BACKGROUND: In spite of interventions, approximately 1000 per 1,000,000 platelet (PLT) collections are contaminated with bacteria at collection. The current prestorage culture procedure at some blood centers is to inoculate a fixed volume from the collection bag (4-8 mL) regardless of collection volume. The sensitivity of early testing varies with the percent of collection volume sampled. We applied the Poisson model to determine whether sampling larger volumes might increase detection at pertinent contamination levels.

STUDY DESIGN AND METHODS: The intervention was testing a fixed proportion of the collection volume from single, double, and triple collections. The Poisson model was applied to blood center data to calculate weighted average detection. Model 1 consisted of inoculating 3.2% of the collection volume from single, 1.6% from double, and 1.2% from triple PLT procedures (8 mL in each case). Model 2 consisted of inoculating 3.8% of the collection volume from all PLT procedures. Volume-related and non–volume-related contamination mechanisms were evaluated.

RESULTS: Testing constant proportions of the collection volume (Model 2) increases percent detection over testing constant volumes (Model 1) (68% vs. 41% detection if contamination is 30 colony-forming units (CFUs)/collection bag and 17% vs. 9% detection if contamination is 5 CFUs/collection bag). At low levels of contamination (approx. 5 CFUs/bag), the intervention might double the number of contaminated units detected.

CONCLUSION: Based on the application of the Poisson model to detection of bacteria in PLT concentrates, inoculating cultures with slightly consistent proportions of the collection volume should lead to a reduction in false negative tests and in the number of contaminated units transfused.