A new study in the European Respiratory Journal suggests that testing patients for the presence of fever, a crackling sound in the lungs, high pulse, and low blood-oxygen levels could help clinicians diagnose pneumonia in patients with lower respiratory infections and thereby avoid prescribing some unnecessary antibiotics.
The results of the large prospective study showed that 86.1% of the patients with a pneumonia diagnosis confirmed by chest x-ray had at least one of these four clinical symptoms, and the positive predictive value of having at least one was 20.2%. Patient characteristics, such as age and medical history, and other symptoms, including shortness of breath and sputum color, did not help clinicians decide who had pneumonia.
The researchers believe that the findings could help reduce antibiotic prescribing for lower respiratory infections, which are commonly caused by viruses and don't require antibiotics.
Unnecessary prescriptions for lower respiratory tract infections are widely considered a contributor to antibiotic resistance, yet prescribing rates remain high in the United Kingdom and other industrialized nations. A 2014 study found that the highest prescribing 10% of UK general practices issued antibiotic prescriptions at 69% of consultations for respiratory tract infections, while the median practices prescribed antibiotics at 54% of consultations.
"Most patients with a lower respiratory tract infection will recover perfectly well without antibiotics but, at the moment, around 60% of patients get a prescription," lead author Michael Moore, BM, BS, a professor of primary healthcare research at the University of Southampton, said in a press release from the European Lung Foundation. "If antibiotic prescribing was restricted to people who had one or more of these signs it could result in a substantial reduction in unnecessary prescriptions for this condition."
Independent predictors of pneumonia
For the study, the researchers recruited 28,883 patients who had visited 5,222 UK medical practices from 2009 to 2013 with acute cough attributed to lower respiratory tract infection, collected data on their symptoms, and followed them for 30 days after their initial visit. A total of 720 of these patients went on to have a chest x-ray within the first 7 days, and 115 (0.4%) were ultimately diagnosed as having pneumonia.
Although most of the patients referred for radiography were older, sicker, and more likely to be a smoker, the researchers found that patient characteristics did not provide useful information for the diagnosis of pneumonia. Nor did many of the symptoms the patients had.
But the researchers identified four symptoms from clinical examination that were diagnostically useful, with a temperature above 37.8ºC (100ºF; risk ratio [RR], 2.65) being the strongest independent predictor of pneumonia, followed by a crackling sound in one or both lungs (RR, 1.82), a pulse rate of more than 100 beats per minute (RR, 1.90), and oxygen saturation in the blood lower than 95% (RR, 1.73). Ninety-nine of the 115 patients with pneumonia exhibited at least one of these symptoms.
When the researchers looked at the sensitivity and specificity of the four independent predictors among the subset of patients who had been referred for an x-ray, they found that the highest sensitivity (83.5%) was achieved when all four were considered.
Moore and his colleagues say the benefit of these four factors is that they all are easily measured clinical symptoms that clinicians routinely test for, with the exception of blood oxygen levels. Oxygen saturation can be easily measured with non-invasive devices called pulse oximeters, which are widely available but not routinely used.
The study's strengths are its size and its large, generalizable cohort, which limited selection bias. But because the patients selected for a chest x-ray represented only a small sample of the full cohort and were generally older and sicker, milder cases of pneumonia may have been missed.
Nov 22 Eur Respir J study
Nov 23 European Lung Foundation press release