Heterogeneity analysis of COVID - 19 clinical phenotype in Wuhan

2019-nCoV, 2019 novel coronavirus; COVID-19, coronavirus disease 2019; HRCT, High resolution computed tomographic; IQR, inter quartile range; CI, condence interval; GGO, ground glass opacity; CRP, C-reactive protein.


Conclusion
The heterogeneity of clinical symptoms for COVID-19 patients in the outpatient is signi cant. We should pay attention to patients without symptoms or those with only extra-respiratory symptoms, who are prone to missed diagnosis.

Background
In December 27th of 2019, we treated three patients with viral pneumonia from one family, and then treated four patients with similar viral pneumonia from South-China seafood market. We then rstly reported these patients with unexplained pneumonia to both Wuhan Municipal Health Commission and Hubei Municipal Health Commission 1 , thus started the battle for this epidemic situation. Among the three patients from one family, the mother rstly visited the respiratory outpatient due to symptoms of fever, dry cough, and wheezing, who was admitted as "community acquired pneumonia". In the same day, the father visited the neurology department due to headache and was transferred to respiratory department based on the "pneumonia" ndings by chest CT examination. We found similar chest CT imaging from the couple. Then we let their son without any symptoms perform chest CT scanning and found pneumonia-like change, which indicated that the clinical symptoms of this disease had signi cant heterogeneity. The pathogen has been named as 2019 novel coronavirus (2019-nCoV) 2 and this disease is named as coronavirus disease 2019 (COVID-19) by World Health Organization 3 .
Up to March 13th 2020, a total of 132 758 patients were diagnosed as COVID-19 worldwide and 122 counties or districts had reported cases 4 . Since the outbreak of COVID-19, some studies reported the clinical characteristics of COVID-19 patients in hospital [5][6][7] , but no study analyzed the clinical features of COVID-19 patients in the outpatient. We found that the clinical symptoms of outpatient COVID-19 patients had signi cant heterogeneity, which could have manifestations from occult infection without symptoms to acute respiratory distress syndrome. This study aimed to describe the difference of clinical features, laboratory examinations and imaging of COVID-19 patients with different syndromes at respiratory outpatient in January and February 2020 and provided early diagnosis reference of 2019-nCoV for physicians of districts with COVID-19 worldwide.

Study Design and Population
COVID-19 patients were included in the respiratory outpatient of our hospital from January 1st to February 28st 2020 based on the diagnosis criteria of Diagnosis and Treatment Protocol of Novel Coronavirus Pneumonia (edition 5) by National Municipal Health Commission 8 . The inclusion criteria included: 1) age > 18 years; 2) rst visit with complete outpatient patient materials; 3) with results of blood routine, CRP ,chest CT imaging and RT-PCR test positive for 2019 n-Cov ; 4) patient with SpO 2 ≤ 93 was diagnosed as severe case in resting condition. Exclusion criteria included: 1) revisit with rst visit in other hospital; 2) incomplete patient materials; 3) without results of blood routine, CRP and chest CT imaging. Institutional review board approval was obtained and informed consent was waived (No.

2020011).
Based on the complaints in visit, the patients were classi ed into four groups including group A (patients without symptoms), group B (fever patients with respiratory or extra-respiratory symptoms), group C (patients with only respiratory symptoms but without fever), and group D (patients with extra-respiratory symptoms such as loss of appetite, muscle soreness, fatigue, and fatigue, but without fever).
Our hospital is a public Tertiary hospital, located in the center of epidemic area, which is only 1.5 miles away from South-China seafood market. Our hospital rstly found and reported the outbreak of this epidemic to the government.

Observational Indicators
The demographic data, history, comorbidities, symptoms, laboratory examinations, and chest HRCT scanning of patients were collected from the electric medical report system in the outpatient. The results of chest HRCT scanning were reviewed by two associate chief physicians. As lots of patients were still in hospital, this study did not include the nal outcomes in the data analysis.

Statistical analysis
Due to all quantitative variables did not t the normal distribution well, the statistical description of these variables was expressed as median and IQR (inter quartile range), and the comparisons among groups were performed by using Kruskal-Wallis test, then Student-News-Keuls test was used for pairwise comparisons based on the rank. For qualitative variables, statistical description was expressed as frequency (percentage), 95% CI (con dence interval) of proportions was estimated by using Clopper-Pearson method, Fisher exact probability method was chosen for comparisons among groups, and stepdown-Bonferroni method was selected for pairwise comparisons. Statistical software was SAS 9.4 (SAS Institute Inc, Cary, NC), and all hypothesis tests were two-sided tests with a signi cance level of 0.05.

