Research Paper
Mortality in patients presenting with fever of unknown origin S. Vanderschueren, T. Eyckmans, P. De Munter, D. Knockaert Clinical Department of General Internal Medicine and the Research Department of Microbiology and Immunology, Laboratory for Clinical Infectious and Immunological Disorders, University Hospital Leuven, Belgium Background: Few data exist on the contemporary prognosis of patients presenting with fever of unknown origin (FUO). Methods: The data of 436 adult immunocompetent patients presenting with FUO between 2000 and 2010 and followed for at least 6 months were analyzed, with a focus on FUO-related deaths. The following variables were assessed in survivors and non-survivors: age, underlying diagnosis, and, in a nested casecontrol design, fever periodicity, selected laboratory parameters (including peripheral blood counts, enzymes, and inflammatory markers) and organomegaly. Results: Thirty FUO-related deaths occurred (6.9%). Malignancy accounted for 11% of fevers but for 60% of deaths. Especially non-Hodgkin lymphoma carried a disproportionally high death toll. In the nonmalignant categories, fatality rates were below 6%. All patients discharged without diagnosis in spite of ample investigations (n5164) survived. Besides malignancy, age, continuous (as opposed to episodic) fever, anaemia, leucopenia, LDH levels, and hepatomegaly were associated with mortality. Conclusions: Fatality rates of FUO have continuously declined over the past decades. Malignancy, including lymphoma, remains a cardinal cause of death. Patients with FUO discharged without diagnosis survive. Keywords: Fever, Fever of unknown origin, Mortality, Outcome
Introduction Fever of unknown origin (FUO) has challenged clinicians for decades and continues to do so today.1 More than 200 disorders, falling into diverse subspecialty categories, can present with FUO.2 Little is known about the current vital outcome of patients presenting with FUO. Do patients with FUO die, and, if so, what do they die of? Can death be predicted or even be prevented? These questions incited the present study.
Methods From January 2000 through December 2010, adult patients with prolonged and perplexing fevers presenting to the department of general internal medicine of the University Hospital, Leuven, Belgium, were prospectively listed. The patients fulfilling the contemporary criteria of FUO were extracted: illness of more than 3 weeks duration, characterized by fever .38.3uC on .3 occasions or lower-grade fevers with C-reactive protein (CRP) levels of .30 mg/l on .3 Correspondence to: S. Vanderschueren, General Internal Medicine, University Hospital Leuven, Herestraat 49, B-3000 Leuven, Belgium. Email: [email protected]
12
ß Acta Clinica Belgica 2014 DOI 10.1179/0001551213Z.0000000005
occasions, and without diagnosis after a careful baseline evaluation minimally consisting of laboratory testing, cultures, chest and abdominal imaging, as detailed elsewhere.3 Patients with known HIIV infection, immunocompromise or with nosocomial fevers were excluded, as were survivors with a followup of less than 6 months after the index contact. Immunocompromise encompassed the following conditions: neutropenia (white blood cell count ,1.06103/ml and/or granulocyte count ,0.56 103/ml); hypogammaglobulinaemia (IgG,7.5 g/l); or the intake of the equivalent of more than 10-mg prednisone or other immunosuppressive agents for at least 2 weeks. The mortality of the patients presenting with FUO was assessed, as was the cause of death. The patients presenting with FUO whose death was attributable to the febrile illness were considered as cases. The data of the patients whose deaths later on during the follow-up period were unrelated to the febrile disease (such as traffic accident or a remote myocardial infarction) were not considered further. For each case, one control from the FUO database, matched for gender and index year, was randomly selected in a nested case-control substudy.
Acta Clinica Belgica
2014
VOL.
69
NO.
1
Vanderschueren et al.
The aetiologies of the FUO were recorded for the entire FUO database and for the cases and controls in particular and divided in the following categories: infectious disorders, malignancies, non-infectious inflammatory disorders, miscellaneous disorders and unknown diagnosis. Clinical pre-mortem diagnosis was used to categorise the fatalities. If the cause of the fever was obscure before death but became evident during autopsy, the patient was retained in the ‘unknown diagnosis’ category. In addition, the following parameters were collected at index contact in patients and controls: demographic data (age and gender), fever characteristics (periodicity, duration), laboratory parameters (including peripheral blood counts, CRP, and liver enzymes), and imaging data (presence of hepato- or splenomegaly). Episodic or recurrent (as opposed to continuous) fever was defined as at least two episodes of fever, with feverfree intervals of at least 2 weeks with seeming remission of the underlying illness.4 The literature was surveyed for articles on FUO that detailed death rates. Contribution of malignancy in particular to FUO related deaths was assessed. Historical data were compared with the present series.
