ICU - Fever Critically Ill - Guidline 2008

Special Article

Guidelines for evaluation of new fever in critically ill adultpatients: 2008 update from the American College of Critical CareMedicine and the Infectious Diseases Society of America

Naomi P. O’Grady, MD; Philip S. Barie, MD, MBA, FCCM; John G. Bartlett, MD; Thomas Bleck, MD, FCCM;Karen Carroll, RN; Andre C. Kalil, MD; Peter Linden, MD; Dennis G. Maki, MD; David Nierman, MD, FCCM;William Pasculle, MD; Henry Masur, MD, FCCM

I n some intensive care units(ICUs), the measurement of anewly elevated temperature trig-gers an automatic order set that

includes many tests that are time con-suming, costly, and disruptive to the pa-

tient and staff. Moreover, the patient mayexperience discomfort, be exposed to un-needed radiation, require transport out-side the controlled environment of theICU, or experience considerable bloodloss due to this testing, which is often

repeated several times within 24 hrs anddaily thereafter. In an era when utiliza-tion of hospital and patient resources isunder intensive scrutiny, it is appropriateto assess how such fevers should be eval-uated in a prudent and cost-effectivemanner.

The American College of Critical CareMedicine of the Society of Critical CareMedicine and the Infectious Diseases So-ciety of America reconvened a Task Forceto update practice parameters for theevaluation of a new fever in adult patients(i.e., �18 yrs of age) in an ICU (1). Thegoal of this update is to continue to pro-mote the rational consumption of re-sources and an efficient evaluation. Thisguideline presumes that any unexplainedtemperature elevation merits a clinicalassessment by a healthcare professionalthat includes a review of the patient’shistory and a focused physical examina-

Objective: To update the practice parameters for the evaluationof adult patients who develop a new fever in the intensive careunit, for the purpose of guiding clinical practice.

Participants: A task force of 11 experts in the disciplinesrelated to critical care medicine and infectious diseases wasconvened from the membership of the Society of Critical CareMedicine and the Infectious Diseases Society of America. Spe-cialties represented included critical care medicine, surgery, in-ternal medicine, infectious diseases, neurology, and laboratorymedicine/microbiology.

Evidence: The task force members provided personal experi-ence and determined the published literature (MEDLINE articles,textbooks, etc.) from which consensus was obtained. Publishedliterature was reviewed and classified into one of four categories,according to study design and scientific value.

Consensus Process: The task force met twice in person, sev-eral times by teleconference, and held multiple e-mail discussionsduring a 2-yr period to identify the pertinent literature and arriveat consensus recommendations. Consideration was given to the

relationship between the weight of scientific evidence and thestrength of the recommendation. Draft documents were com-posed and debated by the task force until consensus was reachedby nominal group process.

Conclusions: The panel concluded that, because fever canhave many infectious and noninfectious etiologies, a new fever ina patient in the intensive care unit should trigger a careful clinicalassessment rather than automatic orders for laboratory and ra-diologic tests. A cost-conscious approach to obtaining culturesand imaging studies should be undertaken if indicated after aclinical evaluation. The goal of such an approach is to determine,in a directed manner, whether infection is present so that addi-tional testing can be avoided and therapeutic decisions can bemade. (Crit Care Med 2008; 36:1330–1349)

KEY WORDS: fever; intensive care unit; critical illness; bloodcultures; catheter infection; pneumonia; colitis; sinusitis; surgicalsite infection; nosocomial infection; temperature measurement;urinary tract infection

From the National Institutes of Health, Bethesda,MD (NPO, HM); Weill Cornell Medical College, NewYork, NY (PSB); Johns Hopkins University Schoolof Medicine, Baltimore, MD (JGB, KC); North-western University, Chicago, IL (TB); University ofNebraska, Omaha, NE (ACK); University of PittsburghMedical Center, Pittsburgh, PA (PL, WP); Universityof Wisconsin Medical School, Madison, WI (DGM);and The Mount Sinai Hospital, New York, NY (DN).

The American College of Critical Care Medicine(ACCM), which honors individuals for their achieve-ments and contributions to multidisciplinary criticalcare medicine, is the consultative body of the So-ciety of Critical Care Medicine (SCCM), which pos-sesses recognized expertise in the practice of crit-ical care. The College has developed administrativeguidelines and clinical practice parameters for

the critical care practitioner. New guidelines andpractice parameters are continually developed, andcurrent ones are systematically reviewed and re-vised.

This guideline was developed in collaboration withthe Infectious Diseases Society of America.

For information regarding this article, E-mail:[email protected]

Dr. Bartlett holds consultancies with HIV-Bristol-Myers, Abbott, Merck, Johnson & Johnson, and Ti-botec; and a patent with Gilead.

The remaining authors have not disclosed anypotential conflicts of interest.

Copyright © 2008 by the Society of Critical CareMedicine

DOI: 10.1097/CCM.0b013e318169eda9

1330 Crit Care Med 2008 Vol. 36, No. 4

tion before any laboratory tests or imag-ing procedures are ordered.

This update specifically addresses howto evaluate a new fever in an adult patientalready in the ICU who has previouslybeen afebrile and in whom the source offever is not initially obvious. This updatewill assist intensivists and consultants asa starting point for developing an effec-tive and cost-conscious approach appro-priate for their patient populations. Thespecific recommendations are rated bythe strength of evidence, using the pub-lished criteria of the Society of CriticalCare Medicine (Table 1).

Initiating a Fever Evaluation:Measuring Body Temperatureand Defining Fever as Thresholdsfor Diagnostic Effort

Definition of Fever. The definition offever is arbitrary and depends on the pur-pose for which it is defined. Some litera-ture defines fever as a core temperatureof �38.0°C (100.4°F) (2– 4), whereasother sources define fever as two consec-utive elevations of �38.3°C (101.0°F). Inpatients who are neutropenic, fever hasbeen defined as a single oral temperatureof �38.3°C (101.0°F) in the absence of anobvious environmental cause, or a tem-perature elevation of �38.0°C (100.4°F)for �1 hr (4). A variety of definitions offever are acceptable, depending on howsensitive an indicator of thermal abnor-mality an ICU practitioner wants to uti-lize. Normal body temperature is gener-

ally considered to be 37.0°C (98.6°F) (4,5). In healthy individuals, this tempera-ture varies by 0.5 to 1.0°C, according tocircadian rhythm and menstrual cycle(6). With heavy exercise, temperature canrise by 2 to 3°C (7). Whereas many bio-logical processes can alter body temper-ature, a variety of environmental forcesin an ICU can also alter temperature,such as specialized mattresses, hot lights,air conditioning, cardiopulmonary by-pass, peritoneal lavage, dialysis, and con-tinuous hemofiltration (8–10). Thermo-regulatory mechanisms can also bedisrupted by drugs or by damage to thecentral or the autonomic nervous sys-tems. Thus, it is often difficult to deter-mine whether an abnormal temperatureis a reflection of a physiologic process, adrug, or an environmental influence.

A substantial proportion of infectedpatients are not febrile: such patientsmay be euthermic or hypothermic. Thesepatients include the elderly, patients withopen abdominal wounds, patients withlarge burns, patients receiving extracor-poreal membrane oxygenation or contin-uous renal replacement therapy (11),patients with congestive heart failure,end-stage liver disease, or chronic renalfailure, and patients taking anti-inflam-matory or antipyretic drugs. A patientwho is hypothermic or euthermic mayhave a life-threatening infection. Othersymptoms and signs in the absence offever, such as otherwise unexplained hy-potension, tachycardia, tachypnea, confu-sion, rigors, skin lesions, respiratorymanifestations, oliguria, lactic acidosis,leukocytosis, leukopenia, immature neu-trophils (i.e., bands) of �10%, or throm-bocytopenia, might appropriately man-date a comprehensive search for infectionand aggressive, immediate empiricaltherapy.

As a broad generalization, it is reason-able in many ICUs to consider everyonewith a temperature of �38.3°C (�101°F)to be febrile and to warrant special atten-tion to determine whether infection ispresent. However, a lower threshold maybe used for immunocompromised pa-tients because they are not able to man-ifest a similar fever response as the oneseen in immunocompetent patients. Ef-fective management of patients in an ICUmandates that infection be considered inpatients regardless of temperature butthat laboratory tests to search for infec-tion should be initiated in febrile patientsonly after a clinical assessment indicates

a reasonable possibility that infectionmight be present.

Site and Technology of TemperatureMeasurement. The ideal system for mea-suring temperature should provide reli-able, reproducible values safely and con-veniently. Any device must be calibratedproperly and checked periodically accord-ing to the manufacturer’s specifications.

Most authorities consider the ther-mistor of a pulmonary artery catheter tobe the standard for measuring core tem-perature against which other devicesmust be compared (6, 12–16). Not allpatients have such a thermistor in place.Even when available, these thermistorsare not all equal in technical perfor-mance. Thermistors in indwelling blad-der catheters provide essentially identicalreadings to thermistors in intravascularsites, are less invasive, provide continu-ous readings, and provide stable mea-surements, regardless of urine flow rate(12, 13, 16–18). However, bladder ther-mistor catheters are costly and require amonitor. Esophageal probes placed in thedistal third of the esophagus providereadings comparable with thermistors inintravascular sites and with bladder cath-eters (19). However, confirming accurateplacement is difficult because they arenot radiopaque. In addition, they are un-comfortable in alert or spontaneouslybreathing patients. The theoretical risk ofan esophageal probe eroding or perforat-ing the esophagus when left in place forextended periods of time makes thisprobe impractical for use in the criticallyill patient.

Rectal temperatures obtained with amercury thermometer or an electronicprobe (intermittent or continuous) aretraditional measurement devices. Read-ings from the rectum are often a fewtenths of a degree higher than core tem-perature (12, 14, 20, 21). The patient of-ten perceives rectal temperature mea-surement as unpleasant and intrusive.Access to the rectum may be limited bypatient position. Moreover, there is asmall risk of trauma or perforation to therectum, which is a particular problem inpatients who are neutropenic, coagulo-pathic, or who have had recent rectalsurgery. Rectal temperature measure-ments have also been implicated inspreading enteric pathogens such asClostridium difficile or vancomycin-resistant enterococci via the device or theoperator (22–24).

Oral temperature measurement issafe, convenient, and familiar for alert

Table 1. Society of Critical Care Medicine’s rat-ing system for references and recommendations

Referencesa) Randomized, prospective, controlled

investigationb) Nonrandomized, concurrent, or historical

cohort investigationc) Peer-reviewed, state-of-the-art articles,

review articles, editorials, or substantial caseseries

d) Non–peer-reviewed published opinions, such astextbook statements or official organizationalpublications

RecommendationsLevel 1: Convincingly justifiable on scientific

evidence aloneLevel 2: Reasonably justifiable by available

scientific evidence and stronglysupported by expert critical careopinion

Level 3: Adequate scientific evidence is lackingbut widely supported by available dataand expert critical care opinion

1331Crit Care Med 2008 Vol. 36, No. 4

and cooperative patients. Mouth breath-ing, heated gases, and hot or cold fluidscan distort the reading (13, 25). Oralprobes can damage oral mucosa, espe-cially in patients with abnormal mucosadue to trauma, thermal injury, infection,surgery, cancer, or cytotoxic drugs. Incritically ill patients, oral temperaturesare often not practical due to intubationor inability of the patient to cooperate.

Tympanic membrane temperature isbelieved to reflect the temperature of thehypothalamus and, thus, the core bodytemperature. Direct measurement of thetympanic membrane temperature re-quires an electronic probe, is painful inawake patients, and risks trauma to thetympanic membrane. Infrared ear ther-mometry is also available to detect radi-ant energy from the tympanic membraneand ear canal through an otoscopicprobe. These devices are not accurate ifinflammation of the auditory canal ortympanic membrane is present or if thereis obstruction of the external canal. Tym-panic membrane and infrared devices donot always agree with other measure-ment devices. Multiple studies haveshown consistently poor agreement be-tween measurements made by infraredear devices and those made by pulmonaryartery catheters (12–16, 18, 26).

Infrared thermometry measurementtechnology used in tympanic membranethermometers has been adapted to a non-invasive temporal artery thermometer(27). Because the temporal artery has ahigh arterial perfusion rate that remainsunchanged under most conditions, mea-surement of temperature via skin areasperfused by the temporal artery providean easy, noninvasive estimate of the coretemperature (28). Environmental tem-perature and sweating have been associ-ated with unreliable temperature mea-surements compared with rectal andesophageal temperature (29), and thetemporal artery thermometer providessimilar accuracy to axillary measure-ments that are not recommended (27).

The chemical dot thermometer is asingle-use, flexible polystyrene plasticstrip with 50 heat-sensitive dots (temper-ature sensors) applied to the forehead;each dot represents a temperature incre-ment of 0.1°C over a range of 35.5–40.4°C; the last dot to turn blue consti-tutes the body temperature (30). Lack ofagreement between measurements madeusing this device and pulmonary arterycatheters limits their usefulness in criti-cally ill patients (26).

Recommendations for MeasuringTemperature

1. Choose the most accurate and reliablemethod to measure temperature basedon the clinical circumstances of thepatient. Temperature is most accu-rately measured by an intravascular,esophageal, or bladder thermistor, fol-lowed by rectal, oral, and tympanicmembrane measurements, in that or-der (Table 2). Axillary measurements,temporal artery estimates, and chem-ical dot thermometers should not beused in the ICU (level 2). Rectal ther-mometers should be avoided in neu-tropenic patients (level 2).

2. Any device used to measure tempera-ture must be maintained and cali-brated appropriately, using the manu-facturer’s guidelines as a reference(level 2).

3. Any device used to measure tempera-ture must be used in a manner thatdoes not facilitate spread of pathogensby the instrument or the operator(level 2).

4. The site of temperature measurementshould be recorded with the tempera-ture in the chart (level 1).

5. A new onset of temperature of�38.3°C is a reasonable trigger for aclinical assessment but not necessarilya laboratory or radiologic evaluationfor infection (level 3).

6. A new onset of temperature of�36.0°C in the absence of a knowncause of hypothermia (e.g., hypothy-roidism, cooling blanket, etc.) is a rea-sonable trigger for a clinical assess-ment but not necessarily a laboratoryor radiologic evaluation for infection(level 3).

7. Critical care units could reduce thecost of fever evaluations by eliminat-ing automatic laboratory and radio-logic tests for patients with new tem-

perature elevation (level 2). Instead,these tests should be ordered based onclinical assessment. A clinical and labo-ratory evaluation for infection, con-versely, may be appropriate in euther-mic or hypothermic patients, dependingon clinical presentation.

Blood Cultures

Because the information provided by apositive blood culture can have such im-portant prognostic and therapeutic impli-cations, blood cultures should be ob-tained in patients with a new fever whenclinical evaluation does not strongly sug-gest a noninfectious cause.

Skin Preparation. The site of veni-puncture should be cleaned with either2% chlorhexidine gluconate in 70% iso-propyl alcohol (2% alcoholic chlorhexi-dine), or 1–2% tincture of iodine (iodinein alcohol). Povidone iodine (10%), al-though acceptable, is a less efficientagent. The access to an intravascular de-vice and to the stopper on all blood cul-ture bottles should be cleaned with 70–90% alcohol (31, 32). Most blood culturebottles should not be swabbed with io-dine-containing antiseptics because thesesolutions may degrade the stoppers.Chlorhexidine and tincture of iodine areequally effective for cleaning the site ofvenipuncture and more effective thanaqueous povidone iodine in reducing therate of blood culture contamination (33–36). Iodophors (aqueous iodine solutions)must be allowed to dry 1.5–2 mins toprovide maximal antiseptic activity andthus to minimize the risk of contamina-tion (37). Both alcoholic chlorhexidineand tincture of iodine have an alcoholbase, so they dry more rapidly (�30 secs)than iodophors. This reduces the timerequired for phlebotomy.

