Diagnosing lung infection in HIV

The majority of lung infections are caused by one of three organisms.

When it comes to figuring out the type of pulmonary infection an HIV patient has, the CD4 count is your secret weapon, said Bradley A. Sharpe, MD, FACP, during the hospital medicine precourse at Internal Medicine 2013.

“CD4 count determines the risk of getting infected and may determine the clinical presentation, so it can be incredibly helpful in diagnosis and management,” said Dr. Sharpe, a professor of medicine at the University of California, San Francisco.

Bradley A Sharpe MD FACP Photo by Kevin Berne
Bradley A. Sharpe, MD, FACP. Photo by Kevin Berne.

Any trick is helpful, since pulmonary infection is 100 times more common in HIV patients than non-HIV patients, he said. Plus, the same symptoms could result from any one of dozens of diagnoses, from influenza to lung cancer to abacavir hypersensitivity.

“There's a long list of organisms that can cause pulmonary infection in HIV. The good news is that the epidemiology helps, and the list [of organisms usually encountered by] a noninfectious disease specialist is much simpler,” Dr. Sharpe said.

Pointing to a pie chart that compiled data from three studies, he noted that 45% of HIV patients with pulmonary infection have bacterial pneumonia, 29% have Pneumocystis jirovecii pneumonia (PCP), and 9% have tuberculosis (TB). “So if you can remember those three, that's getting 85% of the cases,” Dr. Sharpe said. “The other ones are going to be rare diseases we don't see as commonly.”

Bacterial pneumonia

Streptococcus pneumoniae is the most common cause of bacterial pneumonia, and it is more virulent and more likely to be invasive in HIV patients. “Bacteremia is incredibly common in HIV or AIDS patients with Streptococcus pneumoniae,” Dr. Sharpe said. “Remember that rates of bacterial pneumonia do depend on CD4 count—the lower the count goes, the more likely it is to be bacterial.”

Bacterial pneumonia presents in HIV patients the same as in non-HIV patients, he said: acute (over a few days) onset of fever, chills, productive cough and shortness of breath. “The one clinical pearl I can give you is from a classic study [AIDS. 1998;12:885-93] of clinical presentations of different types of infection,” Dr. Sharpe said. “In the setting of suspected pneumonia in an HIV patient, reporting a fever of seven days or less had an odds ratio of 6.6 at predicting bacterial pneumonia.”

So if you ask a patient who has HIV or AIDS how long she has had a fever, and she says just a couple of days, bacterial infection may be more likely than if she reports having a fever for two weeks, he explained.

Chest X-rays in patients with bacterial pneumonia typically will have unilateral, focal or lobal consolidation, he added.

“In fact, a lobar infiltrate also has a fairly high positive odds ratio—5.8—for predicting bacterial pneumonia. So an X-ray doesn't predict the organism, but in the setting of HIV or AIDS, it might push you more towards bacterial infection if they do have a more focal consolidation,” Dr. Sharpe said. “These patients also may present with multifocal pneumonia or parapneumonic effusions.”


Ninety-five percent of patients with PCP have a CD4 count less than 200 cells/µL, Dr. Sharpe said.

Unlike with bacterial infection, which has an acute presentation, the presentation for PCP is often subacute, with patients having days to weeks of symptoms. More than 80% of patients have fever, 95% have cough, and 95% have shortness of breath. Exertional dyspnea is also very common, he said.

“The patient will say, ‘When I'm at home watching TV I'm fine, but if I get up to go upstairs, by the time I'm at the top of the steps I am completely out of breath.’ That may be a predictor of PCP,” Dr. Sharpe said.

On exam, fever, tachypnea and hypoxia are incredibly common, he noted, yet half of patients will have a normal pulmonary exam. “So a normal lung exam in a patient with HIV or suspected HIV shouldn't dissuade you from thinking they might have PCP,” he said.

Along with a CD4 count less than 200 cells/µL, a big help in recognizing PCP is lactate dehydrogenase (LDH) level, which has a sensitivity of greater than 85%, Dr. Sharpe said.

