Early epidemiological analysis of the 2019-nCoV outbreak based on a crowdsourced data

Introduction

As the outbreak of novel 2019 coronavirus (2019-nCoV) is rapidly expanding in China and beyond, threatening to become a worldwide pandemic ¹, there is a need for real-time analyses of epidemiological data to increase situational awareness and inform interventions ². In the past, real-time analyses have shed light on the transmissibility, severity, and natural history of an emerging pathogen in the first few weeks of an outbreak ^3-6. Analyses of detailed line lists of patients are particularly useful to infer key epidemiologic parameters, such as the incubation and infectious periods, and lags in detection, isolation, and reporting of cases ^3,4. However, individual patient data rarely become publicly available in the first weeks of an outbreak, when the information is most needed.

Building on prior experience collating news reports to monitor Ebola transmission ⁷, here we present an effort to compile individual patient information and subnational epidemic curves on 2019-nCoV from a variety of online resources. Data were made publicly available in real-time and were used by the infectious disease modeling community to generate and compare epidemiological estimates relevant for interventions. Here we describe the process to compile data and provide an early analysis of the demographics, hospitalization, and reporting lags, in different provinces of China and internationally as of Jan 30, 2020.

Method

Results

Patients characteristics

Our line list comprised 288 cases reported between Jan 20, 2020 and Jan 29, 2020 including 200 cases from mainland China and 88 international cases (Table 1). The gender ratio was skewed towards male (63%). Within mainland China, there was a large representation of 5 of the 30 provinces, with 17% of cases reported from Hubei (capital, Wuhan), 13% from Shanxi, 8.8% from Tianjin, and 7% each from Yunnan and Beijing. Most cases had a travel history to Wuhan (42%) or where residents of Wuhan (44%), while 14% had no direct relation to Wuhan.

View this table:

• View inline
• View popup
• Download powerpoint

Table 1:

Characteristics of 2019-nCoVpatients included in the line list. Data are available here ¹⁰.

The age distribution of cases was skewed towards older age groups with a median of 44 years for cases who were alive or with unknown outcome at the time of reporting (Figure 1A). The median age of 2019-nCoVcases who had died at the time of reporting was 70 years. Few cases (n=7, 2.4%) were under 15 years of age. Adjustment for the age demographics of China confirmed a deficit of cases among children, with a RR below 0.3. in individuals under 15 years (Figure 1).The RR measure pointed at a sharp increase in the likelihood of reported 2019-nCoV infection starting at age 30 yrs.

• Download figure
• Open in new tab

Figure 1:

Age distribution of 288 2019-nCoVcases from crowdsourced data. Top: by disease outcome (alive/unknown or dead at time of reporting); vertical bars represent means. Bottom: Relative risk (RR) by 5-year age bands. The observed data is in green and the estimated RR in blue (blue dots=estimated RR, blue curve= spline-smoothed curve)

Epidemic timelines in Mainland China, nationally and at provincial level

A timeline of cases in our crowdsourced patient line list is shown by date of onset in Figure 2, indicating an acceleration of reported cases by Jan 13, 2020. The outbreak progression based on the crowdsourced patient line list is consistent with the timeline published by China CDC on Jan 29th, based on a more comprehensive database of more than 6,000 2019-nCoVcases (Figure 2). Comparison of cumulative curves shows that the crowdsourced dataset captured a larger fraction of the cases reported to China CDC before Jan 18, 2020 than in more recent days. The very recent slowing down of cumulative cases apparent in the crowdsourced and China CDC curves is likely a reflection of the delay between disease onset and reporting, estimated at a median of 5 days in our data (range 0-43 days).

• Download figure
• Open in new tab

Figure 2:

Daily timeline of the 2019-nCoVepidemic based on crowdsourced data and official sources, by location. All data are by date of symptom onset. Cumulative curves are shown for the official China CDC data (first report available on Jan 29^th, 2020, light blue), and for the crowdsourced data (dark blue). Histograms represent daily case count based on crowdsourced data for Wuhan (light green), mainland China non-Wuhan (darker green), and cases outside of mainland China (light blue).

The subnational growth of the epidemic in China is shown by reporting date in Figure 3. As of Jan 30, 2020, 14 of the 30 provinces had already reported more than 100 confirmed cases. The apparent rapid growth of newly reported cases in many of the provinces outside of Hubei is consistent with sustained local transmission.

• Download figure
• Open in new tab

Figure 3:

Daily timeline of the nCoV epidemic at provincial level, China, as of 01/30/2020. Curves are by date of reporting and sorted by total number of reported cases. The timeline for each province is reconstructed based on daily outbreak situation reports provided by provincial health authorities and posted on DXY.cn.

Hospitalization and reporting delays in mainland China

The median delay between symptom onset and seeking care at a hospital or clinic was 3 days in mainland China (IQR 0-15 days) (Figure 4). This delay declined from 5 to 2 days before and after Jan 18, 2020 (Wilcoxon test, p=0.03). Some provinces, such as Tianjin and Yunnan had shorter delays, while the early cases from Wuhan were characterized by longer seeking care delays.

• Download figure
• Open in new tab

Figure 4:

Lags between symptoms onset and hospitalization of 2019-nCoVcases, and between hospitalization and reporting, by location.

The median delay between hospitalization and reporting was 3 days (IQR 2-5 days) in mainland China and declined from 9 to 2 days after Jan 18th, 2020 (Wilcoxon test, P<0.0001, Figure 4). Similar to hospitalization delays, reporting was quickest in Tianjin and Yunnan and slowest in Hubei province.

