h1 Fuo Aim03

A Comprehensive Evidence-Based Approachto Fever of Unknown OriginOphyr Mourad, MD, FRCPC; Valerie Palda, MD, MSc; Allan S. Detsky, MD, PhD

Background: Fever of unknown origin (FUO) is de-fined as a temperature higher than 38.3C on several oc-casions and lasting longer than 3 weeks, with a diagno-sis that remains uncertain after 1 week of investigation.

Methods: A systematic review was performed to de-velop evidence-based recommendations for the diagnos-tic workup of FUO. MEDLINE database was searched(January 1966 to December 2000) to identify articles re-lated to FUO. Articles were included if the patient popu-lation met the criteria for FUO and they addressed thenatural history, prognosis, or spectrumof disease or evalu-ated a diagnostic test in FUO. The quality of retrievedarticles was rated as good, fair, or poor, and sen-sitivity, specificity, and diagnostic yield of tests were cal-culated. Recommendationsweremade in accordancewiththe strength of evidence.

Results: The prevalence of FUO in hospitalized patientsis reported tobe2.9%.Eleven studies indicate that the spec-trum of disease includes no diagnosis (19%), infec-tions (28%), inflammatory diseases (21%), andmalignan-cies (17%). Deep vein thrombosis (3%) and temporal

arteritis in the elderly (16%-17%)were important consid-erations. Four good natural history studies indicate thatmost patients with undiagnosed FUO recover spontane-ously (51%-100%).One fair-quality study suggested ahighspecificity (99%) for the diagnosis of endocarditis in FUOby applying the Duke criteria. One fair-quality studyshowed that computed tomographic scanning of the ab-domenhadadiagnostic yieldof 19%.Ten studies of nuclearimaging revealed that technetium was the most promis-ing isotope, showing a high specificity (94%), albeit lowsensitivity (40%-75%) (2 fair-quality studies). Two fair-quality studies showed liver biopsy to have a high diag-nostic yield (14%-17%), but with risk of harm (0.009%-0.12% death). Empiric bone marrow cultures showed alow diagnostic yield of 0% to 2% (2 fair-quality articles).

Conclusions: Diagnosis of FUO may be assisted by theDuke criteria for endocarditis, computed tomographicscan of the abdomen, nuclear scanningwith a technetium-based isotope, and liver biopsy (fair to good evidence).Routine bone marrow cultures are not recommended.

Arch Intern Med. 2003;163:545-551

F EVER OF unknown origin(FUO) identifies a syndromeof fever that does not resolvespontaneously, in which thecause remains elusive after anextensivediagnosticworkup.Petersdorf andBeeson1 first coined the term fever of un-known origin in 1961 and explicitly de-fined it as a temperature higher than38.3Con several occasions and lasting longer than3 weeks, with a diagnosis that remains un-certain after 1weekof investigations inhos-pital. Petersdorf and Beeson chose 3weeksof fever to eliminate self-limited viral ill-nesses and to allow sufficient time for ap-propriate initial investigations to be com-pleted. Over the past 40 years, health carehas shifted from the inpatient to the ambu-latory setting. As a result, it has now be-come widely accepted that the require-ment for a 1-week evaluation in hospitalbemodified so that evaluationsmaynowbecompleted in an outpatient setting.2

Fever of unknown origin is frustrat-ing for patients and physicians because thediagnostic workup often involves numer-ous noninvasive and invasive proceduresthat sometimes fail to explain the fever.There are well over 200 different reportedcauses of FUO.3,4 To date, there are no pub-lished guidelines or evidence-based rec-ommendations for the diagnostic workupof FUO. The body of literature that dis-cusses FUO comprises case series and co-hort studies. In FUO, there is no diagnos-tic gold standard against which otherdiagnostic testsmay bemeasured. Final di-agnoses are determined in a number ofways, includingnatural history, biopsy, sur-gery, postmortem examinations as well asother imaging techniques. For these rea-sons, there is disagreement as to whatshould constitute a comprehensive diag-nostic workup.

To have a structured, sensible, and ef-fective approach, the clinician must have

REVIEW ARTICLE

From the Departments ofMedicine (Drs Mourad, Palda,and Detsky) and Health PolicyManagement and Evaluation(Drs Palda and Detsky),University of Toronto; and theDivision of General InternalMedicine, St Michaels Hospital(Drs Mourad and Palda) andMt Sinai Hospital andUniversity Health Network(Dr Detsky), Toronto, Ontario.

(REPRINTED) ARCH INTERN MED/VOL 163, MAR 10, 2003 WWW.ARCHINTERNMED.COM545

2003 American Medical Association. All rights reserved. at Ucla, on June 21, 2006 www.archinternmed.comDownloaded from

an understanding of the spectrumofdisease and the test characteristics ofthe various diagnostic modalitiesavailable in the evaluation of FUO.A rational approach should also bebased on the relative frequencies ofthe different causes and their impor-tance to the health of the patient. Forthe purpose of this article, FUO is notintended to encompass those indi-viduals with impaired immunity orunexplained fevers in children. Fe-ver of unknown origin in patientswith human immunodeficiency vi-rus infection, patients with knownmalignancy, and children have a dif-ferent diagnostic differential andwillnot be addressed in this article.

