84.2%–100%), in contrast to those having shared a room (19.7%, 95% CrI 6.7%–33.3%). Transmission events tended to cluster by case type: patient_noso were almost twice as likely to be infected by other patient_noso than expected (observed:expected ratio 2.16, 95% CrI 1.17–4.20, p=0.006); similarly, HCW_outbreak were more than twice as likely to be infected by other HCW_outbreak than expected (2.72, 95% CrI 0.87–9.00, p=0.06). The proportion of infectors being HCW_covid was as expected as random. We found a trend towards a greater proportion of high transmitters (≥2 secondary cases) among HCW_outbreak than patient_noso in the late phases (28.6% vs. 11.8%) of the outbreak, although this was not statistically significant.

Conclusions:

Most importation events were linked to HCW. Unexpectedly, transmission between HCW_covid was more limited than transmission between patients and HCW_outbreak. This finding highlights gaps in infection control and suggests the possible areas of improvements to limit the extent of nosocomial transmission.

Funding:

This study was supported by a grant from the Swiss National Science Foundation under the NRP78 funding scheme (Grant no. 4078P0_198363).

Research organism: Viruses

Introduction

Nosocomial acquisition of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) in geriatric institutions and long-term care facilities (LTCFs) may account for large proportions of all declared coronavirus disease 2019 (Covid-19) cases in many countries, and contribute substantially to morbidity and mortality (Thiabaud et al., 2021; Bhattacharya et al., 2021; The Guardian, 2021b; The Guardian, 2021a). Because the reservoir of SARS-CoV-2 in healthcare environments may contribute to amplifying the pandemic (Knight et al., 2022), we need to better understand transmission dynamics in these settings.

The terms healthcare-associated, hospital-onset, and nosocomial Covid-19 reflect the uncertainty around defining and distinguishing community- versus healthcare-acquired Covid-19 cases (Abbas et al., 2021c). Nevertheless, in some settings, such as LTCFs and nursing homes, these definitions are relatively straightforward. In other settings, such as those with a high patient turnover, or where patients are admitted from the community and both Covid-19 and non-Covid-19 cases are hospitalised in the same institution, defining, and more importantly detecting cases is crucial to avoid cross-contamination. Determining sources and transmission pathways of infection may thus help improve infection prevention and control (IPC) strategies.

The role of healthcare workers (HCWs) in nosocomial Covid-19 transmission dynamics is complex, as they can be victims and/or vectors of SARS-CoV-2 infection, and can acquire from or transmit to their peers and patients and the community (Abbas et al., 2021b; Ellingford et al., 2021). There is ongoing controversy and uncertainty surrounding the role of HCWs in infecting patients during nosocomial outbreaks, and findings from acute-care hospitals cannot be applied directly to LTCFs and geriatric hospitals (Abbas et al., 2021a; Klompas et al., 2021; Lucey et al., 2021; Aggarwal et al., 2022).

The aim of this study was to reconstruct transmission dynamics in several nosocomial outbreaks of SARS-CoV-2 involving patients and HCWs in a Swiss university-affiliated geriatric hospital that admitted both Covid-19 and non-Covid-19 patients during the first pandemic wave in Spring 2020.

Methods

We performed a retrospective cohort study of all patients with nosocomial Covid-19 in four outbreak-affected wards, and all SARS-CoV-2 RT-PCR positive HCWs from 1 March to 7 May 2020. This study is reported according to the STROBE (von Elm et al., 2007) (Strengthening the Reporting of Observational Studies in Epidemiology) and ORION (Stone et al., 2007) (Outbreak Reports and Intervention studies Of Nosocomial infection) statements.

Setting

The Hospital of Geriatrics, part of the Geneva University Hospitals (HUG) consortium, has 196 acute-care and 100 rehabilitation beds. During the first pandemic wave, a maximum of 176 acute-care beds were dedicated to admitting geriatric patients with Covid-19 who were not eligible for escalation of therapy (e.g., intensive care unit admission) (Mendes et al., 2020). During the same period, patients were also admitted for non-Covid-19 hospitalisations, and the rehabilitation beds were also open to patients convalescing from Covid-19. Beginning on 1 April 2020, we use RT-PCR to screen all patients on admission for SARS-CoV-2. Between 7 April 2020 and 30 May 2020, we screened patients in non-Covid wards. We encouraged HCWs from outbreak wards to undergo PCR testing on nasopharyngeal swabs between 9 and 16 April 2020, even if they were asymptomatic. We described additional IPC measures in Appendix 1.

Definitions

We defined healthcare-associated (HA) Covid-19 by an onset of symptoms ≥5 days after admission in conjunction with a strong suspicion of healthcare transmission, in accordance with Swissnoso guidelines (Swissnoso, 2021). We classified patients with HA-Covid-19 as ‘patient_noso’, the others were assumed to be community-acquired or ‘patient_community’. An outbreak was declared when ≥3 cases of HA-Covid-19 (HCWs and patients) with a possible temporal-spatial link were identified (Swissnoso, 2021). HCWs were included in the outbreak investigation if they had a positive RT-PCR for SARS-CoV-2. We classified HCWs as ‘HCW_covid’ if they worked in a Covid-19 cohorting ward admitting community-acquired cases, or ‘HCW_outbreak’ if they worked in a ‘non-Covid’ ward (i.e., not admitting community-acquired Covid-19 cases) in which nosocomial outbreaks occurred. HCWs worked in either one or another type of ward, except for one HCW who worked in both Covid and non-Covid wards (although the proportion and/or days worked in each type of ward is unknown). Six HCWs worked across multiple wards (e.g., on-call) and were attributed to Covid-wards for the analyses.

Data sources

The data used for this study were from the sources described previously (Abbas et al., 2021a). First, we used prospectively collected data from the Swiss Federal Office of Public Health-mandated surveillance of hospitalised Covid-19 patients (Thiabaud et al., 2021). We also used prospectively collected data from HUG’s Department of Occupational Health for symptom-onset data and the Department of Human Resources (HR) for HCW shifts. Dates of symptom onset were available for both patients and HCWs from each source, respectively.

Descriptive epidemiology

We produced an epidemic curve using dates of symptom onset; where these were unavailable (e.g., asymptomatic cases), we imputed them with the median difference between date of symptom onset and date of nasopharyngeal swab.

Microbiological methods

All Covid-19 cases in the outbreak were confirmed by RT-PCR on nasopharyngeal swabs. We performed SARS-CoV-2 whole-genome sequencing (WGS) using an amplicon-based sequencing method to produce RNA sequences, as previously described (Abbas et al., 2021a) and summarised in Appendix 1.

Phylogenetic analysis

We performed sequence alignment with MUSCLE (v3.8.31). We employed MEGA X (Kumar et al., 2018) using the Maximum Likelihood method and Tamura three-parameter model (Tamura, 1992) to conduct the evolutionary analyses. We integrated to the phylogenetic analysis all the complete genomes SARS-CoV-2 sequenced by the Laboratory of Virology (HUG) for the purposes of epidemiological surveillance in the community.

Statistical analysis

We performed descriptive statistics with medians and interquartile ranges (IQRs), and counts and proportions, as appropriate.

Reconstruction of transmission trees

We combined epidemiological and genomic data to reconstruct who infected whom using the R package outbreaker2 (Jombart et al., 2014; Campbell et al., 2018), as described elsewhere (Abbas et al., 2021a) and in Appendix 1. Briefly, the model uses a Bayesian framework, combining information on the generation time (time between infections in an infector/infectee pair), and contact patterns, with a model of sequence evolution to probabilistically reconstruct the transmission tree (see Appendix 1).

Because formal contact tracing was limited during the study period, we constructed contact networks based on ward or room presence for patients based on their ward movements, and on HCW shifts obtained from HR. We defined a contact as simultaneous presence on the same ward on a given day (see Appendix 1). The manner by which outbreaker2 handles these contacts is conservative in that it allows for non-infectious contacts to occur (false positives) and incomplete reporting of infectious contacts (false negatives). In addition, the model estimates the proportions of these contacts.

Using the reconstructed transmission trees, we determined the number of imported cases (with minimal posterior support of 10%), and the number of secondary cases generated by the imports. Imported caseswere defined as those that do not have apparent ancestors among the cases included in the outbreak. We also calculated the number of secondary (i.e., onward) infections for each case, that is, the individual reproductive number (R), which we stratified by epidemic phase (early or late with a cut-off on 9 April 2020) and case type (HCW_covid, HCW_outbreak, patient_noso, and patient_community, see Appendix 1).

We assessed the role of each case type in transmission by estimating the proportion of infections attributed to the case type (f_case), which we compared with the random expectation considering the prevalence of each case type (see Appendix 1). To better understand the transmission pathways between and within wards, we also estimated (for outbreak and non-outbreak wards), the proportion of infections attributed to infectors in the same ward. We also constructed a matrix representing ward-to-ward transmission. Patient movements between wards were constructed using the implementation of the vistime package (visualisation tool) as in the publication by Meredith et al., 2020. Statistical analyses were performed in R software version 4.0.3 (https://www.R-project.org/).

Results

During the study period, we included a total of 180 SARS-CoV-2 positive cases: 127 HCWs of whom 91 HCW_covid, and 36 HCW_outbreak, and 53 patients from the four outbreak wards. Of the 53 included patients, post hoc epidemiological analysis showed that 4 of these likely acquired Covid-19 in the community (CA-Covid-19). The remaining 49 nosocomial cases represented 20.2% (49/242) of all patients hospitalised with Covid-19, and 81.7% (49/60) of nosocomial Covid-19 cases in the Geriatric Hospital. The ward-level attack rates ranged from 10% to 19% among patients. Moreover, 21% of all HCWs in the geriatric hospital had a PCR-positive test. The epidemic curve is shown in Figure 1, and ward-level epidemic curves in Figure 1—figure supplement 1. Characteristics of patients and HCWs are summarised in Tables 1 and 2, respectively. Strikingly, the time period between date of onset of symptoms and date of swab was shorter for HCW_covid (mean 1.6 days, SD 1.78) than for HCW_outbreak (mean 2.88 days, SD 4.84).

Figure 1. — Includes eight asymptomatic cases for whom date of onset was inferred (c.f., text). HCW, healthcare worker.

Figure 1—figure supplement 1. — Includes eight asymptomatic cases for whom date of onset was inferred (c.f., text). HCW, healthcare worker.

Table 1. Characteristics of Covid-19 patients with nosocomial acquisition.

Characteristics	All patients(N=49)
Female, n (%)	28 (57.1)
Age, median (IQR)	85.4 (83.5–89.3)

Asymptomatic, n (%)	3 (6.1)
Onset of symptoms before swab date, n (%)	12 (24.5)
Days from onset of symptoms to swab, median (IQR)	0 (0–0)
Days from onset of symptoms to swab, mean (SD)	–0.29 (2.19)

Open in a new tab

Table 2. Characteristics of SARS-CoV-2 RT-PCR positive healthcare workers.

Characteristics	All HCWs(N=127)
Female, n (%)	92 (72.4)
Age, median (IQR)	32.0 (43.3–54.8)

Profession, n (%)
Nurse	57 (44.9)
Nurse assistant	39 (30.7)
Doctor	19 (15.0)
Care assistant	4 (3.2)
Transporter	4 (3.2)
Physical therapist	2 (1.6)
Speech therapist	1 (0.8)
Medical student	1 (0.8)

Asymptomatic, n (%) missing data for 5	5 (3.9)
Days from onset of symptoms to swab, median (IQR)	1 (−2 to 21)
HCWs in Covid-19 wards (HCW_covid)	1 (1–2)
HCWs in non-Covid (outbreak) wards (HCW_outbreak)	1 (0–3)
Days from onset of symptoms to swab, mean (SD)	1.91 (2.86)
HCWs in Covid-19 wards (HCW_covid)	1.60 (1.78)
HCWs in non-Covid (outbreak) wards (HCW_outbreak)	2.88 (4.84)

Open in a new tab

Phylogenetic tree

We obtained SARS-CoV-2 sequences for 148 isolates of the 180 cases (82.2%), including 105 HCWs (82.7%) and 43 patients (81.1%). A rooted phylogenetic tree found substantial genetic diversity, with at least nine clusters and sub-clusters (Figure 2). One cluster (with moderate bootstrap support [BS] 26%) comprised sequences from 17 HCWs, 3 patients, and 6 community isolates in multiple subclusters (e.g., BS 72% with signature mutation C5239T). Another cluster (BS 78% with signature mutations C28854T and A20268G) showed a HCW sequence (H1048) with high similarity with community isolates. There was also a large cluster (BS 68% with signature mutations C8293T, T18488C, and T24739C) with several subclusters includes 19 HCWs, 9 patients, and 3 community cases; ward movements for the patients are shown in Appendix 1—figure 1. A well-defined cluster (BS 100%) included isolates from patients and HCWs from the same ward.

Figure 2. — The tree includes 148 sequences related to the outbreak (patient and employee sequences are named C1xx [blue] and H10xx [red], respectively), alongside the community cases in the canton of Geneva, Switzerland, that were sequenced in March–April 2020 by the Laboratory of Virology (Geneva University Hospitals) and submitted to GISAID (virus names and accession ID [i.e., EPI_ISL_] are indicated) in the context of an epidemiological surveillance. For each sequence the date of the sample collection is mentioned (yyyy-mm-dd).

Imported cases

From the reconstructed trees, we identified 22 imports in total (17 HCWs, 5 patients) with posterior support ≥10%. The 22 imported cases generated 41 secondary cases (posterior support ≥10%), with a median posterior support of 32.4% (IQR 17.0%–53.7%). When restricting to imports with ≥50% posterior support, there were 16 imported cases 16 (12 HCWs, 4 patients), generating 35 secondary cases. There was some degree of uncertainty, reflected by circular transmission pathways, in determination of imported cases and their secondary cases. There were six transmission pairs (C114-C115, C153-H1057, H1008-H1059, H1011-H1019, H1017-H12021, and H1052-H1082) where there was uncertainty as to which of the cases was imported and which was a secondary case, that is, for each case in the pair there was a ≥10% probability of importation, but also ≥10% probability of being a secondary case of an imported case. Therefore, in total, 29 cases were ‘pure’ secondary cases of imported cases (Table 3).

Table 3. Imported cases and secondary infections, patients and HCWs are named C1xx and H10xx, respectively.

Imported case	Posterior probability of importation	Secondary onward transmission by imported case	Posterior probability of onward transmission
C107	100	H1077 C131 C124 H1005 C125 H1034 H1068 C112 C116	100 72.5 39.4 35.0 32.9 27.3 18.5 15.9 11.7
C114^*	42.5	C115^*	42.5
C115^*	57.5	C114^*	57.5
C123	96.4	H1058 H1036 H1047	90.1 16.0 11.1
C153^*	51.7	H1057^*	51.6
H1008^*	85.7	H1059^*	85.7
H1011^*	65.9	H1019^*	61.7
H1012	100	N/A	N/A
H1013	100	N/A	N/A
H1015	100	N/A	N/A
H1017^*	52.3	H1020 H1021^*	100.0 52.3
H1019^*	34.1	H1011^*	28.2
H1021^*	47.7	H1017^*	47.7
H1025	86.6	H1085 H1031	95.7 41.5
H1048	100	N/A	N/A
H1052^*	18.9	H1082^*	18.9
H1057^*	48.3	C153^*	48.3
H1059^*	14.3	H1008^*	14.3
H1073	100	N/A	N/A
H1082^*	81.1	H1052^*	81.1
H1110	85.1	H1063 H1064 H1041 H1024 H1091 H1065	58.0 32.4 30.0 22.5 22.1 17.2
H1122	84.5	C113 C104 H1033 H1003 H1004 H1044	53.7 26.0 17.0 15.5 15.0 10.6

Open in a new tab

N/A: not applicable.

Uncertainty in transmission (i.e., case could either be an imported case or a secondary case).

Reconstructing who infected whom

Figure 3 shows the distribution of posterior support when considering the ancestry from the individual (i.e., Cxxx or Hxxx), case type (HCW_covid, HCW_outbreak, patient_noso, and patient_community), ward, or ward type (outbreak or non-outbreak wards). There was less confidence in attribution of ancestry when considering by individual or ward, when compared with case type or ward type. The output from the ancestry reconstruction is shown in Figure 3—figure supplement 1 and Figure 3—figure supplement 1—source code 1. The model estimated that the reporting probability was 91.2% (95% credible interval [CrI] 86.6%–95.2%), suggesting that only 8.8% of source cases involved in transmission were not identified. For most (90.8%) cases, the model identified the direct infector, without intermediate unobserved cases (Figure 3—figure supplement 2).

