Etiological landscape and co-infection dynamics in Acute Respiratory Infections (ARIs): insights from a multicenter outpatient study in Yunnan, China

Abstract

Background

ARIs pose significant health risks, especially among children, elderly, and immunocompromised populations. This study investigated the etiological spectrum and epidemiological features of ARIs in Yunnan Province from December 2023 to February 2025.

Methods

Throat swab samples and demographic information of ARI patients were collected from fever clinics of 23 sentinel hospitals in Yunnan Province every two weeks during the study period. Twenty-eight respiratory pathogens (13 viruses, 13 bacteria and 2 fungi) were detected by real-time PCR, and their overall detection rate, age distribution, seasonal variation and pathogen interactions were analyzed.

Results

From the 2,386 patient samples tested, the overall detection rate for viruses and Mycoplasma pneumoniae (MP) was 33.86%, while bacteria and fungi were detected in 63.16% of samples. The most frequently detected pathogens varied by age group. Influenza A virus (IAV) and Human rhinovirus (HRV) were among the three most frequently detected viral pathogens across four age groups. Respiratory syncytial virus (RSV) showed the highest positivity rate (12.61%) in the 5–14 years age group and the second-highest rate (14.29%) in children under 5 years. Integrated modeling data revealed distinct temporal patterns among five respiratory pathogens: IAV exhibited bimodal winter peaks, Severe acute respiratory syndrome coronavirus (SARS-CoV-2) a single spring peak, while RSV and HRV showed dispersed low-intensity activity. The cross-correlation analysis indicates a significant short-term temporal correlation between IAV and SARS-CoV-2, limited to lags within ± 0.42 months. Among cases positive for IAV, SARS-CoV-2, or MP, co-infection rates with other tested pathogens (including a panel of respiratory viruses and MP) were 13.26%, 10.88%, and 18.88%, respectively. In the pairwise logistic regression analysis of pathogen co-detection, both IAV and SARS-CoV-2 were significant positively correlated with Aspergillus (False Discovery Rate (FDR)-adjusted p < 0.05).

Conclusions

Pathogens of ARI patients across Yunnan since late 2023 exhibited distinct seasonal fluctuations, diverse population susceptibility, and widespread co-infections. Statistically significant positive correlations with Aspergillus were observed for both SARS-CoV-2 and IAV. Therefore, we recommend the establishment of a seasonal, multi-pathogen surveillance and early warning system, coupled with enhanced clinical protocols for the management of coinfections, particularly among vulnerable populations.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12879-026-12937-z.

Introduction

ARIs are one of the major public health problems leading to increased morbidity, hospitalization, and mortality worldwide. Seasonal influenza viruses are estimated to cause up to 600,000 related deaths each year, posing a serious threat to population health and medical resources [1–3]. The transmission and epidemiological characteristics of ARI pathogens are influenced by a variety of factors, including seasonal variations, geographic environment, demographics, and public health intervention policies [4]. Therefore, the pathogenic spectrum and epidemic patterns of ARI in different countries and regions show significant regional differences and dynamic trends.

In recent years, the Coronavirus Disease 2019 (COVID-19) pandemic has profoundly reshaped the epidemiological patterns of respiratory infections and intensified the interactions among co-circulating pathogens. Beyond triggering an unprecedented public health crisis, it has significantly altered the dynamics of other common respiratory pathogens [5, 6]. A substantial wave of SARS-CoV-2 infections in China during early 2023 likely altered population immunity and the subsequent transmission dynamics of all respiratory pathogens in the following seasons. Several studies have shown that non-pharmaceutical interventions (NPIs) during the COVID-19 pandemic (e.g., facemask wearing, social distancing, etc.) inhibited the spread of pathogens such as influenza virus and RSV in the short term, but then the frequency and seasonality of these pathogens changed to varying degrees. At the same time, the pattern of co-infection between pathogens tends to be more complex, which further increases the difficulty of disease monitoring and control.

Yunnan Province is located in the southwest of China with complex topography and variable climate types.These factors likely influence respiratory pathogen infection intensity and seasonal patterns. However, systematic research on ARI pathogen spectra in Yunnan remains limited, particularly since the COVID-19 pandemic [7].

To this end, based on the surveillance data of several sentinel hospitals in Yunnan Province, this study systematically collected and analyzed the pathogen detection results of ARI patients from December 2023 to February 2025, to characterize respiratory pathogen dynamics, identify epidemic and co-infection patterns, and inform control strategies for similar settings.

