key: cord-0917787-znrk1kgd
authors: Zhang, Yanping; Luo, Wei; Li, Qun; Wang, Xijie; Chen, Jin; Song, Qinfeng; Tu, Hong; Ren, Ruiqi; Li, Chao; Li, Dan; Zhao, Jing; McGoogan, Jennifer M; Shan, Duo; Li, Bing; Zhang, Jingxue; Dong, Yanhui; Jin, Yu; Mao, Shuai; Qian, Menbao; Lv, Chao; Zhu, Huihui; Wang, Limin; Xiao, Lin; Xu, Juan; Yin, Dapeng; Zhou, Lei; Li, Zhongjie; Shi, Guoqing; Dong, Xiaoping; Guan, Xuhua; Gao, George F; Wu, Zunyou; Feng, Zijian
title: Risk Factors for Death Among the First 80 543 COVID-19 Cases in China: Relationships Between Age, Underlying Disease, Case Severity, and Region
date: 2021-05-27
journal: Clin Infect Dis
DOI: 10.1093/cid/ciab493
sha: 360fb514da0c2ef3edbcacbf04d2c2e17b4e7dcc
doc_id: 917787
cord_uid: znrk1kgd

BACKGROUND: Knowledge of COVID-19 epidemiology remains incomplete and crucial questions persist. We aimed to examine risk factors for COVID-19 death. METHODS: A total of 80 543 COVID-19 cases reported in China, nationwide, through April 8, 2020 were included. Risk factors for death were investigated by Cox proportional hazards regression and stratified analyses. RESULTS: Overall national case fatality ratio (CFR) was 5.64%. Risk factors for death were older age (≥80: adjusted hazard ratio [aHR]=12.58, 95% confidence interval [CI]=6.78-23.33), presence of underlying disease (aHR=1.33, CI=1.19-1.49), worse case severity (severe: aHR=3.86, CI=3.15-4.73; critical: aHR=11.34, CI=9.22-13.95), and near-epicenter region (Hubei: aHR=2.64, CI=2.11-3.30; Wuhan: aHR=6.35, CI=5.04-8.00). CFR increased from 0.35% (30-39 years) to 18.21% (≥70 years) without underlying disease. Regardless of age, CFR increased from 2.50% for no underlying disease to 7.72% for 1, 13.99% for 2, and 21.99% for ≥3. CFR increased with worse case severity from 2.80% (mild), to 12.51% (severe) and 48.60% (critical) regardless of region. Compared to other regions, CFR was much higher in Wuhan regardless of case severity (mild: 3.83% versus 0.14% in Hubei and 0.03% elsewhere; moderate: 4.60% versus 0.21% and 0.06%; severe: 15.92% versus 5.84% and 1.86%; and critical: 58.57% versus 49.80% and 18.39%). CONCLUSIONS: Older patients regardless of underlying disease and patients with underlying disease regardless of age were at elevated risk of death. Higher death rates near the outbreak epicenter and during the surge of cases reflect the deleterious effects of allowing health systems to become overwhelmed.

As the first country to encounter COVID-19, experience a large outbreak, and achieve epidemic control, 1,2 a broad range of epidemiologic studies have been conducted in China. [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] These studies use data from China's 4 complementary infectious disease information systems. 2, 22 However, most have included few cases from small areas or single centers during short periods in January-February 2020. [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] To date, the largest COVID-19 case series in China included all 72 314 SARS-CoV-2 patients as of February 11, 2020. 6,22 This report provided first epidemiologic curves, a timeline of discovery and response events, and distributions of cases across age, case severity, geography, and time.

Importantly, it highlighted higher case-fatality ratios (CFRs) among older adults and adults with underlying disease. 6, 22 However, patient data were abruptly cut off on February 11 and analyzed immediately while many remained hospitalized. Therefore, for many of these patients, the dataset did not include follow-up through the entire course of illness to recovery or death. Moreover, in April 2020, Wuhan officials added 1615 records of confirmed COVID-19 cases (325 survivors and 1290 deaths), most of which had occurred in January and February, but had gone unreported due to overwhelmed systems during the case surge. Thus, our understanding of COVID-19 epidemiology in China remains incomplete, and many gaps in the evidence persist. For instance, risk factors for death from COVID-19 in China have been identified but have yet to be quantified and thoroughly investigated in a large cohort. 23, 24 Therefore, our primary aim was to investigate risk factors for death and explore the relationships between them using all confirmed COVID-19 cases nationwide followed for their entire clinical course in the complete first -wave‖ of the epidemic in China. Secondarily, we aimed to more fully characterize COVID-19 epidemiology.

