Empiric Outpatient Management of HIV-Related Pneumonia: Economical or Unwise?

When immunologically normal patients present with pneumonia that is not severe enough to require immediate hospitalization, most health care providers initiate empiric oral therapy with a macrolide or tetracycline after a radiologic and laboratory evaluation that rarely extends further than a chest radiograph, leukocyte count determination, and perhaps a Gram stain of sputum [1]. Empiric therapy is directed against the common, treatable, community-acquired pathogens, such as Streptococcus pneumoniae, Haemophilus influenzae, Mycoplasma pneumoniae, Chlamydia pneumoniae, and Legionella species.

For immunosuppressed patients, health care providers have hesitated to use such an empiric management approach because the range of potential pathogens is so broad that a comprehensive empiric regimen is difficult to administer, especially on an outpatient basis. Moreover, in many immunosuppressed patient populations, patients may deteriorate precipitously, making outpatient management unwise. This potential for rapid deterioration also encourages prompt diagnostic intervention at a time when patients' oxygenation and blood pressure allow them to tolerate sputum induction or bronchoscopy. Finally, early initiation of specific and appropriate therapy maximizes the likelihood of a favorable outcome and minimizes toxicity in patient populations that often have compromised organ function because of underlying disease or various drug therapies.

Many patients with human immunodeficiency virus (HIV) infection are now being followed by primary care practitioners in outpatients settings at which facilities for bronchoscopy and the examination of induced sputum are not conveniently available. There is also severe pressure to minimize the number of costly procedures that do not definitively alter outcome. In the United States, an estimated 1 million persons have HIV infection, and 25% of these persons (250 000) have CD4 counts less than 200 cells/mm³[2]. A decade ago, empiric management of pneumonia in patients with low CD4 counts was anathema to many specialists. In the interim, empiric management of complications related to the acquired immunodeficiency syndrome (AIDS) has gained a foothold in at least one area: Physicians now widely use empiric therapy for mass lesions in the central nervous system that are suspected to be caused by Toxoplasma gondii[3]. For HIV-infected patients, is empiric outpatient management of pneumonia now a reasonable strategy, or are financial pressures pushing health care providers into imprudent and unsafe practices?

For patients with HIV infection and CD4 counts less than 200 cells/mm³, studies in the last decade have provided considerable information about the causes and natural history of pulmonary disease [4, 5]. Although many pathogens can cause pneumonia in such patients, the major concern is pneumocystis pneumonia. This entity has a characteristic presentation (unusual presentations, however, do occur) and usually progresses slowly. Pneumonia caused by S. pneumoniae and H. influenzae can have similar presentations, although the radiologic presentation of these bacterial processes is usually not diffuse and symmetric. Chlamydia pneumoniae, M. pneumoniae, Legionella species, and M. tuberculosis, as well as other fungi and viruses, also cause diffuse infiltrates in patients with HIV infection, and all merit consideration. Overall, however, the opportunity for empiric management in patients with HIV infection and pneumonia is more attractive than opportunities in many other immunosuppressed populations whose differential diagnoses are far broader and less predictable and whose potential for rapid deterioration is greater. Because oral regimens are available that can effectively treat infection with these pathogens, economic outpatient management is possible.

Experience and data now seem to support an empiric approach to treating pneumonia in a sub-population of reliable, compliant patients with HIV infection and CD4 counts less than 200 cells/mm³. The empiric approach should focus on recognizing and treating pneumocystis pneumonia (with broad coverage to treat other reasonable possibilities) using a highly effective oral regimen consisting of antipneumocystis drugs (trimethoprim-sulfamethoxazole, trimethoprim-dapsone, or perhaps clindamycin-primaquine) plus a macrolide (erythromycin, clarithroymcin, or azithromycin). If the patient improves, a 14- to 21-day regimen should be completed. If the patient's condition deteriorates or has not improved after 4 to 5 days, a diagnostic evaluation with some combination of induced sputum, bronchoalveolar lavage, and transbronchial biopsy should be initiated, usually in an inpatient setting. (In our experience, pulmonary function tests and nuclear scanning techniques are not specific enough to be helpful as part of this diagnostic evaluation.) The subpopulation of patients who are eligible for empiric therapy should meet certain well-defined criteria, but these criteria, although logical, have not been validated by large prospective trials.

To be eligible, the patient should have mild disease. Ideally, a room-air arterial blood gas determination should document the PO₂ to be greater than 70 to 80 mm Hg, or a pulse oximeter should show a saturation greater than 95% [6, 7]. Patients should not have severe exercise intolerance or a history suggesting rapid deterioration.

The chest radiograph should be typical of uncomplicated pneumocystis pneumonia, consisting of symmetric, interstitial infiltrates. Lobar infiltrates or bronchopneumonias merit at least a Gram stain of the sputum, and the antimicrobial regimen should focus on bacterial pathogens, not pneumocystis. Admittedly, pneumocystis pneumonia can present with any radiologic manifestation, but unusual manifestations warrant a more precise diagnostic evaluation. Blebs, cavities, and pleural effusions suggest advanced disease, complications of pneumocystis pneumonia, or another process that requires more diagnostic scrutiny. Diffuse alveolar infiltrates might be manageable on an outpatient basis, but many of these patients have substantial hypoxemia or impressive symptoms, indicating a need for hospitalization.

