Association of PM2.5 and its components with lengths of hospital stay for hand foot and mouth disease in children

Hand foot and mouth disease (HFMD) is a widespread public health concern but the studies on air pollution and the lengths of hospital stay (LOS) of HFMD are scarce nevertheless. Clinic demographic features among 5135 hospitalized HFMD cases in Nanjing, China, had been characterized from 2012 to 2017. Then, we had analyzed the association between PM2.5 short-term exposure as well as its components (OM, BC, SO42–, NH4+, NIT, SOIL, and SS) and the LOS of HFMD. Among these cases that were involved in our study, 98.62% were aged 0–6 years old, and 3772 (73.46%) were hospitalized within 1 week or less. The LOS of HFMD patients was different in various age ranges, illness onset years, and illness onset seasons (P < 0.01). For per IQR increase in PM2.5 concentrations, LOS of HFMD increased by 0.52 (0.33, 0.71), 0.50 (95% CI, 0.31–0.69) and 0.46 (95% CI, 0.28–0.65) day in adjusted models at lag 3 days, lag 7 days, and lag 14 days, respectively. In addition, per IQR increase of BC, SO42−, NH4+, NIT, and SOIL was also significantly associated with the LOS of HFMD. Our findings corroborated that short-term PM2.5 exposure was associated with the increased LOS of HFMD, and its components (BC, SO42−, NH4+, NIT, and SOIL) of PM2.5 might play a key role in prolonged LOS of HFMD.


Introduction
Hand foot and mouth disease (HFMD) is a public health concern (Chang et al. 1998;Ma et al. 2010). It is caused by a variety of human intestinal viruses, most commonly by enterovirus 71 (EV71) and coxsackievirus A16 (CVA16) (Blomqvist et al. 2010;Han et al. 2014;Mirand et al. 2012), and is a common infectious disease in infants and young children under 5 years of age with fever and rash or herpes on hands, feet, mouth, and other parts (Repass et al. 2014). This disease is distributed globally and widespread in Asia, especially in the East and South Asia (Duan et al. 2018;Hu et al. 2012;Koh et al. 2016;Wang et al. 2013;Zou et al. 2012). In May 2008, China made HFMD as a statutory infectious disease, and the National Health and Health Committee of China officially issued the "Hand foot and mouth disease Treatment Guide (2018 Edition)." During 2008-2015, a total of 398,010 HFMD patients with > 2 episodes were reported among children in China (Huang et al. 2018).
Due to the long survival period of HFMD-related intestinal virus in ambient air, air pollutants may play a critical role in HFMD epidemics (Hao et al. 2020). Previous studies have indicated that air pollutants are significantly related to Communicated by Lotfi Aleya.
Wei Li and Jieguo Wang contributed equally to this work. HFMD incidence in children (Huang et al. 2019). As PM 2.5 is one of the major air pollutants in the world and severe to health, good air quality might protect against HFMD (Yu et al. 2019). Improving air quality, especially decreasing PM 2.5 , could decrease the risk of HFMD outbreaks (Zhong et al. 2018). PM 2.5 is a complex mixture of inorganic and organic compounds, and the main components of PM 2.5 include organic matter (OM), black carbon (BC), sulfate (SO 4 2− ), ammonium (NH 4 + ), nitrate (NIT), SOIL, sea salt (SS), trace element oxides, and others (Li et al. 2016;Nunez et al. 2021;Tian et al. 2020). Most of the research indicated that an exceedance of PM 2.5 level of the standards could lead to varying degrees of health hazards based on PM 2.5 chemical constituents (Mukherjee and Agrawal, 2018).
Although the most of HFMD continue for 7 to 10 days, some severe cases still require prolonged hospitalization. The longer lengths of hospital stay (LOS) of HFMD will increase the chance of infection in hospital and increase the cost of hospitalization. Previous studies mostly have focused on adult hospital admissions or mortality; however, only a few of studies have focused on the association between PM 2.5 and its chemical components with the LOS of HFMD. In our study, by characterizing the clinical demographic features of hospitalized HFMD on age, gender, illness onset year, illness onset season, and geographical patterns in Nanjing, China, we evaluated PM 2.5 and its main component exposure condition among these chosen cases, which aimed to identify the association between short-term exposure of PM 2.5 and its components and the LOS of HFMD in order to indicate the hypothesis: Specific chemical component of PM 2.5 is responsible for PM 2.5 associated the LOS of HFMD.

Population
The clinical data that had been collected from Children's Hospital of Nanjing Medical University between January 2012 and December 2017 were analyzed retrospectively. Totally, 5135 hospitalized children with HFMD were diagnosed by professional pediatrician. Their medical histories had been abstracted from medical records. The study was approved by the Ethics Committee of Children's Hospital of Nanjing Medical University. Data used in this study were anonymous, and no individually identifiable information was available here.