A c c e p t e d M a n u s c r i p t 5

A nationwide retrospective cohort study design was used to investigate all confirmed COVID-19 cases in China as of April 8, 2020, followed through May 16, 2020 . This study was approved by the Chinese Center for Disease Control and Prevention Institutional Review Board. Individual informed consent was not required.

In China, all COVID-19 cases must be entered into the Notifiable Infectious Disease Reporting System (NIDRS) within 2 hours of discovery. 2, 25 Since all NIDRS case reports contain individuals' unique national identification (ID) numbers, the system contains no duplicate reports. 2, 25 Each NIDRS case record must be investigated by local public health specialists within 24 hours of reporting. 2 Investigation results are collected, stored, and managed in the Epidemiological Investigation Information System (EIIS). 2 Each record in EIIS contains both the individual's national ID number and NIDRS case report number to prevent duplicate records and facilitate records matching. 2

The case definition in China-positive SARS-CoV-2 polymerase chain reaction (PCR) test results and documented symptoms-has been consistent throughout the pandemic and our entire study with one exception: in Hubei Province (including in Wuhan City), early in the outbreak when testing had not yet scaled up sufficiently, some patients were clinically diagnosed based on symptoms and lung imaging. 6 Notably, this case definition excludes asymptomatic infection. Thus, all data from all records in NIDRS and EIIS with positive SARS-CoV-2 PCR test results and documented symptoms through April 8, 2020 were included, but extracted 6 weeks later on May 16, 2020. This delay was required to ensure records of all previously unreported cases and deaths had been completed, quality checked, and released and to allow all patients to be followed through their entire clinical course to recovery or death. M a n u s c r i p t 6

Since all persons with positive SARS-CoV-2 PCR test results were hospitalized in isolation in China, 2 all deaths occurred in hospitals and thus were ascertained by physicians. All deaths among COVID-19 cases were recorded as COVID-19 deaths regardless of any other circumstances surrounding their deaths (eg, co-infection, myocardial infarction, stroke). No deaths were ascertained post-mortem as COVID-19.

For occupation, the health worker category was defined as any type of active employment in any health facility. For case severity, categories were mild, moderate, severe, and critical. For underlying disease, having no major pre-existing clinical diagnosis was categorized as -no‖ while having any was categorized as -yes.‖ Cases with a single underlying disease were further categorized by disease (eg, hypertension) and by number of diseases (eg, ≥3 diseases). Mild cases had mild clinical symptoms and no sign of pneumonia on imaging. Moderate cases had fever and respiratory symptoms with radiological findings of pneumonia. Severe cases were characterized by dyspnea, respiratory rate (RR) ≥30/minute, oxygen saturation ≤93%, PaO 2 /FiO 2 ratio <300 for adults, and any one of the 

Cases and deaths are presented as number, and cases at the national level only as percentages calculated as the number of cases in a category (numerator) divided by the total number of cases (denominator). Incidence and mortality were calculated as the number of cases and deaths, A c c e p t e d M a n u s c r i p t 7 respectively (numerator), divided by the total population (denominator), presented as /100 000. CFR was calculated as the number of deaths (numerator) divided by the total number of confirmed cases (denominator), presented as a percentage. Univariate and multivariate Cox regression models, based on survival analysis, were used to examine risk factors associated with death from COVID-19, producing unadjusted and adjusted hazard ratios (aHRs) and 95% confidence intervals (CIs). Age, sex, residence, underlying disease, case severity, and region variables were adjusted in the multivariate model (STATA v14.0, StataCorp, College Station, TX, USA). Stratified analysis with chi-square test was used to examine the relationships between variables. Regional differences were evaluated using Pearson chi-square and Fisher's exact test. National and regional epidemiological curves were plotted as daily number of confirmed cases versus date of symptom onset. Confirmed cases by province and date of symptom onset were presented as a heat-map time-series to illustrate geographic spread of COVID-19 (RStudio v3.6.3, RStudio, PBC, rstudio.com).

