Introduction

Human parainfluenza virus 3 (HPIV3) is an important pathogen of respiratory tract illness that belongs to the Paramyxovirus family and is one of the main causes of lower respiratory infections1. Similar to respiratory syncytial virus (RSV), severe cases of HPIV3 infection in children predominantly occur in infants under 6 months of age, who are unable to mount a robust antibody response and often develop severe bronchiolitis and pneumonia2. In vulnerable populations, HPIV3 infection may aggravate existing respiratory diseases and increase mortality and morbidity, leading to prolonged hospital stays and a greater burden on healthcare3,4.

Meteorological factors have important impacts on the transmission of HPIV35,6. Climate factors, such as temperature, humidity, and precipitation, have been identified as significant risk factors contributing to the spread of respiratory viral infections7,8. Moreover, air pollutants are also important factors affecting the prevalence of respiratory diseases9,10. Pollutants can damage the respiratory tract mucosa, increasing susceptibility to viral infection and impairing immune function, thus reducing the body’s ability to resist the virus11,12. Children are particularly susceptible to health damage from environmental changes due to their unique physiology and behavior13. Numerous studies have confirmed that changes in environmental factors are closely related to childhood respiratory diseases and allergic asthma.14,15. Understanding the impact of meteorology on the virus’s transmission dynamics can help inform targeted prevention and control measures to reduce spread and safeguard health.

The impact of meteorological factors on disease epidemiology exhibits complex temporal dynamics, often manifesting delayed effects16,17. Existing research has predominantly focused on the acute effects of climate variables on HPIV3 transmission, while lagged temporal relationships have been largely neglected18,19. Furthermore, given substantial regional variation in climatic conditions, analyses must account for local meteorological patterns when evaluating disease dynamics. Suzhou, China, with a subtropical climate, still lacks longitudinal, large-scale epidemiological investigations of HPIV3 infection patterns. The aims were: (1) to identify distinctive clinical features of HPIV3 infections in children; (2) to analyze 13-year HPIV3 epidemiological patterns in Suzhou (2007–2019); and (3) to quantify meteorological/pollutant lag effects using distributed lag nonlinear models (DLNMs).

Methods

Study area and population

This study was approved by the Institutional Human Ethics Committee of Soochow University (2023CS234). Informed consent was waived by the Ethics Review Board of the Children’s Hospital of Soochow University, as no personally identifiable data were collected. This is in accordance with the Declaration of Helsinki guidelines. We performed a time-series analysis in Suzhou, China (population 12.96 million; area 8,657 km2), a region with a subtropical monsoon climate that receives 1100–1200 mm of annual precipitation and 1800–2000 sunshine hours per year. The seasons are defined as follows: spring (March–May), summer (June–August), autumn (September–November), and winter (December–February).

Data collection

A total of 26,202 children with lower respiratory tract infection (LRTI) were admitted to the Department of Respiratory Medicine at the Children’s Hospital of Soochow University from January 1, 2007, to December 31, 2019. Nasopharyngeal aspirates (NPA) and/or sputum samples were collected from these children. Nasal secretions were tested for multiple pathogens, including bacteria via sputum culture and polymerase chain reaction (real-time PCR), for the detection of HPIV3, RSV, and human metapneumovirus (hMPV). Patients with bronchopulmonary dysplasia, bronchial asthma, or other chronic diseases were excluded. All enrolled HPIV3-positive patients had a single infection. The demographic and clinical characteristics of these patients were collected for analysis.

Meteorological data, including monthly mean temperature, mean humidity, rainfall, sunshine, and wind speed, were obtained from the Suzhou Meteorological Bureau. The concentrations of four pollutants—PM2.5, PM10, sulfur dioxide (SO2), and nitrogen dioxide (NO2)—were collected from the Suzhou Ecological Environment Bureau.

