Concordance between upper and lower respiratory tract viral testing in mechanically ventilated ICU patients: A retrospective cohort study.

BACKGROUND

The diagnostic value of upper respiratory tract (URT) viral testing as a surrogate for lower respiratory tract (LRT) viral detection in mechanically ventilated patients remains uncertain. We evaluated URT/LRT concordance in intubated intensive care unit (ICU) patients and identified factors associated with non-concordance.

METHODS

We retrospectively studied adult and pediatric ICU patients receiving invasive mechanical ventilation who underwent paired URT and LRT respiratory viral polymerase chain reaction (PCR) testing between January 2014 and January 2025. Concordance was defined as detection of at least one identical virus in both specimens or negative results in both. Multivariable logistic regression identified factors associated with concordance. A simplified bedside prediction score was developed to predict non-concordance.

RESULTS

Among 112 patients, 70 (62.5%) had concordant results. Specific positive agreement varied by virus: human metapneumovirus (88.9%), respiratory syncytial virus (80.0%), rhinovirus/enterovirus (75.0%), influenza virus (68.4%), adenovirus (34.8%), and human bocavirus (0.0%). A longer symptom-to-URT testing interval (adjusted odds ratio [aOR], 0.74 per day; 95% confidence interval [CI], 0.62-0.87) and chronic lung disease (aOR, 0.22; 95% CI, 0.06-0.80) were associated with lower concordance, whereas positive end-expiratory pressure > 5 cmH₂O was associated with higher concordance (aOR, 2.85; 95% CI, 1.07-7.59). The bedside score showed good discrimination for non-concordance (area under the receiver operating characteristic curve, 0.818).

CONCLUSIONS

URT/LRT concordance was moderate and varied by symptom timing, clinical factors, and viral pathogen. LRT testing may provide additional diagnostic value in patients at high risk of non-concordance. The bedside score may help identify these patients but requires further internal and external validation.