When blood is to be inoculated into aculture or transport tube, the needle usedfor venipuncture should not be replacedby a sterile needle. The risk of a needlestick injury during the switch in needlesis currently thought to outweigh the riskof contamination (37, 38).

Blood Volume and Collection System.One blood culture is defined as a sampleof 20–30 mL of blood drawn at a singletime from a single site, regardless of howmany bottles or tubes the laboratory mayuse to process the specimen. The sensi-tivity of blood cultures for detection oftrue bacteremia or fungemia is related tomany factors, the most important beingobtaining the cultures before the initia-

Table 2. Accuracy of methods used for measur-ing temperature (6, 12–16)

Most accuratePulmonary artery thermistorUrinary bladder catheter thermistorEsophageal probeRectal probe

Other acceptable methods in order of accuracyOral probeInfrared ear thermometry

Other methods less desirableTemporal artery thermometerAxillary thermometerChemical dot


tion of anti-infective therapy and the vol-ume of blood drawn (32, 39–44). The fullamount of blood recommended for eachbottle should be drawn (37). The volumesrequired for blood cultures are typicallydetermined by laboratories based on thevarious media used and on recommenda-tions of the manufacturer of the bloodculture instrument employed (37). A va-riety of commercial blood culture sys-tems can provide excellent results (Table3). Laboratories need to make their owndecisions about the best system, consid-ering their budget, their manpower, andthe patient population they serve. Specialmedia or blood culture systems can beadded in designated circumstances. Theutility of antibiotic removal devices inblood culture systems is controversial,although some laboratories still use themroutinely. Some studies have shown that,when compared with conventional cul-ture systems, these devices can enhancethe recovery of staphylococci and yeasts,regardless of whether the patient is re-ceiving antibiotics (37, 45, 46); they donot seem to enhance the recovery ofGram-negative bacilli. Considering vary-ing experiences related to benefit, and thesubstantial additive costs and increasednumber of contaminants (37, 47), theiruse is considered optional.

In general, commercial pediatric col-lection systems should not be used foradults: the smaller volume of blood sam-pled will diminish the yield of pathogensfor adults (42). If it is not possible tosecure the recommended volume ofblood required for the adult collectionsystem and �5 mL of blood is obtained, itis acceptable to inject the entire amountinto the aerobic blood culture bottle.

Cultures of Blood for Unusual Patho-gens. In special patient populations or inspecial geographic areas, it may be appro-priate in the evaluation of fever to includespecial media or special blood culturesystems for organisms other than com-

mon aerobic and anaerobic bacteria. Forexample, cultures containing resins orlytic agents may be helpful in isolatingyeast, and lysis-centrifugation may beuseful for isolating Bartonella species, di-morphic fungi, Mycobacterium avium,and Mycobacterium tuberculosis. Mostoften, these pathogens are sought in pa-tients with specific underlying diseases(solid organ transplant and stem celltransplant recipients or patients withprolonged granulocytopenia), or they aresought because of epidemiologic circum-stances (Francisella, Bartonella, or His-toplasma). In such situations, communi-cation with the microbiology laboratoryis essential to determine whether specialculture systems, in addition to routineblood cultures, are needed or if incubat-ing the routine culture for a longer pe-riod of time would be useful.

Number of Cultures and Sites. Recentdata suggest that the cumulative yield ofpathogens is optimized when three tofour blood cultures with adequate volume(20–30 mL each) are drawn within thefirst 24 hrs of suspected bacteremia orfungemia (39, 48). Each culture shouldbe drawn by separate venipuncture orthrough a separate intravascular devicebut not through multiple ports of thesame intravascular catheter (49, 50). Ide-ally, blood cultures should not be drawnthrough nonintact or infected skin (e.g.,burns). There is no evidence that theyield of cultures drawn from an artery isdifferent from the yield of cultures drawnfrom a vein. If the patient has limitedaccess for venipuncture, it is acceptableto draw blood cultures from separate in-travascular devices. However, this mayincrease the number of contaminatedblood cultures.

In most cases of true bacteremia/fungemia, when organisms are detected,all specimens, whether drawn through acatheter or through a venipuncture, willyield positive results (51). In the majority

of cases of discordant results, the culturedrawn through the device will be positiveand the culture drawn by venipuncture willbe negative; in such instances, the positiveculture may represent a contaminant or acatheter-related infection, but clinical judg-ment rather than any rigid criteria isneeded to interpret the significance of dis-cordant results (50, 52, 53).

Drawing three to four blood cultureswith appropriate volume from separatesites of access within the first 24 hrs of theonset of fever is the most effective way todiscern whether an organism found inblood culture represents a true pathogen(multiple cultures are often positive), acontaminant (only one of multiple bloodcultures is positive for an organism com-monly found on skin and clinical correla-tion does not support infection), or a bac-teremia/fungemia from an infectedcatheter (one culture from the source cath-eter is positive, often with a positive cathe-ter tip, and other cultures are not) (32, 37,54). Clinical judgment must determinewhich catheter(s) and which lumen(s) areused for drawing the culture(s). Clinicaldata that might determine drawing bloodcultures from certain catheters includecatheter dwell time (carefully inserted cath-eters that have been in place for �3 daysare less likely to be infected than longer-dwelling catheters), conditions of insertion(emergency vs. routine), and local signs ofinflammation. Many experts would use thedistal port of a catheter to obtain bloodcultures. If all blood cultures are obtainedfrom catheters, the blood culture replacingthe venipuncture should be obtained fromthe most recently inserted catheter. Bloodcultures should not be obtained via a pe-ripherally inserted venous catheter at thetime of insertion as this leads to an unac-ceptably high rate of contamination (55).Finally, separating blood cultures by de-fined, spaced intervals (such as every 10mins) has not been shown to enhance mi-crobial recovery, is not practical, and may

Table 3. Blood culture systems

Method Aerobes Anaerobes Yeast Fungi Mycobacteria Comments

Conventional broth-in-bottle �� /�� Slower than automated systemsBroth-in-bottle with

continuous monitoring�� /�� /�� a, ��b Speeds time to detection compared with

intermittent monitoringLysis-centrifugation

(Isolatorc)��/�� Volume cultured is 10 mL

Antibiotic removal systems(resin bottles)

�� /�� Greatest yield on staphylococci and yeastcompared with conventional systems

�, not recommended; ��, acceptable; ��, best available method.aUsing standard blood culture bottles; busing special mycobacterial bottles; cIsolator (Inverness Medical Innovations, Waltham, MA).


lead to a delay in therapy in critically illpatients (56).

Labeling. Blood cultures should beclearly labeled with the exact time, date,and anatomic site or catheter lumen fromwhich blood is drawn and also includeother information (concomitant antimicro-bial therapy) that may be appropriate. Thislabeling is extremely useful for interpretingthe significance of the result (see section oncatheter-related infections below).

Recommendations for ObtainingBlood Cultures

1. Obtain three to four blood cultureswithin the first 24 hrs of the onset offever. Every effort must be made todraw the first cultures before the ini-tiation of antimicrobial therapy. Theycan be drawn consecutively or simul-taneously, unless there is suspicionof an endovascular infection, inwhich case separate venipuncturesby timed intervals can be drawn todemonstrate continuous bacteremia(level 2).

2. Additional blood cultures should bedrawn thereafter only when there isclinical suspicion of continuing orrecurrent bacteremia or fungemia orfor test of cure, 48–96 hrs after ini-tiation of appropriate therapy forbacteremia/fungemia. Additional cul-tures should not be drawn as a singlespecimen but should always bepaired (level 2).

3. For patients without an indwellingvascular catheter, obtain at least twoblood cultures using strict aseptictechnique from peripheral sites by sep-arate venipunctures after appropriatedisinfection of the skin (level 2).

4. For cutaneous disinfection, 2% chlo-rhexidine gluconate in 70% isopropylalcohol is the preferred skin antisep-tic, but tincture of iodine is equallyeffective. Both require �30 secs ofdrying time before proceeding withthe culture procedure. Povidone io-dine is an acceptable alternative, butit must be allowed to dry for �2 mins(level 1).

5. The injection port of the blood cul-ture bottles should be wiped with70–90% alcohol before injecting theblood sample into the bottle to re-duce the risk of introduced contam-ination (level 3).

6. If the patient has an intravascularcatheter, one blood culture should bedrawn by venipuncture and at leastone culture should be drawn through

an intravascular catheter. Obtainingblood cultures exclusively throughintravascular catheters yields slightlyless precise information than infor-mation obtained when at least oneculture is drawn by venipuncture(level 2).

7. Label the blood culture with the ex-act time, date, and anatomic sitefrom which it was taken (level 2).

8. Draw 20–30 mL of blood per culture(level 2).

9. Paired blood cultures provide moreuseful information than single bloodcultures. Single blood cultures arenot recommended, except in neo-nates (level 2).

10. Once blood cultures have been ob-tained after the onset of new fever,additional blood cultures should beordered based on clinical suspicion ofcontinuous or recurrent bacteremiaor fungemia (level 2).

Intravascular Devices and Fever

Stable vascular access is essential tothe management of the critically ill pa-tient. Most patients will have at least onecentral venous catheter, and many mayhave an arterial catheter as well. An in-creasing number of patients will havesome type of tunneled, cuffed central ve-nous catheter or some type of subcutane-ous central venous port.

The relative risk of bloodstream infec-tion caused by various intravascular de-vices ranges widely, depending on thelength of the device, the type of device,the patient population, the techniquesused in insertion, the frequency of ma-nipulation, and the time they have beenin place (57, 58). The highest risk is withshort-term, noncuffed central venouscatheters, in the range of 2–5 per 1,000catheter days, and is especially high withnoncuffed temporary hemodialysis cath-eters. Arterial catheters used for hemody-namic monitoring and peripherally in-serted central venous catheters used inhospitalized patients seem to have a riskof catheter-related bloodstream infection(2–3 per 1,000 catheter days) similar toconventional subclavian, internal jugu-lar, or femoral short-term, noncuffedcentral venous catheters. In contrast, therisk of bloodstream infection with small,peripheral intravenous catheters is �0.1cases per 1,000 catheter days. With goodcare, surgically implanted ports are asso-ciated with a much lower risk of bactere-mia/fungemia than temporary percutane-

ous catheters, approximating 0.2–1 per1,000 catheter days (57).

Location of Infection. All intravascu-lar devices need to be assessed daily, todetermine whether they are still neededand whether there are signs of local in-fection at the site of insertion (manifestedby inflammation or purulence at the exitsite or along the tunnel), or systemically(manifested by positive blood cultures orthrombosis).

It is important to recognize that con-taminated hubs are common portals ofentry for organisms colonizing the en-doluminal surface of the catheter (59,60). In addition, infusate (parenteralfluid, blood products, or intravenousmedications) administered through anintravascular device can become contam-inated and produce device-related bacte-remia or fungemia, which is more likelyto culminate in septic shock than areother catheter-related infections.

Diagnosis. Patients with abrupt onsetof signs and symptoms of sepsis withoutrisk factors for nosocomial infection (e.g.,young, immunocompetent), especiallywithout any local site of infection, shouldprompt suspicion of infection of an intra-vascular device. Difficulty drawing or in-fusing through the catheter may point tothe catheter as a source of infection. Thepresence of inflammation, with or with-out purulence, at the insertion site,though absent in most cases (61), whencombined with signs and symptoms ofsepsis, has been shown to be predictive ofdevice-related bacteremia. Finally, recov-ery of certain microorganisms in multi-ple blood cultures, such as staphylococci,Corynebacterium jeikeium, Bacillus spe-cies, atypical mycobacteria, Candida, orMalassezia species, strongly suggest in-fection of an intravascular device.

Evaluation. As part of the comprehen-sive physical examination, the access siteshould be examined at least daily, and anyexpressible purulence or exudate shouldbe Gram stained and cultured. A mini-mum of two peripheral blood cultures, orone culture drawn percutaneously andone drawn through the catheter, shouldbe obtained. Standard blood culturesdrawn through intravascular devices pro-vide excellent sensitivity for diagnosis ofbloodstream infection (62, 63).

Removal and culture of the catheterhas historically been the gold standardfor the diagnosis of catheter-relatedbloodstream infection, particularly withshort-term catheters. Studies have dem-onstrated the reliability of semiquantita-


tive or quantitative catheter tip culturemethods for the diagnosis of catheter-related bacteremia (64, 65) (Table 4). Thediagnosis of catheter-related bloodstreaminfection is completed when a colonizedcatheter is associated with concomitantbloodstream infection with the identicalorganism, with no other plausible source.Gram-negative stain (66) or acridine or-ange stain (67) of intravascular segmentsof removed catheters also correlates withsemiquantitative or quantitative culturesbut are technically challenging, timeconsuming, and expensive to performand thus are offered by few hospitals.

Some ICU clinicians routinely culturecentral venous catheters on removal, re-gardless of whether there is fever or anyother reason to suspect catheter sepsis.

Because up to 20% of removed centralvenous catheters are colonized at re-moval, most unassociated with local in-fection or bacteremia/fungemia, thispractice adds to microbiology laboratoryexpense and can lead to unnecessarytherapies if interpreted inappropriately.The predictive value of a positive catheterculture is very low when there is a lowpretest probability of catheter sepsis (65),and catheters removed from ICU patientsshould only be cultured if there is strongclinical suspicion of catheter sepsis (68).

Cultures of catheters obviously requiretheir removal, which presents a disadvan-tage in patients with long-term central ve-nous devices. Prospective studies haveshown that only 25–45% of episodes ofsepsis in patients with long-term devices

represent true device-related bloodstreaminfection (69). Thus, it would seem thatdevelopment of methods for detecting de-vice-related bloodstream infection that donot require removal of the device would beof great value. Although exit site swabs arenot sensitive for detecting catheter-relatedinfection, a negative catheter exit site swabculture has a high negative predictive valueand may reduce the proportion of unnec-essary catheter withdrawals (70, 71).

Blood cultures drawn through thecatheter in combination with blood cul-tures drawn percutaneously may be espe-cially useful for detecting catheter-related bloodstream infection. Microbialgrowth can be quantitated using a quan-titative blood culture system, or by usingthe differential time to positivity for pe-

Table 4. Common pathogens and diagnostic tests available

PathogenStaining and RapidDetection Methods Culture Methods Other Tests

Legionella Urinary antigen (serogroup 1 only) Selective media and nonselectiveBCYE agar

Nocardia Gram stainModified acid-fast

BCYE agar Nucleic acid amplification tests forspecies identification

Mycobacteriumtuberculosis

Acid-fast stainFluorochrome stainNucleic acid

amplification tests

Culture in liquid and solid media Nucleic acid amplification tests

Mycobacterium aviumcomplex

Acid-fast stain As above Nucleic acid tests (direct) forculture identification

Mycobacteriumspecies

Acid-fast stain As above Nucleic acid tests (direct) forculture identification

Rhodococcus equi Gram stain Routine media Nucleic acid tests (direct) forculture identification

Pneumocystis jiroveci Fluorescent-labeled antibodyGrocott stainGiemsa stainGomori stainToluidine blue

Aspergillus speciesOther myceliaZygomycosis

KOH wet mountCalcofluor whiteSilver stains

Fungal-selective media Serum ELISA for detection ofgalactomannan or 1-3�—glucan

Herpes simplex Direct fluorescent antibodyWright or Giemsa stain for

intranuclear inclusions ormultinucleated giant cells

Viral culture BAL cytology for inclusion bodiesNucleic acid amplification tests

Cytomegalovirus Shell vialAntigen detectionNucleic acid amplification tests

Viral culture (very slow growth) BAL cytology for inclusionsBlood assay for antigenemiaCMV viral load

Human herpesvirus 6, 7 Nucleic acid amplification tests Viral cultureAdenovirus Rapid antigen detection Viral culture Nucleic acid amplification testsInfluenza A/B Direct fluorescent antibody

Enzyme immunoassayRapid antigen detection kitRT-PCR

Viral culture

Respiratory syncytialvirus

Enzyme immunoassayNucleic acid amplification tests

Viral culture

Strongyloides Wet mount Serum ELISAToxoplasma gondii Giemsa stain

BAL, bronchoalveolar lavage; BYCE, buffered charcoal yeast extract; CMV, cytomegalovirus; ELISA, enzyme-linked immunosorbent assay; RT-PCR,reverse transcription-polymerase chain reaction.


ripheral vs. catheter blood culturesdrawn simultaneously. If both sets of cul-tures are positive for the same organismand the set drawn through the catheterbecomes positive �120 mins earlier thanthe culture drawn peripherally, this ishighly suggestive of catheter-relatedbloodstream infection. This method isless sensitive but as specific as quantita-tive culture.