Infectors are on the vertical axis and infectees are on the horizontal axis. Each bubble represents the posterior probability of each infector-infectee transmission pair. The bottom row denotes the probability that an infectee was in fact an imported case. Patients and employees are named C1xx and H10xx, respectively.

Figure 3—figure supplement 1—source code 1. Interactive ancestry plot (who infected whom) – identical to Figure 3—figure supplement 1.

elife-76854-fig3-figsupp1-code1.zip^{(1.6MB, zip)}

Ward attribution

The epidemiological attribution of the presumptive ward on which patients became infected and that suggested by the model output (see Appendix 1) agree for 95% of the nosocomial cases. The modelling analysis modified the ward attribution for three patients.

Transmission patterns

Among patient-to-patient transmission events, and across all posterior trees, 95.2% (95% CrI 84.2%–100%) involved patients who shared a ward during their hospital stay. In contrast, only 19.7% (95% CrI 6.7%–33.3%) of patient-to-patient transmissions involved patients who had shared a room. The model predicted that C107 infected C131 with a 72.5% probability although they did not share a ward (Appendix 1—figure 1B for ward movements); the probabilities that this was a direct infection and indirect infection with an unreported intermediate infector were 38.3% and 34.2%, respectively (Appendix 1—figure 1B for ward movements).

Secondary infections

The number of secondary infections caused by each infected case (individual reproductive number R, estimated from the transmission tree reconstruction), ranged from 0 to 9 (Figure 4). We compared the proportion of cases with no secondary transmissions (non-transmitters) and of cases with ≥2 secondary transmissions (high transmitters) across case types and outbreak phase. We found that the proportion of non-transmitters among both HCW_outbreak and patient_noso was smaller in the early than in the late stage (approximately 32% in early and 55% in late phase for both groups), suggesting that the contribution of these groups to ongoing transmission decreased over the study period. Conversely, the proportion of non-transmitters among HCW_covid was stable at about 55%–60% across the early and the late phase. The proportions of high transmitters were higher among HCW_outbreak than either patient_noso or HCW_covid in the late phases (28.6% vs. 11.8% and 13.9%) of the outbreak. However, due to small numbers, these differences were not statistically significant. These trends were similar in the sensitivity analyses (Appendix 1—table 2).

Figure 4. — Number of cases in early phase: HCW_outbreak 19, HCW_covid 43, patient_noso 25, patient_community 1. Number of cases in late phase: HCW_outbreak 7, HCW_covid 36, patient_noso 17, patient_community 0. HCW, healthcare worker.

Role of HCWs and patients in transmission events

We found that cases were significantly less likely than expected at random to be infected by HCWs from COVID wards (proportion infected by HCW_covid, f_HCW=42%; 95% CrI 36%–49% vs. 53% expected at random; 95% CrI 44%–62%; p=0.042). This was true across all cases, but particularly among HCWs in outbreak wards (f_HCW=31%, 95% CrI 17%–48%; p=0.07) and patients (f_HCW=24%, 95% CrI 12%–37%, p=0.006). Conversely HCW_outbreak were significantly more likely than expected at random to become infected by other HCW_outbreak (proportion infected by HCW_outbreak, f_outbreak = 38%, 95% CrI 22%–52% v. 18%; 95% CrI 4%–35%, p=0.03). Patients with nosocomial Covid-19 (patient_noso) were significantly more likely than expected at random to be infected by other patient_noso (proportion infected by patient_noso, f_pat=56%, 95% CrI 41%–71% vs. 28%; 95% CrI 14%–45%, p=0.005). Full results are shown in Figure 5.

Figure 5. — The blue histograms indicate the expected random distributions of f_case, given the prevalence of each case type. The red histograms show the observed distribution of f_case, across 1000 transmission trees reconstructed by `outbreaker2`. (A) All cases. (B) Transmission to HCWs in Covid-19 wards only. (C) Transmission to HCWs in non-Covid-19 wards (i.e., outbreak wards) only. (D) Transmission to patients with nosocomial Covid-19 only. HCW, healthcare worker.

Role of within-ward and between-ward transmission

Infected staff or patients in outbreak wards were responsible for significantly more transmission in general (54%; 95% CrI 48%–61%) than expected (43.7%, 95% CrI 35%–53%). This was driven in particular by transmission from staff or patients to other staff or patients in outbreak wards (73%, 95% CrI 63%–82%) (Figure 5). Within-ward transmission was more pronounced in outbreak wards (mean 48%; range: 20%–70% of all infections within a ward) than in non-outbreak wards (mean 14%; range 0%–63%) as shown in Appendix 1—figures 2 and 3.

Discussion

This in-depth investigation of SARS-CoV-2 transmission between patients and HCWs in a geriatric hospital, including several nosocomial outbreaks, has provided many valuable insights on transmission dynamics. First, we showed that the combination of epidemiological and genetic data using sophisticated modelling enabled us to tease out overall transmission patterns. Second, we showed that transmission dynamics among HCWs differed according to whether they worked in Covid wards or in wards where outbreaks occurred. We found that HCW-to-HCW transmission in Covid wards was not higher than expected but the risk of transmission between HCWs in non-Covid wards was twofold higher than expected. Third, we identified excess patient-to-patient transmission events, most of which occurred within the same ward, but not necessarily the same room. Fourth, we identified multiple importation events that led to a substantial number of secondary cases or clusters; most were related to HCWs, but one was related to a patient with community-acquired Covid-19.

These results are particularly important, as settings which care for elderly patients, such as geriatrics and rehabilitation clinics or LTCFs have high attack rates of SARS-CoV-2 for both patients and HCWs (Abbas et al., 2021b). In an institution-wide seroprevalence study in our hospital consortium, the Department of Rehabilitation and Geriatrics, of which the hospital in this study was part, had the highest proportion of HCWs with anti-SARS-CoV-2 antibodies (Martischang et al., 2022).

The different transmission patterns between HCWs in Covid wards and outbreak wards (i.e., meant to be ‘Covid-free’) is intriguing. HCWs' behaviour may have affected transmission. Indeed, a previous study Ottolenghi et al., 2021 found that HCWs caring for Covid-19 patients were concerned about becoming infected while caring for patients, and therefore may apply IPC measures more rigorously than when caring for patients who do not have Covid-19. HCWs working in non-Covid wards may not have felt threatened by Covid-19 patients who were in principle allocated to other wards, and thus not in their direct care. HCWs in non-Covid wards also may have underestimated the transmission risk from their peers to a greater extent than those working in Covid wards, and thus may not have maintained physical distancing well. In addition, we found a higher mean duration (2.9 days) of presenteeism despite symptoms compatible with Covid-19 among HCWs working in non-Covid wards than for those working in Covid wards (1.6 days), which gives credence to the abovementioned possible explanations for the different transmission patterns. Other factors (e.g., work culture, baseline IPC practices) also may have affected transmission patterns.

Most patient-to-patient transmission events involved patients who were hospitalised in the same ward, and were therefore in close proximity. We cannot exclude transmission from a ‘point-source’ or via a HCW’s contaminated hands (i.e., an unidentified HCW who transmitted SARS-CoV-2 to multiple patients in the same ward). To date, we have little evidence to suggest that this is the case; indeed, the transmission patterns were robust to changes in the model assumptions. Mathematical models have suggested that single-room isolation of suspected cases could potentially reduce the incidence of nosocomial SARS-CoV-2 transmission by up to 35% (Evans et al., 2021). In the outbreaks we describe, symptomatic patients were identified promptly, with a median delay of 0 days between symptom onset and their first positive test. However, these precautions may not be sufficient as patients may transmit the virus when they are pre-symptomatic (Ferretti et al., 2020). Thus, infection prevention teams may need to identify patients at high risk of developing nosocomial Covid-19 (Myall et al., 2021) if single rooms are not available for all exposed patients (e.g., in cases of overcrowding). For example, a previous study Mo et al., 2021 found that exposure to community-acquired cases who were identified and segregated or cohorted was associated with half the risk of infection compared with exposure to hospital-acquired cases or HCWs who may be asymptomatic. One possible explanation for this finding is that patients with CA-Covid may have passed the peak of infectiousness when they are admitted, whereas patients with HA-Covid cases have frequent unprotected contact with HCWs and other patients during their period of peak infectiousness.

The current evidence does not support the use of real-time genomics for control of SARS-CoV-2 nosocomial outbreaks (Stirrup et al., 2021; Stirrup et al., 2022). Nevertheless, in this investigation, as in many others, we performed WGS to investigate transmission patterns (Aggarwal et al., 2022). Although we were able to gain considerable insight from the powerful combination of genetic sequencing data and rich epidemiological data, we controlled the outbreaks without using WGS, as was the case in many published reports (Arons et al., 2020; Taylor et al., 2020). Furthermore, WGS may be more useful for ruling out transmission rather than for confirmatory purposes, due to the low number of mutations accumulated in the SARS-CoV-2 genome between transmission pairs (Braun et al., 2021).

Our study has several strengths. To the best of our knowledge, it is the only extensive outbreak investigation that employed WGS and sophisticated modelling to assess SARS-CoV-2 transmission in a geriatric acute-care hospital. We performed WGS on isolates from a high proportion (80%) of cases, including those from HCWs which has been previously shown to improve understanding of transmission dynamics (Ellingford et al., 2021). In addition, we collected the data prospectively, thereby minimising the risk of bias.

Despite these strengths, some limitations must be acknowledged. First, we included only one sequence from a CA-Covid case. However the method we used to reconstruct who infected whom is able to cope with and identify missing intermediate cases, which allowed us to estimate that the overwhelming majority of cases (91.5%) was captured in our sample. Some misclassification may have occurred, for example, whether an imported cases is truly nosocomial or not. For example, the model predicted that patients C107 and C115 were imported cases. The predicted dates of infection were 14–22 March for case C107 and 8–24 March for case C115; their admission dates were 3 March and 18 March, respectively. The probability that infection occurred on or after date of admission for case C115 was 81%. So the fact that the cases are labelled as ‘imported’ does not preclude the fact that these were still nosocomial cases, simply that their infector was not identified in this outbreak. We did not collect data on adherence to Covid-specific IPC recommendations by HCWs in different wards. We were unable to relate the number of secondary infections with the population (ward) size; more complex models would be required to explain the underlying mechanisms in a context of fluctuating denominators, for example, due to ward closures, and so on. In addition, we performed these investigations during the first pandemic wave with a wild-type variant of SARS-CoV-2 in a susceptible population. For these reasons, the results may no longer be applicable in settings with high vaccination coverage and/or substantial natural immunity, or in later stages of the pandemic, also due to accrued experience in managing and preventing outbreaks. Nevertheless, the lessons learned may be useful in a large number of countries with slow vaccine roll-out due to vaccine hesitancy, particularly among HCWs where there is no vaccine mandate, or unequal access to vaccine supplies (The Lancet Infectious Diseases, 2021). Furthermore, nosocomial outbreaks of SARS-CoV-2 still occur despite high vaccination coverage (Burugorri-Pierre et al., 2021; Shitrit et al., 2021). Also, these valuable lessons may be applicable for nosocomial outbreak control in the case of future pandemics due to respiratory viruses with characteristics similar to SARS-CoV-2.

In conclusion, strategies to prevent nosocomial SARS-CoV-2 transmission in geriatric settings should take into account the potential for patient-to-patient transmission and the transmission dynamics between HCWs in non-Covid wards, which our study suggests may differ from those in dedicated Covid-19 wards.

Acknowledgements

The authors would like to thank Frédéric Bouillot (Geneva University Hospitals) for providing us with HR data, Aurore Britan (Geneva University Hospitals) for data management, Rachel Goldstein for data collection (Geneva University Hospitals), Daniel Teixeira for data extraction (Geneva University Hospitals). In addition, the authors would like to thank Stéphanie Baggio, Frédérique Jacquerioz Bausch, Hervé Spechbach from the AMBUCoV study (Geneva University Hospitals), Amaury Thiabaud (University of Geneva), and the IPC team members from Geneva University Hospitals involved in SARS-CoV-2 outbreak management (Pascale Herault, Didier Pittet, Walter Zingg). Thibaut Jombart (London School of Hygiene and Tropical Medicine) and Finlay Campbell (World Health Organization) provided input on technical aspects of the outbreaker2 model development. The study team also wishes to thank the dedicated HCWs who cared for the patients, and the Geriatric Hospital Covid-19 crisis management team (Gabriel Gold, Charline Couderc, Etienne Satin). This work was supported by a grant from the Swiss National Science Foundation under the NRP78 funding scheme (Grant no. 4078P0_198363). Mohamed Abbas and Anne Cori are supported by the National Institute for Health Research (NIHR) Health Protection Research Unit in Modelling and Health Economics, a partnership between Public Health England, Imperial College London and LSHTM (grant code NIHR200908). Anne Cori also acknowledges funding from the MRC Centre for Global Infectious Disease Analysis (reference MR/R015600/1), jointly funded by the UK Medical Research Council (MRC) and the UK Foreign, Commonwealth & Development Office (FCDO), under the MRC/FCDO Concordat agreement and is also part of the EDCTP2 programme supported by the European Union. Ashleigh Myall was supported by a scholarship from the Medical Research Foundation National PhD Training Programme in Antimicrobial Resistance Research (MRF-145-0004-TPG-AVISO). The views expressed are those of the author(s) and not necessarily those of the NIHR, Public Health England or the Department of Health and Social Care.

Appendix 1

Appendix 1—table 1. Composition of baseline `outbreaker2` model and different sensitivity analyses.

Scenario type	Onset of symptoms	Genetic data	Contact data	Short serial interval	Longer serial interval	Low outlier threshold	Medium outlier threshold	Default outlier threshold
Baseline scenario	X	X	X	X		X

Sensitivity analyses
1.	X	X		X		X
2.	X	X	X		X	X
3.	X	X	X^*	X		X
4.	X	X	X^†	X		X
5.	X	X	X	X			X
6.	X	X	X	X				X

Open in a new tab

For this model, we used the HR data for healthcare worker presence (with some corrections).

^†

For this model, we assumed that patients were no longer infectious after the date of positive RT-PCR.

Appendix 1—table 2. Proportions of secondary infections (i.e., individual R) for each case type (patient_noso, HCW_outbreak, nd HCW_covid) in early (up to 9 April 2020) and late (as of 10 April 2020) phases of the study.

The p-values are for chi-squared tests on these proportions.

