Estimating the risk of incident SARS-CoV-2 infection among healthcare workers in quarantine hospitals: the Egyptian example

Introduction

Healthcare settings have faced multiple challenges related to the currently ongoing severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic. They have notably had to deal with influxes of infected patients across successive epidemic waves while controlling the risk of nosocomial spread to other patients and staff. As a result, healthcare workers (HCWs) have been a population of interest in terms of risk assessment and control measure implementation [1].

At early stages of the pandemic, Egypt was identified as one of the African countries most vulnerable to SARS-CoV-2 importation [2]. On February 14, 2020, Egypt reported the first confirmed case of coronavirus disease 2019 (COVID-19) in Africa and remained among the five African countries most impacted by the COVID-19 epidemic up to the end of 2020 [3]. From February 14 to August 31, 2020 (the period to which we refer to as the first wave), the COVID-19 epidemic in Egypt resulted in about 9,700 confirmed infections and 5,500 deaths reported nationally [4]. These numbers most certainly reflect underreporting of the real number of infections, due to the high proportion of asymptomatic SARS-CoV-2 infection [5, 6] and limited testing.

To mitigate the high risk of SARS-CoV-2 spreading, Egypt established a unique care model under supervision from the World Health Organization, whereby specific hospitals were assigned as quarantine hospitals for patients with COVID-19, and where dedicated medical teams stayed in the hospital days and nights during working shifts of various durations [7]. The quarantine-hospital protocol required HCWs to be screened for SARS-CoV-2 infection before starting a working shift, and to be systematically tested at the end of a shift, as well as during shifts, in the case of developing symptoms and/or outbreak suspicion.

While the quarantine-hospital strategy has the potential to be highly efficient in terms of patient care, as well as in limiting the potential spread of the virus from hospitals into the community, its impact on infection risk for HCWs remains understudied. One theoretical modelling study assessing healthcare working force organization found that alternating HCW teams by 1-week periods may reduce the overall number of infected HCWs [8]. On the other hand, to the best of our knowledge, previous epidemiological studies focusing on the infection risk faced by Egyptian HCWs have been conducted in non-quarantine settings exclusively [9–11]. Yet, the quarantine strategy was adopted again to face the second wave of the COVID-19 epidemic at the national level, and could be adopted for future epidemic waves in Egypt and/or in other countries [12].

Here, we used mathematical modelling to estimate the risk of SARS-CoV-2 infection among HCWs participating in quarantine-hospital interventions, relying on detailed longitudinal data collected in three Egyptian healthcare facilities (hereafter denoted by Hosp1, Hosp2 and Hosp3) during the first wave of the COVID-19 epidemic.

Results

The study period of the quarantine-hospital intervention covered ten 2-week shifts in Hosp1, nine ~ 2-week shifts in Hosp2 and five 1-week shifts in Hosp3, during the first COVID-19 wave in Egypt (March–July, 2020). The mean (min–max) number of HCWs per shift was 46 (34–63) in Hosp1, 15 (5–26) in Hosp2 and 19 (16–20) in Hosp3 (Table 1). Over a total follow-up of 8,733 person-days (PD), 54 SARS-CoV-2 infections were observed (ie., confirmed SARS-CoV-2 infection diagnosis) among 722 HCWs showing no evidence of SARS-CoV-2 antibodies at the beginning of their shifts, across the three hospitals. This represented an overall incidence rate of 0.62 (95% CI: 0.45–0.78) diagnosed SARS-Cov-2 infections per 100 PD (Table 2). A significantly higher incidence rate was observed for HCWs working at Hosp3 as compared to the other two hospitals; and HCWs working in intensive care units (ICU) tended to face a lower risk than HCWs working in non-ICU units, though this difference was not statistically significant (Table 2).

Table 1. Hospital characteristics.

	Hosp1	Hosp2	Hosp3
Quarantine organization period	March 14 – August 1	April 1 – July 31	June 6 – July 11
Location	Cairo	Fayoum	Cairo
Mean daily number of patients (min–max)	62 (0–108)	37 (0–103)	8 (0–20)
Mean number of HCWs per shift (min–max)	46 (34–63)	15 (5–26)	19 (16–20)
Shift duration	14 days	7–14 days	7 days

Quarantine organization period, location and mean number of HCWs and patients by hospital (Hosp1, Hosp2 and Hosp3). Abbreviations: HCWs = Healthcare workers, PD = person-days.