Statistical methods Categorical and continuous variables are expressed as number (percentage) and as mean6standard deviation, respectively. We used Pearson Chi Square or Fisher’s exact test, as appropriate, to compare categorical variables and the Mann–Whitney test to compare continuous variables. Statistical analyses were performed using SPSS 19.0 software. All statistical testing was performed using unpaired 2tailed tests, with significance set at P,0.05. The study protocol was approved by the ethical committee of the University Hospitals, Leuven, Belgium. As the study was descriptive and did not demand deviation from standard clinical care, the need for informed consent was waived.
Results Out of 579 adult patients presenting between 2000 and 2010 with prolonged fevers to the in-or outpatient department of General Internal Medicine of the University Hospital Leuven (Leuven, Belgium), 436 fulfilled the contemporary criteria of FUO3 and had no exclusion criteria (Fig. 1). Forty-four patients (10.1%) died during follow-up. In 14 patients (seven with and seven without a diagnostic explanation for the FUO), the deaths were both remote from the index contact and not judged to be attributable to the febrile illness (e.g. fatal haemorrhagic stroke 4 years after febrile episode). Thus, 30 patients (6.9%) died because of the febrile illness, 16 during the index admission, a total of 26 within the first year and a total of 28 within the first 2
Mortality in FUO
years. Half of the non-survivors were male, half female. Table 1 shows the breakdown in diagnostic categories for the entire population with FUO and for the fatal cases. Overall, a diagnosis was established in 268 patients (61%), including non-infectious inflammatory disorders (in 24%), infections (in 17%), malignancies (in 11%) and miscellaneous disorders (in 9.9%). Ninety seven different aetiologies were uncovered. Eighteen of the 30 deaths (60%) occurred in patients with malignancies. Thus, the fatality in patients with malignancy (38%) clearly surpassed that of all other categories (fatality rates (5.4%) (Table 1). Causes of death are given in Table 1. In the group without pre-mortem diagnosis (n5168), four fatalities occurred. All four patients died during the index admission. Thus, all 164 patients discharged from the hospital without evident diagnosis survived. In the four fatal cases, autopsy was performed. In two patients, both with meningoencephalitis of unknown cause, autopsy failed to produce additional diagnostic information. In the two other patients, post-mortem examination revealed a diagnosis of intravascular lymphoma and of bilateral pneumonia, respectively. We compared survivors and non-survivors. In the entire cohort, non-survivors (n530) were significantly older than survivors (n5392) (60.8613 years versus 49.5617 years, P,0.0005). The diagnostic case-mix was significantly different (P50.001), with a disproportional representation of malignancies among nonsurvivors, as already alluded too. The comparisons of cases (n530) and controls (n530) are given in Table 2. Continuous (as opposed to episodic) fever, anaemia, leucopenia, elevated lactate dehydrogenase, and hepatomegaly emerged as additional variables significantly associated with mortality. Inflammatory parameters such as CRP levels and erythrocyte sedimentation rates were remarkably similar in cases and controls.
Discussion In this large series of patients presenting in the 2000s with FUO, the attributable mortality rate was 6.9% (first-year mortality: 6.0%). Malignancy was the main cause of death, being responsible for 11% of the fevers, but for 60% of deaths. Half of the deaths due to malignancy were caused by non-Hodgkin lymphoma. In the other categories (infections, noninfectious inflammatory disorders, miscellaneous disorders and the no-diagnosis category), the death toll was below 6%. No patient discharged from the hospital without diagnosis died as a cause of the febrile illness. Most deaths occurred during the index admission or within the first year. Apart from disease category, the following baseline variables were associated with poor vital outcome: age (nonsurvivors being 10 years older in the entire cohort), continuous (as opposed to episodic) fever, anaemia,
Acta Clinica Belgica
2014
VOL .
69
NO .
1
13
Vanderschueren et al.
Mortality in FUO
Figure 1 Study algorithm.
leucopenia, elevated lactate dehydrogenase, and hepatomegaly. Although comparison with historical series is prone to bias, owing in part to differences in casemix, it is rewarding to find out that FUO-related mortality has been continuously declining over the past decades (Table 3).1,5–14 While malignancy, the ‘emperor of all maladies’,15 is still the number one cause of death, the prognosis of this category has improved markedly too. While a diagnosis of FUOrelated malignancy represented a virtual death sentence only 3 decades ago, the malignancy-related death rate in the most recent FUO series dropped to 40% or lower. Yet, the experience in patients with FUO underscores that fever remains an important Bsymptom and prognostic marker especially in haematological malignancies.16
14
Acta Clinica Belgica
2014
VOL .
69
NO .
1
The present series confirms that the outcome of patients with FUO depends upon the underlying diagnosis and that the majority of fevers are caused by treatable conditions. Apart from disease category, age has been noted as a prognostic variable in other series.7,10,11,13 In the studies of Knockaert et al., for instance, in-hospital mortality was 12% in patients with FUO over 65 years versus 2% in younger patients.10,11 In the study of Goto et al., the vast majority of non-survivors harboured malignant disease and non-survivors were significantly older (median age: 58 years versus 49 years).13 In that series, neutrophil count and CRP levels were significantly higher in deceased patients, but not all patients had genuine FUO.13 In our study, leucopenia portended a poor prognosis and CRP levels were similar in survivors and non-survivors.
