What causes fever of unknown origin?

That might sound like “Who’s buried in Grant’s tomb?” or “When was the war of 1812?” but it’s actually a really good question.

First, a definition. Before you can officially “diagnose” someone with FUO, you have to meet a few criteria. The original criteria (back in 1961) were:

A fever of 101º F or more on several occasions
…persisting for over three weeks
…with no diagnosis despite a week-long, in-hospital investigation.

These days, we don’t keep people in the hospital for a week just to “investigate” things, unless they are really sick. Plus, things have changed a bit with the emergence of HIV, and with the development of new chemotherapeutic and immune-modulating drugs.

So there are now four different categories of FUO – classic, nosocomial, immune-deficient, and HIV-associated – each with its own definition and differential diagnosis. This is a good way to do it, because it limits the number of possibilities you need to investigate.

Classic FUO
Patients in this category meet the original criteria for fever (>101º F on several occasions over a three-week period), but they don’t have to be in the hospital for a week. They just have to have no diagnosis despite:

3 outpatient visits
3 days in the hospital, or
1 week of intensive outpatient testing.

The most common causes of FUO in these patients are:

Infections (30-40% of patients), such as abdominal or pelvic abscesses, tuberculosis, urinary tract infections, endocarditis, osteomyelitis, and liver infections.
Tumors (20-30% of patients), such as leukemia, lymphoma, and renal cell carcinoma (these can be difficult to diagnose)
Autoimmune disorders (10-20% of patients), such as Still’s disease and temporal arteritis (lupus and rheumatoid arthritis used to head up this list, but they’ve become much easier to diagnose)

After those are ruled out, there are a list of miscellaneous disorders, like drug-related fever, alcoholic hepatitis, and granulomatous disorders, that aren’t really grouped into a single etiologic category. And then, of course, there remain a small number of patients in whom no diagnosis can be made.

Nosocomial FUO
This category includes patients with a fever who have been in the hospital for at least a day, and who had no obvious source of infection before admission. You have to have looked carefully for at least three days before you can call it an official nosocomial FUO. There are a bunch of things that can cause fever in this group of patients, including catheter-related infections, drug-related fever, and pulmonary embolism.

Immune-deficient FUO
This is defined as recurrent fever in a patient with a neutrophil count less than 500/mm³. Again, you have to look for three days without finding anything before you can officially call it immune-deficient FUO. Most of the time, these patients turn out to have – not surprisingly – an opportunistic infection.

HIV-associated FUO
Patients with HIV can get fevers for all kinds of reasons. To call it FUO, the patient needs to have had recurrent fevers over a 4-week period as an outpatient, or over a 3-day period as an inpatient. The differential diagnosis in this group of patients includes infectious things (like Mycobacterium avium-intracellulare complex, Pneumocystis jirovecii, and CMV), tumors (lymphoma and Kaposi sarcoma), and drug-related fevers.

The workup of a patient with FUO involves taking a really detailed history, doing an extensive physical exam, and ordering appropriate labs and diagnostic studies. Here’s a readable article that summarizes the process.