“LDH is a relatively inexpensive test and the evidence supports its use. If the level is high, it is actually not helpful, because LDH can be elevated in any pulmonary infection. But if it's normal, the infection is much less likely to be PCP, since PCP causes an elevated LDH in almost every patient,” Dr. Sharpe said.

The typical chest X-ray with PCP shows diffuse, bilateral, symmetrical interstitial infiltrates. However, patients could also have pneumothorax, lobar infiltrate or nodules—and up to 10% of patients have a normal chest X-ray, he said.

“So, think of that patient you are admitting to the hospital who you think might have PCP because of fever, shortness of breath, cough, but then the X-ray is normal. Well, the X-ray doesn't rule it out,” Dr. Sharpe said.


Although TB is less common than other causes of pulmonary infection, it's something hospitalists need to be vigilant about because it's so highly infectious, Dr. Sharpe noted.

While more common at lower CD4 counts, TB can be present at any CD4 count, he added. “So if you're admitting a patient on [antiretroviral therapy] with a CD4 level of 900, it's reasonable to say it's not PCP because of that CD4, but you can't say it's not TB,” Dr. Sharpe said.

As with non-HIV patients, the exposure history matters. Find out whether the patient has been homeless, in a correctional facility, or exposed to someone with TB when trying to determine the diagnosis.

TB presentation is similar to that in non-HIV patients, with fever, cough, weight loss and night sweats for weeks to months. Dr. Sharpe said, “I now have built into my review of systems for HIV and pulmonary infection, ‘Tell me, has your weight changed, have you lost a few pounds?’”

At 350 cells/µL or above, the chest X-ray for TB is similar to non-HIV patients, with upper-lobe infiltrates, often with cavitation. At a lower count, “basically the X-ray can show anything, which makes our clinical work hard,” Dr. Sharpe said.

According to research on chest X-rays in TB, 36% of the time the findings suggest primary TB, 28% of the time they indicate reactivation TB, 13% of the time they indicate atypical infiltrates that look like PCP, and 19% of the time they show minimal change or a normal X-ray. “So my take-home point for TB in patients with HIV/AIDS or suspected HIV/AIDS is that your suspicion needs to be quite high, especially as a CD4 count gets lower,” Dr. Sharpe said.

Hospitalists should have the Centers for Disease Control and Prevention (CDC) recommendations for isolation of suspected TB patients handy (available online, “but my recommendation would be to err on the side of isolation, because these patients may not present in an obvious manner,” he added.


Empiric treatment of bacterial pneumonia is the same for HIV patients as non-HIV patients, Dr. Sharpe said. If the patient is admitted to a non-ICU bed, you give ceftriaxone and macrolide. “You might use a fluoroquinolone, although if you are worried about TB you might want to avoid it, as it can impact future therapy or breed resistance,” he said.

And, he noted, “A run-of-the-mill patient who presents with HIV or AIDS and what looks like bacterial pneumonia doesn't need broader coverage than a non-HIV/AIDS patient, as a general principle.”

Treatment of a patient with HIV who requires ICU admission for their pneumonia is also the same as in non-HIV patients. Dr. Sharpe did provide the caveat that if the CD4 level is really low, “like under 50,” the patient might be at risk for Pseudomonas and would need to be treated with piperacillin/tazobactam plus a fluoroquinolone and vancomycin, Dr. Sharpe said.

If you suspect PCP clinically, you should treat it empirically, he added. Trimethoprim/sulfamethoxazole is the treatment of choice; add steroids if the partial pressure of arterial oxygen (Pao2) is less than 70 mm Hg or the alveolar-arterial gradient is greater than 35 mm Hg, he said.

“Unlike CAP, where seven days is enough, PCP actually requires 21 days of treatment,” Dr. Sharpe said.

Treatment selection and duration for other pulmonary infections depend on the diagnosis, he added. “The decision to treat TB acutely is complex. You might want to get help from [an] ID consult, especially if it seems complicated,” he said.