The association between diurnal temperature range and clinic visits for upper respiratory tract infection among college students in Wuhan, China

The effects of daily mean temperature on health outcomes have been discussed in many previous studies, but few have considered the adverse impacts on upper respiratory tract infection (URTI) due to variance of temperature in one day. Diurnal temperature range (DTR) was a novel indicator calculated as maximum temperature minus minimum temperature on the same day. In this study, generalized additive model (GAM) with quasi-Poisson distribution was used to investigate the association between DTR and the number of daily outpatient visits for URTI among college students. Data about meteorological factors and air pollutants were provided by Hubei Meteorological Bureau and Wuhan Environmental Protection Bureau, respectively. Outpatient visits data were collected from the Hospital of Wuhan University from January 1, 2016, to December 31, 2018. Short-term exposure to DTR was associated with the increased risk of outpatient for URTI among all college students. Per 1 °C increased in DTR was associated with 0.73% (95%CI: 0.24, 1.21) increased in outpatient visits of all college students for URTI at lag 0 day. The greatest effect values were observed in males [1.35% (95%CI: 0.33,2.39)] at lag 0-6 days, and in females [0.86% (95%CI: 0.24, 1.49)] at lag 0-1 days. DTR had more adverse health impact in autumn and winter. Public health departments should consider the negative effect of DTR to formulate more effective prevention and control measures for protecting vulnerable people.


Introduction
Upper respiratory tract infection (URTI) is a common disease among people, with respiratory symptoms including cough, stuffy nose, sniffling, and sore throat. According to the Global Burden of Disease Study 2016(GBD 2016, the incidence of URTI is the highest among 318 diseases, even higher than dental caries and diarrhea (Disease et al. 2017). A previous study found that in Italy, URTI accounted for about four-fifths of the outpatients for respiratory diseases in emergency department visits in people aged 0-18 years (Bono et al. 2016). Cross-sectional studies pointed out that URTI took up 30.1% of respiratory disorders and 26.4% of all hospital outpatient cases in Hong Kong (Lo et al. 2011;Nelson et al. 2007). Although the mortality caused by URTI was relatively low, it would have negative impacts on daily life, resulting in days lost from work and school (Mourtzoukou and Falagas 2007).
Ambient temperature was widely used to investigate the associations with the mortality and admission rate of respiratory disease in epidemical researches for the past few years (Zhao et al. 2019;Yang et al. 2018;Michelozzi et al. 2009). A study conducted in Iran found that the mortality rate increased as the daily mean temperature decreased (Khanjani and Bahrampour 2013). Some researchers observed that optimal daily mean temperatures were existing (18.1°C for Wuhan, 19.0°C for London, and 29°C for Taiwan), which harmful effects were the lowest at these temperatures (Zhang et al. 2016;Patz et al. 2005;Hajat et al. 2002). However, most of them chose daily mean temperature as personal exposure index, which would ignore the variance of temperature within one day. Extreme temperature was confirmed to be associated with adverse health outcomes in a large number of epidemiological studies (Moghadamnia et al. 2017;Wang et al. 2017;Wu et al. 2016;Zhang et al. 2017a). Diurnal temperature range (DTR) was a novel index calculated as maximum temperature minus minimum temperature in the same day, to estimate the effects of many adverse outcomes. DTR was a useful indicator of climate change and variable and can provide more information due to the consideration of the highest temperature and the lowest temperature (Braganza et al. 2004). Byun et al. found that a 1°C increased in DTR corresponded to an 8.81% (95% CI: 3.46%-14.44%) increase in emergency department visits for multiple sclerosis at lag 0-1 days (Byun et al. 2020). Wang et al. observed the acute effect of DTR on respiratory and digestive emergency room admissions among the elderly in Beijing [per 1°C increased in DTR associated with 2.08% (95% CI: 0.88%-3.29%) and 2.14% (95% CI: 0.71%-3.59%) changed in relative risk at lag 0-7 days, respectively] (Wang et al. 2013).
Owing to the factors related to climate change, such as aerosols and clouds, the global average DTR has declined in recent decades because the minimum temperature has increased faster than the maximum temperature (Easterling et al. 1997;Makowski et al. 2008). However, there were also some studies that found that the average DTR will be projected to increase in most countries and regions in the future (Lindvall and Svensson 2015;Lee et al. 2020). In a warmer climate, the risks of DTR and mortality were higher especially in eastern Asia (Lee et al. 2019;Lee et al. 2018a, b).
