Antibiotic resistance · PLoS Med · 2024
Longer antibiotic use, fewer resistance genes: the summary’s p-value says convincing; its table, could be chance
How often chance alone would produce this gap if treatment length changed nothing
The paper says
The summary: gut bacteria from communities given the antibiotic azithromycin for 4.5 years, not 2.5, carried 0.80 times as many genes for resisting it. Its p-value, 0.010: with no real difference, chance alone would produce such a gap only about 1 time in 100.
Its own data show
The paper’s own Table 2 gives 0.187 for the same comparison, and its main text 0.19: about 1 time in 5, so it could easily be chance. Every other p-value the text gives for these genes matches the table. The summary’s own range, 0.50 to 1.00 times as many, reaches 1.00: no difference.
A p-value is roughly how often chance alone would produce a gap this big if there were no real difference; under 1 in 20 (0.05) usually counts as convincing. 1 in 100 passes; 1 in 5 fails. The summary’s words, “no significant increase”, fit the table. Anyone copying its number would record a convincing drop the study did not find.