Vanderschueren et al.
Mortality in FUO
Table 1 Diagnostic categories of the total group of 436 patients with fever of unknown origin and of non-survivors
Diagnostic category
Number of patients (% of total)
Patients succumbed to febrile illness (fatality rate)
Infectious diseases Cytomegalovirus Tuberculosis Pyelonephritis Endocarditis Q Fever Bartonellosis Other Malignancies Lymphoma (non-Hodgkin/Hodgkin) Myelodysplastic syndromes Multiple myeloma Solid tumours Non-infectious inflammatory disorders Adult onset still disease Giant cell arteritis Granulomatosis with polyangiitis Sarcoidosis Behc¸et’s disease Systemic lupus erythematosus Polyarteritis nodosa Polymyalgia rheumatica Other Miscellaneous disorders Habitual hyperthermia Drug fever Familial mediterranean fever Common variable immunodeficiency Other No diagnosis
74 (17%) 10 9 7 6 5 5 32 48 (11%) 30 (20/10) 3 1 14 103 (24%) 27 10 8 7 6 6 6 3 30 43 (9.9%) 11 6 4 3 19 168 (39%)
4 (5.4%) 0 1 0 1 0 0 2 (twice Leishmaniasis) 18 (38%) 12 (9/3) 1 1 4 2 (1.9%) 0 1 0 0 0 0 0 0 1 (idiopathic pulmonary fibrosis) 2 (4.7%) 0 0 0 0 2 (1 thyrotoxicosis, 1 haemophagocytic syndrome) 4 (2.4%)
Our findings also validate the excellent vital outcome in patients with FUO discharged without diagnosis in spite of an extensive evaluation. Several studies have documented that these patients not only survive but mostly become symptom-free shortly after discharge.1,5,6,11,12,17 This outcome indicates that, in these patients, potentially lethal occult diseases are very unusual and that empirical therapy with antibiotics, antituberculosis agents, or corticosteroids
is rarely required in stable patients. Instead, a wait-and-see approach may be preferred. On the other hand, in patients with clinical deterioration and with bad prognostic markers, a more invasive diagnostic pursuit may be warranted to rule out treatable disease. The case-mix of the present series is congruent with those from recent series from affluent countries.1,10,12,14,18 In other regions, infectious diseases
Table 2 Selected baseline characteristics of non-survivors (cases) and survivors (controls) Characteristic
Non-survivors (n530)
Survivors (n530)
P-value
Age, years Diagnostic category Infections Malignancies Non-infectious inflammatory disorders Miscellaneous disorders No diagnosis Fever pattern Continuous Episodic C-reactive protein, mg/l Erythrocyte sedimentation rate, mm/h Haemoglobin (g/dl) White blood cell count, 6109/l Leukocytosis Leucopenia Monocytosis Lactate dehydrogenase, U/l Hepatomegaly Splenomegaly
60.8613
54.5617
0.06 0.001
4 (13%) 18 (60%) 2 (6.7%) 2 (6.7%) 4 (13%)
4 (13%) 3 (10%) 9 (30%) 3 (10%) 11 (37%)
25 (83%) 5 (17%) 97674 62635 11.161.7 7.964.8 9 (30%) 8 (27%) 11 (37%) 7486675 11 (37%) 13 (43%)
15 (50%) 15 (50%) 97673 62631 12.261.5 8.863.4 7 (23%) 1 (3.3%) 4 (13%) 3866190 2 (6.7%) 6 (20%)
0.006
0.95 0.75 0.02 0.18 0.77 0.03 0.07 0.005 0.01 0.09
Note: Data are expressed as number (percentage) or as mean6standard deviation. P-values were computed using Fisher’s exact test or Mann–Whitney test.
Acta Clinica Belgica
2014
VOL .
69
NO .
1
15
Mortality in FUO
16
are still a major cause of FUO and outcomes may be different. This study has other limitations. It is restricted to the experience of a single centre. The nested casecontrol study consisted of a relatively small number of patients and the minimum length of follow-up for inclusion was 6 months, implying that some late deaths may have been missed. Also, the follow-up was not identical in all subjects. However, we do not expect that using patient-days instead as patients as the main unit of measurement would alter the key findings. In conclusion, mortality in patients presenting with FUO is at an all-time low. The best predictor of survival remains disease category, with malignancy incurring the highest mortality. The vital prognosis of patients discharged without diagnosis is excellent.
89% 86% 88% 54% 52% Not stated Not stated 67% Not stated 40% 38% 2014
VOL .
