Fever of unknown origin in the elderly

Authors


 Sari Tal MD, Subacute Department, Harzfeld Geriatric Hospital, Kaplan Medical Center, Gedera 70750, Israel (fax: 972 8859 4227; e-mail: ysmg_tal@netvision.net.il).

Abstract

Fever of unknown origin (FUO) means fever that does not resolve itself in the period expected for self-limited infection and whose cause cannot be ascertained despite considerable diagnostic efforts. The differential diagnosis is often different in older patients, and presentation of disease is frequently nonspecific and symptoms are difficult to interpret. Multisystem disease has emerged as the most frequent cause of FUO in the elderly, and temporal arteritis is the most frequent specific diagnosis. Infections, particular tuberculosis, remain an important group. FUO is often associated with treatable conditions in this age group. Early recognition and prompt initiation of appropriate empirical therapy are cornerstones of the strategy.

Introduction

The approach to the medical evaluation of the elderly person requires a perspective different from that needed for the medical evaluation of the younger person. The spectrum of symptoms is different, the manifestations of distress are more subtle, the implications for maintenance of function are more important, and improvement is sometimes less dramatic and slower to appear. The differential diagnosis is often different in older patients, and presentation of disease is frequently nonspecific and symptoms are difficult to interpret. Fever of unknown origin (FUO) in the elderly is an example of a classicmedical syndrome that requires a specific approach.

Fever of unknown origin means fever that does not resolve itself in the period expected for self-limited infection and whose cause cannot be ascertained despite considerable diagnostic effort [1]. In1961, Petersdorf and Beeson introduced the definition that subsequently became standard – namely, illness of more than 3-week duration, fever higher than 38.3°C on several occasions, and diagnosis uncertain after 1 week of study in hospital [2]. Recently, Durack and Street have proposed a new system for classification of FUO: (i) classic FUO; (ii) nosocomial FUO; (iii) neutropenic FUO; and (iv) FUO associated with HIV infection. Because hospital admission is so expensive and as thorough diagnostic testing now can be performed in an outpatient setting, the definition was recently modified to remove the requirement that a hospital be the setting for 1 week of evaluation [1, 3, 4].

Based on these criteria, studies of FUO in the general population showed that infections were the most common cause of FUO; intra-abdominal abscesses, endocarditis and tuberculosis predominated [5]. Neoplasms and multisystem/collagen vascular disease comprised the remaining major aetiologies of FUO. The proportion of undiagnosed causes is 30%[6].

Past studies of FUO in the elderly have revealed that unlike the young, a precise diagnosis can be made 87–95% of the time [7]. In many cases, FUO in the elderly is the result of atypical presentation of common disease. Infection is the aetiology in 25–35% of cases, with tuberculosis occurring much more commonly in elderly than young patients with FUO (Table 1). Connective-tissue disease such as temporal arteritis, rheumatoid arthritis, and polymyalgia rheumatica account for 25–31% of causes in elderly patients, and malignancy accounts for 12–23% of cases [7]. As many of these diseases are treatable, it is well worth pursuing the aetiology of FUO in the elderly.

Table 1.  Diagnosis of FUO in elderly patients (more than 65 years old) compared with the diagnosis in younger patients (Percentages represent pooled data from a study of Knockaert et al. – JAGS 1993; 41: 1187–92)
DiagnosesElderly (n = 47)Young (n = 152)
  1. Fuo, fever of unknown origin.

Infections (%)12 (25)33 (21)
 Abscess 2 6
 Endocarditis 1 2
 Tuberculosis 6 4
 Viral infections 1 8
Multisystem disease15 (31)27 (17)
Tumours 6 (12) 8 (5)
Miscellaneous 5 (10)24 (15)
Drug-related fever 3 (6) 3 (1)
Habitual hyperthermia 0 5 (3)
Factitious 0 7 (4)
No diagnosis 6 (12)45 (29)

Infections

In most published series of FUO in the general population, infection has been the most common diagnosis. This was also the case in the published series of FUO in the elderly [7–10]. Compared with the younger population, the elderly have increased susceptibility to infection and are at significantly increased risk for morbidity and mortality because of many common infections. Possible explanations for the observed higher morbidity and mortality rates amongst older patients include low physiological reserves because of the biological changes that accompany ageing and the frequent presence of comorbid illnesses [8].