Methods

Patients and materials

The subjects of this study were patients with ARI who attended fever clinics in several sentinel hospitals in Yunnan Province from December 2023 to February 2025. A case was defined as an individual presenting to the fever clinic with a disease onset within 14 days prior to the visit, characterized by fever (body temperature ≥ 37.3℃) and at least one respiratory symptom, such as cough, sputum production, sore throat, rhinorrhea, or dyspnea. Throat swab samples were collected on a biweekly schedule, with sampling always conducted on a Monday. On each sampling day, samples were obtained from patients presenting with ARIs at the fever clinics of the sentinel hospitals and were then sent to the central laboratory for analysis, and demographic and clinical information were recorded simultaneously (Table S1). Inclusion criteria include the presence of typical symptoms of ARI and the ability to cooperate with sampling and information collection. The exclusion criteria were as follows: missing important sample information (n = 16), including sampling time, hospital source, age, or sex; the throat swab sample collection is unqualified (n = 34), such as sampling tube is not sealed, leaking, contaminated, label off; duplicate samples from the same patient (n = 2). To prevent repeat enrollment, we used the patient’s hospital unique identification number to link with the sample number to prevent duplicate samples from being included in the testing. A flowchart detailing patient enrollment and specimen screening is depicted in Figure S1.This study has been approved by the Ethics Committee of the First People’s Hospital of Yunnan Province (ethics number: KHLL2024-KY155).

Specimen collection and detection

The posterior pharyngeal wall and bilateral tonsil areas of the subject were scraped 3–5 times with a sterile sampling swab and immediately placed in a sterile sampling tube containing the preservation solution. The preservative solution contains guanidinium salts (such as guanidinium thiocyanate or guanidinium hydrochloride). In addition, it also contains phosphate buffer solution (to maintain stable pH), surfactant (to help lyse the virus), phenol red (as a pH indicator, making the liquid pink or red for easy observation), and stabilizers such as bovine serum albumin. Samples are temporarily stored in a 2–8 °C refrigerator and sent to the central laboratory within 24 h. A throat swab sample was considered qualified only if the sample tube was intact without leakage and the label was complete and legible. If the sample cannot be tested within 48 h, it is transferred to a − 80 °C ultra-low temperature freezer for long-term storage. All testing was completed within 72 h of the samples’ arrival at the laboratory. All patients filled out a uniformly designed epidemiological questionnaire and completed it after on-site interviews with the attending doctors.

The laboratory used multiplex real-time PCR technology to detect 28 common respiratory pathogens, including 13 viruses, 13 bacteria and 2 fungi (Table S2). All pathogens detection assays in this study were performed using commercially available detection kits. The detection kits included both multiplex pathogen detection kits and single-pathogen detection kits. The Sixteen Respiratory Pathogens Nucleic Acid Multiplex Detection Kit (Fluorescent PCR) was purchased from Xiamen Amplly Bioengineering Co., Ltd. (China), and the single-pathogen detection kits were purchased from Sansure Biotech Inc. (China) (Table S3). All tests were conducted strictly in accordance with the manufacturers’ instructions, including sample preparation, PCR reaction mixture setup, cycling parameter settings, and result interpretation. According to the manufacturer’s kit instructions, we defined a sample detection Ct value of ≤ 38 as positive, and a Ct value of > 38 as negative, RNase P is used as an internal reference gene for quality control. The positive (detection) rate was calculated as the number of positive samples (Ct ≤ 38) divided by the total number of tested samples for the overall rate, or by the number of samples within the respective age subgroup for stratified rates.

Statistical methods

The geospatial analyses and map generation were performed using ArcGIS. R (version 4.4.2) and Origin 2024 (version 4.4.1) were used for statistical analysis and data visualization. Count data are expressed as frequency (n) and percentage (%). Non-normally distributed continuous variables (e.g., nucleic acid Ct values) were reported as median (interquartile range, IQR).

Generalized Additive Models (GAMs) were employed to model the monthly periodicity of each virus’s positivity rate, using a cyclic cubic regression spline s (Month, bs="cc”) to ensure annual continuity. Smoothing complexity was optimized via generalized cross-validation to avoid overfitting. Models were fitted using restricted maximum likelihood (REML), yielding predicted monthly rates with 95% confidence intervals (95%CIs) from which the annual peak month was identified.