Table 2

Observed risk of SARS-CoV-2 infection among HCWs, by hospital and by hospital unit.
	Hosp1	Hosp2	Hosp3
Quarantine organization period	March 14 – August 1	April 1 – July 31	June 6 – July 11
Location	Cairo	Fayoum	Cairo
Mean daily number of patients (min–max)	62 (0–108)	37 (0–103)	8 (0–20)
Mean number of HCWs per shift (min–max)	46 (34–63)	15 (5–26)	19 (16–20)
Shift duration	14 days	7–14 days	7 days
A total of 54 infections were observed (ie., diagnosed before or at the end of a shift) over 8,733 person-days, in the three hospitals (Hosp1, Hosp1 and Hosp3). Crude rates of SARS-CoV-2 infection among HCWs by hospital and by hospital unit: intensive and non-intensive care units and 95% confidence intervals. A Poisson regression adjusted by hospital and by hospital unit was performed on the observed infections to obtain the incidence rate ratios and 95% confidence intervals. Abbreviations: CI = confidence interval, ICU = intensive care unit, IRR = incidence rate ratio, PD = person-days.

Healthcare-associated outbreaks among HCWs (observed attack rates of $\ge 20\%$) represented ~70% (38/54) of all SARS-CoV-2 infections observed over the study period (Fig. 1). Two of these outbreaks occurred in Hosp2 with 30% (3/10) and 24% (5/21) of HCWs being infected over two different working shifts. An outbreak where 36% of HCWs were infected (16/44) occurred in Hosp1 around the same time period. Two outbreaks took place later on in Hosp3, leading to, respectively, 30% (6/20) and 40% (8/20) of susceptible HCWs being infected over two different shifts. Of note, each outbreak resulted in infections in both ICU and non-ICU. For the last shift in Hosp3, staff recruitment relied exclusively on HCWs who had been infected during the first or the second shifts. This likely prevented HCWs in Hosp3 from becoming infected during this shift (cf. Figure 1).

The main outcomes estimated by of our mathematical model are the mean incidence rates of both observed (ie., diagnosed) and unobserved (ie., undiagnosed) SARS-CoV-2 infections in each hospital, over the whole study period. We found significantly different risk levels between the three hospitals (Kolmogorov-Smirnov test), with an incidence rate of 1.05 (95% CrI: 0.58–1.65) per 100 PD in Hosp1, 1.92 (95% CrI: 0.93–3.28) per 100 PD in Hosp2 and 7.62 (95% CrI: 3.47–13.70) per 100 PD in Hosp3.

The probability for a HCW to be infected at the end of a working shift was estimated at 13.7% (95% CrI: 7.8–20.8%) and 23.8% (95% CrI: 12.2–37.3%) for 2-week shifts at Hosp1 and Hosp2, respectively, whereas a much higher probability of 42.6% (95% CrI: 21.9–64.4%) was found for a 7-day shift at Hosp3 (Table 3). These probabilities become 7.1% (95% CrI: 4.0–11.0%) and 12.7% (95% CrI: 6.3–20.8%), for Hosp1 and Hosp2, respectively, if 7-day shifts are considered as well, for easier comparison (Table 3).

Table 3

Model-based estimates of all SARS-CoV-2 infections
		Observations		Crude rates		Adjusted Poisson Regression
		Events	PD	Rate per 100 PD	95% CI	IRR	95% CI	$\text{p}$
Hospital	Hosp1	28	6258	0.45	0.28–0.61	1	ref	–
	Hosp2	11	1808	0.61	0.25–0.97	1.49	0.74–2.87	0.28
	Hosp3	15	667	2.25	1.1–3.39	5.59	3.13–10.01	< 0.001
Hospital Unit	ICU	15	2628	0.57	0.28–0.86	1	ref	–
Hospital Unit	Non-ICU	39	6105	0.64	0.44–0.84	1.49	0.85–2.59	0.16
Model-based estimates of the incidence rate, probability of infection for a 7-day working shift and estimated number of SARS-CoV-2 infections in each hospital, over the whole study period. Running our model for each type of care unit (ICU and non-ICU) in each hospital unveils the difference in the risk faced by HCWs between the three hospitals of our study. Indeed, HCWs working in Hosp1 and Hosp2 face a higher risk when assigned to non-ICU, as compared with those working in the ICU; whereas HCWs in Hosp3 face a slightly higher risk when assigned to the ICU. All estimates concern both diagnosed and undiagnosed infections (all), unless otherwise stated (diagnosed). Abbreviations: CrI = credibility interval, ICU = intensive care unit, PD = person-days.