Two earlier studies published in 1978 [9] and 1982 [10] reported that infections (37% and 41%, respectively) and multisystem/collagen vascular disease (23% and 30%, respectively) were the primary causes of prolonged perplexing fevers in the elderly (Table 2). In a more recent study by Knockaert et al. infections were less common than multisystem disease, and this trend was even more pronounced in patients older than 70 years [7]. Although the global frequency of infections was the same in older and younger patients, some important differences were noted.

Table 2.  Percentages represent pooled data from three studies of FUO in the elderly (JAGS 1993; 41: 1187–92; Concours Medical 1982; 104: 4679–89; Arch Intern Medical 1979; 139: 575–79)
DiagnosisKnockaertBarrierEsposito
  1. Fuo, fever of unknown origin.

Infections (%)2541.336.9
Tumour1213.023.4
Multisystem3130.325.2
Miscellaneous10 2.1 7.0
No diagnosis1213.0 5.4

Intra-abdominal abscesses were found in 4% of the elderly with FUO in the study by Knockaert et al. with same proportion as in the younger group. A small number of abscesses were found compared with previous studies. The most probable explanation is the widespread use of ultrasonography and computed tomography (CT) that allows early detection of this previously common cause of FUO [7].

The diverse sites of intra-abdominal abscesses give rise to different clinical features. Localizing symptoms, such as abdominal pain, nausea, vomiting or diarrhoea, are common in liver or intraperitoneal abscesses or chronic cholecystitis. Tenderness on examination is reported in most cases of liver, splenic or intraperitoneal abscess. Elderly patients typically have a more subacute course with few signs and symptoms and a long illness [1].

Tuberculosis was found in 12% of elderly with FUO, whilst it accounted for only 2% of the diagnoses in younger patients in the study by Knockaert et al. carried out in Belgium. Tuberculosis accounted for 50% of the infections in the elderly in this study [7]. In two previous series of FUO in the elderly, tuberculosis caused 20% of the infections [9, 10]. Elderly individuals appear to be at increased risk of infection, particularly because of reactivation of earlier disease. The nursing home patient is at the highest risk for this disease. A number of studies confirm the differing presentations of tuberculosisbetween the elderly and younger adults. Haemoptysis, night sweats and a positive purified protein derivative (PPD) test response are less common in the elderly. Pleural effusion may be the sole manifestation of tuberculosis. There is no difference in bacteriologically proven disease or radiologic findings between the two groups, apart from the more common occurrence of miliary tuberculosis in the elderly [11]. Disseminated disease to lymph nodes, bones, kidneys, gastrointestinal tract and skin may cause diagnostic confusion and delay.

Osteomyelitis may be a rare cause of FUO in elderly patients. Infection reaches via the blood stream or is spread from adjoining tissue. Haematogenous osteomyelitis most often affects the vertebrae but may also affect the long bones (e.g. femur, tibia, humerus). Haematogenous osteomyelitis is almost always a monomicrobial infection affecting predominantly the older population. Staphylococcus aureus is the most common microorganism isolated; other common pathogens include group β-streptococci. However, Gram-negative aerobic bacilli, most likely caused by genitourinary tract infection or instrumentation, are also found [12]. Vascular insufficiency (e.g. diabetes mellitus, atherosclerosis, vasculitis) and neuropathy (e.g. diabetes mellitus) are common contributing factors to osteomyelitis of the foot in elderly patients. The small bones of the feet and toes are most often involved. Pressure sores may be associated with underlying osteomyelitis that is difficult to differentiate clinically from infection or colonization of adjacent soft tissue. Staphylococcus aureus is a pathogen in more than half of the cases of contiguous-focus osteomyelitis. However, in contrast to haematogenous osteomyelitis, these infections often are polymicrobial and are more likely to involve Gram-negative and anaerobic bacteria. Although several bacteria are often cultured, not all of them are necessarily pathogenic. Cultures of bone specimens are frequently contaminated with organisms residing in adjacent soft tissue [13]. The bacteria that are commonly found include S. aureus, S. epidermidis, streptococci, Gram-negative aerobic bacilli, and anaerobic organisms. A bacteriological diagnosis of chronic osteomyelitis which is based on isolation of common pathogens other than S. aureus from sinus tracts does not reflect the pathogen isolated from bone and must be verified by an appropriate operative bone culture [14]. Pain and fever are the hallmark symptoms of osteomyelitis in the general population, as well as in the elderly. However, pain may be absent in some ofthe debilitated patients with osteomyelitis because of an overlying pressure sore that does not heal andin diabetic patients with osteomyelitis of the foot[15].