Cross-correlation analysis was conducted to assess the temporal association between the monthly positivity rates of two viruses. We calculated the cross-correlation function (CCF) for lags up to ±6months. Correlation coefficients exceeding the 95% CIs (± 2/√n, where n is the number of monthly observations in the time series) were considered statistically significant (p < 0.05). The lag at which the maximum significant correlation occurred and its amplitude were used to interpret the relationship.

Binary logistic regression was used to analyze the pairwise associations between respiratory pathogens. The detection status of pathogen 2 was modeled as a function of the detection status of pathogen 1, with adjustment for age group, sex, sampling month, chronic structural lung disease (including chronic obstructive pulmonary disease, bronchiectasis, cavitary pulmonary tuberculosis, and pulmonary fibrosis), and smoking history as fixed effects. A random intercept for reporting hospital was included to account for inter-site variability. The model formula is as follows:

Where Yij indicates whether individual i in hospital j was positive for pathogen 2, and Xij indicates whether the same individual was positive for pathogen 1. Models were fitted using the glmer () function from the R package lme4. Odds ratios (ORs) with 95%CIs were derived from exponentiated regression coefficients. Statistical significance was defined as an False Discovery Rate (FDR)-adjusted p-value < 0.05 using the Benjamini–Hochberg method. This regression model were employed as an exploratory screening tool to identify candidate associations across a wide range of pathogens; therefore, the results should be interpreted as hypothesis-generating rather than definitive causal inferences.

Results

General information

As of February 2025, a total of 2,438 throat swab samples were collected, and 52 samples with missing demographic or clinical information and unqualified sample quality were excluded, and a total of 2,386 valid samples were finally included in the analysis. All samples were from 23 sentinel hospitals in 16 prefectures and cities in Yunnan Province. Figure 1 shows the spatial distribution of surveyed hospitals and population density within the study area.The top three cities with the largest number of provided samples were Kunming (n = 704, 29.5%), Yuxi (n = 262, 11.0%) and Baoshan (n = 246, 10.3%). Diqing (n = 29, 1.2%) provided the smallest sample size. There were 1137 males (47.7%) and 1249 females (52.3%). The number of patients in each age group was shown as follows: 126 (5.3%) cases in the 0 ~ < 5 years group, 119 (5.0%) cases in the 5 ~ 14 years group, 1612 (67.6%) cases in the 15 ~ 59 years group, and 529 (22.2%) cases in the ≥ 60 years group. In terms of ethnic composition, there were 1,857 cases (77.8%) of the Han population, and the rest were ethnic minorities such as Yi, Bai, Dai, Hani, and Hui.

Fig. 1.

Population Density and Surveillance Hospital Sentinel Distribution in Yunnan Province, China. Population Density Gradient: The map illustrates the population density across various cities in Yunnan Province, categorized by color gradients. The legend on the right indicates the population density per square kilometer, ranging from less than 50 (lightest shade) to over 400 (darkest shade).Surveillance Hospital Sentinels: Blue dots represent the locations of surveillance hospital sentinels. Generated by ArcGIS software

Overall results of the pathogens detection

The overall detection rate of 13 respiratory viruses and MP was 33.86% (808/2386). IAV had the highest detection rate (12.32%, 294/2386), followed by MP (6.66%, 159/2386), HRV (6.58%, 157/2386), SARS-CoV-2 (3.86%, 92/2386) and RSV (2.22%,53/2386). The detection rate of other respiratory viruses ranged from 0.13% to 1.51%, including Influenza B virus (IBV), Human coronaviruses (HCoV-HKU1, 229E, NL63, OC43), Human metapneumovirus (HMPV), Human bocavirus (HBoV), and Human parainfluenza virus (HPIV). Among these, HMPV was detected in 20 cases, primarily distributed in children under 14 years (8/20, 40.00%), HPIV was primarily detected among individuals under 5 years (9/28, 32.14%). HCoV and HBoV showed sporadic distribution across all age groups. Despite their low overall prevalence, these viruses contributed to the complex viral spectrum observed in pediatric patients under 14 years of age (Fig. 2a). Of the samples positive for viruses and MP, 3.19% (76/2386) detected two pathogens and 0.13% (3/2386) detected three pathogens (Table 1).

Fig. 2.

