Aerobic gram negative rod, it is second to Escherichia coli as a pathogen in humans. Includes K. pneumonia, K. oxytoca, K. ozaenae, and K. rhinoscleromatis.
Part of the human gut. Curiously, bloodstream infections are most common in the summer (PubMed). There has been one outbreak in a hospital due to transmission with hospital food, so there can be more than just the flavor that is off (PubMed).
K. pneumonia is best known for lobar pneumonia, but can cause other infections: cystitis, liver abscess in diabetics etc. The hyperviscous strains (they are like snot on a plate) have a particular habit of causing eye and liver abscess. In Taiwan (and elsewhere) (PubMed) it is a cause on mono-microbial necrotizing fasciitis (Pubmed). There may be an association between liver abscesses and malignancy (PubMed).
K. rhinoscleromatis causes a chronic granulomatous infection (PubMed) of the upper airway, not uncommon (does this mean common?) in Eastern Europe, central Africa, South America, and Asia. More common in them what mate with their cousins.
Given increasing resistance of this organism to antibiotics, you have to know your local resistance patterns to give empiric therapy. Sorry. You actually have to know what you are doing. The most reliable agents are aminoglycosides, carbapenems, quinolones and third generation cephalosporins.
Be wary. ESBL (extended spectrum beta lactamase that hydrolyse most b-lactams (sparing only cephamycins and carbapenems) carrying Klebsiella are increasing in frequency and the only reliable antibiotic are carbapenems. Risks for ESBL include health care, urinary catheters and prior antibiotics and if you choose wrong there is increased mortality (PubMed). If the MIC for an ESBL to piperacillin/tazobactam is <= 2 the patient will do fine but if higher MICs expect death (PubMed). For bacteremia (or suspected) from and ESBL, carbapenems have less mortality than piperacillin/tazobactam (PubMed). Maybe (PubMed).
If the organism is an ESBL, DO NOT use cefepime even if susceptible. Won't work as well as carbapenems (PubMed) and avoid piperacillin/tazobactam. Remember that sensitive in the lab does not always mean effective in the patient.
There is also the New Delhi metallo-beta-lactamase, which gets all beta-lactams including carbapenems aka CRE (PubMed); often these are only sensitive to colistin and tigecycline. Cetazadime-avibactam.html has a high failure rate with CRE.
CRE are perhaps more virulent as well as hard to kill(PubMed).