	Patient_noso	HCW_outbreak	HCW_covid	p-value(HCW_outbreak vs. patient_noso)
Main analysis
≥1 secondary transmission
Early phase	0.68 (0.52–0.8)	0.684 (0.526–0.789)	0.442 (0.349–0.535)	0.576
Late phase	0.471 (0.294–0.647)	0.429 (0.286–0.714)	0.417 (0.306–0.528)	0.552
≥2 secondary transmissions
Early phase	0.32 (0.2–0.44)	0.316 (0.158–0.474)	0.186 (0.093–0.256)	0.459
Late phase	0.118 (0–0.294)	0.286 (0–0.571)	0.139 (0.056–0.222)	0.807

Sensitivity analysis #1 (no assumptions about contacts)
≥1 secondary transmission
Early phase	0.64 (0.48–0.76)	0.632 (0.474–0.789)	0.442 (0.326–0.512)	0.527
Late phase	0.529 (0.353–0.706)	0.429 (0.143–0.714)	0.389 (0.278–0.5)	0.322
≥2 secondary transmissions
Early phase	0.36 (0.24–0.52)	0.263 (0.105–0.421)	0.163 (0.093–0.256)	0.190
Late phase	0.176 (0.059–0.353)	0.143 (0–0.429)	0.111 (0.028–0.195)	0.527

Sensitivity analysis #2 (long serial interval)
≥1 secondary transmission
Early phase	0.64 (0.52–0.76)	0.684 (0.526–0.791)	0.442 (0.349–0.535)	0.669
Late phase	0.471 (0.294–0.647)	0.429 (0.282–0.714)	0.389 (0.278–0.5)	0.561
≥2 secondary transmissions
Early phase	0.36 (0.2–0.48)	0.316 (0.158–0.474)	0.186 (0.116–0.279)	0.445
Late phase	0.118 (0–0.294)	0.286 (0–0.571)	0.139 (0.056–0.222)	0.790

Sensitivity analysis #3 (calibrating contacts based on assumptions on infectiousness)
≥1 secondary transmission
Early phase	0.68 (0.56–0.8)	0.684 (0.526–0.789)	0.442 (0.349–0.535)	0.433
Late phase	0.471 (0.235–0.647)	0.286 (0.143–0.571)	0.361 (0.25–0.472)	0.249
≥2 secondary transmissions
Early phase	0.4 (0.24–0.52)	0.368 (0.211–0.474)	0.186 (0.116–0.256)	0.360
Late phase	0.118 (0–0.294)	0.143 (0–0.429)	0.083 (0.028–0.167)	0.702

Sensitivity analysis #4 (patients no longer infectious after date of swab)
≥1 secondary transmission
Early phase	0.64 (0.48–0.8)	0.684 (0.526–0.842)	0.465 (0.349–0.535)	0.627
Late phase	0.471 (0.294–0.706)	0.429 (0.286–0.714)	0.417 (0.306–0.528)	0.522
≥2 secondary transmissions
Early phase	0.32 (0.16–0.44)	0.316 (0.158–0.475)	0.186 (0.116–0.279)	0.541
Late phase	0.118 (0–0.294)	0.286 (0–0.571)	0.111 (0.028–0.194)	0.781

Sensitivity analysis #5 (higher value for outlier threshold)
≥1 secondary transmission
Early phase	0.64 (0.48–0.721)	0.684 (0.526–0.789)	0.419 (0.326–0.512)	0.683
Late phase	0.471 (0.294–0.647)	0.429 (0.143–0.714)	0.417 (0.306–0.528)	0.542
≥2 secondary transmissions
Early phase	0.32 (0.16–0.48)	0.316 (0.158–0.474)	0.186 (0.093–0.256)	0.407
Late phase	0.118 (0–0.294)	0.286 (0–0.571)	0.111 (0.028–0.194)	0.774

Sensitivity analysis #6 (default value for outlier threshold)
≥1 secondary transmission
Early phase	0.64 (0.48–0.76)	0.684 (0.526–0.789)	0.442 (0.349–0.512)	0.682
Late phase	0.471 (0.294–0.647)	0.429 (0.286–0.714)	0.417 (0.306–0.528)	0.522
≥2 secondary transmissions
Early phase	0.32 (0.2–0.48)	0.316 (0.158–0.474)	0.186 (0.116–0.279)	0.424
Late phase	0.118 (0–0.294)	0.286 (0–0.571)	0.139 (0.056–0.222)	0.790

Open in a new tab

Appendix 1—figure 1. — Each row corresponds to a patient, and the solid lines indicate hospitalisation dates. The lines are coloured according to which ward a patient was in on a particular day. Outbreak wards (A–D) are coloured differently from non-outbreak wards (Q–Z).

Appendix 1—figure 2. — The matrix indicates the sum of transmission events across all posterior trees from cases in ‘infector’ wards (vertical axis) to cases in ‘infectee’ wards (horizontal axis). The degree of shading is proportional to the estimated posterior number of transmissions for each ward-to-ward pair. Outbreak wards: A–D; non-outbreak wards: P–Z (Z is ‘all wards’).

Appendix 1—figure 3. — The blue histograms indicate the expected random distributions of f_ward, given the proportion of HCWs amongst cases. The red histograms show the observed distribution of f_ward, across 1000 transmission trees reconstructed by `outbreaker2`. (A). All wards. (B) Transmission to outbreak wards only. (C) Transmission to non-outbreak wards only.

Infection prevention and control measures during the first pandemic wave

A multidisciplinary Covid-19 response unit was formed in the Geriatric hospital, with daily meetings and involvement of the IPC team. Direct coaching of HCWs working in Covid-19 wards by the IPC practitioner was undertaken once a ward had been attributed as such. The IPC practitioner assisted front-line HCWs in streamlining tasks.

HCWs caring for Covid-19 patients applied ‘contact’ and ‘droplet’ precautions, in line with Swissnoso and FOPH guidelines. Universal masking of all front-line HCWs was implemented on 11 March 2020. From 1 April 2020, the use of ocular protection (eye shields) was encouraged for contact with all Covid-19 patients, and masking of HCWs in non-clinical areas (e.g., offices) was recommended.

RT-PCR screening of SARS-CoV-2 on admission for all patients, even if asymptomatic or without clinical suspicion of Covid-19, started on 1 April 2020. From 7 April 2020, weekly screening surveys were performed in non-Covid wards, until 30 May 2020.

As of 11 April 2020 non-Covid wards were closed to new admissions, and room occupancy was decreased (e.g., four-bed rooms were limited to three patients).

Due to shortages in PPE, HCWs were instructed to wear surgical masks for 4 hr continuously (and 8 hr if not humid), and N95 respirators (FFP2 masks) for as long as possible. Gowns were used for >1 patient in the same room, except in the case of patients carrying multidrug-resistant bacteria.

HCWs from outbreak wards were encouraged to undergo PCR testing on nasopharyngeal swabs, even if asymptomatic between 9 and 16 April 2020. Including tests that were performed since mid-March, a total of 83 out of 124 eligible HCWs (67%) in the four outbreak wards underwent testing.

We split the outbreak into two phases, up to 9 April 2020 and after 9 April 2020 for several reasons. First, many preventive measures were implemented around that date (e.g., weekly screening of patients on 7 April enhanced testing of HCWs on 9 April, decrease in bed-occupancy and closures on 11 April). Second, approximately half of the cases occurred in each phase of the outbreak. Third, the overall trend in the epidemic curve can be interpreted as increasing in the first phase, and decreasing in the second phase.

Ward architecture and room sizes in the geriatric hospital

The architectural layout of the hospital means that there are two wings per floor, each with a central corridor, and two wards per wing (without separation). On either side of the central corridor are patient rooms and/or offices (nursing, medical, etc.). Therefore, the wards themselves are crowded areas, and it is sufficient for a patient to step outside their room in the corridor to potentially come in contact with another patient.

The mean room sizes (for rooms that were shared) were small, being 23.1 m² (standard deviation 6.5 m²) and 27.8 m² (standard deviation 0.2 m²) for two bed and three bed rooms, respectively.

Microbiological methods

Amplicon-based high-throughput sequencing analysis

All nasopharyngeal swabs (NPS) tested positive for SARS-CoV-2 by RT-PCR (Cobas 6800 SARS-CoV2 RT-PCR), the Charite or the BD SARS-CoV2 reagent kit for BD Max system assays, and for which sufficient volume remained, were selected for whole-genome sequencing analysis.

NPS were sequenced with an amplicon-based sequencing method. Thus, nucleic acids were extracted using the NucliSENS easyMAG (bioMérieux, Geneva, Switzerland) and then sequenced using an updated version of the nCoV-2019 sequencing protocol (https://www.protocols.io/view/ncov-2019-sequencing-protocol-bbmuik6w) (Microsynth, Balgach, Switzerland) on a MiSeq instrument (Illumina) with a 2×250 bp protocol.

Duplicate reads were removed using cd-hit (v4.6.8). Low-quality and adapter sequences were trimmed out using Trimmomatic (v0.33). Reads were then mapped against the reference sequence MN908947 using snap-aligner (v1.0beta.18). Consensus for sequences with at least 10-fold coverage were then generated using custom script.

Phylogenetic analysis

Sequence alignment was performed with MUSCLE (v3.8.31). Evolutionary analyses were conducted in MEGA X (Kumar et al., 2018) using the Maximum Likelihood method and Tamura three-parameter model (Tamura, 1992). The tree includes also all SARS-CoV-2 complete genomes sequenced by our laboratory and submitted to GISAID from respiratory samples from COVID-19 positive patients presenting to our institution or other medical centres in Geneva, Switzerland, during the same period.

Statistical analyses

Implementation of the `outbreaker2` models

We combined epidemiologic and genetic data using the outbreaker2 package in the R software, which has been used successfully in the reconstruction of the 2003 SARS-CoV-1 outbreak in Singapore (Jombart et al., 2014; Campbell et al., 2018) and a nosocomial outbreak of SARS-CoV-2 in a rehabilitation clinic (Abbas et al., 2021a). The model uses a Bayesian framework, which combines information on the generation time (time between infections in an infector/infectee pair), with a model of sequence evolution to probabilistically reconstruct the transmission tree.

As dates of onset of symptoms are known, and because dates of infection (i.e., acquisition) are not known with certainty, we imputed serial intervals based on estimates from the work by Ali et al., 2020, who showed that the serial interval decreased from the early stages of the pandemic due to improved control using non-pharmaceutical measures. For the primary analysis we used a short serial interval (mean 3.0, standard deviation [SD] 4.1), under the assumption of swift isolation of patients following onset of symptoms. We also performed a sensitivity analysis using a longer serial interval (mean 5.2 days, SD 4.7) to allow for potentially slower isolation of symptomatic patients. We used the incubation period as estimated by Bi et al., 2020, which follows a lognormal distribution with parameters mu of 1.57 and sigma 0.65 (corresponding to a mean of 5.95 days and SD 4.31). Where dates of onset were unavailable, we imputed them by using the median of the difference between symptom onset and date of swab.

Imported cases are detected by outbreaker2 during a preliminary run of the model, where cases are removed in turn (with a leave one out approach) and the ‘global influence’ of each case is measures as the extent to which removing it affects the genetic log-likelihood. Cases with high global influence (whose removal dramatically affects the genetic likelihood) are considered to be genetic outliers, and classified as imported cases. This approach is described in more detail in the original outbreaker2 manuscript (Jombart et al., 2014). By default, a case is determined as being imported if its global influence is five times the average global influence. While this approach has excellent specificity, because of the limited genetic diversity of SARS-CoV-2, it may lack in sensitivity. We therefore ran the model over a range of lower thresholds (from 2 to 5 times the average global influence). We selected for our main model a threshold of 2, and present one-way sensitivity analyses with a threshold of 3 and the default (9).

The outbreaker2 package is designed to use contact tracing data to inform who infected whom. These data were unavailable for our outbreak, but we made a series of assumptions to generate a matrix of possible contacts between cases. Initially, we aimed to construct this matrix using dates of presence in the hospital/ward for patients based on administrative data, and based on human resources shift rota for HCWs. However, we identified potential inconsistencies in the latter, in particular stemming from multiple changes as a result of many HCWs self-isolating due to possible or confirmed COVID-19. We therefore used these data in a sensitivity analysis, but for our main analysis, we reverted to a simpler set of assumptions to build our contact matrix. In our main analysis, we assumed that HCWs were present in the hospital every day until the date of their first positive swab, included. We further assumed that patients only interacted with patients in their own ward. We assumed that all HCWs were able to infect each other, but HCWs could have contact with patients only in the wards they were attributed to; HCWs such as physical therapists or doctors who worked across multiple wards could have significant contact with all HCWs and patients. Under these assumptions, we are likely to capture many contacts which did not happen, and it is possible that we miss a few contacts which in fact did happen. However outbreaker2 does account for imperfect sensitivity and specificity of contact data, the levels of which are estimated as part of the model. To account for this uncertainty in potential contact patterns, we ran two sensitivity analyses where we input different contact data in the model.

Sensitivity analyses

We conducted a number of different analyses, including a base scenario and several (n=5) one-way sensitivity analyses (Appendix 1—table 1):

Sensitivity analysis 1:
- We did not make any assumptions about contact patterns, and therefore all cases in the outbreak could potentially infect all other cases.
Sensitivity analysis 2:
- We used a longer serial interval (mean 5.2 days, SD 4.7) to allow for potentially slower isolation of symptomatic patients.
Sensitivity analysis 3:
- We used the HCW shift data from the human resources department, with minor corrections (removing HCWs that were mislabelled as ‘present’ after date of positive RT-PCR). For both patients and HCWs we categorised days of ‘susceptibility’ (fifth percentile of the cumulative incubation period from Bi et al., 2020) and days of ‘infectiousness’ (2 days before symptom onset based on the study by He et al., 2020). The last day of ‘infectiousness’ was the date of swab for HCWs.
Sensitivity analysis 4:
- We assumed that isolation precautions prescribed for patients on date of positive RT-PCR were effective, and that from that date patients were no longer infectious.
Sensitivity analysis 5:
- We used a higher threshold of 3 for the determination of imported cases.
Sensitivity analysis 6:
- We used the default threshold (5) used by outbreaker2 to detect imported cases.

For each model, we used a uniform prior between 0.55 and 1 for the reporting probability (pi). Indeed, we had a comprehensive screening and testing strategy, including of asymptomatic cases, and are therefore confident that we captured a near-total proportion of cases. We obtained sequences for 82% of all identified cases, and these are the cases used in the model. The lower bound of the prior for ‘pi’ thus allows us to have missed six cases in addition to the 14 that were not sequenced. Posterior estimates for ‘pi’ were visually compared to our prior choice to assess the validity of this assumption.

We allowed for a maximum of two unobserved cases on a transmission chain between any two observed cases (maximum ‘kappa’ of 3 in outbreaker2). This allows for identification of missed cases.

We used the default priors for the mutation rate (mu) for all models (uninformative exponential prior with mean 1), and, where relevant, those for non-infectious contact rate (lambda) and contact reporting coverage (eps), which were uniform on [0, 1] (Campbell et al., 2018; Campbell et al., 2019). We used the default likelihoods for all models, except for the model without contact data where this was disabled (Campbell et al., 2018; Campbell et al., 2019).

We ran each outbreaker2 model over 2,000,000 iterations of the MCMC (500,000 for model without contact data), with a thinning of 1 in 2000 (1 in 500 for model without contact data), in order to obtain a sample of 1000 posterior parameter sets, after a burn-in of 2000 iterations (500 for model with contact data). Each of these parameter sets corresponds to a posterior transmission tree. Convergence was assessed visually as well as through the Gelman-Rubin convergence diagnostic (using the gelman.diag function in the R package coda v0.19-4) (Gelman and Rubin, 1992), concluding that the chains converged appropriately if the upper limit of the confidence interval was <1.1.

To assess the role of each type of case in transmission by estimating the proportion of infections attributed to the type of case (f_case), we compared the posterior distribution of direct infections caused by each infector type to that obtained by a matching number of random draws of the infector types (expected proportion of infections), drawn according to the prevalence of each type among cases. We concluded that f_case was significantly higher than expected by chance if at least 95% of the posterior samples had higher proportions of cases infected by an infector type than that obtained by the random draws. The corresponding p-values were calculated as 1 minus the proportion of posterior samples with values higher than random draws. This was done for all infectees (i.e., whole outbreak), as well as for each type of infectee.

We evaluated the differences in the distribution of secondary cases across case types in the early (up to 9 April 2020) and later phases (as of 10 April 2020) of the study period using a chi-squared test (for proportions of cases with no secondary transmissions [non-transmitters] and of cases with ≥2 secondary transmissions [high transmitters]).

Funding Statement

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Contributor Information

Mohamed Abbas, Email: mohamed.abbas@hcuge.ch.

Joshua T Schiffer, Fred Hutchinson Cancer Research Center, United States.

Wendy S Garrett, Harvard T.H. Chan School of Public Health, United States.

Funding Information

This paper was supported by the following grants:

Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung 4078P0_198363 to Mohamed Abbas, Samuel Cordey, Florian Laubscher, Tomás Robalo Nunes, Anne Iten, Stephan Harbarth, Anne Cori.
National Institute for Health Research Health Protection Research Unit NIHR200908 to Mohamed Abbas, Anne Cori.

Additional information

Competing interests

No competing interests declared.

received honoraria (which was paid to the institution) from Pfizer for lecturing on a course on mathematical modelling of infectious disease transmission and vaccination book. The author has no other competing interests to declare.

No competing interests declared.

Author contributions

Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Visualization, Writing – original draft, Writing – review and editing.

Formal analysis, Investigation, Methodology, Writing – original draft, Writing – review and editing.

Funding acquisition, Investigation, Methodology, Writing – review and editing.

Formal analysis, Investigation, Methodology, Writing – review and editing.

Investigation, Methodology, Writing – review and editing.

Investigation, Writing – review and editing.

Investigation, Methodology, Writing – review and editing.

Investigation, Writing – review and editing.

Investigation, Methodology, Writing – review and editing.

Conceptualization, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Supervision, Writing – original draft, Writing – review and editing.

Ethics

Human subjects: The Ethics Committee of the Canton of Geneva (CCER), Switzerland, approved this study (CCER no. 2020-01330 and CCER no. 2020-00827). Written informed consent was obtained from HCWs. Written informed consent was not required for patients as data were generated in a context of mandatory surveillance.

