Spreading good words around...
Spreading good words around...

Writing can be evidence-based - and we did the research that proved it

Submitting a scientific article for publication is a marketing activity. Authors must provide a 'product' that the 'customer' (the editor, advised by reviewers) will 'purchase'. As with all markets, the preferences of customers can differ in important ways (1), and the more authors can find out about these, the better their chances of acceptance. In this study we examined 300 original research articles to see whether basic variables (such as voice, title, opening and closing sentences, length and authorship) varied according to the journal.


One of us (KH) examined 50 consecutive research articles, dating from June 1, 1997, in three general journals (New England Journal, Lancet and BMJ) and three specialist paediatric journals (Journal of Pediatrics, Pediatric Research, Archives of Disease in Children). Three of these journals (BMJ, Lancet, Archives) are UK-based; the other three are US-based. In each article she counted the number of words in each headline, the presence or absence of a verb, and the presence or absence of a colon. She counted the number of authors, paragraphs in each section, use of 'we' in the methods section, figures, tables and references. She noted whether the last sentence of the introduction answered the question: 'what we did?' and the first sentence of the discussion answered the question 'what we found'. Using our previous classifications (2), she allocated the first sentence of the introduction to one of three categories ('seminar', 'alarmist' and 'much discussion recently') and the last sentence of the discussion to one of three categories ('a puzzle solved', 'perhaps possibly' and 'more research is indicated').

Opening sentences 

  • Seminar: 'Onchocerciasis is a major public health and socioeconomic problem in many rural areas of Africa'.
  • Alarmist: 'During the current decade in China there will be about 15 millions deaths from strokes, plus much disability'.
  • Much Discusssion Recently (MDR): 'Medical discussion about the end of life for neonates and infants is under debate in the Netherlands.'

Last sentences

  • Another puzzle solved: 'It is clear, however, that care guidelines for people with IDDM should now be included the treatment of early-stage renal disease with ACE inhibitors, even in normotensive patients.'
  • More research is indicated: 'Further studies will be needed to determine optimum doses and duration of therapy.'
  • Perhaps possibly; 'The prevention of rickets through such means as health education, exposrure of infants to sunlight and vitanmin D, and/or calcium supplementation, as well as early treatment of rickets, may be an important part of the global efforts in the control of acute respiratory infection'.

The results are shown in table 1. Thirty three of the 50 BMJ articles contained a colon in the title (proportion = .66, 95% CI .529 to .791) compared with 28 of the 250 other journal articles (proportion = .112, 95% CI .073 to .151). Forty six of the 150 US articles contained a verb in the title (Proportion = .307, 95% CI .233 to .380) compared with 10 of the 150 UK articles (proportion = .067, 95% CI .023 to .107). The average number of authors in NEJM and Lancet articles (the two with the highest impact factor) was 7.72 (SD 3.45, 95% CI of mean 7.04 to 8.40). For articles in Archives of Disease in Childhood and Pediatric Research (those with the lowest impact factor) the average was 4.74 (SD 2.30, 95% CI of mean 4.28 to 5.20). A total of 97 of the 150 articles from general journals used 'we' in the methods section (proportion =.647, 95% CI .570 to ..723) compared with 41 of the 150 specialist journals (proportion = .273, 95% CI .202 to .345). Ten of the 50 Lancet articles had an alarmist opening sentence (proportion = .20, 95% CI .089 to .311) compared with 15 of the other 250 journal articles (proportion + .06, 95% CI .031 to .089). Nine of the 50 BMJ articles had a 'much discussions recently' opening sentence (proportion = .18, 95% CI .074 to .286) compared with 16 of the other 250 journal articles (proportion = .064, 95% CI .034 to .094).


The findings throw up some challenging questions: Is the prestige of a journal the main determinant of the number of authors listed? Why do US journals have longer headlines, more paragraphs and more references? Do general journals prefer the active because they have more full-time staff working on the copy? Why is the Lancet more sensational than other journals? More importantly, the findings show that there are some clear differences in the journal 'market', and this has important implications. Intending authors should study their target journal closely and use this market research to inform the way they prepare the paper. We can describe this process as evidence-based writing.



1. Cameron H, Robertson A. The colon in medicine: nothing to do with the intestinal tract. BMJ 1997: 315: 1657-8.

2. Albert T. Winning the publications game. Oxford: Radcliffe Medical Press, 1997: 28 and 33.

This research was done done in 1996-7 by Kathryn Hampson and Tim Albert. 

Funding: Tim Albert Training

Conflict of interest: Tim Albert Training was providing courses in writing scientific papers.

Examples taken from The Lancet, June 1997

'Beautifully written ... a rare good autobiography'


For more details, click on picture

Talks now available: topics include the rise of the baby boomers and memories of a country newspaper 50 years ago. Click for details

Winning the publications game: the groundbreaking book on writing scientific papers.

Click for details

Print Print | Sitemap
© Tim Albert 2016