Fig. 4: Evidence S. pneumoniae and clindamycin MIC dynamics are the result of database curation methodologies.
From: Seeking patterns of antibiotic resistance in ATLAS, an open, raw MIC database with patient metadata
Show an analogous structural shift to Fig. 1B, C for S. pneumoniae and erythromycin: the correlelogram (in A) and MIC dynamics (in B) shift towards lower MICs around 2011. The result of separating ATLAS into its (C)) TEST and (D)) INFORM components are then shown. These indicate the loss of the most-resistant cluster from B around 2011 is due to TEST and INFORM both having bimodal MIC distributions with different high-MIC clusters (see C(right) and D(right)). When both the latter are amalgamated to form ATLAS, noting (D) has data only after 2012, this appears (in (B)) to change the structure and dynamics of the MIC distributions when, in fact, these are artefacts reflecting the merging of 2 datasets.