Positive rates of ARIs pathogens in different age groups. Bar chart of each color displaying the positivity rates of viral pathogens (a), bacterial pathogens (b), and fungal pathogens (c) across different age groups (< 5, 5–14, 15–59, and ≥ 60 years). The statistical analyses and visualizations were performed using R software (version 4.4.2)

Table 1.

Positive detection count, percentage of the total sample size, and median nucleic acid amplification Ct value (IQR) for respiratory viruses and mycoplasma

Pathogen Type	N	Positive Rate (%)	Ct value, median (IQR)
MP	159	6.66	30.50 (27.90, 33.00)
IAV	294	12.32	27.17 (24.92, 28.88)
IBV	25	1.05	28.21 (25.38, 30.22)
RSV	53	2.22	25.80 (24.13, 28.79)
HAdV	36	1.51	33.74 (28.78, 35.16)
HBoV	9	0.38	27.30 (26.46, 34.31)
HCoV-HKU1	3	0.13	28.11 (25.65, 28.46)
HCoV-229E	7	0.29	28.55 (18.63, 33.99)
HCoV-NL63	4	0.17	24.98 (23.75, 28.19)
HCoV-OC43	6	0.25	24.96 (22.14, 27.11)
HRV	157	6.58	32.52 (29.07, 34.70)
HMPV	20	0.84	29.83 (28.04, 32.02)
HPIV	28	1.17	29.49 (24.41, 32.20)
SARS-CoV-2	92	3.86	30.78 (28.44, 32.18)
Dual Detection	76	3.19	-
Triple Detection	3	0.13	-
Overall Positive Detection	808	33.86	-

Overall positive rate of bacteria and fungi was 63.16% (1507/2386). The most frequently detected bacteria across all samples was S. pneumoniae, with a detection rate of 35.71% (852/2386), followed by H. influenzae (13.83%, 330/2386), Klebsiella pneumoniae (K. pneumoniae, 8.13%, 194/2386), Staphylococcus aureus (S. aureus, 8.13%, 194/2386) and A. baumannii (7.59%, 181/2386). The positive rate of the other bacteria was between 0.63%~3.10%.In addition, some specific bacterial pathogens were detected, including Legionella pneumophila (L. pneumophila, 1.76%, 42/2386), Bordetella pertussis (B. pertussis, 0.84%, 20/2386), and Chlamydia pneumoniae (C. pneumoniae, 0.63%, 15/2386).

Among the fungi, the detection rate of Candida albicans (C. albicans) was 11.19% (267/2386) and that of Aspergillus was 5.99%(143/2386). The co-detection rate of bacteria and fungi (≥ 2 species) was 26.70% (637/2386) (Table 2).

Table 2.

Positive detection count, percentage of the total sample size, and median nucleic acid amplification Ct value (IQR) for respiratory bacteria and fungi

Pathogen Type	N	Positive Rate (%)	Ct value, median (IQR)
S. pneumoniae	852	35.71	31.46 (28.09, 33.94)
H. influenzae	330	13.83	30.84 (27.78, 32.65)
S. aureus	194	8.13	30.72 (27.93, 33.87)
A. baumannii	181	7.59	32.75 (29.72, 34.66)
K. pneumoniae	194	8.13	29.69 (23.65, 33.86)
P. aeruginosa	55	2.31	34.77 (30.97, 36.46)
M. catarrhalis	74	3.10	31.80 (28.95, 33.91)
L.pneumophila	42	1.76	31.75 (29.39, 32.95)
B. pertussis	20	0.84	31.11 (28.54, 32.81)
GAS	29	1.22	32.11 (29.65, 33.50)
C. pneumoniae	15	0.63	29.33 (27.98, 33.35)
GBS	19	0.80	29.34 (28.04, 31.01)
C. albicans	267	11.19	27.81 (24.57, 30.49)
Aspergillus	143	5.99	36.16 (33.98, 36.71)
Multiple detection	637	26.70	-
Overall Positive Detection	1507	63.16	-

Respiratory pathogens in different age groups

Stratified analyses showed significant differences in the positivity rates of respiratory viruses and bacteria across age groups (Fig. 2). Among viral pathogens, IAV and HRV were among the three most frequently detected pathogens across all age groups. RSV demonstrated high prevalence, with the second-highest positivity rate (14.29%) in children under 5 years old and the highest rate (12.61%) in the 5–14 years age group. In the 15 ~ 59-year-old group and the ≥ 60-year-old group, IAV, HRV, and SARS-CoV-2 were overwhelmingly dominant, while most viruses such as RSV, HAdV, HPIV, HMPV, and HCoV had low detection rates. Overall, children under the age of 14 appear to show a more complex viral profile (Fig. 2a) .

