Clinical relevance: This study investigates the effect of an initial specimen diversion device on the rate of 19 culture contamination in hospitalized patients. It finds that the device is associated with a significant reduction in 20 contamination. This intervention may result in a reduction in costs, antibiotic use and duration of hospitalization. 21
22Word count: Abstract -236 words; Introduction through Discussion -1724 words. 23 without the device, with 24 (5.2%) contaminated cultures (p < 0.008). No significant difference was shown in the 36 rate of true positive cultures. Thus, use of a diversion device was associated with reduced culture contamination in 37 hospitalized patients over a six month period without concomitant reduction in true positive cultures. This 38 intervention may result in a reduction in costs, antibiotic use and duration of hospitalization. 39 40 41 42 Ease of use of the device was evaluated using a standardized survey of phlebotomists and others involved in 108 sample collection. 109 Data was analyzed using Win-Pepi (version 11.65). Proportions were compared using the χ 2 -score or Fisher's 110 exact test, where appropriate. Continuous variables were compared by Student's t-test or Mann-Whitney-111 Wilcoxon test where appropriate. Two-tailed p values were taken and a p value <0.05 was defined as significant. 112 The study was approved by the Institutional Review Board of Shaare Zedek Medical Center. 113 114 115 Of the control group cultures, 44 were defined as true positives resulting in a true positive rate of 9.5%. Of the 128 study group cultures, 16 were defined as true positives resulting in a true positive rate of 7.7%. Enterobacteriaceae 129 and staphylococci were the most common causes of true bacteremia in both groups. No significant difference in 130 the true positive rate or in the microbiological characteristics of true positive cultures were noted between the 131 groups (Table 2 and Figure 1). 132In a survey of ease of device use, all respondents who had used the device more than once considered it 133 slightly or moderately more difficult to use than standard practice (grade 2 or 3 out of 5). 134
Discussion
135A substantial fraction of positive blood cultures represent contamination, rather than true bloodstream 136 infection 2,3 , including in the control group of our study (Figure 1), where 24 out of 464 (5.2%) cultures 137 represented contamination and 44 (9.5%) represented true infection. These false positive cultures, at the