METHODS

MEDLINE database was searched toidentify articles related to FUO. Thesearch included English-language ar-ticles published between January 1966to December 2000 using the MedicalSubject Heading fever of unknown ori-gin and the text words FUO, PUO, andpyrexia of unknown origin. Articles wereincluded if the patient population wasclearly defined and met the criteria setforth by Petersdorf and Beeson1 for FUOand if they addressed the natural his-tory, prognosis, or spectrum of diseaseor evaluated a diagnostic test in FUO.Articles were excluded if they focusedon immunosuppressed patients, thoseyounger than 18 years, and patientswithhuman immunodeficiency virus infec-tion or cancer. To identify a group of pa-tients similar to our own, only patientpopulations fromNorth America,West-ern Europe, and Scandinavia were in-cluded.ACochrane review failed to iden-tify any relevant articles. References ofselected articles were reviewed to iden-tify further relevant articles

We define the diagnostic yield asthe number of patients with positive testresults divided by the number of alltested patients. The absolute value of thediagnostic yield should not be viewed in-dependently, but rather togetherwith in-formation about the ability of the test toidentify serious and potentially curabledisease and all clinically important toxiceffects of the diagnostic test.

HARMS

Adverse effects of diagnostic tests wereextracted from each individual studyand from a separate literature review.MEDLINE database was searched for ar-ticles that identified complications andadverse effects of invasivediagnostic tests.

SYSTEMATIC PROCESSUSED TO ARRIVE AT

FINAL RECOMMENDATIONS

Articles thatmet theselectioncriteriaweresummarized in tabular format. Criteriawere developed to assess methodologi-cal quality for diagnostic tests and natu-ral history studies based on publishedmethods of the US Preventive ServicesTask Force.5 The evidence was system-atically reviewed by assigning a qualityrating to each article according to aprioricriteria. While the importance of re-search design remains the main basisby which to assess strength of evidence,not all studies within a research designhave equal internal validity. To moreclearly assess the internal validity of in-dividual studies within research de-signs, design-specific criteria were usedthat allow rating of studies into 3 inter-nal validity categories: good, fair, andpoor. Thus all individual studies re-ceive 2 codes: 1 for research design and1 (good, fair, poor) for internal validitywithin its design.

The body of evidence available foreach topicwas then synthesized, and rec-ommendations were made based on thefollowing considerations: publishedprevalence of disease, performance char-acteristics of the test (diagnostic yield,sensitivity, specificity, positive and nega-tive likelihood ratios), harms of the test,strength of the evidence supporting theuse of the test (study design and qual-ity rating), and harms of the diagnostictest. For example, elements likely to re-sult in a recommendation to perform thetest would be good performance char-acteristics and no harms, even in thepresence of limited evidence, or a testwith moderate performance character-istics but multiple fair-quality studiesdemonstrating somebenefit. Tests aimedat detecting commondisorderswere alsomore likely to be recommended. Finalrecommendations used language de-fined by The Canadian Task Force onPreventive Health Care, which wasamended to apply to a diagnostic test.

Summary tables of the natural his-tory, prognosis, and diagnostic studiesas well as appendixes that describe cri-teria developed for study selection andassessment of the methodological qual-ity of diagnostic tests and natural his-tory studies are available on request fromthe authors.

PREVALENCE AND SPECTRUMOF DISEASE

Iikuni et al6 documented that of 5245patients admitted to the Department ofInternal Medicine between 1982 and1992 at Kitasato University Hospital,

Kanagawa, Japan, 153 (2.9%) had FUO.Kazanjian7 reported that of 6250 infec-tious disease consults performed at 3community hospitals in Rhode Islandbetween 1984 and 1990, 86 met thecriteria for FUO (ie, 1 FUO in every 73consults requested). Because FUO en-compasses a wide spectrum of both in-fectious and noninfectious diseases, webelieve that a significant proportion ofpatientswill be investigatedbygeneral in-ternists, with subspecialist (eg, infec-tious diseases, oncology, or rheumatol-ogy) consulting thereafter.

It is important to understand thespectrum of disease before addressingthe utility of the diagnostic tools in theevaluation of FUO. The causes of FUOhave traditionally been grouped into1 of 4 categories: infectious, malignant,inflammatory, and undetermined.

There are 11 series that includeover1000 patients that have reported the di-agnostic entities that constituteFUO.1,7-16

Grouping all the patients collected from1952 until 1994 reveals that the spec-trum of disease includes infections in28% and inflammatory diseases in 21%.Malignancies account for a smaller pro-portion (17%). A cause is never identi-fied in a significant proportion (19%) ofpatients.

The spectrum of disease has alsochangedconsiderably fromthe timeof thefirst prospectively collected series of 100patients. Over the past 40 years, the pro-portion of cases of FUO caused by infec-tions and neoplasms has decreased. Theeasy detection of solid tumors and ab-normal lymph nodes via ultrasonogra-phy and computed tomography (CT)hasresulted in thedeclineof tumors as a com-mon cause of FUO, with the conse-quence that malignancies are less likelyto present with prolonged undiagnosedfever. Patients with undiagnosed FUOused to make up the smallest propor-tion. At present, the largest proportion ofpatients who present with FUO neverhave a cause identified (Figure 1).