The per-hospital incidence rate estimates yields the following model-based median numbers of total SARS-CoV-2 infections: 59 (95% CrI: 45–74) at Hosp1, 31 (95% CrI: 14–52) at Hosp2 and 41 (95% CrI: 19–63) at Hosp3 (Table 3), which unveils a proportion of undiagnosed infections among HCWs of 46.4% (95% CrI: 18.8–66.7%), 45.0% (95% CrI: 5.6–70.8%) and 59.2% (95% CrI: 34.8–78.8%), respectively.

We further estimated the incident risk by hospital unit. Given the notable differences between hospitals (Fig. 1), we did not assess infection risk by care unit across hospitals, but within each hospital. This confirmed a notable difference between Hosp1–2 and Hosp3: our results suggest that HCWs working in Hosp1 and Hosp2 faced a higher risk when assigned to non-ICU, as compared to those assigned to the ICU; whereas HCWs in the ICU of Hosp3 faced higher risk (see Fig. 1 and Table 3).

Discussion

Here, we studied the risk of incident SARS-CoV-2 infection among HCWs residing days and nights in quarantine hospitals, relying on detailed longitudinal data collected during the first wave of COVID-19 in Egypt. We found that most diagnosed infections (70%) occurred during what we defined as healthcare-associated outbreaks (as compared to isolated infections). We observed high variability in nosocomial incidence, ranging from 0.45 to 2.25 infections per 100 PD across the three hospitals. Using a model-based approach, we estimated the risk of both diagnosed and undiagnosed SARS-CoV-2 infections among HCWs, by hospital and by hospital unit (ICU and non-ICU). We further estimated that a substantial proportion of infections may have remained undetected, ranging from 45.0–59.2% across the three hospitals.

Our study design has several limitations. First, HCWs were screened using rapid serological tests before starting their working shifts. This may have allowed recently infected – and probably infectious – HCWs to start a working shift, as antibodies are detected by serological tests in less than 40% of infected individuals within 7 days since symptoms onset [13]. The Egyptian public health authorities’ choice of a protocol relying on rapid serological tests before working shifts was mainly driven by material constraints. Indeed, the use of serological tests as a diagnostic tool was and remains of particular interest in limited-resource contexts due to their lower cost, the minimal equipment required, and faster results, as compared with reverse transcription-polymerase chain reaction (RT-PCR) tests [14]. For this study, this may have led us to slightly underestimate the infection risks, as some HCWs may not actually have been at-risk. Second, the data collected for this study was scarce. The urge of implementing control measures under limited resources forced Egyptian hospitals to implement the quarantine-hospital strategy rapidly and, in some cases, during very short periods of time. Scarce data limits the strength and reach of our methods, and thus, our modeling choices were driven by parsimony. In particular, we assumed a constant overall risk of infection among HCWs over time, in each hospital unit, even though this disregards dynamic fluctuations in risk driven by variation in the number of HCWs and/or patients in the hospitals, as well as the frequency and nature of HCWs’ contacts during their working shifts. Indeed, explicitly modelling different transmission routes (patient-to-HCW vs HCW-to-HCW) would have required additional parameters in the model, leading to identifiability issues due to the scarcity of our data. In addition, this assumption allowed us to compare our results between hospitals and with existing literature on similar contexts or settings (see below). We would like to however stress the importance of developing modelling tools applied to low- and middle- income contexts, despite the difficulties that may come with limited resources. In particular, it is of great interest to evaluate what was the impact of interventions such as the quarantine-hospitals, which was implemented only in a few settings, to accompany future decision-making.

