Contact Settings and Risk for Transmission in 3410 Close Contacts of Patients With COVID-19 in Guangzhou, China: A Prospective Cohort StudyFREE

Index cases were identified through surveillance testing, screening individuals with symptoms who presented to a health care facility, or tracing and screening close contacts (Part 1 of the Supplement). During the COVID-19 outbreak period, health care facilities at all levels and of all types in China were obliged to identify COVID-19 cases and report them immediately via the online direct reporting system to Centers for Disease Control and Prevention (CDCs) in China. Upon receiving a report, the CDCs reviewed cases within 2 hours in the online direct reporting system. Once suspected cases (symptomatic, but not yet confirmed by laboratory testing), confirmed cases (symptomatic cases), and asymptomatic infected persons (asymptomatic cases) were verified and reported, the county or district CDCs completed an epidemiologic investigation of these cases within 24 hours, including contact tracing (13, 14). Close contacts are individuals who have had contact, without effective protection regardless of duration of exposure, with 1 or more persons with suspected or confirmed COVID-19 any time starting 2 days before onset of symptoms in persons with a suspected or confirmed case, or 2 days before sampling for laboratory testing of asymptomatic infected persons (14).

For close contact tracing on public transportation, a cell phone database based on the movements of the users was developed in China (12). By measuring and recording proximity events between individuals, it can immediately trace close contacts of diagnosed cases upon case confirmation (15). In this way, almost all of the close contacts of public transportation can be traced. A small proportion of close contacts was difficult to trace owing to lack of cell phone data. Detailed information on definitions and close contact tracing is provided in Part 2 of the Supplement.

From 28 December 2019 to 5 March 2020, 3410 close contacts of 391 index cases (244 in Guangzhou and 147 in other regions) were identified and followed up by the Guangzhou CDC (GZCDC) from 13 January 2020 to 6 March 2020. The close contacts included in this study are the close contacts living in Guangzhou of these 391 index cases during the study period and do not include additional close contacts managed by other regional CDCs in China. Figure 1 shows the flowchart of COVID-19 cases and close contact management.

Our study was based on the data from the work for an ongoing public health response to COVID-19 by GZCDC as required by the National Health Commission of China, and hence individual informed consent was waived. The study was determined not to be human subjects research and therefore was considered exempt from ethical approval after consultation with the ethics committee of GZCDC. Analytical data sets were constructed in an anonymized manner, and all analysis of personally identifiable data took place onsite at the GZCDC.

The information collected for 244 index cases in Guangzhou included demographic characteristics (age and sex), severity, clinical symptoms at onset (fever, dry cough, expectoration, fatigue, myalgia, and diarrhea), radiologic examinations (chest computed tomography), and blood examinations (leukocyte count, lymphocyte percentage, and neutrophilic granulocyte percentage). Clinical data on the 147 index cases from regions other than Guangzhou were unavailable, and therefore their 800 close contacts inside Guangzhou could not be categorized by severity and symptoms of the index cases. In addition, no information on symptoms for 21 index cases in Guangzhou, and thus their 797 close contacts, could be categorized by symptoms of the index cases.

Upon identification, information about the setting of exposure to the index case was obtained and classified into 5 categories: household, public transportation, health care settings, entertainment venues or workplaces, and multiple settings. “Multiple settings” indicates exposure in more than 1 contact setting. Index cases were quarantined before diagnosis if they were suspected cases, and household contacts were separated from the index cases before diagnosis was made in the index cases.

After identification, close contacts were quarantined for 14 days from last contact with index cases and followed up clinically after quarantine until 6 April 2020. For example, if an individual was traced on the fourth day after the last unprotected contact, this individual was required to be medically observed for 10 days. The quarantine period was extended for 337 close contacts for whom real-time fluorescence-based reverse transcriptase polymerase chain reaction (RT-PCR) testing was delayed.

