Introduction

Acute respiratory tract infection (ARI) is one of the most common infectious diseases in children. It has been estimated that, regardless the geographic location or economic status, young children experience 3–6 episodes of ARI on average every year [1]. ARI may affect children’s upper respiratory tract (from the nostrils to the vocal cords in the larynx including the sinuses and the middle ear) or lower respiratory tract (from the trachea and bronchi to bronchioles and alveoli). Children with ARI may also be subject to the risks of extended infection, inflammation, and reduced lung function. It remains one of the major causes of pediatric outpatient and inpatient care, hospital referral and admission, mortality, and morbidity in children under 5 years of age.

Viral pathogens are largely responsible for ARI in children. The positive rate of viral infections among ARI in children ranges from 32.3 to 70% [2,3,4,5,6] and can reach 80% or above in upper respiratory tract infection [7, 8]. More than 200 viral serotypes were found to be associated with human respiratory diseases [9] and the most frequently reported virus found in ARI among young children included influenza A (IFV-A), influenza B (IFV-B), respiratory syncytial viruses A and B (RSV), human coronaviruses 229E, OC43, HKU1 and NL63 (hCoV), human metapneumoviruses (hMPV), human parainfluenza virus types 1, 2, 3, and 4 (PIV-1, PIV-2, PIV-3, and PIV-4), human enteroviruses (EV)/human rhinoviruses (HRV), human adenoviruses (ADV), and human bocavirus (hBoV) [10,11,12]. The prevalence of respiratory virus is region-specific [13,14,15]. For instance, in China, the most common virus causing ARI in children varied across different areas and was found to be HRV in Chengdu, RSV in Beijing, HRV/EV in Shanghai, and HCoV in Guangzhou [16,17,18,19].

Previous studies have shown that the viral etiology of ARI in children is subject to the impact of outdoor environments including meteorological factors (temperature, humidity, solar radiation and wind speed) [20, 21] and air pollutants (such as Particulate Matter < 10 μm (PM10), Particulate Matter < 2.5 μm (PM2.5), nitric dioxide (NO2), and ozone (O3)) [22, 23]. However, the results about the effect of meteorological factors and air pollutants on the ARI viral etiology are inconclusive. For instance, while epidemiology studies in China and the US showed that positive associations existed between PM2.5 exposure and ARI in children [23, 24], other studies found no association between them [25, 26]. At present, few studies have been conducted on children in the subtropical marine monsoon climate zone. Therefore, this study aimed to explore the viral etiology among children hospitalized for ARI and the association of meteorological factors and air pollutants with children hospitalization due to viral ARI.

Methods

Study site

Macao, a territory with only 32.8 km2, is located on the subtropical coast of China and has a typical subtropical marine monsoon climate featured with rich heat, high humidity, warm temperature, and heavy seasonal rainfall, featuring an average annual temperature of 22.6 °C, mean relative humidity of 78.8% and total precipitation of 2058.1 mm [27]. The air quality of Macao is heavily influenced by external factors such as human movement and activities in the neighboring Guangdong province [28].

Data sources and participants

The demographic, epidemiological and clinical information of eligible patients was extracted from the electronic health records of Kiang Wu Hospital—one of the major hospitals in Macao [29]. Inclusion criteria were: (1) inpatients aged between 1 month and 14 years; (2) admitted to Kiang Wu Hospital for ARI between 2014 and 2017; and (3) the patient profiles must have complete results of the nasopharyngeal swab tests. The diagnosis of ARI was made according to the clinical guidelines recommended by the World Health Organization (with at least two of the following: fever, sore throat, cough, rhinorrhea, nasal congestion and hoarseness). Patients with congenital pneumonia (ICD-10 code: P23), nosocomial infections (ICD-10 code: Y95) or chronic tuberculosis (ICD-10 code: A15) were excluded.

