REPLY

Reply: The authors thank Dr Myrianthefs et al. for the interest in our recently published article on γ-globulin levels in patients with community-acquired septic shock (1). These authors reported a lower incidence (32% vs. 57%) of low immunoglobulin G (IgG) levels in a cohort of 38 patients than we did in our study population. In addition, they did not find any difference in the duration of vasopressor therapy or mechanical ventilation between the two populations. These discrepancies could be explained by the different populations. First, they studied intensive care unit (ICU)-acquired infections, whereas we studied community-acquired infections. Second, we excluded patients with monoclonal production of IgG (4 of 29 patients screened) or with hematologic diseases. It is quite possible that in a hospital population with multiple comrplications and pre-existing diseases, some patients could have monoclonal production of IgG and, thus, be, perhaps wrongly, considered as having normal IgG levels. We can expect that IgG levels will eventually return to the reference ranges some time after the onset of the septic shock. We did not document this because the ICU stay was quite short (median, 6 days) and we did not check immunoglobulin levels after ICU discharge. In contrast, the median ICU stay in the study from Myrianthefs et al. was more than 30 days. Mortality in the study by Myrianthefs et al. was much higher than that in our study (42% vs. 29%). Interestingly, the combined mortality for the two studies was 9 (26%) of 35 in the "normal IgG" level group compared with 13 (54%) of 24 in the "low IgG" level group (P = 0.03), with a relative risk of death of 2.11 (1.07 - 4.13) and an odds ratio of 3.41 (1.13 - 10.30). This supports our conclusion that patients with septic shock and low IgG levels have a higher mortality rate than patients with normal IgG levels, and that this population could be a target for the administration of i.v. immunoglobulins. NB: Table 1: The mortality in the "normal IgG" level group is 9 of 26, i.e., 34%, and not 26.9%. We could not confirm the P value presented by the authors (P = 0.29 using the Fisher exact test and P = 0.17 using the chi-square test). We suppose that there was an error in the Table. Fabio Silvio Taccone Daniel De Backer Jean-Louis Vincent Erasme Hospital Université Libre de Bruxelles Bruxelles, Belgium