A quantitative blood culture drawnthrough an infected catheter characteris-tically shows a marked step-up in concen-tration of organisms, usually ten-fold orgreater, as compared with a quantitativeblood culture drawn concomitantly froma peripheral vein. Due to the additionalexpense and expertise necessary for pro-cessing, quantitative cultures of catheter-drawn blood are not routinely performedas part of the usual evaluation of fever.They can, however, provide useful infor-mation, especially for surgically im-planted catheters that cannot easily beremoved.

For patients with fever alone who arenot septic (i.e., temperature of �38°C or�36°C, white blood cell count of�12,000 cells/mm3 or �4,000 cells/mm3,heart rate of �90 beats/min, respiratoryrate of �20 breaths/min, or PaCO2 of �32[3]), there is usually no need to removeor change all indwelling catheters imme-diately, although such an approach wouldbe the most cautious management strat-egy and might be desirable in a patientwith a prosthetic heart valve or a fresharterial graft (68). If patients have sepsisor septic shock refractory to vasopressors,peripheral embolization, disseminated in-travascular coagulation, or acute respira-tory distress syndrome, removal of all in-travascular catheters and reinsertion atnew sites is indicated, even if the cathetersare cuffed or tunneled devices (72). If thereis clinical evidence of vascular compromise(i.e., signs of occlusive venous thrombosisor arterial insufficiency), the cathetershould be removed. If a radiographic studyis performed and there is evidence ofthrombosis, clinical judgment is needed todetermine whether anticoagulation,thrombolysis, or surgical interventions arenecessary.

Septic phlebitis of a central vein dueto a centrally placed catheter is unusual(73, 74). With suppurative phlebitis,bloodstream infection characteristicallypersists after the catheter has been re-moved, producing a clinical picture ofoverwhelming sepsis with high-gradebacteremia or fungemia or with septic

embolization. This syndrome is most of-ten encountered in burn patients or otherICU patients who develop catheter-related infection that goes unrecognized,permitting microorganisms to proliferateto high levels within an intravascularthrombus (73–75). In patients with per-sistent bacteremia with Staphylococcusaureus or persistent fungemia, echocar-diography is appropriate to assess for en-docarditis and guide further therapy (68).

Recommendations for Management ofIntravascular Catheters

1. Examine the patient at least daily forinflammation or purulence at the exitsite or along the tunnel, and assess thepatient for signs of venous thrombosisor evidence of embolic phenomena(level 2).

2. Any expressed purulence from the in-sertion site should be Gram stainedand cultured (level 2).

3. If there is evidence of a tunnel infec-tion, embolic phenomenon, vascularcompromise, or septic shock, thecatheter should be removed and cul-tured and a new catheter inserted at adifferent site (level 2).

4. With short-term temporary cathe-ters—peripheral venous catheters,noncuffed central venous catheters, orarterial catheters—if catheter-relatedsepsis (i.e., source of the infection is acolonized catheter) is consideredlikely, the suspect catheter or cathe-ters should be removed and a cathetersegment cultured. Blood culturesshould be obtained as well. With allshort-term catheters, a 5- to 7-cm in-tracutaneous segment should be cul-tured to document the source of bac-teremia; with short peripheral venousor arterial catheters, the tip should becultured; with longer central venouscatheters, the intracutaneous segmentand tip should be cultured; and withpulmonary artery catheters, the intro-ducer and the pulmonary artery cath-eter should be cultured (level 1).

5. At least two blood cultures should beobtained. At least one blood cultureshould be obtained peripherally by ve-nipuncture. One specimen should beobtained from the suspected catheter(level 1). If a quantitative culture sys-tem is available, it should be used todiagnose the catheter as the source ofbacteremia/fungemia. Alternatively,differential time to positivity can beused if both blood cultures are positivefor the same organism. The distal port

is the logical port from which to drawcultures. When short-term, uncuffedcentral venous catheters are suspectedof infection, it is usually more efficientto remove the existing catheter andreplace it than to draw quantitativecultures (level 2).

6. Do not routinely culture all cathe-ters removed from ICU patients. Cul-ture only those catheters suspectedof being the source of infection(level 2).

7. It is not necessary to routinely cultureinfusate specimens as part of the eval-uation for catheter-related infections,unless there is clinical suspicion forinfected infusate or blood products(level 2).

Pulmonary Infections andICU-Acquired Pneumonia

Pneumonia is the second most com-mon cause of infection acquired in theICU and a ubiquitous cause of fever, withthe majority of cases occurring in me-chanically ventilated patients (76, 77). Inan ICU, it can be difficult to determinewhether fever is due to pneumonia whenpatients commonly have other noninfec-tious processes producing abnormalchest radiographs and gas exchange (i.e.,congestive heart failure, atelectasis, acuterespiratory distress syndrome). Many pa-tients in an ICU are intubated and se-dated, cannot cough, and have other rea-sons for abnormal secretions. In addition,immunocompromised patients, such assolid organ transplant recipients, mayhave severe pneumonia without fever,cough, sputum production, or leukocyto-sis (78, 79).

Diagnostic Evaluation. Physical ex-amination, chest radiograph, and exami-nation of pulmonary secretions comprisethe initial evaluation. For initial feverevaluations, portable chest radiographsare generally adequate. All radiographsshould be performed in an erect sittingposition, during deep inspiration if possi-ble. Among all radiographic signs in ICUpatients, unilateral air bronchogramshave been shown to have the best predic-tive value for pneumonia; however, nosingle radiographic finding is highly pre-dictive (80 – 82). The absence of infil-trates, masses, or effusions does not ex-clude pneumonia, abscess, or empyemaas a cause of fever. Clinical judgment isneeded to determine whether the suspi-cion of infection is high enough to war-rant transporting the patient to the radi-


ology suite for a higher resolution studysuch as computerized tomographic (CT)imaging. Such studies are particularlysensitive for demonstrating parenchymalor pleural disease in the posterior-inferior lung bases (83, 84), althoughthere is only a fair correlation with adiagnosis of pneumonia in complex pa-tients (85). CT imaging is also valuable inimmunocompromised patients, in whomsmall nodular or cavitary lesions aremore prevalent and difficult to detect bystandard chest radiographs (86, 87). Re-spiratory secretions can be obtained forexamination by a variety of techniques.Expectoration, nasopharyngeal washing,saline induction, deep tracheal suction-ing, bronchoscopic protected specimenbrush samples or aspiration, or broncho-scopic or nonbronchoscopic (mini-BAL)bronchoalveolar lavage are the principlediagnostic options. Each of these tech-niques has advantages and disadvantages(77, 88). For initial evaluation of fever inthe nonintubated patient, before it is ap-parent whether pneumonia is present,evaluation of an expectorated sputum,nasotracheal or endotracheal aspirate, asopposed to a more invasively obtainedspecimen, is usually sufficient. In the in-tubated patient, saline should be instilledin the endotracheal tube only if an ade-quate specimen cannot be obtained bydeep suctioning in the absence of saline.There is concern that saline dilutes thespecimen and could introduce pathogenspresent in the tube biofilm or upper air-way into the lower airway.

The utility of fiberoptic bronchoscopyis variable and depends on the patientpopulations, causative organism, andcurrent and previous use of antibiotics.Aspirates from the inner channel of thebronchoscope are characteristically con-taminated by upper respiratory flora (89,90). If bronchoscopic sampling is not im-mediately available, nonbronchoscopicsampling (mini-BAL) can reliably obtainlower respiratory secretions for micro-scopic evaluation (88, 91). Bronchoscopy,however, may be especially useful for thedetection of selected pathogens such asPneumocystis jiroveci, Aspergillus andother filamentous fungi, Nocardia, Legio-nella, cytomegalovirus (CMV), and Myco-bacterium species (92, 93).

When pulmonary secretions are ac-quired for analysis, the specimen shouldbe transported to the laboratory and pro-cessed within 2 hrs so that fastidious or-ganisms such as Streptococcus pneu-moniae remain viable (94). For any

expectorated specimen, it is importantfor the laboratory to perform direct mi-croscopy on the specimen to determinewhether it represents saliva (i.e., if thepredominant cells are squamous epithe-lial) or lower respiratory secretions (i.e.,if the predominant cells are leukocytes,assuming the patient is nonneutropenic).

If the specimen is of lower respiratoryorigin, in most situations, a Gram-negative stain should be performed, andthe specimen should be cultured for rou-tine aerobic bacterial pathogens. In ap-propriate circumstances, principally de-fined by the host category, rapidity ofclinical presentation, and radiographicfeatures, it may be desirable to performother direct tests, such as a potassiumhydroxide with calcofluor stain for fun-gus, an enzyme-immunoassay, or directfluorescent antibody tests for respiratoryviruses and P. jiroveci, and an acid-faststain for mycobacteria. It may also bedesirable to culture the specimen forfungi, mycobacteria, Legionella, and re-spiratory viruses. Urinary antigen testsare currently available for Legionellapneumophila type 1 and S. pneumoniae.Some laboratories have developed andverified nucleic acid amplification testsfor some of these pathogens. However,there are currently no assays approved bythe Food and Drug Administration, ex-cept those available for M. tuberculosis.The use of such assays must be driven byinstitution-specific expertise and in consul-tation with clinicians. Because local andreferral center testing methods may varyacross different centers, the most commonlaboratory technique options for eachpathogen are summarized in Table 4.

Clinicians need to be aware of the or-ganisms that are virtually always pathogenswhen recovered from respiratory secre-tions. Although not all-inclusive, these or-ganisms might include Legionella, Chla-mydia, M. tuberculosis, Rhodococcusequi, influenza virus, respiratory syncy-tial virus, parainfluenza virus, Strongy-loides, Toxoplasma gondii, P. jiroveci,Histoplasma capsulatum, Coccidioidesimmitis, Blastomyces dermatitidis, andCryptococcus neoformans. Conversely,isolation of enterococci, viridans strepto-cocci, coagulase-negative staphylococci,and Candida species (93, 95, 96) shouldrarely if ever be considered the cause ofrespiratory dysfunction. Potential bacte-rial pathogens such as Pseudomonasaeruginosa, Enterobacteriaceae, S. pneu-moniae, S. aureus, and Haemophilus in-fluenzae are frequently encountered in

respiratory specimens and may representcontaminants that colonize the upper air-ways, or they may be true pathogens ofpneumonia. The distinction betweenpathogen and colonizer is facilitated bydetection of pathogenic organisms as thedominant flora on direct Gram-negativestain or their recovery in respiratory se-cretions in moderate or heavy growth.Quantitative cultures of bronchoscopic orother specimens from lower airways mayalso facilitate the distinction between col-onizing bacteria and pathogens. The useof these quantitative techniques has beenshown to increase the specificity of thediagnosis of healthcare-associated pneu-monia in some studies but requires bron-choscopic expertise, considerable laborby the microbiology department, andwell-standardized methodology (88).

Clinical interpretation of quantitativecultures is likely to be hampered by pre-vious antibiotic administration, whichmay lower the observed quantitative in-oculum after 24 hrs of ongoing antibiotictherapy and up to 72 hrs after cessation ofantibiotics (97, 98). The utility of quanti-tative cultures to identify the causativepathogen has been reviewed elsewhere(99–101).

It is appropriate to draw blood cul-tures to attempt to identify the cause ofpneumonia. Blood tests other than cul-tures may also yield the etiology of pneu-monia, especially in immunocompro-mised patients: a) antigenemia for CMVin non–human immunodeficiency virus-infected patients, histoplasmosis, andcryptococcosis; b) polymerase chain reac-tion for CMV, varicella-zoster virus, hu-man herpes virus-6, and adenovirus; andc) galactomannan and beta-D-glucan foraspergillosis and Candida may be usefulas supportive evidence of infections butmay be most useful to exclude invasivefungal infection, given their high nega-tive predictive value.

Many febrile patients in an ICU havesmall amounts of pleural fluid due tocongestive heart failure, hypoalbumine-mia, or postoperative processes. It is notnecessary to obtain a sample of such fluidfor culture in every febrile patient. Tho-racentesis to obtain fluid for stain, cul-ture, and cytology (and for measurementof pH, protein, glucose, and lactate dehy-drogenase) would be especially appropri-ate if there is enough fluid to aspiratesafely using ultrasound guidance and ifthere is either an adjacent pulmonary in-filtrate, suspicion of tuberculosis, or pos-sible contamination of the pleural space


by surgery, trauma, or a fistula (102,103). The fluid should be cultured foraerobic and anaerobic bacteria. Culturesfor fungi and mycobacteria should be per-formed as epidemiologically appropriate.

Once cultures are obtained and thelaboratory grows pathogens, antimicro-bial susceptibility tests should be per-formed on isolates of aerobic and facul-tative bacteria, including S. pneumoniae.Susceptibility tests should be performedand interpreted using the most recentcriteria published by the Clinical andLaboratory Standards Institute. Suscepti-bility tests for anaerobic bacteria, fungi,or viruses are not routinely indicated.

Recommendations for Evaluation ofPulmonary Infections. If a febrile patientis suspected of having a lower respiratorytract infection by clinical or radiographicassessment:

1. A chest imaging study should be ob-tained. In most cases, an upright porta-ble anteroposterior chest radiograph isthe most feasible study to obtain. Poste-rior-anterior chest radiographs with lat-eral view or CT scan offer more infor-mation and should be obtained whenclinically indicated, especially to ruleout opportunistic infections in immu-nocompromised patients (level 1).

2. Obtain one sample of lower respira-tory tract secretions for direct exami-nation and culture before initiation ofor change in antibiotics. Expectoratedsputum, induced sputum, tracheal se-cretions, or bronchoscopic or nonbron-choscopic alveolar lavage material canbe used effectively. If pneumonia isdocumented by physical examinationand radiographic evaluation, a deci-sion to employ bronchoscopy or otherinvasive diagnostic approaches shouldbe considered based on an individualbasis and the availability of local ex-pertise (level 2).

3. Respiratory secretions obtained formicrobiological evaluation should betransported to the laboratory and pro-cessed in �2 hrs (level 2).

4. Respiratory secretions that are judgedto be appropriate samples by the lab-oratory should be evaluated by Gram-negative stain and cultured for routineaerobic and facultative bacteria. Addi-tional stains, rapid tests, cultures, andother tests should be performed as ep-idemiologically appropriate (level 2).

5. Quantitative cultures can provide use-ful information in certain patient pop-ulations when assessed in experienced

laboratories; however, quantitativecultures have not yet been sufficientlystandardized nor have they beenshown to alter outcome for this tech-nique to be considered part of routineevaluation (level 2).

6. Pleural fluid should be obtained withultrasound guidance for Gram-nega-tive stain and routine culture (withother studies as clinically indicated) ifthere is an adjacent infiltrate or anotherreason to suspect infection and the fluidcan be safely aspirated (level 2).