Additional files

Transparent reporting form

elife-76854-transrepform1.docx^{(111.5KB, docx)}

Data availability

Due to small size of the various clusters, the raw clinical data will not be shared to safeguard anonymity of patients and healthcare workers. Processed data of the output of the model, which will comprise the posterior distribution of infectors, will be made available in an anonymized version. This will allow reproduction of the analyses looking at the proportion of healthcare workers among infectors, and the number of secondary infections. This data will not allow reconstruction of the transmission tree, which would require the raw data. The raw data in an anonymized format will be made available upon reasonable and justified request, subject to approval by the project's Senior Investigator. The genomic sequencing data have been submitted to the Genbank repository (GenBank accession numbers: ON209723-ON209871).

References

Abbas M, Robalo Nunes T, Cori A, Cordey S, Laubscher F, Baggio S, Jombart T, Iten A, Vieux L, Teixeira D, Perez M, Pittet D, Frangos E, Graf CE, Zingg W, Harbarth S. Explosive nosocomial outbreak of SARS-CoV-2 in a rehabilitation clinic: the limits of genomics for outbreak reconstruction. The Journal of Hospital Infection. 2021a;117:124–134. doi: 10.1016/j.jhin.2021.07.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
Abbas M, Robalo Nunes T, Martischang R, Zingg W, Iten A, Pittet D, Harbarth S. Nosocomial transmission and outbreaks of coronavirus disease 2019: the need to protect both patients and healthcare workers. Antimicrobial Resistance and Infection Control. 2021b;10:7. doi: 10.1186/s13756-020-00875-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
Abbas M, Zhu NJ, Mookerjee S, Bolt F, Otter JA, Holmes AH, Price JR. Hospital-onset COVID-19 infection surveillance systems: a systematic review. The Journal of Hospital Infection. 2021c;115:44–50. doi: 10.1016/j.jhin.2021.05.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
Aggarwal D, Myers R, Hamilton WL, Bharucha T, Tumelty NM, Brown CS, Meader EJ, Connor T, Smith DL, Bradley DT, Robson S, Bashton M, Shallcross L, Zambon M, Goodfellow I, Chand M, O’Grady J, Török ME, Peacock SJ, Page AJ, COVID-19 Genomics UK (COG-UK) Consortium The role of viral genomics in understanding COVID-19 outbreaks in long-term care facilities. The Lancet. Microbe. 2022;3:e151–e158. doi: 10.1016/S2666-5247(21)00208-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
Ali ST, Wang L, Lau EHY, Xu X-K, Du Z, Wu Y, Leung GM, Cowling BJ. Serial interval of SARS-CoV-2 was shortened over time by nonpharmaceutical interventions. Science (New York, N.Y.) 2020;369:1106–1109. doi: 10.1126/science.abc9004. [DOI] [PMC free article] [PubMed] [Google Scholar]
Arons MM, Hatfield KM, Reddy SC, Kimball A, James A, Jacobs JR, Taylor J, Spicer K, Bardossy AC, Oakley LP, Tanwar S, Dyal JW, Harney J, Chisty Z, Bell JM, Methner M, Paul P, Carlson CM, McLaughlin HP, Thornburg N, Tong S, Tamin A, Tao Y, Uehara A, Harcourt J, Clark S, Brostrom-Smith C, Page LC, Kay M, Lewis J, Montgomery P, Stone ND, Clark TA, Honein MA, Duchin JS, Jernigan JA, Public Health–Seattle and King County and CDC COVID-19 Investigation Team Presymptomatic SARS-CoV-2 Infections and Transmission in a Skilled Nursing Facility. The New England Journal of Medicine. 2020;382:2081–2090. doi: 10.1056/NEJMoa2008457. [DOI] [PMC free article] [PubMed] [Google Scholar]
Bhattacharya A, Collin SM, Stimson J, Thelwall S, Nsonwu O, Gerver S, Robotham J, Wilcox M, Hopkins S, Hope R. Healthcare-associated COVID-19 in England: A national data linkage study. The Journal of Infection. 2021;83:565–572. doi: 10.1016/j.jinf.2021.08.039. [DOI] [PMC free article] [PubMed] [Google Scholar]
Bi Q, Wu Y, Mei S, Ye C, Zou X, Zhang Z, Liu X, Wei L, Truelove SA, Zhang T, Gao W, Cheng C, Tang X, Wu X, Wu Y, Sun B, Huang S, Sun Y, Zhang J, Ma T, Lessler J, Feng T. Epidemiology and transmission of COVID-19 in 391 cases and 1286 of their close contacts in Shenzhen, China: a retrospective cohort study. The Lancet. Infectious Diseases. 2020;20:911–919. doi: 10.1016/S1473-3099(20)30287-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
Braun KM, Moreno GK, Buys A, Somsen ED, Bobholz M, Accola MA, Anderson L, Rehrauer WM, Baker DA, Safdar N, Lepak AJ, O’Connor DH, Friedrich TC. Viral Sequencing to Investigate Sources of SARS-CoV-2 Infection in US Healthcare Personnel. Clinical Infectious Diseases. 2021;73:e1329–e1336. doi: 10.1093/cid/ciab281. [DOI] [PMC free article] [PubMed] [Google Scholar]
Burugorri-Pierre C, Lafuente-Lafuente C, Oasi C, Lecorche E, Pariel S, Donadio C, Belmin J. Investigation of an Outbreak of COVID-19 in a French Nursing Home With Most Residents Vaccinated. JAMA Network Open. 2021;4:e2125294. doi: 10.1001/jamanetworkopen.2021.25294. [DOI] [PMC free article] [PubMed] [Google Scholar]
Campbell F, Didelot X, Fitzjohn R, Ferguson N, Cori A, Jombart T. outbreaker2: a modular platform for outbreak reconstruction. BMC Bioinformatics. 2018;19:363. doi: 10.1186/s12859-018-2330-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
Campbell F, Cori A, Ferguson N, Jombart T. Bayesian inference of transmission chains using timing of symptoms, pathogen genomes and contact data. PLOS Computational Biology. 2019;15:e1006930. doi: 10.1371/journal.pcbi.1006930. [DOI] [PMC free article] [PubMed] [Google Scholar]
Ellingford JM, George R, McDermott JH, Ahmad S, Edgerley JJ, Gokhale D, Newman WG, Ball S, Machin N, Black GC. Genomic and healthcare dynamics of nosocomial SARS-CoV-2 transmission. eLife. 2021;10:e65453. doi: 10.7554/eLife.65453. [DOI] [PMC free article] [PubMed] [Google Scholar]
Evans S, Agnew E, Vynnycky E, Stimson J, Bhattacharya A, Rooney C, Warne B, Robotham J. The impact of testing and infection prevention and control strategies on within-hospital transmission dynamics of COVID-19 in English hospitals. Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences. 2021;376:1829. doi: 10.1098/rstb.2020.0268. [DOI] [PMC free article] [PubMed] [Google Scholar]
Ferretti L, Wymant C, Kendall M, Zhao L, Nurtay A, Abeler-Dörner L, Parker M, Bonsall D, Fraser C. Quantifying SARS-CoV-2 transmission suggests epidemic control with digital contact tracing. Science (New York, N.Y.) 2020;368:eabb6936. doi: 10.1126/science.abb6936. [DOI] [PMC free article] [PubMed] [Google Scholar]
Gelman A, Rubin DB. Inference from Iterative Simulation Using Multiple Sequences. Statistical Science. 1992;7:457–472. doi: 10.1214/ss/1177011136. [DOI] [Google Scholar]
He X, Lau EHY, Wu P, Deng X, Wang J, Hao X, Lau YC, Wong JY, Guan Y, Tan X, Mo X, Chen Y, Liao B, Chen W, Hu F, Zhang Q, Zhong M, Wu Y, Zhao L, Zhang F, Cowling BJ, Li F, Leung GM. Temporal dynamics in viral shedding and transmissibility of COVID-19. Nature Medicine. 2020;26:672–675. doi: 10.1038/s41591-020-0869-5. [DOI] [PubMed] [Google Scholar]
Jombart T, Cori A, Didelot X, Cauchemez S, Fraser C, Ferguson N. Bayesian reconstruction of disease outbreaks by combining epidemiologic and genomic data. PLOS Computational Biology. 2014;10:e1003457. doi: 10.1371/journal.pcbi.1003457. [DOI] [PMC free article] [PubMed] [Google Scholar]
Klompas M, Baker MA, Rhee C, Tucker R, Fiumara K, Griesbach D, Bennett-Rizzo C, Salmasian H, Wang R, Wheeler N, Gallagher GR, Lang AS, Fink T, Baez S, Smole S, Madoff L, Goralnick E, Resnick A, Pearson M, Britton K, Sinclair J, Morris CA. A SARS-CoV-2 Cluster in an Acute Care Hospital. Annals of Internal Medicine. 2021;174:794–802. doi: 10.7326/M20-7567. [DOI] [PMC free article] [PubMed] [Google Scholar]
Knight GM, Pham TM, Stimson J, Funk S, Jafari Y, Pople D, Evans S, Yin M, Brown CS, Bhattacharya A, Hope R, Semple MG, Read JM, Cooper BS, Robotham JV, ISARIC4C Investigators. CMMID COVID-19 Working Group The contribution of hospital-acquired infections to the COVID-19 epidemic in England in the first half of 2020. BMC Infectious Diseases. 2022;22:2021. doi: 10.1186/s12879-022-07490-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
Kumar S, Stecher G, Li M, Knyaz C, Tamura K. MEGA X: Molecular Evolutionary Genetics Analysis across Computing Platforms. Molecular Biology and Evolution. 2018;35:1547–1549. doi: 10.1093/molbev/msy096. [DOI] [PMC free article] [PubMed] [Google Scholar]
Lucey M, Macori G, Mullane N, Sutton-Fitzpatrick U, Gonzalez G, Coughlan S, Purcell A, Fenelon L, Fanning S, Schaffer K. Whole-genome Sequencing to Track Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) Transmission in Nosocomial Outbreaks. Clinical Infectious Diseases. 2021;72:e727–e735. doi: 10.1093/cid/ciaa1433. [DOI] [PMC free article] [PubMed] [Google Scholar]
Martischang R, Iten A, Arm I, Abbas M, Meyer B, Yerly S, Eckerle I, Pralong J, Sauser J, Suard J-C, Kaiser L, Pittet D, Harbarth S. Severe acute respiratory coronavirus virus 2 (SARS-CoV-2) seroconversion and occupational exposure of employees at A Swiss university hospital: A large longitudinal cohort study. Infection Control and Hospital Epidemiology. 2022;43:326–333. doi: 10.1017/ice.2021.117. [DOI] [PMC free article] [PubMed] [Google Scholar]
Mendes A, Serratrice C, Herrmann FR, Genton L, Périvier S, Scheffler M, Fassier T, Huber P, Jacques M-C, Prendki V, Roux X, Di Silvestro K, Trombert V, Harbarth S, Gold G, Graf CE, Zekry D. Predictors of In-Hospital Mortality in Older Patients With COVID-19: The COVIDAge Study. Journal of the American Medical Directors Association. 2020;21:1546–1554. doi: 10.1016/j.jamda.2020.09.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
Meredith LW, Hamilton WL, Warne B, Houldcroft CJ, Hosmillo M, Jahun AS, Curran MD, Parmar S, Caller LG, Caddy SL, Khokhar FA, Yakovleva A, Hall G, Feltwell T, Forrest S, Sridhar S, Weekes MP, Baker S, Brown N, Moore E, Popay A, Roddick I, Reacher M, Gouliouris T, Peacock SJ, Dougan G, Török ME, Goodfellow I. Rapid implementation of SARS-CoV-2 sequencing to investigate cases of health-care associated COVID-19: a prospective genomic surveillance study. The Lancet. Infectious Diseases. 2020;20:1263–1272. doi: 10.1016/S1473-3099(20)30562-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
Mo Y, Eyre DW, Lumley SF, Walker TM, Shaw RH, O’Donnell D, Butcher L, Jeffery K, Donnelly CA, Cooper BS, Oxford COVID infection review team Transmission of community- and hospital-acquired SARS-CoV-2 in hospital settings in the UK: A cohort study. PLOS Medicine. 2021;18:10. doi: 10.1371/journal.pmed.1003816. [DOI] [PMC free article] [PubMed] [Google Scholar]
Myall A, Price JR, Peach RL, Abbas M, Mookerjee S, Zhu N, Ahmad I, Ming D, Ramzan F, Teixeira D, Graf C, Weiße AY, Harbarth S, Holmes A, Barahona M. Predicting Hospital-Onset COVID-19 Infections Using Dynamic Networks of Patient Contacts: An Observational Study. medRxiv. 2021 doi: 10.1101/2021.09.28.21264240. [DOI] [PMC free article] [PubMed]
Ottolenghi J, McLaren RA, Bahamon C, Dalloul M, McCalla S, Minkoff H. Health Care Workers’ Attitudes Toward Patients With COVID-19. Open Forum Infectious Diseases. 2021;8:ofab375. doi: 10.1093/ofid/ofab375. [DOI] [PMC free article] [PubMed] [Google Scholar]
Shitrit P, Zuckerman NS, Mor O, Gottesman BS, Chowers M. Nosocomial outbreak caused by the SARS-CoV-2 Delta variant in a highly vaccinated population, Israel, July 2021. Euro Surveillance. 2021;26:2100822. doi: 10.2807/1560-7917.ES.2021.26.39.2100822. [DOI] [PMC free article] [PubMed] [Google Scholar]
Stirrup O, Hughes J, Parker M, Partridge DG, Shepherd JG, Blackstone J, Coll F, Keeley A, Lindsey BB, Marek A, Peters C, Singer JB, COVID-19 Genomics UK (COG-UK) consortium. Tamuri A, de Silva TI, Thomson EC, Breuer J. Rapid feedback on hospital onset SARS-CoV-2 infections combining epidemiological and sequencing data. eLife. 2021;10:e65828. doi: 10.7554/eLife.65828. [DOI] [PMC free article] [PubMed] [Google Scholar]
Stirrup O, Blackstone J, Mapp F, MacNeil A, Panca M, Holmes A, Machin N, Shin GY, Mahungu T, Saeed K, Saluja T, Taha Y, Mahida N, Pope C, Chawla A, Cutino-Moguel MT, Tamuri A, Williams R, Darby A, Robertson D, Flaviani F, Nastouli E, Robson S, Smith D, Loose M, Laing K, Monahan I, Kele B, Haldenby S, George R, Bashton M, Witney A, Byott M, Coll F, Chapman M, Peacock S, Hughes J, Nebbia G, Partridge DG, Parker M, Price J, Peters C, Roy S, Snell LB, de Silva TI, Thomson E, Flowers P, Copas A, Breuer J. Evaluating the Effectiveness of Rapid SARS-CoV-2 Genome Sequencing in Supporting Infection Control Teams: The COG-UK Hospital-Onset COVID-19 Infection Study. medRxiv. 2022 doi: 10.1101/2022.02.10.22270799. [DOI] [PMC free article] [PubMed]
Stone SP, Cooper BS, Kibbler CC, Cookson BD, Roberts JA, Medley GF, Duckworth G, Lai R, Ebrahim S, Brown EM, Wiffen PJ, Davey PG. The ORION statement: guidelines for transparent reporting of outbreak reports and intervention studies of nosocomial infection. The Lancet. Infectious Diseases. 2007;7:282–288. doi: 10.1016/S1473-3099(07)70082-8. [DOI] [PubMed] [Google Scholar]
Swissnoso Prevention & control of healthcare-associated COVID-19 outbreaks. Covid-19. 2021 https://www.swissnoso.ch/fileadmin/swissnoso/Dokumente/5_Forschung_und_Entwicklung/6_Aktuelle_Erreignisse/200515_Prevention_and_control_of_healthcare-associated_COVID-19_outbreaks_V1.0_ENG.pdf
Tamura K. Estimation of the number of nucleotide substitutions when there are strong transition-transversion and G+C-content biases. Molecular Biology and Evolution. 1992;9:678–687. doi: 10.1093/oxfordjournals.molbev.a040752. [DOI] [PubMed] [Google Scholar]
Taylor J, Carter RJ, Lehnertz N, Kazazian L, Sullivan M, Wang X, Garfin J, Diekman S, Plumb M, Bennet ME, Hale T, Vallabhaneni S, Namugenyi S, Carpenter D, Turner-Harper D, Booth M, Coursey EJ, Martin K, McMahon M, Beaudoin A, Lifson A, Holzbauer S, Reddy SC, Jernigan JA, Lynfield R, Minnesota Long-Term Care COVID-19 Response Group Serial Testing for SARS-CoV-2 and Virus Whole Genome Sequencing Inform Infection Risk at Two Skilled Nursing Facilities with COVID-19 Outbreaks - Minnesota, April-June 2020. MMWR. Morbidity and Mortality Weekly Report. 2020;69:1288–1295. doi: 10.15585/mmwr.mm6937a3. [DOI] [PMC free article] [PubMed] [Google Scholar]
The Guardian 40,600 people likely caught Covid while hospital inpatients in England. Covid-19. 2021a https://www.theguardian.com/society/2021/mar/26/40600-people-likely-caught-covid-while-hospital-inpatients-in-england
The Guardian Up to 20% of hospital patients with Covid-19 caught it at hospital. Covid-19. 2021b https://www.theguardian.com/world/2020/may/17/hospital-patients-england-coronavirus-covid-19
The Lancet Infectious Diseases COVID-19 vaccine equity and booster doses. Lancet (London, England) 2021;21:1193. doi: 10.1016/S1473-3099(21)00486-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
Thiabaud A, Iten A, Balmelli C, Senn L, Troillet N, Widmer A, Flury D, Schreiber PW, Vázquez M, Damonti L, Buettcher M, Vuichard-Gysin D, Kuhm C, Cusini A, Riedel T, Nussbaumer-Ochsner Y, Gaudenz R, Heininger U, Berger C, Zucol F, Bernhard-Stirnemann S, Corti N, Zimmermann P, Uka A, Niederer-Loher A, Gardiol C, Roelens M, Keiser O. Cohort profile: SARS-CoV-2/COVID-19 hospitalised patients in Switzerland. Swiss Medical Weekly. 2021;151:w20475. doi: 10.4414/smw.2021.20475. [DOI] [PubMed] [Google Scholar]
von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP, STROBE Initiative Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. BMJ (Clinical Research Ed.) 2007;335:806–808. doi: 10.1136/bmj.39335.541782.AD. [DOI] [PMC free article] [PubMed] [Google Scholar]