Clinical demographic features of participants
After admission, a face-to-face interview was conducted by professional pediatricians to collect data on children's characteristics, including age, gender, illness onset season, and illness onset year. All HFMD cases were diagnosed, excluding other infectious diseases. The LOS was recorded for each participant.

Modeling PM 2.5 and its component concentrations
The geo-coded residential addresses of all 5135 HFMD cases were linked to the average modeled concentrations of PM 2.5 and its constituents. We evaluated the residential children to PM 2.5 and 7 main constituent (OM, BC, SO4 2-, NH4 + , NIT, SOIL, and SS) exposures among residential children had been measured in 3, 7, and 14 days prior to hospitalization by using the V4.CH.02 product of the Dalhousie University Atmospheric Composition Analysis Group (ACAG) (van Donkelaar et al. 2016(van Donkelaar et al. , 2019.

Statistical analysis
Firstly, the characteristics of age, gender, illness onset season (spring (from March to May), summer (from June to August), autumn (from September to November), and winter (from December to February)), illness onset year, and the LOS of 5135 hospitalized HFMD children who live in Nanjing were estimated. Cases' residential addresses were collected from hospital medical records. The geographic location of children who were at home was converted to latitude and longitude and was described in Fig. 1. Then, we explored age, gender, illness onset season, and illness onset year parameters by specific LOS. Correlation between each component of PM 2.5 was calculated by Pearson correlation analysis. Multiple linear regression models were performed to examine the associations between (1) per inter-quartile range (IQR) increase in PM 2.5 and its components and the LOS of HFMD in lag 3 days; (2) per IQR increase in PM 2.5 and its components and the LOS of HFMD in lag 7 days; and (3) per IQR increase in PM 2.5 and its components and the LOS of HFMD in lag 14 days. Model 1 was unadjusted; however, Model 2 was adjusted for age, gender, illness onset year, and illness onset season. FDR corrections were conducted to correct the P-values due to multiple comparisons. Meanwhile, we also use generalized weighted quantile sum (gWQS) method to further identify harmful components. All analyses were performed by using R software (version 3.6.1, R Core Team 2019).

Study population characteristics
We had described the characteristics of entire 5135 hospitalized HFMD children. Most cases were boys (64.32%). The mean age was 1.99 years (standard deviation [SD]: 1.48) (Table 1), and 5064 (98.62%) were aged 0 to 6. Among all patients, 3772 (73.46%) were hospitalized for 7 days or less. There were two epidemic cycles during summer and autumn. All of the patients who were involved in the study were discharged from the hospital with a zero mortality.
According to Table 2, the results showed that the LOS of HFMD children was different in various age ranges, illness onset years, and illness onset seasons (P < 0.01). The LOS of HFMD between genders was not significantly different (P = 0.91). Table 3 shows the average distribution of PM 2.5 and its component (OM, BC, SO 4 2− , NH 4 + , NIT, SOIL, and SS) exposure among children in 3, 7, and 14 days prior to hospitalization. During 3, 7, and 14 days before hospitalization, the median concentrations were 50.00 μg/m 3 of PM 2.5 , 50.00 μg/ m 3 , and 50.02 μg/m 3 , respectively. The detailed distributions of PM 2.5 and its components for 3, 7, and 14 days before hospitalization are presented in Table 3. The correlations between PM 2.5 and its components are shown in Fig. S1,

The association between PM 2.5 and its component exposure and the LOS of HFMD
As shown in Fig. 2 and Table S1, Tables S2 and S3, the crude and adjusted estimate values and 95% confidence interval (95% CI) per IQR of PM 2.5 increase on the hospital LOS were 0.15 (0.02, 0.28) and 0.52 (0.33, 0.71) at lag 3 days. For lag 7 days, per IQR increase of PM 2.5 corresponded to a 0.15 (95% CI, 0.02-0.29) and 0.50 (95% CI, 0.31-0.69) increase in the LOS in crude and adjusted models, respectively. For lag 14 days, an IQR increase in PM 2.5 concentrations corresponded to a 0.17 (95% CI, 0.04-0.30) and 0.46 (95% CI, 0.28-0.65) increase in the hospital LOS, respectively.
In unadjusted model, we noted that positive associations of per IQR increase of OM, SO 4 2− , NH 4 + , and SS with the LOS of HFMD at lag 3, 7, and 14 days. There was no association between BC, NIT, and SOIL with the LOS of HFMD at lag 3, 7, and 14 days. After adjusting age, gender, illness onset year, and illness onset season, we found that per IQR increase of BC, SO 4 2− , NH 4 + , NIT, and SOIL was significantly associated with the LOS of HFMD at lag 3, 7, and 14 days, but OM and SS were not associated with the LOS of HFMD at above three time lags. The detailed data of the relationship between PM 2.5 and its components and the LOS of HFMD at lag 3, 7, and 14 days are shown in Tables S1, S2, and S3. The results of the subgroup analysis according to gender are shown in Tables S4 and S5. The boy group showed a positive association between PM 2.5 and the LOS of HFMD; however, there is no significant relationship between PM 2.5 and the LOS of HFMD in the girl group.