Stool Evaluation in the FebrilePatient in the ICU

Many patients in the ICU have diarrhea,which is often caused by enteral feedings ordrugs. By far the most common entericcause of fever in the ICU is Clostridiumdifficile, which should be suspected inany patient with fever or leukocytosis anddiarrhea who received an antibacterialagent or chemotherapy within 60 daysbefore the onset of the diarrhea (104,105). C. difficile accounts for 10–25% ofall cases of antibiotic-associated diarrheaand virtually all of the cases of antibiotic-associated pseudomembranous colitis(106). Other organisms that can causefever and diarrhea include Salmonella,Shigella, Campylobacter jejuni, Aeromo-nas, Yersinia, Escherichia coli, Entam-oeba histolytica, and multiple virusesthat are not usually identified by standardtechniques. In general, these are commu-nity-acquired diseases and only rarelycause infectious diarrhea acquired after apatient has been admitted to the ICU. Ifthe patient was not initially admitted tothe hospital with diarrhea and is not in-fected with HIV, it is unlikely that theseorganisms would produce diarrhea andfever in the ICU. Thus, sending stools forbacterial cultures or ova and parasite ex-amination should generally be avoided aspart of a fever evaluation unless the pa-tient was admitted to the hospital withdiarrhea, is infected with HIV, or is a partof an outbreak evaluation.

Presentation. Most patients with C.difficile as the cause for their feverpresent with diarrhea. (Diarrhea is de-fined in this document as more than twostools per day that conform to the con-tainer in which they are placed.) How-ever, some patients, especially those whoare postoperative, may present with ileusor toxic megacolon or leukocytosis with-out diarrhea as the manifestation of C.difficile disease. In these patients, the di-

agnosis of C. difficile disease may be dif-ficult to establish because stool speci-mens are not accessible (107). C.difficile–associated diarrhea may occurwith any antibacterial agent, but themost common causes are clindamycin,cephalosporins, and fluoroquinolones(108). Other clinical clues are fever, toxicmegacolon, leukocytosis (especially witha leukemoid reaction), and hypoalbumin-emia (109).

Evaluation for C. difficile. Testing al-gorithms differ markedly among institu-tions. The most sensitive and specific testmost laboratories can perform to estab-lish this diagnosis is the tissue cultureassay (110, 111). Disadvantages with thistest are the 24- to 48-hr delay in results,lack of experience with tissue culturetechniques in many laboratories, andcost. As a result, most laboratories nowuse enzyme immunoassay (EIA) tests,which are commercially available, pro-vide results within minutes to hours, andare technically easy to perform. They areless sensitive than the tissue culture as-say and consequently may require two tothree repeat tests to document disease(112). The EIA for toxin A is available, but2–3% of stains produce only toxin B, sothe EIA for toxin A and B is preferred(113). Once toxin is demonstrated, thereis no utility in follow-up assays to dem-onstrate cure, as patients may shed toxinlong after they are clinically cured. A two-step algorithm that screens samples forglutamate dehydrogenase, the C. difficilecommon antigen, followed by testingonly positives for toxin has been demon-strated to have high negative predictivevalue (114, 115).

Cultures for C. difficile are technicallydemanding, require 2 to 3 days forgrowth, and are not specific in distin-guishing toxin-positive strains, toxin-negative strains, and asymptomatic car-riage (112, 116). Cultures may be usefulin the setting of nosocomial outbreakswhen isolates are needed for strain typingfor epidemiologic purposes (105). TheNAP1 strain has been epidemic in manyhospitals in the United States, Canada,and Europe; it is associated with seriouscomplications (toxic megacolon, leuke-moid reactions, sepsis, and death) and isoften refractory to standard therapy (108,117, 118). There are strain differences invirulence, which may be important torecognize for epidemiologic purposes(119). However, the methods to identifyspecific strains are not generally avail-


able, and strain identification does notalter management of individual patients.

Direct visualization of pseudomem-branes is nearly diagnostic for C. difficilecolitis (120). Most diagnoses by visualiza-tion come from proctoscopic evaluationrather than more extensive lower gastro-intestinal evaluation. In terms of sensi-tivity, in patients with severe disease,only 71% of patients had pseudomem-branes documented by direct visualiza-tion, whereas pseudomembranes werepresent in only 23% of patients with milddisease (121). Concerns about cost, riskof perforation during the examination,and the relative ease of cytotoxin assayshave removed flexible sigmoidoscopy andcolonoscopy from routine use for diagno-sis. However, a role for direct visualiza-tion may exist in cases requiring rapiddiagnosis if laboratory results will be de-layed or if false-negative C. difficile toxinassays are suspected (105). Many clini-cians would treat such patients empiri-cally rather than perform sigmoidoscopyor colonoscopy.

Evaluation for Other Enteric Patho-gens. Infection with Klebsiella oxytocashould be considered in patients with an-tibiotic-associated colitis who are nega-tive for C. difficile (122). Patients whohave a recent and significant travel his-tory to developing countries, patientswith HIV disease, and patients with un-usual domestic exposures may require amore extensive evaluation. Patients witha recent travel history should have theirevaluation directed by the most likelypathogens that occur in their area oftravel, although the most common causeof travelers’ diarrhea, enterotoxigenic E.coli, is not detected with the usual labo-ratory tests. Patients who have traveled toareas where parasitic disease is commonshould have their stool evaluated by stoolova and parasite examination for otherorganisms such as Cyclospora, E. histo-lytica, and Strongyloides. Norovirusshould be considered in the setting ofemployee illness, patients with a travelhistory, or when multiple patients on aparticular hospital unit are having fever,vomiting, and diarrhea.

Patients with HIV disease and CD4�100/�L often have chronic diarrheacaused by Salmonella, Microsporidium,CMV, or M. avium complex. Microspo-ridium detection requires special stainsof the stool or small bowel biopsy. Thediagnosis of CMV should be made endo-scopically by means of a biopsy. CMV co-litis should be highly suspected in the

solid organ transplant recipient with fe-ver and diarrhea, particularly in the re-cipient who received an allograft from adonor with positive CMV serology. Acuteneutropenic enterocolitis or typhlitiscaused by enteric Gram-negative bacilli(i.e., Pseudomonas species) or anaerobes(i.e., Clostridium septicum) should besought in cancer or stem cell transplantrecipients by imaging studies or endos-copy.

Recommendations for Evaluation ofthe Gastrointestinal Tract. If more thantwo stools per day conform to the con-tainer in which they are placed in a pa-tient at risk for C. difficile and if clinicalevaluation indicates that a laboratoryevaluation is necessary:

1. Send one stool sample for C. difficilecommon antigen, EIA for toxin A andB, or tissue culture assay (level 2).

2. If the first specimen for C. difficile isnegative and testing is performed byan EIA method, send an additionalsample for C. difficile EIA evaluation.A second specimen is not necessary ifthe common antigen test was negative(level 2).

3. If severe illness is present and rapidtests for C. difficile are negative orunavailable, consider flexible sigmoid-oscopy (level 3).

4. If severe illness is present, considerempirical therapy with vancomycinwhile awaiting diagnostic studies. Em-pirical therapy is not generally recom-mended if two stool evaluations arenegative using a reliable assay. Al-though it may be more cost-effectivethan making the diagnosis, the empir-ical use of antibiotics, especially van-comycin, is discouraged because of therisk of producing resistant pathogens(level 2).

5. Stool cultures for other enteric patho-gens are rarely indicated in a patientwho did not present to the hospitalwith diarrhea or in patients who arenot HIV infected. Send stool culturesfor other enteric pathogens and exam-ine for ova and parasites only if epide-miologically appropriate or evaluatingan immunocompromised host (level 2).

6. Test stool for norovirus if the clinicaland epidemiologic setting is appropri-ate. Testing for norovirus is usuallyonly available in state laboratories andis usually performed in outbreak set-tings. Obtain consultation with infec-tion control and public health author-ities (level 3).

Urinary Tract Infection

Catheter-associated bacteriuria orcandiduria usually represents coloniza-tion, is rarely symptomatic, and is rarelythe cause of fever or secondary blood-stream infection (123), even in immuno-compromised patients (124), unless thereis urinary tract obstruction, the patienthas had recent urologic manipulation orsurgery, or is granulocytopenic (125,126). The traditional clinical signs andsymptoms (dysuria, urgency, pelvic orflank pain, fever or chills), that correlatewell with significant bacteriuria in non-catheterized patients are rarely reportedin ICU patients with documented cathe-ter-associated bacteriuria or candiduriaof �105 colony-forming units (cfu)/mL(123, 124).

Etiology. In the ICU, the majority ofurinary tract infections are related to uri-nary catheters and are caused by multire-sistant nosocomial Gram-negative bacilliother than E. coli, Enterococcus species,and yeasts (123, 127, 128).

Diagnosis. When clinical evaluationindicates the urinary tract is a source offever, a specimen of urine should be ob-tained and evaluated by direct micros-copy, Gram-negative stain, and quantita-tive culture.

The specimen should not be collectedfrom the drainage bag because multipli-cation of bacteria to high levels can occurwhile the urine is in the bag (129).Rather, a specimen of urine should beaspirated from the catheter samplingport. Healthcare personnel should wearclean gloves whenever manipulating aurinary device and should scrupulouslyclean the port with 70–90% alcohol be-fore collecting the specimen. For patientswithout a catheter in place, a conven-tional midstream clean-catch urine spec-imen should be obtained. Urine collectedfor culture should be transported to thelaboratory and processed promptly toprevent the multiplication of insignifi-cant numbers of microorganisms to highlevels within the receptacle, which mightlead to the misdiagnosis of infection. Ifthe transport of urine will be delayedlonger than approximately 1 hr, the spec-imen should be refrigerated. For trans-port to a remote laboratory site, the useof a urine preservative device containingboric acid is recommended (130).

In contrast to community-acquiredurinary tract infections, where pyuria ishighly predictive of significant bacteri-uria, pyuria may be absent in patients


with catheter-associated urinary tract in-fection and, even if present, is not reliablypredictive of infection or associated withsymptoms referable to the urinary tract(123).

The concentration of urinary bacteriaor yeast needed to cause symptomaticurinary tract infection or fever is unclear.Whereas it is clear that counts of �103

cfu/mL represent true bacteriuria or can-diduria in catheterized patients (131),there are no data to show that highercounts are more likely to represent symp-tomatic infection than lower ones. Gram-negative stain of a centrifuged urinaryspecimen, however, will show microor-ganisms most of the time if infection ispresent (132) and can provide valuableinformation to the clinician selecting em-pirical antimicrobial therapy for sus-pected urosepsis.

Although it is appropriate to collecturinary specimens in the investigation offever, routine monitoring or “surveil-lance” cultures of urine contribute littleto patient management (133). The rapiddipstick tests, which detect leukocyte es-terase and nitrite, are unreliable tests inthe setting of catheter-related urinarytract infection. The leukocyte esterasetest correlates with significant pyuria,which may or may not be present in acatheter-related urinary tract infection.The nitrite test corresponds to Enter-obacteriaceae, which convert nitrate tonitrite. In the setting of active bacteriuriacaused by Enterococcus species, Candidaspecies, and coagulase-positive Staphylo-coccus species, it is not reliable and is notrecommended for use in patients withurinary catheters in place (134, 135).

Recommendations for Evaluation ofthe Urinary Tract

1. For patients at high risk for urinarytract infection (kidney transplant pa-tients, granulocytopenic patients, orpatients with recent urologic surgeryor obstruction), if clinical evaluationsuggests a patient may have symptom-atic urinary tract infection, a labora-tory evaluation is necessary. Obtainurine for microscopic exam, Gram-negative stain, and culture (level 2).

2. Patients who have urinary catheters inplace should have urine collected fromthe sampling port of the catheter andnot from the drainage bag (level 2).

3. Urine should be transported to thelaboratory and processed within 1 hrto avoid bacterial multiplication. Iftransport to the laboratory will be de-

layed for �1 hr, the specimen shouldbe refrigerated. Alternatively, a preser-vative could be used but is less prefer-able to refrigeration (level 2).

4. Cultures from catheterized patientsshowing �103 cfu/mL represent truebacteriuria or candiduria, but neitherhigher counts nor the presence ofpyuria alone are of much value in de-termining if the catheter-associatedbacteriuria or candiduria is the causeof a patient’s fever; in most cases, it isnot the cause of fever (level 1).

5. Gram stains of centrifuged urine willreliably show the infecting organismsand can aid in the selection of anti-infective therapy if catheter-associatedurosepsis is suspected (level 1).

6. Rapid dipstick tests are not recom-mended for patients with urinarycatheters in the analysis of possiblecatheter-associated infection (level 1).

Sinusitis

In the ICU setting, nosocomial sinus-itis is a closed-space infection that is in-frequent and may be clinically occult but,when it occurs, can have serious conse-quences (136). Whereas sinusitis is oftenpart of the differential diagnosis of fever,the prevalence is low in comparison withother nosocomial infections in the ICU,and the diagnosis can be difficult to doc-ument convincingly.

Etiology. The most common risk fac-tor for sinusitis is anatomic obstructionof the ostia draining the facial sinuses,especially the maxillary sinuses. Transna-sal intubations of the stomach and espe-cially of the trachea are the leading riskfactors, with transnasal intubation of theairway carrying a prevalence of sinusitisestimated to be 33% after 7 days of intu-bation. Maxillofacial trauma, with ob-struction of drainage by retained bloodclots, is another clear risk factor. Theparanasal sinuses are normally sterile,but bacterial overgrowth occurs whendrainage is impeded. The etiologicalagents responsible for most cases of nos-ocomial sinusitis are those that colonizethe naso-oropharynx (137–139), whichoccur at high frequency among hospital-ized patients, and especially among crit-ically ill patients. Gram-negative bacilli(particularly P. aeruginosa) constitute60% of bacteria isolated from nosocomialsinusitis, whereas Gram-positive cocci (typi-cally S. aureus and coagulase-negative staph-ylococci) comprise one third of isolates and

fungi the remaining 5–10% (138–141). Infec-tions are often polymicrobial.

Diagnosis of Sinusitis. Either two ma-jor criteria (cough, purulent nasal dis-charge) or one major and two minor crite-ria (headache or earache, facial or toothpain, fever, malodorous breath, sore throat,wheezing) suggest acute bacterial sinusitisin outpatients when these manifestationshave been present for �7 days (142). Adiagnostic challenge even among outpa-tients, the diagnosis of sinusitis in criticallyill, intubated patients is even more difficult.Complaints of facial pain or headache maybe impossible to elicit, and purulent nasaldischarge is present in only 25% of provedcases of sinusitis (139).

Acute sinusitis can be suggested byplain radiographs, ultrasound, CT, ormagnetic resonance imaging (143). Sinusopacification by plain radiography is sen-sitive but nonspecific for the diagnosis(139, 141, 144, 145). A combination ofnasal endoscopy and plain radiographycan increase accuracy but depends on theskill of the endoscopist (146). However,obtaining good-quality plain sinus radio-graphs is a practical impossibility usingportable equipment in the ICU. CT scan-ning provides better sensitivity than plainradiography alone (147, 148).

Microbial analysis of fluid obtained byminimally invasive sinus puncture andaspiration under antiseptic conditions isdefinitive for the diagnosis of infectioussinusitis. In addition, tissue biopsy mayneed to be performed to rule out invasivefungal sinusitis in the immunocompro-mised patient. Although less well studied,endoscopically guided middle meatal tis-sue culture has been shown to be a safealternative for septic patients who are notcandidates for antral puncture, such aspatients with coagulopathies (149). Sam-pling is mandatory because of the discor-dant accuracy between radiographic andmicrobiological testing, whereas patho-gen identification and susceptibility test-ing permit focused, narrow-spectrum an-timicrobial therapy. However, specimencollection is susceptible to contaminationby bacteria colonizing the overlying mu-cosa if rigorous antisepsis is not practicedwhen obtaining the specimen.