Statistical analysis

The monthly positivity rates of respiratory viruses were calculated. Categorical variables were compared using chi-square tests, while continuous variables were assessed for normality (Shapiro–Wilk test) and homogeneity of variance (Levene’s test). Normally distributed data are presented as mean ± standard deviation (x̄ ± s), and non-normally distributed data as median (interquartile range [IQR]). Spearman’s correlation analysis evaluated associations between meteorological parameters and HPIV3 incidence after temporal alignment of monthly cases with environmental data.

DLNMs20 were employed to assess nonlinear exposure-lag-response relationships, with a maximum lag of 7 months selected based on observed data periodicity to capture delayed environmental effects. The primary DLNM was structured as follows:

$$Log\left( {Yt} \right) = \alpha + cb\left( {Meteorological variable,lag} \right) + ns\left( {Timet,df} \right) + ns\left( {{\text{Air pollutants}},df} \right)$$

where t denotes the monthly observation interval, Yt represents the HPIV3 case count, α is the model intercept, cb indicates the cross-basis function modeling nonlinear exposure-lag-response relationships, and ns denotes natural cubic splines for confounder adjustment. Temporal trends were controlled using 7 degrees of freedom per year, while air pollutant covariates (SO2, NO2, PM10, PM2.5) were incorporated with 3 degrees of freedom.

For air pollution effect assessment, the model was reconfigured as:

$$Log\left( {Yt} \right) = \alpha + cb\left( {{\text{Air pollutants}},lag} \right) + ns\left( {Timet,df} \right) + ns\left( {Meteorological variable,df} \right)$$

Variables with Spearman correlation coefficients > 0.7 (P < 0.05) were excluded to mitigate multicollinearity. Model outputs provided relative risks (RR) comparing 1st vs 99th percentile exposures (P1 vs P99) at different lags, with optimal specifications determined by Akaike Information Criterion (AIC) minimization. Robustness was evaluated through sensitivity analyses varying the degrees of freedom for temporal control (ranging from 7 to 9/year) and environmental confounders (ranging from 3 to 5). All computations were performed using R statistical software (version 4.3.3) with the dlnm package (version 2.4.7), implementing rigorous model diagnostics throughout the analytical process.

Results

Demographic characteristics and seasonality of HPIV3 incidence

The demographic characteristics and time distributions during the study period in Suzhou, China are shown in Fig. 1. From 2007 to 2019, a total of 26,202 cases of lower respiratory infection were collected, including 934 cases of HPIV3 infection, with a positive rate of 3.56%. The male-to-female ratio was 2.06:1 (67.3% males and 32.7% females). The age distribution followed WHO pediatric classifications21: infants (29 days to < 1 year; 4.3%), toddlers (1 to < 3 years; 4.0%), preschoolers (3 to < 6 years; 1.6%), and school-age children (≥ 6 years; 1.3%). The HPIV3 detection rate was highest in the infant group. In 2010, the annual positivity rate reached 6.0% (110/1829), notably higher than all other study years. Regarding seasonal patterns, HPIV3 was most frequently detected in summer (46.4%), while the lowest detection rate was noted in winter (5.9%). The month with the highest incidence of HPIV3 was June (8.8%), while the lowest incidence was observed in December (0.7%).

Fig. 1
Fig. 1
Full size image

HPIV3 detection rates of hospitalized children in Suzhou. (A) The age distribution of HPIV3. (B) The number and percentage of HPIV3-positive samples per season. (C) The distribution of case numbers across different months and years from 2007 to 2019. Each point on the plot corresponds to a specific month and year, with the size of the point indicating the number of cases.

Clinical symptoms and laboratory findings in HPIV3-positive children

The most common symptoms among HPIV3-positive children were cough (97.86%), fever (46.15%), rhinorrhea (34.69%), and wheezing (39.94%). After matching the cohorts based on age, sex, and season of onset, a sample of 645 children infected with RSV was selected for comparison. Wheezing and dyspnea were significantly less frequent in HPIV3-positive patients than in those infected with RSV. Furthermore, the incidence of tachypnea was markedly higher in children infected with RSV. Blood analysis revealed that the HPIV3 group had a lower proportion of neutrophils, but a significantly greater median platelet count and a higher proportion of patients with C-reactive protein (CRP) levels exceeding 8 mg/L compared to the RSV group (Table 1).

