Basic epidemiological parameter values from data of real-world in mega-cities: the characteristics of COVID-19 in Beijing, China

Protocols for COVID-19 diagnosis and treatment, surveillance, epidemiological investigation, management of close contacts, and laboratory testing were formulated, and relevant surveillance activities and epidemiological investigations conducted. All individuals with suspected SARS-CoV-2 infection were isolated in designated hospitals for treatment. The epidemiological investigation, contact tracing, and quarantine were conducted by the Centers for Disease Prevention and Control (CDCs) at district level or municipal level. Close contacts were defined based on the protocol released by National Health Commission of the People’s Republic of China (NHCPRC). Without any efficient personal protective equipment, a person who had contact within one meter with a confirmed case after onset of symptoms was defined as close contact. The definition changed on Feb 10th in accordance with the WHO guidelines [4], from the onset to 4 days prior to symptoms, given evidence of presymptomatic transmission. Close contacts were required to be quarantined at home or in designated sites for cases who have no separate room or imported cases from overseas for medical observation till 14 days after the last exposure. Besides traditional contact tracing, big data and artificial intelligence (AI) were used to strengthen contact tracing. For enhanced surveillance to detect COVID-19 in influenza-like illness (ILI) cases, all samples from ILI cases reported in routine influenza virological surveillance system [5] were tested for SARS-CoV-2 using real-time reverse transcription-polymerase chain reaction (Real-time RT-PCR) since January 28th.

The case definitions of suspected and confirmed cases of COVID-19 were based on the protocol released by the NHCPRC [6]. Some modifications in case definitions were made according to recent studies in China. The definition modifications of suspected cases and confirmed cases are shown in Supplementary Table 1. Among all confirmed cases, mild cases were defined cases without evidence of pneumonia, and moderate cases as those with pneumonia but no oxygen therapy required. Severe cases were defined as those with dyspnea, respiratory frequency ≥ 30/min, blood oxygen saturation ≤ 93%, PaO₂/FiO₂ ratio < 300, and/or lung infiltrates > 50% within 24–48 h. Critical cases were those that exhibited respiratory failure, septic shock, and/or multiple organ dysfunction/failure.

Asymptomatic laboratory-confirmed cases were defined a person who were positive for SARS-CoV-2 while showing no clinical symptoms and signs.

The demographic and epidemiological information, clinical symptoms and outcomes were obtained using a standardized questionnaire by interviewing patients and/or their family members/relatives, attending doctors and other health care providers, supplemented by patient medical records [7]. All data were checked by two public health professionals.

Upper and lower respiratory tract specimens, urine, blood, stool, and sputum specimens were obtained from patients. For all RNA extractions, RNA was extracted from 200 μL of sample and eluted in 90 μL elution buffer by KingFisher Flex Purification System (Thermo Fisher, USA). The real-time RT-PCR assay was performed by TaqMan Fast Virus 1-Step Master Mix (ThermoFisher) in ABI 7500 fast system (ABI, USA). 0.5 μM of forward primer, 0.5 μM of reverse primer, 0.25 μM of probe, and 5 μl of RNA sample were mixed in a 25 μL monoplex quantitative RT-PCR reaction. The primer and probe were generated following the national guideline [7], and the reaction condition was set according to the manufacturer’s protocol. Cases testing positive for both target genes (open reading frame 1ab and nucleocapsid protein) were determined as laboratory-confirmed cases. Meanwhile, commercial kit (Bio-germ Inc., Shanghai, China and ABT Inc., Beijing, China), which were approved by China Food and Drug Administration (CFDA), were also used in testing. For those cases tested by a commercial kit, the results were determined following the manufacturer’s instructions. Upper and lower respiratory tract specimens from confirmed cases were used for viral genome sequencing. Next-generation sequencing (NGS) was achieved based on a metagenomics strategy followed by Illumina sequencing (Illumina Inc., San Diego, CA, USA). Output data were assembled by a viral genome-targeted assembly pipeline with a homology search E-value of 1e^− 10. Full genome sequences used in the current study are available on request. Sequence alignment was achieved by MAFFT [8]. The Neighbor-Joining (N-J) method with a bootstrap of 1000 was used for phylogenetic analysis.

Data collection and analysis of cases and close contacts were determined by the NHCPRC to be part of a continual public health outbreak investigation, and as such was granted exemption from the institutional review board.