We found a ~ 5-fold higher observed risk for Hosp3 as compared with the risk found in Hosp1; whereas the risk of infection at Hosp2 was just slightly higher. The higher risk of SARS-CoV-2 infection we report for Hosp3 and the higher risk faced by HCWs assigned to the ICU of Hosp3 may be partly explained by the short period over which this hospital adopted the quarantine organization, which coincided with the country's highest epidemic activity of the first wave [4]. This may have led to a high proportion of severe and thus highly contagious COVID-19 patients referred to quarantine hospitals together with a higher workload for HCWs. Contrarily, Hosp1 adopted the quarantine organization in mid-March 2020, and thus experienced several weeks of lower epidemic intensity that may have improved preparedness for intense COVID-19 activity and implementation of infection control measures for invasive procedures (e.g., intubation in the ICU). Third, despite the quarantine strategy being adopted nationally, the differences between the organizations of each of the three hospitals (working shifts lengths, quarantine-organization period) should be accounted for when comparing risk estimates across the three hospitals. Data from other hospitals that have adopted a quarantine strategy will be beneficial to better understand occupational risks and eventually study the performance of such a strategy in comparison to standard care settings.

Our model-based approach, accounting for false-negativity rates of testing, as well as the right-truncation of our data, allowed us to estimate the risk of both diagnosed and undiagnosed SARS-CoV-2 infection among HCWs, by hospital and by hospital care unit. We estimated that a rigorous quarantine organization with systematic testing of HCWs at the end of their working shifts captures only about half of all HCW infections. Of note, the higher proportion of infections that remained undetected in Hosp3 (59.2%, versus 46.4% and 45.0% in Hosp1 and Hosp2, respectively) may be explained by the shorter duration of the working shifts (7 days), which may result in false-negative tests for HCWs infected a short time before the end of their shift. Moreover, these results suggest that in the absence of systematic testing at the end of quarantine working shifts, an even larger proportion of infections among HCWs may remain undetected, thus putting HCWs’ close contacts at risk of infection [15] and, consequently, putting themselves at risk for adverse psychological symptoms [16].

The model-based estimates we found for the SARS-CoV-2 infection risk in Hosp1 and Hosp2 are consistent with infection point-prevalence reported in earlier studies performed in non-quarantine Egyptian hospitals (varying from 4.2–14.3%) and with incidence estimates reported among front-line HCWs in the UK (13% infection rate after one month) [10, 11, 17]. For comparison, a summary of previous results obtained early in the Covid-19 pandemic and/or specifically in the Egyptian context is presented in Table 4. Moreover, our results suggest that HCWs assigned to non-ICU face a higher risk than those assigned to the ICU, in Hosp1 and Hosp2, which is in line with what was observed in previous studies addressing the occupational risk of SARS-CoV-2 infection for HCWs in non-quarantine settings [18]. Overall, our findings on Hosp1 and Hosp2 suggest that, providing sufficient preparedness, HCWs working in quarantine hospitals may not face a higher infection risk, and thus highlight the benefits of implementing a quarantine-hospital strategy.

Table 4

Our results in context.
	Median estimated incidence rate per 100 PD (95% CrI)			Probability of infection for a 7-day shift (95% CrI)	Median number of SARS-CoV-2 infections (95% CrI)
	ICU	Non-ICU	Overall	Probability of infection for a 7-day shift (95% CrI)	All	Diagnosed	Proportion of undiagnosed infections
Hosp1	0.59 (0.11–1.61)	1.27 (0.68–2.01)	1.05 (0.58–1.65)	7.1% (4.0–11.0%)	59 (45–74)	31 (15–51)	46.4% (18.8–66.7%)
Hosp2	1.33 (0.33–3.33)	2.75 (1.12–5.02)	1.92 (0.93–3.28)	12.7% (6.3–20.8%)	31 (14–52)	17 (7–31)	45.0% (5.6–70.8%)
Hosp3	9.37 (3.48–19.66)	7.12 (2.06–18.71)	7.62 (3.47–13.70)	42.6% (21.9–64.4%)	41 (19–63)	16 (6–30)	59.2% (34.8–78.8%)
Comparison of the per-shift probability of infection estimated in Hosp1–3 with earlier estimates of the point-prevalence in other Egyptian studies and the observed cumulative incidence rate among HCWs in an English study.