Table 1 The clinical characteristic comparation between HPIV3 and RSV.
Full size table

Analysis of the correlation between HPIV3 and meteorological factors

The monthly mean temperature, mean humidity, rainfall, sunlight, and wind speed in Suzhou city were 17.47 °C (ranging from 1.1 to 32.3 °C), 71.21% (ranging from 56 to 85%), 3.4 mm (ranging from 0.1 to 13.65 mm), 4.82 h (ranging from 1.42 to 10.71 h), and 2.36 m/s (ranging from 1.20 to 3.50 m/s), respectively (Supplementary Table S1). The time series distribution of environmental variables and HPIV3 cases is shown in Fig. 2a. The trends in changes of meteorological factors and air pollutants from 2007 to 2019 are shown in Supplementary Figure S1. From 2007 to 2019, while temperature, humidity, and wind speed exhibited relatively stable trends, the annual average concentrations of air pollutants, including SO2, NO2, PM10, and PM2.5, showed a significant downward trend.

Fig. 2
Fig. 2
Full size image

The correlation between meteorological factors and number of HPIV3 cases in Suzhou, 2007–2019. (A) Time series diagram between HPIV3-positive cases, meteorological factors and air pollutants. (B) The correlation heatmap between HPIV3-positive cases, meteorological factors and air pollutants.

Spearman correlation coefficients among meteorological factors, air pollutants, and HPIV3 incidence are presented in Fig. 2b. Temperature, relative humidity, rainfall, sunshine duration, and wind speed showed positive correlations with HPIV3 incidence, whereas SO2, NO2, PM10, and PM2.5 exhibited negative correlations. The strongest positive correlation was observed for mean temperature (r = 0.58), while relative humidity showed the weakest positive association (r = 0.08). Among pollutants, NO2 demonstrated the strongest negative correlation (r = − 0.42). All meteorological factors were inversely correlated with air pollutants. Notably, PM2.5 showed strong collinearity with PM10 (r = 0.90, p < 0.001). To avoid multicollinearity in DLNMs, we excluded PM2.5 due to its high correlation with PM10, SO2, and NO2.

Distributed lag nonlinear model

The exposure-lag-effect relationships between meteorological factors and HPIV3 are shown in Fig. 3 via contour plots. Extreme high temperatures (32 °C, P99) had the highest estimated RR for HPIV3 during the current month (month 0, RR = 4.258 (95% CI = 1.387–13.074)), while extreme low temperatures (3 °C, 1st percentile) exhibited maximal risk at a 4-month lag (RR = 3.958 (95% CI = 1.858–8.430)) (Fig. 3A). The contour plot suggests an increasing trend in RR values with higher relative humidity (RH) and longer lag periods, although this trend was not statistically significant (Fig. 3B). Notably, high wind speeds have a more pronounced effect on HPIV3 than low wind speeds, with a notably promotional effect observed when wind speeds exceed 3.0 m/s (Fig. 3C). The contour plots for rainfall and sunlight duration are presented in Supplementary Figure S2.

Fig. 3
Fig. 3
Full size image

Contour plots and plots showcasing the lag effect of meteorological factors on HPIV3 infection. (A) Temperature. (B) Relative humidity. (C) Wind speed.