From Jan 1st 2020 to Apr 3rd 2020, respiratory specimens from 7432 suspected cases were tested for SARS-CoV-2, of which 602 (8.1%) were positive. Among the laboratory-confirmed SARS-CoV-2 infections, 585 were confirmed case-patients with symptoms and signs and 17 were asymptomatic COVID-19 cases. The proportion of asymptomatic infection was 2.8% (95% CI: 1.5–4.2).

The epidemic curve in Beijing by date of confirmation and onset is shown in Fig. 1a. The first case was confirmed on Jan 19th. The number of infection rapidly grew to Jan 30th and peaked between Jan 31st and Feb 1st. The infections steadily declined between Jan 30th and Feb 18th. However, the number of reported cases fluctuated due to a local cluster (14 cases) that occurred in a company on Feb 25th in Beijing and imported cases from abroad.

Since the first imported case from Iran on Feb 29th, the number of cases from overseas increases rapidly. To effectively treat and manage these cases, Xiaotangshan Hospital was launched on Mar 16th. On Mar 23rd, nine cities were designated as the first entry sites for international passenger flights bound for Beijing. To detect more potential cases, all international passengers arriving in Beijing were required to be tested and quarantined at designated facilities on Mar 25th. The number of imported case from overseas in Beijing ranked first in China on Mar 29th. As of Apr 3rd, a total of 169 cases had been identified in Beijing. However, only 4 local cases were identified at the same time. And besides one local case was related to one imported case from abroad, no local cases had been reported for 27 days.

As shown in Fig. 1b, among 585 confirmed case-patients, 124 (21.2%) were imported from Wuhan, 76 (13.0%) from other regions of China, 169 (28.9%) from overseas, 1 (0.2%) was a secondary case of imported cases from overseas, 201 (34.4%) were local cases and 14 (2.4%) were under investigation. Among the 169 cases from overseas, most were from the United Kingdom (55), Spain (47), Italy (18), the United States of America (17) and France (7). The epidemic in Beijing has been undertaken in two main phases. The first phase started on Jan 19th when the first two cases were confirmed and ended on Feb 28th. The second phase started on Feb 29th when the first imported case from overseas was identified. At the beginning of the first phase (before Feb 1st), the majority of confirmed cases were imported from Wuhan (61.0%, 97/159), while the proportion of imported cases from Wuhan had decreased since Feb 1st (1 week after the lockdown of Wuhan) (16.1%, 27/168) (χ² = 70.065, P < 0.001). At the beginning of the second phase (before Mar 6th when the last local cases identified), imported cases from overseas and local cases coexisted, and then the imported cases from overseas were dominated.

Of 585 confirmed case-patients, 268(45.8%) cases were male. The male-to-female ratio was 0.9:1. The incidence of males was slightly lower than females, 2.5/100,000 and 2.9/100,000, respectively (Fig. 2). The median age of confirmed cases was 39 years old (range, 0.5 to 94; interquartile range, 27 to 56). About 19% (114/585) were 60 or above, and 7.9% (46/585) were children < 18 years old, among which 34.8% (16/46) were children under 5 years old (Table 1). A total of 17 (2.9%) cases aged 80 years or above. The incidence rate of the population 60 years old or above (6.9/100,000) was the highest compared to the other three groups, followed by the 18–59 years old group (5.6/100,000), the 5–17 years old group (0.5/100,000), and children under 5 years old (0.3/100,000).

The median age of 46 children< 18 years was 7 years old (interquartile range, 3 to 13). The sex ratio was 0.8:1. The proportion of asymptomatic infection among children< 18 years and adults was 4.2 and 2.7% (χ² = 0.343, P = 0.558). Among 585 confirmed case-patients, 17 cases (2.9%) were healthcare workers (HCW). Epidemiological investigations suggested that 7 cases were infected due to health care activities and the remaining 10 were infected due to close contact with household cases rather than in a health care setting according to data from epidemiological investigation.

Of all confirmed case-patients, 36.3% were mild, 49.7% were moderate, 11.3% (64/585) were severe cases and 2.7% (16/585) were critically ill. The proportion of severe and critical cases decreased from 21.4% before Feb 1st to 7.2% after Feb 1st (Fig. 3a). The association between illness severity and age was shown in Fig. 3b. It was shown that illness severity aggravated with age (Supplementary Table 2, χ² = 50.576, P < 0.001). A total of 8 cases deceased (with a crude case-fatality rate of 1.4%), among which 7 deaths were elderly adults over 60 years and 1 death was a 50-year-old man. The case fatality rate for males was 1.9% (5/268) and 1.0% (3/317) for females. The case fatality rate for ≥80 age group was 29.4% (5/17). All of the deaths had comorbid conditions, of which 75.0% (6/8) had hypertension or cardiovascular disease. Among 46 children < 18 years, 28 (60.9%) were mild, 17 (37.0%) were moderate and 1 (2.2%) was severe. The severe case was a three-year-old child with leukemia.