A total of 80 543 confirmed COVID-19 cases were used in the analysis (asymptomatic infections excluded but summarized in eTable 1 in the Supplement). In Table 1 , most were aged 40-69 years (60.68%) and urban residents (75.81%). A total of 34 320 cases (42.61%) had information on underlying disease in their records. Among them, 25 743 had no underlying disease (75.01%), 4923 had 1 underlying disease (14.34%), 1458 had 2 (17.00%) and 391 had ≥3 (4.56%). Among those with 1 underlying disease, hypertension was most common (53.14%). Most cases were mild (41.96%) or moderate (39.17%) at diagnosis; severe cases were less common (14.89%) and critical cases were rare (3.46%).

[Insert Table 1 

A c c e p t e d M a n u s c r i p t 8

In Table 2 , compared to <30 years, older age was associated with greater risk of death-3. 5 [Insert Table 2 

In Table 3 , among those with no underlying disease, CFR increased significantly with age from 0.7% for 40-49 years to 1.9% for 50-59 years, 6.7% to 60-69 years, and 18.2% for ≥70 years (P<0.001).

Significant increases in CFR with increasing age were observed for those with 1 underlying disease (P<0.001), 2 diseases (P<0.001), and ≥3 diseases (P=0.001). Conversely, CFR also increased significantly with the number of underlying diseases for single age groups. For example, for those 50-59, CFR significantly increased from 1.9% for no underlying disease to 2.8% for 1 underlying disease, 6.2% for 2 diseases, and 10.2% for ≥3 diseases (P<0.001). Significant increases in CFR with increasing age were also observed for single, specific underlying conditions (eTable 2 in the Supplement). Among those with a single underlying disease, CFR remained highest in the oldest age groups. For example, in the ≥70 age group, CFR was 33.3% for liver disease, 26 A c c e p t e d M a n u s c r i p t 9 Case Severity and Region Figure 1 shows progression from initial case severity at diagnosis through worst status reached during hospitalization to final outcome of recovery or death. Among those diagnosed at mild (56.2%) or moderate (30.0%), most did not progress further (74.2% and 93.3%, respectively) and recovered (97.2% and 97.8%, respectively). However, among those already severe (12.1%) or critical (1.6%) at diagnosis and among those who were ever severe (15.5%) or critical (3.7%) during hospitalization, CFRs were high (≥12% and >47%, respectively; eTable 3 in the Supplement). In Table 4 , those diagnosed at critical case status had very high CFRs in all regions (58.87% in Wuhan, 49.80% in Hubei, and 18.39% in the rest of China) regardless of proximity to the outbreak epicenter. However, highest regional CFR was observed in Wuhan (7.59%) and increasing CFR with increasing proximity to the outbreak was observed across all age groups and regardless of underlying disease or case severity.

[Insert Figure 1 and Table 4 .]

Nationally, incidence was 5.79/100 000. Incidence was highest for 60-69-year-olds (13.38/100 000) and urban residents (7.38/100 000). Altogether, 4545 deaths were recorded for national mortality of 0.33/100 000 and CFR of 5.64%. Elevated mortality and CFR were observed for ≥80-year-olds (3.38/100 000, 32.08%), retirees (12.90%), severe and critical cases (12.51% and 48.60%), cases with underlying disease (8.69%), and cases with symptom onset early in the epidemic (15.84%; Table 1) .

Notably, CFR increased with age, numbers of underlying diseases, and case severity, and decreased with later epidemic stage in all regions ( Table 4 ). Additional data on cases, incidence, deaths, mortality, and CFRs by region are presented in eTable 4 in the Supplement. 

This study of all 80 543 confirmed COVID-19 cases in China through April 8, 2020 is the first to provide an epidemiological description of a complete cycle from initial outbreak through rapid epidemic expansion to achievement of long-term control at a national level. The main result was the independent nature of 4 important risk factors for death: older age, greater numbers of underlying diseases, worse case severity at diagnosis, and close proximity to the outbreak epicenter. This important finding indicates that even otherwise healthy older adults and even young adults but with pre-existing disease are at elevated risk of death from COVID-19, which has implications for prevention and vaccine prioritization strategies. Moreover, it highlights the deleterious consequences of late diagnosis and overwhelmed health systems.

CFRs were higher in our cohort compared to the prior report of cases up to February 11, 2020, 6 both overall (5.6% versus 2.3%, respectively) and when disaggregated, for two reasons. First, we included follow-up time sufficient to ensure that all cases achieved either recovery or death whereas the prior paper did not. The prior paper's data cutoff occurred very early in the epidemic and many patients were still hospitalized at the time the data were analyzed. Second, we included 1300 deaths not A c c e p t e d M a n u s c r i p t 11 previously reported, most of which occurred during the prior report's study period but were not included.