Our lagged exposure analysis revealed distinct temporal patterns of association between ambient air pollutants and HPIV3 infection risk (Fig. 4). The exposure–response relationships exhibited significant variations across different lag periods and pollutant species. For PM10, the effect was most pronounced at lag0, with a maximum RR of 3.335 (95% CI 1.236–8.999) observed at the 99th percentile concentration of 140 μg/m3. As the lag period increased, the magnitude of the effect progressively decreased (Fig. 4A). In contrast, gaseous pollutants (SO2 and NO2) primarily manifested delayed effects. The risk elevation peaked at 5 months post-exposure for both pollutants. The maximum risks were observed at month 5 (lag5): SO2: RR = 2.047 (95% CI 1.247–3.362) at 56 μg/m3 (P99); NO2: RR = 2.596 (95% CI 1.577–4.273) at 78 μg/m3 (P99) (Fig. 4B, C).

Fig. 4
Fig. 4
Full size image

The lag effect plot for extreme concentration of air pollutants on HPIV3 infection. (A) PM10. (B) SO2. (C) NO2.

Discussion

Compared with adults, children’s immune systems are still developing, resulting in weaker resistance to viral infections. As a vulnerable population, children are highly sensitive to environmental changes, which significantly impact their health13. Given these vulnerabilities, health departments can leverage seasonal virus trend data to develop targeted strategies, such as boosting vaccinations, increasing health awareness, and enhancing medical services. These measures can effectively reduce virus spread and mitigate its societal impact. Our study determined that HPIV3 epidemics exhibit pronounced seasonality, peaking in the summer and predominantly affecting infants. Although HPIV3 shares similar clinical features with RSV, it is distinguished by a lower incidence of wheezing and dyspnea. Meteorological factors, particularly temperature, significantly influence HPIV3 transmission. Among pollutants, PM10 primarily exerts immediate effects, whereas SO2 and NO2 are characterized by lagged effects. These findings underscore the importance of considering both meteorological and environmental factors in developing public health interventions.

In this study, HPIV3 infections predominantly occurred in infants and young children, particularly those aged 0–12 months, which aligns with findings from other reports1,6. Both HPIV3 and RSV belong to the Paramyxovirus family and share similar high-risk populations, especially infants under 6 months of age. The clinical symptoms in these high-risk populations are also similar, including wheezing and dyspnea, making it difficult to distinguish between the two infections in young infants3,22. In this study, the primary symptoms of HPIV3 infection were cough, rhinorrhea and fever. Although wheezing can occur, it is less prevalent compared to RSV infections. The underlying mechanisms for this difference may be due to variations in viral tropism, host immune responses, and viral pathogenicity. RSV evades the human adaptive immune system by skewing the Th1/Th2 cytokine balance towards increased levels of Th2 cytokines and IgE, leading to heightened mucus secretion and airway narrowing23,24. In contrast, the impact of HPIV3 on respiratory epithelial cells is relatively benign and is rapidly repaired in most case25. Viremia and infections beyond the respiratory epithelium can also cause death, although these are rare and typically found only in immunocompromised patients26. Consequently, compared with RSV, HPIV3 has a greater tendency to infect the upper respiratory tract, causing milder symptoms such as cough and nasal congestion. This may further explain its lower incidence of wheezing, dyspnea, and tachypnea symptoms.

The study revealed that HPIV3 exhibits clear seasonality and periodicity, with a peak in summer and the lowest incidence in winter. This finding aligns with our team’s 10-year study from 2001 to 201127. A similar trend has also been reported in Yamagata, Japan28. However, Ji Yoon Han et al. reported that HPIV3 was most common in spring in Daejeon, Republic of Korea18, and Hsieh et al. noted clusters of HPIV3 infections in Taiwan, mainly during spring and early summer29. In general, HPIV3 shows a strong seasonal distribution with high prevalence during the warm season. Therefore, the seasonal epidemiology of HPIV infections should be assessed independently in each country over sufficient years, and its changing trends should be closely watched.

