During June 30–July 3, 2018, a total of 10 cases of diarrheal disease were reported at 2 hospitals in the Nanshan District of Shenzhen, China. This outbreak was suspected to be foodborne illness and was notified to the Shenzhen Center for Disease Control and Prevention (Shenzhen CDC). We collected details of food exposure histories, clinical manifestations, and demographic data through interviews with ill persons.

A total of 21 samples were collected during laboratory and environmental investigations, comprising anal swab specimens from 7 case-patients and 14 samples from the implicated restaurant (6 from chicken legs, 4 from restaurant staff, 2 from kitchenware items, and 2 from other foods). All samples were forwarded to the laboratory-based testing of common foodborne pathogens at Shenzhen CDC as previously described (7). Salmonella Enteritidis isolates were subtyped by pulsed-field gel electrophoresis (PFGE) using XbaI according to standard PulseNet protocols (https://www.cdc.gov/pulsenet/pathogens/protocols.html).

We sequenced genomes on the BGISEQ-500 platform (MGI Tech Co., Ltd., https://en.mgitech.cn) to generate 100-bp paired-end reads and deposited short-read sequence data in GenBank under BioProject PRJNA565566. We used Snippy version 4.3.8 (https://github.com/tseemann/snippy) for single-nucleotide polymorphism (SNP) calling, with Salmonella Enteritidis P125109 (GenBank accession no. NC_011294) as the reference genome. We excluded SNPs located in repetitive and recombinogenic regions before phylogenetic analysis, identified repetitive regions using TRF version 4 (https://tandem.bu.edu/trf/trf.html) and self-aligning by blastn (https://blast.ncbi.nlm.nih.gov), and determined recombinogenic regions by using Gubbins version 2.3.4 (8). We constructed a maximum-likelihood tree based on genomewide SNPs from an alignment of 300 SNPs using RAxML version 8.2.12 (9) under the general time-reversible with gamma distribution model (100 bootstraps).

All 10 case-patients were university students who had diarrhea (>3×/24 h), fever (>37.5°C), and high leukocyte counts (>10.0 × 10⁹ cells/L [reference 4.0–10.0 × 10⁹ cells/L]); 7 case-patients also reported nausea and vomiting. Case-patients were from 6 different colleges of the same university but lived in different dormitories and did not know each other. However, on the afternoon of June 30, all had eaten food delivery (chicken leg with rice) from the same restaurant near the university, ordered through an online delivery platform during a 6-hour period (noon–6 pm). Foods were precooked an hour before anticipated orders and left at room temperature and then dispatched upon receipt of orders and delivered within 1 hour in ambient temperature (29°C) using a linen storage bag.

From a total of 21 samples, 9 were positive for Salmonella Enteritidis, which was isolated from 5 chicken legs and from 4 case-patients and belonged to an indistinguishable XbaI PFGE pattern (JEGX01.SZ0001). Interrogation of existing PFGE patterns within the Shenzhen CDC PulseNet local database showed that 5 Salmonella Enteritidis isolates from sporadic cases within 1 month before the outbreak shared the same PFGE pattern. Routine surveillance further identified 5 additional sporadic isolates with the same PFGE pattern within 1 month after the outbreak. However, no clear epidemiologic links were found between any of the 10 sporadic cases and the outbreak.

Figure

Figure. Clustering of 19 outbreak and sporadic Salmonella enterica serovar Enteritidis isolates from an outbreak linked to online food delivery, Shenzhen, China, 2018. Clusters were inferred by constructing a maximum-likelihood tree based...

WGS SNP-based cluster analysis showed that all 9 outbreak-associated isolates were genetically closely related to each other (Figure). Isolates from 5 chicken legs and 4 case-patients differed by <1 SNP, confirming chicken legs as the food source of the outbreak. In comparison, the minimum distance between the 10 sporadic isolates and any of the outbreak-associated isolates was 59 SNPs, larger than the common threshold (<3 SNPs) used for delineating outbreak clusters (10), indicating that the sporadic cases were not part of the outbreak.

The burgeoning online food delivery industry has resulted in a landscape of change in food consumption behavior and lifestyle in China. In contrast to traditional restaurant dining, online food delivery could send potentially contaminated food across wide geographic areas throughout a city within a short time to cause large-scale outbreaks. Online food delivery also poses additional food safety risks, including improper handling and storage temperature during transport. As illustrated in this outbreak, the total time elapsed was 2 hours from food preparation to delivery at ambient temperature, potentially enabling Salmonella Enteritidis to sufficiently multiply and cause illnesses.

On the basis of the fine-scale delineation by whole-genome SNP-based cluster analysis, we differentiated genetically closely related Salmonella Enteritidis isolates that were indistinguishable by PFGE. WGS confirmed chicken legs as the outbreak source and excluded the possible links of additional sporadic isolates to the outbreak and highlights the advantages of using WGS for differentiating highly clonal Salmonella Enteritidis isolates. Several studies have demonstrated high epidemiologic concordance and superior discriminatory power of WGS in retrospective outbreak analyses involving Salmonella Enteritidis (10,11).

In this report of a Salmonella Enteritidis outbreak investigation linked to online food delivery guided by WGS, we highlighted the food safety challenges posed by a new mode of foodborne disease transmission. One of the new features of this outbreak is that the case-patients who ate food from the same source were isolated from each other, which contrasts with typical foodborne outbreaks that usually involved dining in restaurants or catering events within a family or group settings. Additional cases might therefore be more likely to be missed during the epidemiologic investigation. However, detailed information associated with food orders, such as the ordering and delivery time, the food items ordered, and the names and addresses of the merchant and the consumer, would all be electronically recorded over the online food delivery platform. Therefore, such remarkable level of detail would be highly valuable for prospective outbreak investigations. Close collaboration between public health agencies and online food delivery platforms would be essential to facilitate timely intervention of disease propagation and limit the scale of outbreaks efficiently.

Given the continued rapid growth anticipated for the online food delivery industry, in-depth risk assessments should be a research priority to inform appropriate food safety strategies. We explored a rare opportunity to investigate and gained new insights into the transmission dynamics of a Salmonella Enteritidis outbreak over an Internet platform, complemented with the pragmatic use of WGS analysis.