Up to now, the best available evidence on risk factors for COVID-19 death have come from metaanalyses of mostly small studies from China. 23, 24, [26] [27] [28] But, they were limited by heterogeneity, particularly with respect to how age groups were categorized and comorbidities were defined (ie, disease that pre-existed SARS-CoV-2 infection were often not differentiated from disease that emerged during COVID-19 disease progression). This presents challenges in isolating risk factors for death. More recently, a very large study in Mexico reported odds of death 7-fold greater for 61-80, and 12-fold greater odds for aged >80 and 24-31% greater for those with hypertension, obesity, diabetes, chronic obstructive pulmonary disease, or immunosuppression, and 85% greater for those with chronic kidney disease. However, the relationship between age and underlying disease was not explored. 29 Our results provide a clearer picture of groups more vulnerable to death, facilitating targeted prevention intervention and vaccine distribution.

We also found that although overall incidence was 5.79/100 000 and mortality was 0.33/100 000, these national level measures hid dramatic geographic and demographic differences such as very high incidence and mortality in Hubei and in Wuhan among older and urban adults. Overall, for the 81% of cases that were mild or moderate, CFR was under 3%, but the 15% who were severe faced a CFR of 12.5% and the 3.5% of cases who were critical had a CFR of nearly 50%. Yet in Wuhan, where the health system quickly became overwhelmed, 17% of cases were severe and had a CFR of 16% and nearly 3% were critical with a CFR of almost 60%.

Even for countries that, like China, took aggressive action to try to contain the infection and break the chains of human transmission, 2 enormous numbers of people fell ill and died. For instance, in Italy, the first western nation to be affected by COVID-19, the number of cumulative cases rose 100 000fold and deaths rose 429-fold in March 2020 alone. 30 Still lacking therapeutics and vaccines, the only defenses were traditional public health methods. 2 China provides not only an example of how decisive action to thoroughly implement these countermeasures can bring COVID-19 under control, 2 but also a A c c e p t e d M a n u s c r i p t 12 cautionary tale of how quickly a new emerging infectious disease can overwhelm a health system.

Evidence of China's struggles during the January and February 2020 -surge‖ of cases are observable in the data. Higher incidence values were to be expected in Wuhan and outlying areas in Hubei, but these areas also suffered higher mortality and CFRs compared to elsewhere in China, which reflects the enormous strain on the local health system during the surge. These areas also suffered reporting delays (ie, 325 survivors and 1290 deaths never-before included in other studies) and incompleteness of records due to stresses on the public health system. These issues have also been observed, for example, in Italy where the CFR in Milan was much greater than some parts of southern Italy and in the US, the CFR in New York was greater than other states like Minnesota. Many national, state, and city public health systems have also experienced disruption to timely case and death reporting due to the overwhelming stress of case surges.

However, CFR was not only elevated because of health system strain. Delayed testing because of insufficient capacity and inexperience with a new disease meant late presentation and suboptimal clinical management early in the pandemic. CFRs were elevated early in the outbreak in all regions of China compared to later in the outbreak because diagnosis and clinical management improved over time. This has been observed in many nations as they began dealing with COVID-19 for the first time.

As physicians and nurses learned from their own experience (and that of others) how to care for COVID-19 patients and as best practices were developed and more broadly shared, CFR naturally declined.

Major strengths of this study were its nationwide scope, very large size, and complete inclusion of all cases in China's COVID-19 epidemic wave followed for their entire clinical course. Nevertheless, our study had some important limitations. First, because of China's official case definition, characteristics of asymptomatic infections were not examined. Second, the data were compiled under the crushing pressure of a rapidly spreading epidemic. This limited our ability to explore other variables (eg, smoking, obesity) and outcomes (eg, rehospitalizations, post-acute phase complications), which are now of interest. Third, this also meant some records were incomplete or may have contained errors.

A c c e p t e d M a n u s c r i p t 13 For example, 57% of records contained no information on underlying conditions, requiring our analyses on this variable to be limited to a subset of cases. Fourth, clinically diagnosed cases during the early outbreak when testing was scarce may have included some misdiagnoses, thereby exaggerating the total case count. Finally, although all confirmed cases were immediately hospitalized, and therefore all deaths among them occurred in hospital and were recorded as COVID-19 deaths regardless of other contributing factors, there remained a small possibility of underascertainment of COVID-19 deaths.

