• 
  

  Empiric antimicrobial therapy for acute severe urosepsis should initially include two agents with activity against gram-negative bacilli, such as a third- or fourth-generation cephalosporin, aztreonam, or extended-spectrum penicillin in combination with either a fluoroquinolone or an aminoglycoside.

  
  
• 
  

  Where local epidemiology indicates significant prevalence of extended-spectrum ß-lactamases among Enterobacteriaceae, then a carbapenem such as imipenem, meropenem, ertapenem, or doripenem is preferred while awaiting definitive cultures.

  
  
• 
  

  Where local epidemiology indicates significant prevalence of carbapenem-resistant Enterobacteriaceae, then colistin and a carbapenem should be chosen while awaiting definitive cultures.

  
  
• 
  

  Urine and blood cultures should be obtained prior to the first antimicrobial doses, which should be given without delay.

  
  
• 
  

  Once the pathogen is identified by a positive urine or blood culture, the antimicrobial regimen should be tailored to a single, least toxic agent with the narrowest spectrum, based on susceptibility data.

  
  
• 
  

  Patients with severe urosepsis requiring ICU admission should have imaging of the urinary tract on an urgent basis, preferably by computed tomography with intravenous contrast, because suppurative complications require drainage as a priority.

  
  
• 
  

  Percutaneous drainage by an interventional radiologist is generally preferred to drain definitively or stabilize temporarily a patient with suppurative complications.

  
  
• 
  

  Urinary catheters cause a high incidence (3%-7% per day) of bacteriuria and candiduria; the latter associated with broad-spectrum antimicrobial therapy.

  
  
• 
  

  Asymptomatic catheter–associated bacteriuria or candiduria should not be treated; the only exceptions are transplant and neutropenic patients, and before instrumentation of the urinary tract.

  
  
• 
  

  The continued usefulness of a urinary catheter should be reassessed on a regular basis, and removal in selected patients should be considered.

  
  
• 
  

  Fever or sepsis should only be attributed to catheter-associated bacteriuria and treated only after exclusion of other potential causes of infection.

Community-acquired pyelonephritis sometimes leads to sepsis syndrome and intensive care unit (ICU) admission, especially when it arises in an obstructed urinary tract or when the host defense is compromised by poorly controlled diabetes. Bacteremia arises in about 15% of cases, at a rate of 50 per 100,000 person years, with a 28-day mortality of about 5%.¹ Urinary tract infection (UTI) is also a common sequel of ICU, because of the use of urinary catheters for in excess of 70% of ICU patient days² and ranks in the top three or four of ICU-acquired infections.^3,4 Although older data suggested that catheter-associated UTI caused mortality,^5,6 more recent studies that control for confounding factors show no such link.^7-10 In addition the evidence that bacteriuria prolongs ICU stay or increases cost has also been challenged.¹¹ Unfortunately asymptomatic bacteriuria is frequently screened for and treated, resulting in harmful and unnecessary antimicrobial therapy.

Quantitative culture methods distinguish true bacterial multiplication within the urinary tract from a false result due to procurement contamination. Significant bacteriuria is defined as ≥10⁵ organisms/mL. In the presence of urinary symptoms, a count of ≥10² organisms/mL from a woman with pyuria represents true infection.¹² In a catheterized patient with symptoms or signs of UTI without other explanation, a criterion of ≥10³ organisms/mL represents catheter-associated urinary tract infection (CAUTI).¹³ Pyuria, the presence of white blood cells in urine, is best measured with a counting chamber, using a criterion of ≥10 cells/µL; alternatively a criterion of ≥5pus cells/high powered field is used. Pyuria has a sensitivity of 96% for symptomatic UTI, but should not be used to support a diagnosis of UTI in the catheterized patient. In assessing a patient with sepsis of unknown source, whether in the community or health care system, whether catheterized or not, the absence of pyuria is useful at excluding a urinary source of sepsis.¹³