Multiplex polymerase chain reaction tests

The nasopharyngeal swab specimens were collected from the patients at admission using a special swab (ESWAB from Copan Italia SpA) by nursing staff and transported to the laboratory immediately at a storing temperature of 20–25 °C for testing within 1 h of collection. A qualitative nucleic acid multiplex test (The xTAG® Respiratory Viral Panel FAST v2, Luminex) was performed to test the following respiratory viruses: IFV-A, IFV-B, RSV, hCoV, hMPV, PIV-1, PIV-2, PIV-3, and PIV-4, EV/HRV, ADV, and hBoV. Nucleic acid extraction was performed using the EZ1 DSP Virus Kit (Qiagen, Germany).

Meteorological and air pollutant data

Meteorological and air pollutant data was obtained from the Macao Meteorological and Geophysical Bureau. Meteorological data included daily mean temperature (°C), daily mean relative humidity (%), daily mean solar radiation duration (h), and daily mean wind speed (km/h). Air pollutant data included daily level of PM10 (μg/m3), PM2.5 (μg/m3), NO2 (ppb) and O3 (ppb).

Statistical analysis

Chi-square test was used for comparisons between groups. Spearman correlation analysis was used to evaluate the associations between the meteorological and air pollution factors. A value of P < 0.05 was considered statistically significant. All the analyses were performed with the Statistical Package for the Social Sciences (SPSS) V23.0. A distributed lag nonlinear model (DLNM) was used to assess the nonlinear and lagged effects of meteorological factors and air pollutants on ARI hospitalizations as demonstrated in Eq. 1 shown in the following:

$${\text{LogE}}\left[ {{\text{Yt}}} \right] = {\text{cb}}\left( {{\text{Temp}}} \right) + {\text{cb}}\left( {{\text{RH}}} \right) + {\text{cb}}\left( {\text{Wind speed}} \right) + {\text{cb}}\left( {{\text{solar}}} \right) + {\text{cb}}\left( {{\text{PM}}_{{{1}0}} } \right) + {\text{cb}}\left( {{\text{O}}_{{3}} } \right) + {\text{s}}\left( {{\text{day}}} \right) + \alpha {\text{Dowt}}$$
(1)

where Yt refers to the counts of cases occurred on day t; cb refers to the cross-basis function built up using the DLMN() package in R, combining functions for both exposure and lag dimensions; a maximum lag of 14 days was applied to reflect the incubation period of 3–10 days for ARI; s() refers to the natural cubic spline function to depict long-term time trend; Dowt refers to the day of the week on day t; α is the coefficient of the corresponding terms.

Sensitivity analyses were conducted by applying different df for the lag terms to the models to observe the robustness of the results. All analyses were performed with the DLNM packages in R software version 3.5.3 (https://www.r-project.org/). The significance level was set at P < 0.05.

Ethics

The study was approved by the Ethics Committee of Kiang Wu Hospital, Macao (Reference number: 2020-003). Considering that all data was collected retrospectively and anonymized in a standardized case report form in the hospital database, informed consent from the patients was neither feasible nor deemed necessary, and was therefore exempted by the committee.

Results

As shown in Table 1, there were 4880 children admitted for ARI to Kiang Wu Hospital between Jan 2014 and Dec 2017, of them 2070 cases (55%) were male, 2462 cases (50.5%) were of toddler age, and 2712 cases (55.6%) were admitted in spring or summer.

Table 1 Descriptive statistics of ARI admissions
Full size table

Viral etiology

Out of the 4880 ARI-related admissions, 3767 cases (77.2%) were tested positive for at least one virus (Table 1). Single infection was detected in 3091 (63.3%) cases. The prevalence of viral infection was higher among male (n = 2120, 56.3%). The toddlers’ group had the highest detection rate and the school-age group had the lowest (χ2 = 119.092, P < 0.001). The major diagnosis of viral ARIs included upper respiratory tract infection (n = 1805, 47.9%) and pneumonia (n = 833, 22.1%). Overall, the detection of viral infection peaked in spring (n = 1059, 29.1%) (Fig. 1).

