We read with interest the study by Paul Little and colleagues1 that confirmed the findings of previous smaller studies2 and support the present recommendations that acute bronchitis should not be treated with antibiotics. Because the authors used a threshold of 1% for statistical significance, we wonder why one of the secondary outcomes—that fewer individuals experienced new or worsened symptoms in the amoxicillin group—was interpreted as significant (p=0·043). We think that, according to the methods of the authors, none of the primary or secondary outcomes reached statistical significance, which would further corroborate the authors' conclusions to discourage the use of antibiotics in patients with acute non-pneumonic lower-respiratory tract infection. Moreover, these data emphasise the need to develop strategies to identify which patients with lower-respiratory tract infections need antibiotic treatment and which do not. For example, the use of biomarkers such as procalcitonin has been suggested to provide such a discriminative method,3 and thus it might be interesting to know whether the patients included in the present study had low procalcitonin concentration.