Recommendations for Evaluation ofthe Sinuses

1. If clinical evaluation suggests that si-nusitis may be a cause of fever, a CTscan of the facial sinuses should beobtained (level 2).

2. If the patient has not responded to


empirical therapy, puncture and aspi-ration of the involved sinuses underantiseptic conditions should be per-formed (level 2).

3. Aspirated fluid should be sent forGram-negative stain and culture foraerobic and anaerobic bacteria andfungi to determine the causativepathogen and its antimicrobial sus-ceptibility (level 1).

Postoperative Fever

Fever is a common phenomenon dur-ing the initial 48 hrs after surgery. It canbe useful to remember that fever in thisearly postoperative period is usually non-infectious in origin (150), presuming thatunusual breaks in sterile technique orpulmonary aspiration did not occur. Con-siderable money can be wasted in over-zealous evaluation of early postoperativefever. However, once a patient is �96 hrspostoperative, fever is likely to representinfection.

A chest radiograph is not mandatoryfor evaluation of postoperative fever un-less respiratory rate, auscultation, abnor-mal blood gas, or pulmonary secretionssuggest a high probability of utility. Atel-ectasis is often considered to be a cause ofpostoperative fever, although this shouldbe a diagnosis of exclusion. The clinicianmust be alert to the possibility that thepatient could have aspirated during theperioperative period or that the patient wasincubating a community-acquired process,for instance, influenza A or Legionellapneumonia, before the operation.

Urinary tract infection is commonpostoperatively because of the use of uri-nary drainage catheters (151). The dura-tion of catheterization is the most impor-tant risk factor for the development ofnosocomial cystitis or pyelonephritis. Aurinalysis or culture is not mandatory toevaluate fever during the initial 2 to 3days postoperatively unless there is rea-son by history or examination to suspectan infection at this site.

Fever can be related to hematoma orinfection of the surgical field. Wound in-fection is rare in the first few days afteroperation, except for group A streptococ-cal infections and clostridial infections,which can develop 1–3 days after surgery.These causes should be suspected on thebasis of inspection of the wound.

Many emergency abdominal opera-tions are performed for control of an in-fection (e.g., peritonitis due to perforateddiverticulitis). Even under optimal cir-

cumstances (definitive surgical sourcecontrol and timely administration of ap-propriate broad-spectrum antibiotics), itmay take �72 hrs for such patients todefervesce. New or persistent fever �4days after surgery should raise a strongsuspicion of persistent pathology or anew complication. Thus, it is mandatoryto remove the surgical dressing to inspectthe wound. Swabbing the wound for cul-ture is rarely helpful if clinical assess-ment reveals no symptoms or signs sug-gesting infection (152). When erysipelasor myonecrosis is present, the diagnosisis often suspected by inspection alone,and such patients are usually “toxic” ap-pearing. Muscle compression injury (ei-ther direct trauma or as a result of com-partment syndrome) and tetanus are twoother rare complications of traumaticwounds that may cause fever. Toxic shockmay accompany infection with group A�-hemolytic streptococci or S. aureus.

Other potentially serious causes ofpostoperative fever include deep venousthrombosis, suppurative phlebitis, pul-monary embolism (153–156), drug-induced fever, anesthesia-induced malig-nant hyperthermia, acute allograftrejection, and catheter-related infection.

Recommendations for Evaluation ofFever Within 72 Hours of Surgery

1. A chest radiograph is not mandatoryduring the initial 72 hrs postopera-tively if fever is the only indication(level 3).

2. A urinalysis and culture are not man-datory during the initial 72 hrs post-operatively if fever is the only indica-tion. Urinalysis and culture should beperformed for those febrile patientshaving indwelling bladder cathetersfor �72 hrs (level 3).

3. Surgical wounds should be examineddaily for infection. They should not becultured if there is no symptom orsign suggesting infection (see below)(level 2).

4. A high level of suspicion should bemaintained for deep venous thrombo-sis, superficial thrombophlebitis, andpulmonary embolism, especially in pa-tients who are sedentary, have lowerlimb immobility, have a malignantneoplasm, or are taking an oral con-traceptive (level 2).

Surgical Site Infections

Surgical site infections alone accountfor approximately 25% of overall costs

related to treatment of nosocomial infec-tions. The rate of surgical site infection isapproximately 3%, based on infectionsand operative case loads sampled frommany hospitals (157, 158). This variesbased on the degree of contamination ofthe incision, the medical comorbidity ofthe patient (e.g., diabetes mellitus andobesity increase the risk), whether sur-gery is prolonged or an emergency, andwhether antimicrobial prophylaxis is ad-ministered correctly (e.g., appropriatenarrow spectrum of activity, administra-tion just before incision, and discontinu-ation within 24 hrs [48 hrs for cardiacsurgery]), if indicated (152).

Microbiology. In clean surgical proce-dures, S. aureus from the patient’s skinflora or the exogenous environment is themost common cause of surgical site infec-tion (159, 160). However, Gram-negativebacilli may be causative for surgical siteinfections of clean infra-inguinal opera-tions. For all other categories of proce-dures, the indigenous polymicrobial aer-obic-anaerobic flora of the organ or tissuebeing operated on are the most commonpathogens of surgical site infection. In-fections of contaminated operations ofperioral, perirectal, and vulvovaginal tis-sues yield bacteria that are similar to thenormal microbial flora of the adjacentmucous membrane. Infections of areasremote from those sites are caused pri-marily by indigenous skin microflora, es-pecially if no body cavity has been en-tered. Rarely, necrotizing incisionalsurgical site infection (superficial ordeep) may be caused by group A Strepto-coccus or Clostridium species, which, ifidentified or suspected, carries the samediagnostic and therapeutic urgency asother manifestations of necrotizing soft-tissue infection.

Although patients commonly receiveantibiotics when surgical site infection isfirst diagnosed, there is little or no evi-dence supporting this practice. Studies ofsubcutaneous abscesses found no benefitfor antibiotic therapy when combinedwith drainage (161, 162). A trial of anti-biotic administration for surgical site in-fections found no clinical benefit associ-ated with this treatment (163). Acommon practice endorsed by most ex-perts is to open all infected wounds (152).

Recommendations for Evaluation ofSurgical Site Infection

1. Examine the surgical incision at leastonce daily for erythema, purulence, or


tenderness as part of the fever evalua-tion (level 2).

2. If there is suspicion of infection, theincision should be opened and cul-tured (level 2).

3. Gram-negative stain and culturesshould be obtained from any expressedpurulence obtained from levels withinthe incision consistent with a deepincisional or organ/space surgical siteinfection. Tissue biopsies or aspiratesare preferable to swabs (level 3).

4. Drainage from superficial surgical siteinfections may not require Gram-negative stain and culture because in-cision, drainage, and local care may besufficient treatment and antibiotictherapy may not be required. Superfi-cial swab cultures are likely to be con-taminated with commensal skin floraand are not recommended (level 2).

5. Standard guidelines should be used todefine burn wound infection (level 3).

Central Nervous System Infection

A prospective study of fever in neuro-critical care patients indicates that al-though fever occurs in about 25% of suchpatients, almost half are noninfectious inetiology (164). Hence, the intensivistneeds to maintain a high index of suspi-cion for central nervous system infection.

Diagnostic Evaluation. Central ner-vous system infection rarely causes en-cephalopathy without focal abnormalitieson neurologic examination. However, inany febrile ICU patient, even without fo-cal findings, infection must be consideredbecause of the inherent limitations of theneurologic examination in critically illpatients. However, the yield of lumbarpuncture in patients without immunecompromise or central nervous systeminstrumentation who develop mental sta-tus changes in the ICU may be low (165).

Imaging studies and culture of the ce-rebrospinal fluid are the cardinal featuresof a diagnostic evaluation. Patients withfocal neurologic findings suggesting dis-ease above the foramen magnum willgenerally require an imaging study beforea lumbar puncture. A noncontrast CTscan is adequate to exclude mass lesionsor obstructive hydrocephalus, whichmight contraindicate a lumbar puncture.If bacterial meningitis is suspected andthe lumbar puncture is delayed for anyreason, including an imaging study, thenappropriate empirical antibiotic therapyfor meningitis due to rapidly fatal etiolo-gies (such as S. pneumoniae) should be

started after blood cultures are obtained.The usual contraindications to lumbarpuncture detected by CT scanning in-clude lateral shift of midline structures,loss of the suprachiasmatic and basilarcisterns, obliteration of the fourth ventri-cle, or obliteration of the superior cere-bellar and quadrigeminal plate cisternswith sparing of the ambient cisterns(166). If the physical examination sug-gests involvement of the spinal cord, con-sultation with a neurologist or neurosur-geon should be obtained because of thepotential for spinal cord herniation withan intra-axial mass. Whether to postponea lumbar puncture for an imaging studyin a patient who is unresponsive withoutfocal findings is a clinical decision.

Patients with suspected brain abscessesshould not undergo a lumbar puncture un-til the degree of cerebral edema and intra-cranial hypertension are determined (167).Aspiration of the suspected abscess is thediagnostic procedure of choice. The opti-mal timing of aspiration is currently de-bated; if it is delayed, empirical antibiotictherapy should be considered.

Patients with Intracranial Devices.When a patient with an intracranial device,such as a ventriculostomy catheter or aventriculoperitoneal shunt, becomes fe-brile, cerebrospinal fluid (CSF) should al-most always be obtained for analysis. Thesite of CSF access in the patient with aventriculostomy is straightforward. The pa-tient with a shunt system that includes aCSF reservoir should have the reservoiraspirated; this is also the case for patientswith Ommaya reservoirs. Patients in whomCSF flow to the lumbar subarachnoid spaceis obstructed may also need a lumbar punc-ture because one space may be infectedwhile the other is sterile. In patients withventriculostomies who develop stupor orsigns of meningitis, the catheter should beremoved and the tip cultured.

Tests to Be Performed on CSF. Basictests to be performed on CSF from pa-tients with suspected central nervous sys-tem infection include cell counts and dif-ferential, glucose and proteinconcentrations, Gram-negative stain, andbacterial cultures. Whether to performcryptococcal antigen testing, fungalstaining and cultures, acid-fast bacillussmears and cultures, cytologic examina-tion for neoplasia, polymerase chain re-action tests, or a serologic test for syphilisdepends on the clinical situation. The immu-nocompromised patient may require addi-tional tests such as polymerase chain reac-tion for herpes simplex virus, CMV,

Epstein-Barr virus, human herpes virus-6,JC virus, West Nile virus, adenovirus, andenterovirus. The precise combination oftests with the greatest likelihood of detect-ing bacterial infection depends on the pa-tient population. The normal protein con-tent of CSF varies with the site from whichit was withdrawn; ventricular fluid usuallyhas an upper protein content limit of 15mg/dL, cisternal fluid of 20 mg/dL, andlumbar fluid of 45 mg/dL.

Patients with bacterial meningitis typ-ically have a CSF glucose of �35 mg/dL,a CSF-blood glucose ratio of �0.23, aCSF protein level concentration of �220mg/dL, �2,000 total white blood cells/�L, or �1,180 neutrophils/�L (168).Conversely, in immunologically normalhosts, the presence of a normal openingpressure, �5 white blood cells/�L, and anormal CSF protein concentration essen-tially exclude meningitis (169). The ap-plicability of these findings to the criti-cally ill immunocompromised patient isuncertain, and a high index of suspicionfor infection should be maintained, re-gardless of cell count and glucose con-centration, until cultures are final. CSFlactate measurements may be useful inneurosurgical patients to distinguish in-fection from postoperative aseptic men-ingitis (170, 171).

Recommendations for Evaluation ofCentral Nervous System Infections

1. If altered consciousness or focal neu-rologic signs are unexplained, lumbarpuncture should be considered in anypatient with a new fever, unless thereis a contraindication to lumbar punc-ture (level 3).

2. For a patient with a new fever and newfocal neurologic findings suggestingdisease above the foramen magnum,an imaging study is usually requiredbefore lumbar puncture. If a mass ispresent, neurology/neurosurgery con-sultation is required to determine theoptimal diagnostic approach (level 2).

3. In febrile patients with an intracranialdevice, CSF should be obtained foranalysis from the CSF reservoir. IfCSF flow to the subarachnoid space isobstructed, it may be prudent to alsoobtain CSF from the lumbar space(level 3).

4. In patients with ventriculostomieswho develop stupor or signs of men-ingitis, the catheter should be re-moved and the tip cultured (level 3).

5. CSF should be evaluated by Gram-negative stain and culture, glucose,


protein, and cell count with differen-tial. Additional tests for tuberculosis,viral and fungal disease, neoplasia,etc., should be performed as dictatedby the clinical situation (level 2).

Other Considerations

Mononucleosis syndrome after bloodtransfusion has been broadly recognizedas a cause of hectic fever in postsurgicalpatients for more than four decades. In-fection may be caused by transmission ofCMV by way of transfused white bloodcells in the blood product or from reac-tivation of CMV by transfusion-inducedantigenic stimulation (172). The spec-trum of disease related to posttransfusionCMV ranges from asymptomatic conver-sion to a self-limited mononucleosis syn-drome. In symptomatic patients, symp-toms usually begin 1 month aftertransfusion. The classic features of mono-nucleosis are usually absent. Rather, pa-tients develop high fever, which oftenleads to empirical antimicrobial therapyfor sepsis. This syndrome should be sus-pected when patients with spiking feversdo not respond to antimicrobial therapyor when cultures for bacterial pathogensare negative. Immunocompetent patientswith this syndrome lack clinical toxicity,despite daily fever as high as 40°C. Thepresence of pancytopenia with atypicallymphocytosis and mild elevations ofliver function tests may also be clues tothis entity (173, 174). In contrast, immu-nocompromised patients can develop se-rious consequences, including fatal dis-seminated disease or diffuse interstitialpneumonia, especially in the setting ofprimary infection (seronegative patienttransfused with seropositive blood). Ingeneral, transplant services attempt toadminister screened, filtered, or leuko-cyte-reduced blood components to thisgroup of seriously ill patients to preventCMV disease. This has substantially re-duced but not completely eliminated therisk of this infection. Diagnosis can bemade by using quantitative molecularnucleic acid amplification tests to mea-sure serum viral load.

Silent sources of infection, includingbut not limited to otitis media, decubitusulcers at the sacrum or the back or thehead, perineal or perianal abscesses, andundetected retained tampons need to becarefully ruled out. The recognition andidentification of these other less commoninfectious causes of fever require a care-ful history and physical examination, in-

cluding the examination of all devices,drains, and hardware on a daily basis.

Use of Adjunctive Markers for theEvaluation of Fever. Several biomarkershave been investigated for their utility inrapidly discriminating true infectionfrom other inflammatory processes caus-ing fever. Several generations of serumprocalcitonin assays with variable cutoffpoints have been examined (175–180). Amore sensitive serum procalcitonin assayis now approved for the early detection ofbacterial infections and sepsis in patientsduring the first day of ICU admission.Procalcitonin level elevations of �0.5ng/mL occur within 2–3 hrs of onset,with higher levels observed along thecontinuum from systemic inflammatoryresponse syndrome (0.6–2.0 ng/mL), se-vere sepsis (2–10 ng/mL), and septicshock (�10 ng/mL). Most importantly,viral infections, recent surgery, andchronic inflammatory states are not as-sociated with an increment in procalcito-nin levels.

Traditional endotoxin detection sys-tems, such as the chromogenic limuluslysate assay, have been hampered by con-tamination concerns, cumbersome tech-niques, and variable sensitivity, with in-consistent correlation to the presence oftrue Gram-negative infection (181, 182).However, kinetic luminometric anti-assay (endotoxin activity assay) uses anantibody to the conserved lipid A compo-nent of endotoxin to form an antibody-antigen complex capable of stimulating adetectable host neutrophil respiratoryburst in an ex vivo assay (183). In a mul-ticenter ICU study, the endotoxin activityassay demonstrated a high negative pre-dictive value (98.6%) for Gram-negativeinfection (184).