eLife. doi: 10.7554/eLife.76854.sa0

Editor's evaluation

Joshua T Schiffer ¹

Congratulations on this useful and technically impressive paper demonstrating that phylogenetic and epidemiologic data can be used in a retrospective cohort to reconstruct that chain of events in terms of case importation into a high risk geriatrics ward. The conclusion that HCW (healthcare worker) transmission in non-COVID wards was particularly important is critical for hospital epidemiologists. The methodology advances will hopefully push the field forward in terms of tracking outbreaks in various settings.

eLife. doi: 10.7554/eLife.76854.sa1

Decision letter

Editor: Joshua T Schiffer¹

Reviewed by: Lulla Opatowski², Thi Pham

Our editorial process produces two outputs: (i) public reviews designed to be posted alongside the preprint for the benefit of readers; (ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "Reconstruction of transmission chains of SARS-CoV-2 amidst multiple outbreaks in a geriatric acute-care hospital: a combined retrospective epidemiological and genomic study" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Wendy Garrett as the Senior Editor. The following individuals involved in the review of your submission have agreed to reveal their identity: Lulla Opatowski (Reviewer #1); Thi Pham (Reviewer #3).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

1) Clarify and improve Figure 3.

2) If possible, add any available details about transmissions between ward-sharing individuals in terms of distance between rooms and beds.

3) Add discussion about which results may or may not be generalizable to other health care settings, as well as the issue of misclassifying the infection source.

Reviewer #1 (Recommendations for the authors):

– Abstract line 63, sentence starting with "the proportion of infectors…" should be clarified.

– Because the analysed data provides from the first pandemic wave, several variables may have changed substantially over the studied period. In particular, as described in the methods section, sampling policy varied over time. Sampling was generally achieved differently in HCWs and patients. Were these two aspects accounted for in the statistical model? How does that affect uncertainty on estimates?

– What is the level of certainty for provided dates of symptom onset? Were they extracted from medical records or collected in real-time? To what extent does the reconstruction of the transmission chain rely on this variable and was uncertainty considered on that variable?

– Do all patients share rooms, and with how many others? What is this 44% as a proportion of those who do?

– There is a long-tailed distribution of infection by each individual (Figure 4). How much do heavy spreaders influence the results e.g. would the results from Figure 5 all hold if individual superspreader events were removed from the dataset?

– When calculating expectation in Figure 5, this is calculated by the prevalence of Covid-19 in each type of individual on a given ward. In our understanding, the ward was not considered here. Could the authors justify that choice? Would it be feasible to calculate an expectation which took into account the higher risk of transmission when sharing a room, and use this as the comparator against observed infections?

– Page 11 line 231, the referenced figure doesn't indicate these probabilities. What is the source?

– How was the date of April 9 fixed to define a cut-off between the early phase and late phase? Is it just based on the peak incidence? Or due to a change in screening practice?

– The large outbreak size makes figure 3 quite hard to interpret for the reader. Would it be possible to improve it possibly by zooming on some specific cases or wards, and adding global statistics informing on the precision/quality of the inference of transmission routes?

– How does the number of secondary infections relate to the population's sizes (i.e. at-risk patients and HCWs from the studied wards). Is that relationship expected?

– We may have missed this: (apologies if so): were HCWs cohorted by wards or were there HCWs shared among wards? How may that affect the results here?

Reviewer #2 (Recommendations for the authors):

This paper addresses an important question about the dynamics of SARS-CoV-2 in a geriatric acute-care hospital. This reviewer's comments address primarily manuscript clarity.

General Comments:

Patients should not be equated with their conditions. Please replace Covid-19 and non-Covid 19 patients with patients who have Covid-19 and patients who do not have Covid-19 throughout the manuscript.

Compare takes "with" not "to".

The manuscript includes many "there is," "there are," "it is," xxx. These statements are in passive voice and are usually wordier and more convoluted than a similar statement in active voice. This reviewer suggests revising most or all of these statements. For example, the following sentence could be revised as shown.

"The reservoir of SARS-CoV-2 in healthcare environments may contribute to amplifying the pandemic [5], and as such, it is important to understand transmission dynamics in these settings."

"Because the reservoir of SARS-CoV-2 in healthcare environments may contribute to amplifying the pandemic [5], we need to better understand transmission dynamics in these settings."

Statements like "in order to" and "as well as" have extra words that don't add to the meaning. They can be shortened to "to" and "and".

Specific Comments

Line 67: Is the word "unexpectedly" needed? This reviewer was confused by that statement. I would assume HCWs knowing that they were caring for patients with Covid-19 might be more in adherence with PPE use and hand hygiene than those on non-Covid-19 units.

Lines 93-95: The meaning of "in nosocomial Covid-19" is vague and should be clarified. Does this mean HCWs' role in transmitting SARS-CoV-2 within healthcare facilities is complex?

Does it mean HCWs role in the epidemiology of nosocomial SARS-CoV-2 infection is

complex? This reviewer does not like the word "victim" (although the alteration is nice) in this setting. This reviewer thinks the last part of this sentence will work better if revised to: "as HCWs can acquire SARS-CoV-2 in the community or from their peers and patients and they can transmit SARS-CoV-2 to peers and patients [7, 8].

Methods

The authors switch frequently between active and passive voice. This reviewer suggests picking one and sticking with it. This reviewer prefers an active voice for clarity and brevity.

Line 162: This reviewer does not understand what the authors mean by "trajectories" in this sentence.

Lines 165-6: Revise to something like: "conservative in that it accounts for non-infectious contacts and incomplete reporting of infectious contacts and it estimates the proportion of the total contacts in these categories."

Lines 18-91: Suggested revision. A post hoc epidemiological analysis of the 53 patients found that 4 of these patients likely acquired Covid-19 in the community (CA-Covid-19). The 49 nosocomial cases represented 20.2% (49/242) of all patients with COVID-19, and 81.7% (49/60) of nosocomial Covid-19 cases identified in the Geriatric Hospital. The ward-level attack rates ranged from 10 to 19% among patients and 21% of all HCWs in the geriatric..."

Line 203: This reviewer does not understand this sentence given that the verb appears to be "shows".

Lines: 273-5: Suggest revising to something like: "Fourth, we identified multiple importation events that led to a substantial number of secondary cases or clusters. Most transmission events were related to HCWs, but one (?? or some) were related to a patient with community-acquired Covid-19." Note: the authors did not make clear how many transmissions were related to this patient.

Lines 282-292: Suggest revising to something like: "HCWs' behavior may have affected transmission. Indeed, Ottolenghi et al., found that HCWs caring for Covid-19 patients were concerned about becoming infected while caring for patients, and therefore may apply IPC measures more rigorously than when caring for patients who do not have Covid-19 [23]. HCWs working in non-Covid wards may not have felt threatened by Covid-19 patients on other wards. HCWs in non-Covid wards also may have underestimated the transmission risk from their peers to a greater extent than those working in Covid wards, and thus did not maintain physical distancing well. In addition, we found a higher mean duration (2.9 days) of presenteeism despite symptoms compatible with Covid-19 among HCWs working in non-Covid wards than for those working in Covid wards (1.6 days), which gives credence to the abovementioned possible explanations for the different transmission patters. Other factors (e.g., work culture, baseline IPC practices) also may have affected transmission patterns.

This reviewer wonders if the following sentence is necessary? "HCWs working in non-Covid wards may not have felt threatened by Covid-19 patients on other wards."

Line 294: This author is wondering how the same ward is close proximity when patients are not in the same room. How would the patients have contact with each other?

Lines 294-309: Suggested revisions: We cannot exclude transmission from a "point-source" or via an HCW's contaminated hands (i.e., an unidentified HCW who transmitted SARS-CoV-2 to multiple patients in the same ward). To date, we have little evidence to suggest that this is the case; indeed, the transmission patterns were robust to changes in the model assumptions. Mathematical models have suggested that single-room isolation of suspected cases could potentially reduce the incidence of nosocomial SARS-CoV-2 transmission by up to 35% [24]. In the outbreaks we describe, symptomatic patients were identified promptly, with a median delay of 0 days between symptom onset and their first positive test. However, these precautions may not be sufficient as patients may transmit the virus when they are pre-symptomatic [25]. Thus, infection prevention teams may need to identify patients at high risk of developing nosocomial Covid-19 [26] if single rooms are not available for all exposed patients (e.g., in cases of overcrowding). For example, Mo et al., found that exposure to community-acquired cases who were identified and segregated or cohorted was associated with half the risk of infection compared with exposure to hospital-acquired cases or HCWs who may be asymptomatic [27]. One possible explanation for this finding is that patients with CA-Covid may have passed the peak of infectiousness when they are admitted, whereas patients with HA-Covid cases have frequent unprotected contact with HCWs and other patients during their period of peak infectiousness.

Line 310: Suggested edit: The current evidence does not support the use of real-time genomics for control of SARS-CoV-2 nosocomial outbreaks [28].

Lines 324-328: Suggested edits: Despite these strengths, our study had some limitations. First, we included only one sequence from a CA-Covid. However, the method we used to reconstruct who infected who is able to cope with and identify missing intermediate cases, which allowed us to estimate that our sample included the overwhelming majority of cases (91.5%). In addition, we performed these investigations during the first pandemic wave in a susceptible population...."

Line 334: Replace "in the case of" with "during".

Lines 336-338: The main point of the conclusion is not clear. What does it mean to take into account the complex interplay between HCWs in dedicated Covid-19 wards vs non-Covid wards? The way this is worded makes it sound like the HCW on the two different kinds of wards are interacting, which likely is not what the authors intended.

Table 1: This reviewer doesn't understand what is meant by "Onset of symptoms before swab date, n (%)". Is this the number of people who were symptomatic before they were tested?

Table 2: Should HCWs be written out in the title? Readers in the US will likely not know what a logopedist is. We will call this person a speech therapist. Perhaps the authors could include a footnote defining this term. Did the authors address the longer time from onset of symptoms to testing for HCWs on non-Covid wards compared with those on Covid wards in the results and discussion?

Table 3: Dose "Secondary onward transmission" refer to transmission from the case in the first column? Consider putting NA (not applicable) in the blank cells.

Line 515: Suggested edit: "Histograms displaying the distributions of secondary cases...."

Line 525: Is something missing between "... patient community and "proportion"?

Reviewer #3 (Recommendations for the authors):

Methods:

The authors present the IPC measures on page 6 in the main text and in the supplementary material. However, there seems to be no information on how HCWs adhered to the recommendations that were not mandatory and whether adherence differed between wards. This is probably because the authors did not have any information on that. This is a limitation and should be explained more explicitly or highlighted in the discussion.

It's not clear to me whether the HCWs only worked in either COVID or non-COVID wards during the study period. Please clarify in the Methods section.

On page 7, line 139: The authors state that they substituted the missing date of symptom onsets with the median difference between symptom onset date and swab date? The epidemic curve in Figure 1 shows the incidence on the y-axis but does the x-axis (date of symptom onset) make sense if not all of the cases were symptomatic?

Page 8, line 169 and Figure 4: How was the epidemic phase determined?

Results:

At the beginning of the Results section, the authors write that post hoc epidemiological analysis showed that were likely 4 patients that had a community-acquired infection. What exactly did the post-hoc epidemiological analysis entail? Later on (page 10, lines 206ff), they present their analysis of the imported cases. How are these related?

Discussion:

Could the authors discuss whether misclassifications (either of community- or hospital-acquired cases) even after the analysis is still possible and what impact it could have on their results?

Supplement

Page 5: Could the authors elaborate on "global influence" for imported cases?

eLife. 2022 Jul 19;11:e76854. doi: 10.7554/eLife.76854.sa2

Author response

Essential revisions:

1) Clarify and improve Figure 3.

We agree with Reviewer #1 that the initial Figure 3 was difficult to read. We have now created an interactive chart where one can hover over points and zoom in selected areas. We have also removed the colours, which did not have a particular meaning. Following a suggestion from Reviewer #1, we have decided to move this new interactive figure as Figure 1 – Supplementary figure 1, and have replaced it with a more informative figure demonstrating the extent to which our model is able to infer of transmission routes (i.e., showing the distribution of the posterior support of ancestries based on different criteria (individual, case type, ward), see new Figure 3).

Legend for New Figure 3: Distribution of posterior support of maximum posterior ancestry for all cases, according to identity of A individual ancestor, B ancestor’s case type (i.e., “HCW_covid”, “HCW_outbreak”, “patient_noso”, “patient_community”), C ancestor’s ward, and D ancestor’s ward type (i.e., “outbreak ward”, “non-outbreak ward”).

2) If possible, add any available details about transmissions between ward-sharing individuals in terms of distance between rooms and beds.

We have added information about room sizes in the Appendix (page 3): “Indeed, the room sizes (for rooms that were shared) were small, being 23.1 m² (standard deviation 6.5 m²) and 27.8 m² (standard deviation 0.2 m²) for 2-bed and 3-bed rooms, respectively.” Unfortunately, we do not have precise measurements on distance between beds, but in general, beds next to each other were separated by approximately 1.5 – 2 metres.

When responding to a related comment from from Reviewer #1, we noticed an error in the pre-processing stage of contact data (n = 839 contacts [4%] were included that should not have been). We therefore re-ran all our analyses after correcting this error. This has led to minor changes in our numeric results but has not altered our main findings and conclusions. We apologise for this inconvenience, and are grateful to have had the opportunity to double-check our work as a result of the high-quality in-depth reviews.

3) Add discussion about which results may or may not be generalizable to other health care settings, as well as the issue of misclassifying the infection source.

We have added discussion points on the lack of generalisability of our results: “In addition, we performed these investigations during the first pandemic wave with a wild-type variant of SARS-CoV-2 in a susceptible population. For these reasons, the results may no longer be applicable in settings with high vaccination coverage and/or substantial natural immunity, or in later stages of the pandemic with different variants, also due to accrued experience in managing and preventing outbreaks. Nevertheless, the lessons learned may be useful in a large number of countries with slow vaccine roll-out due to vaccine hesitancy, particularly among HCWs where there is no vaccine mandate, or unequal access to vaccine supplies [33]. Furthermore, nosocomial outbreaks of SARS-CoV-2 still occur despite high vaccination coverage [34, 35]. Also, these valuable lessons may be applicable for nosocomial outbreak control in the case of future pandemics due to respiratory viruses with characteristics similar to SARS-CoV-2.” Part in italics was already in the text.

