Lessons learned from the investigation of a COVID-19 cluster in Creil, France: effectiveness of targeting symptomatic cases and conducting contact tracing around them

SARS-CoV-2 is a novel coronavirus that emerged in 2019 in Wuhan (China) and spread rapidly to become a pandemic via airborne, droplet, and hand-to-hand transmission [1]. It is responsible for an upper and/or lower acute respiratory infection of infected cases, ranging from mild and/or nonspecific to severe coronavirus disease called COVID-19 [1‐3]. The first European cases were detected in France in January 2020 [4]. The French outbreak response included a three-stage strategy, with the first two stages being the early detection and isolation of every imported and then locally-acquired case, with rigorous contact tracing to contain local transmission. When the extent of the epidemic made the first two stages unmanageable, we moved on to the third stage, which focused strictly on the management of the most severe cases in healthcare facilities with an increase in hospital resources. The third stage also included a national lockdown starting on March 17, 2020 to slow down the reproduction rate and the geographic progression of the virus [5, 6]. After a remission stage, incidence has been significantly increasing in France and in other European countries, especially with the spread of SARS-CoV-2 variants [7]. Knowledge on the durability of the specific immune response following infection with coronavirus as SARS-CoV-2 currently remains uncertain [8, 9]. Furthermore, the possible ability of variants to escape post-infection or vaccine-induced immunity is concerning [10]. That is why we should not rely solely on the expectation of reaching a supposed minimal natural or vaccinal herd immunity to control the spread of the virus. Indeed, in a non- or partially immune population, a single infected person could restart an epidemic. In that hypothesis, the virus is at high risk for local re-circulation or re-importation if the same comprehensive, strict strategy is not implemented worldwide, and this could last several months or years [11]. Thus, management of emergences, clusters, outbreaks and contact tracing around them will still be part of the strategy. Feedback from recent experiences can improve the future response to the re-circulation of SARS-CoV-2.

In February 2020, a local COVID-19 cluster occurred in the French department of Oise [12]. It was the first cluster in France not linked to an imported case. Nested within this cluster, the transmission chain spread into a military support facility located at the air base in the city of Creil (MSFAC). We present here the investigation and management of this COVID-19 cluster. The main objective is to show that prioritizing symptomatic COVID-19 diagnosis and contact tracing led to successful containment.

The investigation design combined active case finding and contact tracing around each case: 1/ Anyone from MSFAC with any symptoms was asked to report to the military health facility to undergo a RT-PCR test; 2/ Contact persons with any evocative symptoms had a RT-PCR test. Once the perimeter of the cluster was well defined (as of March 1), we considered the entire MSFAC staff as contacts and extended the RT-PCR indication to any symptomatic patients from MSFAC regardless of their contact history.

Confirmed cases were patients with positive RT-PCR test results and/or positive serology. Contact persons were persons with moderate to high risk of exposure to SARS-CoV-2, i.e. at-risk contact with a confirmed case according to French national recommendations (same household, room, team, etc., or about 15 min face to face < 1 m). Suspected cases were excluded after one or two negative RT-PCR tests (when the first sample was collected within 48 h of symptom onset or in case of real clinical suspicion).

A field sampling unit was set up in the military health facility to support the epidemiological investigation. It respected technical, biosafety, and biocleaning conditions and could perform 16–20 medical evaluations and test samples a day. Two swabs (nasopharyngeal and oropharyngeal) per case were sampled (Sigma Virocult, Medical Wire Instrument, Corsham) and pooled, stored at + 4 °C, and sent under biosafety conditions to the microbiology facility near Paris. We used the automated EZ1 XL (Qiagen France SAS, Courtaboeuf) for RNA extraction, following the manufacturers’ instructions, and a LightCycler 480 System (Roche Diagnostics, Meylan, France) for SARS-CoV-2 RT-PCR. Primers and probe sequences corresponded to the RdRp-IP2 and RdRp-IP4 assay designed at the CNR, and they provided positive SARS-CoV-2 control. For each specimen, the quality of the initial sampling, quality of RNA extraction, and the absence of PCR inhibitors were checked by two other PCRs, using a cellular control (CELL control R-gene, Argene, Biomérieux, France) and an internal control (RICO Extra-R-gene, Argene, Biomérieux, France). We sent positive samples to the CNR for sequencing of viral genomes to contribute to molecular epidemiological studies and, in particular, to investigate the link with a community cluster in Oise. We also collected the results of RT-PCR tests when they were performed in other laboratories.