Tumor necrosis factor-�, interleukin-6,C-reactive protein, and triggering receptorexpressed on myeloid cells-1 (TREM-1)have been tested as methods to discrimi-nate true infection from other inflamma-tory states but have not yet been validated.

Recommendation for Using Biomark-ers to Determine the Cause of Fever

1. Serum procalcitonin levels and endo-toxin activity assay can be employed asan adjunctive diagnostic tool for dis-criminating infection as the cause forfever or sepsis presentations (level 2).

Noninfectious Causes of Feverin the ICU

Drug-Related Fever. Any drug cancause fever due to hypersensitivity (185–

187). Hypersensitivity reactions maymanifest as fever alone to life-threateninghypersensitivity, such as those that havebeen associated with abacavir and nevi-rapine. In addition, some drugs cause fe-ver by producing local inflammation atthe site of administration (phlebitis, ster-ile abscesses, or soft-tissue reaction): am-photericin B, erythromycin, potassiumchloride, sulfonamides, and cytotoxicchemotherapies are prime examples.Drugs or their delivery systems (diluent,intravenous fluid, or intravascular deliv-ery devices) may also contain pyrogensor, rarely, microbial contaminants (188).Some drugs may also stimulate heat pro-duction (e.g., thyroxine), limit heat dissi-pation (e.g., atropine or epinephrine), oralter thermoregulation (e.g., butyrophe-none tranquilizers, phenothiazines, anti-histamines, or antiparkinson drugs).

Among drug categories, fever is mostoften attributed to antimicrobials (espe-cially �-lactam drugs), anti-epilepticdrugs (especially phenytoin), antiar-rhythmics (especially quinidine and pro-cainamide), and antihypertensives (meth-yldopa). There is nothing characteristicabout the fevers induced by these drugs(186). Fevers do not invariably occur im-mediately after drug administration: itmay be days after administration that fe-ver occurs and many more days beforethe fever abates. In one series, the lagtime between initiating a drug and feverwas a mean of 21 days (median, 8). Feveroften took 1 to 3 days to return to normalbut can take �7 days to return to normalafter removing the offending agent (189).Rash occurs in a small fraction of cases.Eosinophilia is also uncommon.

The diagnosis of drug-induced fever isusually established by temporal relation-ship of the fever to starting and stoppingthe drug. Patients can be rechallengedwith the drug to confirm the diagnosis,but this is rarely done unless the drug inquestion is essential and alternatives arenot available. Patients who had anaphy-laxis or toxic epidermal necrolysis as aresult of drug exposure should not berechallenged.

Two important syndromes, malignanthyperthermia and neuroleptic malignantsyndrome, deserve consideration whenfever is especially high because the re-sults can be devastating if left untreated(190, 191). Malignant hyperthermia ismore often identified in the operatingroom than in the ICU, but onset can bedelayed for as long as 24 hrs, especially ifthe patient is on steroids. It can be caused


by succinylcholine and the inhalation an-esthetics, of which halothane is the mostfrequently identified. This hyperthermicsyndrome is believed to be a geneticallydetermined response mediated by a dys-regulation of cytoplasmic calcium con-trol in skeletal muscle. The result of thiscalcium dysregulation is intense musclecontraction, generating fever and in-creasing creatinine phosphokinase con-centrations.

The neuroleptic malignant syndrome israre but more often identified in the ICUthan malignant hyperthermia. It has beenstrongly associated with antipsychotic neu-roleptic medications—phenothiazines,thioxanthenes, and butyrophenones (190,192). In the ICU, haloperidol is perhaps themost frequently reported drug. It manifestsas muscle rigidity, generating fever and in-creasing creatinine phosphokinase concen-trations. However, unlike malignant hyper-thermia, the initiator of muscle contractionis central.

The serotonin syndrome may be con-fused with neuroleptic malignant syn-drome, but it is a distinct entity (193).The mechanism is related to excessivestimulation of the 5-HT1A–receptor andis increasingly seen with the popularity ofserotonin reuptake inhibitors in thetreatment of various psychiatric disor-ders. The serotonin syndrome may be ex-acerbated with concomitant use of lin-ezolid (194).

Importantly, withdrawal of certaindrugs may be associated with fever, oftenwith associated tachycardia, diaphoresis,and hyperreflexia. Alcohol, opiates (in-cluding methadone), barbiturates, andbenzodiazepines have all been associatedwith this febrile syndrome. It is impor-tant to recognize that a history of use ofthese drugs may not be available whenthe patient is admitted to the ICU. With-drawal and related fever may thereforeoccur several hours or days after admis-sion. Fever related to other therapies andinflammatory states are listed in Table 5.

Recommendations for RecognizingNoninfectious Causes of Fever

1. Consider all new medications andblood products the patient has re-ceived. Ideally, if the suspected drugcan be stopped, do so. If the drug can-not be stopped, consider a comparablesubstitute (level 2).

2. Fever induced by drugs may take severaldays to resolve. Establishing a temporalrelationship between fever and the of-

fending agent may be helpful in estab-lishing the diagnosis (level 3).

Considerations for EmpiricTherapy During Diagnostics

Initiation of therapy may be necessaryfor unstable or high-risk patients whilethe diagnostic evaluation is ongoing, andcertainly before the results of cultures areavailable. Usually, this entails antimicro-bial therapy, but treatment may also haveto be considered for noninfectious causesof fever as well.

If an infectious cause of fever is sus-pected, empirical antimicrobial therapymay be urgent. Delay of effective antimi-crobial therapy has been associated withincreased mortality from infection andsepsis (195–201). Antibiotic therapyshould begin within 1 hr after the diag-nosis of sepsis is considered (202). Thechoice of regimen depends on the sus-pected infectious etiology, whether theinfection is community-, healthcare-, orhospital-onset, and whether the patient isimmunocompromised. If drug-resistantpathogens are suspected, initial broad-spectrum empirical antimicrobial ther-apy directed against both resistant Gram-positive cocci (including methicillin-resistant S. aureus) and Gram-negativebacilli is indicated. This may require sev-eral agents to ensure that resistant patho-gens are covered. In addition, empiricalantifungal coverage may be appropriatein selected patients.

Recommendations for Empiric Ther-apy of Fever

1. When clinical evaluation suggests thatinfection is the cause of fever, consid-eration should be given to administer-ing empirical antimicrobial therapy assoon as possible after cultures are ob-tained, especially if the patient is seri-ously ill or deteriorating (level 1).

2. Initial empirical antibiotic therapyshould be directed against likelypathogens, as suggested by the sus-pected source of infection, the patientrisk for infection by multidrug-resis-tant pathogens, and local knowledgeof antimicrobial susceptibility pat-terns (level 1).

REFERENCES

Letters after the reference numbers indicate So-ciety of Critical Care Medicine’s rating system(see Table 1).

1c. O’Grady NP, Barie PS, Bartlett JG, et al:Practice guidelines for evaluating new feverin critically ill adult patients: Task Force ofthe Society of Critical Care Medicine andthe Infectious Diseases Society of America.Clin Infect Dis 1998; 26:1042–1059

2b.Arbo MJ, Fine MJ, Hanusa BH, et al: Feverof nosocomial origin: Etiology, risk factors,and outcomes. Am J Med 1993; 95:505–512

3d.Bone RC, Balk RA, Cerra FB, et al: Defini-tions for sepsis and organ failure and guide-lines for the use of innovative therapies insepsis: The ACCP/SCCM Consensus Confer-ence Committee. American College of ChestPhysicians/Society of Critical Care Medi-cine. Chest 1992; 101:1644–1655

4c. Hughes WT, Armstrong D, Bodey GP, et al:2002 guidelines for the use of antimicrobialagents in neutropenic patients with cancer.Clin Infect Dis 2002; 34:730–751

5c. Mackowiak PA, Wasserman SS, Levine MM:A critical appraisal of 98.6 degrees F, theupper limit of the normal body tempera-ture, and other legacies of Carl ReinholdAugust Wunderlich. JAMA 1992; 268:1578–1580

6c. Dinarello CA, Cannon JG, Wolff SM: Newconcepts on the pathogenesis of fever. RevInfect Dis 1988; 10:168–189

7b.Waterhouse J, Edwards B, Bedford P, et al:Thermoregulation during mild exercise atdifferent circadian times. Chronobiol Int2004; 21:253–275

8c. Insler SR, Sessler DI: Perioperative thermo-regulation and temperature monitoring.Anesthesiol Clin 2006; 24:823–837

9d.van der Sande FM, Kooman JP, LeunissenKM: Haemodialysis and thermoregulation.Nephrol Dial Transplant 2006; 21:1450–1451

10b.van der Sande FM, Rosales LM, Brener Z, etal: Effect of ultrafiltration on thermal vari-

Table 5. Fever related to other therapies andnoninfectious inflammatory states

Acalculous cholecystitisAcute myocardial infarctionAdrenal insufficiencyBlood product transfusionCytokine-related feverDressler syndrome (pericardial injury syndrome)Drug feverFat emboliFibroproliferative phase of acute respiratory

distress syndromeGoutHeterotopic ossificationImmune reconstitution inflammatory syndromeIntracranial bleedJarisch-Herxheimer reactionPancreatitisPulmonary infarctionPneumonitis without infectionStrokeThyroid stormTransplant rejectionTumor lysis syndromeVenous thrombosis


ables, skin temperature, skin blood flow,and energy expenditure during ultrapurehemodialysis. J Am Soc Nephrol 2005; 16:1824–1831

11c. Le Blanc L, Lesur O, Valiquette L, et al: Roleof routine blood cultures in detecting un-apparent infections during continuous re-nal replacement therapy. Intensive CareMed 2006; 32:1802–1807

12b.Erickson RS, Kirklin SK: Comparison ofear-based, bladder, oral, and axillary meth-ods for core temperature measurement.Crit Care Med 1993; 21:1528–1534

13b.Erickson RS, Meyer LT: Accuracy of infra-red ear thermometry and other tempera-ture methods in adults. Am J Crit Care1994; 3:40–54

14b.Schmitz T, Bair N, Falk M, et al: A compar-ison of five methods of temperature mea-surement in febrile intensive care patients.Am J Crit Care 1995; 4:286–292

15b.Milewski A, Ferguson KL, Terndrup TE:Comparison of pulmonary artery, rectal,and tympanic membrane temperatures inadult intensive care unit patients. Clin Pe-diatr (Phila) 1991; 30(4 Suppl):13–16

16b.Nierman DM: Core temperature measure-ment in the intensive care unit. Crit CareMed 1991; 19:818–823

17b.Fallis WM: The effect of urine flow rate onurinary bladder temperature in critically illadults. Heart Lung 2005; 34:209–216

18b.Moran JL, Peter JV, Solomon PJ, et al: Tym-panic temperature measurements: Are theyreliable in the critically ill? A clinical studyof measures of agreement. Crit Care Med2007; 35:155–164

19b.Lefrant JY, Muller L, de La Coussaye JE, et al:Temperature measurement in intensive carepatients: Comparison of urinary bladder,oesophageal, rectal, axillary, and inguinalmethods versus pulmonary artery coremethod. Intensive Care Med 2003; 29:414–418

20b.Eichna LW, Berger AR, Rader B, et al: Com-parison of intracardiac and intravasculartemperatures with rectal temperatures inman. J Clin Invest 1951; 30:353–359

21b. Ilsley AH, Rutten AJ, Runciman WB: Anevaluation of body temperature measure-ment. Anaesth Intensive Care 1983; 11:31–39

22b.Brooks S, Khan A, Stoica D, et al: Reductionin vancomycin-resistant Enterococcus andClostridium difficile infections followingchange to tympanic thermometers. InfectControl Hosp Epidemiol 1998; 19:333–336

23b.Livornese LL Jr, Dias S, Samel C, et al:Hospital-acquired infection with vancomycin-resistant Enterococcus faecium transmittedby electronic thermometers. Ann InternMed 1992; 117:112–116

24d.Gerding DN, Johnson S, Peterson LR, et al:Clostridium difficile–associated diarrheaand colitis. Infect Control Hosp Epidemiol1995; 16:459–477

25b.Cranston WI, Gerbrandy J, Snell ES: Oral,rectal and oesophageal temperatures and

some factors affecting them in man.J Physiol 1954; 126:347–358

26b.Farnell S, Maxwell L, Tan S, et al: Temper-ature measurement: Comparison of non-invasive methods used in adult critical care.J Clin Nurs 2005; 14:632–639

27b.Hebbar K, Fortenberry JD, Rogers K, et al:Comparison of temporal artery thermome-ter to standard temperature measurementsin pediatric intensive care unit patients. Pe-diatr Crit Care Med 2005; 6:557–561

28b.Harioka T, Matsukawa T, Ozaki M, et al:“Deep-forehead” temperature correlateswell with blood temperature. Can J Anaesth2000; 47:980–983

29b.Kistemaker JA, Den Hartog EA, Daanen HA:Reliability of an infrared forehead skin ther-mometer for core temperature measure-ments. J Med Eng Technol 2006; 30:252–261

30b.Potter P, Schallom M, Davis S, et al: Eval-uation of chemical dot thermometers formeasuring body temperature of orally intu-bated patients. Am J Crit Care 2003; 12:403–407

31a. Strand CL, Wajsbort RR, Sturmann K: Ef-fect of iodophor vs iodine tincture skinpreparation on blood culture contamina-tion rate. JAMA 1993; 269:1004–1006

32b.Weinstein MP, Reller LB, Murphy JR, et al:The clinical significance of positive bloodcultures: A comprehensive analysis of 500episodes of bacteremia and fungemia inadults. I. Laboratory and epidemiologic ob-servations. Rev Infect Dis 1983; 5:35–53

33b.Trautner BW, Clarridge JE, Darouiche RO:Skin antisepsis kits containing alcohol andchlorhexidine gluconate or tincture of io-dine are associated with low rates of bloodculture contamination. Infect Control HospEpidemiol 2002; 23:397–401

34b.Barenfanger J, Drake C, Lawhorn J, et al:Comparison of chlorhexidine and tinctureof iodine for skin antisepsis in preparationfor blood sample collection. J Clin Micro-biol 2004; 42:2216–2217

35a. Little JR, Murray PR, Traynor PS, et al: Arandomized trial of povidone-iodine com-pared with iodine tincture for venipuncturesite disinfection: Effects on rates of bloodculture contamination. Am J Med 1999;107:119–125

36a. Mimoz O, Karim A, Mercat A, et al: Chlo-rhexidine compared with povidone-iodineas skin preparation before blood culture: Arandomized, controlled trial. Ann InternMed 1999; 131:834–837

37d.Clinical and Laboratory Standards Institute:Principles and procedures for blood cul-tures: Proposed Guideline. Wayne, PA, Clin-ical and Laboratory Standards Institute,CLSI Document M47-P 2006

38b.Leisure MK, Moore DM, Schwartzman JD,et al: Changing the needle when inoculat-ing blood cultures: A no-benefit and high-risk procedure. JAMA 1990; 264:2111–2112

39b.Cockerill FR III, Wilson JW, Vetter EA, et al:

Optimal testing parameters for blood cul-tures. Clin Infect Dis 2004; 38:1724–1730

40b.Tenney JH, Reller LB, Mirrett S, et al: Con-trolled evaluation of the volume of bloodcultured in detection of bacteremia andfungemia. J Clin Microbiol 1982; 15:558–561

41b.Salventi JF, Davies TA, Randall EL, et al:Effect of blood dilution on recovery of or-ganisms from clinical blood cultures in me-dium containing sodium polyanethol sulfo-nate. J Clin Microbiol 1979; 9:248–252

42b.Mermel LA, Maki DG: Detection of bactere-mia in adults: Consequences of culturing aninadequate volume of blood. Ann InternMed 1993; 119:270–272