CLINICAL FEATURES OF ACUTE PYELONEPHRITIS

Acute pyelonephritis is a syndrome of fever with evidence of renal inflammation, such as costovertebral angle tenderness or flank pain, usually accompanied by signs of systemic toxicity. In the very elderly or cognitively impaired, acute confusion may be the sole manifestation. Patients with spinal cord injury are especially prone to silent and complicated pyelonephritis associated with urinary obstruction due to calculi. They often will have fever with nonspecific abdominal discomfort, increased spasms, and autonomic dysreflexia. Patients admitted with acute pyelonephritis who subsequently deteriorate should be urgently investigated for urinary tract obstruction and for a suppurative focus. Alternatively, such deterioration may be due to a resistant urinary pathogen. Features that suggest obstruction include renal colic and severe costovertebral angle tenderness. Blood and urine cultures should always be obtained promptly before therapy, if necessary by a single “in-out” urinary catheterization. Acute renal failure caused directly by pyelonephritis or renal suppuration is rare¹⁴ and if present, is usually due to sepsis, hypotension, or drug toxicity.

MICROBIOLOGY

Most (80%) community-acquired UTIs in women are caused by Escherichia coli. Staphylococcus saprophyticus is the second most common community urinary pathogen in younger women, but it virtually never gives rise to sepsis and ICU admission. Enterobacteriaceae other than E coli (eg, Klebsiella, Proteus, Enterobacter, Citrobacter, Morganella, Providentia, Serratia species), Enterococcus species, Pseudomonas aeruginosa, and Candida species are each uncommon. In men, E coli is also the commonest community-acquired urinary pathogen, but other Enterobacteriaceae and enterococci are more frequent than in women. The spectrum changes to more resistant species when UTI arises in the ICU. Gram-negative bacilli account for about half, consisting of Enterobacteriaceae such as E coli (21%-23%), Klebsiella (9%), and Enterobacter species (4%), along with the nonfermenters P aeruginosa (10%) and Acinetobacter baumannii (1%); Candida species account for 21% to 29% and enterococci for 15%.¹⁵ This predominance of more resistant species reflects the more widespread use of broad-spectrum antimicrobial agents and the use of urinary catheters in the ICU. Bacteriuria or candiduria, acquired through urinary catheterization, constitutes a reservoir of resistant pathogens, which can occasionally be a source of epidemic spread of resistant infection within the ICU.

Almost all UTIs arise by the ascending route, but in the case of Candida species and S aureus, the kidneys are sometimes seeded through the bloodstream. Over 3 years, only four such bacteremic/fungemic infections were identified, compared to 356 ascending UTIs in the same period in a Canadian ICU.⁸ Isolation of S aureus in a urine culture should prompt a search for evidence of invasive staphylococcal infection elsewhere. Multiple or single cortical renal abscesses (renal carbuncle) may be present.

EMERGING ANTIMICROBIAL RESISTANCE IN ENTEROBACTERIACEAE

In the 21st century, the prevalence of antimicrobial resistance in gram-negative bacilli has increased dramatically worldwide so that many previously effective antimicrobial regimens are no longer reliable for UTI. In particular both extended-spectrum-ß-lactamase (ESBL)–producing and carbapenem-resistant Enterobacteriaceae (CRE) have emerged as major challenges. The genes for both ESBLs and CRE are on plasmids, which also commonly carry genes for resistance against most non-ß-lactam antimicrobial agents. Some are prone to interspecies transfer, resulting in widespread dissemination. Globalization has facilitated spread, especially for Enterobacteriaceae, which constitute such a huge proportion of the normal colonic flora. The absence of new agents for gram-negative bacilli in development raises the prospect of increased mortality for patients with urosepsis in the second decade of the 21st century. However, in 2011, there is great geographic variation in resistance and in many regions community-acquired urinary pathogens still remain susceptible to standard antimicrobial regimens (Table 75-1).