Although the frequency of infective endocarditis as a cause of FUO is quite low, it has become more prevalent amongst the elderly. Epidemiological studies have shown an upswing in the average age of patients with infective endocarditis. Fifty per cent of the cases occur in patients above 60 years of age. The disease is more common amongst men [16]. Prevalence has increased because the number of elderly persons with prosthetic valves has increased, hospital-acquired bacteraemia has become more prevalent, and persons with valvular heart disease survive longer. Streptococci and staphylococci are the predominant organisms, which are isolated from about 80% of elderly patients with endocarditis. Studies have noted a higher prevalence of enterococci in the elderly. In addition, Streptococcus bovis, an organism associated with colonic malignancy, is more common in elderly patients with endocarditis [17, 18]. However, a recent study did not show any significant difference in the appearance of the various pathogens within the overall general population [19, 20]. In older patients, endocarditis occurs more frequently on the mitral valve than it does on the aortic valve [16]. Sites of primary infection include the mouth, the genitourinary tract (particularly after procedures involving instrumentation), the skin, decubitus ulcers, surgical wound, catheters and rarely the gastrointestinal tract [20]. The presenting symptoms of infective endocarditis in the older patients are nonspecific, such as lethargy, fatigue, malaise, anorexia and weight loss. Heart murmurs in elderly may be wrongly attributed to the underlying valvular calcification and therefore not taken into consideration. Sometimes, endocarditis in the elderly may be present with a stroke syndrome, rheumatologic complaints, or peripheral nervous system abnormalities [21, 22]. Studies that used the Duke criteria for diagnosis of endocarditis have not found any relevant differences between elderly and young patients with regard to the frequency of fever, heart failure, embolic events, neurological symptoms, distribution of causative organisms and cerebral deficit at the time of discharge from the hospital [19, 20]. However, renal insufficiency at admission and malignancy are significantly more common amongst elderly patients with infective endocarditis [19, 20].

Viral diseases are rarely found in elderly patient with FUO, and only a limited number of viruses such as cytomegalovirus and Epstein–Barr virus, and human immunodeficiency virus (HIV) may cause FUO. Cytomegalovirus and Epstein–Barr virus infection are typical diseases of children, adolescents and young adults, but it should not be forgotten that mononucleosis occurs in elderly patients too [23].

Human immunodeficiency virus infection is primarily a disease of young sexually active people and was rarely found in the elderly population. However, more recent examination of the issues reveals that older persons are increasingly being affected by HIV. In fact, it has been argued that older adults should be considered at risk for HIV infection because they express themselves sexually, may be (or have been) intravenous drug users, may have received blood transfusions, and may already have a compromised immune system as a result of other age- and health-related conditions [24]. Because the elderly, may, experience chronic and acute age-related disease to which problems can be attributed, HIV and AIDS may be the last disease considered, if considered at all. The most commonly documented causes of fever in HIV infected patients include: primary HIV-infection (PHI), disseminated Mycobacterium avium infection, Pneumocystis carinii pneumonia, cytomegalovirus infection, disseminated histoplasmosis and lymphoma. The various infections accounting for HIV-associated FUO change according to their prevalence in the different global locations [25]. HIV symptoms can also mimic symptoms of other diseases common amongst the elderly, making it difficult to diagnose. For example, the symptoms of HIV dementia can mimic Alzheimer's or Parkinson's disease. Similarly, pneumocystic pneumonia may be mistaken for heart failure in the elderly with chronic heart disease [26].

Multisystem disease

The collagen vascular disorders, particularly temporal arteritis, seem to be the most frequent cause of FUO in the elderly. A number of disorders of immune-mediated injury, such as temporal arteritis, polymyalgia rheumatica (PMR), Wegener's disease, polyarteritis nodosa, rheumatoid arthritis and sarcoidosis constitute the most important disease category in the study by Knockaert et al. and thistrend is more pronounced as age increases. Temporal arteritis and PMR represent 60% of cases in this category [7].

