In family medicine, the incidence of acute bronchitis is estimated at 24 to 46 per 1,000 patients per year and that of pneumonia as five to 11 per 1,000 patients per year. While pneumonia must be treated with antibiotics, bronchitis now is believed to be self-limiting for the most part and better managed symptomatically. As evidence continues to show that many episodes of bronchitis are better managed without antibiotics, the ability to accurately differentiate cases of pneumonia from other lower respiratory tract infections becomes increasingly significant. The classic diagnostic criteria derived from hospital studies are of limited applicability in primary care. In addition, family physicians face increasing challenges to accurately diagnose pneumonia with limited time and laboratory resources. Hopstaken and colleagues studied the diagnostic value of various parameters to derive a prediction rule for pneumonia in 15 general practices in the Netherlands.

They studied all adult patients consulting 25 family physicians with symptoms of lower respiratory infection. All patients had new or increasing cough within the previous 29 days plus symptoms such as dyspnea, wheeze, chest pain, fever, perspiration, headache, and myalgias. To be included, patients had to be diagnosed clinically by the physician as having a lower respiratory tract infection. Patients with recent antibiotic use, current serious illness, recent hospitalization for a respiratory condition, or inability to tolerate antibiotics were excluded. Pregnant or lactating women were also excluded from the study.

The physicians obtained an extensive standardized medical history and performed a physical examination for each of the 246 patients and developed a diagnosis of pneumonia or “non-pneumonia” for each patient. Blood samples were taken from each patient for erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) determination. Chest radiographs were obtained on the third day after assessment to ensure that infiltrates would be detectable. Radiographs were interpreted independently by two senior radiologists who were unaware of the clinical diagnosis of each patient. Radiologic evidence of infiltrate was taken as the diagnostic gold standard for pneumonia.

During the 16-month study period, pneumonia was diagnosed radiographically in 32 (13 percent) of the 246 patients assessed. The physicians made a clinical diagnosis of pneumonia in 21 (8.5 percent) cases. As shown in the table on page 682, recent cough and mental confusion were predictive symptoms that occurred in relatively few patients. Conversely, nausea and diarrhea were common complaints that at least doubled the probability of pneumonia. Classic diagnostic characteristics of pneumonia, such as dyspnea, tachypnea, and clinical signs, were not discriminatory for pneumonia.

The physicians had a high index of suspicion for pneumonia, and their clinical impression for severe illness and the diagnosis was associated with increased probability of radiologic diagnosis. In logistic regression analysis, dry cough, diarrhea, and fever had statistically significant odds ratios. When ESR and CRP were added to the diagnostic model, the ability to predict pneumonia was significantly improved. ESR and CRP values alone had higher diagnostic odds ratios than any of the clinical symptoms or signs.

The authors calculate that using a CRP value of less than 20 can help differentiate patients with pneumonia from those with other lower respiratory tract infections. In practice, the presence of dry cough, diarrhea, temperature higher than 38°C (100.4°F) and elevated ESR and CRP levels can be considered significant factors in the clinical diagnosis of pneumonia and the decision to prescribe antibiotics.