A notable strength of our study lies in the specific nature of the quarantine hospital set-up: because HCWs resided continuously in the hospitals over their entire working shifts, we were able to exclude risk of infection in the community and specifically quantify the nosocomial risk for HCWs. Conversely, results reported from non-quarantine hospitals worldwide are generally unable to distinguish between the nosocomial vs. community risk of infection. However, as most studies conducted in healthcare settings, we were unable to distinguish between patient-to-HCW and HCW-to-HCW routes of transmission. A previous study on several nosocomial Covid-19 healthcare-associated outbreaks in Germany reported that HCW-to-HCW transmission could represent an outsized risk as compared to the one due to infected patients [19]. Still, assessing the relative contribution between patient-to-HCW versus HCW-to-HCW SARS-Cov-2 transmission was left for future work. However, it is worth noting that during the outbreaks observed in our study, infections occurred in different care units, which may be indicative of HCW-to-HCW transmission, rather than simultaneous independent events of patient-to-HCW transmission. In early stages of the pandemic, personal protective equipment (PPE) placed a focus on the risk induced by patients, at times underestimating the risk of infection from –infected and undiagnosed– colleagues. Indeed, a modeling study based on a non-quarantine hospital setting found that the contribution of HCW-to-HCW transmission may highly exceed that of patient-to-HCW [20]. This may be especially true for quarantine hospitals in which HCWs share resting and conviviality rooms for longer times than in standard care settings. Adapting the use of PPE and social distancing between colleagues while ensuring sufficient social interaction and support to maintain HCWs’ mental health thus constitutes a specific challenge for quarantine hospitals [16].

In conclusion, the key role played by healthcare settings during pandemic waves since 2020 worldwide [21] highlights the need for innovative strategies to control the nosocomial risk. Quarantine hospitals, which effectively isolated non-COVID patients from COVID patients and allowed to closely monitor at-risk HCWs, may provide interesting solutions to this challenge. Our results suggest that, with sufficient anticipation and infection control measures, the risk faced by HCWs working in such quarantine hospitals can be brought down to levels similar to those observed in standard COVID-19 care settings. However, a comprehensive assessment of quarantine hospital care models should also include their impact on HCWs’ mental health as well as the potential benefits of earlier infection diagnosis, which is likely to reduce further hospital, household and community transmission [16].

Methods

Study settings

Data was collected within three Egyptian hospitals located in Cairo (Hosp1 and Hosp3), and Fayoum (middle Egypt, Hosp2), that were temporarily transformed into quarantine hospitals during the first COVID-19 wave. During the quarantine-organization period (Hosp1: March 14th to August 1st, 2020; Hosp2: April 1st to July 31th; Hosp3: June 6th to July 11th, 2020, a 4-fold shorter period), only externally-referred COVID-19 confirmed patients were admitted to these hospitals for medical care. Multidisciplinary medical teams fully dedicated to COVID-19 patient management worked in total isolation, organized by shifts (Hosp1: 2-week shifts; Hosp2: 1 to 2-week shifts; Hosp3: 1-week shifts). During their shift, HCWs were assigned either to ICU or non-ICU units. HCWs were screened for SARS-CoV-2 infection before starting a working shift using rapid serological IgM/IgG antibody tests (Artron laboratories Burnaby, Canada; sensitivity: 83.3%, specificity: 100% [22]). Only HCWs with no SARS-CoV-2 antibodies were allowed to start working in the hospital; except for the 5th and last shift in Hosp3, where staff recruitment relied exclusively on HCWs who had been previously infected (positive serological tests), and were additionally tested using RT-PCR tests on nasopharyngeal swabs, obtaining negative results. During the working shifts, HCWs were tested for SARS-CoV-2 infection using RT-PCR tests: i) routinely at the end of the shift, ii) upon symptoms, and iii) in case of outbreak suspicion (> 2 positive tests among HCWs). HCWs testing negative at the end of the shift were then released for self-isolation at home for two weeks. HCWs testing positive before or during their shift self-isolated at home in the case of presenting no or mild symptoms, or were admitted to the same quarantine hospital for medical care in the case of presenting moderate to severe symptoms.

Observed risk of SARS-CoV-2 infection

Crude incidence rates were obtained from the number of incident SARS-CoV-2 infections observed (i.e., diagnosed during or at the end of a shift) among HCWs screened as seronegative for SARS-CoV-2 before their working shift. Incidence rate ratios were obtained through a Poisson regression adjusted by hospital and by type of care unit (ICU/ non-ICU). In addition, we computed the attack rates over each working shift, based on the observed infections among HCWs having a negative screening test at the beginning of the working shift. Observed per-shift attack rates of $\ge 20\%$ were considered to be healthcare-associated outbreaks among HCWs.