Temporal arteritis represented 17% of the cases of FUO in elderly patients [7]. The disease is almost always confined to Caucasians. The incidence is higher in Scandinavia and north Europe (between 17 and 18 cases per 100 000 population aged over 50 years) than in middle France, Spain and Israel [27]. Clinical suspicion can be based, in part, on the American College of Rheumatology study that determined highly sensitive parameters for diagnosis of temporal arteritis. These parameters include age more than 50 years, a Westergren erythrocyte sedimentation rate of more than 50 mm h−1 and an abnormal temporal artery biopsy. The highest specific clinical features include jaw or tongue claudication, visual abnormalities and temporal artery abnormalities (e.g. decreased pulse, tenderness or nodules). High clinical suspicion includes any one of these three criteria in a patient more than 50 years of age with an elevated erythrocyte sedimentation rate (ESR), or if three ormore of the following criteria are met: (i) new localized headache; (ii) temporal artery abnormality; (iii) elevated ESR (more 50 mm h−1) and (iv) abnormal temporal artery biopsy (e.g. necrotizing arteritis or multinucleated giant cells) [28]. Temporal arteritis affects many arteries throughout the body, producing symptoms and signs, which mimic many other medical and surgical conditions. A systemic illness with malaise, anorexia, night sweats, weight loss and depression is not uncommon. This mode of onset may be confused with infection and malignancy, leading to intensive investigation with the possibility of blindness or stroke developing in the mean time. Arteritic involvement by inflammation is most frequently noticed in superficial temporal arteries, which may stand out and are tender on brushing the hair. There is debate about the value of the presentation of the pulsation of temporal artery, as atherosclerosis seen on temporal artery biopsy may be responsible for reduced or absent pulsation. Histological detection of giant cell arteritis remains the only diagnostic investigation in temporal arteritis. Normal biopsy appearances do not exclude the diagnosis because of possible skip lesions.

Polymyalgia rheumatica is a painful condition generally seen in patients older than 50 years of age and may be associated with temporal arteritis. Similar to temporal arteritis PMR is found to be more frequent in Sweden and Denmark than in North Italy [27]. Its hallmark is pain and stiffness in the shoulders, neck and hips that may appear suddenly. Morning or late evening stiffness is common in both muscles and joints. Headache, fatigue, depression and generalized weakness may be confused with other diseases. Physical findings are usually unremarkable, although tenderness on palpation of axial muscles or pain on motion of joints may be seen. A markedly elevated sedimentation rate is commonly associated with, but not diagnostic of, PMR; its absence should not rule out this disease [29]. In several reports, a sizable proportion of patients with PMR up to 22% had an ESR that was either normal or slightly increased at the diagnosis, supporting the notion that an increased ESR should not be a requirement for its diagnosis. In these studies the diagnosis was based on an otherwise typical presentation and a rapid response to a corticosteroid given in a low dose [30, 31].

Tumours

Tumours are said to be a common cause of FUO in the elderly. In the study by Knockaert et al. tumours were slightly more prevalent in elderly than in younger patients but clearly less common than in previous reports [7, 9, 10]. The tumours were leukaemia, Hodgkin's disease, multiple myeloma and colon cancer. Colon cancer emerged as arelatively important cause (17% of all tumours) [7]. The relative importance of tumours as a cause ofFUO is declining, probably because of the easydetection of solid tumours and enlarged lymphnodes by ultrasonography and computed tomography [5].

Miscellaneous group

Pulmonary embolism is an important cause of FUO in the elderly, particularly in bedridden patients and represented 4% of the cases of FUO in this population [7]. The clinical presentations can be classified into three large groups. The first and most common presentation is dyspnoea with or without pleuritic chest pain and haemoptysis. The second presentation is haemodynamic instability and syncope, which is usually associated with massive embolism; the third and least common presentation mimics indolent pneumonia or heart failure, especially in the elderly [32, 33]. Temperatures generally higher than 38.3°C were observed in patients with extensive pulmonary infarction or in whom secondary pneumonitis had developed distal to the embolus [34, 35].

Hyperthyroidism is relatively common amongst elderly individuals. In the study by Knockaert et al.only one patient had hyperthyroidism [7]. In manycases, the symptoms are typical and include nervousness, palpitations, sweating, tremors and weight loss. However, often the most striking characteristic of hyperthyroidism in older patients is the paucity of symptoms, which they describe. They frequently do not have enlarged thyroids and do not complain of nervousness or heat intolerance. Rather than the usual picture of weight loss despite an increased appetite, older patients may lose their desire for food so that weight loss is very common. In fact, many older patients with hyperthyroidism are evaluated for depression or cancer before the correct diagnosis ismade. Therefore, it is apparent that hyperthyroidism can be easily overlooked in the elderly, as the presenting features may not conform to the usual finding in younger people.