The patient should have been receiving either no prophylaxis or a preventive regimen other than trimethoprim-sulfamethoxazole. Although pneumocystis pneumonia can occur despite an appropriate trimethoprim-sulfamethoxazole regimen, especially in patients with CD4 counts less than 100 cells/mm³, development of disease under these circumstances is sufficiently unusual that attention should focus on other pathogens [8-11]. Streptococcus pneumoniae and H. influenzae are also less likely to develop in patients receiving trimethoprim-sulfamethoxazole, thus directing attention to “atypical organisms,” such as chlamydia, mycoplasma, or legionella, or to other opportunistic processes. Dapsone, dapsone-pyrimethamine, and aerosol pentamidine are effective antipneumocystis regimens, but breakthrough episodes of pneumocystis pneumonia are much more common with these regimens than with trimethoprim-sulfamethoxazole [8-11].

The patient should be reliable, compliant, and able to tolerate oral regimens. In some settings, establishing a specific diagnosis, teaching the patient about the disease, and formulating a specific plan is so important that hospitalization may be warranted and outpatient management may not be practical. Nausea, vomiting, or diarrhea render oral regimens imprudent: Patients with these conditions require hospitalization, and the economics of empiric management thus become minor compared with the cost of hospitalization. Patients must also comply with their regimens and must be likely to promptly contact their health care provider if they cannot tolerate their drug regimen or if they become clinically worse. This caveat means that, in some health care settings, few patients will be good candidates for outpatient management.

Finally, unusual epidemiologic features in terms of causative organisms may make an empiric approach unwise in certain communities. If tuberculosis, histoplasmosis, or coccidiomycosis are especially common, as is the case in certain cities in North America or in some areas in the developing world, optimal management might necessitate more emphasis on specific diagnosis for all patients.

What are the disadvantages of this approach? A major concern with severely immunocompromised patients is that the empiric regimens will not have antimicrobial activity against the true causative pathogen and that the patient's prognosis will be adversely affected by the delay in therapy. If patients with mild disease and typical presentations are chosen and monitored closely, the current approach seems reasonable [12-16]. Testing this approach in a prospective study is desirable, but funding a study large enough to yield useful results is problematic.

Disease caused by Mycobacterium tuberculosis is a special concern because the recommended empiric regimen is not active against this pathogen; patients will not receive a diagnosis and thus will be contagious for a few additional days [16, 17]. Because tuberculosis progresses slowly, a delay in therapy of a few days is not likely to adversely affect outcome. Because exposure has already occurred, the failure to use respiratory precautions is less of a problem at home (although more prolonged exposure and additional community contacts are hardly desirable) than in a health care setting, where new patients, visitors, and employees will be exposed. In some communities that have a high frequency of tuberculosis, empiric management is inappropriate: Identifying cases so that appropriate isolation precautions and therapy are initiated may take precedence [17].

If a patient does have pneumocystis pneumonia but deteriorates despite therapy or does not improve, will 5 days of therapy reduce the likelihood that examination of bronchoalveolar lavage fluid can help identify pneumocystis as the causative pathogen? Although the data are not extensive, the literature suggests that organisms are detectable during the first week [18].

Empiric management of pneumonia in some patients with HIV infection is reasonable if patients are selected judiciously [12-15]. Such approaches should be validated by high-quality studies in various community settings. Even more important, however, is identifying patients with HIV infection before they are severely immunodeficient and instituting the type of comprehensive preventive strategy recently recommended by the U.S. Public Health Service and Infectious Diseases Society of America and endorsed by other leading professional societies [2]. Such a comprehensive preventive program should substantially reduce the frequency of pulmonary complications.

If empiric management is reasonable for patients with clinically mild pneumonia and CD4 counts less than 200 cells/mm³, is it reasonable for patients with HIV infection and CD4 counts greater than 200 cells/mm³? In this population, the differential diagnosis is broad: There is no single pathogen, such as Pneumocystis carinii, that warrants focused attention. In many ways, it is reasonable to approach this population as one would an immunocompetent population, with attention focused on community-acquired respiratory viruses, chlamydia, mycoplasma, legionella, haemophilus, pneumococcus, and, in some communities, tuberculosis [1]. Clinicians must keep in mind, however, that endemic fungi (Cryptococcus, Histoplasma, Coccidioides, and perhaps Blastomyces) enter the differential diagnosis, as do other unusual processes. Clinicians must also recognize that the CD4 count is not an absolute marker for susceptibility to infection with the more ominous pathogens. Even if a recent CD4 count was greater than 200 cells/mm³, a patient with oral candidiasis, prolonged fever, and perhaps weight loss should be treated as if his or her CD4 count was less than 200 cells/mm³. Similarly, CD4 counts can decrease precipitously, especially if the results of quantitative HIV studies (available to some clinicians) show increasing viral titers.

In 1996, empiric therapy for AIDS-related pneumonia seems appropriate in well-defined subpopulations of patients if convenient, cost-effective diagnostic facilities are not readily available. The approach is not without risk, however, and requires careful evaluation of patients to assure that they are suitable candidates.

• Copyright ©2004 by the American College of Physicians