Due to the influence of geographical environment and socio-economic factors, the health risks of climate change show obvious differences on the spatial and temporal scales (Gasparrini et al. 2017;Li et al. 2019;Messina et al. 2019;Barreca and Schaller 2020). Additionally, few studies focused on the relationship between DTR and the outpatient visits of URTI in college students. Therefore, it was necessary to carry out relevant researches in places with different weather patterns and different populations.
In this study, we collected the data about outpatient visits from the Hospital of Wuhan University and environmental factors from January 1, 2016, to December 31, 2018, and used a generalized additive model with quasi-Poisson distribution to estimate the associations between short-term exposure to DTR and the risk of URTI. A deep understanding of the relationship between DTR and URTI will help the relevant authorities to take appropriate measures to control the occurrence of the disease.

Study area
In the current study, our study area is Wuhan University, located in the Wuchang district of Wuhan, the capital of Hubei Province. Wuhan is situated in the central part of China and the Yangtze River which has a typical subtropical monsoon climate with high temperatures in summer and low temperatures in winter. The Hospital of Wuhan University is a university-level hospital affiliated with Wuhan University, consisting of four branches located in each of the four academic departments (Faculty of Medicine, Faculty of Engineering, Faculty of Arts and Sciences, and Faculty of Information Science) ( Figure S1). As the nearest and designated hospital for university students' medical insurance, the first choice for nearly 60,000 college students of Wuhan University is to seek medical support and public health services.

Data collection
The data about outpatient visits were collected from the Hospital of Wuhan University from January 1, 2016, to December 31, 2018. The medical information data included date of visit, gender, age, occupation, and diagnosis. We extracted the part of college students according to their occupation (only students) and diagnosis (only URTI). Finally, we established a new database to record the number of daily outpatient visits in college students for URTI.
Data about meteorological factors in this study period were provided by Hubei Meteorological Bureau (http:// hb.cma.gov.cn/), including daily maximum, mean, and minimum temperature, as well as relative humidity. DTR was defined as maximum temperature minus minimum temperature on the same day. Data about air pollution including the daily average concentration of fine particulate matter (PM 2.5 ), particulate matter (PM 10 ) , sulfur dioxide (SO 2 ), nitrogen dioxide (NO 2 ), and ozone (O 3 ) were collective from Wuhan Environmental Protection Bureau (http://hbj.wuhan.gov.cn/), which had 10 monitoring stations distributed throughout the Wuhan city. Based on the geographical location of the study area, we chose the average values of these two nearest sites called Liyuan-Donghu and Ziyang-Wuchang as exposure indicators for the college students of Wuhan University.

Statistical analysis
Statistical description analysis (mean, standard deviation, and quartiles) was used to exhibit the daily number of outpatient visits, meteorological factors, and air pollution. Spearman rank correlation analysis was conducted to explore the relationships between paired data of air pollutants and meteorological factors. The generalized additive model (GAM) with quasi-Poisson distribution was used to estimate the increased number of outpatient visits of college students for URTI due to DTR. Cubic smoothing functions were applied to control multi-annual and seasonal trends in time-series design, as well as the weather factors (i.e., daily mean air temperature and relative humidity). Degree of freedom (df) was selected by referring to Akaike Information Criterion (AIC) and previous studies. Indicators for the day of week (DOW), the public holidays (holiday), and the summer and winter vacations (vacation) were adjusted as dummy variables in the GAM. Besides, considering the potential confounding effects of influenza epidemics, influenza epidemic (influenza) was defined as the number of outpatient visits in a week larger than the 75% quantile of the total number of outpatient visits per week in that year and also controlled as a dummy variable in the GAM (Wong et al. 2002;Lin et al. 2016).
The naturally smooth functions of two variables, average temperature and relative humidity, were included in the model to control the confounding effects of meteorological factors on the association between DTR and outpatient visits for URTI. The model was as follows: Where E(Yt) is the estimated daily outpatient visits of college students for URTI; α is the intercept of the model; β is the log of excess relative risk in upper respiratory clinic visits corresponding to each 1°C increase in DTR; Zt represents the values of DTR on day t,°C; ns is a spline smoothing function for the nonlinear variables such as time, temperature, and humidity; df means the degree of freedom; in the final model, we defined 7 df per year for time trends, 6 and 3 df per year for daily mean temperature and relative humidity, respectively Zanobetti and Schwartz 2009;Peng et al. 2006;Chen et al. 2010).