In conclusion, this study provides new, high-quality evidence of the epidemiology of COVID-19 from M a n u s c r i p t 21 

Report of the WHO-China Joint Mission on coronavirus disease 2019 (COVID-19). Available at

One hundred days of COVID-19 prevention and control in China

The 2019-nCoV Outbreak Joint Field Epidemiology Investigation Team

Notes from the field: an outbreak of NCIP (2019-nCoV) infection in China-Wuhan

Clinical characteristics of 138 hospitalized patients with 2019 novel coronavirus-infected pneumonia in Wuhan

Clinical findings in a group of patients infected with the 2019 novel coronavirus (SARS-CoV-2) outside of Wuhan, China: retrospective case series

The Novel Coronavirus Pneumonia Emergency Response Epidemiology Team. The epidemiological characteristics of an outbreak of 2019 novel coronavirus diseases (COVID-19)-China

Clinical characteristics of coronavirus disease 2019 in China

Clinical characteristics of imported cases of COVID-19 in Jiangsu Province: a multicenter descriptive study

Epidemiological and clinical features of 125 hospitalized patients with COVID-19 in

Clinical characteristics of patients with 2019 coronavirus disease in a non-Wuhan area of Hubei Province, China: a retrospective study

A cross-sectional comparison of epidemiological and clinical features of patients with coronavirus disease (COVID-19) in Wuhan and outside Wuhan, China

Clinical features and short-term outcomes of 221 patients with COVID-19 in Wuhan

Clinical characteristics and drug therapies in patients with the common-type coronavirus disease 2019 in Hunan, China

Epidemiological characteristics and incubation period of 7015 confirmed cases with coronavirus disease 2019 outside Hubei Province in China

A retrospective study of the initial 25 COVID-19 patients in Luoyang, China

Clinical characteristics and co-infections of 354 hospitalized patients with COVID-19 in Wuhan, China: a retrospective cohort study

Clinical characteristics and outcomes of hospitalised patients with COVID-19 treated in Hubei (epicentre) and outside Hubei (non-epicentre): a nationwide analysis of China

Clinical characteristics of coronavirus disease

Epidemiologic and clinical characteristics of 91 hospitalized patients with COVID-19 in Zhejiang, China: a retrospective, multi-centre case series

Epidemiological and clinical characteristics of a familial cluster of COVID-19

Clinical characteristics of different subtypes and risk factors for the severity of illness in patients with COVID-19 in Zhejiang, China

Characteristics of and important lessons from the coronavirus disease 2019 (COVID-19) outbreak in China: summary of a report of 72 314 cases from the Chinese Center for Disease Control and Prevention

The age-related risk of severe outcomes due to COVID-19 infection: a rapid review, meta-analysis, and meta-regression

Quantification of death risk in relation to sex, preexisting cardiovascular diseases and risk factors in COVID-19 patients: let's take stock and see where we are

A nationwide web-based automated system for outbreak early detection and rapid response in China

Risk factors of critical and mortal COVID-19 cases: a systematic literature review and meta-analysis

Risk factors for mortality in patients with coronavirus disease 2019 (COVID-19) infection: a systematic review

A c c e p t e d M a n u s c r i p t 15 A c c e p t e d M a n u s c r i p t 22 Changes in case severity status (disease progression) over time from initial diagnosis to final outcome in China. Status at initial diagnosis is shown at left, worst status reached during hospitalization is shown in the center, and final outcome is shown at right. Case status is color-coded as mild in green, moderate in yellow, severe in orange, and critical in red. Likewise, final outcome is color-coded as recovery in blue and death in purple. Flow between states is indicated and thickness of bands is proportional to numbers of cases. It is important to note that, in China, all confirmed COVID-19 cases were hospitalized in isolation regardless of the severity of their condition and that all cases remained until either recovery or death, with recovery defined as completely free of all COVID-19-associated symptoms and consistently negative on SARS-CoV-2 PCR testing. Finally, among a total of 80 543 participants, 459 had missing data for case severity status (299 at diagnosis only and 160 at diagnosis and during hospitalization), and therefore, 80 084 participants (99.4%) were included in this analysis.A c c e p t e d M a n u s c r i p t 26 Figure 1