69
NO .
1
References
Note: *Not all patients suffered from FUO; body temperature >37.0uC in a hospitalized patient sufficed for inclusion.
59% 86% 83% 43% 79% Not stated Not stated 62% 84% 40% 60% 19% 21% 31% 18% 24% 7.0% 13% 11% 7.1% 7.0% 11% 100 34 105 133 86 199 167 223 226 73 436 Not stated Not stated >12 months Not stated Average: 22 months 5 years in group without diagnosis .2 years in group without diagnosis Median: 810 days In-hospital Median: 12 months >6 months United States United States United States Spain United States Belgium The Netherlands Belgium Japan The Netherlands Belgium 1952–1957 1960s 1970–1980 1968–1981 1984–1990 1980–1990 1992–1994 1990–1999 1994–2002 2003–2005 2000–2010 Petersdorf5 Deal6 Larson7 Barbado8 Kazanjian9 Knockaert10,11 De Kleijn12 Vanderschueren1 Goto13* Bleeker-Rovers14 Present series (–)
Length of follow-up Country Study period
Acta Clinica Belgica
32% 21% 33% 23% 16% 6.5% 11.3 12% 8.4% 6.8% 6.9%
Percentage of malignant cases in non-survivors Series first author (reference)
Table 3 FUO related deaths: comparison with historical series
Number of cases
FUO-related mortality rate
Percentage of malignant cases in total group
Malignancy-associated mortality rate
Vanderschueren et al.
1 Vanderschueren S, Knockaert D, Adriaenssens T, Demey W, Durnez A, Blockmans D, et al. From prolonged febrile illness to fever of unknown origin. The challenge continues. Arch Intern Med. 2003;163:1033–41. 2 Arnow PM, Flaherty JP. Fever of unknown origin. Lancet. 1997;350:575–80. 3 de Kleijn EMHA, Knockaert DC, van der Meer JWM. Fever of unknown origin: a new definition and proposal for diagnostic work-up. Eur J Intern Med. 2000;11:1–3. 4 Knockaert DC, Vanneste LJ, Bobbaers HJ. Recurrent or periodic fever of unknown origin: review of 45 cases and survey of the literature/Medicine (Baltimore). 1997;76:392–400. 5 Petersdorf RB, Beeson PB. Fever of unexplained origin: report on 100 cases. Medicine (Baltimore). 1961;40:1–30. 6 Deal WB. Fever of unknown origin: analysis of 34 patients. Postgrad Med. 1971;50:182–8. 7 Larson EB, Featherstone HJ, Petersdorf RG. Fever of undetermined origin: diagnosis and follow-up of 105 cases, 1970–1980. Medicine (Baltimore). 1982;61:269–92. 8 Barbado FJ, Vazquez JJ, Penˇa JM, Seoane JG, Arnalich F, Gil A, et al. Fever of unknown origin: a survey on 133 patients. J Med. 1984;15:185–92. 9 Kazanjian PH. Fever of unknown: review of 86 patients treated in community hospitals. Clin Infect Dis. 1992;15:968–73. 10 Knockaert DC, Vanneste LJ, Vanneste SB, Bobbaers HJ. Fever of unknown origin in the 1980s: an update of the diagnostic spectrum. Arch Intern Med. 1992;152:51–5. 11 Knockaert DC, Dujardin KS, Bobbaers HJ. Long-term followup of patients with undiagnosed fever of unknown origin. Arch Intern Med. 1996;156:618–20. 12 de Kleijn EMHA, Vandenbroucke JP, van der Meer JWM, and the Netherlands FUO Study Group. Fever of unknown origin (FUO). I. A prospective multicenter study of 167 patients with FUO, using fixed epidemiologic entry criteria. Medicine (Baltimore). 1982;61:269–92. 13 Goto M, Koyama H, Takahashi O, Fukui T. A retrospective review of 226 hospitalized patients with fever. Intern Med. 2007;46:17–22. 14 Bleeker-Rovers CP, Vos FJ, de Kleijn EMHA, Mudde AH, Dofferhoff TS, Richter C, et al. A prospective multicenter study on fever of unknown origin. The yield of a structured diagnostic protocol. Medicine (Baltimore). 2007;86:26–38. 15 Mukherjee S. The emperor of all maladies. A biography of cancer. New York: Scribner; 2010. 16 Tironi FM, Dierickx D, Knockaert D, Vanderschueren S. Lymphoma presenting as fever of unknown origin. Acta Clin Belg. 2009;64–6 (Abstract 11). 17 Iikuni Y, Okada J, Kondo H, Kashiwazaki. Current fever of unknown origin 1982–1992. Intern Med. 1994;33:67–73. 18 Zenone T. Fever of unknown origin in adults: evaluation of 144 cases in a non-university hospital. Scan J Infect Dis. 2006;38:632–9.