The most common infectiouscauses documented in the literature aretuberculosis and intra-abdominal ab-scesses.1,7,9,10,13,14 Themost commonma-lignancies are Hodgkin disease and non-Hodgkin lymphoma.1,7,9,10,13,14 Temporalarteritis accounts for 16% to 17% of allcauses of FUO in the elderly.17,18

Outcomes of patients with FUO isa functionof theunderlying cause.1,7,10,12,13

Overall, 12% to 35% of patients will diefromFUO-related causes1,7,10,12,13; 52% to100%of patientswith a final diagnosis ofmalignancywill die within 5 years of thediagnosis.1,7,10,12 Mortality ismuch lowerif an infection is identified as the causeof FUO (8%-22%).1,7,10,12 Therefore, thebest predictor of survival is disease cat-



egory, with malignancy incurring thehighest mortality. The prognosis of pa-tients with FUO in whom a cause can-not be identified is excellent.1,6,10,13,19Mostof these patients have a spontaneous re-covery (51%-100%), andonly a small pro-portion have persistent fever (0%-30%).

RESULTS

LIMITATIONS OFTHE LITERATURE ON FUO

The body of literature that dis-cusses FUO comprises case or co-hort studies. There are no random-ized controlled trials in the FUOliterature. Most of these patientswere identified in tertiary care cen-ters; however, a number of studiesreport their experience from com-munity hospitals.

LIMITATIONS OFTHE LITERATURE ON

DIAGNOSTIC TESTS FOR FUO

In FUO, there is no diagnostic goldstandard against which other diag-nostic tests may be measured. Fi-nal diagnoses are determined in anumber of ways, including naturalhistory, biopsy, surgery, and post-mortem examinations as well asother imaging techniques. The di-agnostic tests being assessed havebeen performed at various stages ofthe investigation. The definitions oftrue positives, false positives, truenegatives, and false negatives varyfrom study to study. Therefore, cal-culation and significance of sensi-tivity, specificity, positive predic-tive value, negative predictive value,and likelihood ratios should beviewed with caution.

INITIAL INVESTIGATIONS

Although there is no substitute fora thoroughhistory review andphysi-cal examination, the yield from acomplete history review and me-ticulous physical examination is notknown, since it has never been stud-ied. Review articles and articlesevaluating a diagnostic test in FUOstate explicitly that a certain num-ber of investigations must be com-pleted for a case to qualify as FUO.These have varied over the years, andwe have compiled the following list

of minimum diagnostic evalua-tions based on reviewing all of theliterature (Table 1).

One of the first steps thatshould be undertaken is to confirmthat a true fever exists. Patientsshould be instructed to record andmeasure their temperature daily. Thefever pattern adds little to the diag-nostic workup.20 Also, all medica-tions should, if possible, be discon-tinued early in the evaluation to ruleout a drug-induced fever. Persis-tence of fever beyond 72 hours af-ter the suspected drug has been re-moved allows one to conclude thatthe drug is not the offending agentin producing the fever.21

RECOMMENDEDDIAGNOSTIC TESTS FORWHICH EVIDENCE EXISTS

Abdominal CT

ACT of the abdomen should be oneof the first investigations in FUO,

since it has ahighdiagnostic yield andis likely to identify 2 of themost com-mon causes of FUO: intra-abdomi-nal abscesses and lymphoprolifera-tive disorders. A retrospective caseseries of an abdominal CT in theworkup of FUO reported a diagnos-tic yield of 19%.22 Clinical fol-low-up in 32 of the 47 cases inwhichtheCT scan of the abdomenwas nor-mal identified only 1 patient with anintra-abdominal pathologic cause(lymphoma).

Nuclear Imaging

Ten studies of fair methodologicalquality have assessed the test char-acteristics of nuclear imaging stud-ies inFUO.23-32Technetium(99m-TcBW 250/183)based studies reportthe highest specificity (93%-94%)but are insensitive (40%-75%).24,25

Indium 111 IgG and indium 111labeled white blood cell scans havepoor sensitivity (45%-82%) and aspecificity that ranges from 69% to

Infection Malignancy Inflammatory Other No Diagnosis

40

30

35

25

20

15

10

5

0

Patie

nts,

%

24.5 14.5 23.5 7.5 301990s

29 15.6 25.4 12.7 17.31980s

30.8 23.9 15.1 13 17.21970s

36 19 18 18 91950s

Figure 1. The percentage of patients with fever of unknown origin by cause over the past 40 years.

Table 1. Minimal Diagnostic Workup to Qualify as Fever of Unknown Origin

Comprehensive historyPhysical examinationComplete blood cell count + differentialBlood film reviewed by hematopathologistRoutine blood chemistry (including lactic dehydrogenase, bilirubin, and liver enzymes)Urinalysis and microscopyBlood (3) and urine culturesAntinuclear antibodies, rheumatoid factorHuman immunodeficiency virus antibodyCytomegalovirus IgM antibodies; heterophil antibody test (if consistent with mononucleosis-like

syndrome)Q-fever serology (if exposure risk factors exist)Chest radiographyHepatitis serology (if abnormal liver enzyme test result)



86%.26-30 Gallium 67 nuclear scan-ning is less well studied.23,31,32 Thebest quality study of gallium scan-ning reportedasensitivityof67%anda specificity of 78% but only in-cluded 20 patients.23 Fludeoxyglu-cose F 18 is a promising new alter-native tracer that accumulates inboth malignant tumors and at sitesof inflammation. One recent smallfair-quality study reported a sensi-tivity and specificity of 84%and86%respectively.23 Fludeoxyglucose F18based scans hold promise, butfurther studies are required to vali-date its utility.