DTR was incorporated into the model to evaluate the relationship between the outpatient visits for URTI and DTR. Due to the potential delayed effects of DTR on the outpatient visits for URTI, this study used the single-day lags (lag 0, lag 1, lag 2, lag 3, lag 4, lag 5, lag 6, and lag 7 day) and cumulative lags (lag 0-1, lag 0-2, lag 0-3, lag 0-4, lag 0-5, lag 0-6, and lag 0-7 days) to estimate the effects of DTR at different lag days. We also analyzed the association between DTR and outpatient visits for URTI in different gender (males and females) and season [spring (March to May), summer (June to August), autumn (September to November), and winter (December to February in the next year)] groups (Kim et al. 2014;Wang et al. 2018). After evaluating the health effects of DTR in the initial model, two sensitivity analyzes were conducted to assess the robustness of the effects of DTR on outpatient visits for URTI. One was constructing new models by adding each air pollutant into models (only one air pollutant at a time), and another was changing the dfs for time (from 7 to 6 or 8 per year), temperature (from 6 to 5 or 7), and relative humidity (from 3 to 2 or 4) to examine the robustness of the results in our study.
Consistent with previous studies, results in this study were reported as excess relative risk (ERR: RR-1) and 95% confidence intervals (CI). All statistical analyzes were conducted in R software (version 4.0.3), and the "mgcv" and "splines" packages were utilized to fit the model. Results with a 2sided and P value < 0.05 were statistically significant. Figure 1 illustrates the distribution of daily diurnal temperature range (DTR) and the number of outpatient visits for URTI stratified by gender in the Hospital of Wuhan University, China. Similar trends between outpatient visits and DTR were observed. Relatively higher values for outpatient visits and DTR were found in cold seasons. Due to most college students leaving university during the winter vacation, in particular, Chinese New Year (Usually occurred in January or February), the number of outpatient visits was extremely low even though DTR remained high level. Table 1 shows the basic descriptions of demographics and environmental factors (meteorological factors and air pollution). There were 44,449 outpatient visits for URTI were recorded in hospital of Wuhan University from January 1, 2016, to December 31, 2018, including 23,654 females (53.2%) and 20,845 males (46.8%). More than 90% of college students are 18-29 years old ( Figure S2). Daily mean outpatient visits in this study period were about 41, 19, and 22 for all, males, and females, respectively. For meteorological factors, the minimum, mean, and maximum DTR were 0.9°C, 8.8°C, and 20.5°C. Daily mean temperature and relative humidity were 8.8°C and 79.5%, respectively. For air pollution, the daily average concentrations of PM 2.5 , PM 10 , SO 2 , NO 2 , and O 3 were 50.6 ug/m 3 , 80.1 ug/m 3 , 10.8 ug/m 3 , 44.3 ug/m 3 , and 88.0 ug/m 3 , separately. Four season-specific descriptive statistics both for meteorological variables and number of outpatient visits for URTI were shown in supplementary Table S1. There were 13,477, 6,693, 13,852, and 10,477 outpatient visits in spring, summer, autumn, and winter, respectively.

Results
The values of daily mean DTR were highest in spring (9.6°C) and lowest in summer (7.9°C). As seen in Figure 2, DTR was negatively correlated with meteorological factors and positively correlated with air pollutants. The largest values of Spearman correlation coefficient about DTR were found between DTR and PM 10 (r=0.53, P<0.001), and an extremely weak correlation was observed between DTR and temperature (r=-0.01, P<0.001), as seen in Table S2.   Figure 3 reveals the exposure-response (E-R) curves between DTR and outpatient visits for URTI stratified by gender at lag 0-6 days (for all and males) and 0-1 days (for females).