During the quarantine period, monitoring of clinical symptoms and RT-PCR assay of SARS-CoV-2 nucleic acid was performed approximately every 24 hours. Specimens for RT-PCR testing were obtained from the upper (nasopharyngeal and oropharyngeal swabs) and lower (sputum) respiratory tracts by trained CDC staff and were sent to GZCDC for testing. The RT-PCR testing was performed by qualified staff, and results were identified through open reading frame 1ab (ORF1ab) and nucleocapsid protein (N) in accordance with the protocol established by the Chinese CDC (16). Details of the laboratory processes are provided in Part 3 of the Supplement. For close contacts who tested positive on RT-PCR, radiologic and blood examinations were conducted; these included chest radiography or computed tomography, complete blood count, coagulation profile, serum biochemical tests (renal and liver function, creatine kinase, lactate dehydrogenase, and electrolytes), myocardial enzymes, interleukin-6, serum ferritin, and procalcitonin (17).

The symptoms that were monitored mainly included fever and respiratory symptoms (for example, dry cough and expectoration). Monitoring was performed by the staff of the medical observation institution during daily in-person visits (14).

Close contacts who tested negative on RT-PCR and remained symptom-free were released after the quarantine period. Those who developed symptoms but tested negative throughout the quarantine period on RT-PCR received radiologic examinations and serologic COVID-19–specific IgM and IgG examinations. If the examinations were all negative, they were released from quarantine and followed for the development of symptoms by staff in their communities via weekly in-person visits through 6 April 2020. Close contacts who developed symptoms during the follow-up period after release from quarantine were retested.

The main outcome was COVID-19 infection in close contacts during the follow-up period. These included asymptomatic cases and symptomatic cases, which were defined on the basis of guidelines issued by the National Health Commission of the People's Republic of China (13, 17). Detailed definitions of these 2 outcomes are provided in Part 1 of the Supplement.

In brief, an asymptomatic case (13) was defined as an individual without clinical manifestations and with etiologic detection of SARS-CoV-2 in respiratory specimens (that is, RT-PCR–positive) or specific IgM detected in serum. After a quarantine period of 14 days, asymptomatic infected persons were followed up until 6 April 2020 to confirm whether symptoms occurred. A symptomatic case (13) was defined as a suspected case on the basis of symptoms and epidemiologic history (Part 1 of the Supplement) and 1 of the following etiologic or serologic findings: 1) a positive result for SARS-CoV-2 nucleic acid on RT-PCR; 2) a viral gene identified by gene sequencing that was highly homologous with known SARS-CoV-2; or 3) detectable SARS-CoV-2–specific IgM and IgG in serum, or at least a 4-fold increase in IgG between paired acute and convalescent sera. Symptomatic cases were categorized as mild, moderate, severe, or critical (17) (Part 1 of the Supplement).

To show the difference between identifying cases through presenting symptoms versus intensive surveillance of a close contact, we attempted to pair every secondary case with their index case and compared their clinical characteristics. If a secondary case was in contact with 2 or more index cases, we linked the secondary case to the earliest index case. Secondary cases were unlikely to have been co–index cases because they did not share the same primary infection exposure as the index case during the contact tracing episode.

The secondary attack rate of COVID-19 was estimated by dividing the number of asymptomatic or symptomatic secondary cases by the number of close contacts for that exposure setting compared with different contact exposure groups. Categorical variables were calculated as numbers (percentages). Skewed and normally distributed continuous variables were calculated as the median (interquartile range [IQR]) or mean (SD), respectively. To minimize the potential for inferential bias and to maximize the decreased statistical power due to exclusion of participants with missing covariate data, we used multiple imputation with chained equations to impute missing covariate values, and 20 data sets were imputed (18). All variables including outcomes were included in the multiple imputation model.

Generalized estimating equations (19) taking into account clustering by index cases were performed to estimate odds ratios (ORs) and 95% CIs for associations of potential factors with the risk for SARS-CoV-2 infection:

where V_i = working covariance matrix (), R_i (α) = working correlated matrix, and A_i = T-dimensional diagonal matrix.

Age (0 to 17 years, 18 to 44 years, 45 to 59 years, or ≥60 years), sex, exposure setting, severity of index cases, and symptoms in index cases (fever, dry cough, expectoration, fatigue, and myalgia) were included in the multivariable model.

All analyses were performed by using SAS software, version 9.4, for Windows (SAS Institute). Statistical tests were 2-sided, and P values less than 0.05 were considered statistically significant.