As soon as serology assays were available, 1 month after this outbreak, we offered a serological test to all MSFAC staff to search for recent contact with SARS-CoV-2 and asymptomatic SARS-CoV-2-infected patients. Sera were tested by SARS-CoV-2 IgG and IgA assay (Euroimmun AG, Lübeck, Germany) on automated microtiter plate analyzers, Etimax (Diasorin SA, Antony, France) or Elispeed (Euroimmun). Positive sera for IgG or IgA with Euroimmun assay were controlled with Elecsys Anti-SARS-CoV-2 immunoassay (Roche Diagnostics, Meylan, France) on Cobas 6000. An in-house seroneutralization assay was performed [13]. If a discrepancy was found, a new serological test was proposed 2 weeks later.

Viral genome sequencing was attempted with a highly multiplexed PCR amplicon approach [14] using the ARTIC Network multiplex PCR primers set v1 (https://​artic.​network/​ncov-2019), with modification as suggested in previous research [15]. Synthesized cDNA was used as a template, and amplicons were generated using two pooled primer mixtures for 35 rounds of amplification. Libraries were then prepared using the Nextera XT DNA Library Prep Kit (Illumina) and sequenced on an Illumina NextSeq500 (2 × 150 cycles).

Raw reads were trimmed using Trimmomatic v0.36 [16] to remove Illumina adaptors and low quality reads as well as primer sequences corresponding to the PCR amplicons. We performed iterative mapping against the reference genome Wuhan/Hu-1/2019 (NCBI Nucleotide – NC_045512, GenBank – MN908947) and then on the extracted consensus using the CLC Genomics Suite v5.1.0 (QIAGEN). We used SAMtools v1.3 to sort the aligned bam files and generate alignment statistics [17]. Aligned reads were manually inspected using Geneious prime v2020.1.2 (2020) (https://​www.​geneious.​com/​), and consensus sequences were generated using a minimum of 3X read-depth coverage to make a base call. No genomic deletions were detected in the genomes analyzed.

In addition, SARS-CoV-2 RT-PCR tests were performed on environmental swabs sampled from several common objects or surfaces inside the MSFAC building on March 1, before disinfection.

All confirmed cases were isolated for 14 days and medically monitored daily at home or at the hospital, according to symptom severity, individual risk of worsening illness, and proximity with persons in their household at risk for severe COVID-19 (vulnerable individuals). Contact tracing was performed remotely by phone. Cases were interviewed at length about symptoms and date of onset and about their activities and contacts in the 14 days prior to symptom onset, to determine the most probable source of the contamination, and in the last 48 h to identify at-risk contact persons. The interviewers used a standardized questionnaire.

After the investigation, strict epidemiological surveillance of COVID-19 was maintained to detect any new cases and respond rapidly. We used R software to calculate 95% confidence intervals.

This field investigation on patients infected with SARS-CoV-2 made it possible to observe the true reality of a SARS-CoV-2 epidemic and provided a good understanding of how to manage clusters or outbreaks. Such results also enable theoretical modeling to be validated or adjusted. Despite a lack of resources at that time (masks, testing), taking adequate and well-adapted measures stopped viral transmission in a few days. However, the results of this investigation should be confronted with those of other observational studies.

All cases in MSFAC were linked to each other within a single chain of transmission. The epidemiological investigation and viral genome sequencings were concordant and compatible with an introduction originating from a previous local circulation in Oise, from the Crépy-en-Valois cluster, which is described elsewhere by Fontanet et al. [12].

Airborne, droplet, and hand-to-hand transmissions were suspected, as 23 cases worked at the same place or together, and an environmental sample was positive for SARS-CoV-2 (despite the late collection date). Viral density in a confined area might have increased the reproductive rate, and viral inoculum at the time of infection could play a crucial role in the expression of the disease and the occurrence of severe cases (2/27, 7% in this cluster), [18].