Although there were small plateau periods in the middle of the curves, nearly similar linear upward trends were found in all, males, and females, which indicated that DTR was a risk factor for the increase in the number of outpatient visits for URTI. Table 2 shows the estimated effects with 95%CI in outpatient visits of college students for URTI associated with per 1°C increase in DTR at different lag days. The largest effect values were found at lag 0 day in single-day lags. A 1°C increased in DTR was associated with 0.73% (95%CI: 0.24, 1.21) increased in outpatient visits of college students for URTI in all, 0.60% (95%CI: 0.03, 1.18) in males, and 0.84% (95%CI: 0.32, 1.36) in females. The estimates with statistically significant were detected in some cumulative lag days, and the greatest effect values were observed in all [1.08% (95%CI: 0.22, 1.95)] and males [1.35% (95%CI: 0.33, 2.39)] at lag 0-6 days, and in females [0.86% (95%CI: 0.24, 1.49)] at lag 0-1 days. As seen in Table S3, the effect of DTR on URTI in students aged 24+ years [0.96% (95%CI: 0.37, 1.57)] at current day was larger than aged 15-23 years [0.58% (95%CI: 0.05, 1.11)]. No matter at single-day or cumulative lags, it was worth noting that the associations of DTR in males and aged 15-23 years were stronger than females and aged 24+ years at longer lag days, respectively. Figure 4 shows the percent changes and 95%CI of the outpatient visits of college students for URTI associated with per 1°C  Fig. 3 The exposure-response curves between outpatient visits of URTI and DTR for all and males at lag 0-6 days, and for females at lag 0-1 days    (Table S4). In spring and summer, the effects of DTR with statistical significance were only observed at lag 0, 3, and 0-1 days. For gender, significant associations of DTR-URTI were only found in females at cumulative lag days in winter, and the effect of DTR on URTI in females [3.07% (95%CI: 0.77, 5.43)] was larger than males [2.15% (95%CI: -0.67, 5.04)] at lag 0-6 days in winter. Table 3 describes the percent changes and 95%CI of the outpatient visits of college students for URTI associated with a 1°C increase in DTR at lag 0-6 days (for all and males) and lag 0-1 days (for females) in sensitivity analysis. Although the effect values were slightly decreased to varying degrees compared with the unadjusted models when adding each particulate matter (PM 2.5 and PM 10 ) one by one, the vast majority of effect values remained statistically significant. However, the associations of DTR-URTI became insignificant when adjusting gaseous pollution (SO 2 , NO 2 , and O 3 ). After changing the degrees of freedom of time, temperature, and relative humidity from 7 to 6 or 8, 6 to 5 or 7, and 3 to 2 or 4, respectively, all the effect values also remained statistically significant. The above tests illustrated the robustness of our model and results to some degree.

Discussion
In the current study, GAM was used to investigate the link of short-term exposure to DTR and the daily number of (A) All  Note: bold results are statistically significant (P<0.05); temp: temperature, relh: relative humidity; based on the effects of diurnal temperature range at lag 0-6 days for all and male, lag 0-1 days for female outpatient visits of college students for URTI at Hospital of Wuhan University in Wuhan, China. After adjusting for multiple covariates in this model, we noticed that with the increase of DTR, the number of outpatients was increasing accordingly. We also found the strongest associations occurred at lag 0 day in single-day lags whatever for all, males, and females. The greatest effect values were observed in all and males at lag 0-6 days and in females at lag 0-1 days. Females and older college students were more vulnerable to extremely acute exposure to DTR than males and students aged 15-23 years. In general, the estimates of the associations between DTR and URTI were larger in autumn and winter, especially in the cumulative lag days. This study might have great significance to the prevention of upper respiratory tract infection in college students and provide scientific epidemiological evidence and reference for relevant administration department to take measures to control temperature change.
Our results showed that per 1°C increase in DTR would elevate 0.73% (95% CI: 0.24, 1.21) and 1.08% (95% CI: 0.22, 1.95) in the daily number of outpatient visits of all college students for URTI at lag 0 and 0-6 days, respectively. These findings were consistent with some previous studies Phosri et al. 2020). For instant, Ge et al. found per 1°C increment of DTR could increase 0.94% (95% CI: 0.34, 1.55) and 1.60% (95% CI: 0.62, 2.58) in emergency-room visits for respiratory tract infection in Shanghai at lag 0 day and at lag 0-5 days, separately (Ge et al. 2013). Li et al. observed that a 1°C increase in DTR corresponded to 2.55% (95% CI: 1.97, 4.01) increase in the number of daily emergency-room visits for respiratory tract infection in children aged ≤5 years old at lag 1 day .