Fig. 1
figure 1

Number of ARI-related admissions tested positive for viral infection during the study period (2014–2017)

Full size image

In descending order of detection rate, the most common viruses were EV/HRV, ADV, RSV (RSV-A and RSVB), IFV, PIV, hMPV, hBOV and hCOV (Fig. 2a). On a seasonal basis, EV/HRV, ADV and RSV remained the most common viruses across throughout the study period (Fig. 2b). Proportionally, some viruses were detected more often at times during the study period. For instance, hMPV was less common in autumn than in other seasons while hBOV was more common in autumn than in other seasons.

Fig. 2
figure 2

Viral etiology of children hospitalization due to viral ARI

Full size image

The detection rates of single infection and multiple infections are shown in Fig. 3. hBOV has the highest multiple infection rates (67.3%, 169/251), followed by hCOV (49.6%, 65/131) and EV/HRV (35.2%, 470/1336).The most frequent combination of multiple infections is hBOV and EV/HRV (32.7%, 82/251).

Fig. 3
figure 3

Detection rates of single and multiple viral infections among children hospitalization due to viral ARI

Full size image

Meteorological factors and air pollutants

The official metrological and air pollutant data dated between 2014 and 2017 (including temperature, humidity, solar radiation, wind speed, and the air pollutants) is presented as the average value per unit, and the average of the 5th, 25th, 75th and 95th percentiles as shown in Table 2.

Table 2 Descriptive statistics of meteorological and air pollution factors
Full size table

The association of meteorological factors and air pollutants with the incidence of viral infection among children hospitalized for ARI.

Using spearman analysis, the linear relationship between virus infection and the metrological factors was not significant and the correlations identified were only weak positive or weak negative (Additional file 1: Appendix 1). Considering the linear analysis could not reflect the impact of changes in weather and air factors on the risk of virus infection, DLNM was, therefore, employed to analyze the association of meteorological and air pollutant factors with children hospitalization due to viral ARI. With DLNM, our study showed a nonlinear relationship and lagged effect between daily meteorological variables and viral infection incidence.

The effect of temperature change on the incidence of viral infection among children hospitalized for ARI

Daily mean temperature had the most significant influence. In Fig. 4, the overall effect of temperature is shown with a 3-D graph of the relative risk (RR) along with temperature and lags compared with a reference value of 10 °C, the point of overall minimum incidence.

Fig. 4
figure 4

A 3D plot of RR along with temperature and lags

Full size image

Figure 5 showed the RR by lag at 4 specific temperatures (13.3, 18.2, 24.3, 29.5 °C), corresponding approximately to 5th, 25th, 75th and 95th percentiles of temperature distribution. Figure 6 confirmed a delayed effect, which suggested that the effect of short-term changes in temperature on viral infection was 1–5 days. By using the appropriate lag time, the association between daily mean temperature and the risk of viral respiratory infection was discovered. The maximum effect of temperature was reached on day 1 (Fig. 5). The RR of viral infection incidence increased with the daily mean temperature, and then plateaued for a temperature higher than 23 °C. The lagged effect lasted for about 5 days.

Fig. 5
figure 5

A plot of RR by temperature at 5th, 25th, 75th and 95th percentiles of temperature distribution

Full size image
Fig. 6
figure 6

A plot of RR by temperature at specific lags

Full size image

The effect of relative humidity on the incidence of viral infection among children hospitalized for ARI

Figure 7 shows the 3D plot of RR along with humidity and lags. The plot showed a rapid effect of relative humidity on the incidence of viral infection. The RR increased when the relative humidity was < 70% and peaked at 50% (Fig. 8). The lagged effect lasted for about 3 days.

