Fig. 2: The impact of antibiotic prescription and geographic overlap on patient microbiomes.

a Aggregated prescription history of 20 of the most frequently prescribed antibiotic, antifungal and antiviral drugs. Each panel shows the percentage of patients who were prescribed a drug at the given day, relative to the date of HCT. Shannon diversity at the species (b) or genus (c) level compared to total antibiotic-days in the 7 days prior to sample collection. (d) Samples with or without a single species dominant (≥30%), compared with total antibiotic-days in the prior seven days. Boxes extend to the first and third quartile, whiskers extend to the upper and lower value within 1.5*interquartile range (IQR) from the box. Outliers are shown as points. A two-sided Wilcoxon rank-sum test was used to compute the p-value. n = 133 samples from 71 patients in FALSE category; n = 81 samples from 60 patients in TRUE category. Taxonomic similarity at the species level (1−Bray–Curtis dissimilarity) between samples from different patients, evaluated against days of hospital overlap (e) or hours of roommate overlap (f) prior to the earlier sample. Maximum inStrain popANI achieved by comparing all strains in all samples from two patients, evaluated against hours of roommate (g) or days of hospital overlap (h) prior to the earlier sample. In all panels, trend lines are calculated as the best-fit linear regression between the X and Y variables. R and p-values are the Pearson correlation coefficient and correlation p-value, respectively. Pip-Tazo Piperacillin-Tazobactam, TMP-SMX trimethoprim-sulfamethoxazole, PJP Pneumocystis jirovecii pneumonia. Source Data for this figure are provided in Supplementary Data 4–9.