Former studies pointed out that DTR might be an independent risk factor for human health (Kim et al. 2016;Ding et al. 2015). The potential mechanism of action of DTR was not well understood yet, but one research had been suggested that rapid changes in temperature can increase respiratory load and thus induce respiratory health events (Imai et al. 1998). Air with suddenly altered temperature was inhaled into the human body related to upregulate the release of mast cell-associated inflammatory mediators (Togias et al. 1985). In addition, the rapid decrease of temperature in respiratory epithelial would lead to the decrease of the effectiveness of local respiratory defense (for example, mucociliary clearance and leukocyte phagocytosis) (Diesel et al. 1991;Bang 1961). DTR might be an additional independent environmental influence, and its variation can result in the increment of respiratory stress (Ge et al. 2013). The high value of DTR can promote the spread and reproduction of bacteria and viruses (Cheng et al. 2014;Onozuka 2015). Besides, when the temperature of respiratory epithelial fluctuated violently, it might affect the host defense function of the respiratory system, nasal response, and airway mucosal cilia clearance, thus, increasing the incidence of respiratory infections (Graudenz et al. 2006;Diesel et al. 1991). Based on the complexity and diversity of DTR impacts, we called for more researches to explore its exact mechanism.
In our results, it was worth noting that the effect values of per 1°C increment of DTR in females [0.84%, 95%CI: (0.32, 1.36)] for URTI at current day was higher than males [0.60%, 95%CI: (0.03, 1.18)]. However, in a longer lag period, the males emerged greater effect values than females. For instant, a 1°C increased in DTR associated with 1.35% (95% CI: 0.33, 2.39) elevated in the daily number of outpatient visits of male college students for URTI, but only 0.84% (95% CI: -0.09, 1.77) in female college students at lag 0-6 days. One study observed the relative risks of DTR and hospital admission for cardiovascular in males from lower than females at lag 0 day [1.044 (0.991, 1.101) for males vs. 1.058 (1.004, 1.116) for females] to higher than females at lag 0-7 days [1.151 (1.022, 1.296) vs. 1.115 (0.988, 1.257)], which were similar to our results (Phosri et al. 2020). Nevertheless, researchers also found females were more likely to be vulnerable to DTR than males at all lag days (Zhang et al. 2017b;Zhou et al. 2014). The possible reason for such inconsistent results was that the study area (climate conditions) and population (economy and education) were different (Basu 2009;Ding et al. 2015).
Among all the results with statistical significance in this study, we observed that the effect values in autumn and winter were higher than the other two seasons. Our findings were similar to some recent studies investing the associations of DTR and respiratory diseases (Kan et al. 2007;Ma et al. 2018). A research conducted in Shanghai pointed out the effects of DTR on respiratory mortality were only observed in cold days (daily mean temperature <23°C) (Kan et al. 2007). Ma et al. found that per 1°C increase in DTR was associated with increasing the risk of hospital admissions for COPD by 11.5% (95%CI: 10.0,12.9) in the cold season (October to April in the next year), which was stronger than 9.0% (95%CI: 7.7, 10.3) in the warm season ). However, a positive and stronger relationship was revealed between DTR and childhood asthma exacerbations in the warm season (May to September) , which was inconsistent with our conclusion. The results that appeared in the current study can be partially explained as follows: (1) the value of DTR was at a relatively high level in autumn, which usually had a stronger association with the increasing risks of human health (He et al. 2018); (2) the prevail peak period of human metapneumovirus and human parainfluenza virus causing URTI usually occurred in autumn and winter (Debur et al. 2010;Liu et al. 2013).
In some previous epidemical studies focused on DTR, air pollution was not included in the main model and sensitive analyzes Zhang et al. 2017a, b). Some researches regarded air pollution as a part of main model, others only added in sensitive analyzes (Ge et al. 2013;Luo et al. 2013;Phosri et al. 2020). Considering the confounding/modification effects of air pollution on DTR-mortality/morbidity association remain mixed and there was no widely accepted statistical model, we did not add air pollution in the main model in this study and just adjusted them as sensitive analyzes to estimate the robustness of our model (Byun et al. 2020). As seen in Table 3, although the effect values were slightly decreased, the most effects of DTR remain significant after adding particular matter (PM 2.5 and PM 10 ), which might imply that the relativity robustness in our study and DTR was an independent risk factor for the number of outpatient visits for URTI to some degree. However, due to the potential existence of collinearity, the effects of DTR became insignificant after adding gaseous pollution (SO 2 , NO 2 , and O 3 ).