Subacute thyroiditis with thyrotoxicosis may manifest as FUO [36, 37]. It is not a common cause of FUO, but it is found consistently across several series of FUO and, along with hyperthyroidism, is the most common endocrinologic cause. The authors suggest that subacute thyroiditis should be considered in elderly patients with FUO and elevations of ESR and alkaline phosphatase levels, even in the absence of symptoms suggestive of thyroid disease.

Drug fever

This finding is not unexpected because elderly patients take more drugs than younger ones. Commonly, several weeks elapse between initiation of the drug and onset of fever. Once the causative drug is stopped, fever almost always resolves itself within 2 days. The list of implicated agents is lengthy andincludes some drugs given to treat fever (e.g. aspirin, nonsteroidal anti-inflammatory drugs and antibiotics) [1].

Habitual hyperthermia

This entity has never been reported as a cause of FUO in the elderly, and only rarely meets the criteria of FUO, because most patients with habitual hyperthermia have a body temperature lower than 38°C [38].

Factitious fever

Factitious fever is a rare cause of FUO, it is more prevalent in young female patients allied with health professions, and was not a cause of FUO in the elderly population in most previous studies [7, 9, 10].

Approach to the elderly patient with FUO

Fever of unknown origin in elderly patients is a difficult diagnostic problem for a clinician who provides care to older people. The investigation of elderly patients, who have low tolerance to the long and exhausting FUO investigation, should be directed by the spectrum of underlying diseases rather than by reported results of selected diagnostic tests [7]. A reasonable approach to FUO in the elderly is to perform a thorough history and physical examination, focusing on symptoms and signs of intra-abdominal diseases, cardiac disorders, tuberculosis, musculoskeletal disorders and cancers (Table 3). After chest X-rays, and basic laboratory studies areperformed, imaging studies of the abdomen, repeated blood cultures and echocardiography should be carried out [39]. Elderly patients are more likely to have predisposing valvular conditions (e.g. degenerative and calcific lesions) and prosthetic valves, which decrease the sensitivity of transthoracic echocardiography to 45%. Transesophageal echocardiography has increased the diagnostic yield of infective endocarditis in the elderly patients by 45%[17]. A high index of suspicion and an aggressive diagnostic approach are required for early detection and treatment.

Table 3.  Investigation of FUO in the elderly
  1. Fuo, fever of unknown origin.

Suggested approach to FUO in the elderly
Confirm fever by serial measures
Through meticulous history, complete physical examination,  complete blood count with differential, erythrocyte  sedimentation rate, a chemistry panel (consisted liver enzymes),  urinalysis, multiple cultures of urine and blood, TSH.  Consider HIV, EBV, CMV and ANA in specific cases
Discontinue all nonessential drugs
Abdominal ultrasonography
Echocardiography
Search for tuberculosis
Computed tomography of abdomen and chest
Consider temporal artery biopsy, particularly  if the ESR exceeds 40 mm h−1
Search for pulmonary embolism, particularly in bedridden  patients
Scanning with gallium-67 or indium-111 labelled  autologous leucocytes
Consider liver and bone marrow biopsy
Explorative laparotomy

All nonessential drugs the patient takes should be immediately discontinued as well as other essential drugs if fever persists. The low cost and wealth of information makes abdominal ultrasonography a first choice technique in investigation of FUO.

Computed tomography of the chest and the abdomen constitutes the next step [7]. Abdominal CT scan has a major role in the detection of intra-abdominal pathology. These can be rewarding tests for patients with signs and symptoms suggesting an abdominal process but can also occasionally be useful for cases with no indication of disease elsewhere. This study rarely establishes a definitive diagnosis; instead it helps to identify abnormal tissue. This ultimately helps the clinician to identify the site where invasive procedures such as needle biopsy, aspiration or catheter placement are likely to be worthwhile [40]. The usefulness of abdominal CT and occasionally ultrasound scanning of the gallbladder and hepatobiliary system has resulted in application of these tests to virtually all cases of FUO. Because of their extensive use, tabulated in one study at more than three CT and/or ultrasound examinations per patient [5], the yield per test has been only about 10%[5, 7].

