New algorithm may help guide empiric therapy for patients with any type of pneumonia

The algorithm classified patients into four groups based on severity of illness (i.e., need for mechanical ventilation or ICU admission) and the presence of other risk factors for multidrug-resistant pathogens.


A single algorithm based on risk factors for multidrug-resistant pathogens, rather than site of pneumonia acquisition, may help guide empiric therapy for all patients with pneumonia, a recent study suggested.

From November 2013 to April 2017, researchers prospectively applied a new therapeutic algorithm to a cohort of 1,089 patients with pneumonia at 12 hospitals in Japan: 656 with community-acquired pneumonia (CAP), 238 with health care-associated pneumonia (HCAP), 140 with hospital-acquired pneumonia (HAP), and 55 with ventilator-associated pneumonia (VAP). All patients were hospitalized with radiographically confirmed pneumonia with appropriate clinical findings.

The algorithm classified patients into four groups based on severity of illness (i.e., need for mechanical ventilation or ICU admission) and the presence of other risk factors for multidrug-resistant pathogens. The six risk factors included antibiotic therapy in the past 180 days, poor functional status (Barthel Index <50 or performance status ≥3), hospitalization for more than two days in the past 90 days, occurrence five days or more after admission to an acute hospital, hemodialysis, and immunosuppression.

Patients with zero or one risk factor (groups 1 and 3) received therapy for CAP (a beta-lactam in combination with a macrolide). Those with two or more risk factors (groups 2 and 4) received HAP therapy (two- or three-drug regimen including an anti-pseudomonal beta-lactam in combination with a quinolone or aminoglycoside, plus either optional linezolid or vancomycin).

Results were published online on Aug. 1 by Clinical Infectious Diseases.

Overall, 898 (82.5%) patients were treated according to the algorithm. When the recommended therapy was followed, it led to an inappropriate choice in only 4.3% of patients. The frequency of multidrug-resistant pathogens varied across pneumonia types (VAP, 50.9%; HAP, 27.9%; HCAP, 10.9%; and CAP, 5.2%). Patients with two or more risk factors had multidrug-resistant pathogens more often than those with zero or one risk factor (25.8% vs. 5.3%; P<0.001).

The 30-day mortality rate was 7% for all patients (VAP, 18.2%; HAP, 13.6%; HCAP, 6.7%; and CAP, 4.7%). Mortality rates were lower in those with zero or one risk factor than in those with two or more risk factors (4.5% vs. 12.5%; P<0.001). In a multivariate logistic regression analysis, hypotension (systolic blood pressure ≤90 mm Hg), inappropriate therapy, and five risk factors (age ≥75 years, hematocrit <30%, albumin level <3.0 g/dL, blood urea nitrogen level ≥21 mg/dL, and chronic liver disease) were significantly correlated with 30-day mortality, whereas pneumonia type was not.

The authors noted limitations of the study, such as the need to apply the algorithm in countries other than Japan and the fact that the algorithm does not include therapy for viral infections even though the cohort included patients with viral infection and suspected bacterial co-infection. They added that the application of the algorithm may be limited for terminally ill patients who do not receive mechanical ventilation or ICU admission.

The authors added that once the algorithm is verified in other clinical settings, it could be supplemented with hospital-specific antibiograms to optimize therapy for patients with HAP and VAP. “Use of our algorithm has the possibility to eliminate excessive use of antibiotics and simplify pneumonia treatment, while leading to a high rate of appropriate empiric therapy with an associated good outcome,” they wrote.