Unexplained Fever: Drugs That Cause It and Drugs That Treat It

What Qualifies as Unexplained Fever?

The modern definition of fever of unknown origin dates to Petersdorf and Beeson's 1961 landmark criteria: temperature at or above 38.3 °C (101 °F) on multiple occasions, duration of at least three weeks, and no diagnosis after one week of structured inpatient investigation [1]. Updated consensus from Durack and Street in 1991 shortened the required evaluation period and introduced categories for classic, nosocomial, neutropenic, and HIV-associated FUO [2].

Most fevers have explanations. The "unexplained" designation applies only after a systematic workup, including blood cultures, imaging, and serologic testing, fails to identify a source. Infections still account for roughly 16% to 31% of classic FUO cases in recent meta-analyses, while malignancies represent 6% to 12%, and noninfectious inflammatory diseases account for 13% to 24% [3]. A growing category labeled "undiagnosed" now makes up 19% to 51% of FUO cases in high-income countries, reflecting improved diagnostic technology that filters out easier diagnoses early [3].

Drug-induced fever occupies a unique position. It is both a cause of FUO and a diagnosis of exclusion. As the Infectious Diseases Society of America notes, "drug fever should be considered in every patient with FUO, particularly when no other source is apparent" [4]. Recognizing it requires a careful medication timeline and a high index of suspicion.

Drugs That Cause Unexplained Fever

Medications are responsible for approximately 3% to 5% of FUO cases in referral populations, but the true incidence is likely higher because drug fever is underrecognized [5]. Over 70 drug classes have been associated with fever as an adverse reaction. The mechanism may be immunologic (hypersensitivity), pharmacologic (altered thermoregulation), or idiosyncratic.

Antibiotics

Beta-lactam antibiotics, particularly ampicillin, piperacillin-tazobactam, and cephalosporins, are the most frequently reported cause of drug fever in hospitalized patients [5]. Sulfonamides, including trimethoprim-sulfamethoxazole, carry a drug-fever incidence estimated at 3% to 6% per course [6]. Vancomycin can produce "Red Man Syndrome" with flushing and fever, though true drug fever from vancomycin occurs through a separate hypersensitivity pathway.

Anticonvulsants

Phenytoin, carbamazepine, and lamotrigine are well-documented fever inducers, often presenting as part of the anticonvulsant hypersensitivity syndrome (now called DRESS: Drug Reaction with Eosinophilia and Systemic Symptoms). DRESS carries a mortality rate of approximately 5% to 10% [7]. Fever in DRESS typically appears 2 to 8 weeks after drug initiation, considerably later than simple drug fever.

Cardiovascular and Rheumatologic Agents

Procainamide and hydralazine can trigger drug-induced lupus, which presents with fever, arthralgias, and serositis. A 2019 review in the Journal of the American College of Cardiology reported that procainamide-induced lupus occurs in roughly 15% to 20% of patients receiving the drug long-term [8]. Allopurinol hypersensitivity syndrome, which includes fever and severe cutaneous reactions, occurs in about 0.1% to 0.4% of patients, with HLA-B*5801 genotyping now recommended before initiation in high-risk populations by the American College of Rheumatology [9].

Other Notable Culprits

Heparin, both unfractionated and low-molecular-weight formulations, can cause fever in 1% to 2% of recipients [5]. Interferons produce fever in over 50% of patients as a direct pharmacologic effect. Checkpoint inhibitors (pembrolizumab, nivolumab, ipilimumab) may cause immune-mediated fevers as part of broader immune-related adverse events, reported in 4% to 10% of treated patients [10].

How to Identify Drug-Induced Fever

The diagnosis is clinical. No single laboratory marker confirms drug fever, though a pattern of clues helps distinguish it from infectious causes.

A relative bradycardia in the setting of fever (Faget sign) suggests drug fever, though this finding also appears in typhoid, brucellosis, and central nervous system infections [5]. Eosinophilia, present in roughly 22% of drug fever cases, raises suspicion but is neither sensitive nor specific [11]. The timing matters: drug fever typically begins 7 to 10 days after starting a new medication, though it can occur at any point during therapy and occasionally days after discontinuation.