The median duration from onset of symptoms to their first medical visit was estimated to be 3 days (P_2.5-P_97.5: 0–17). The median duration from onset to case confirmation was estimated to be 5 days (P_2.5-P_97.5: 1.0–19.5).

We reviewed the records of the confirmed cases and found 37.4% (219/585) had clear history of contacts with cases prior to symptom onset; based on which we estimated that the mean incubation period was 6.3 days (95% CI: 6.0–6.6) and the median was 5.7 days (P_2.5-P_97.5: 5.2–6.1).

By Apr 3rd, a total of 4007 close contacts were quarantined, 186 were confirmed with SARS-CoV-2 infection, with an overall secondary attack rate of 4.6% (95% CI: 4.0–5.3). The secondary rate was higher among family members or relatives (15.6%, 111/714) than that among social contacts (2.2%, 75/3363) (χ² = 239.852, P < 0.001).

Among 441 close contact of HCWs, 32 were confirmed with SARS-CoV-2 infection, with an overall secondary attack rate of 7.3%, which was higher than that of non-HCWs’, with a secondary attack rate of 4.2% (154/3636) (χ² = 8.243, P = 0.004).

Till Apr 3rd, a total of 117 clusters occurred, involving 391 confirmed cases. Among 391 cases, 246 (66.3%) occurred in family, 56 from abroad (15.1%), 28 (7.6%) in health care facilities, 28 (7.6%) in public areas and 13 (3.5%) in mixed areas. Before Feb 18th, clusters mainly occurred in family and then were predominantly from abroad after Feb 28th (Fig. 4a).

There were 246 confirmed cases were involved in 91 family clusters. The median number of involved cases of family clusters was 3 (Range, 2–7; IQR, 2–3). To estimate the basic reproduction number and serial interval in families, we identified 38 family clusters where one single index case was introduced. Since the 38 index cases caused 76 secondary cases among 193 family close contacts, we estimated that R₀ in family clusters was 2.0 (95% CI: 1.6–2.4). And the average serial interval was estimated at 7.6 days (95% CI: 6.4–8.9). The median serial interval was 6.0 days (range, 0–34; IQR, 3.3–11.0) (see Fig. 4b).

Two clusters occurred in hospitals, involving 38 confirmed cases. One cluster occurred in the cardiac intensive care unit (CICU) and intensive care unit (ICU) of a general hospital in Beijing, involving 35 cases, among which 7 were HCWs. Another clusters in hospital involved 3 cases including one hospitalized patient and her two relatives providing paramedical assistants.

From Jan 28th to Apr 3rd 2020, a total of 3267 specimens were collected from 3267 individuals, and no SARS-CoV-2 viral RNA was identified.

Among all 585 confirmed case-patients, 936 specimens from 243 cases were available for analysis, including 612 pharyngeal swabs, 7 nasal swabs, 5 saliva specimens, 219 sputa specimens, 2 serum/plasma specimens, 22 urine specimens, and 69 fecal specimens. 354 pharyngeal swabs, 7 nasal swabs, 156 sputa specimens, and 28 fecal specimens showed positive results in real-time RT-PCR tests. The RNA positive rate of throat swabs in mild, ordinary, severe and critical cases was 58.73, 55.17, 62.07, and 76.47%, respectively (χ² = 4.050, P = 0.256). In contrast to these, the RNA positive rate of sputum in mild, ordinary, and severe cases was 63.08, 75.86, and 70.27%, respectively (χ² = 3.329, P = 0.189). No viral RNA was found in all 22 urine specimens, 2 serum/plasma specimens and 5 saliva specimens.

A total of five viral full genomes were obtained during the study period, including four sequences from imported cases from Whuan and one from a secondary case. Phylogenetic analysis suggested all tested viruses belonged to lineage B of the genus beta-coronavirus, and is genetically closely related to SARS-CoV-2 isolates in Wuhan (Fig. 5). It showed high genetic similarity among all tested viruses of 99.97–99.99%. It is worth noting that all five viruses carried 442 L, 472F, 479Q, 487 N, and 491Y in viral S gene receptor-binding subdomain.