Microbial spectrum, co-detection patterns, and clinical correlations in 10,153 hospitalized children with acute respiratory infections: a large-scale tNGS analysis.

OBJECTIVE

This study characterized the microbial spectrum and clinical manifestations in hospitalized children with acute respiratory infections (ARI) using targeted next-generation sequencing (tNGS) and identified predictors for severe disease requiring ICU admission.

METHODS

10,153 hospitalized children with ARI were enrolled. Microbial detection was performed using tNGS on pediatric respiratory tract samples. Epidemiological trends, clinical features, laboratory and imaging findings, and outcomes were analyzed. Logistic regression identified factors associated with ICU admission.

RESULTS

The microbial detection rate was 97.77%, with a co-detection rate of 82.87%. Distinct clinical profiles emerged: human respiratory syncytial virus (HRSV) and human bocavirus were associated with younger age and more severe disease; Mycoplasma pneumoniae (MP) with frequent respiratory complications yet lower overall severity, particularly in older children; and adenovirus and Haemophilus influenzae with milder illness. A bidirectional severity pattern was identified: monodetections of higher-virulence agents (e.g., HRSV, Acinetobacter baumannii, human bocavirus) correlated with more severe disease than their co-detections, whereas the opposite was true for lower-pathogenicity microbes such as MP, adenovirus, and Haemophilus influenzae. ICU patients had significantly higher detection rates of HRSV (29.01% vs. 17.53%), cytomegalovirus (25.66% vs. 19.75%), and Acinetobacter baumannii (19.25% vs. 12.51%) (all P < 0.001). Independent predictors for ICU admission included respiratory complications (OR = 8.289), fever (OR = 7.338), and other system complications (OR = 5.564).

CONCLUSION

This large-scale study uncovers complex co-detection patterns and microbe-dependent outcomes in ARI. A novel bidirectional severity pattern and distinct ICU microbial profiles with robust predictive factors have been identified, aiding early risk stratification and tailored management.