Model-based estimates of the incidence rate of SARS-CoV-2 infection

An important feature arising from COVID-19 surveillance within quarantine hospitals is potential right-truncation of data: HCWs infected a short time before the end of a shift are likely to remain undiagnosed despite the systematic testing at the end of shifts, especially if the test is performed early in the incubation period. We thus developed the following mathematical model to estimate the incidence rate of both diagnosed and undiagnosed SARS-CoV-2 infections.

We simulated the daily number of incident infections among HCWs, ${X}_{k,i}^{A}$ based on an unobserved binomial process:

where $k$denotes the working shift, $i$denotes the day within the shift, $A$is the hospital unit of HCWs’ assignments ($A\in \left\{\text{ICU}\text{, }\text{Non-ICU}\right\}$),${S}_{k,i}^{A}$denotes the number of susceptible HCWs assigned to A, at the beginning of the $i$-th day of shift $k$and $\lambda$ denotes the daily probability of SARS-CoV-2 infection faced by HCWs, assumed to be constant over the whole study period in each hospital unit.

The number of susceptible HCWs was updated on a daily basis:

$${S}_{k,i+1}^{A}={S}_{k,i}^{A}-{X}_{k,i}^{A}$$

To model right-truncation in the surveillance data, we simulated an observation process representing the systematic RT-PCR testing of HCWs at the end of each shift, accounting for documented variation in test sensitivity as a function of time since infection [23]. The period of time since infection was computed as ${j}_{k}-i,$where ${j}_{k}$ denotes the day at the end of shift $\text{k}$ (i.e., the day when testing was performed) and $i$denotes the day of infection. Then, the daily number of incident infections eventually diagnosed at the end of the $k$-th shift was given by

$${I}_{k,i}^{A}={X}_{k,i}^{A}{\theta }_{{j}_{k}-i}$$

where ${\theta }_{{j}_{k}-i}$ denotes the probability of testing positive to RT-PCR, ${j}_{k}-i$ days after infection.

We used a Bayesian Markov Chain Monte-Carlo (MCMC) approach to estimate the parameter $\lambda$ that best fitted the observed number of infections. We considered a non-informative uniform prior distribution for $\lambda$ and the following binomial likelihood function:

$$L\left(\lambda |\text{data}\right)=\prod _{A}\left(\begin{array}{c}\sum _{k,i}{N}_{k,i}^{S,A}\\ \sum _{k,i}{N}_{k,i}^{I,A}\end{array}\right){\left(\frac{\sum _{k,i}{I}_{k,i}^{A}}{\sum _{k,i}{S}_{k,i}^{A}}\right)}^{\sum _{k,i}{N}_{k,i}^{I,A}}{\left(1-\frac{\sum _{k,i}{I}_{k,i}^{A}}{\sum _{k,i}{S}_{k,i}^{A}}\right)}^{\sum _{k,i}{N}_{k,i}^{S,A}-\sum _{k,i}{N}_{k,i}^{I,A}},$$

where $\sum _{k,i}{N}_{k,i}^{S,A}$ and $\sum _{k,i}{N}_{k,i}^{I,A}$ denote the total number of susceptible HCWs (denoted by the superscript $S$) and the total number of HCW infected and diagnosed with COVID-19 (denoted by the superscript $\text{I}$), observed in hospital unit $A$, during the study period, respectively. The model was thus fitted to the total number of observed infections in each hospital unit.

Medians and 95% credibility intervals (CrI) for $\lambda$were computed from posterior samples obtained after 10,000 model runs. The chains were visually inspected for convergence (Supplementary Fig. 1). For each hospital, we further estimated the probability for a susceptible HCW to be infected at the end of a shift as

$$P\left(\text{Infected at shift end}\right)=1-{\left(1-\lambda \right)}^{d}$$

where$d$ is the length of the working shift, in days.

Finally, to assess the proportion of undetected infections among HCWs, we ran our model on each hospital, using the posterior distributions for $\lambda$, to obtain an estimate for the real number of SARS-CoV-2 infections in each hospital (both diagnosed and undiagnosed), over the study period.

The numeric implementation of the model was coded in R 4.0.3 [24], using the FME package [25] for parameter estimation. The code is available at https://github.com/sjijon/SARS-CoV-2_QuarantineHospitals.

