Carbapenem-Resistant Klebsiella pneumoniae Drove Severe Adult Invasive Disease in Johannesburg
Adults hospitalized with Klebsiella pneumoniae invasive disease in Johannesburg faced a heavy burden of antimicrobial resistance and an in-hospital death rate that leaves little room for complacency. Across a multicentre observational cohort, 524 of 617 people with culture-confirmed bloodstream or cerebrospinal fluid infection were enrolled over 12 months, and most infections were already healthcare-associated by the time they were detected. The median age was 48 years, placing the syndrome squarely in working-age adults rather than only in the very elderly or the profoundly immunocompromised.
The microbiology tells the story quickly. Carbapenem resistance was present in 55.0% of isolates, a level that sharply narrows empiric treatment options and raises the stakes for rapid susceptibility testing. Among isolates tested for carbapenemase production, OXA-48 was detected in 71.5%, while OXA-48 and NDM co-detection appeared in 20.9%. Colistin resistance was also present, identified in 7.6% of tested isolates, a troubling finding because colistin often remains one of the last therapeutic options when resistance erodes standard beta-lactam regimens. In practical terms, these organisms were not only resistant to first-line agents; many were resistant to the fallback drugs clinicians often reach for when the initial regimen fails.
A Hospital-Acquired Pattern With Narrowing Treatment Options
Healthcare exposure shaped the epidemiology. Of the 524 enrolled patients, 84.4% had presumed healthcare-associated infection, making inpatient care itself part of the transmission and selection environment. Comorbidity burden was substantial, including HIV in 26.9% and diabetes mellitus in 16.4% of patients. Those conditions do not explain the infection burden on their own, but they create a biologically plausible setting for invasive Gram-negative disease by weakening host defenses, increasing prior antibiotic exposure, and amplifying the likelihood of device use and prolonged hospitalization.
The invasive phenotype here was not subtle. Blood and cerebrospinal fluid infections mark a high-acuity syndrome in which delays in effective treatment can be rapidly fatal, especially when the organism is multidrug-resistant. In a setting where more than half of isolates were carbapenem-resistant, empiric therapy is forced into a precarious balance between over-treatment and under-treatment. The resistance pattern also suggests sustained selective pressure in the hospital ecosystem, where antibiotic exposure, invasive devices, and transmission opportunities converge.
Mortality Exceeded 50% and Two Bedside Signals Stood Out
The in-hospital case fatality risk reached 56.5%, meaning more than half of the patients died before discharge. That figure is stark on its own, but it becomes more concerning when paired with the clinical factors linked to death. Urethral catheterization was independently associated with mortality, with an adjusted odds ratio of 3.30 and a 95% confidence interval of 1.51 to 7.23. An admission quick sepsis-related organ failure assessment score of 1 to 3 was also independently associated with death, with an adjusted odds ratio of 2.14 and a 95% confidence interval of 1.25 to 3.68.
One sentence captures the clinical reality: device exposure and early organ dysfunction tracked with death.
The association with urethral catheterization likely reflects more than a single procedural risk. Indwelling devices can serve as portals for infection, reservoirs for biofilm formation, and markers of severe underlying illness requiring intensive inpatient support. The quick SOFA signal aligns with that same biology. Even modest early organ dysfunction in sepsis can indicate a narrow physiologic reserve, and when the infecting organism is carbapenem-resistant, every hour of delay to effective therapy can further increase the chance of irreversible deterioration.
Source Control Remained the Clearest Protective Factor
The strongest modifiable signal in the cohort came from source control. Achieving source control was associated with markedly lower odds of death, with an adjusted odds ratio of 0.18 and a 95% confidence interval of 0.10 to 0.30. In invasive bacterial disease, source control can mean removal of infected devices, drainage of collections, or interruption of a nidus that keeps seeding the bloodstream. Here, that association suggests that bedside intervention retained real value even in a cohort dominated by resistant organisms.
This finding matters because antimicrobial resistance can create the impression that treatment is largely out of the clinician’s hands. It is not. When source control is delayed or impossible, antibiotics are asked to do all the work against a pathogen already equipped with multiple resistance mechanisms. When the nidus is removed early, the bacterial burden falls and host recovery becomes more plausible, even when the organism is difficult to treat. The data fit that physiology cleanly.
The Resistance Pattern Raises Broader Infection-Control Questions
The dominance of OXA-48 and the substantial co-detection of OXA-48 and NDM point to a complicated resistance ecology in Johannesburg hospitals. Carbapenemase combinations often leave clinicians with fewer reliable options and may also signal successful dissemination of highly adapted strains within and between facilities. The presence of colistin resistance, even at a lower frequency, further compresses the therapeutic window and heightens the importance of antimicrobial stewardship, rapid laboratory diagnostics, and infection prevention practices that limit spread.
For hospitalized adults with suspected invasive Gram-negative infection, these results argue for fast microbiological confirmation and early attention to device management. They also support systems that can identify carbapenemase production without delay, because the interval between culture collection and directed therapy may determine whether a patient survives to discharge. In a cohort where healthcare-associated disease predominated and mortality exceeded half of cases, empiric optimism has little clinical value.
Why These Data Matter Beyond One City
The study fills a major evidence gap for Africa, where the burden of Klebsiella pneumoniae invasive disease has been undercharacterized despite clear signals of rising antimicrobial resistance. Johannesburg’s hospital network revealed a disease pattern that is both microbiologically aggressive and clinically unforgiving. The median patient was not elderly, and many had pre-existing conditions that are common in routine inpatient practice, which means the syndrome is not confined to a rarefied specialist population.
The practical implication is blunt: hospitals facing similar resistance patterns need faster diagnostics, stronger source control pathways, and tighter device stewardship. The mortality signal is already here, and the resistance profile leaves little margin for late action. Future work will need to determine how much of this burden can be reduced through earlier effective therapy, better catheter avoidance, and infection-control measures that interrupt transmission before these strains reach the bloodstream or central nervous system.