Misclassifications may still occur, and it is even part of the method (outbreaker2) to quantify the uncertainty in determining the ancestry, as it operates within a Bayesian framework. Indeed, in most cases (c.f. new Figure 3) there are very few patients for whom the ancestry is identified with 100% posterior probability. For example, the model predicted that patients C107 and C115 were most likely imported cases with posterior probabilities of 100% and 57.5%, respectively. The predicted dates of infection were in the range March 14-22 for case C107 and March 8-24 for case C115, which was after their admission dates (March 3 and March 18, respectively). The probability that infection occurred on or after date of admission for case C115 was 81%. Therefore the model acknowledges the uncertainty in the classification of these cases as imported. We have modified the section on imported cases (under the heading “imported cases”), and added a sentence after referencing the new Figure 3 to highlight the uncertainty in the model.

Finally, the fact that a case has a probability of being “imported” does not preclude the fact that it is still nosocomial cases, simply that their infector was not identified in this outbreak c.f. update in the manuscript (lines 241-259).

Reviewer #1 (Recommendations for the authors):

– Abstract line 63, sentence starting with "the proportion of infectors…" should be clarified.

We agree that the sentence was unclear, and have modified it to “The proportion of infectors being HCW_covid was as expected as random”. We hope that it is now clearer.

– Because the analysed data provides from the first pandemic wave, several variables may have changed substantially over the studied period. In particular, as described in the methods section, sampling policy varied over time. Sampling was generally achieved differently in HCWs and patients. Were these two aspects accounted for in the statistical model? How does that affect uncertainty on estimates?

The reviewer raises an interesting and important point. Screening policies were indeed different between patients and HCWs. For patients, weekly screening surveys were performed in non-Covid wards from April 07, 2020 until May 30, 2020 (Appendix, page 3). In contrast, HCWs from outbreak wards were encouraged to undergo PCR testing, even if asymptomatic, between April 09 and April 16, 2020. These aspects were not explicitly accounted for in the statistical model. However, we are confident that the intensity of screening we performed during that period, led to a good case ascertainment. We note that the method we rely on, outbreaker2, does allow for undetected cases. It assumes a constant case reporting probability, which is estimated. In our study, this reporting probability estimated to be 91.5% (95% credible interval 91.2-91.9%), which confirms that case ascertainment was very high.

Outbreaker2 also estimates, for each observed case, the probability that one or more cases were unobserved in the transmission chain leading to their closest observed ancestry. We compared the posterior distribution of this number of “missed” generations between the early and late phases of the epidemic (i.e. before and after April 09, 2020). We have plotted the distribution of the number of missed generations across posterior trees, stratified by the phase. This suggests very little difference in the distribution of missed generations between the two phases. We have included this figure as Figure 3 – Supplementary Figure 2.

– What is the level of certainty for provided dates of symptom onset? Were they extracted from medical records or collected in real-time? To what extent does the reconstruction of the transmission chain rely on this variable and was uncertainty considered on that variable?

The dates of symptom onset were collected prospectively in real-time through several sources, minimising risk of bias. For patients, data were generated by the prospective surveillance of hospitalised patients with Covid-19 mandated by the Swiss Federal Office of Public Health, and for HCWs from both the hospital’s department of Occupational Health and from the prospective cohort study of outpatient Covid-19 testing in the hospital. We have clarified this in the “Data sources” section: “Dates of symptom onset were available for both patients and HCWs each source, respectively.”

The date of symptom onset is a main input of the model, as it is the starting point from which the date of infection is inferred. Although we did not directly account for potential uncertainty in the date of onset, our model considers whole distributions for the serial interval and incubation period which would allow absorbing a small level of uncertainty in onset dates, assuming such uncertainty is similar to those in studies which estimated the serial interval and incubation period.

– Do all patients share rooms, and with how many others? What is this 44% as a proportion of those who do?

Out of all 43 patients, there were 5 patients (9.4%) who did not share a room with another Covid patient; this number increases to 11 patients (20.7%) if a “significant” room sharing is considered (i.e. when at least one room occupant is during an infectious period, defined as the period between 3 days before onset of symptoms and 14 days after). Of the 319 days where a room was shared, 256 days (80.3%) involved a room being shared by 2 patients, and 63 (19.7%) by 3 patients. The proportion of patient-to-patient transmissions that involved patients sharing a room was calculated for each posterior transmission tree (by determining whether the ancestor/infector had ever shared a room with the infectee), with the denominator being all patient-to-patient transmissions.

When responding to this comment we noticed an error in the pre-processing stage of contact data (n = 839 contacts, 4%, that should not have been included). We re-ran all the analyses after correcting this error. This has led to minor changes in our numeric results does not alter our major findings and conclusions. We apologise for this inconvenience, and thank the referee for their in-depth review which has given us the opportunity to double-check our work.

– There is a long-tailed distribution of infection by each individual (Figure 4). How much do heavy spreaders influence the results e.g. would the results from Figure 5 all hold if individual superspreader events were removed from the dataset?

We thank the Reviewer for this interesting observation and question. Altogether, the proportion of super-spreading events (defined as generating ≥5 secondary cases) among all posterior trees is very low (0.4%). The majority of posterior trees (54.1%) have no infectors with ≥5 secondary cases. The proportion of posterior trees with 1, 2, and ≥3 infectors with ≥5 secondary cases (super-spreaders) are 37.3%, 7.9% and 0.7%, respectively. The overwhelming majority of posterior trees (88.6%) do not have infectors who are the plausible ancestors of ≥6 infectees. For these reasons, we anticipated the influence of the superspreading events across all posterior transmission trees to be negligible. To confirm this, we performed a sensitivity analysis, estimating the proportion of transmissions (f_case) attributed to each case type (HCW_covid, HCW_outbreak, patient_noso), i.e., the results from Figure 5, but excluding posterior transmission trees where there is at least one infector with ≥5 secondary cases. In Author response image 1, we show the original figure 5 and in Author response image 2, the corresponding results from our sensitivity analysis. Results are very similar. Should the editor find it useful, we would be happy to add this sensitivity analysis to the Appendix.

– When calculating expectation in Figure 5, this is calculated by the prevalence of Covid-19 in each type of individual on a given ward. In our understanding, the ward was not considered here. Could the authors justify that choice? Would it be feasible to calculate an expectation which took into account the higher risk of transmission when sharing a room, and use this as the comparator against observed infections?

The scenario chosen for the comparison is the most agnostic in that it does not make any assumptions regarding contact patterns. Instead, it assumes that all individuals are able to infect each other with the same probability, and, as a consequence, a group is able to infect an individual based on the proportion of individuals belonging to that group. However, this may be seen as an upper bound of the effect, as indeed it does not account for ward/room sharing in addition to the case type. Indeed, the analysis estimates excess transmission, but it does not specify what the excess is due to. For this reason, we performed the analysis of the ward type as well as the ward-to-ward matrix. Unfortunately, it is not feasible to account for both ward/room sharing and case type, as the combination of multiple characteristics would decrease the sample size and may potentially lead to type II error. As highlighted by the Reviewer, to calculate an expectation which takes into account the higher risk of transmission by sharing a room, we would need to estimate this, yet there is little data to inform this parameter.

We apologise but there was a small error in the Legend for Figure 5: instead of “f_ward”, the legend should read “f_case”. We have now corrected this.

– Page 11 line 231, the referenced figure doesn't indicate these probabilities. What is the source?

We thank the Reviewer for pointing this out. Indeed, the Figure was intended to show the ward movements of the patients and the dates where the patients overlapped. The probabilities of infection were derived from the output of the model (basically from Figure 3). We have added that the referenced figure shows the ward movements, hoping this makes it clearer.

– How was the date of April 9 fixed to define a cut-off between the early phase and late phase? Is it just based on the peak incidence? Or due to a change in screening practice?

The epidemic phase was determined on the basis of several factors. First, many preventive measures were implemented around that date (e.g., weekly screening of patients on April 07, enhanced testing of HCWs on April 09, decrease in bed-occupancy and closures on April 11). Second, approximately half of the cases occurred in each phase of the outbreak. Third, the overall trend in the epidemic curve can be interpreted as increasing in the first phase, and decreasing in the second phase.

– The large outbreak size makes figure 3 quite hard to interpret for the reader. Would it be possible to improve it possibly by zooming on some specific cases or wards, and adding global statistics informing on the precision/quality of the inference of transmission routes?

We agree with Reviewer 1 that Figure 3 was difficult to read. We have now created an interactive version of this figure where one can hover over points and zoom in selected areas. We have also removed the colours, which did not have a particular meaning. We have decided to move this new interactive figure as Figure 3 – Supplementary Figure 1, and have replaced it with a more informative figure demonstrating the extent to which our model is able to infer of transmission routes (i.e., showing the distribution of the posterior support of ancestries based on different criteria (individual, case type, ward), see new Figure 3).

– How does the number of secondary infections relate to the population's sizes (i.e. at-risk patients and HCWs from the studied wards). Is that relationship expected?

The Reviewer raises a valid and interesting point. Generally, and under normal circumstances, the ward sizes are comparable, as are the number of HCWs working in each ward. However, the number of patients in a ward is constantly fluctuating due to admissions/discharges as well as ward closures. Similarly, the number of HCWs fluctuates due to sick leave or annual leave. Therefore, the wards are not “closed systems”, and as the denominator fluctuates, more complex models would be required in order to explain the underlying mechanisms. We have added a sentence in the discussion to acknowledge this limitation:

“We were unable to relate the number of secondary infections with the population (ward) size; more complex models would be required to explain the underlying mechanisms in a context of fluctuating denominators, e.g., due to ward closures, etc.”

– We may have missed this: (apologies if so): were HCWs cohorted by wards or were there HCWs shared among wards? How may that affect the results here?

We thank the Reviewer for pointing this out. HCWs did work exclusively in one ward type or another. We have clarified this in the Methods (“Definitions” section):

“HCWs worked in either one or another type of ward, except for one HCW who worked in both Covid and non-Covid wards (although the proportion and/or days worked in each type of ward is unknown). Six HCWs worked across multiple wards (e.g. on-call) and were attributed to Covid-wards for the analyses.”

Reviewer #2 (Recommendations for the authors):

This paper addresses an important question about the dynamics of SARS-CoV-2 in a geriatric acute-care hospital. This reviewer's comments address primarily manuscript clarity.

General Comments:

Patients should not be equated with their conditions. Please replace Covid-19 and non-Covid 19 patients with patients who have Covid-19 and patients who do not have Covid-19 throughout the manuscript.

Compare takes "with" not "to".

We absolutely agree with the reviewer that patients should not be equated with their conditions. Nevertheless, we feel that the use of “patients who have Covid-19” may unnecessarily lengthen the manuscript and complicate reading. Also, we believe that the use of “Covid-19 patients” is neither disrespectful or stigmatising, and has been used in official documents from highly respected institutions (e.g., WHO, https://apps.who.int/iris/rest/bitstreams/1292529/retrieve; NHS England, https://www.england.nhs.uk/coronavirus/publication/assessment-monitoring-and-management-of-symptomatic-covid-19-patients-in-the-community/). Nevertheless, should the Editor feel the same way as the Reviewer, we would be happy to reword.

We have corrected “compared to” with “compared with”.

The manuscript includes many "there is," "there are," "it is," xxx. These statements are in passive voice and are usually wordier and more convoluted than a similar statement in active voice. This reviewer suggests revising most or all of these statements. For example, the following sentence could be revised as shown.

"The reservoir of SARS-CoV-2 in healthcare environments may contribute to amplifying the pandemic [5], and as such, it is important to understand transmission dynamics in these settings."

"Because the reservoir of SARS-CoV-2 in healthcare environments may contribute to amplifying the pandemic [5], we need to better understand transmission dynamics in these settings."

We thank the Reviewer for their comment, which we have taken into consideration wherever possible. We do, however, defer to the Editor and the Editorial team for further editorial and wording matters.

Statements like "in order to" and "as well as" have extra words that don't add to the meaning. They can be shortened to "to" and "and".

We understand the Reviewer’s point, and have modified accordingly.

Specific Comments

Line 67: Is the word "unexpectedly" needed? This reviewer was confused by that statement. I would assume HCWs knowing that they were caring for patients with Covid-19 might be more in adherence with PPE use and hand hygiene than those on non-Covid-19 units.

We understand the Reviewer’s point, and indeed, in 2022 it is conventional wisdom to make the assumption raised. However, at the time these outbreaks occurred, we did not think there would be such a stark difference in terms of cross-infections amongst HCWs in Covid vs. non-Covid wards. Indeed, the finding is unexpected. For this reason, should the Editor agree, we prefer to keep the word “unexpectedly”.

Lines 93-95: The meaning of "in nosocomial Covid-19" is vague and should be clarified. Does this mean HCWs' role in transmitting SARS-CoV-2 within healthcare facilities is complex?

Does it mean HCWs role in the epidemiology of nosocomial SARS-CoV-2 infection is

complex? This reviewer does not like the word "victim" (although the alteration is nice) in this setting. This reviewer thinks the last part of this sentence will work better if revised to: "as HCWs can acquire SARS-CoV-2 in the community or from their peers and patients and they can transmit SARS-CoV-2 to peers and patients [7, 8].

We thank the Reviewer for their comment. We have modified the sentence “in nosocomial Covid-19” to “in nosocomial Covid-19 transmission dynamics”, and hope that the sentence is clearer. We understand that the reviewer does not like the word victim; however, we believe that its use is justified in the present context, and unlike many health conditions, we feel that in the case of Covid-19 in most societies around the world there is no particular stigma associated with it. We also respectfully refer the Reviewer to the book entitled “The Patient as Victim and Vector: Ethics and Infectious Disease” by Margaret P. Battin, Leslie P. Francis, Jay A. Jacobson, and Charles B. Smith (Oxford Scholarship Online, 2009. DOI:10.1093/acprof:oso/9780195335842.001.0001).

Methods

The authors switch frequently between active and passive voice. This reviewer suggests picking one and sticking with it. This reviewer prefers an active voice for clarity and brevity.

We thank the Reviewer for these suggestions that we have implemented.

Line 162: This reviewer does not understand what the authors mean by "trajectories" in this sentence.

We meant “ward movements”, which has now replaced “trajectories”.

Lines 165-6: Revise to something like: "conservative in that it accounts for non-infectious contacts and incomplete reporting of infectious contacts and it estimates the proportion of the total contacts in these categories."

We have replaced the sentence to the following, hoping that the Reviewer finds it clearer:

“The manner by which outbreaker2 handles these contacts is conservative in that it allows for non-infectious contacts to occur (false positives) and incomplete reporting of infectious contacts (false negatives). In addition, the model estimates the proportions of the proportions of these contacts.”

Lines 18-91: Suggested revision. A post hoc epidemiological analysis of the 53 patients found that 4 of these patients likely acquired Covid-19 in the community (CA-Covid-19). The 49 nosocomial cases represented 20.2% (49/242) of all patients with COVID-19, and 81.7% (49/60) of nosocomial Covid-19 cases identified in the Geriatric Hospital. The ward-level attack rates ranged from 10 to 19% among patients and 21% of all HCWs in the geriatric..."

We thank the Reviewer for these suggestions that we have implemented.

Line 203: This reviewer does not understand this sentence given that the verb appears to be "shows".

We have modified this sentence in order for it to be grammatically correct, and hopefully easier to understand:

“There is also a large cluster (BS 68% with signature mutations C8293T, T18488C, and T24739C) with several subclusters includes 19 HCWs, nine patients, and three community cases; ward movements for the patients are shown in Appendix Figure 1”

Lines: 273-5: Suggest revising to something like: "Fourth, we identified multiple importation events that led to a substantial number of secondary cases or clusters. Most transmission events were related to HCWs, but one (?? or some) were related to a patient with community-acquired Covid-19." Note: the authors did not make clear how many transmissions were related to this patient.