Declarations

Ethical statements

This study was approved by the Research Ethics Committee (REC) of the Central Directorate of Research and Health Development and Reviews at the Egyptian Ministry of Health and Population (Serial: 25-2020/16), and by REC for human subject research at the Faculty of Medicine, Helwan University (Serial: 50-2020). All methods were performed in accordance with the relevant guidelines and regulations.

In addition, all participant healthcare workers provided informed consent during enrolment.

Acknowledgments

The authors thank all the dedicated HCWs who contributed to this study.

Author contributions

SJ conducted the study, built and published the numerical applications, SJ and KJ conducted the statistical and mathematical analyses, SJ, LT and KJ conceived the mathematical model, AAS, EH and MEK collected, managed and interpreted the data, KJ, MEK and LT conceived and designed the study. SJ wrote the first draft of the article. All authors reviewed and approved the final version of the article.

Data availability statement

Data (in aggregate form) and codes are available at: https://github.com/sjijon/SARS-CoV-2_QuarantineHospitals.

Competing interests

The authors declare no conflicts of interest.

Funding

This study was supported by INSERM-ANRS (France Recherche Nord and Sud Sida-HIV Hépatites, grant number ANRS-COV-19).

EMAE-MESuRS working group on nosocomial SARS-CoV-2 modeling

Audrey Duval^6,7,8, Kenza Hamzi¹, Niels Hendrickx⁶, Kévin Jean^1,2,5, Sofía Jijón^1,2, Ajmal Oodally^6,7,1, Lulla Opatowski^6,7, George Shirreff^6,7,8, David RM Smith^6,7,1, Cynthia Tamandjou⁷ and Laura Temime^1,2

⁶Institut Pasteur, Epidemiology and Modelling of Antibiotic Evasion (EMAE), Paris, France

⁷Université Paris-Saclay, UVSQ, Inserm, CESP, Anti-infective evasion and pharmacoepidemiology team, Montigny-Le-Bretonneux, France