In this study, the exposure-response curves between DTR and the daily outpatient visits of URTI in all, male, and female college students were close to linear upward trends. These linear-like trends were also found in previous studies (Wang et al. 2020;Zheng et al. 2016). Therefore, the generalized additive models were used by us to investigate the associations of DTR and the outpatient visits for URTI. However, in some other studies, U-or J-shaped exposure-response curves between DTR and the occurrence of adverse outcomes were revealed Hu et al. 2018;Zhao et al. 2017). A study conducted in Hefei even showed a slightly M-shaped exposure-response curve between DTR and the risk of admissions for tuberculosis ). The reasons for such inconsistent results might be that the types of population and disease in these studies were very different.
The strongest effect values of DTR were found at the current day in the models of single-day lags, whether for male, female, or all college students. Likewise, a time-series study in Beijing indicated that the adverse effects of DTR peaked at lag 0 day [per 1°C increased in DTR corresponded to 0.58% (95%CI: 0.02, 1.15) increase in emergency room admissions for respiratory disease] (Wang et al. 2013). A study conducted in Changchun, one northeast city of China, reported that the largest relative risk of 1.023 (95%CI: 1.004, 1.042) about hospital admissions for the chronic obstructive pulmonary disease was observed at lag 7 day in the males when 1°C increased in DTR , which was inconsistent with our results.
In summary, our study has several advantages: (1) according to the information I have, this is the first study exploring the associations between DTR and upper respiratory tract infection among college students in China. (2) University is one of the most important stages of learning. College students with upper respiratory tract infection will seriously affect the efficiency of learning. Therefore, it is necessary to investigate the relationship between DTR and upper respiratory tract infection. (3) The population of this study all live in Wuhan University, an area of about 3.5 square kilometers, and they will stay in school most of the year because of their studies. So, it is reasonable to suppose that they have similar exposure backgrounds. (4) The hospital of Wuhan University had set up four branches in four academic departments to facilitate college students with different majors and faculties to seek basic medical treatment. In addition, benefit by the medical insurance policy of Wuhan University, the medical expenses of our college students in the Hospital of Wuhan University are all reduced 90%, which greatly avoids the bias of not choosing to seek medical treatment due to the cost (Zhang et al. 2021) and makes the exposure-response relationship of DTR and the daily outpatient visits of URTI in this study more genuine and believable.
Nevertheless, there are several limitations of this study that must be mentioned: (1) data about meteorological factors and air pollutants were collected from monitoring stations with a certain distance from Wuhan University, and individuals' indoor and outdoor activity duration cannot be measured, which would inevitably lead to misclassification in exposure to DTR to a certain degree. (2) Due to the unavailability of data about more detailed personal information, for example, body mass index (BMI), eating habits, smoking status, and history of diseases, which would result in unavoidable errors. (3) The outpatient visit data provided from the hospital of Wuhan University do not contain patients' name because of medical ethics, therefore, we cannot distinguish the cases of repeated visits and regard each visit record as a new case. (4) Since not available for obtaining the other comorbid status of outpatients, it might cause a bias. But considering college students rarely suffered from basic diseases, the bias was slight. (5) Owing to the only one hospital was selected, the results might be a little less representative for the whole city (Chen et al. 2019). More comprehensive researches were warranted to reveal the associations between DTR and URTI.

Conclusions
In conclusion, we investigate the short-term effects of DTR on the daily number of outpatient visits in college students for URTI from 2016 to 2018 in Wuhan University, China. Results in this study show that short-term exposure to DTR was linked with the increased risk of outpatient for URTI among all college students at the Hospital of Wuhan University, which meant DTR was a novel indicator to estimate the relationship between environment and disease. Males were more susceptible to DTR than females in the longer lag days. DTR had more adverse health impact in autumn and winter. Consequently, public health departments should consider not only the health impact of daily mean temperature but also the DTR, to formulate better preventive measures.
Availability of data and materials Not applicable.
Author contribution Wei Zhu conceived and designed the study; Chuangxin Wu and Miaoxuan Zhang collected and cleaned the data; Faxue Zhang performed the data analysis and drafted the manuscript. Huan Feng and Han Zhang helped revise the manuscript. All authors read and approved the final manuscript.
Funding Not applicable.

Declarations
Ethics approval This study was approved by the Ethics Committee of Wuhan University.
Consent to participate Not applicable.

Consent for publication Not applicable.
Competing interests The authors declare no competing interests.