Aside from ageusia and anosmia, which were not described at the start of the epidemic, clinical signs of SARS-CoV-2 infection are not specific, and COVID-19 mimicked co-circulating flu or colds. The first case identified (Case 7), who was not the index case but the starting point of this alert, was tested 14 days after the onset of symptoms, and 8 days after he deteriorated clinically, as he did not meet the epidemiological criteria of a possible case at that date. A lesson learned from this episode is that epidemiological case definitions are useful to standardize counting of cases everywhere, but they must not replace clinical diagnosis, especially during the emergence of new pathogens, because the infection and symptoms they cause are not well known and have not yet been described. When Case 7 was finally diagnosed, other primary, secondary, tertiary, and quaternary generations of cases had already been infected. The entire chain of transmission had to be identified backwards and forwards as early as possible to prevent an uncontrolled amplification of the outbreak [19, 20].

Deployment of molecular diagnostic tests was crucial. Due to a shortage of swabs and viral transport medium, our testing capability was limited (as was the case throughout France at that time). We defined the cluster area based on active case finding and contact tracing results and tested all symptomatic individuals within that perimeter. This testing strategy, based on medical evaluation of at-risk exposure, symptoms, and RT-PCR quantitative results, made it possible to limit false positives (and inappropriate lockdowns or closings) [21] and false negatives (we collected a second RT-PCR sample if any doubt existed). The effective use of targeted testing according to pre-test probability seems to be a key point in the successful management of outbreaks.

Regarding the possibility of missing asymptomatic cases, serology results confirmed the low percentage of patients who remained asymptomatic (13%), and contact tracing results did not identify any transmission from asymptomatic to symptomatic cases in this cluster. In a meta-analysis, Madewell et al. compared the secondary attack rate around symptomatic (18%) and asymptomatic COVID-19 cases (0.7%, p < 0.001) [22]. Nowadays, questions are being raised about the real role of asymptomatic infected people in the spread of the disease [23‐27]. Among SARS-CoV-2 positive individuals, there is a confusion between those who are asymptomatic (= no symptoms at all), which is the case for 20% of COVID-19 patients, CI 95%, 17–25 according to the review of Buitrago-Garcia et al. [28], and those who are pre-symptomatic (= no symptoms at the time of sample collection for RT-PCR) [3, 29]. Indeed, Case 2 probably contaminated the five following cases during her pre-symptomatic stage, and 10 pre-symptomatic contaminations (about 40%) occurred in this cluster, highlighting the need for contact tracing and identifying, testing, quarantining and/or mask wearing for asymptomatic at-risk contact persons [30]. This is a major difference with SARS-CoV-1, where infectiousness started after symptom onset and was proportional to the clinical expression, making it easier to control [31].

Once the cluster area was defined, we placed the MSFAC staff under lockdown to slow down transmission within military settings during the investigation and contact tracing period. Ideally, increasing the use of digital tools and staff for timely contact tracing and early identification of the entire chain of transmission could reduce the duration of a lockdown [20, 32]. Six symptomatic cases (22% of the cases) did not go to the military health facility to be diagnosed and isolated, but no secondary cases occurred around them, probably due to the lockdown. Symptomatic cases that are not isolated are responsible for spreading the outbreak and underline the importance of raising awareness about COVID-19 symptoms within the population [33]. They also justify systematic mask wearing by at-risk populations [34].

Regarding the air base outside the cluster, since no significant viral circulation was identified, a lockdown was not required, but staff turnover, physical distancing, and hand hygiene measures were strongly implemented, without halting all activities.

All cases with a positive RT-PCR result and who underwent serological testing had a positive anti-SARS-CoV-2 serology with neutralizing antibodies, which confirmed a specific immunity against SARS-CoV-2, even if its duration and protective effect against further re-infection with variants are still unknown [10]. A longer follow-up of these patients would be necessary to conclusively establish that immunity.

A rigorous investigation strategy based on systematic testing of at-risk symptomatic patients, with isolation of cases and at-risk contact persons, made it possible to quickly stop the spread of this COVID-19 cluster.