Mortality in patients presenting with fever of unknown origin S. Vanderschueren, T. Eyckmans, P. De Munter, D. Knockaert Clinical Department of General Internal Medicine and the Research Department of Microbiology and Immunology, Laboratory for Clinical Infectious and Immunological Disorders, University Hospital Leuven, Belgium Background: Few data exist on the contemporary prognosis of patients presenting with fever of unknown origin (FUO). Methods: The data of 436 adult immunocompetent patients presenting with FUO between 2000 and 2010 and followed for at least 6 months were analyzed, with a focus on FUO-related deaths. The following variables were assessed in survivors and non-survivors: age, underlying diagnosis, and, in a nested casecontrol design, fever periodicity, selected laboratory parameters (including peripheral blood counts, enzymes, and inflammatory markers) and organomegaly. Results: Thirty FUO-related deaths occurred (6.9%). Malignancy accounted for 11% of fevers but for 60% of deaths. Especially non-Hodgkin lymphoma carried a disproportionally high death toll. In the nonmalignant categories, fatality rates were below 6%. All patients discharged without diagnosis in spite of ample investigations (n5164) survived. Besides malignancy, age, continuous (as opposed to episodic) fever, anaemia, leucopenia, LDH levels, and hepatomegaly were associated with mortality. Conclusions: Fatality rates of FUO have continuously declined over the past decades. Malignancy, including lymphoma, remains a cardinal cause of death. Patients with FUO discharged without diagnosis survive. Keywords: Fever, Fever of unknown origin, Mortality, Outcome
Introduction Fever of unknown origin (FUO) has challenged clinicians for decades and continues to do so today.1 More than 200 disorders, falling into diverse subspecialty categories, can present with FUO.2 Little is known about the current vital outcome of patients presenting with FUO. Do patients with FUO die, and, if so, what do they die of? Can death be predicted or even be prevented? These questions incited the present study.
Methods From January 2000 through December 2010, adult patients with prolonged and perplexing fevers presenting to the department of general internal medicine of the University Hospital, Leuven, Belgium, were prospectively listed. The patients fulfilling the contemporary criteria of FUO were extracted: illness of more than 3 weeks duration, characterized by fever .38.3uC on .3 occasions or lower-grade fevers with C-reactive protein (CRP) levels of .30 mg/l on .3 Correspondence to: S. Vanderschueren, General Internal Medicine, University Hospital Leuven, Herestraat 49, B-3000 Leuven, Belgium. Email: [email protected]
12
ß Acta Clinica Belgica 2014 DOI 10.1179/0001551213Z.0000000005
occasions, and without diagnosis after a careful baseline evaluation minimally consisting of laboratory testing, cultures, chest and abdominal imaging, as detailed elsewhere.3 Patients with known HIIV infection, immunocompromise or with nosocomial fevers were excluded, as were survivors with a followup of less than 6 months after the index contact. Immunocompromise encompassed the following conditions: neutropenia (white blood cell count ,1.06103/ml and/or granulocyte count ,0.56 103/ml); hypogammaglobulinaemia (IgG,7.5 g/l); or the intake of the equivalent of more than 10-mg prednisone or other immunosuppressive agents for at least 2 weeks. The mortality of the patients presenting with FUO was assessed, as was the cause of death. The patients presenting with FUO whose death was attributable to the febrile illness were considered as cases. The data of the patients whose deaths later on during the follow-up period were unrelated to the febrile disease (such as traffic accident or a remote myocardial infarction) were not considered further. For each case, one control from the FUO database, matched for gender and index year, was randomly selected in a nested case-control substudy.
Acta Clinica Belgica
2014
VOL.
69
NO.
1
Vanderschueren et al.
The aetiologies of the FUO were recorded for the entire FUO database and for the cases and controls in particular and divided in the following categories: infectious disorders, malignancies, non-infectious inflammatory disorders, miscellaneous disorders and unknown diagnosis. Clinical pre-mortem diagnosis was used to categorise the fatalities. If the cause of the fever was obscure before death but became evident during autopsy, the patient was retained in the ‘unknown diagnosis’ category. In addition, the following parameters were collected at index contact in patients and controls: demographic data (age and gender), fever characteristics (periodicity, duration), laboratory parameters (including peripheral blood counts, CRP, and liver enzymes), and imaging data (presence of hepato- or splenomegaly). Episodic or recurrent (as opposed to continuous) fever was defined as at least two episodes of fever, with feverfree intervals of at least 2 weeks with seeming remission of the underlying illness.4 The literature was surveyed for articles on FUO that detailed death rates. Contribution of malignancy in particular to FUO related deaths was assessed. Historical data were compared with the present series.