Comparisons Of Characteristics Among The Groups
The percentages of all the groups included 2.91% (95% CI: 1.34-5.46%) in group A, 50.81% (95% CI: 45.09-56.51%) in group B, 18.12% (95% CI: 13.99-22.88%) in group C, 28.16% (95% CI: 23.21-33.53%) in group D. All the patients in group D had no fever and most of them had fatigue, loss of appetite, muscle soreness and others ( Table 2). The ages among the groups were signi cantly different (P = 0.0001). The median age of group A was 33.00 years (IQR: 31.00-49.00 years), which was younger than the other three groups. The gender among the groups were signi cantly different (P = 0.0386). The percentage of male in group was the highest (66.67%), but no signi cant difference was identi ed between the groups after P-value adjustment. The percentage of patients with hypertension history among the groups were signi cantly different (P = 0.0196). The percentage of patients with hypertension in group D was signi cantly higher than that of group C, but no signi cant difference was found in other group comparisons. The levels of CRP among groups were signi cantly different (P = 0.0167). The level of CRP in group A was signi cantly lower than that of group C, but no signi cant difference was found in other group comparisons. The comparisons for the level of SpO2 found that the hypoxia condition of group A was better than other three groups, but no signi cant difference was found between other three groups. Only one side and multiple GGO of imaging signs among the groups were signi cantly different (P = 0.0381), but it did not maintain statistically difference after further P value adjustment. The percentages of severe patients among the groups were signi cantly different (P = 0.0156). No patient in group A met with the severe criteria, but no signi cant difference was found in other group comparisons. Analysis of chest CT signs for all the patients in the rst visit found common changes included bilateral lung multiple ground glass opacity (GGO) (71.52%, 95% CI: 66.14-76.49%), interlobular septal thickening (68.28%, 95% CI: 62.78-73.44%), and thickening of broncho-vascular tracts (22.98%, 95% CI: 18.40-28.08%), but no signi cant difference of other signs were found among the group comparisons (Table 3).