Fig. 7
figure 7

A 3D plot of RR along with humidity and lags

Full size image
Fig. 8
figure 8

A plot of RR by humidity at specific lags

Full size image

The effect of solar radiation on the incidence of viral infection among children hospitalized for ARI

Figure 9 was the 3D plot of RR along with solar radiation and lags. The 3D plot and the plots of different lags showed that the effect of solar radiation on viral activity was insignificant (Fig. 10).

Fig. 9
figure 9

A 3D plot of RR along with solar radiation and lags

Full size image
Fig. 10
figure 10

A plot of RR by solar radiation at specific lags

Full size image

The effects of PM10, PM2.5 and NO2 on the incidence of viral infection among children hospitalized for ARI

Since PM10, PM2.5 and NO2 are highly correlated according to the above spearman’s correlation analysis, we used PM10 to represent the 3 pollutants in this study. The 3D plot of RR along PM10 and lags is provided in Fig. 11. The plot showed a very strong and immediate effect of PM10. The RR increased significantly when the PM10 level was over 150 ug/m3. The lagged effect lasted for about 5 days (Fig. 12).

Fig. 11
figure 11

A 3D plot of RR along PM10 and lags

Full size image
Fig. 12
figure 12

A plot of RR by PM10 at specific lags

Full size image

The effects of O3 on the incidence of viral infection among children hospitalized for ARI

The 3D plot of RR along O3 and lags was provided in Fig. 13. The plot showed a very strong and immediate effect of O3. The RR increased significantly when the O3 level was over 80 ppb. The lagged effect lasted for about 3 days (Fig. 14).

Fig. 13
figure 13

A 3D plot of RR along O3 and lags

Full size image
Fig. 14
figure 14

A plot of RR by O3 at specific lags

Full size image

Discussion

In this study, we found that the detection rate of viral respiratory infection among children hospitalized for ARI was 77.2%, while previous studies showed that about 31.2–86% of respiratory infection was caused by virus [13, 15, 30]. The most prevalent pathogens among admitted children in this study were EV/HRV (1336, 27.4%), ADV (771, 15.8%), and RSV (756, 15.4%). To analyze the association of meteorological and environmental factors with viral etiology, we constructed a DLNM model that revealed the lagged effect of short-term changes in meteorological environmental factors on the incidence of virus infection among children hospitalized due to ARI in Macao. Collectively, the study findings not only provide the health authority with local information to inform health policies, but also supplement the current understanding about the impact of meteorological and environmental factors on the viral etiology of ARI among children.

The high incidence of viral ARI found in the toddler group was similar to the results of the Korean study [13] and was probably related to the immature immune system of younger cases and the high risk of cross-infection in nurseries. Multiple infections were found to be 13.9% (676 cases) which fell within the range 4.9–42% determined previously [13,14,15, 30, 31]. hBOV had the highest rate of multiple infections, which was also consistent with previous research [32, 33]. The toddler group had the highest multiple infection rates, similar to a Spanish study [34].

Scotta et al. found that multiple viral infection did not influence clinically important outcomes such as length of stay, length of supplemental oxygen, need for hospitalization, supplemental oxygen, need of intensive care, mechanical ventilation and death [35]. Similarly in this study, no statistical significance was found between the multiple infections and the clinical presentation of bronchiolitis, pneumonia or croup. Recent findings suggested that one virus infection could be blocked by the presence of another through different mechanisms, such as resource competition of host cells, immune response, or interference through viral proteins [36]. Therefore, multiple viral infection was not necessarily associated with more severe illness [37].

Meteorological and environmental factors have been shown to have important influence on human health, especially the respiratory system and the relationship between them were not necessarily linear [38]. As found in this study, the effect of meteorological and pollution factors was often delayed in time. This situation occurs frequently when assessing the short-term effects of environmental changes and may cause a harvesting effect: a phenomenon that arises when a stressor affects a specific population. After some periods with excess mortality, however, a decrease in overall mortality during the subsequent periods had also been observed [39]. Therefore, linear analysis and correlation analysis may not be very useful for reflecting the effects of meteorological factors on respiratory diseases. Recently, the distributed non-linear model has been widely used to describe the non-linear relationship and delayed effect between meteorological factors and health outcomes [40, 41].