Technetium (99m-Tc BW250/183)based scans are thereforemostlikely to be diagnostically helpful(positive likelihood ratio, 5.7-12.5) because of their high specific-ity (93%-94%). The other tests havebeen shown to be either poorly dis-criminating (gallium 67) or incon-clusive because the studies were ofpoor overall quality.

The only potential toxic effectrelated to imaging studies such asCTand nuclear studies appears to be ra-diation exposure. The levels of ra-diation involved in nuclear medi-cine studies are usually considerablylower than a patient would receivein a conventional radiographic studyor CT scan. Owing to its minimaltoxicity and overall good test char-acteristics, nuclear imaging studiesare helpful in localizing a potentialinfectious or inflammatory focus.Technetium should be the tracer ofchoice.

The Duke Criteria

Infective endocarditis is an impor-tant cause of FUO and accounts for1% to 5% of all causes.1,7,8,9,12-14 TheDuke criteria have a very high speci-ficity in patients with FUO (99%;95% confidence interval, 97%-100%), and thus should be used toidentify patients with suspected in-fectiveendocarditis.33Thedesignandretrospectivenature of the study thatassessed the utility of the Duke cri-teria in identifying thosewith infec-tive endocarditismayhavebiased theresults toward a higher specificity.Sensitivity data aremore difficult todetermine from the literature. Thesame authors determined that in 27patients without FUO in whom the

diagnosis of infective endocarditiswas histologically and/or bacterio-logically confirmed, the sensitivityof the Duke criteria was 82%.34

Liver Biopsy

The diagnostic yield from liver bi-opsy is 14% to 17%.35,36 Physical ex-amination findings of hepato-megaly or abnormal liver profile arenot helpful in predicting which pa-tients will have an abnormal liver bi-opsy result. In patientswithout FUO,complications from liver biopsies arereported in 0.06% to 0.32%.37-40

Death as a direct result of the liverbiopsy occurs in 0.009% to 0.12%.We believe that the benefits of a liverbiopsy outweigh what we considerare minimal risks.

Temporal Artery Biopsy

There is no single large series com-posed solely of elderly patients withFUO. Two studies (Esposito andGleckman17 and Knockaert et al18)identified temporal arteritis as thecause of FUO in 16% and 17%, re-spectively.17,18 A decision analysis inthe management of suspected giantcell arteritis concluded that a bi-opsy and treat positive cases is thepreferred strategy when the likeli-hood of disease is intermediate.41

Temporal artery biopsy is a safe sur-gical procedure42-44 with rare com-plications including damage of thefacial nerve,45 skin necrosis,46 anddrooping of the eyebrow.47 Color du-plex ultrasonography of the tempo-ral arteries may be a helpful alter-native to temporal artery biopsy inthe diagnosis of temporal arteritis,with a reported sensitivity and speci-ficity of 93% when a halo, stenosis,or occlusion is identified.48 Tempo-ral arteritis accounts for a large pro-portion of causes of FUO in the el-derly, and thus a temporal arterybiopsy should be performed in el-derly patientswith unresolved FUO.

Leg Doppler Imaging

Venous thrombosis canpresentwithprolonged fever.Three series6,10,49 re-ported a deep vein thrombosis as thecauseofFUOin2%to6%ofpatients.Although deep vein thrombosis ac-countsforasmallpercentageofcauses

of FUO, leg Doppler imaging is safeand may identify a treatable cause.

DIAGNOSTIC TEST FORWHICH EVIDENCE EXISTSTO RECOMMEND AGAINST:BONE MARROW CULTURES

The diagnostic yield of bone mar-row cultures in immunocompetentindividuals was found to be 0% to2%.50,51 Owing to the very low diag-nostic yield from bone marrow cul-tures in FUO, bone marrow cul-tures are not recommended in thediagnostic workup. Physiciansmustuse their discretion in determiningwhether there are other indicationsto perform a bone marrow biopsy.

AREAS OF UNCERTAINTY

Surgical Explorationof the Abdomen

All of the studies reporting the diag-nostic yield of exploratory lapa-rotomy in FUO are of poor method-ological quality. Most of the studieswereperformed in thepre-CTera,52-57

whereas only 1 study examined therole of surgery in the post-CT era.58

In that study,CTof the abdomenwasperformed in 14 of 25 patients, and10 had abnormal findings on CT(hepatomegaly, splenomegaly, and/orretroperitoneal nodes). The diagnos-tic yield in those who had a normalCT and thosewho did not have aCTwas not reported. Themortality ratewas 4%,with 12%experiencingpost-operative complications.

The diagnostic yield of lapa-roscopy was evaluated in 1 study inthe pre-CT era and determined to be44% with no mortality and mini-mal morbidity reported.59 Liver bi-opsy was performed in 63 of 70 ofthese patients at laparoscopy, and itis not clear what proportion of fi-nal diagnoseswere contributed to bythe liver biopsy results alone. Therole of surgery in the post-CT era re-mains unclear.