We thank the Reviewer for these suggestions that we have implemented. We would like to point out that the number of secondary transmissions related to the patient with community-acquired Covid-19 was mentioned in the Results, under the heading “Imported cases”, which, incidentally, has been amended to better reflect the uncertainty associated with the reconstruction of the transmission chains:

“Imported Cases

From the reconstructed trees, we identified 22 imports in total (17 HCWs, five patients) with posterior support ≥10%. The 22 imported cases generated 41 secondary cases (posterior support ≥10%), with a median posterior support of 32.4% (IQR 17.0-53.7%). When restricting to imports with ≥50% posterior support, there were 16 imported cases 16 (12 HCWs, four patients), generating 35 secondary cases. There was some degree of uncertainty, reflected by circular transmission pathways, in determination of imported cases and their secondary cases. There were 6 transmission pairs (C114-C115, C153-H1057, H1008-H1059, H1011-H1019, H1017-H12021, H1052-H1082) where there was uncertainty as to which of the cases was imported and which was a secondary case, i.e. for each case in the pair there was a ≥10% probability of importation, but also ≥10% probability of being a secondary case of an imported case. Therefore, in total, 29 cases were “pure” secondary cases of imported cases (Table 3).”

Lines 282-292: Suggest revising to something like: "HCWs' behavior may have affected transmission. Indeed, Ottolenghi et al., found that HCWs caring for Covid-19 patients were concerned about becoming infected while caring for patients, and therefore may apply IPC measures more rigorously than when caring for patients who do not have Covid-19 [23]. HCWs working in non-Covid wards may not have felt threatened by Covid-19 patients on other wards. HCWs in non-Covid wards also may have underestimated the transmission risk from their peers to a greater extent than those working in Covid wards, and thus did not maintain physical distancing well. In addition, we found a higher mean duration (2.9 days) of presenteeism despite symptoms compatible with Covid-19 among HCWs working in non-Covid wards than for those working in Covid wards (1.6 days), which gives credence to the abovementioned possible explanations for the different transmission patters. Other factors (e.g., work culture, baseline IPC practices) also may have affected transmission patterns.

This reviewer wonders if the following sentence is necessary? "HCWs working in non-Covid wards may not have felt threatened by Covid-19 patients on other wards."

We thank the Reviewer for taking the time to make these detailed suggestions. We have largely changed the paragraph accordingly:

“HCWs' behaviour may have affected transmission. Indeed, Ottolenghi et al., found that HCWs caring for Covid-19 patients were concerned about becoming infected while caring for patients, and therefore may apply IPC measures more rigorously than when caring for patients who do not have Covid-19 [23]. HCWs working in non-Covid wards may not have felt threatened by Covid-19 patients who were in principle allocated to other wards, and thus not in their direct care. HCWs in non-Covid wards also may have underestimated the transmission risk from their peers to a greater extent than those working in Covid wards, and thus may not have maintained physical distancing well. In addition, we found a higher mean duration (2.9 days) of presenteeism despite symptoms compatible with Covid-19 among HCWs working in non-Covid wards than for those working in Covid wards (1.6 days), which gives credence to the abovementioned possible explanations for the different transmission patterns. Other factors (e.g., work culture, baseline IPC practices) also may have affected transmission patterns.”

Line 294: This author is wondering how the same ward is close proximity when patients are not in the same room. How would the patients have contact with each other?

The architectural layout of the hospital means that there are 2 wings per floor, each with a central corridor, and 2 wards per wing (without separation). On either side of the central corridor are patient rooms and/or offices (nursing, medical, etc.). Therefore, the wards themselves are crowded areas, and it is sufficient for a patient to step outside their room in the corridor to potentially come in contact with another patient. We have added this information in the Appendix, alongside data on room sizes, under the heading “Ward architecture and room sizes in the geriatric hospital”.

Lines 294-309: Suggested revisions: We cannot exclude transmission from a “point-source” or via an HCW’s contaminated hands (i.e., an unidentified HCW who transmitted SARS-CoV-2 to multiple patients in the same ward). To date, we have little evidence to suggest that this is the case; indeed, the transmission patterns were robust to changes in the model assumptions. Mathematical models have suggested that single-room isolation of suspected cases could potentially reduce the incidence of nosocomial SARS-CoV-2 transmission by up to 35% [24]. In the outbreaks we describe, symptomatic patients were identified promptly, with a median delay of 0 days between symptom onset and their first positive test. However, these precautions may not be sufficient as patients may transmit the virus when they are pre-symptomatic [25]. Thus, infection prevention teams may need to identify patients at high risk of developing nosocomial Covid-19 [26] if single rooms are not available for all exposed patients (e.g., in cases of overcrowding). For example, Mo et al., found that exposure to community-acquired cases who were identified and segregated or cohorted was associated with half the risk of infection compared with exposure to hospital-acquired cases or HCWs who may be asymptomatic [27]. One possible explanation for this finding is that patients with CA-Covid may have passed the peak of infectiousness when they are admitted, whereas patients with HA-Covid cases have frequent unprotected contact with HCWs and other patients during their period of peak infectiousness.

We thank the Reviewer for taking the time to make these detailed suggestions, which we have implemented in the manuscript.

Line 310: Suggested edit: The current evidence does not support the use of real-time genomics for control of SARS-CoV-2 nosocomial outbreaks [28].

We thank the Reviewer for their suggestion. We have also taken the opportunity to include a reference to the preprint by Stirrup et al., (medRxiv 2022.02.10.22270799) which reports the COG-UK HOCI study.

Lines 324-328: Suggested edits: Despite these strengths, our study had some limitations. First, we included only one sequence from a CA-Covid. However, the method we used to reconstruct who infected who is able to cope with and identify missing intermediate cases, which allowed us to estimate that our sample included the overwhelming majority of cases (91.5%). In addition, we performed these investigations during the first pandemic wave in a susceptible population...."

We thank the Reviewer for these suggestions, some of which we have implemented.

Line 334: Replace "in the case of" with "during".

We believe that “in the case of” is a better fit, as although it is likely that there will be another pandemic, it is still an unknown.

Lines 336-338: The main point of the conclusion is not clear. What does it mean to take into account the complex interplay between HCWs in dedicated Covid-19 wards vs non-Covid wards? The way this is worded makes it sound like the HCW on the two different kinds of wards are interacting, which likely is not what the authors intended.

We have replaced the conclusion by the following sentence, hoping it is clearer:

“In conclusion, strategies to prevent nosocomial SARS-CoV-2 transmission in geriatric settings should take into account the potential for patient-to-patient transmission and the transmission dynamics between HCWs in non-Covid wards, which our study suggests may differ from those in dedicated Covid-19 wards.”

Table 1: This reviewer doesn't understand what is meant by "Onset of symptoms before swab date, n (%)". Is this the number of people who were symptomatic before they were tested?

Yes, the Reviewer is correct.

Table 2: Should HCWs be written out in the title? Readers in the US will likely not know what a logopedist is. We will call this person a speech therapist. Perhaps the authors could include a footnote defining this term. Did the authors address the longer time from onset of symptoms to testing for HCWs on non-Covid wards compared with those on Covid wards in the results and discussion?

HCWs was written out in full. Logopedist was replaced by “speech therapist”.

The longer time from onset of symptoms to testing has been added in the Results section. We had already brought up the issue in the third paragraph of the Discussion.

Table 3: Dose "Secondary onward transmission" refer to transmission from the case in the first column? Consider putting NA (not applicable) in the blank cells.

We have replaced “Secondary onward transmission” by “Secondary onward transmission by imported case” in order to make it clearer. Blank cells were replaced by N/A.

Line 515: Suggested edit: "Histograms displaying the distributions of secondary cases...."

Thank you for this suggestion which we have taken into account.

Line 525: Is something missing between "... patient community and "proportion"?

Apologies as the phrase “proportion of infections attributed to the type of case” was accidentally inserted. We hope that after deletion the sentence is clearer.

Reviewer #3 (Recommendations for the authors):

Methods:

The authors present the IPC measures on page 6 in the main text and in the supplementary material. However, there seems to be no information on how HCWs adhered to the recommendations that were not mandatory and whether adherence differed between wards. This is probably because the authors did not have any information on that. This is a limitation and should be explained more explicitly or highlighted in the discussion.

We agree with the Reviewer that this is a limitation of our study. We have added this in the Discussion:

“We did not collect data on adherence to Covid-specific IPC recommendations by HCWs in different wards.”.

It's not clear to me whether the HCWs only worked in either COVID or non-COVID wards during the study period. Please clarify in the Methods section.

We thank the Reviewer for pointing this out. HCWs did work exclusively in one ward type or another. We have clarified this in the Methods (“Definitions” section):

On page 7, line 139: The authors state that they substituted the missing date of symptom onsets with the median difference between symptom onset date and swab date? The epidemic curve in Figure 1 shows the incidence on the y-axis but does the x-axis (date of symptom onset) make sense if not all of the cases were symptomatic?

The Reviewer has rightly pointed out that the Epidemic curve is based on symptom onset and that for the asymptomatic cases (n = 8, 4.4%) there is no date available. It is common practice to use date of symptom onset to construct epidemic curves, and this explains our choice. So yes, the epidemic curve includes the asymptomatic cases, and we have clarified this in the legend of Figure 1 (“Includes 8 asymptomatic cases for whom date of onset was inferred (c.f. text).”). We could have, instead, used the date of the first positive sample. While using the date of the first positive sample might ensure that there were no missing data points, it does provide a completely different picture of the progression of the outbreak. For example, the earliest positive swab was on March 11, 2020, whereas the earliest onset of symptoms was 6 days earlier on March 05, 2020. An additional reason is that the outbreaker2 model uses date of symptom onset as the starting point from which the date of infection is inferred statistically. Should the Editor request that we construct an epidemic curve using date of swab or removing the patients with the missing date of symptom onset, we will gladly do it. We can also highlight in the epidemic curve the 8 asymptomatic cases, should the Editor and/or the Reviewer request this.

Page 8, line 169 and Figure 4: How was the epidemic phase determined?

Results:

At the beginning of the Results section, the authors write that post hoc epidemiological analysis showed that were likely 4 patients that had a community-acquired infection. What exactly did the post-hoc epidemiological analysis entail? Later on (page 10, lines 206ff), they present their analysis of the imported cases. How are these related?

We thank the Reviewer for their careful reading of our paper. The post hoc epidemiological analysis consisted of a chart review performed by the investigators at the start of the study, as opposed to conclusions that were arrived at during the management of the outbreak. The classification of cases as being imported is based on the model output, and they are defined as those that do not have apparent ancestors among the cases included in the outbreak. We have included this information in the manuscript, hoping that this has been clarified.

Discussion:

Could the authors discuss whether misclassifications (either of community- or hospital-acquired cases) even after the analysis is still possible and what impact it could have on their results?

The Reviewer rightly points out that misclassifications may still occur. It is even part of the method (outbreaker2) to quantify the uncertainty in determining the ancestry, as it operates within a Bayesian framework. Indeed, in most cases (c.f. new Figure 3) there are very few patients for whom the ancestry is identified with 100% posterior probability. For example, the model predicted that patients C107 and C115 were most likely imported cases with posterior probabilities of 100% and 57.5%, respectively. The predicted dates of infection were March 14-22 for case C107 and March 8-24 for case C115, which was after their admission dates (March 3 and March 18, respectively). The probability that infection occurred on or after date of admission for case C115 was 81%. Therefore the model acknowledges the uncertainty in the classification of these cases. We have modified the section on imported cases (under the heading “imported cases”), and added a sentence after referencing the new Figure 3 to highlight the uncertainty in the model.

Supplement

Page 5: Could the authors elaborate on "global influence" for imported cases?

The global influence is a way of measuring the impact of a single observation on the log-likelihood of a parameter. During a preliminary run of the model, cases are removed in turn (with a leave one out approach) and the “global influence” of each case is measures as the extent to which removing it affects the genetic log-likelihood. Cases with high global influence (whose removal dramatically affects the genetic likelihood) are considered to be genetic outliers, and classified as imported cases. This approach is described in more detail in the original outbreaker manuscript (Jombart et al., PLoS Comput Biol 10(1): e1003457). We have added these details in the Appendix.

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Figure 3—figure supplement 1—source code 1. Interactive ancestry plot (who infected whom) – identical to Figure 3—figure supplement 1.

elife-76854-fig3-figsupp1-code1.zip^{(1.6MB, zip)}

Transparent reporting form

elife-76854-transrepform1.docx^{(111.5KB, docx)}

Data Availability Statement

[bib1] Abbas M, Robalo Nunes T, Cori A, Cordey S, Laubscher F, Baggio S, Jombart T, Iten A, Vieux L, Teixeira D, Perez M, Pittet D, Frangos E, Graf CE, Zingg W, Harbarth S. Explosive nosocomial outbreak of SARS-CoV-2 in a rehabilitation clinic: the limits of genomics for outbreak reconstruction. The Journal of Hospital Infection. 2021a;117:124–134. doi: 10.1016/j.jhin.2021.07.013. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib2] Abbas M, Robalo Nunes T, Martischang R, Zingg W, Iten A, Pittet D, Harbarth S. Nosocomial transmission and outbreaks of coronavirus disease 2019: the need to protect both patients and healthcare workers. Antimicrobial Resistance and Infection Control. 2021b;10:7. doi: 10.1186/s13756-020-00875-7. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib3] Abbas M, Zhu NJ, Mookerjee S, Bolt F, Otter JA, Holmes AH, Price JR. Hospital-onset COVID-19 infection surveillance systems: a systematic review. The Journal of Hospital Infection. 2021c;115:44–50. doi: 10.1016/j.jhin.2021.05.016. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib4] Aggarwal D, Myers R, Hamilton WL, Bharucha T, Tumelty NM, Brown CS, Meader EJ, Connor T, Smith DL, Bradley DT, Robson S, Bashton M, Shallcross L, Zambon M, Goodfellow I, Chand M, O’Grady J, Török ME, Peacock SJ, Page AJ, COVID-19 Genomics UK (COG-UK) Consortium The role of viral genomics in understanding COVID-19 outbreaks in long-term care facilities. The Lancet. Microbe. 2022;3:e151–e158. doi: 10.1016/S2666-5247(21)00208-1. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib5] Ali ST, Wang L, Lau EHY, Xu X-K, Du Z, Wu Y, Leung GM, Cowling BJ. Serial interval of SARS-CoV-2 was shortened over time by nonpharmaceutical interventions. Science (New York, N.Y.) 2020;369:1106–1109. doi: 10.1126/science.abc9004. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib6] Arons MM, Hatfield KM, Reddy SC, Kimball A, James A, Jacobs JR, Taylor J, Spicer K, Bardossy AC, Oakley LP, Tanwar S, Dyal JW, Harney J, Chisty Z, Bell JM, Methner M, Paul P, Carlson CM, McLaughlin HP, Thornburg N, Tong S, Tamin A, Tao Y, Uehara A, Harcourt J, Clark S, Brostrom-Smith C, Page LC, Kay M, Lewis J, Montgomery P, Stone ND, Clark TA, Honein MA, Duchin JS, Jernigan JA, Public Health–Seattle and King County and CDC COVID-19 Investigation Team Presymptomatic SARS-CoV-2 Infections and Transmission in a Skilled Nursing Facility. The New England Journal of Medicine. 2020;382:2081–2090. doi: 10.1056/NEJMoa2008457. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib7] Bhattacharya A, Collin SM, Stimson J, Thelwall S, Nsonwu O, Gerver S, Robotham J, Wilcox M, Hopkins S, Hope R. Healthcare-associated COVID-19 in England: A national data linkage study. The Journal of Infection. 2021;83:565–572. doi: 10.1016/j.jinf.2021.08.039. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib8] Bi Q, Wu Y, Mei S, Ye C, Zou X, Zhang Z, Liu X, Wei L, Truelove SA, Zhang T, Gao W, Cheng C, Tang X, Wu X, Wu Y, Sun B, Huang S, Sun Y, Zhang J, Ma T, Lessler J, Feng T. Epidemiology and transmission of COVID-19 in 391 cases and 1286 of their close contacts in Shenzhen, China: a retrospective cohort study. The Lancet. Infectious Diseases. 2020;20:911–919. doi: 10.1016/S1473-3099(20)30287-5. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib9] Braun KM, Moreno GK, Buys A, Somsen ED, Bobholz M, Accola MA, Anderson L, Rehrauer WM, Baker DA, Safdar N, Lepak AJ, O’Connor DH, Friedrich TC. Viral Sequencing to Investigate Sources of SARS-CoV-2 Infection in US Healthcare Personnel. Clinical Infectious Diseases. 2021;73:e1329–e1336. doi: 10.1093/cid/ciab281. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib10] Burugorri-Pierre C, Lafuente-Lafuente C, Oasi C, Lecorche E, Pariel S, Donadio C, Belmin J. Investigation of an Outbreak of COVID-19 in a French Nursing Home With Most Residents Vaccinated. JAMA Network Open. 2021;4:e2125294. doi: 10.1001/jamanetworkopen.2021.25294. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib11] Campbell F, Didelot X, Fitzjohn R, Ferguson N, Cori A, Jombart T. outbreaker2: a modular platform for outbreak reconstruction. BMC Bioinformatics. 2018;19:363. doi: 10.1186/s12859-018-2330-z. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib12] Campbell F, Cori A, Ferguson N, Jombart T. Bayesian inference of transmission chains using timing of symptoms, pathogen genomes and contact data. PLOS Computational Biology. 2019;15:e1006930. doi: 10.1371/journal.pcbi.1006930. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib13] Ellingford JM, George R, McDermott JH, Ahmad S, Edgerley JJ, Gokhale D, Newman WG, Ball S, Machin N, Black GC. Genomic and healthcare dynamics of nosocomial SARS-CoV-2 transmission. eLife. 2021;10:e65453. doi: 10.7554/eLife.65453. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib14] Evans S, Agnew E, Vynnycky E, Stimson J, Bhattacharya A, Rooney C, Warne B, Robotham J. The impact of testing and infection prevention and control strategies on within-hospital transmission dynamics of COVID-19 in English hospitals. Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences. 2021;376:1829. doi: 10.1098/rstb.2020.0268. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib15] Ferretti L, Wymant C, Kendall M, Zhao L, Nurtay A, Abeler-Dörner L, Parker M, Bonsall D, Fraser C. Quantifying SARS-CoV-2 transmission suggests epidemic control with digital contact tracing. Science (New York, N.Y.) 2020;368:eabb6936. doi: 10.1126/science.abb6936. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib16] Gelman A, Rubin DB. Inference from Iterative Simulation Using Multiple Sequences. Statistical Science. 1992;7:457–472. doi: 10.1214/ss/1177011136. [DOI] [Google Scholar]