Scanning with gallium-67 or indium-111 labelled autologous leucocytes may be helpful when infection or malignancy is the cause of FUO, despite the rather high frequency of nonspecific findings. Indium-111 labelled leucocyte scanning showed poor sensitivity but good specificity both for infective and noninfective causes of FUO and is useful when combined with ultrasound. In one study sensitivities of leucocyte scan and ultrasound were 25% and 23% with specificities of 100% and 83%, respectively [41].

High pretest probability of temporal arteritis suggests that the diagnostic yield of temporal artery biopsy in older patients with FUO (with initial negative evaluations for other causes) would be high. In particular, FUO patients older than 60 years with anaemia, elevated sedimentation rate and an elevated alkaline phosphatase level should invoke a high index of suspicion for temporal arteritis. A markedly elevated sedimentation rate is considered to be a hallmark of temporal arteritis, but a sedimentation rate of <30 mm h−1 is found in 10% of patients [42, 43]. If initial evaluation rules out the more common diagnoses of endocarditis, intra-abdominal abscess and tuberculosis in these patients, then prompt temporal artery biopsy is warranted [7, 44]. A positive temporal artery biopsy result confirms the diagnosis but a negative result does not exclude it because of possible skip lesions. Temporal artery biopsy samples are positive only in 60–80% of patients [27]. It may exclude other systemic vasculitides such as polyarteritis nodosa and Wegener's granulomatosis [45]. If the biopsy is negative but clinical suspicion remains high, a contralateral biopsy should be considered but a low diagnostic yield was found [28, 46]. Recent studies suggest that, at some centres, colour duplex ultrasonography may be a useful noninvasive technique for the diagnosis of temporal arteritis [47]. A delay in diagnosis of temporal arteritis may lead to catastrophic results, such as blindness or stroke. It is therefore essential to maintain a high index of suspicion when managing older patients at risk. It is recommended to begin corticosteroid therapy soon after obtaining initial investigations and even prior to performing a temporal artery biopsy in elderly patients with FUO where there is a strong clinical suspicion of temporal arteritis and serious complications, such as imminent visual loss or other vascular events are suspected [48]. A dramatic response to glucocorticoid therapy can confirm the diagnosis.

A routine search for tuberculosis is advocated, because the presentation of tuberculosis in the elderly is often uncharacteristic in terms of both symptoms and chest radiographic features [11, 49]. Diagnosis may be difficult or may be missed, with the result being that it is too late to achieve a cure. The wide variation in presentation of disease in the elderly should alert the clinical team to perform appropriate diagnostic tests at an early stage in the disease process. The Mantoux method of administering the tuberculin test is the preferred method. Absence of reactions to the test does not exclude the diagnosis of tuberculosis. The tuberculin skin tests require careful interpretation because the healthy elderly individual may have positive test results as a consequence of previous infection, whereas the severely ill elderly patient with tuberculosis may be anergic and therefore have a negative test result. In one study skin test responses to 5TU PPD were positive in 86.2% of young adults with tuberculosis and only in 67.6% of elderly patients tested [11]. Impairment of the immunological responses of T lymphocytes can explain the lower frequency of positive tuberculin skin tests in older patients [50]. The tests should always include sending sputum for smear and culture and possibly bronchoscopy with lavage and biopsy or, rarely, open lung biopsy. Three sputum specimens, collected in the morning, should be submitted to the laboratory to microscopic examination. The sputum smears may be positive for acid-fast bacilli in only one-half of the cases. The use of specimens of urine and gastric lavage fluid are limited by the presence of mycobacterial commensals, which can cause false-positive results [1].

If disease is suspected at an extrapulmonary site, appropriate specimens should be sent for histology and culture. Two potentially useful tests are often unhelpful: the tuberculin skin test is frequently negative in the miliary and peritoneal forms, and the chest radiograph is normal in approximately 50% of most types extrapulmonary tuberculosis [11]. Lung and liver biopsy each demonstrate granulomas in 80–90% of cases of miliary tuberculosis, and bone-marrow biopsy is likely to show granulomas in only half the cases [1]. DNA amplification techniques such as polymerase chain reaction and gene probes may have an increasing part to play in the diagnosis of tuberculosis in the elderly [50]. The importance of tuberculosis as a cause of FUO in the elderly warrants empirical treatment with antituberculous agents if a rapid deterioration of the clinical condition occurs. Treatment should be started before diagnosis can be confirmed, because many patients are diagnosed on the basis of response to therapy alone [50, 51].