Empiric Therapy

The utility of empiric therapy, suchas antibiotics, antituberculosisagents, or corticosteroids has notbeen studied in FUO. This, how-ever, is not an uncommon practice



for the frustrated physician.We be-lieve that empiric therapy should notbe given to patients with FUO be-cause it often obscures or confusesthe diagnosis.

COMMONLY PERFORMEDDIAGNOSTIC TESTS FOR

WHICH NO EVIDENCE EXISTS

There is no literature assessing theutility of erythrocyte sedimenta-tion rate, C-reactive protein, mag-netic resonance imaging, bone scan,and echocardiography in FUO.

Transthoracic echocardiogra-phy (sensitivity, 63%; specificity,98%) and transesophageal echocar-diography (sensitivity, 100%; speci-ficity, 98%) may allow for early de-tection of vegetations on valves andmay help to identify infective endo-carditis.60 Transesophageal echocar-diography is important in the diag-nosis of culture-negative endocarditisand performs better than transtho-racic echocardiography.61 TheDukecriteria have incorporated echocar-diography as an important tool in thediagnosis of endocarditis. It thusseems reasonable to include echo-cardiography in the diagnosticworkup of FUO.

It is important to appreciatethat there is no evidence to supportor refute the utility of diagnostic testssuch as echocardiography, mag-netic resonance imaging, bone scan,and D-dimer assay in patients withFUO. Their potential utility may beextrapolated from the non-FUOliterature.

PROPOSED ALGORITHM

The proposed algorithm (Figure2)was derived by taking into accountthe spectrum of disease, the clinicalimportanceof the various causes, andthe test characteristics of the vari-ous diagnosticmodalities available inthe evaluation of FUO. The proce-dures that were least invasive andthose that reported the highest diag-nostic yield appear early in the algo-rithm.Risks and complications of thevarious procedures were also takeninto account. The algorithmwas notderived through a formal process.

The proposed algorithm needsto be evaluated prospectively beforeits validity can be ascertained. Infor-

mationobtained fromathoroughhis-tory review, repeatedphysical exami-nations, and initial laboratory studiesmaydirect the physician to tests thatdonot conformto thealgorithm.Thealgorithm is meant only as a frame-work,withnecessaryadjustmentsandprovisions made according to pre-test probability. The frameworkwasderived fromconsiderable evidence;however, one should not neglect theimpact of the art of medicine andclinical experience on pretest prob-abilities, thus allowing for devia-tions from the proposed algorithm.

CONCLUSIONS

The diagnostic workup of FUO re-mains complex; however, consider-able evidence exists to guide em-piric testing. Historically, thespectrum of disease includes no di-agnosis (19%), infection (28%), in-flammatory diseases (21%), andma-lignancies (17%), with deep veinthrombosis (3%) and temporal arte-ritis in the elderly (16%-17%) beingimportant considerations. The diag-

nostic workup should begin with athorough history review and physi-cal examination. Routine noninva-sive investigations (Table 1) are rec-ommended in all patients prior toidentifying a patient as having FUO.The Duke criteria have a very highspecificity (99%) in patients withFUO and suspected infective endo-carditis, and thus should be used toidentify endocarditis as the cause ofFUO. When the initial investiga-tions are not helpful in identifying acause, the clinician should then pro-ceed to imaging. These should in-clude a CT of the abdomen and atechnetium-basednuclear scan.ACTof the abdomen has a high diagnos-tic yield (19%) and carries a low risk.Two fair-quality studies show thattechnetium-based scans have a highspecificity but are insensitive. LegDoppler imaging should be consid-ered the next step in identifying deepvein thrombosis as a potential revers-ible and easily treatable cause. A tem-poral artery biopsy should be con-sidered in elderly patientswith FUO.There is fair evidence to suggest that

Initial Evaluation

Abdominal CT

Laparoscopy

Liver Biopsy

TA Biopsy

Obtain Tissueto ConfirmDiagnosis

Use Duke Criteria toRule In or Rule Out IE

Leg Doppler Imaging

Follow Clinically

Tc-Based Nuclear Scan

Fever Chartto Document Fever

Discontinue All NonessentialMedications

Undiagnosed FUO

Focus Identified

Fever Persists

Positivefor DVT

No

No

Yes

Yes

Yes

No

Drug Fever

LMWH

FUO

Infective EndocarditisSuspected?

>Age 50 y

Clinical StatusDeteriorating?

Fever Persists >3 wk

Persistence of Fever >72 h

Fever Resolvesin

a liver biopsy has a high diagnosticyield with minimal toxicity. Bonemarrowcultures are of lowyield (0%-2%) and arenot recommended in im-munocompetent patients with FUO.The literature evaluating the role oflaparoscopy and laparotomy is over-all poor, and the risk of these inter-ventions is high. These surgical pro-cedures may be considered forindividuals who have persistent fe-ver andadeterioratingclinical course.