43b. Ilstrup DM, Washington JA Jr: The impor-tance of volume of blood cultured in thedetection of bacteremia and fungemia. Di-agn Microbiol Infect Dis 1983; 1:107–110

44b.Hall MM, Ilstrup DM, Washington JA Jr.Effect of volume of blood cultured on de-tection of bacteremia. J Clin Microbiol1976; 3:643–645

45b.Mirrett S, Everts RJ, Reller LB: Controlledcomparison of original vented aerobic fanmedium with new nonvented BacT/ALERTFA medium for culturing blood. J Clin Mi-crobiol 2001; 39:2098–2101

46b. Jorgensen JH, Mirrett S, McDonald LC, etal: Controlled clinical laboratory compari-son of BACTEC plus aerobic/F resin me-dium with BacT/Alert aerobic FAN mediumfor detection of bacteremia and fungemia.J Clin Microbiol 1997; 35:53–58

47b.Weinstein MP, Mirrett S, Reimer LG, et al:Controlled evaluation of BacT/Alert stan-dard aerobic and FAN aerobic blood culturebottles for detection of bacteremia and fun-gemia. J Clin Microbiol 1995; 33:978–981

48b.Lee A, Mirrett S, Reller LB, et al: Detectionof bloodstream infections in adults: Howmany blood cultures are needed? J Clin Mi-crobiol 2007; 45:3546–3548

49b.Hudson-Civetta JA, Civetta JM, MartinezOV, et al: Risk and detection of pulmonaryartery catheter-related infection in septicsurgical patients. Crit Care Med 1987; 15:29–34

50b.Bates DW, Goldman L, Lee TH: Contami-nant blood cultures and resource utiliza-tion: The true consequences of false-positive results. JAMA 1991; 265:365–369

51b.Wormser GP, Onorato IM, Preminger TJ, etal: Sensitivity and specificity of blood cul-tures obtained through intravascular cath-eters. Crit Care Med 1990; 18:152–156

52b.Tokars JI: Predictive value of blood culturespositive for coagulase-negative staphylo-cocci: Implications for patient care andhealth care quality assurance. Clin InfectDis 2004; 39:333–341

53b.Bryant JK, Strand CL: Reliability of bloodcultures collected from intravascular cath-eter versus venipuncture. Am J Clin Pathol1987; 88:113–116

54c. Washington JA Jr, Ilstrup DM: Blood cul-


tures: Issues and controversies. Rev InfectDis 1986; 8:792–802

55c. Weinstein MP: Current blood culture meth-ods and systems: Clinical concepts, technol-ogy, and interpretation of results. Clin In-fect Dis 1996; 23:40–46

56b.Li J, Plorde JJ, Carlson LG: Effects of vol-ume and periodicity on blood cultures.J Clin Microbiol 1994; 32:2829–2831

57c. Maki DG, Kluger DM, Crnich CJ: The risk ofbloodstream infection in adults with differ-ent intravascular devices: A systematic re-view of 200 published prospective studies.Mayo Clin Proc 2006; 81:1159–1171

58d.Crnich CJ: Infections caused by intravascu-lar devices: Epidemiology, pathogenesis, di-agnosis, prevention, and treatment. In:APIC Text of Infection Control and Epide-miology. Second Edition. Carrico R (Ed).Washington, DC, Association for Profes-sionals in Infection Control and Epidemiol-ogy, 2005, pp 24.1–24.6

59a. Leon C, Alvarez-Lerma F, Ruiz-Santana S,et al: Antiseptic chamber-containing hubreduces central venous catheter-related in-fection: A prospective, randomized study.Crit Care Med 2003; 31:1318–1324

60b.Mermel LA, McCormick RD, Springman SR,et al: The pathogenesis and epidemiology ofcatheter-related infection with pulmonaryartery Swan-Ganz catheters: A prospectivestudy utilizing molecular subtyping.Am J Med 1991; 91(3B):197S–205S

61b.Safdar N, Maki DG: Inflammation at theinsertion site is not predictive of catheter-related bloodstream infection with short-term, noncuffed central venous catheters.Crit Care Med 2002; 30:2632–2635

62b.Blot F, Nitenberg G, Chachaty E, et al: Di-agnosis of catheter-related bacteraemia: Aprospective comparison of the time to pos-itivity of hub-blood versus peripheral-bloodcultures. Lancet 1999; 354:1071–1077

63b.DesJardin JA, Falagas ME, Ruthazer R, et al:Clinical utility of blood cultures drawn fromindwelling central venous catheters in hos-pitalized patients with cancer. Ann InternMed 1999; 131:641–647

64b.Maki DG, Weise CE, Sarafin HW: A semi-quantitative culture method for identifyingintravenous-catheter-related infection.N Engl J Med 1977; 296:1305–1309

65b.Safdar N, Fine JP, Maki DG: Meta-analysis:Methods for diagnosing intravascular de-vice-related bloodstream infection. Ann In-tern Med 2005; 142:451–466

66b.Cooper GL, Hopkins CC: Rapid diagnosis ofintravascular catheter-associated infectionby direct Gram staining of catheter seg-ments. N Engl J Med 1985; 312:1142–1147

67b.Zufferey J, Rime B, Francioli P, et al: Simplemethod for rapid diagnosis of catheter-associated infection by direct acridine or-ange staining of catheter tips. J Clin Micro-biol 1988; 26:175–177

68c. Mermel LA, Farr BM, Sherertz RJ, et al:Guidelines for the management of intravas-

cular catheter-related infections. Clin InfectDis 2001; 32:1249–1272

69c. Tacconelli E, Tumbarello M, Pittiruti M, etal: Central venous catheter-related sepsis ina cohort of 366 hospitalised patients. EurJ Clin Microbiol Infect Dis 1997; 16:203–209

70c. Raad II, Baba M, Bodey GP: Diagnosis ofcatheter-related infections: The role of sur-veillance and targeted quantitative skin cul-tures. Clin Infect Dis 1995; 20:593–597

71b.Catton JA, Dobbins BM, Kite P, et al: In situdiagnosis of intravascular catheter-relatedbloodstream infection: A comparison ofquantitative culture, differential time topositivity, and endoluminal brushing. CritCare Med 2005; 33:787–791

72c. Mayhall CG: Diagnosis and management ofinfections of implantable devices used forprolonged venous access. Curr Clin Top In-fect Dis 1992; 12:83–110

73c. Kaufman J, Demas C, Stark K, et al: Cath-eter-related septic central venous thrombo-sis: Current therapeutic options. WestJ Med 1986; 145:200–203

74c. Strinden WD, Helgerson RB, Maki DG: Can-dida septic thrombosis of the great centralveins associated with central catheters:Clinical features and management. AnnSurg 1985; 202:653–658

75c. Verghese A, Widrich WC, Arbeit RD: Centralvenous septic thrombophlebitis: The role ofmedical therapy. Medicine (Baltimore)1985; 64:394–400

76b.Rello J, Ollendorf DA, Oster G, et al: Epide-miology and outcomes of ventilator-associ-ated pneumonia in a large US database.Chest 2002; 122:2115–2121

77c. Chastre J, Fagon JY: Ventilator-associatedpneumonia. Am J Respir Crit Care Med2002; 165:867–903

78b.Sawyer RG, Crabtree TD, Gleason TG, et al:Impact of solid organ transplantation andimmunosuppression on fever, leukocytosis,and physiologic response during bacterialand fungal infections. Clin Transplant1999; 13:260–265

79b.Pelletier SJ, Crabtree TD, Gleason TG, et al:Characteristics of infectious complicationsassociated with mortality after solid organtransplantation. Clin Transplant 2000; 14(4Pt 2):401–408

80b.Combes A, Figliolini C, Trouillet JL, et al:Factors predicting ventilator-associatedpneumonia recurrence. Crit Care Med2003; 31:1102–1107

81c. Winer-Muram HT, Rubin SA, Ellis JV, et al:Pneumonia and ARDS in patients receivingmechanical ventilation: Diagnostic accu-racy of chest radiography. Radiology 1993;188:479–485

82c. Wunderink RG, Woldenberg LS, Zeiss J, etal: The radiologic diagnosis of autopsy-proven ventilator-associated pneumonia.Chest 1992; 101:458–463

83c. Barkhausen J, Stoblen F, Dominguez-Fernandez E, et al: Impact of CT in patients

with sepsis of unknown origin. Acta Radiol1999; 40:552–555

84c. Miller WT Jr, Tino G, Friedburg JS: Tho-racic CT in the intensive care unit: Assess-ment of clinical usefulness. Radiology 1998;209:491–498

85c. Winer-Muram HT, Steiner RM, Gurney JW,et al: Ventilator-associated pneumonia inpatients with adult respiratory distress syn-drome: CT evaluation. Radiology 1998; 208:193–199

86c. Franquet T: High-resolution computed to-mography (HRCT) of lung infections innon-AIDS immunocompromised patients.Eur Radiol 2006; 16:707–718

87c. Hiorns MP, Screaton NJ, Muller NL: Acutelung disease in the immunocompromisedhost. Radiol Clin North Am 2001; 39:1137–1151

88c. Guidelines for the management of adultswith hospital-acquired, ventilator-associ-ated, and healthcare-associated pneumonia.Am J Respir Crit Care Med 2005; 171:388–416

89c. Bartlett JG, Alexander J, Mayhew J, et al:Should fiberoptic bronchoscopy aspiratesbe cultured? Am Rev Respir Dis 1976; 114:73–78

90c. Jain P, Sandur S, Meli Y, et al: Role offlexible bronchoscopy in immunocompro-mised patients with lung infiltrates. Chest2004; 125:712–722

91c. Campbell GD Jr: Blinded invasive diagnos-tic procedures in ventilator-associatedpneumonia. Chest 2000; 117(4 Suppl 2):207S–211S

92c. Peikert T, Rana S, Edell ES: Safety, diagnos-tic yield, and therapeutic implications offlexible bronchoscopy in patients with fe-brile neutropenia and pulmonary infil-trates. Mayo Clin Proc 2005; 80:1414–1420

93c. Rello J, Esandi ME, Diaz E, et al: The role ofCandida sp isolated from bronchoscopicsamples in nonneutropenic patients. Chest1998; 114:146–149

94b.Rishmawi N, Ghneim R, Kattan R, et al:Survival of fastidious and nonfastidious aer-obic bacteria in three bacterial transportswab systems. J Clin Microbiol 2007; 45:1278–1283

95b.el-Ebiary M, Torres A, Fabregas N, et al:Significance of the isolation of Candida spe-cies from respiratory samples in criticallyill, non-neutropenic patients: An immediatepostmortem histologic study. Am J RespirCrit Care Med 1997; 156(2 Pt 1):583–590

96b.Barenfanger J, Arakere P, Cruz RD, et al:Improved outcomes associated with limit-ing identification of Candida spp. in respi-ratory secretions. J Clin Microbiol 2003;41:5645–5649

97a. Heyland D, Dodek P, Muscedere J, et al: Arandomized trial of diagnostic techniques forventilator-associated pneumonia. N Engl J Med2006; 355:2619–2630

98b.Souweine B, Veber B, Bedos JP, et al: Diag-nostic accuracy of protected specimenbrush and bronchoalveolar lavage in noso-


comial pneumonia: Impact of previous an-timicrobial treatments. Crit Care Med 1998;26:236–244

99b.Fagon JY, Chastre J, Hance AJ, et al: Detec-tion of nosocomial lung infection in venti-lated patients: Use of a protected specimenbrush and quantitative culture techniquesin 147 patients. Am Rev Respir Dis 1988;138:110–116

100c. Niederman MS, Torres A, Summer W: Inva-sive diagnostic testing is not needed rou-tinely to manage suspected ventilator-associated pneumonia. Am J Respir CritCare Med 1994; 150:565–569

101b. Pugin J, Auckenthaler R, Mili N, et al: Di-agnosis of ventilator-associated pneumoniaby bacteriologic analysis of bronchoscopicand nonbronchoscopic “blind” bronchoal-veolar lavage fluid. Am Rev Respir Dis 1991;143(5 Pt 1):1121–1129

102b. Lipscomb DJ, Flower CD, Hadfield JW: Ul-trasound of the pleura: An assessment of itsclinical value. Clin Radiol 1981; 32:289–290

103a. Grogan DR, Irwin RS, Channick R, et al:Complications associated with thoracente-sis: A prospective, randomized study com-paring three different methods. Arch InternMed 1990; 150:873–877

104c. Fekety R: Guidelines for the diagnosis andmanagement of Clostridium difficile–associated diarrhea and colitis: AmericanCollege of Gastroenterology, Practice Pa-rameters Committee. Am J Gastroenterol1997; 92:739–750

105d. DeMaio J, Bartlett JG: Update on diagnosisof Clostridium difficile–associated diarrhea.Curr Clin Top Infect Dis 1995; 15:97–114

106c. Bartlett JG: Clostridium difficile: History ofits role as an enteric pathogen and the cur-rent state of knowledge about the organism.Clin Infect Dis 1994; 18(Suppl 4):S265–S272

107c. Fekety R, Shah AB: Diagnosis and treat-ment of Clostridium difficile colitis. JAMA1993; 269:71–75

108b. Pepin J, Saheb N, Coulombe MA, et al:Emergence of fluoroquinolones as the pre-dominant risk factor for Clostridiumdifficile–associated diarrhea: A cohort studyduring an epidemic in Quebec. Clin InfectDis 2005; 41:1254–1260

109b. Pepin J, Valiquette L, Alary ME, et al: Clos-tridium difficile–associated diarrhea in a re-gion of Quebec from 1991 to 2003: A chang-ing pattern of disease severity. CMAJ 2004;171:466–472

110c. Bartlett JG: Narrative review: The new epi-demic of Clostridium difficile–associatedenteric disease. Ann Intern Med 2006; 145:758–764

111c. Borek AP, Aird DZ, Carroll KC: Frequencyof sample submission for optimal utiliza-tion of the cell culture cytotoxicity assay fordetection of Clostridium difficile toxin.J Clin Microbiol 2005; 43:2994–2995

112b. Manabe YC, Vinetz JM, Moore RD, et al:

Clostridium difficile colitis: An efficientclinical approach to diagnosis. Ann InternMed 1995; 123:835–840

113c. Johnson S, Kent SA, O’Leary KJ, et al: Fatalpseudomembranous colitis associated witha variant Clostridium difficile strain not de-tected by toxin A immunoassay. Ann InternMed 2001; 135:434–438

114b. Zheng L, Keller SF, Lyerly DM, et al: Mul-ticenter evaluation of a new screening testthat detects Clostridium difficile in fecalspecimens. J Clin Microbiol 2004; 42:3837–3840

115c. Ticehurst JR, Aird DZ, Dam LM, et al: Ef-fective detection of toxigenic Clostridiumdifficile by a two-step algorithm includingtests for antigen and cytotoxin. J Clin Mi-crobiol 2006; 44:1145–1149

116b. Walker RC, Ruane PJ, Rosenblatt JE, et al:Comparison of culture, cytotoxicity assays,and enzyme-linked immunosorbent assayfor toxin A and toxin B in the diagnosis ofClostridium difficile–related enteric dis-ease. Diagn Microbiol Infect Dis 1986;5:61–69

117c. Loo VG, Poirier L, Miller MA, et al: A pre-dominantly clonal multi-institutional out-break of Clostridium difficile–associated di-arrhea with high morbidity and mortality.N Engl J Med 2005; 353:2442–2449

118c. McDonald LC, Killgore GE, Thompson A, etal: An epidemic, toxin gene-variant strain ofClostridium difficile. N Engl J Med 2005;353:2433–2441

119c. Warny M, Pepin J, Fang A, et al: Toxinproduction by an emerging strain of Clos-tridium difficile associated with outbreaksof severe disease in North America and Eu-rope. Lancet 2005; 366:1079–1084

