Stephen J.LurieMD, PhD, Senior EditorIndividualAuthorJody W.ZylkeMD, Contributing EditorIndividualAuthor
In Reply: Although the definitions we used to select case and control patients could imply that some patients may have developed infection prior to exposure to risk, this actually was not the case in any of the patients. All case and control patients received ventilatory support within the first 24 hours after admission.
Drs Kagramanov and Lyman are concerned about restriction of the sample size resulting from the matching process. Without this restriction, which resulted from a careful and strict matching of control with case patients, the comparison of 2 cohorts may have resulted in larger differences between the groups in terms of infection or other outcome variables, but the resulting bias would have made the comparison difficult to interpret. Indeed, severity of the patients' illnesses on admission as well as other important variables would not have been comparable between the 2 groups, and the evidence of a protective effect of NIV on infections would have been much less convincing. In addition, we indicated that case patients who could not be matched with control patients had similar characteristics and outcomes to those who could be matched.
Girou E, Brochard L. Noninvasive Ventilation and Nosocomial Infection—Reply. JAMA. 2001;285(7):881. doi:10.1001/jama.285.7.879