Statistical methods Categorical and continuous variables are expressed as number (percentage) and as mean6standard deviation, respectively. We used Pearson Chi Square or Fisher’s exact test, as appropriate, to compare categorical variables and the Mann–Whitney test to compare continuous variables. Statistical analyses were performed using SPSS 19.0 software. All statistical testing was performed using unpaired 2tailed tests, with significance set at P,0.05. The study protocol was approved by the ethical committee of the University Hospitals, Leuven, Belgium. As the study was descriptive and did not demand deviation from standard clinical care, the need for informed consent was waived.
Results Out of 579 adult patients presenting between 2000 and 2010 with prolonged fevers to the in-or outpatient department of General Internal Medicine of the University Hospital Leuven (Leuven, Belgium), 436 fulfilled the contemporary criteria of FUO3 and had no exclusion criteria (Fig. 1). Forty-four patients (10.1%) died during follow-up. In 14 patients (seven with and seven without a diagnostic explanation for the FUO), the deaths were both remote from the index contact and not judged to be attributable to the febrile illness (e.g. fatal haemorrhagic stroke 4 years after febrile episode). Thus, 30 patients (6.9%) died because of the febrile illness, 16 during the index admission, a total of 26 within the first year and a total of 28 within the first 2
Mortality in FUO
years. Half of the non-survivors were male, half female. Table 1 shows the breakdown in diagnostic categories for the entire population with FUO and for the fatal cases. Overall, a diagnosis was established in 268 patients (61%), including non-infectious inflammatory disorders (in 24%), infections (in 17%), malignancies (in 11%) and miscellaneous disorders (in 9.9%). Ninety seven different aetiologies were uncovered. Eighteen of the 30 deaths (60%) occurred in patients with malignancies. Thus, the fatality in patients with malignancy (38%) clearly surpassed that of all other categories (fatality rates (5.4%) (Table 1). Causes of death are given in Table 1. In the group without pre-mortem diagnosis (n5168), four fatalities occurred. All four patients died during the index admission. Thus, all 164 patients discharged from the hospital without evident diagnosis survived. In the four fatal cases, autopsy was performed. In two patients, both with meningoencephalitis of unknown cause, autopsy failed to produce additional diagnostic information. In the two other patients, post-mortem examination revealed a diagnosis of intravascular lymphoma and of bilateral pneumonia, respectively. We compared survivors and non-survivors. In the entire cohort, non-survivors (n530) were significantly older than survivors (n5392) (60.8613 years versus 49.5617 years, P,0.0005). The diagnostic case-mix was significantly different (P50.001), with a disproportional representation of malignancies among nonsurvivors, as already alluded too. The comparisons of cases (n530) and controls (n530) are given in Table 2. Continuous (as opposed to episodic) fever, anaemia, leucopenia, elevated lactate dehydrogenase, and hepatomegaly emerged as additional variables significantly associated with mortality. Inflammatory parameters such as CRP levels and erythrocyte sedimentation rates were remarkably similar in cases and controls.
Discussion In this large series of patients presenting in the 2000s with FUO, the attributable mortality rate was 6.9% (first-year mortality: 6.0%). Malignancy was the main cause of death, being responsible for 11% of the fevers, but for 60% of deaths. Half of the deaths due to malignancy were caused by non-Hodgkin lymphoma. In the other categories (infections, noninfectious inflammatory disorders, miscellaneous disorders and the no-diagnosis category), the death toll was below 6%. No patient discharged from the hospital without diagnosis died as a cause of the febrile illness. Most deaths occurred during the index admission or within the first year. Apart from disease category, the following baseline variables were associated with poor vital outcome: age (nonsurvivors being 10 years older in the entire cohort), continuous (as opposed to episodic) fever, anaemia,
Acta Clinica Belgica
2014
VOL .
69
NO .
1
13
Vanderschueren et al.
Mortality in FUO
Figure 1 Study algorithm.
leucopenia, elevated lactate dehydrogenase, and hepatomegaly. Although comparison with historical series is prone to bias, owing in part to differences in casemix, it is rewarding to find out that FUO-related mortality has been continuously declining over the past decades (Table 3).1,5–14 While malignancy, the ‘emperor of all maladies’,15 is still the number one cause of death, the prognosis of this category has improved markedly too. While a diagnosis of FUOrelated malignancy represented a virtual death sentence only 3 decades ago, the malignancy-related death rate in the most recent FUO series dropped to 40% or lower. Yet, the experience in patients with FUO underscores that fever remains an important Bsymptom and prognostic marker especially in haematological malignancies.16
14
Acta Clinica Belgica
2014
VOL .
69
NO .
1
The present series confirms that the outcome of patients with FUO depends upon the underlying diagnosis and that the majority of fevers are caused by treatable conditions. Apart from disease category, age has been noted as a prognostic variable in other series.7,10,11,13 In the studies of Knockaert et al., for instance, in-hospital mortality was 12% in patients with FUO over 65 years versus 2% in younger patients.10,11 In the study of Goto et al., the vast majority of non-survivors harboured malignant disease and non-survivors were significantly older (median age: 58 years versus 49 years).13 In that series, neutrophil count and CRP levels were significantly higher in deceased patients, but not all patients had genuine FUO.13 In our study, leucopenia portended a poor prognosis and CRP levels were similar in survivors and non-survivors.