The role of erythrocyte sedi-mentation rate, C-reactive protein,magnetic resonance imaging, bonescan, and echocardiography has notbeen determined. The clinicianmustuse judgment to determine whetherthese investigations would be ofvalue for each individual patient(Table 2)

The prognosis of FUO is de-pendent on the etiological cat-egory. Undiagnosed FUOhas a veryfavorable outcome. Patients inwhomthe above diagnostic investigationsfail to identify a cause should be fol-lowed clinically with serial historyreviews and physical examinationsuntil the fever resolves or new di-agnostic clues are found.

Accepted for publication June 19, 2002.Nofinancialsupportwasrequested

or rendered for this body of work.We thank David McNeely, MD,

and Hillar Vellend, MD, from the Di-visions of Infectious Disease and Gen-eral Internal Medicine, UniversityHealthNetwork, Toronto, Ontario, forreviewing the manuscript and provid-ing valuable feedback.

Corresponding author: OphyrMourad, MD, Division of GeneralInternal Medicine, St Michaels Hos-pital, Room 4-140, Cardinal CarterWing, 30 Bond St, Toronto, Ontario,C a n a d a M5B 1W8 ( e -m a i l :[email protected]).

REFERENCES

1. Petersdorf RG, Beeson P. Fever of unexplained ori-gin: report on 100 cases. Medicine (Baltimore).1961;40:1-30.

2. PetersdorfRG.Feverofunknownorigin: anold friendrevisited. Arch Intern Med. 1992;152:21-22.

3. Arnow P, Flaherty J. Fever of unknown origin. Lan-cet. 1997;350:575-580.

4. Hirschmann JV. Fever of unknown origin in adults.Clin Infect Dis. 1997;24:291-300.

5. Harris RP, Helfand M, Woolf SH, et al. Currentmethods of the US Preventive Services Task Force:a review of the process. Am J Prev Med. 2001;3(suppl):21-35.

Table 2. Recommendations for Diagnostic Testing in FUO

Maneuver Effectiveness* Level of Evidence Recommendation

The Duke criteria Specificity = 99% Fair (1 fair-quality study) The Duke criteria has a very high specificity in patientswith FUO and suspected infective endocarditis andthus should be used to identify endocarditis as thecause of FUO. (Recommend)

Abdominal CT Diagnostic yield = 19%Sensitivity = 71%Specifity = 71%

Fair (1 fair-quality study) A CT of the abdomen has a high diagnostic yield and islikely to contribute to identifying the cause of FUO.(Recommend)

Tc 99m BW 250/183nuclear scan

Specificity = 93%-94%Sensitivity = 40%-75%+ve LR = 5.7-12.5

Fair (2 fair-quality studies) Technetium-based scans have a high specificity andpoor sensitivity. Technetium is the tracer of choice inthe evaluation of FUO. (Recommend)

In 111 IgG nuclear scan Specificity = 69%-79%Sensitivity = 47%-82%

Fair (1 fair- and 1poor-quality study)

In 111 IgGbased scans have a poor sensitivity andspecificity and are thus not the nuclear tracer ofchoice. (Recommend against)

In 111labeled WBC scan Specificity = 78%-86%Sensitivity = 45%-60%+ve LR = 2.7-3.2

Fair (1 fair- and 2poor-quality studies)

In 111labeled WBC scan is helpful in identifyingsuspected infectious processes. (Recommend)

Gallium 67 scan Specificity = 70%-78%Sensitivity = 54%-67%

Fair (1 fair- and 1poor-quality study)

Gallium-67based scans have a poor sensitivity andspecificity and are thus not the nuclear tracer ofchoice. (Recommend against)

ESR, C-reactive protein,MRI, echocardiography

. . . . . . There is no evidence to make recommendations for oragainst the use of ESR, C-reactive protein, MRI, bonescan, and echocardiography in the evaluation of FUO.(Insufficient evidence to recommend)

Empiric therapy . . . . . . Empiric therapy with antibiotics, anti-TB agents, orcorticosteroids should not be given to patients withFUO because they often obscure or confuse thediagnosis. (Insufficient evidence to recommend)

Liver biopsy Diagnostic yield = 14%-17% Fair (2 fair-quality studies) Liver biopsy has a high diagnostic yield and minimaltoxicity. (Recommend)

Bone marrow cultures Diagnostic yield = 0%-2% Fair (2 fair-quality studies) Owing to the very low diagnostic yield from bonemarrow cultures in FUO, bone marrow cultures arenot recommended in the diagnostic workup.(Recommend against)

Laparotomy/laparoscopy . . . Poor (8 poor-qualitystudies)

Eight poor-quality studies revealed a high diagnosticyield in the pre-CT era. Surgical exploration of theabdomen is associated with significant morbidity andmortality. There are no studies evaluating the utility ofsurgical exploration in the post-CT era. (Insufficientevidence to recommend)

Abbreviations: CT, computed tomography; ESR, erythrocyte sedimentation rate; FUO, fever of unknown origin; In, indium; MRI, magnetic resonance imaging;TB, tuberculosis; Tc, technetium; WBC, white blood cell.*Diagnostic yield is defined as number of positive tests divided by the number of tests performed; +ve LR is sensitivity/1specificity.



6. Iikuni Y, Okada J, Kondo H, Kashiwazaki S. Cur-rent fever of unknown origin 1982-1992. InternMed. 1994;33:67-73.

7. Kazanjian PH. Fever of unknown origin: review of86 patients treated in community hospitals. ClinInfect Dis. 1992;15:968-973.