After this initial approach reveals no clues, bone marrow and liver biopsy should be considered [39]. Liver biopsy is essential for diagnosis and should be considered whenever a systemic granulematous disorder is suspected. Biopsy demonstrates granulomas and may provide histological evidence of the specific aetiology (e.g. caseation of tuberculosis, schistosomiasis, fungal organisms and primary biliary cirrhosis) [52–54].

Bone marrow biopsy is a useful procedure for the diagnosis of FUO in patients with advanced HIV disease, particularly in areas where tuberculosis is prevalent. Involvement of the marrow may be the first indication of the existence of extranodal lymphomas [55].

Pulmonary embolism must be actively searched for, particularly in bedridden patients. One schedules routine colon investigation after all these tests because this is a poorly tolerated procedure in elderly patients. Further investigation should mainly be directed by clues provided by preceding tests [7]. Exploratory laparotomy in the absence of localizing features is unusual these days. Laparoscopy, including laparoscopic liver biopsy, is a less traumatic alternative. It is most helpful when other features point to abdominal disease and has had a yield of only 20% when such features are absent [1].

Alternatively, if the patient is clinically stable, it is reasonable and prudent (especially in elderly patients who are more frail) to observe the patient in the ambulatory setting and repeat all noninvasive diagnostic studies at some later time or more specific manifestations appear. In elderly patients who are deteriorating clinically, a trial of 10–14 days of broad-spectrum antibiotics is warranted; after all diagnostic studies have been completed [39].

Should the criteria of FUO be changed for the elderly?

Is it necessary to modify the definition of FUO in the elderly?

Fever in the elderly can be defined as a persistent oral or tympanic membrane temperature ≥37.2°C or persistent rectal temperature ≥37.5°C. Moreover, an increase over baseline temperature ≥1.3°C, independent of site measured or device used, also indicates the presence of fever [8, 39]. Fever is a primary manifestation of infection, but the physiology of thermoregulation and pathogenesis of fever appears to be altered with advancing age. Elderly patients with serious infections may have blunted febrile response, which may delay diagnosis and treatment [39]. The physiological mechanisms for the blunted temperature responses to infection observed in the elderly have not been clearly elucidated. However, the basic pathogenesis of fever is becoming clearer and suggests alterations that possibly accompany ageing. Diminished thermoregulatory responses, such as sudomotor and vasomotor responses, as well as quantitative and qualitative abnormalities in both the production of and response to endogenous pyrogens, such as interleukin (IL)-1, IL-6 and tumour necrosis factor, may be possible explanations for the differences between elderly and young patients in fever response to infection [8]. With respect to these changes in thermoregulation it may be reasonable to consider as a FUO in the elderly, a body temperature of <38.3°C (at least the interval of 37.5–38.3°C). Within the modified definition of FUO the proportion of the various aetiologies in this population might be changed. Despite the apparent lack of consensus, we believe that the elderly require a redefinition of FUO.

Conclusions

Elderly patients with FUO have a different spectrum of underlying conditions.

Multisystem disease emerged as the most frequent cause of FUO in the elderly, and temporal arteritis is the most frequent specific diagnosis. Infections, particularly tuberculosis, remain an important group. The number of undiagnosed cases is significantly lower in elderly patients with FUO compared with younger individuals and diagnosis can be made in the majority of the cases. Evaluation of FUO in the elderly is complex and challenging. Atypical disease presentation is further complicated by the fact that multiple diseases commonly exist. Investigative procedures may be less well tolerated by older people. Thus the investigative pathway is more complex, with decision-making dependent upon clinical presentation, sensitivity and specificity of tests, side-effects and discomfort of the tests. In addition, invasive procedures are often not possible because the elderly patients are unable to cooperate in the test. Each patient with this syndrome requires an individual approach and the use of algorithms is inappropriate. One common mistake is to follow a preconceived protocol rather than investigate the abnormalities already uncovered and focus efforts around these.

In spite of all this, as FUO is often associated with treatable conditions in this age group, an intensive, accelerated evaluation is necessary as their lack of reserves within their various systems makes this population vulnerable to irreversible changes and functional deterioration. Therefore a delay in diagnosis and initiation of appropriate treatment may lead to increased morbidity and mortality. In elderly patients who are deteriorating clinically, early recognition and prompt initiation of appropriate empirical therapy are the cornerstones of the strategy.

Finally, with respect to the changes in physiology of thermoregulation in the elderly we believe that the definition of FUO must be changed in this population.

Ancillary