Distinguishing Bacterial Pneumonia from 2009 H1N1 Influenza Pneumonia
Amesh A. Adalja, MD, March 18, 2011
Distinguishing at presentation which patients with community-acquired pneumonia require antiviral treatment and droplet isolation for influenza is a difficult task made more so by the lack of a solid evidence base or prediction model for guidance. Determining whether a patient with pneumonia during a flu outbreak has influenza-related pneumonia or community-acquired bacterial pneumonia (CAP) has important implications for treatment and isolation decisions. A team of researchers from the UK recently published results of their attempt to develop a clinical rule that could distinguish between these 2 types of pneumonia in the setting of 2009 H1N1 influenza.
H1N1 and Bacterial Pneumonia Cohorts
Bewick and colleagues’ study population consisted of an H1N1 group and a CAP group.
H1N1 group: 254 patients from 75 hospitals. Patients in this group had PCR-confirmed H1N1 and radiographic evidence of pneumonia.
CAP group: 648 patients with community-acquired pneumonia (CAP) from 1 National Health Service Hospital; these patients were part of an observational cohort study of pneumonia patients.
All patients were 16 years of age or older. H1N1 patients were identified between May 2009 and January 2010; CAP patients were identified between September 2008 and April 30, 2009. Patients admitted after 2009 H1N1 appeared were excluded from this study.
Several statistically significant findings were noted upon comparison of the 2 groups—findings that formed the basis for the prediction model. Specifically, H1N1 patients were:
younger, with median age 42 vs. 75 years;
more likely to be febrile, tachypneic, and tachycardic;
more likely to have lower blood urea nitrogen (BUN) levels;
more likely to have bilateral consolidation on chest radiographs;
less likely to be confused;
less likely to have leukocytosis and lower C-reactive protein (CRP) levels; and
less likely to have comorbid illnesses.
Clinical Model Derived
Based on the differences between the cohorts, a 5-point regression model was developed to identify the clinical variables most predictive of the presence of H1N1 pneumonia. The 5 factors and their associated odds ratios (OR) in multivariate analysis were:
age <65 years (OR 12.7);
white blood cell count <12,000 (OR 9.7);
bilateral radiographic changes (OR 3.3);
oriented mental status (OR 2.6); and
temperature >38°C (OR 1.9).
When all 5 variables were present, 87.3% of H1N1 patients were captured by the model. If age was dropped from the model—because of its highly significant predictive value—78.9% of H1N1 patients were captured. The authors also developed a quick emergency department–targeted model in which age, temperature, chest radiograph appearance, and mental status orientation could be used. With this model, 81.1% of H1N1 patients were captured.
H1N1 Pneumonia Score
The authors further derived a scoring system in which 1 point was assigned for each of the 5 variables identified. A score of 0 or 1 gave a clinically significant positive likelihood ratio for excluding H1N1 of 75.7, while a score of 4-5 gave a likelihood ratio of 9 for predicting H1N1 pneumonia.
Can Clinical Scoring Improve the Care of H1N1 Pneumonia Patients?
The results of this study have the potential to inform better clinical decisions about isolation and antiviral and antibiotic use when managing patients with pneumonia during flu season. While the clinical model has several limitations—the most important of which is its applicability only to 2009 H1N1—it has, after further validation, a role in safely ruling out influenza in hospitalized pneumonia patients who can then forgo antiviral treatment and isolation. During a pandemic there may be strict utilization guidelines for antiviral therapy because of regional or national scarcity, and being able to clinically justify foregoing antiviral treatment in certain hospitalized patients may relieve some of the demand on strained stockpiles without compromising clinical care. Also, unburdening hospitals by lessening the need to place patients in private or influenza cohort rooms can potentially improve hospital patient flow at a time when patient surge is occurring.
Further development of this model, especially to assess its applicability to other strains of influenza viruses, would be a useful tool for clinicians, hospital administrators, and infection control practitioners.
Bewick T, Myles P, Greenwood S, et al. Clinical and laboratory features distinguishing pandemic H1N1 influenza-related pneumonia from interpandemic community-acquired pneumonia in adults. Thorax 2011;66:247-252. http://thorax.bmj.com/content/66/3/247.full
. Accessed March 14, 2011.http://www.upmc-cbn.org/report_archive/ ... 82011.html