Copyright 2003 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2003
My colleagues and I read with interest the letter from Dr Rozzini and colleagues. We believe their findings generally concur with our observations and look forward to seeing more information on other nonfatal outcomes from this important cohort study. Of interest, Dr Rozzini and colleagues found that community-acquired pneumonia (CAP) per se was associated with increased 6-month mortality in the crude analysis (I presume this was univariate Cox modeling) but that the association was no longer statistically significant in the adjusted analysis (I presume this was the multivariate Cox modeling). Although this may suggest CAP is not an independent risk factor, it is also possible that there is some collinearity between CAP and other measures such as APS-APACHE II. I also note that the confidence intervals for the other risk factors are very wide, raising concern about the stability of the model and/or sample size. Finally, we and others1- 3 have found that many hazards after severe infection or sepsis are not proportional, making the use of the standard Cox model problematic. Nevertheless, the findings certainly support the contention that patients who develop CAP remain at high risk for a wide set of problems that extend considerably beyond hospital discharge.
Angus DC. Is Pneumonia Still the Old Man's Friend?—Reply. Arch Intern Med. 2003;163(12):1492. doi:10.1001/archinte.163.12.1491