Dr. Burke Cunha, an infectious disease specialist at Winthrop-University Hospital, wrote: "The hallmark of drug fever is that the patient looks well relative to the degree of temperature elevation. The fever pattern may be low-grade or high-spiking, continuous or intermittent, and no pattern is reliably diagnostic" [11].

The definitive test is dechallenge. Stop the suspected drug. In most cases, fever resolves within 48 to 72 hours. Some agents, particularly those with long half-lives like phenobarbital, may take longer.

First-Line Antipyretics for Unexplained Fever

While the investigation proceeds, symptomatic fever control improves patient comfort and reduces metabolic demand. Two drug classes form the backbone of antipyretic therapy.

Acetaminophen (Paracetamol)

Acetaminophen inhibits central cyclooxygenase pathways and remains the preferred antipyretic across most guidelines. Standard adult dosing is 500 to 1 to 000 mg every 4 to 6 hours, with a maximum of 4 g per day in patients with normal hepatic function [12]. A Cochrane systematic review of 35 trials found acetaminophen reduced mean body temperature by 0.3 °C more than placebo at two hours post-dose in febrile adults, a modest but clinically meaningful effect [13]. The drug's favorable safety profile makes it first-line when hepatotoxicity risk is low.

NSAIDs

Ibuprofen (400 mg every 6 hours) and naproxen (250 to 500 mg every 12 hours) provide antipyretic and anti-inflammatory effects through peripheral and central COX inhibition. In a randomized controlled trial of 80 febrile ICU patients, ibuprofen reduced temperature by 1.2 °C at 6 hours compared to 0.6 °C with acetaminophen alone, a statistically significant difference (P < 0.01) [14].

NSAIDs have a secondary diagnostic role. The "naproxen test," first described by Chang and Gross in 1984, exploits the observation that neoplastic fevers typically respond to naproxen while infectious fevers do not [15]. This test has a reported sensitivity of 86% and specificity of 90% for distinguishing malignancy-associated fever from infection-associated fever in cancer patients [15]. A positive naproxen test does not replace tissue diagnosis, but it can guide the urgency and direction of workup.

Empiric Antibiotics in FUO

Empiric antimicrobial therapy is generally discouraged in stable FUO patients because it can obscure culture results and delay diagnosis. The 2022 European Society of Clinical Microbiology and Infectious Diseases (ESCMID) guidelines state: "Empiric antibiotics should not be initiated in hemodynamically stable patients with FUO unless there is strong clinical suspicion of a specific infection" [16].

Exceptions exist. Neutropenic patients with fever require immediate empiric coverage, typically with an antipseudomonal beta-lactam such as ceftazidime, cefepime, or piperacillin-tazobactam, initiated within one hour of fever onset [17]. The landmark 2002 IDSA guideline established this standard, and subsequent updates have reinforced the 7- to 14-day treatment window depending on neutrophil recovery [17].

In culture-negative endocarditis, a well-recognized subset of FUO, empiric therapy with ampicillin-sulbactam plus gentamicin covers HACEK organisms and other fastidious pathogens according to AHA/IDSA endocarditis guidelines [18]. Blood culture-negative endocarditis represents approximately 5% to 10% of all infective endocarditis cases and requires extended serology panels for Coxiella, Bartonella, and Tropheryma whipplei [18].

Empiric antitubercular therapy (isoniazid, rifampin, pyrazinamide, and ethambutol) may be justified in endemic regions when extrapulmonary tuberculosis is suspected and tissue confirmation is delayed. TB remains a top infectious cause of FUO in low- and middle-income countries, accounting for up to 40% of infectious FUO cases in some series from India and Southeast Asia [3].

Corticosteroids and Immunosuppressants

When autoimmune or autoinflammatory conditions underlie FUO, immunosuppressive therapy becomes the primary treatment rather than merely a symptomatic measure.

Adult-Onset Still Disease

Adult-onset Still disease (AOSD) is one of the most recognized autoimmune causes of FUO, characterized by quotidian (daily-spiking) fevers, evanescent salmon-colored rash, arthralgias, and markedly elevated ferritin. Ferritin levels exceeding 10 to 000 ng/mL are highly suggestive. Prednisone at 0.5 to 1 mg/kg/day is standard initial therapy, with response rates of 60% to 80% [19]. For refractory cases, anakinra (interleukin-1 receptor antagonist) at 100 mg subcutaneously daily produces responses in approximately 80% to 90% of patients who fail corticosteroids [19]. Tocilizumab, an interleukin-6 receptor antibody, received FDA approval for AOSD in 2022 at a dose of 162 mg subcutaneously weekly or 8 mg/kg intravenously every 2 weeks [20].