Prior to 2000, resistance to fluoroquinolones in gram-negative bacilli was rare in the community and was only significant in P aeruginosa in the ICU. Fluoroquinolones were ideal urinary agents, with a broad spectrum of activity against urinary pathogens, excellent bioavailability facilitating oral switch, and high volume of distribution favoring tissue penetration. Predominantly renal excretion results in high urinary levels of ciprofloxacin, ofloxacin, and levofloxacin, but not of moxifloxacin or gemifloxacin; the latter two agents are not recommended for UTI.¹⁶ With the emergence of widespread resistance, fluoroquinolones can no longer be relied upon empirically as monotherapy for severe urosepsis. The prevalence of resistance to ciprofloxacin in gram-negative bacilli, including P aeruginosa, increased from 14% in 1994 to 24% in 2000, in the ICUs of large US hospitals. This corelated with a 2.5-fold increase in national fluoroquinolone use.¹⁷ By 2011, nearly half of all E coli urinary isolates were resistant to fluoroquinolones in the Asia-Pacific region.¹⁸ By contrast in 2007 to 2009, the rate of such resistance in urinary pathogens in Canadian ICUs was 19%¹⁹ and in the community was 6%.²⁰ Resistance to third- and fourth-generation cephalosporins in Enterobacteriaceae was previously rare but abruptly increased in the first decade of this century and is now common worldwide. By 2010, in excess of 650 molecular varieties of ESBLs had been described, which had evolved from older narrower spectrum ß-lactamases, such as TEM-1, TEM-2, and SHV-1, by one to four amino acid substitutions. The resultant increased resistance in third- and fourth-generation cephalosporins and aztreonam is mainly due to Group-1 CTX-M enzymes, principally CTX-M-15. In addition, these strains are usually resistant to fluoroquinolones, aminoglycosides, and all ß-lactams except carbapenems. This multidrug resistance is due to a linkage on the plasmid of the CTX-M-15 gene^* to a gene for aminoglycoside resistance^† that also deactivates fluoroquinolones and to a gene^‡ encoding an inhibitor-resistant penicillinase. Klebsiella pneumoniae organisms with ESBLs have caused common source ICU outbreaks, including UTI. Of greater concern, E coli isolates with ESBLs have become increasingly common worldwide causing infection in both the hospital and community, including pyelonephritis. One-third of recent E coli urinary isolates in the Asia-Pacific region had ESBLs, including 60% of isolates from India.¹⁸ An ICU in China reported that 80% of E coli isolates between 2003 and 2007 were resistant to third-generation cephalosporins.²¹ In the United Kingdom and Ireland, the proportion of E coli bacteremia resistant to third-generation cephalosporins began to increase steadily from 2% in 2001 to 12% by 2007.²² Many of these were of urinary origin in the community due to spread of CTM-15, often due to the international O25:H4-ST131 clone,²³ which is commonly fluoroquinolone resistant. It had also been reported in many other countries by 2010, including France, Switzerland, Spain, Portugal, Turkey, Lebanon, Korea, Canada, and USA.^24,25 The need to cover ESBLs has driven a more extensive use of carbapenems for empiric therapy, sometimes even for pyelonephritis, eroding the reserve role of carbapenems.

Carbapenem-resistant Enterobacteriaceae consist mainly of strains producing carbapenemases, which are subdivided into metallo-β-lactamases (IMP, VIM, NDM) or nonmetallo enzymes (KPC, OXA-48). Additionally, a small minority of carbapenem resistance in Enterobacteriaceae is due to a chromosomal mutation leading to a porin loss in combination with hyperexpression of an ESBL or AmpC enzyme. Most CRE strains are resistant to all other β-lactams and often have multiple aminoglycoside-modifying enzymes; those with NDM-1 carbapenemase typically also have 16S rRNA methylases, conferring high-level resistance to all aminoglycosides. The vast majority are also fluoroquinolone, trimethoprim-sulfamethoxazole, chloramphenicol, and tetracycline resistant.