Vanderschueren et al.
Mortality in FUO
Table 1 Diagnostic categories of the total group of 436 patients with fever of unknown origin and of non-survivors
Diagnostic category
Number of patients (% of total)
Patients succumbed to febrile illness (fatality rate)
Infectious diseases Cytomegalovirus Tuberculosis Pyelonephritis Endocarditis Q Fever Bartonellosis Other Malignancies Lymphoma (non-Hodgkin/Hodgkin) Myelodysplastic syndromes Multiple myeloma Solid tumours Non-infectious inflammatory disorders Adult onset still disease Giant cell arteritis Granulomatosis with polyangiitis Sarcoidosis Behc¸et’s disease Systemic lupus erythematosus Polyarteritis nodosa Polymyalgia rheumatica Other Miscellaneous disorders Habitual hyperthermia Drug fever Familial mediterranean fever Common variable immunodeficiency Other No diagnosis
74 (17%) 10 9 7 6 5 5 32 48 (11%) 30 (20/10) 3 1 14 103 (24%) 27 10 8 7 6 6 6 3 30 43 (9.9%) 11 6 4 3 19 168 (39%)
4 (5.4%) 0 1 0 1 0 0 2 (twice Leishmaniasis) 18 (38%) 12 (9/3) 1 1 4 2 (1.9%) 0 1 0 0 0 0 0 0 1 (idiopathic pulmonary fibrosis) 2 (4.7%) 0 0 0 0 2 (1 thyrotoxicosis, 1 haemophagocytic syndrome) 4 (2.4%)
Our findings also validate the excellent vital outcome in patients with FUO discharged without diagnosis in spite of an extensive evaluation. Several studies have documented that these patients not only survive but mostly become symptom-free shortly after discharge.1,5,6,11,12,17 This outcome indicates that, in these patients, potentially lethal occult diseases are very unusual and that empirical therapy with antibiotics, antituberculosis agents, or corticosteroids
is rarely required in stable patients. Instead, a wait-and-see approach may be preferred. On the other hand, in patients with clinical deterioration and with bad prognostic markers, a more invasive diagnostic pursuit may be warranted to rule out treatable disease. The case-mix of the present series is congruent with those from recent series from affluent countries.1,10,12,14,18 In other regions, infectious diseases
Table 2 Selected baseline characteristics of non-survivors (cases) and survivors (controls) Characteristic
Non-survivors (n530)
Survivors (n530)
P-value
Age, years Diagnostic category Infections Malignancies Non-infectious inflammatory disorders Miscellaneous disorders No diagnosis Fever pattern Continuous Episodic C-reactive protein, mg/l Erythrocyte sedimentation rate, mm/h Haemoglobin (g/dl) White blood cell count, 6109/l Leukocytosis Leucopenia Monocytosis Lactate dehydrogenase, U/l Hepatomegaly Splenomegaly
60.8613
54.5617
0.06 0.001
4 (13%) 18 (60%) 2 (6.7%) 2 (6.7%) 4 (13%)
4 (13%) 3 (10%) 9 (30%) 3 (10%) 11 (37%)
25 (83%) 5 (17%) 97674 62635 11.161.7 7.964.8 9 (30%) 8 (27%) 11 (37%) 7486675 11 (37%) 13 (43%)
15 (50%) 15 (50%) 97673 62631 12.261.5 8.863.4 7 (23%) 1 (3.3%) 4 (13%) 3866190 2 (6.7%) 6 (20%)
0.006
0.95 0.75 0.02 0.18 0.77 0.03 0.07 0.005 0.01 0.09
Note: Data are expressed as number (percentage) or as mean6standard deviation. P-values were computed using Fisher’s exact test or Mann–Whitney test.
Acta Clinica Belgica
2014
VOL .
69
NO .
1
15
Mortality in FUO
16
are still a major cause of FUO and outcomes may be different. This study has other limitations. It is restricted to the experience of a single centre. The nested casecontrol study consisted of a relatively small number of patients and the minimum length of follow-up for inclusion was 6 months, implying that some late deaths may have been missed. Also, the follow-up was not identical in all subjects. However, we do not expect that using patient-days instead as patients as the main unit of measurement would alter the key findings. In conclusion, mortality in patients presenting with FUO is at an all-time low. The best predictor of survival remains disease category, with malignancy incurring the highest mortality. The vital prognosis of patients discharged without diagnosis is excellent.
89% 86% 88% 54% 52% Not stated Not stated 67% Not stated 40% 38% 2014
VOL .