WQS regression models to assess the association between PM2.5 components and the LOS of HFMD
The WQS indexes were statistically associated with PM 2.5 component co-exposure and the LOS of HFMD at lag 3, 7, and 14 days. In the adjusted models, the WQS index was statistically significantly associated with PM 2.5 component co-exposure and the LOS of HFMD (β (95% CI) = 1.78 (1.39, 2.27), 1.63 (1.30, 2.04), 1.53 (1.24, 1.90)) at lag 3, 7, and 14 days, respectively. Detailed results of robust and adjusted models are presented in Table S6. To further analyze WQS model regression index weights of each component of PM 2.5 , we found that SOIL and SO4 2− may have a closer association with the LOS of HFMD positivity than the other components, and the weight of each component is reported in Fig. 3.

Discussion
Pollutants such as PM 2.5 are polluting the environment seriously, causing numerous health problems (Manisalidis et al. 2020). Many time-series studies have used PM 2.5 as an exposure indicator (Atkinson et al. 2014). However, HFMD hospitalization caused by PM 2.5 has been reported rarely so far. To our knowledge, the topic of the clinical demographic features by the LOS is scarce in Chinese children. Long-term LOS will not only seriously reduce the turnover rate of scarce bed resources in the hospitals and reduce the throughput of patients but also increase the financial burden on the families of children (Cosgrove 2006;Rahmqvist et al. 2016). Analyzing the LOS of HFMD is important for hospital administrators to formulate countermeasures, improve hospital work efficiency, and control excessive medical expenses. Therefore, we described the clinical demographic features and the LOS of HFMD in Children's Hospital of Nanjing Medical University during 2012-2017. Meanwhile, we conducted the relationship between the exposure of PM 2.5 and its components and the LOS of HFMD. The incidence of HFMD in China has been reported to be 1-2 per 1000 people (Xing et al. 2014). In our research, the incidence of hospitalized HFMD peaked in children who aged 1 and then decreased with age, and over 98% of hospitalized HFMD cases occur in children younger than 6 years old, which was consistent with the findings in other studies (Baek et al. 2020b;Chen et al. 2007;Podin et al. 2006). Therefore, measures must be taken to prevent  HFMD in these key population groups. There was a difference in gender-specific hospital admission; the same results were reported in previous study Hoang et al. 2019;Wang et al. 2020). Although infection rates between males and females are comparable, males are more likely to develop symptoms, more involved in the propagation of outbreaks, and more likely to be brought in for medical care than females (Baek et al. 2020b), and the reason for the differences observed in gender is not known exactly. The cases of hospitalized HFMD tended to arise in the warmer season (summer and autumn) of the year. Nanjing has strong sunshine, high temperature, and heavy rainfall in summer and autumn, and it has a more serious greenhouse effect. We estimate the sunshine, temperature, and humidity which might explain the seasonality of HFMD (Huang et al. 2013, Onozuka andHashizume, 2011).
The number of hospitalized HFMD cases rose to the highest peak in 2013 and then slightly decreased from 2014 to 2017; especially in 2017, the number of hospitalized HFMD was the lowest. The promulgation of the guidelines for HFMD can reduce the number of inpatients through standardized diagnosis and treatment in outpatient. Meanwhile, the guidelines introduce measures for personal protection, family protection, kindergarten, or school prevention and management, thus reducing the incidence of hand foot and mouth disease (Ma et al. 2011). EV71 infections are one of the main etiological agents of HFMD; on December 2015, the China Food and Drug Administration (CFDA) approved the first inactivated EV71 whole virus vaccine for preventing severe HFMD. EV71 vaccination could decrease HFMD incidence significantly among children aged 2 to 5 years (Wang et al. 2021). In children, the EV71 vaccine elicited EV71-specific immune response, and less EV71-associated HFMD cases have been observed. One real-world study has provided evidence of EV71 vaccine effectiveness for preventing EV71 and "other" viruses associated with HFMD (Du et al. 2021).
In our research, most of inpatient children with HFMD were hospitalized for less than 7 days, followed by 8 to 14 days, and only a few of them were hospitalized for more than 14 days. Children who aged 1 to 6, with illness onset in summer and 2012, were most likely to be hospitalized for more than 14 days. In 2012, the number of inpatient children with HFMD exceeded to 14 days was the largest, followed by 2013 and 2014, then it decreased significantly since 2015, which was related to the gradual improvement of the diagnosis and treatment of HFMD. The number of children hospitalized for more than 14 days in summer may be related to the increase of nosocomial infections that were caused by high temperature and humidity in summer (Eber et al. 2011). The number of children hospitalized for more than 14 days is also higher in aged 1 to 6 than that of other age groups. In terms of hospitalization management, more attentions should be paid to these individuals.
Previous studies showed that PM 2.5 has a higher influence on hospital admission than other air pollutants (Buczyńska et al. 2014). PM 2.5 was positively associated with LOS among children (Baek et al. 2020a). There was a short-term increase in hospital admission rates associated with PM 2.5 for all of the health outcomes except injuries (Dominici et al. 2006). Our study showed that short-term PM 2.5 exposure was positively associated with the LOS of HFMD. Our findings were broadly consistent with those studies (Bell et al. 2009;Zhong et al. 2018). It has been demonstrated that PM 2.5 exposure could adversely affect vascular endothelial function, the activity of the sympathetic nervous system, and systemic inflammation, leading to vasoconstriction, increased plasma viscosity, and a risk of blood clotting and thrombosis (Liu et al. 2017). These adverse effects can exacerbate children's symptoms, therefore increasing the LOS of HFMD.
In addition, in our research, SO 4 2− was the most significantly associated component of PM 2.5 with the LOS of HFMD followed by NH 4 + , SOIL, NIT, and BC. SO 4 2− and NH 4 + were the secondary pollutants and mainly concentrated in power plant dust, motor vehicle exhaust, and construction dust; generally, higher values were found within summer and spring. SO 4 2− is the mostly regionally transported in summer, and various studies have linked sulfate exposure with adverse respiratory and cardiovascular effects, as well as mortality (Nunez et al. 2021).
The BC component of PM 2.5 consists of soot, charcoal, char, and other light-absorbing refractory matter. Although the health impacts of BC have been extensively studied (Clifford et al. 2016;Nunez et al. 2021), its association with HFMD has not been characterized as well. In Nanjing, we found an expected association between BC exposure and the LOS of HFMD. Our research has enriched the epidemiological information on the health impacts of BC on HFMD. NIT is mainly a secondary particle found in the atmosphere; unlike the adverse health effects of BC exposure, the exposure of NIT has been less explored (Cao et al. 2012).
The SS originate Cl − , Na + , and Mg 2+ and is variable during winter months because of unsettled weather conditions in winter season (Kocak et al. 2015); in our study, the cases of hospitalized HFMD tended to arise in warmer season. The OM is a highly complex mixture of hundreds of compounds such as organic carbon, polycyclic aromatic hydrocarbons, alkanes, and fatty acids. However, the health effects of OM remain largely uncharacteristic, and we did not find any associations between OM exposure and the LOS of HFMD. These findings need to be further confirmed.
Our study had advantages because of focusing on the epidemiological features of the LOS of HFMD among Chinese populations and the relationship between the PM 2.5 and its component exposure and the LOS of HFMD. This is a topic that has not received enough attentions and will provide a basis for hospital management. PM 2.5 exposure is associated with the increased LOS of HFMD, and its components (BC, SO 4 2− , NH 4 + , NIT, and SOIL) of PM 2.5 might play a key role in the prolonged LOS of HFMD. Our findings call for greater awareness of environmental protection and the implementation of effective measures to improve the quality of air, which may reduce the risks of adverse effects on children. Policy changes to reduce outdoor air pollutant exposure may lead to improved HFMD outcomes and substantial savings in healthcare spending.