120b. Tedesco FJ, Corless JK, Brownstein RE: Rectalsparing in antibiotic-associated pseudomembra-nous colitis: A prospective study. Gastroenterol-ogy 1982; 83:1259–1260

121b. Talbot RW, Walker RC, Beart RW Jr: Chang-ing epidemiology, diagnosis, and treatmentof Clostridium difficile toxin–associated co-litis. Br J Surg 1986; 73:457–460

122c. Hogenauer C, Langner C, Beubler E, et al:Klebsiella oxytoca as a causative organismof antibiotic-associated hemorrhagic colitis.N Engl J Med 2006; 355:2418–2426

123c. Tambyah PA, Maki DG: Catheter-associatedurinary tract infection is rarely symptom-atic: A prospective study of 1,497 catheter-ized patients. Arch Intern Med 2000; 160:678–682

124b. Safdar N, Slattery WR, Knasinski V, et al:Predictors and outcomes of candiduria inrenal transplant recipients. Clin Infect Dis2005; 40:1413–1421

125c. Bryan CS, Reynolds KL: Hospital-acquiredbacteremic urinary tract infection: Epide-miology and outcome. J Urol 1984; 132:494–498

126c. Quintiliani R, Klimek J, Cunha BA, et al:Bacteraemia after manipulation of the uri-nary tract: The importance of pre-existing

urinary tract disease and compromised hostdefences. Postgrad Med J 1978; 54:668–671

127b. Laupland KB, Zygun DA, Davies HD, et al:Incidence and risk factors for acquiringnosocomial urinary tract infection in thecritically ill. J Crit Care 2002; 17:50–57

128b. Laupland KB, Bagshaw SM, Gregson DB, etal: Intensive care unit-acquired urinarytract infections in a regional critical caresystem. Crit Care 2005; 9:R60–R65

129b. Shah PS, Cannon JP, Sullivan CL, et al:Controlling antimicrobial use and decreas-ing microbiological laboratory tests for uri-nary tract infections in spinal-cord-injurypatients with chronic indwelling catheters.Am J Health Syst Pharm 2005; 62:74–77

130d. Thongboonkerd V, Saetun P: Bacterial over-growth affects urinary proteome analysis:Recommendation for centrifugation, tem-perature, duration, and the use of preserva-tives during sample collection. J ProteomeRes 2007; 6:4173–4181

131c. Stark RP, Maki DG: Bacteriuria in the cath-eterized patient: What quantitative level ofbacteriuria is relevant? N Engl J Med 1984;311:560–564

132c. Cornia PB, Takahashi TA, Lipsky BA: Themicrobiology of bacteriuria in men: A5-year study at a Veterans’ Affairs hospital.Diagn Microbiol Infect Dis 2006; 56:25–30

133c. Warren JW, Muncie HL Jr, Bergquist EJ, etal: Sequelae and management of urinaryinfection in the patient requiring chroniccatheterization. J Urol 1981; 125:1–8

134b. Schwartz DS, Barone JE: Correlation of uri-nalysis and dipstick results with catheter-associated urinary tract infections in surgi-cal ICU patients. Intensive Care Med 2006;32:1797–1801

135b. Schiotz HA: The value of leucocyte stix re-sults in predicting bacteriuria and urinarytract infection after gynaecological surgery.J Obstet Gynaecol 1999; 19:396–398

136c. Stein M, Caplan ES: Nosocomial sinusitis: Aunique subset of sinusitis. Curr Opin InfectDis 2005; 18:147–150

137c. Westergren V, Forsum U, Lundgren J: Pos-sible errors in diagnosis of bacterial sinus-itis in tracheal intubated patients. Acta An-aesthesiol Scand 1994; 38:699–703

138b. Rouby JJ, Laurent P, Gosnach M, et al: Riskfactors and clinical relevance of nosocomialmaxillary sinusitis in the critically ill. Am JRespir Crit Care Med 1994; 150:776–783

139b. Caplan ES, Hoyt NJ: Nosocomial sinusitis.JAMA 1982; 247:639–641

140b. Grindlinger GA, Niehoff J, Hughes SL, et al:Acute paranasal sinusitis related to nasotra-cheal intubation of head-injured patients.Crit Care Med 1987; 15:214–217

141c. Aebert H, Hunefeld G, Regel G: Paranasalsinusitis and sepsis in ICU patients withnasotracheal intubation. Intensive CareMed 1988; 15:27–30

142c. Shapiro GG, Rachelefsky GS: Introductionand definition of sinusitis. J Allergy ClinImmunol 1992; 90(3 Pt 2):417–418

143c. Vargas F, Bui HN, Boyer A, et al: Transnasal


puncture based on echographic sinusitis ev-idence in mechanically ventilated patientswith suspicion of nosocomial maxillary si-nusitis. Intensive Care Med 2006; 32:858–866

144c. Chidekel N, Jensen C, Axelsson A, et al:Diagnosis of fluid in the maxillary sinus.Acta Radiol Diagn (Stockh) 1970; 10:433–440

145b. Hamory BH, Sande MA, Sydnor A Jr, et al:Etiology and antimicrobial therapy of acutemaxillary sinusitis. J Infect Dis 1979; 139:197–202

146b. Roberts DN, Hampal S, East CA, et al: Thediagnosis of inflammatory sinonasal dis-ease. J Laryngol Otol 1995; 109:27–30

147c. Osguthorpe JD, Hadley JA: Rhinosinusitis:Current concepts in evaluation and man-agement. Med Clin North Am 1999; 83:27–41

148c. Zinreich SJ: Rhinosinusitis: Radiologic di-agnosis. Otolaryngol Head Neck Surg 1997;117(3 Pt 2):S27–S34

149b. Kountakis SE, Skoulas IG: Middle meatal vsantral lavage cultures in intensive care unitpatients. Otolaryngol Head Neck Surg2002; 126:377–381

150b. Garibaldi RA, Brodine S, Matsumiya S, et al:Evidence for the non-infectious etiology ofearly postoperative fever. Infect Control1985; 6:273–277

151c. Cheadle WG: Current perspectives on anti-biotic use in the treatment of surgical in-fections. Am J Surg 1992; 164(4A Suppl):44S–47S

152c. Stevens DL, Bisno AL, Chambers HF, et al:Practice guidelines for the diagnosis andmanagement of skin and soft-tissue infec-tions. Clin Infect Dis 2005; 41:1373–1406

153b. Agnelli G, Bolis G, Capussotti L, et al: Aclinical outcome-based prospective studyon venous thromboembolism after cancersurgery: The @RISTOS project. Ann Surg2006; 243:89–95

154b. Jaffer AK, Barsoum WK, Krebs V, et al:Duration of anesthesia and venous throm-boembolism after hip and knee arthro-plasty. Mayo Clin Proc 2005; 80:732–738

155b. White RH, Gettner S, Newman JM, et al:Predictors of rehospitalization for symp-tomatic venous thromboembolism after to-tal hip arthroplasty. N Engl J Med 2000;343:1758–1764

156b. White RH, Zhou H, Romano PS: Incidence ofsymptomatic venous thromboembolism afterdifferent elective or urgent surgical proce-dures. Thromb Haemost 2003; 90:446–455

157b. Haley RW, Culver DH, White JW, et al: Theefficacy of infection surveillance and con-trol programs in preventing nosocomial in-fections in US hospitals. Am J Epidemiol1985; 121:182–205

158b. Haley RW: Measuring the costs of nosoco-mial infections: Methods for estimatingeconomic burden on the hospital. Am J Med1991; 91(3B):32S–38S

159c. Brook I, Frazier EH: Aerobic and anaerobic

bacteriology of wounds and cutaneous ab-scesses. Arch Surg 1990; 125:1445–1451

160c. File TM Jr, Tan JS: Treatment of skin andsoft-tissue infections. Am J Surg 1995;169(5A Suppl):27S–33S

161c. Meislin HW, Lerner SA, Graves MH, et al:Cutaneous abscesses: Anaerobic and aerobicbacteriology and outpatient management.Ann Intern Med 1977; 87:145–149

162b. Macfie J, Harvey J: The treatment of acutesuperficial abscesses: A prospective clinicaltrial. Br J Surg 1977; 64:264–266

163b. Huizinga WK, Kritzinger NA, Bhamjee A:The value of adjuvant systemic antibiotictherapy in localised wound infectionsamong hospital patients: A comparativestudy. J Infect 1986; 13:11–16

164b. Commichau C, Scarmeas N, Mayer SA: Riskfactors for fever in the neurologic intensivecare unit. Neurology 2003; 60:837–841

165c. Jackson WL Jr, Shorr AF: The yield of lum-bar puncture to exclude nosocomial men-ingitis as aetiology for mental statuschanges in the medical intensive care unit.Anaesth Intensive Care 2006; 34:21–24

166c. Gower DJ, Baker AL, Bell WO, et al: Con-traindications to lumbar puncture as de-fined by computed cranial tomography.J Neurol Neurosurg Psychiatry 1987; 50:1071–1074

167c. Schliamser SE, Backman K, Norrby SR: In-tracranial abscesses in adults: an analysis of54 consecutive cases. Scand J Infect Dis1988; 20:1–9

168b. Spanos A, Harrell FE Jr, Durack DT: Differ-ential diagnosis of acute meningitis: Ananalysis of the predictive value of initialobservations. JAMA 1989; 262:2700–2707

169b. Hayward RA, Shapiro MF, Oye RK: Labora-tory testing on cerebrospinal fluid: A reap-praisal. Lancet 1987; 1:1–4

170b. Leib SL, Boscacci R, Gratzl O, et al: Predic-tive value of cerebrospinal fluid (CSF) lac-tate level versus CSF/blood glucose ratio forthe diagnosis of bacterial meningitis follow-ing neurosurgery. Clin Infect Dis 1999; 29:69–74

171c. Tunkel AR, Hartman BJ, Kaplan SL, et al:Practice guidelines for the management ofbacterial meningitis. Clin Infect Dis 2004;39:1267–1284

172c. Adler SP: Transfusion-associated cytomega-lovirus infections. Rev Infect Dis 1983;5:977–993

173c. Drew WL, Miner RC: Transfusion-related cyto-megalovirus infection following noncardiac sur-gery. JAMA 1982; 247:2389–2391

174c. Lerner PI, Sampliner JE: Transfusion-associated cytomegalovirus mononucleosis.Ann Surg 1977; 185:406–410

175b. Ugarte H, Silva E, Mercan D, et al: Procal-citonin used as a marker of infection in theintensive care unit. Crit Care Med 1999;27:498–504

176b. Suprin E, Camus C, Gacouin A, et al: Pro-calcitonin: A valuable indicator of infectionin a medical ICU? Intensive Care Med 2000;26:1232–1238

177b. Selberg O, Hecker H, Martin M, et al: Dis-crimination of sepsis and systemic inflam-matory response syndrome by determina-tion of circulating plasma concentrations ofprocalcitonin, protein complement 3a, andinterleukin-6. Crit Care Med 2000; 28:2793–2798

178b. Rothenburger M, Markewitz A, Lenz T, et al:Detection of acute phase response and in-fection: The role of procalcitonin and C-re-active protein. Clin Chem Lab Med 1999;37:275–279

179b. Muller B, Becker KL, Schachinger H, et al:Calcitonin precursors are reliable markersof sepsis in a medical intensive care unit.Crit Care Med 2000; 28:977–983

180b. Aouifi A, Piriou V, Bastien O, et al: Useful-ness of procalcitonin for diagnosis of infec-tion in cardiac surgical patients. Crit CareMed 2000; 28:3171–3176

181c. Cohen J: The detection and interpretationof endotoxaemia. Intensive Care Med 2000;26(Suppl 1):S51–6

182c. Cohen J, McConnell JS: Observations on themeasurement and evaluation of endotox-emia by a quantitative limulus lysate mi-croassay. J Infect Dis 1984; 150:916–924

183b. Marshall JC, Walker PM, Foster DM, et al:Measurement of endotoxin activity in criti-cally ill patients using whole blood neutro-phil dependent chemiluminescence. CritCare 2002; 6:342–348

184b. Marshall JC, Foster D, Vincent JL, et al:Diagnostic and prognostic implications ofendotoxemia in critical illness: Results ofthe MEDIC study. J Infect Dis 2004; 190:527–534

185c. Lipsky BA, Hirschmann JV: Drug fever.JAMA 1981; 245:851–854

186d. Mackowiak PA: Drug fever: Mechanisms,maxims and misconceptions. Am J Med Sci1987; 294:275–286

187c. Mackowiak PA, LeMaistre CF: Drug fever: Acritical appraisal of conventional concepts.An analysis of 51 episodes in two Dallashospitals and 97 episodes reported in theEnglish literature. Ann Intern Med 1987;106:728–733

188c. Mermel LA: Bacteriology, safety and preven-tion of infection associated with continuousintravenous infusions. Blood Coagul Fibri-nolysis 1996; 7(Suppl 1):S45–S51

189c. Cunha BA: Drug fever: The importance ofrecognition. Postgrad Med 1986; 80:123–129

190c. Heiman-Patterson TD: Neuroleptic malig-nant syndrome and malignant hyperther-mia: Important issues for the medical con-sultant. Med Clin North Am 1993; 77:477–492

191c. Nimmo SM, Kennedy BW, Tullett WM, et al:Drug-induced hyperthermia. Anaesthesia1993; 48:892–895

192c. Caroff SN, Mann SC: Neuroleptic malignantsyndrome and malignant hyperthermia. An-aesth Intensive Care 1993; 21:477–478


193c. Mason PJ, Morris VA, Balcezak TJ: Seroto-nin syndrome: Presentation of 2 cases andreview of the literature. Medicine (Balti-more) 2000; 79:201–209

194c. Lawrence KR, Adra M, Gillman PK: Seroto-nin toxicity associated with the use of lin-ezolid: A review of postmarketing data. ClinInfect Dis 2006; 42:1578–1583

195b. Garnacho-Montero J, Garcia-Garmendia JL,Barrero-Almodovar A, et al: Impact of ade-quate empirical antibiotic therapy on theoutcome of patients admitted to the inten-sive care unit with sepsis. Crit Care Med2003; 31:2742–2751

196b. Garnacho-Montero J, Aldabo-Pallas T,Garnacho-Montero C, et al: Timing of ade-

quate antibiotic therapy is a greater deter-minant of outcome than are TNF and IL-10polymorphisms in patients with sepsis. CritCare 2006; 10:R111

197b. Ibrahim EH, Sherman G, Ward S, et al: Theinfluence of inadequate antimicrobial treat-ment of bloodstream infections on patientoutcomes in the ICU setting. Chest 2000;118:146–155

198b. Harbarth S, Garbino J, Pugin J, et al: Inap-propriate initial antimicrobial therapy andits effect on survival in a clinical trial ofimmunomodulating therapy for severe sep-sis. Am J Med 2003; 115:529–535

199b. Kollef MH, Sherman G, Ward S, et al: Inade-quate antimicrobial treatment of infections: A

risk factor for hospital mortality among criti-cally ill patients. Chest 1999; 115:462–474

200b. Micek ST, Lloyd AE, Ritchie DJ, et al:Pseudomonas aeruginosa bloodstream in-fection: Importance of appropriate initialantimicrobial treatment. AntimicrobAgents Chemother 2005; 49:1306–1311

201b. Kumar A, Roberts D, Wood KE, et al: Du-ration of hypotension before initiation ofeffective antimicrobial therapy is the criti-cal determinant of survival in human septicshock. Crit Care Med 2006; 34:1589–1596

202b. Ibrahim EH, Ward S, Sherman G, et al:Experience with a clinical guideline for thetreatment of ventilator-associated pneumo-nia. Crit Care Med 2001; 29:1109–1115


ICU - Fever Critically Ill - Guidline 2008

Documents

Transcript of ICU - Fever Critically Ill - Guidline 2008