In terms of bacterial pathogens, S. pneumoniae had the highest detection rate among all age groups, especially in children aged <5years, with a positive rate of 70.63%. With the increase of age, the detection rate gradually decreased, and was 25.71% in the ≥ 60-year-old group. In addition, the dominant bacteria detected in each age group varied. For example, H. influenzae was primarily distributed in children under 14 years of age, and the highest detection rates of (A) baumannii occur in ≥ 60-year-olds. Meanwhile, the 5 ~ 14 years age group showed the highest occurrence of MP, with a rate of 12.61%, whereas the other groups ranged from 4.54% to 9.52%(Fig. 2b). Positivity of C. pneumoniae ranged from 0.19% to 2.52%, but was not detected in children aged < 5 years. L. pneumophila was mainly detected in people aged 15 ~ 59 years (2.36%) and ≥ 60 years (0.76%), while no cases were identified in those under 15 years old. (B) pertussis was detected in all age groups(0.38%~3.36%).

In terms of fungi, Aspergillus. was distributed in all age groups, and the positive rate was between 4.76%~7.56%. C. albicans had the highest detection rate of 26.09% in ≥ 60-year-old population, and the other groups ranged from 5.88%~7.14%(Fig. 2c).

Epidemic trends of four viruses and MP

By integrating surveillance with predictive modeling, we demonstrate distinct temporal heterogeneity and alternation in the circulation of the five major respiratory pathogens (Fig. 3). IAV exhibited a distinct epidemic peaks in December 2023, followed by a rapid decline and a subsequent gradual rebound after October 2024. In contrast, SARS-CoV-2 peaked in April 2024, followed by a gradual monthly decline. The other pathogens (HRV, MP, RSV) maintained largely stable trends throughout the year. Deviations from this baseline included transient elevations of HRV in December 2023 and September 2024, and a sustained increase in MP cases from November 2024 onward (Fig. 3a). The Generalized Additive Model (GAM)-smoothed trend (Fig. 3b) indicates distinct periodic patterns for the five pathogens. IAV demonstrated a bimodal distribution with peaks in January and December. The peak for MP occurred near December. SARS-CoV-2 formed a single peak in spring (March-April). In contrast, both RSV and HRV demonstrated dispersed activity during summer and autumn, with relatively lower peak intensities.

Fig. 3.

Monthly actual surveillance and predicted trends of five respiratory pathogens. (a) Monthly distribution of the test volume and the observed positive rate (with 95% confidence intervals) from December 2023 to February 2025. (b) Predicted positive rate (with 95% confidence intervals) based on the Generalized Additive Model (GAM), representing the fitted epidemic trends. Generated by R software

The cross-correlation analysis between IAV and SARS-CoV-2 revealed a sequential, rather than synchronous relationship, which was confined to lags within ± 0.42 months (Fig. 4; Table S4). A peak in SARS-CoV-2 was significantly associated with a subsequent peak in IAV, the strongest correlation overall was observed at a lag of -0.25 months (r = 0.70, 95%CI[0.18, 1.22], p < 0.05) and − 0.33 months (r = 0.69, 95%CI[0.18, 1.21], p < 0.05). Conversely, a peak in IAV was followed by a significant decline in SARS-CoV-2 activity, a significant negative correlation was found at a lag of + 0.08 months (r=-0.55, 95% CI[-1.07, -0.03], p < 0.05) and + 0.17 months (r=-0.54, 95% CI[-1.06, -0.03], p < 0.05). The correlation at zero lag was weak(r=-0.45, 95%CI[-0.97, 0.06], p = 0.07), indicating minimal synchronous co-circulation (Table S4).

Fig. 4.

