Case-fatality rates for treated Legionnaires disease depend on various factors, but the chief determinants seem to be 1) the premorbid immune and general health status of the patient and 2) whether the disease is community-acquired or nosocomial [3-6]. Higher fatality rates for nosocomial infection are probably due to host-related rather than organism-related factors. Previously healthy patients with community-acquired Legionnaires disease who do not require hospitalization respond well to erythromycin therapy about 95% to 99% of the time. Quite a different picture emerges for immunocompromised patients with Legionnaires disease or for patients with nosocomial Legionnaires disease. Fatality rates of up to 50% have been reported for treated patients in these categories, with a very wide range; fatality rates as high as 80% have been found in untreated patients [4-10]. Prolonged illness before therapy and respiratory failure caused by pneumonia are especially poor prognostic factors. Relapse after cessation of erythromycin therapy can be a serious problem in some immunosuppressed patients and can require prolonged and repeated courses of therapy [3, 11, 12]. The occasional ineffectiveness of erythromycin therapy was recently highlighted in a report of fatal Legionella micdadei pneumonia in an immunocompromised patient who died despite receiving multiple courses of erythromycin therapy [13].

In addition to giving rise to potentially poor outcomes in patients with severe Legionnaires disease, erythromycin causes numerous unpleasant side effects. These include high rates of phlebitis and infusion pain with intravenous administration and frequent gastrointestinal problems with oral administration. Large intravenous infusion volumes are often required to increase patient tolerance, and this can exacerbate fluid management problems in critically ill patients. Other possible problems include cardiac arrhythmias and conduction abnormalities, hepatitis, hearing loss, interference with the metabolism of numerous drugs, and poor patient tolerance overall. Of adults who receive outpatient therapy with erythromycin, 12% to 33% do not complete therapy because of intolerance; this contrasts with the much lower dropout rates for most other antibiotics [14, 15]. Because of adverse effects, hospitalized patients receiving intravenous erythromycin may have significantly increased lengths of stay and increased costs of hospitalization compared with patients receiving other agents [16]. In some outpatients, the frequency of adverse effects may mean that the long courses of therapy required to cure Legionnaires disease are not completed, resulting in relapse.

If valid comparative clinical studies cannot be done, how can the value of new drugs be assessed? Luckily, certain types of laboratory studies have correlated very well with clinical response to infection [3]. For example, animal models and intracellular infection models correctly predict that erythromycin is clinically effective and that cefoxitin is not. The success of these correlations is based on the intracellular location of L. pneumophila and the ability of drugs to achieve high, bioactive, intracellular concentrations. It is less certain that an antimicrobial agent that is highly active in such experiments is more effective clinically than a drug (such as erythromycin) that is marginally effective in laboratory tests. The lack of controlled clinical trials means that this question is not easily answered. One retrospective study [10] showed that pefloxacin is more active clinically than erythromycin; this finding correlates with the superior laboratory activity of pefloxacin.

In intracellular models of L. pneumophila infection, bacterial growth is inhibited by erythromycin and rifampin but recurs when the drugs are removed from the cells [3]. Thus, these drugs differ from many newer fluoroquinolones, which can kill most, if not all, of the bacteria present in the infected cells and can be removed from the cells without causing rapid bacterial regrowth [3]. Against intracellular L. pneumophila, levofloxacin, sparfloxacin, and trovafloxacin are the most active fluoroquinolone antimicrobial agents in the laboratory that are licensed by the U.S. Food and Drug Administration (FDA). Azithromycin is also more active than erythromycin against intracellular L. pneumophila.

Studies of animal models of L. pneumophila pneumonia confirm the superior activity of most fluoroquinolone antimicrobial agents and azithromycin compared with erythromycin [3, 17]. In these experiments, a single low dose of azithromycin not only is curative but also markedly reduces lung concentrations of L. pneumophila. In contrast, multiple, high doses of erythromycin are needed to achieve a similar cure rate without the benefit of rapid lung sterilization. Extremely low dosages of trovafloxacin or sparfloxacin also have had curative and sterilizing effects in animal studies. The failure of erythromycin treatment to completely eradicate L. pneumophila from guinea pigs with experimental pneumonia correlates with disease relapse in some heavily immunosuppressed patients after cessation of erythromycin therapy [3, 17].

Uncontrolled clinical trials show that many new fluoroquinolone antimicrobial agents effectively treat both community-acquired and nosocomial Legionnaires disease; these agents include ciprofloxacin, pefloxacin, ofloxacin, levofloxacin, and trovafloxacin (trovafloxacin has been studied against community-acquired disease only) [3, 18, 19]. Similar clinical data exist for azithromycin [3]. On the other hand, treatment failures have been reported for these newer antimicrobial agents, as they have for erythromycin, and use of these agents is not a guarantee of therapeutic success [3]. Laboratory evidence suggests that the most active fluoroquinolone antimicrobial agents are trovafloxacin, sparfloxacin, and pefloxacin. Levofloxacin, ofloxacin, and probably grepafloxacin possess intermediate but still high activity. Ciprofloxacin is slightly less active than the other compounds but is still more active than erythromycin. Azithromycin is substantially more active in animal experiments than is clarithromycin. Azithromycin (intravenous formulation), levofloxacin, and trovafloxacin are currently licensed by the FDA for the treatment of Legionnaires disease, as are erythromycin and dirithromycin.

In my opinion, azithromycin or one of the more active fluoroquinolones should be used in preference to erythromycin for the treatment of Legionnaires disease in hospitalized and immunocompromised patients. The potentially high mortality and morbidity rates of the disease in such patients, combined with the experimental evidence discussed above, is reason enough to use more active antimicrobial agents in this setting. In addition to producing a potentially better outcome, these agents will often improve patient convenience as a result of fewer side effects and shorter durations of therapy. Adding rifampin to the above therapies probably provides little further benefit; no convincing laboratory data show that adding rifampin to fluoroquinolone or more active macrolide therapy improves bacterial killing [3]. In our institution, levofloxacin (500 mg/d for 7 to 10 days) or azithromycin (500 mg/d for 3 days; 500 mg on day 1 and 250 mg/d for 4 more days; or 500 mg/d for 5 days) are recommended for the treatment of inpatients with Legionnaires disease [20].

For nonimmunocompromised outpatients with community-acquired Legionnaires disease, erythromycin or tetracycline therapy remain perfectly acceptable. These drugs are much less expensive than the newer agents, and treatment outcomes in this population are likely to be the same with new and old therapies. Ten to 14 days of treatment with erythromycin or doxycycline should be sufficient for such patients. Drug intolerance or expected poor compliance with a longer duration of therapy are the main reasons to prescribe the newer agents. As with the newer antimicrobial agents, the possibility that a patient whose illness seems unresponsive to these other agents has pneumonia caused by an agent other than Legionella species, especially an antimicrobial-resistant bacterium, should be considered.