Discussion
The results of this study indicated that the percentage of patients with occult infection in group A was 2.91% with median age of 33 years (IQR: 31.00-49.00 years), which was signi cantly lower than that of other three groups. All of them had no symptoms and basic diseases as well as normal blood routine, CRP and SpO 2 . Only new single or multiple patch or cluster shadows were found in chest CT (Figure A &   B). Occult infection is an important feature for 2019-nCoV infection. A previous news media reported that a COVID-19 patient in a Wuhan hospital had no symptoms before surgery, who caused the infections of 14 medical staffs. This indicated that occult infection was an important infection source and added the di culty for the prevention and control of COVID-19.
In previous published clinical studies on 2019-nCoV, the percentages of patients with fever were 83-98.6% 5-7 , but Guan et al. 9 reported that only 43.1% of patients had fever symptom in early stage. Our study showed that the percentage of patients with complaint of fever in visit was 50.81%, which was similar to the results of the later study. As all the patients in the previous studies were in hospital, they had already progressed into moderate or advanced stage and fever symptom was more common.
However, all the patients in our study were in the early stage, therefore our result was similar to the later one. The incidences of cough in relevant reports were 59.4-82% 6,7,9 , and our study found that the percentage of dry cough was 35.28%. This difference of results may be related with the difference of included subjects. All the subject in our study were patients in the outpatients, who was less severe than patients in hospital and had less fever and respiratory symptoms in the early stage. This indicated that the percentage of pneumonia patients without fever after the 2019-nCoV infection was signi cantly higher than that of SARS 10 and MERS (middle east respiratory syndrome) 11 . If we use fever to screen early infected patients, it will lead to the missed diagnosis of many patients, which indicated that it was more di cult to prevent and control 2019-nCoV infection.
During this COVID-19 epidemic, many pneumonia patients visited with extra-respiratory symptoms such as fatigue, fatigue, loss of appetite, and muscle soreness, which are patients of group D in our study. The percentage of group D was 28.16% with median age of 58.00 years (IQR: 50.00-65.00 years) who was older than those of other three groups of patients without statistically signi cant difference. 67.82% (59/87) of them were female. All of them had no fever. The time from disease onset to visit was almost around 10 days. Their blood routines were almost normal. The scope of effusion lesions were relatively wide in chest CT ( Figure  As the infection of 2019-nCoV is occult without symptom or with rst symptoms beyond respiratory tract, this group of patients may visit other non-infection departments, which causes missed diagnosis, thus increases the risk of virus spread. The daily new cases of China mainland gradually decreased recently, but this numbers in other countries such as European Region are continuously increasing 12 . In order to control the epidemic, we suggested that 1) physicians in epidemic areas should pay attention to patients with atypical and extra-respiratory symptoms and people with close contact history, and perform chest CT scanning to screen COVID-19 especially in areas with limited capability to detect 2019-nCoV nucleic acid; 2) it is not enough for disease prevention and control organization of epidemic area as well as control and prevention staffs in public place to just monitor body temperature to screen COVID-19, but also need to collect information of atypical symptoms such as fatigue, loss of appetite and muscle soreness.
The results of study by Wang et al from Zhong Nan Hospital of Wuhan University indicated that decreased lymphocytes was one of the features for COVID-19 and consistent decreasing for this number indicated poor prognosis 7 . Due to the consistent in ammatory response, the apoptosis of lymphocytes increased with the dramatical decreasing for the number of lymphocytes, thus started "immune inhibition" or "immune paralysis" condition 13 . During the rescue treatment for patients with severe infection, except for pathogen, we should pay attention to immune imbalance secondary to excessive in ammatory response, which is an important feature of pathophysiological process, and main cause of the aggravation and death of patients with severe infection 14 . Our study showed that the percentage of patients with decreased lymphocytes was 12.94% (95% CI: 9.41-17.21%) and the median number of lymphocytes was 1.20 × 10 9 /L (IQR: 0.95-1.53) × 10 9 /L, which were signi cantly different from the report of Wang et al 7 . This may be related with subjects with rst visit to the outpatient. Most patients in this group were in the early stage and some patients probably had no decreased lymphocytes.
Although the imaging of viral pneumonia was not non-speci c 15 , chest CT scanning had convenient and rapid advantages, especially during initial stage that it was di cult to detect viral nucleic acid. Although we found that lung imaging of each group had their own features ( Figure A-H), our study did not nd statistical difference of imaging signs among each group probably due to lack of digital tools to describe the imaging features and not able to describe the location and scope of signs distribution. We though that in area with high incidence of disease, HRCT had imaging features of viral pneumonia, which could make the clinical diagnosis of COVID-19 together with epidemiological materials and clinical symptoms.
However, Wáng et al. 16 thought that the clinical signi cance of CT examination for 2019-nCoV was not de ned, avoided as much as possible, and recommended chest X-ray examination to observe the progression and regression of lung lesions for patients with moderate/severe patients. We thought that if this strategy causes mild or early patients not able to be diagnosed timely, it will increase the risk to spread virus for this group of patients as infection source.
Our study had several limitations. Firstly, all the included subjects were outpatient patients. The information of outpatient medical record system is relatively simple, which may miss some important information, and the number of severe patients in the outpatient is small, which may have selection bias. Secondly, as the number of patients in group A is small, and the power of test is small when group A is compared with other groups, which may explain why the difference among the groups was statistically signi cant but comparisons between each group were negative. Thirdly, as lots of patients were still in hospital, we could not analyze the risk factors of adverse consequence and need to continuously follow up these patients.
In conclusion, the heterogeneity of early clinical symptoms for COVID-19 patients was signi cant. It was impossible to early screen these patients by only symptoms of fever and cough, which may lead to the missed diagnosis of patients without symptoms or only with extra-respiratory symptoms. In the epidemic area, it is necessary to screen COVID-19 for patients without fever, patients with symptoms such as fatigue, loss of appetite, and muscle soreness as well as people with close contact history.
16.Wáng YX, Liu WH, Yang M, Chen W. The role of CT for Covid-19 patient's management remains poorly de ned. Ann Transl Med. 2020. In press. doi: 10.21037/atm.2020.02.7111:10 AM 2/25/2020. Figure 1 HRCT scan of the lung area in a young woman of who is a NCP close contact with no symptoms showing ground-glass opacities, consolidation, the blood vessel thickening.

Figure 2
HRCT scan of the lung area in a old man whose chief complaint is shortness of breath and weak for a week showing multiple ground-glass opacities and diffusely thickened interlobular septa giving crazypaving appearance in both lung subpleural areas.

Figure 3
HRCT scan of the lung area in a young man whose chief complaint is fever and cough for a week showing multiple ground-glass opacities, and some area showing mixed ground-glass opacities with blood vessel thickening inside.

Figure 4
HRCT scan of the lung area in a old woman whose chief complaint is weak, anorexia and muscle soreness for 10 days showing multiple ground-glass opacities and diffusely thickened interlobular septa giving crazy-paving appearance in both lung subpleural areas. Some areas showing consolidation, oblique ssure thickening on the left side of pleura, traction deformation of oblique ssure on both sides, anterior displacement, and multiple tentorial adhesion of pleura.