Conclusion
In our research, we aimed to identify the association between short-term exposure of PM 2.5 and its components and the LOS of HFMD. The incidence of hospitalized HFMD peaked in children aged 1 and then decreased with age. Most of hospitalized HFMD cases had occurred in children younger than 6 years old. Parents should not let children attend kindergartens/schools until febrile and all vesicles have dried up and keep children at home while symptomatic and do not let them attend activities which mix with other children, e.g., birthday party, interest classes, and swimming pool. Short-term PM 2.5 and its component (BC, SO 4 2− , NH 4 + , NIT, and SOIL) exposure might increase the LOS of HFMD. Improving the quality of air could help to reduce the burden of HFMD. In the future, it is necessary to carry out prospective cohort study about short-term PM2.5 exposure and the LOS of HFMD.

Limitations
Our study had several limitations. Firstly, some covariates were unmeasured, such as prematurity, health conditions at birth, information of disease severity, laboratory confirmation status, and enterovirus serotype. Secondly, although our observational study was completed in a representative city in China and lasted for 7 years, the sample size included is still relatively small, which requires larger sample size to support research results. Thirdly, in our study, we focused on PM 2.5 , and other air pollutants such as nitrogen dioxide (NO 2 ), ozone (O 3 ), and carbon monoxide (CO) might also contribute to the LOS of HFMD. Finally, due to the temporal resolution of the model which is a monthly average, we consider each day of the month as the average exposure level for that month. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.