8. de Kleijn EM, van der Meer JW. Fever of un-known origin (FUO): report on 53 patients in aDutch university hospital. Neth J Med. 1995;47:54-60.

9. Barbado FJ, Vazquez JJ, Pena JM, Arnalich F, Ortiz-Vazquez J. Pyrexia of unknown origin: changingspectrum of diseases in two consecutive series.Postgrad Med J. 1992;68:884-887.

10. Larson EB, Featherstone HJ, Petersdorf RG. Fe-ver of undetermined origin: diagnosis and followup of 105 cases, 1970-1980. Medicine (Balti-more). 1982;61:269-292.

11. Barbado FJ, Vazquez JJ, Pena JM, et al. Fever ofunknown origin: a survey on 133 patients. JMed.1984;15:185-192.

12. Deal WB. Fever of unknown origin: analysis of 34patients. Postgrad Med. 1971;50:182-188.

13. de Kleijn EM, Vandenbroucke JP, van der MeerJM. Fever of unknown origin (FUO), IA: prospec-tive multicenter study of 167 patients with FUOusing fixed epidemiological criteria.Medicine (Bal-timore). 1997;76:392-400.

14. Knockaert DC, Vanneste LJ, Vanneste SB, Bob-baers HJ. Fever of unknown origin in the 1980s:an update of the diagnostic spectrum. Arch In-tern Med. 1992;152:51-55.

15. Smith JW. Southwestern Internal Medicine Con-ference: fever of undetermined origin: not what itused to be. Am J Med Sci. 1986;292:56-64.

16. Gleckman R, Crowley M, Esposito A. Fever of un-known origin: a view from the community hospi-tal. Am J Med Sci. 1977;274:21-25.

17. Esposito AL, Gleckman RA. Fever of unknown ori-gin in the elderly. J Am Geriatr Soc. 1978;26:498-505.

18. Knockaert DC, Vanneste LJ, Bobbaers HJ. Feverof unknown origin in elderly patients. J Am Geri-atr Soc. 1993;41:1187-1192.

19. Knockaert DC, Dujardin KS, Bobbaers HJ. Long-term follow-up of patients with undiagnosed fe-ver of unknown origin. Arch Intern Med. 1996;156:618-620.

20. Musher D, Fainstein V, Young EJ, Pruett TL. Fe-ver patterns: their lack of clinical significance.ArchIntern Med. 1979;139:1225-1228.

21. Johnson DH, Cunha BA. Drug fever. Infect Dis ClinNorth Am. 1996;10:85-91.

22. Quinn MJ, Sheedy PF, Stephens DH, Haltery RR.Computed tomography of the abdomen in evalu-ation of patients with fever of unknown origin.Ra-diology. 1980;136:407-411.

23. Meller J, Altenvoerde G, Munzel U, et al. Fever ofunknown origin: prospective comparison of[18F]FDG imaging with a double-head coinci-dence camera and gallium-67 citrate SPET. EurJ Nucl Med. 2000;27:1617-1625.

24. Meller J, Ivancevic V, Conrad M, Gratz S, MunzDL, Becker W. Clinical value of immunoscintig-raphy in patients with fever of unknown origin.J Nucl Med. 1998;39:1248-1253.

25. Becker W, Dolkemeyer U, Gramatzki M, SchneiderMU, Scheele J, Wolf F. Use of immunoscintigra-phy in the diagnosis of fever of unknown origin.Eur J Nucl Med. 1993;20:1078-1083.

26. de Kleijn EM, Oyen W, Corstens F, van der MeerJ. Utility of indium-111-labeled polyclonal immu-noglobulin G scintigraphy in fever of unknown ori-gin. J Nucl Med. 1997;38:484-489.

27. de Kleijn EM, Oyen W, Claessens R, Corstens F,van der Meer J. Utility of scintigraphic methodsin patients with fever of unknown origin. Arch In-tern Med. 1995;155:1989-1994.

28. MacSweeney JE, Peters AM, Lavender JP. In-dium labeled leukocyte scanning in pyrexia of un-known origin. Clin Radiol. 1990;42:414-417.

29. Davies SG, Garvie NW. The role of indium-labeled leukocyte imaging in pyrexia of unknownorigin. Br J Radiol. 1990;63:850-854.

30. Schmidt KG, Rasmussen JW, Sorensen PG, Wede-bye IM. Indium-111-granulocyte scintigraphy in theevaluation of patients with fever of undeterminedorigin. Scand J Infect Dis. 1987;19:339-345.

31. Knockaert DC, Mortelmans A, De Roo MC. Clini-cal value of gallium-67 scintigraphy in evalua-tion of fever of unknown origin. Clin Infect Dis.1994;18:601-605.

32. Knockaert DC, Mortelmans A, De Roo MC, Bob-baers HJ. Clinical value of gallium scintigraphy inthe investigation of fever or inflammation of un-known origin in the ultrasound and computed to-mography era. Acta Clinica Belg. 1989;44:91-98.

33. Hoen B, Beguinot I, Rabaud RJ, Selton-Suty C, MayT, Canton P. The Duke criteria for diagnosing in-fective endocarditis are specific: analysis of 100patients with acute fever or fever of unknown ori-gin. Clin Infect Dis. 1996;23:298-302.

