Longitudinal prescribing patterns tracked alongside strain shifts in Canadian bacteraemia
A two-decade genomic cohort from Calgary offers a rare view of how antibiotic pressure and bacterial population structure move together in real-world practice. By combining prescribing data with whole-genome sequences from five major causes of bacteraemia, investigators mapped resistance trends across Staphylococcus aureus, Enterococcus faecalis, Enterococcus faecium, Escherichia coli, and Klebsiella pneumoniae in a Canadian metropolitan health zone between 2006 and 2022.
The study moves beyond static resistance surveillance. It places clinical microbiology, genomic epidemiology, and antimicrobial use into the same frame, which matters because resistance rarely behaves as a single organism-wide trait. Instead, selective pressure acts on particular lineages, resistance genes, and ecological niches. In bloodstream infection, where treatment delays and empiric decisions carry immediate consequences, the composition of circulating strains can alter the practical meaning of every prescribing trend.
Between 2010 and 2022, fluoroquinolone prescribing declined in both community and hospital settings, falling by an average annual percentage change of −6.8% in the community and −5.1% in hospitals. Resistance to fluoroquinolones in Gram-positive species fell in parallel. The clearest signal appeared in S aureus, where resistance to cloxacillin, ciprofloxacin, erythromycin, and clindamycin declined from 2006 to 2022, with the steepest reductions in hospital-onset cases.
The hospital-onset decline was abrupt in epidemiologic terms.
Gram-positive resistance fell as fluoroquinolone use declined
For S aureus, the resistance trajectory suggests that reduced fluoroquinolone exposure may have eased selection pressure on hospital-adapted clones carrying multi-drug resistance profiles. Fluoroquinolones are not the only force shaping methicillin-resistant or otherwise resistant S aureus, but they can co-select for mobile resistance elements and broaden the fitness advantage of strains already equipped to survive repeated antimicrobial exposure. When use drops across a population, the advantage of those lineages can erode.
The study’s stratification by infection onset sharpens the interpretation. Hospital-onset bacteraemia often reflects concentrated antimicrobial exposure, invasive procedures, and sustained transmission chains, while community-onset cases track broader outpatient prescribing and background carriage. In this dataset, hospital-onset S aureus showed the sharpest fall in clinically important resistance, consistent with the possibility that stewardship in acute care settings can influence local strain ecology more quickly than outpatient trends alone.
The analysis also included Enterococcus species, organisms that often occupy a different resistance niche. Although the summary emphasizes the clearest resistance changes in S aureus and E coli, the inclusion of E faecalis and E faecium matters because enterococci frequently mirror the cumulative impact of hospital antimicrobial pressure. Their genetic architecture is distinct, and their resistance pathways do not always track with the same drugs or the same ecological rules as staphylococci and Enterobacterales.
E coli resistance clustered in ST131 and the expanding ST1193 lineage
The most consequential Gram-negative finding centered on E coli. Ceftriaxone and ciprofloxacin resistance clustered in ST131 and the emerging ST1193 lineage, but the two strains did not behave identically over time. ST1193 increased steadily, especially in community-onset cases, with an average annual percentage change of 17.7%. That pattern is clinically important because community-onset bloodstream infection is where outpatient prescribing patterns, colonization reservoirs, and first-line empiric choices intersect.
The genomic detail deepens the story. CTX-M-27-producing ST131 strains increased substantially from 2008 to 2022, while CTX-M-14-producing ST131 declined from 2013 to 2022. These are not interchangeable shifts. Different CTX-M variants reflect different evolutionary paths within a successful clonal background, and changes in their relative frequency can signal changing antibiotic selection or altered transmission advantages. The rise of CTX-M-27 within ST131 suggests that the lineage is not static; it is adapting internally while retaining its broader epidemic success.
One line in the dataset runs in the opposite direction from the fluoroquinolone story: community cephalosporin prescribing increased by 7.3% per year between 2010 and 2022. In a bacterium like E coli, where extended-spectrum beta-lactamases can reshape susceptibility to ceftriaxone and related agents, rising cephalosporin use can reinforce the dominance of resistant clones rather than displace them. The community setting becomes especially relevant when a lineage such as ST1193 expands outside the hospital walls, because outpatient exposure can feed the same resistance circuitry that hospitals later confront at admission.
K pneumoniae showed a different epidemiologic pattern
In K pneumoniae, hypervirulent ST23 was the most common strain, with 88 isolates identified. Its incidence increased modestly between 2006 and 2019, although the trend estimate crossed uncertainty bounds. Even without a definitive directional claim, the repeated recovery of ST23 signals that hypervirulent lineages deserve attention alongside classical multidrug-resistant clones. Hypervirulence and antimicrobial resistance are not the same phenomenon, yet when they converge in a bloodstream pathogen, the therapeutic stakes rise quickly.
This pathogen-specific divergence matters because it resists any simple idea that lower antibiotic use alone will uniformly reduce resistance across all species. Gram-positive organisms in this cohort moved in one direction, while the Gram-negative ecology of E coli moved in another. K pneumoniae followed its own lineage pattern, driven by strain success as much as by drug pressure. The biology explains why stewardship programs cannot rely on a single lever. Different bacteria carry different resistance mechanisms, different fitness costs, and different routes into the bloodstream.
Genomic surveillance now sits at the center of resistance control
The value of this study lies in its breadth and its time span. It is the first regional analysis to integrate longitudinal clinical, genomic, and antibiotic prescribing data across five key pathogens over nearly two decades. That combination makes it possible to see when resistance falls with prescribing, when it decouples from drug use, and when a successful lineage expands despite broader ecological change.
For clinical teams, the practical consequence is straightforward. Resistance surveillance should no longer stop at species-level antibiograms, because species labels hide the strain dynamics that drive real-world treatment failure. A ceftriaxone-resistant E coli burden, for example, may be dominated by a small number of expanding clones rather than by random accumulation of resistance across the species. The same logic applies to S aureus and the enterococci, where hospital and community reservoirs interact in ways that routine summaries cannot fully capture.
The study also points to a geographic truth about antimicrobial resistance control: stewardship confined to hospitals will miss part of the problem. Community prescribing shaped the rise of cephalosporin exposure, and community-onset infections carried major portions of the resistant E coli signal. Effective control will need surveillance that follows strains across care settings, because the same lineage can circulate silently in the community before arriving as a bloodstream infection in hospital.
The remaining uncertainty is not whether resistance evolves, but how quickly health systems can detect strain-level shifts before they alter empiric therapy. This cohort shows that the answer depends on linking prescribing data with genomic surveillance, not on either measure alone. Without both, the rise of a lineage such as ST1193 or the fall of a resistant S aureus clone can look like noise instead of an early warning signal.