An audit of the BMJ's correspondence columns.

OBJECTIVE: To see whether some sections of the BMJ attract more comment than others, whether letters submitted in response to different sections of the journal are rejected at different rates, and whether the balance between letters that agree and disagree with articles in the published correspondence reflects that in submitted letters. DESIGN: Retrospective audit of letters submitted for publication in the correspondence columns of the BMJ in response to articles published between 1 January and 21 May 1989. SUBJECTS: A total of 1319 letters received by the journal, 974 submitted in response to the 1501 published articles and a further 345 raising new issues. MAIN OUTCOME MEASURES: The total numbers of letters submitted in response to the four main sections of the journal--editorials, news, papers, and middles--and the numbers published. Submitted and published letters were analysed according to whether they agreed or disagreed with articles. RESULTS: The overall rejection rate was 63% (831/1319), but among letters relating to articles it was 56% (543/974). Editorials and middles attracted proportionately more letters than papers, but letters relating to papers had a lower rejection rate (43% v 57% for editorials and 43% v 66% for middles). For all sections more letters disagreed than agreed, but a higher proportion of letters in response to editorials and middles disagreed than those submitted in response to papers (64% and 72% v 53%). Among the published letters, however, broadly equal numbers of letters agreed and disagreed with articles, irrespective of section. CONCLUSION: Those sections of the journal that aim at stimulating debate succeeded in attracting the most comment. The relative importance of original scientific research papers was reflected by the priority given to letters submitted in response to papers, and the final correspondence column was a balanced platform of debate despite an unequal submitted response in terms of letters that agreed and disagreed with different sections of the journal.