Cross-correlation Analysis Between IAV and SARS-CoV-2. The plot depicts the cross-correlation coefficient across different time lags (months). A positive lag denotes that SARS-CoV-2 cases followed IAV cases, while a negative lag indicates the opposite. A lag of zero represents synchronous variation. The red dashed lines indicate the 95% confidence interval bounds. Correlation peaks exceeding these bounds are statistically significant (p < 0.05). Generated by R software

Co-infection and interaction

Among samples positive for IAV, SARS-CoV-2, and MP, the co-infection rates with other viruses and MP were 13.26%, 10.88%, and 18.88%, respectively (Fig. 5a-c). The most common combinations of viruses-MP co-infection are IAV and HRV, SARS-CoV-2 and HRV, MP and HRV. Co-infections involving other respiratory viruses such as RSV, HAdV, HCoV, and HPIV were less frequently detected. The co-detection of viruses with bacteria/fungi is more prevalent. Specifically, 68.37% of IAV-positive samples and 70.65% of SARS-CoV-2-positive samples showed bacterial/fungal co-detection (Fig. 5d-f) .

Fig. 5.

Analysis of co-infections/detection and constituent ratios among patients positive for IAV, SARS-CoV-2, or MP. Pie charts represent the distribution of samples co-detected with (a, b) IAV, (c, d) SARS-CoV-2, and (e, f) MP alongside other respiratory pathogens. The graph was created and data were plotted using Origin2024 (version 4.4.1)

Figure 6a displays the co-detection rates for all pairwise pathogen combinations in the overall cohort. The top 15 combinations are listed in Table S5, with co-detection rates for IAV and S. pneumoniae and for IAV and Aspergillus were 4.86% and 1.72%, respectively. Pairwise logistic regression analysis of pathogen co-detection, controlled for potential confounders (age group, sex, sampling month, chronic structural lung diseases, and smoking history), revealed that most virus-virus interactions exhibited OR below 1.00. However, these negative associations did not retain statistical significance after FDR correction (Table S4). Conversely, significant positive associations were identified between SARS-CoV-2 and Aspergillus, as well as between IAV and Aspergillus (FDR-adjusted p < 0.05) (Fig. 6, Table S6), which may reflect increased risk of fungal infection in the context of virus-induced host immune modulation. These results indicate complex potential interactions among pathogens, which may influence host susceptibility, disease severity, and clinical outcomes.

Fig. 6.

Co-detection and correlation heatmap for pairwise respiratory pathogens of the whole cohort. (a) This heatmap illustrates the co-detection rates between various respiratory pathogens. Each cell represents the co- detection rate (%) between two pathogens, with the color intensity indicating the rate. (b) This heatmap shows the correlation between various respiratory pathogens. Positive associations were denoted in orange and the negative associations were denoted in green color. Asterisks (*) denoted statistical significance (FDR-adjusted p < 0.05). Generated by R software

Discussion

ARIs continue to be one of the leading causes of hospitalization and death worldwide, posing a serious public health burden to people of all ages. As the COVID-19 pandemic has erupted and eased globally, the epidemic patterns of respiratory pathogens have changed dramatically. Especially in Chinese mainland, since the gradual relaxation of NPIs measures at the end of 2022, the number of respiratory tract infection cases has increased significantly, and the activity level of viral pathogens has risen again, showing complex seasonal characteristics and co-infection combinations [8].

Based on the data of ARI cases from 23 medical institutions in 16 cities of Yunnan Province, our findings revealed key shifts in the epidemiological landscape of ARIs since the onset of the COVID-19 pandemic. These changes are mainly reflected in: (1) the altered composition of dominant pathogen combinations, (2) seasonal dynamics of respiratory viruses after the emergence of SARS-CoV-2, and (3) new evidence supporting virus-Aspergillus associations. The winter-spring epidemic of most respiratory viruses in our study mirrors the classic timing seen in temperate regions [9–11]. Aligning with large-scale national studies in China, our research found that IAV and SARS-CoV-2 exhibited staggered epidemic patterns, characterized by intermittent periods of overlap [8]. Consistently, we observed that children presented with a more diverse viral spectrum, including a high prevalence of RSV [7, 8, 12]. Across all age groups, S. pneumoniae, IAV, and HRV were identified as the most prevalent pathogens [8, 12, 13]. Besides, elevated detections of A. baumannii and K. pneumoniae in the elderly likely reflect factors such as comorbid structural lung disease and healthcare-associated exposure [12].