Discussion

Information from patient line-lists is crucial but difficult to obtain at the beginning of an outbreak. Here we have shown that careful compilation of crowdsourced reports curated by a long-standing Chinese clinician network provides a valuable picture of the outbreak in real-time. The outbreak timeline is consistent with aggregated case counts provided by health authorities (for comparison, China CDC published the first epidemic curve by symptom onset on Jan 29, 2020 ¹⁴). The line list provides information on the lag between symptom onset and detection by the healthcare system, reporting delays, and travel history, which is not available from aggregated case counts published by official sources. Crowdsourced data can contribute to assessment of the effectiveness of interventions and the potential for widespread local transmission beyond the initial foci of infection. In particular, shorter delays between symptoms onset and hospitalization accelerate detection and isolation of cases, effectively shortening the infectious period.

Perhaps the most useful feature of our crowdsourced database was a careful curation of travel histories for patients returning from Wuhan, which, along with dates of symptom onset, allowed for precise estimation of the incubation period ^15,16. A narrow window of exposure can be defined for a subset of cases who had a short stay in Wuhan, at a time when the epidemic was still localized to Wuhan. Several teams have used ours and other datasets to estimate an incubation period of 5-6 days for 2019-nCoV (95% CI 2-11 dys) ^15-18. This is a useful parameter to guide isolation and contact tracing; the disease status of a contact should be known with near certainty after a period of observation of 14 days ¹⁵. Availability of a public dataset enables independent estimation of important epidemiological parameters by several modeling teams, allowing for confirmation and cross-checking at a time when information can be conflicting and noisy.

Another interesting finding in our data is the age distribution of cases. There is a heavy skew of reported 2019-nCoV cases towards older age groups, with a marked deficit of children. This pattern could point at age-related differences in susceptibility to infection, severe outcomes, or behavior. On one hand, a substantial portion of the cases in our database are from travelers, a population which tends to be skewed towards adults (although it does not exclude children). Further, because cases in our dataset were captured by the health system, they are biased towards the more severe spectrum of the disease, especially for cases from Mainland China. Clinical reports have shown that 2019-nCoV severity is associated with the presence of chronic conditions ^18,19, which are more frequent in older age groups. Nevertheless, we would also expect very young patients to be at risk of severe outcomes and to be reported to the healthcare system, as is seen for other respiratory infections ²⁰.

Biological differences could play a role in driving these age profiles. Detailed analysis of one of the early 2019-nCoV clusters suggests that all household members were infected, except for a baby, hinting at true biological differences in the risk of infection driven by age ²¹. It has been speculated that prior immunity from infection with related coronaviruses may protect children from SARS infection ²², which could also play a role for 2019-nCoV. In any case, if the age distribution of cases reported here was to be confirmed and the epidemic were to progress regionally and globally, we would expect a rise in respiratory mortality concentrated among ages 30 and above. This mortality pattern would be markedly different from the profile of 2009 influenza pandemic, where excess mortality was concentrated in those under 65 years ²³.

The age profile of 2019-nCoVcases is reminiscent of SARS infections, for which a deficit of children was also noted ^22,24. The SARS epidemic was characterized by a concentration of cases among health care workers due to super-spreading events occurring in healthcare facilities ²². We do not see a large fraction of health care workers in the data currently available for 2019-nCoV, although one large cluster was reported in Wuhan (not captured in our data).

Based on our data, the timeline of the outbreak indicates rapid growth in several provinces of China, consistent with local transmission outside of Wuhan. At this point, province-level epidemic curves are only available by date of reporting, rather than date of symptoms, which tends to inflate recent case counts if detection has increased. Further, these province-level curves include both returning travelers from Wuhan (importations) and locally acquired cases, which contribute to overestimate the potential for local transmission. It is worth noting that other lines of evidence suggest that local transmission is now well-established outside of Wuhan, as travel increased just before the Chinese New Year and before the ban was implemented in Wuhan ^10,25. Accordingly, in our line list, we have information on a few local transmission clusters reported outside of Wuhan. For instance, there is evidence of second-generation transmission in Shanxi on Jan 21, 2020. We recognize that our data source is useful and timely but should not replace official statistics. Manual compilation of detailed line lists from media sources is highly time consuming, and it is not sustainable when case counts reach several thousands. Here we provide detailed data on 288 patients when the official case count is over 7000 (as of Jan 30, 2020) ¹⁰, representing a sample of ∼4% of all reported cases. At the time of this writing, efforts are underway to coordinate compilation of 2019-nCoV cases from on-line sources across various modeling teams. Ultimately, we expect that a line list of patients will be shared by government sources with the global community; however, data cleaning and access issues may take prohibitively long to resolve. For the West African Ebola outbreak, it took 2 years until publication of a line list ²⁶. Given the progression of the 2019-nCoV outbreak, such a long delay would be counterproductive.

In conclusion, crowdsourced epidemiological data can be useful to monitor emerging outbreaks, such as 2019-nCoV (see also ⁷ for a related study in the context of Ebola). These efforts can help generate and disseminate detailed information to the scientific community, particularly in the early stages of an outbreak when little else is available, enabling independent estimation of key parameter estimates that affect interventions. Based on our small sample of 2019-nCoV cases, we see an intriguing age distribution, reminiscent of that of SARS, which warrants further epidemiologic and serologic studies. We also report early signs that the response is strengthening in China based on a decrease in case detection time, and rapid management of travel-related cases that are identified internationally. As a caveat, this is an early report of a rapidly evolving situation and the parameters discussed here could change quickly. In the coming weeks, we will continue to monitor the epidemiology of this outbreak using data from news reports and official sources.