Abstract
In clinical trials of antibacterial drugs for treatment of acute exacerbation of chronic bronchitis (AECB), a high proportion of patients who have bacterial infection is essential. In 83 randomised clinical trials of antibacterial drugs in AECB or acute exacerbation of chronic obstructive pulmonary disease (AECOPD), published between 1977 and 2007, the percentage of patients who were bacteriologically-positive ranged from 13.9% to 96.0%. There was no evidence that any particular combination of signs and symptoms was associated with a bacterial exacerbation. A study of patients with signs and symptoms of AECB was undertaken to identify entry criteria that would optimise the number of enrolled patients with bacteriologically-proven (culture-positive) exacerbations. This failed to demonstrate that patients with Type I exacerbations (increased sputum purulence, sputum volume, and dyspnoea) had a higher proportion of culture-positive sputum samples than patients with Type II exacerbations (two of these three symptoms). Fisher's exact test produced a borderline result suggesting that patients with Type II exacerbations possibly had a higher chance of culture-positive sputum samples. From a range of demographic, symptomatic and laboratory investigations, the only factor which emerged as highly significant in distinguishing culture-positive from culture-negative patients was the semi-quantitative Gram stain analysis. Although this technique may offer a useful method of predicting bacterial infection in around 90% of patients, it may not be practical for all clinical trial settings. Nevertheless, identification of bacterial infection in patients with AECB or AECOPD remains an important target for the most cost-effective and scientifically robust clinical trials.