Co-infection with respiratory pathogens is often commonly observed clinically, and predicts more severe disease and a worse prognosis [14–17]. Our study showed that virus-virus/MP co-infection mainly occurred between IAV, MP, SARS-CoV-2, and HRV, while virus-bacteria/fungi co-detections was more common overall. Previous reports suggest cooperative interactions between IAV and S. pneumoniae [18, 19], virus-virus interactions are predominantly negatively correlated, supporting the hypothesis of competitive interference [20, 21]. More importantly, clinical evidence indicates that critically ill patients with influenza or COVID-19 are at significantly elevated risk of developing invasive pulmonary aspergillosis, which is often associated with prolonged hospitalization, higher need for mechanical ventilation, and worse clinical outcomes [22–25]. Our study identified positive correlations between both IAV and SARS-CoV-2 and the detection of Aspergillus, suggesting that viral infections may increase the risk of Aspergillus colonization or infection in the upper respiratory tract. These findings carry important implications: during influenza or COVID-19 outbreaks, heightened vigilance for secondary fungal infections is warranted, and enhanced pathogen surveillance and early intervention should be considered for high-risk populations, such as the elderly and immunocompromised individuals.

Our fever-clinic cohort primarily captured mild-to-moderate ARIs, providing a pathogen spectrum that better reflects the community disease burden than hospitalized-patient studies. This etiological profile is crucial for informing primary care and public health interventions in similar settings globally. Therefore, we advocate for an optimized control strategy: in the multi-ethnic border provinces represented by Yunnan, a cross-seasonal and multi-pathogen sentinel surveillance system should be established to enhance regional disease forecasting and resource allocation.

Limitations

Although this study covers a large sample size and is widely distributed in the region, there are still some limitations. First, the reliance on a narrow specimen spectrum (exclusively throat swabs) and the application of a uniform cutoff (e.g., Ct < 38), while operationally straightforward, inevitably overestimates the clinical relevance of detected pathogens, particularly for frequent colonizers like S. pneumoniae. Second, although the PCR method has high sensitivity, the multiplex primer system may lead to non-specific amplification and false positive results, which affects the specificity. RNA virus testing is also affected by sample storage and transportation, which may lead to an underestimation of the positivity rate. Third, the fever clinic-based design, while ideal for capturing community-level ARI burden, inherently focuses on symptomatic individuals seeking care. This may lead to an overestimation of the prevalence of pathogens that typically cause moderate-to-severe symptoms (e.g., IAV, SARS-CoV-2) and an underestimation of pathogens associated with a higher proportion of asymptomatic or mild infections (e.g., HCoV, HPIV). Furthermore, other limitations of this study include the lack of detailed clinical data to assess pathogen-specific severity and, despite our multi-city design across Yunnan’s complex topography, an absence of synchronized, high-resolution environmental data to quantify climatic effects.

Conclusions

This study presents a systematic analysis of the dynamic characteristics of the ARIs pathogen profile in Yunnan Province after the containment of the COVID-19 pandemic. We identified distinct seasonal fluctuations, age-related variations, and pathogen co-infection patterns within the regional respiratory pathogen spectrum. Our findings underscore the predominance of IAV, SARS-CoV-2, and MP in the current epidemic activity. Statistical analyses significant positive correlations with Aspergillus were observed for both SARS-CoV-2 and IAV. We recommend the establishment of a seasonal, multi-pathogen surveillance and early warning system, coupled with enhanced clinical protocols for the management of coinfections, particularly among vulnerable populations.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

JC, JH: study design and manuscript; JC, JA, DL: review and editing; JC, DL: project administration; JL, NW, SMY: data analysis and visualization; YF, GL: sample management and experiments; JC, HQ, ZL: collecion of samples and information; JG, JH: supervision and funding acquisition.

Funding

This study was supported by a grant from the National Key Research and Development Program of China (2024YFC2309202), National Key Research and Development Program of China (202402AA310055), Yunnan Provincial Department of Science and Technology Major Science and Technology Special Program (2023YFC3043513), Science and Technology plan project of the First People’s Hospital of Yunnan Province (Project number: KHYJ-2025-6-05).

Data availability

All the data involved in this article were originally generated from this study.Data can be reasonably obtained after consulting the corresponding author byemail.

Declarations

Ethics approval and consent to participate

All study protocols and procedures were approved by the Medical Ethics Board of the First People’s Hospital of Yunnan Province, China, and were carried out in accordance with all relevant provincial, national, and international guidelines, including the Declaration of Helsinki. Written informed consent was obtained from all participants prior to their inclusion in the study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.