⁸IAME, UMR 1137, Université Paris 13, Sorbonne Paris Cité, France

References

Lai X et al. 2020 Coronavirus Disease 2019 (COVID-2019) Infection Among Health Care Workers and Implications for Prevention Measures in a Tertiary Hospital in Wuhan, China. JAMA Netw. Open 3, e209666–e209666. (doi:10.1001/jamanetworkopen.2020.9666)
Gilbert M et al. 2020 Preparedness and vulnerability of African countries against importations of COVID-19: a modelling study. The Lancet 395, 871–877. (doi:10.1016/S0140-6736(20)30411-6)
Salyer SJ et al. 2021 The first and second waves of the COVID-19 pandemic in Africa: a cross-sectional study. The Lancet 397, 1265–1275. (doi:10.1016/S0140-6736(21)00632-2)
WHO. 2020 Coronavirus Disease (COVID-19) Dashboard. Egypt. See https://covid19.who.int/region/emro/country/eg (accessed on 24 May 2022).
Buitrago-Garcia D, Egli-Gany D, Counotte MJ, Hossmann S, Imeri H, Ipekci AM, Salanti G, Low N. 2020 Occurrence and transmission potential of asymptomatic and presymptomatic SARS-CoV-2 infections: A living systematic review and meta-analysis. PLOS Med. 17, e1003346. (doi:10.1371/journal.pmed.1003346)
Yanes-Lane M, Winters N, Fregonese F, Bastos M, Perlman-Arrow S, Campbell JR, Menzies D. 2020 Proportion of asymptomatic infection among COVID-19 positive persons and their transmission potential: A systematic review and meta-analysis. PLOS ONE 15, e0241536–e0241536. (doi:10.1371/journal.pone.0241536)
Medhat MA, El Kassas M. 2020 COVID-19 in Egypt: Uncovered figures or a different situation? J. Glob. Health 10. (doi:10.7189/jogh.10.010368)
Sánchez-Taltavull D, Castelo-Szekely V, Candinas D, Roldán E, Beldi G. 2021 Modelling strategies to organize healthcare workforce during pandemics: Application to COVID-19. J. Theor. Biol. 523, 110718. (doi:10.1016/j.jtbi.2021.110718)
Kassem AM et al. 2020 SARS-CoV-2 infection among healthcare workers of a gastroenterological service in a tertiary care facility. Arab J. Gastroenterol. 21, 151–155. (doi:10.1016/j.ajg.2020.07.005)
Abdelmoniem R et al. 2021 SARS-CoV-2 infection among asymptomatic healthcare workers of the emergency department in a tertiary care facility. J. Clin. Virol. 134, 104710–104710. (doi:10.1016/j.jcv.2020.104710)
Mostafa A et al. 2020 Universal COVID-19 screening of 4040 health care workers in a resource-limited setting: an Egyptian pilot model in a university with 12 public hospitals and medical centers. Int. J. Epidemiol. , 1–12. (doi:10.1093/ije/dyaa173)
Egypt Independent. 2020 Egypt reopens 21 coronavirus quarantine hospitals. See https://www.egyptindependent.com/egypt-reopens-21-coronavirus-quarantine-hospitals/ (accessed on 24 May 2022).
Zhao J et al. 2020 Antibody Responses to SARS-CoV-2 in Patients With Novel Coronavirus Disease 2019. Clin. Infect. Dis. 71, 2027–2034. (doi:10.1093/cid/ciaa344)
Jiang J-C, Zhang Y. 2020 Serological antibody testing in the COVID-19 pandemic: their molecular basis and applications. Biochem. Soc. Trans. 48, 2851–2863. (doi:10.1042/BST20200744)
Krastinova E et al. 2021 Household transmission and incidence of positive SARS-CoV-2 RT-PCR in symptomatic healthcare workers, clinical course and outcome: a French hospital experience. Occup. Environ. Med. 78, 479–485. (doi:10.1136/oemed-2020-106866)
Youssef N, Mostafa A, Ezzat R, Yosef M, El Kassas M. 2020 Mental health status of health-care professionals working in quarantine and non-quarantine Egyptian hospitals during the COVID-19 pandemic. East. Mediterr. Health J. 26, 1155–1164. (doi:10.26719/emhj.20.116)
Houlihan CF et al. 2020 Pandemic peak SARS-CoV-2 infection and seroconversion rates in London frontline health-care workers. The Lancet 396, e6–e7. (doi:10.1016/S0140-6736(20)31484-7)
Eyre DW et al. 2020 Differential occupational risks to healthcare workers from SARS-CoV-2 observed during a prospective observational study. eLife 9, e60675. (doi:10.7554/eLife.60675)
Schneider S, Piening B, Nouri-Pasovsky PA, Krüger AC, Gastmeier P, Aghdassi SJS. 2020 SARS-Coronavirus-2 cases in healthcare workers may not regularly originate from patient care: lessons from a university hospital on the underestimated risk of healthcare worker to healthcare worker transmission. Antimicrob. Resist. Infect. Control 9, 192. (doi:10.1186/s13756-020-00848-w)
Huang Q et al. 2021 SARS-CoV-2 transmission and control in a hospital setting: an individual-based modelling study. R. Soc. Open Sci. 8, rsos.201895, 201895. (doi:10.1098/rsos.201895)
Cooper B et al. 2021 The burden and dynamics of hospital-acquired SARS-CoV-2 in England. (doi:10.21203/rs.3.rs-1098214/v1)
Lassaunière R, Frische A, Harboe ZB, Nielsen ACY, Fomsgaard A, Krogfelt KA, Jørgensen CS. 2020 Evaluation of nine commercial SARS-CoV-2 immunoassays. medRxiv (doi:10.1101/2020.04.09.20056325)
Kucirka LM, Lauer SA, Laeyendecker O, Boon D, Lessler J. 2020 Variation in False-Negative Rate of Reverse Transcriptase Polymerase Chain Reaction-Based SARS-CoV-2 Tests by Time Since Exposure. Ann. Intern. Med. 173, 262–267. (doi:10.7326/M20-1495)
R Core Team. 2020 R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. See https://www.r-project.org/.
Soetaert K, Petzoldt T. 2010 Inverse Modelling, Sensitivity and Monte Carlo Analysis in R Using Package FME. J. Stat. Softw. 33. (doi:10.18637/jss.v033.i03)
Mostafa A et al. 2021 SARS-CoV-2 seroconversion among 4040 Egyptian healthcare workers in 12 resource-limited healthcare facilities: A prospective cohort study. Int. J. Infect. Dis. 104, 534–542. (doi:10.1016/j.ijid.2021.01.037)