In the 1990s, a metallo-ß-lactamase termed IMP, emerged in Japan in the context of widespread use of imipenem²⁶ and subsequently occurred sporadically worldwide but remained rare. Verona imipenemase (VIM), a new metallo-ß-lactamase, was first isolated in Europe from 1996, but initially was confined to P aeruginosa. VIM-1–producing K pneumoniae was first reported extensively in 2002 in the ICUs of three Athens teaching hospitals and by the end of the decade had been reported from the majority of Greek hospitals, with 50% of K pneumoniae demonstrating carbapenem resistance in ICUs by 2007.²⁷ Some Greek ICUs reported that few of their K pneumoniae isolates were susceptible to carbapenems.²⁸ The gene for this enzyme, bla_VIM-1, is on a plasmid, usually linked to other resistance genes, including an ESBL and an aminoglycoside resistance gene. Many isolates with bla_VIM-1 were initially reported as susceptible to carbapenems by automated laboratory systems, leading to an underappreciation of the Infection Control challenge.

In the USA, a single case of carbapenem-resistant Enterobacteriaceae was reported first in 1991, due to a combination of AmpC hyperproduction with a porin mutation.²⁹ A limited outbreak (eight cases, six deaths) of carbapenem-resistant K pneumoniae by this mechanism was next reported in a surgical ICU in New York City in 1999. This was superimposed on a prolonged ESBL outbreak that had necessitated extensive use of imipenem for empiric therapy.³⁰ The outbreak was contained and terminated by vigorous Infection Control and limitation of cephalosporin use.

Klebsiella pneumoniae carbapenemase (KPC), a new nonmetallic ß-lactamase was first detected in North Carolina in 1999.³¹ The gene, bla_KPC, is on a plasmid and most are part of a genetically distinct clone, sequence type (ST) 258. This is currently the main mechanism of carbapenem resistance in the United States, concentrated in New York City and the North Eastern States. By 2010, KPCs had been reported from 36 states, Washington DC, and Puerto Rico and had also spread worldwide, most extensively in Israel and Greece.

By 2010, KPCs were endemic in most Israeli and Greek hospitals, adding to the already established problem due to VIM-1 in Greece.^32-34 In Israel, a National Infection Control response had reduced the number of new KPC cases to 30 per month by 2010. Sporadic cases and outbreaks of both KPC and VIM have occurred throughout Europe, usually linked to the larger Greek outbreak.^35,36 A hospital outbreak of KPC in Warsaw began in 2008 and was followed by extensive outbreaks throughout Poland.³⁵ Klebsiella pneumoniae carbapenemases have also emerged in various Latin American and Asian countries. Infections due to KPC and VIM-producing organisms are associated with a higher mortality despite usually remaining susceptible to colistin and tigecycline. As a result of widespread use of colistin in Greek ICUs, colistin resistance has emerged, spread outside of Greece, and been associated with very high mortality.^35,37,38 Isolates demonstrating resistance to all available agents, including colistin and tigecycline, have also been reported.^28,39 Carbapenem-resistant K pneumoniae has followed the pathways of patient referral, causing hospital outbreaks along the way, and has remained largely a health care–associated pathogen, including CAUTI. Interspecies transfer of bla_KPC to E coli and other Enterobacteriaceae has occurred but not commonly and therefore organisms with KPCs are not yet significant community uropathogens.