The term ‘pneumonia temperature’ is valid30, whereas different regions exhibit varying infection patterns. Our previous research suggested a positive correlation between temperature and human parainfluenza virus activity27. Utilizing Distributed Lag Nonlinear Models, we found that high temperatures were associated with an immediate increase in infection risk, while the impact of low temperatures became evident after a lag of four months. This pattern suggests that high temperatures within the month may increase the risk of HPIV3 infections due to increased use of indoor air conditioning, which reduces ventilation and enhances the risk of indoor virus transmission. In contrast, low temperatures may compromise immune system function by reducing vitamin D production, thus increasing host susceptibility to seasonal viruses31.

Our findings indicate that HPIV3 infection is less influenced by relative humidity and wind speed than by temperature, which is in line with previous studies conducted in South China17. However, there is considerable heterogeneity in the reported associations between HPIV3 and meteorological factors across different regions. For instance, a study in Edinburgh, Scotland, demonstrated a positive correlation between HPIV3 incidence and relative humidity32, whereas research in Singapore revealed an inverse relationship between HPIV3 risk and relative humidity5. Such discrepancies may stem from variations in the study populations, geographical locations, and local climate conditions. These observations underscore the importance of considering regional differences when examining the impact of meteorological factors on HPIV3 transmission dynamics.

There is extensive evidence linking short- and long-term exposure to air pollution with an increased risk of respiratory infections. In our study, SO2, NO2, and PM10 exhibited significant lag effects and nonlinear relationships regarding their impact on HPIV3 infections. For PM10, the concentration consistently exerts a positive effect on HPIV3 infections, with the impact intensifying as the concentration increases, while the effect diminishes with an extended time lag. This may be attributed to the persistent irritant effect of PM10 on the respiratory tract, as this pollutant has been shown to increase susceptibility and inflammatory responses to viruses in primary human bronchial epithelial cells33. Luong et al34 reported a relationship between PM10 concentrations and respiratory disease-related hospital admissions in the polluted city of Hanoi, Vietnam. In other cities in China, both in warm southern regions and cold industrial cities, studies have shown that PM10 concentrations serve as a risk indicator for the incidence of respiratory diseases among children10,35. Increased exposure to SO2 and NO2 is also associated with a greater risk of respiratory diseases12,36. Past studies have reported that an ordinary person can withstand only 2.62 μg/m3 SO2 in ambient air without any respiratory problems37. Orellano et al38, in a more recent and extensive review and meta-analysis, confirmed that short-term exposure to SO2, ranging from a few hours to a few days, can lead to an increased risk of respiratory morbidity and mortality. Our research indicates that SO2 and NO2 have a delayed impact on HPIV3 infections, which contrasts with the immediate effects typically associated with PM10. Iwasawa et al.39 reported that SO2 exposure-related respiratory symptoms can persist up to two years. A study in Northeast China has highlighted the association between the cumulative lag effect of NO2 and an increased risk of influenza40. This suggests that the effects of these gases are prolonged, underscoring the importance of considering long-term exposure in respiratory health assessments.

However, our study has certain limitations. First, as a time series analysis, this study is inherently ecological and may be susceptible to ecological fallacy. By focusing solely on the overall population without stratifying by sex or geographical region, we may have overlooked important nuances in the data. Second, our study’s reliance on data exclusively from hospitalized children, with outpatient cases excluded, limits the generalizability of our conclusions to the broader population. This limitation underscores the need for future research to include a more diverse sample to ensure the robustness of the findings. Third, our study only compared the clinical differences between HPIV3 and RSV infections and did not further compare the epidemiological differences of the two infections. Future research should include a comparison of the epidemiological characteristics of HPIV3 and RSV to provide a more comprehensive understanding of their respective transmission dynamics. Finally, the single-center nature of our research may have introduced biases due to geographical, environmental, and other factors. To address this limitation and gain a more comprehensive understanding of the disease’s epidemiological patterns, future studies should involve multicenter research, which can provide a more accurate and reliable picture of the meteorological impact on HPIV3 infection.

In conclusion, while our study provides valuable insights into the impact of temperature and air pollutants on HPIV3 infection, further research is needed to overcome these limitations and refine our understanding of the disease.