In this study, the daily mean temperature, relative humidity, and air pollutants demonstrated important influence on the incidence of viral respiratory infection in children, while the effect of solar radiation was found to be insignificant. The effect of temperature was most significant on day 0–5 and peaked on day 1. The RR increased with the daily mean temperature, and then plateaued for a temperature higher than 23 °C. A rapid effect of relative humidity on viral activity was also noted in which the RR increased when the relative humidity was < 70% and peaked at 50%. This suggests that low humidity can be a risk factor for viral respiratory infection in children. However, the effect of solar radiation on the risk of viral infection incidence is insignificant.

High air pollutant level has been associated with increased asthma attacks and emergency department visits caused by respiratory diseases [42,43,44]. In recent years, studies of viral respiratory infection among children in different regions had been published [45, 46]. Since the environment varied in different regions, for the first time we tried to explore the association between meteorological factors and viral infection among children hospitalized for ARI in Macao. Air pollutants were known to have great influence on the risk of viral respiratory infection among children. In Macao, the levels of PM10, PM2.5 and NO2 were positive correlated. Elevated level of air pollutants can be harmful to children’s respiratory tracts and cause increased risk of viral infection [23, 45,46,47]. Fernando et al. described that fine particulate matter can cause inflammation of airway mucosa and airway clearance dysfunction [48].

Fine particulate matters in human alveolar epithelial cell A549 can induce oxidative stress and inflammation. Oxidative stress was characterized by elevated intracellular levels of reactive oxygen species (ROS) and production of antioxidant enzyme (superoxide dismutase (SOD), catalase (CAT) and hemeoxygenas-1 (HO-1)). During oxidative stress, TNF-α and IL-6 levels increase and cause cell damage. Cell damage is characterized by decline in cell activity, release of LDH increases, formation of apoptotic body and necrosis. All these eventually lead to inflammation of lung tissue and decrease local resistance to infection. Our study confirmed that high levels of air pollutants were associated with an increased risk of viral infection and the lag effect lasted for 3–5 days. Hence, if high level of air pollution was detected, the government could advise the public to avoid unnecessary outdoor activities in the next 5–7 days, in order to reduce the delayed effect on the incidence of respiratory viral infection.

This study has several limitations. Since only a relatively small sample size was included in this study, more regional information and studies are needed to support our conclusion. Second, we were unable to gather individual-level data such as socioeconomic status or secondary smoking status due to the nature of the data. Third, we were not able to control the unmeasured confounding factors. The study findings may only provide a snapshot of the temporal fluctuations in the associated identified. Thus, our results may be at best generalized to the population in Macao.

In order to inform the management of viral ARI among children in the future, further studies may focus on the following areas: (1) as some common viruses had been identified, the association of meteorological factors and air pollutants with specific virus responsible for viral ARI may be assessed; (2) the severity of the symptoms caused by viral ARI among hospitalized children should also be taken into account when analyzing the viral etiology and its associations with meteorological factors and air pollutants; (3) similar analysis of the associations may be conducted by season to determine if there are any seasonal variations; and (4) the effect of meteorological factors and air pollutants on the symptoms experienced by children hospitalized for viral ARI.

Conclusion

In conclusion, EV/HRV, ADV, RSV and IFV are common viral pathogens of respiratory infection among children hospitalized for ARI in Macao. Multiple infections of viruses are very common and hBOV has the highest multiple infection rate. Daily mean temperature, relative humidity and air pollutants (PM10, PM2.5, NO2 and O3) have significant effect on the risk of viral respiratory infection. The risk of viral infection increases with the level of air pollutants. As the impact of meteorological factors and air pollutants on the viral etiology of ARI in children is region-specific, the findings of this study have important implications for public health policy to mitigate the risk of viral infections under certain environmental influences.