More Comments on IV Filter Debate

Abstract

To the Editor: Frankly, an exchange of letters in your September 1987 issue' and the Product Commentary concerned seem to cloud important issues more than clarify them. Brief mention is made of studies, and inferences are drawn without detailed summary of supportive evidence from those studies. Considerable attention is drawn to the report by Quercia et al,' in which a highly sensitive method is used to detect contamination of IV administration sets containing real or blank filter cartridges. The study is doubleblinded and stated to be randomized, although the method of random assignment is not provided. It purports to support a conclusion that IV filters could be supplied to all 1CU patients in that center for an annual cost of approximately $5,700 versus annual treatment costs of $168,000 for currently unprevented septicemic episodes. However, it is difficult to interpret the data of Quercia et al for several reasons. First, on what basis can one assume that all bacteremias detected during the study were attributable to the IV sets? Thirteen episodes were detected; the authors state that five of ten invol\,ed blank filterset patients with blood culture isolate(s) matching their filterset isolate(s). What of the other five? "The microorganisms isolated from the blood of the remaining bacteremic patients with blank filtersets were the same as those isolated from other contaminated f-iltersets." 'I-his finding suggests a definite lack of concordance, complicated by an absence of clear data stratification in 2 x 2 tables and/or statistical analysis. If one generously assumes that all bacteremic episodes were IV-attributable, then a statistically significant relationship (by Fisher's exact test) is dem