A statistical model for assessing sample size for bacterial colony selection: a case study of Escherichia coli and avian cellulitis

A general problem for microbiologists is determining the number of phenotyp ically similar colonies growing on an agar plate that must be analyzed in o rder to be confident of identifying all of the different strains present in the sample. If a specified number of colonies is picked from a plate on wh ich the number of unique strains of bacteria is unknown, assigning a probab ility of correctly identifying all of the strains present on the plate is n ot a simple task. With Escherichia coli of avian cellulitis origin as a cas e study, a statistical model was designed that would delineate sample sizes for efficient and consistent identification of all the strains of phenotyp ically similar bacteria in a clinical sample. This model enables the microb iologist to calculate the probability that all of the strains contained wit hin the sample are correctly identified and to generate probability-based s ample sizes for colony identification. The probability of cellulitis lesion s containing a single strain of E, coli was 95.4%. If one E. coli strain is observed out of three colonies randomly selected from a future agar plate, the probability is 98.8% that only one strain is on the plate. These resul ts are specific for this cellulitis E. coli scenario. For systems in which the number of bacterial strains per sample is variable, this model provides a quantitative means by which sample sizes can be determined.