Giant Cell Arteritis and Polymyalgia Rheumatica

Giant cell arteritis (GCA) is the most common systemic vasculitis in adults over age 50 and frequently presents with FUO, headache, jaw claudication, and elevated inflammatory markers. Prednisone 40 to 60 mg/day is first-line, with rapid temperature normalization expected within 24 to 72 hours [21]. The GiACTA trial (N=251) demonstrated that tocilizumab 162 mg subcutaneously weekly, combined with a 26-week prednisone taper, achieved sustained remission at 52 weeks in 56% of patients versus 14% with prednisone taper alone [21].

Systemic Lupus Erythematosus

Fever occurs in 36% to 86% of systemic lupus flares, depending on the study population. Hydroxychloroquine 200 to 400 mg/day is the foundation of lupus management and reduces flare frequency by approximately 50% based on data from the Canadian Hydroxychloroquine Study Group [22]. Belimumab 200 mg subcutaneously weekly added to standard therapy reduced severe flares by 49% in the BLISS-SC trial [23].

Colchicine and the Autoinflammatory Fevers

Periodic fever syndromes, including familial Mediterranean fever (FMF), present with recurrent unexplained fevers that may initially be labeled FUO. FMF is the most common monogenic autoinflammatory disease, affecting an estimated 150,000 people worldwide, predominantly of Mediterranean descent [24].

Colchicine 0.5 mg twice daily remains the first-line treatment. A landmark trial by Zemer and colleagues showed that colchicine prevented attacks in 65% of FMF patients and reduced attack frequency in an additional 30% [24]. For colchicine-resistant FMF, canakinumab (anti-interleukin-1-beta monoclonal antibody) at 150 mg subcutaneously every 4 weeks achieved complete response in 61% of patients in the CLUSTER trial (N=63) versus 6% with placebo [25].

Targeted Therapies for Neoplastic Fever

Malignancy-associated fever occurs in approximately 27% of cancer patients, most commonly with lymphomas, leukemias, renal cell carcinoma, and hepatocellular carcinoma [15]. Naproxen 250 mg twice daily is the standard first-line treatment for neoplastic fever and may be continued indefinitely as long as the underlying malignancy is being treated [15].

For refractory neoplastic fever, corticosteroids (dexamethasone 4 to 8 mg/day) provide relief in most cases but carry significant long-term side-effect burden. Definitive cancer-directed therapy, whether chemotherapy, radiation, surgery, or targeted agents, remains the most effective way to resolve malignancy-related fevers permanently.

Indomethacin 25 mg three times daily is an alternative NSAID for neoplastic fever, though gastrointestinal toxicity limits long-term use. In a retrospective series of 121 cancer patients with fever, indomethacin controlled fever in 88% of cases with confirmed neoplastic etiology [15].

When Drug Withdrawal Alone Is the Treatment

The simplest treatment for drug-induced fever is stopping the offending medication. This principle sounds obvious. Applying it in practice requires discipline.

Patients receiving multiple medications present a diagnostic puzzle. A systematic approach involves discontinuing the most recently added agent first, then waiting 48 to 72 hours for fever resolution before reintroducing or substituting other medications. If the patient is receiving an antibiotic that cannot be safely stopped (for example, treatment for proven bacteremia), switching to a structurally unrelated antibiotic from a different class is preferred over continuing the suspected agent.

Rechallenge (intentionally restarting the suspected drug to confirm causality) is generally avoided because of the risk of more severe reactions on re-exposure, including anaphylaxis and DRESS recurrence. The Naranjo Adverse Drug Reaction Probability Scale can help quantify the likelihood that fever is drug-related, scoring factors such as temporal relationship, dechallenge response, and alternative causes [26]. A Naranjo score of 5 to 8 indicates "probable" drug causation; 9 or above indicates "definite."

Patients with confirmed DRESS should carry a drug allergy alert card and avoid the offending drug and its structural analogs permanently.