69
NO .
1
References
Note: *Not all patients suffered from FUO; body temperature >37.0uC in a hospitalized patient sufficed for inclusion.
59% 86% 83% 43% 79% Not stated Not stated 62% 84% 40% 60% 19% 21% 31% 18% 24% 7.0% 13% 11% 7.1% 7.0% 11% 100 34 105 133 86 199 167 223 226 73 436 Not stated Not stated >12 months Not stated Average: 22 months 5 years in group without diagnosis .2 years in group without diagnosis Median: 810 days In-hospital Median: 12 months >6 months United States United States United States Spain United States Belgium The Netherlands Belgium Japan The Netherlands Belgium 1952–1957 1960s 1970–1980 1968–1981 1984–1990 1980–1990 1992–1994 1990–1999 1994–2002 2003–2005 2000–2010 Petersdorf5 Deal6 Larson7 Barbado8 Kazanjian9 Knockaert10,11 De Kleijn12 Vanderschueren1 Goto13* Bleeker-Rovers14 Present series (–)
Length of follow-up Country Study period
Acta Clinica Belgica
32% 21% 33% 23% 16% 6.5% 11.3 12% 8.4% 6.8% 6.9%
Percentage of malignant cases in non-survivors Series first author (reference)
Table 3 FUO related deaths: comparison with historical series
Number of cases
FUO-related mortality rate
Percentage of malignant cases in total group
Malignancy-associated mortality rate
Vanderschueren et al.
1 Vanderschueren S, Knockaert D, Adriaenssens T, Demey W, Durnez A, Blockmans D, et al. From prolonged febrile illness to fever of unknown origin. The challenge continues. Arch Intern Med. 2003;163:1033–41. 2 Arnow PM, Flaherty JP. Fever of unknown origin. Lancet. 1997;350:575–80. 3 de Kleijn EMHA, Knockaert DC, van der Meer JWM. Fever of unknown origin: a new definition and proposal for diagnostic work-up. Eur J Intern Med. 2000;11:1–3. 4 Knockaert DC, Vanneste LJ, Bobbaers HJ. Recurrent or periodic fever of unknown origin: review of 45 cases and survey of the literature/Medicine (Baltimore). 1997;76:392–400. 5 Petersdorf RB, Beeson PB. Fever of unexplained origin: report on 100 cases. Medicine (Baltimore). 1961;40:1–30. 6 Deal WB. Fever of unknown origin: analysis of 34 patients. Postgrad Med. 1971;50:182–8. 7 Larson EB, Featherstone HJ, Petersdorf RG. Fever of undetermined origin: diagnosis and follow-up of 105 cases, 1970–1980. Medicine (Baltimore). 1982;61:269–92. 8 Barbado FJ, Vazquez JJ, Penˇa JM, Seoane JG, Arnalich F, Gil A, et al. Fever of unknown origin: a survey on 133 patients. J Med. 1984;15:185–92. 9 Kazanjian PH. Fever of unknown: review of 86 patients treated in community hospitals. Clin Infect Dis. 1992;15:968–73. 10 Knockaert DC, Vanneste LJ, Vanneste SB, Bobbaers HJ. Fever of unknown origin in the 1980s: an update of the diagnostic spectrum. Arch Intern Med. 1992;152:51–5. 11 Knockaert DC, Dujardin KS, Bobbaers HJ. Long-term followup of patients with undiagnosed fever of unknown origin. Arch Intern Med. 1996;156:618–20. 12 de Kleijn EMHA, Vandenbroucke JP, van der Meer JWM, and the Netherlands FUO Study Group. Fever of unknown origin (FUO). I. A prospective multicenter study of 167 patients with FUO, using fixed epidemiologic entry criteria. Medicine (Baltimore). 1982;61:269–92. 13 Goto M, Koyama H, Takahashi O, Fukui T. A retrospective review of 226 hospitalized patients with fever. Intern Med. 2007;46:17–22. 14 Bleeker-Rovers CP, Vos FJ, de Kleijn EMHA, Mudde AH, Dofferhoff TS, Richter C, et al. A prospective multicenter study on fever of unknown origin. The yield of a structured diagnostic protocol. Medicine (Baltimore). 2007;86:26–38. 15 Mukherjee S. The emperor of all maladies. A biography of cancer. New York: Scribner; 2010. 16 Tironi FM, Dierickx D, Knockaert D, Vanderschueren S. Lymphoma presenting as fever of unknown origin. Acta Clin Belg. 2009;64–6 (Abstract 11). 17 Iikuni Y, Okada J, Kondo H, Kashiwazaki. Current fever of unknown origin 1982–1992. Intern Med. 1994;33:67–73. 18 Zenone T. Fever of unknown origin in adults: evaluation of 144 cases in a non-university hospital. Scan J Infect Dis. 2006;38:632–9.