34. Hoen B, Selton-Suty C, Danchin N, et al. Evalua-tion of the Duke criteria versus the Beth Israel cri-teria for the diagnosis of infective endocarditis.Clin Infect Dis. 1995;21:905-909.

35. Mitchell DP, Hanes TE, Hoyumpa AM, SchenkerS. Fever of unknown origin: assessment of thevalue of percutaneous liver biopsy. Arch InternMed. 1977;137:1001-1004.

36. Holtz T, Moseley RH, Scheiman JM. Liver biopsyin fever of unknown origin: a reappraisal. J ClinGastroenterol. 1993;17:29-32.

37. Tobkes A, Nord H. Liver biopsy: review of meth-odology and complications. Dig Dis. 1995;13:267-274.

38. Garcia-Tsao G, Boyer G. Outpatient liver biopsy: howsafe is it? Ann Intern Med. 1993;118:150-153.

39. McGill D, Rakela J, Zinsmeister AR, Ott BJ. A 21year experience with major hemorrhage after per-cutaneous liver biopsy.Gastroeneterology. 1990;99:1396-1400.

40. Piccinino F, Sagnelli E, Pasquale G, Giusti G. Com-plications following percutaneous liver biopsy: amulticentre retrospective study of 68,276 biop-sies. J Hepatol. 1986;2:165-173.

41. Buchbinder R, Detsky AS. Management of sus-pected giant cell arteritis: a decision analysis.J Rheumatol. 1992;19:1220-1228.

42. Stuart RA. Temporal artery biopsy in suspectedtemporal arteritis: a five year survey. N Z Med J.1989;102:431-433.

43. Kent RB III, Thomas L. Temporal artery biopsy.Am Surg. 1990;56:16-21.

44. Albertini JG, Ramsey ML, Marks VJ. Temporal ar-tery biopsy in a dermatologic surgery practice.Der-matol Surg. 1999;25:501-508.

45. Scott KR, Tse DT, Kronish JW. Temporal arterybiopsy technique: a clinico-anatomical ap-proach. Ophthalmic Surg. 1991;22:519-525.

46. Siemssen SJ. On the occurrence of necrotizing le-sions in temporal arteritis: review of the litera-ture with a note on the potential risk of a biopsy.Br J Plast Surg. 1987;40:73-82.

47. Slavin ML. Brow droop after superficial temporalartery biopsy [letter]. Arch Ophthalmol. 1986;104:1127.

48. Schmidt WA, Kraft HE, Vorpahl K, Volker L, Grom-nica-Ihle EJ. Color duplex ultrasonography in thediagnosis of temporal arteritis.NEngl JMed. 1997;337:1336-1342.

49. AbuRahma AF, Saiedy S, Robinson PA, Boland JP,Cotrell DJ IV, Stuart C. Role of venous duplex im-aging of the lower extremities in patients with fe-verofunknownorigin.Surgery.1997;121:366-371.

50. Volk E, Miller ML, Kirskley BA, Washington JA.The diagnostic usefulness of bone marrow cul-tures in patients with fever of unknown origin. AmJ Clin Pathol. 1998;110:150-153.

51. Riley U, Crawford S, Barrett SP, Abdalla SH. De-tection of mycobacteria in bone marrow biopsyspecimens taken to investigate pyrexia of un-known origin. J Clin Pathol. 1995;48:706-709.

52. Rothman DL, Schwartz SI, Adams JT. Diagnos-tic laparotomy for fever or abdominal pain of un-known origin. Am J Surg. 1977;133:273-275.

53. Howard PH Jr, Hardin WJ. The role of surgery infever of unknown origin.SurgClinNorth Am. 1972;52:397-403.

54. Barr J, Magnusson P, Wallensten S. Diagnosticlaparotomy in cases of obscure fever. Acta ChirScand. 1972;138:153-158.

55. Moossa AR, Skinner DB. The place of lapa-rotomy in the management of pyrexia of un-known origin. Ann R Coll Surg Engl. 1972;51:396-401.

56. Ben-Shoshan M, Gius JA, Smith I. Exploratorylaparotomy for fever of unknown origin. Surg Gy-necol Obstet. 1971;132:994-996.

57. Keller JW, Williams RD. Laparotomy for unex-plained fever. Arch Surg. 1965;90:494-498.

58. Takahashi T, Herrera MF, Onuma L, et al. Diag-nostic laparotomy in fever of unknown origin.RevInvest Clin. 1991;43:25-30.

59. Solis-Herruzo JA, Benita J, Morillas JD. Laparos-copy in fever of unknown originstudy of sev-enty cases. Endoscopy. 1981;13:207-210.

60. Erbel R, Rohmann S, Drexler M, Mohr-Kahaly S,Gerharz CD, Iversen S. Improved diagnostic valueof echocardiography in patients with infective en-docarditis by transesophageal approach: a pro-spective study. Eur Heart J. 1988;9:43-53.

61. Kupferwasser LI, Darius H, Muller AM, Martin C,Mohr-Kahaly S, Erbel R. Diagnosis of culture nega-tive endocarditis: the role of the Duke criteria andthe impact of transesophageal echocardiogra-phy. Am Heart J. 2001;142:146-152.



Documents

h1 Fuo Aim03