Definitions and Clinical Adjudication of LRTI.

Because admission diagnoses made by treating clinicians at the time of study enrollment were by necessity based on incomplete clinical, microbiologic, and treatment outcome information, a post hoc adjudication approach was carried out to enhance accuracy of LRTI diagnosis. For this, two attending physicians [one from infectious disease (C.L.) and one from pulmonary medicine (F.M.)] blinded to mNGS results, retrospectively reviewed each patient’s medical record following hospital discharge or death to determine whether they met the CDC/NHSN surveillance definition of pneumonia, with respect to clinical and/or microbiologic criteria (Dataset S1) (19). Chart review consisted of in-depth analysis of complete patient histories, including laboratory and radiographic results, inpatient notes, and postdischarge clinic notes. Using this approach, subjects were assigned to one of four groups, consistent with a recently described approach (16): (i) LRTI defined by both clinical and laboratory criteria; (ii) no evidence of respiratory infection and with a clear alternative explanation for respiratory failure (no-LRTI); (iii) LRTI defined by clinical criteria only (LRTI^+C); and (iv) unknown, LRTI possible (unk-LRT). A determination of noninfectious etiology was made only if an alternative diagnosis could be established and results of standard clinical microbiological testing for LRTI were negative.