[bib17] He X, Lau EHY, Wu P, Deng X, Wang J, Hao X, Lau YC, Wong JY, Guan Y, Tan X, Mo X, Chen Y, Liao B, Chen W, Hu F, Zhang Q, Zhong M, Wu Y, Zhao L, Zhang F, Cowling BJ, Li F, Leung GM. Temporal dynamics in viral shedding and transmissibility of COVID-19. Nature Medicine. 2020;26:672–675. doi: 10.1038/s41591-020-0869-5. [DOI] [PubMed] [Google Scholar]

[bib18] Jombart T, Cori A, Didelot X, Cauchemez S, Fraser C, Ferguson N. Bayesian reconstruction of disease outbreaks by combining epidemiologic and genomic data. PLOS Computational Biology. 2014;10:e1003457. doi: 10.1371/journal.pcbi.1003457. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib19] Klompas M, Baker MA, Rhee C, Tucker R, Fiumara K, Griesbach D, Bennett-Rizzo C, Salmasian H, Wang R, Wheeler N, Gallagher GR, Lang AS, Fink T, Baez S, Smole S, Madoff L, Goralnick E, Resnick A, Pearson M, Britton K, Sinclair J, Morris CA. A SARS-CoV-2 Cluster in an Acute Care Hospital. Annals of Internal Medicine. 2021;174:794–802. doi: 10.7326/M20-7567. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib20] Knight GM, Pham TM, Stimson J, Funk S, Jafari Y, Pople D, Evans S, Yin M, Brown CS, Bhattacharya A, Hope R, Semple MG, Read JM, Cooper BS, Robotham JV, ISARIC4C Investigators. CMMID COVID-19 Working Group The contribution of hospital-acquired infections to the COVID-19 epidemic in England in the first half of 2020. BMC Infectious Diseases. 2022;22:2021. doi: 10.1186/s12879-022-07490-4. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib21] Kumar S, Stecher G, Li M, Knyaz C, Tamura K. MEGA X: Molecular Evolutionary Genetics Analysis across Computing Platforms. Molecular Biology and Evolution. 2018;35:1547–1549. doi: 10.1093/molbev/msy096. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib22] Lucey M, Macori G, Mullane N, Sutton-Fitzpatrick U, Gonzalez G, Coughlan S, Purcell A, Fenelon L, Fanning S, Schaffer K. Whole-genome Sequencing to Track Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) Transmission in Nosocomial Outbreaks. Clinical Infectious Diseases. 2021;72:e727–e735. doi: 10.1093/cid/ciaa1433. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib23] Martischang R, Iten A, Arm I, Abbas M, Meyer B, Yerly S, Eckerle I, Pralong J, Sauser J, Suard J-C, Kaiser L, Pittet D, Harbarth S. Severe acute respiratory coronavirus virus 2 (SARS-CoV-2) seroconversion and occupational exposure of employees at A Swiss university hospital: A large longitudinal cohort study. Infection Control and Hospital Epidemiology. 2022;43:326–333. doi: 10.1017/ice.2021.117. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib24] Mendes A, Serratrice C, Herrmann FR, Genton L, Périvier S, Scheffler M, Fassier T, Huber P, Jacques M-C, Prendki V, Roux X, Di Silvestro K, Trombert V, Harbarth S, Gold G, Graf CE, Zekry D. Predictors of In-Hospital Mortality in Older Patients With COVID-19: The COVIDAge Study. Journal of the American Medical Directors Association. 2020;21:1546–1554. doi: 10.1016/j.jamda.2020.09.014. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib25] Meredith LW, Hamilton WL, Warne B, Houldcroft CJ, Hosmillo M, Jahun AS, Curran MD, Parmar S, Caller LG, Caddy SL, Khokhar FA, Yakovleva A, Hall G, Feltwell T, Forrest S, Sridhar S, Weekes MP, Baker S, Brown N, Moore E, Popay A, Roddick I, Reacher M, Gouliouris T, Peacock SJ, Dougan G, Török ME, Goodfellow I. Rapid implementation of SARS-CoV-2 sequencing to investigate cases of health-care associated COVID-19: a prospective genomic surveillance study. The Lancet. Infectious Diseases. 2020;20:1263–1272. doi: 10.1016/S1473-3099(20)30562-4. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib26] Mo Y, Eyre DW, Lumley SF, Walker TM, Shaw RH, O’Donnell D, Butcher L, Jeffery K, Donnelly CA, Cooper BS, Oxford COVID infection review team Transmission of community- and hospital-acquired SARS-CoV-2 in hospital settings in the UK: A cohort study. PLOS Medicine. 2021;18:10. doi: 10.1371/journal.pmed.1003816. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib27] Myall A, Price JR, Peach RL, Abbas M, Mookerjee S, Zhu N, Ahmad I, Ming D, Ramzan F, Teixeira D, Graf C, Weiße AY, Harbarth S, Holmes A, Barahona M. Predicting Hospital-Onset COVID-19 Infections Using Dynamic Networks of Patient Contacts: An Observational Study. medRxiv. 2021 doi: 10.1101/2021.09.28.21264240. [DOI] [PMC free article] [PubMed]

[bib28] Ottolenghi J, McLaren RA, Bahamon C, Dalloul M, McCalla S, Minkoff H. Health Care Workers’ Attitudes Toward Patients With COVID-19. Open Forum Infectious Diseases. 2021;8:ofab375. doi: 10.1093/ofid/ofab375. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib29] Shitrit P, Zuckerman NS, Mor O, Gottesman BS, Chowers M. Nosocomial outbreak caused by the SARS-CoV-2 Delta variant in a highly vaccinated population, Israel, July 2021. Euro Surveillance. 2021;26:2100822. doi: 10.2807/1560-7917.ES.2021.26.39.2100822. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib30] Stirrup O, Hughes J, Parker M, Partridge DG, Shepherd JG, Blackstone J, Coll F, Keeley A, Lindsey BB, Marek A, Peters C, Singer JB, COVID-19 Genomics UK (COG-UK) consortium. Tamuri A, de Silva TI, Thomson EC, Breuer J. Rapid feedback on hospital onset SARS-CoV-2 infections combining epidemiological and sequencing data. eLife. 2021;10:e65828. doi: 10.7554/eLife.65828. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib31] Stirrup O, Blackstone J, Mapp F, MacNeil A, Panca M, Holmes A, Machin N, Shin GY, Mahungu T, Saeed K, Saluja T, Taha Y, Mahida N, Pope C, Chawla A, Cutino-Moguel MT, Tamuri A, Williams R, Darby A, Robertson D, Flaviani F, Nastouli E, Robson S, Smith D, Loose M, Laing K, Monahan I, Kele B, Haldenby S, George R, Bashton M, Witney A, Byott M, Coll F, Chapman M, Peacock S, Hughes J, Nebbia G, Partridge DG, Parker M, Price J, Peters C, Roy S, Snell LB, de Silva TI, Thomson E, Flowers P, Copas A, Breuer J. Evaluating the Effectiveness of Rapid SARS-CoV-2 Genome Sequencing in Supporting Infection Control Teams: The COG-UK Hospital-Onset COVID-19 Infection Study. medRxiv. 2022 doi: 10.1101/2022.02.10.22270799. [DOI] [PMC free article] [PubMed]

[bib32] Stone SP, Cooper BS, Kibbler CC, Cookson BD, Roberts JA, Medley GF, Duckworth G, Lai R, Ebrahim S, Brown EM, Wiffen PJ, Davey PG. The ORION statement: guidelines for transparent reporting of outbreak reports and intervention studies of nosocomial infection. The Lancet. Infectious Diseases. 2007;7:282–288. doi: 10.1016/S1473-3099(07)70082-8. [DOI] [PubMed] [Google Scholar]

[bib33] Swissnoso Prevention & control of healthcare-associated COVID-19 outbreaks. Covid-19. 2021 https://www.swissnoso.ch/fileadmin/swissnoso/Dokumente/5_Forschung_und_Entwicklung/6_Aktuelle_Erreignisse/200515_Prevention_and_control_of_healthcare-associated_COVID-19_outbreaks_V1.0_ENG.pdf

[bib34] Tamura K. Estimation of the number of nucleotide substitutions when there are strong transition-transversion and G+C-content biases. Molecular Biology and Evolution. 1992;9:678–687. doi: 10.1093/oxfordjournals.molbev.a040752. [DOI] [PubMed] [Google Scholar]

[bib35] Taylor J, Carter RJ, Lehnertz N, Kazazian L, Sullivan M, Wang X, Garfin J, Diekman S, Plumb M, Bennet ME, Hale T, Vallabhaneni S, Namugenyi S, Carpenter D, Turner-Harper D, Booth M, Coursey EJ, Martin K, McMahon M, Beaudoin A, Lifson A, Holzbauer S, Reddy SC, Jernigan JA, Lynfield R, Minnesota Long-Term Care COVID-19 Response Group Serial Testing for SARS-CoV-2 and Virus Whole Genome Sequencing Inform Infection Risk at Two Skilled Nursing Facilities with COVID-19 Outbreaks - Minnesota, April-June 2020. MMWR. Morbidity and Mortality Weekly Report. 2020;69:1288–1295. doi: 10.15585/mmwr.mm6937a3. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib36] The Guardian 40,600 people likely caught Covid while hospital inpatients in England. Covid-19. 2021a https://www.theguardian.com/society/2021/mar/26/40600-people-likely-caught-covid-while-hospital-inpatients-in-england

[bib37] The Guardian Up to 20% of hospital patients with Covid-19 caught it at hospital. Covid-19. 2021b https://www.theguardian.com/world/2020/may/17/hospital-patients-england-coronavirus-covid-19

[bib38] The Lancet Infectious Diseases COVID-19 vaccine equity and booster doses. Lancet (London, England) 2021;21:1193. doi: 10.1016/S1473-3099(21)00486-2. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib39] Thiabaud A, Iten A, Balmelli C, Senn L, Troillet N, Widmer A, Flury D, Schreiber PW, Vázquez M, Damonti L, Buettcher M, Vuichard-Gysin D, Kuhm C, Cusini A, Riedel T, Nussbaumer-Ochsner Y, Gaudenz R, Heininger U, Berger C, Zucol F, Bernhard-Stirnemann S, Corti N, Zimmermann P, Uka A, Niederer-Loher A, Gardiol C, Roelens M, Keiser O. Cohort profile: SARS-CoV-2/COVID-19 hospitalised patients in Switzerland. Swiss Medical Weekly. 2021;151:w20475. doi: 10.4414/smw.2021.20475. [DOI] [PubMed] [Google Scholar]

[bib40] von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP, STROBE Initiative Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. BMJ (Clinical Research Ed.) 2007;335:806–808. doi: 10.1136/bmj.39335.541782.AD. [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Reconstruction of transmission chains of SARS-CoV-2 amidst multiple outbreaks in a geriatric acute-care hospital: a combined retrospective epidemiological and genomic study

Mohamed Abbas

Anne Cori

Samuel Cordey

Florian Laubscher

Tomás Robalo Nunes

Ashleigh Myall

Julien Salamun

Philippe Huber

Dina Zekry

Virginie Prendki

Anne Iten

Laure Vieux

Valérie Sauvan

Christophe E Graf

Stephan Harbarth

Roles

Abstract

Background:

Methods:

Results:

Conclusions:

Funding:

Introduction

Methods

Setting

Definitions

Data sources

Descriptive epidemiology

Microbiological methods

Phylogenetic analysis

Statistical analysis

Reconstruction of transmission trees

Results

Figure 1. Epidemic curve of the nosocomial COVID-19 outbreak in a geriatric hospital involving HCWs and patients.

Figure 1—figure supplement 1. Ward-level epidemic curve.

Table 1. Characteristics of Covid-19 patients with nosocomial acquisition.

Table 2. Characteristics of SARS-CoV-2 RT-PCR positive healthcare workers.

Phylogenetic tree

Figure 2. Phylogenetic tree of SARS-CoV-2 genome sequences.

Imported cases

Table 3. Imported cases and secondary infections, patients and HCWs are named C1xx and H10xx, respectively.

Reconstructing who infected whom

Figure 3—figure supplement 1. Ancestry reconstruction (who infected whom) of the outbreaker2 model.

Figure 3—figure supplement 2. Distribution of number of missed generations across posterior trees, stratified by phase of outbreak.

Figure 3—figure supplement 3. Comparison of the accuracy of ancestry attribution of each sensitivity analysis.

Ward attribution

Transmission patterns

Secondary infections

Role of HCWs and patients in transmission events

Figure 5. Proportions of transmissions (fcase) attributed to each case type (HCWcovid, HCWoutbreak, patientnoso, and patientcommunity) for each of the 1000 posterior trees retained.

Role of within-ward and between-ward transmission

Discussion

Acknowledgements

Appendix 1

Appendix 1—table 1. Composition of baseline outbreaker2 model and different sensitivity analyses.

Appendix 1—table 2. Proportions of secondary infections (i.e., individual R) for each case type (patientnoso, HCWoutbreak, nd HCWcovid) in early (up to 9 April 2020) and late (as of 10 April 2020) phases of the study.

Appendix 1—figure 1. Ward movements for patients involved in a cluster.

Appendix 1—figure 2. Ward-to-ward transmission matrix.

Appendix 1—figure 3. Proportions of transmissions attributed to (A) outbreak (foutbreak-ward) and (B) non-outbreak (fnon-outbreak-ward) wards.

Infection prevention and control measures during the first pandemic wave

Ward architecture and room sizes in the geriatric hospital

Microbiological methods

Amplicon-based high-throughput sequencing analysis

Phylogenetic analysis

Statistical analyses

Implementation of the outbreaker2 models

Sensitivity analyses

Funding Statement

Contributor Information

Funding Information

Additional information

Competing interests

Author contributions

Ethics

Additional files

Data availability

References

Editor's evaluation

Figure 3—figure supplement 1. Ancestry reconstruction (who infected whom) of the `outbreaker2` model.

Figure 5. Proportions of transmissions (f_case) attributed to each case type (HCW_covid, HCW_outbreak, patient_noso, and patient_community) for each of the 1000 posterior trees retained.

Appendix 1—table 1. Composition of baseline `outbreaker2` model and different sensitivity analyses.

Appendix 1—table 2. Proportions of secondary infections (i.e., individual R) for each case type (patient_noso, HCW_outbreak, nd HCW_covid) in early (up to 9 April 2020) and late (as of 10 April 2020) phases of the study.

Appendix 1—figure 3. Proportions of transmissions attributed to (A) outbreak (f_{outbreak-ward}) and (B) non-outbreak (f_{non-outbreak-ward}) wards.

Implementation of the `outbreaker2` models