The study was funded by the Guangdong Province Higher Vocational Colleges and Schools Pearl River Scholar Funded Scheme, the National Natural Science Foundation of China, the Construction of High-level University of Guangdong, and the Zhejiang University Special Scientific Research Fund for COVID-19 Prevention and Control. The funders had no role in study design, data collection and analysis, preparation of the manuscript, or the decision to submit the manuscript for publication.

We collected data from 3410 close contacts in Guangzhou who were linked to 391 index cases (244 index cases in Guangzhou and 147 in other regions). Table 1 shows demographic characteristics of index cases, close contacts, and secondary cases. Among 3410 close contacts, 1799 (52.8%) were male, and the median age was 38.0 years (IQR, 26.0 to 51.0 years). Of these 3410 close contacts, 127 (3.7% [95% CI, 3.1% to 4.4%]) were secondarily infected; 8 (0.2% [CI, 0.1% to 0.4%]) were asymptomatic, and 119 (3.5% [CI, 2.9% to 4.1%]) were symptomatic. Of the 119 symptomatic cases, 20 (16.8%) presented with mild symptoms, 87 (73.1%) with moderate symptoms, and 12 (10.1%) with severe or critical symptoms. Of the secondary cases, the majority (105 [82.7%]) occurred via household contact (Table 2). The basic characteristics and distribution of clusters in each exposure setting are summarized in Appendix Tables 1 and 2. The distribution of clusters according to the number of index cases is shown in Figure 2.

Because of the lag in identification of close contacts, the median time from last contact to the start of quarantine was 4.0 days (IQR, 1.0 to 12.0 days); thus, the 3410 close contacts were only quarantined for a median of 10.0 days (IQR, 2.0 to 14.0 days) (Table 2; Appendix Figure 1). Furthermore, the 127 close contacts in whom asymptomatic or symptomatic cases were diagnosed ended quarantine after a median of 2.0 days (IQR, 1.0 to 5.0 days) because most of them received a diagnosis within 2 days of the start of quarantine and were transferred to designated hospitals in a timely manner. For close contacts in whom symptomatic cases were diagnosed, the median time from last contact to symptom onset was 2.0 days (IQR, 1.0 to 3.0 days) (Appendix Figure 2). All secondary cases were confirmed by RT-PCR testing, and the mean number of RT-PCR assays was 2.8 (SD, 2.4).

Table 3 shows the association between various factors and risk for SARS-CoV-2 infection among close contacts of index cases before and after adjustment. A higher percentage of females than males (4.4% [CI, 3.4% to 5.4%] vs. 3.1% [CI, 2.3% to 3.9%]) were infected, but this difference was not statistically significant on multivariate analysis. In terms of exposure setting, household contacts had a higher risk for secondary infection (10.3% [CI, 8.5% to 12.2%]) than did persons who were exposed in health care settings (1.0% [CI, 0.3% to 1.8%]; OR, 0.09 [CI, 0.04 to 0.20]) and those who were exposed on public transportation (0.1% [CI, 0.0% to 0.4%]; OR, 0.01 [CI, 0.00 to 0.08]).

The secondary attack rate increased with the severity of index cases, from 0.3% (CI, 0.0% to 1.0%) for asymptomatic to 3.3% (CI, 1.8% to 4.8%) for mild, 5.6% (CI, 4.4% to 6.8%) for moderate, and 6.2% (CI, 3.2% to 9.1%) for severe or critical cases (P for trend < 0.001). Manifestation of certain symptoms, such as fever (6.7% [CI, 5.3% to 8.0%] vs. 3.3% [CI, 1.6% to 4.9%]) and expectoration (13.6% [CI, 10.6% to 16.7%] vs. 3.0% [CI, 2.1% to 3.9%]), in the index cases was associated with an increased risk for infection in their close contacts. The frequency of contact and number of index cases contacted were not separately assessed owing to multicollinearity with household contacts; the incidence of COVID-19 by these 2 variables is shown in Appendix Table 3.

A total of 121 secondary cases were successfully linked to their 68 index cases, and 6 secondary cases were not successfully linked because their index cases were outside Guangzhou and we had no detailed information for them. Compared with their 68 index cases, the 121 secondary cases were in general clinically milder and were less likely to manifest such common symptoms as fever, cough, expectoration, fatigue, myalgia, and diarrhea (Table 4). Secondary cases were also less likely than index cases to demonstrate radiologic and laboratory alterations related to the infection (Table 4).