In 2008, the first of a new carbapenemase, termed New Delhi metallo-β-lactamase (NDM-1), was identified in a urinary isolate of K pneumoniae in Sweden. The patient had recently returned from India with a urinary catheter following extensive hospitalization. An E coli with the same NDM-1 was subsequently found in his stool, consistent with interspecies transfer.⁴⁰ In 2010, a landmark report of NDM-1 isolates from throughout the United Kingdom and Indian subcontinent identified 180 isolates with NDM-1, mostly K pneumoniae or E coli, 37 from the United Kingdom, and the rest from 20 different centers in India, Pakistan, and Bangladesh.⁴¹ At least 17 of the UK patients had traveled to India or Pakistan over the previous year, 14 of who were hospitalized there. Subsequent reports emerged from elsewhere in Asia, Europe, Australia, Kenya, Canada, and USA, many linked to travel and some to hospitalization in India. Many of the isolates were urinary from patients with UTI or from those who had been catheterized. A large proportion of the Indian isolates were from community-acquired infections, especially UTI. A study of stool carriage at two Pakistani military hospitals showed NDM in 27% of inpatients and 13% of outpatients.⁴² Enterobacteriaceae isolated in 2009 from India showed that 28% were CRE, half of which were NDM-1.^42,43 New Delhi metallo-ß-lactamase is likely widespread in K pneumoniae and E coli in both hospitals and community in India and Pakistan and has been reported from remote districts.^44,45 However, the absence of population-based surveillance data and a National Reference Laboratory for resistant bacteria in both India and Pakistan means that uncertainty remains about the precise epidemiology. In 2011, the extent to which these various forms of resistance, especially NDM-1, will impact on both community- and hospital-acquired UTIs throughout the world over the second decade of the 21st century remains uncertain, but the negative potential is enormous.

ENTEROCOCCI AS URINARY PATHOGENS

Enterococcus species are the third most frequent cause of CAUTI in the ICU but are uncommonly a source of bacteremia and sepsis. A Spanish prospective multicenter observational study of 21,979 ICU admissions between 1997 and 2001 found enterococcal UTI in 0.15% of the total.⁴⁶ In ICUs the proportion of vancomycin-resistant enterococci (VRE), mostly E faecium, is increasing but varies geographically. In many countries, VRE has become endemic in hospitals, with a median ICU colonization rate of 10% in the United States, with some units as high as 59%.⁴⁷ In contrast VRE in France has only occurred in limited outbreaks, curtailed by effective Infection Control measures. Vancomycin-resistant E faecium also commonly exhibits resistance to ampicillin and high-level aminoglycosides.⁴⁸ A US tertiary center reported a rate of bacteriuria due to VRE to be 23 per 10,000 admissions. Of 107 positive cultures, there were 13 symptomatic UTIs, two associated with bacteremia and one associated with death.⁴⁹ Almost all VRE are susceptible to linezolid, daptomycin, nitrofurantoin, tigecycline, and chloramphenicol.⁵⁰

ANTIMICROBIAL THERAPY

The choice of antimicrobial therapy for urosepsis raises a conflict between the need to cover all possible pathogens and to avoid the adverse effects of excessively broad-spectrum regimens. The imperative of wise antimicrobial stewardship applies, tempered by the danger that failure to provide therapy active against the subsequently identified pathogen is associated with increased mortality, at least for patients with compromised host defense or septic shock. Studies of septic shock associate delay in the administration of the first dose of effective antimicrobial therapy with death, with each hour of delay causing an increment in mortality of almost 8% per hour over the first 6 hours.^51,52 If necessary, delay should be overcome by the physician administering the initial doses. Current local and international surveillance reports of resistance in gram-negative bacilli should inform the choice of empiric therapy. A history of recent antimicrobial consumption or of international travel involving hospitalization should also be taken into account. Two agents from different antimicrobial classes active against Enterobacteriaceae in blood and urine are required, with subsequent tailoring of the regimen to a single agent with the narrowest spectrum, least toxicity and cost, based on the results of cultures.

Historically, a combination of an aminoglycoside and ampicillin provided cost-effective, empiric therapy for pyelonephritis in the community. However, because 30% to 80% of E coli urinary isolates are now resistant, ampicillin can no longer be recommended. The addition of a β-lactamase inhibitor, such as clavulanic acid or sulbactam, eliminates most resistance, but not that due to ESBL or CRE. Ampicillin-sulbactam or clavulanic acid (available intravenously in many countries but not in the United States) combined with an aminoglycoside is an appropriate choice in regions in which ESBL prevalence remains low (Table 75-2).