|
O problema que o Renato Assuncao levanta, ocorre com qualquer conjunto de dados observados de comportamento. O experimento que ele sugere, embora correto do ponto de vista formal, é inviável do ponto de vista metodológico na maioria dos casos.
On Thu, 18 Sep 2008 11:35:12 -0300, Renato Martins Assuncao wrote
> Olha ai, de novo, as agruras da vida do estatistico que noa lida com
> experimentos.
>
> Eu imagino que os ensaios clinicos da industria sao muito bem escolhidos entre os
> mais promissores. Ja os outros (non-profit) sao de risco mais elevado e por isto
> mesmo deixados de lado pelka industria. Qual a surpresa sobre as taxas de sucesso
> academico posteriores?
>
> OK, maquinacoes da insustria farmaceutica e a hipotese do sistema malevolo
> comandoadno nossas vidas sao tambem hipoteses explicativas. Mas eu so compraria
> esta se pudessemos fazer um experiemnto:
> - selecione um grupo de ensaios clinicos a serem realizados.
> - com uma moeda decida quem recebe financiamento for profit e non-profit
> - compare depois.
>
> Da forma como sao feitos os estudos atualmente, os dois grupos nao sao
> comparaveis desde o instante zero.
>
> Renato Assunção
> Universidade Federal de Minas Gerais
> Instituto de Ciencias Exatas
> Departamento de Estatistica
> Campus Pampulha
> Belo Horizonte MG 31270-901 - Brasil
> assuncao@est.ufmg.br
> FAX: 55-31-3409-5924 PHONE: 55-31-3409-5940
> http://www.est.ufmg.br/~assuncao
>
> -----"Luis Paulo Vieira Braga" <lpbraga@im.ufrj.br> escreveu: -----
>
>
Para: abe-l@ime.usp.br
> De: "Luis Paulo Vieira Braga" <lpbraga@im.ufrj.br>
> Data: 18/09/2008 9.25
> Assunto: [ABE-L]: citações versus financiamento
>
> Industry-Sponsored Trials More Widely Cited Than Not-for-Profit Studies
>
> (from Heart wire a professional news service of WebMD)
> http://www.theheart.org/index.do
>
> Michael O'Riordan
>
> September 10, 2008 (Boston, MA) Cardiovascular clinical trials sponsored by
> industry are more likely to be cited in future medical publications than
> studies performed by not-for-profit organizations, a new study has shown.
> Efforts should be made to ensure that important trials conducted by
> government agencies, such as the National Institutes of Health (NIH), are
> more widely disseminated to the cardiovascular community, say investigators.
>
> "As a researcher, my core belief is that high-quality research gets done by
> the NIH and by industry, and the quality of our patients' lives will improve
> if this information is transmitted to the medical community," senior
> investigator Dr Paul Ridker (Brigham and Women's Hospital, Boston, MA) told
> heartwire. "What we observed is that the translation to practice is more
> rapid and thorough for industry-funded studies. We hope there can be a
> mechanism to do a better job promoting the findings of federally funded
> studies as well."
>
> The results of the study, with first author Dr David Conen (Brigham and
> Women's Hospital), are published online September 8, 2008 in Circulation.
>
> In 2006, Ridker and colleagues published a study showing that cardiovascular
> clinical trials funded by for-profit sponsors were significantly more likely
> to have positive results than trials reported by not-for-profit funding
> sources. He told heartwire that while the initial publication has an impact
> on clinical practice, how often the paper is cited in subsequent medical
> publications also has an effect on physician behavior.
>
> The purpose of this study, explained Ridker, was to examine how the
> cardiovascular community responded to high-quality research funded by
> different sources. To do so, the Harvard researchers analyzed 303 consecutive
> superiority trials of cardiovascular medicine published between January 1,
> 2000 and July 30, 2005 in the Journal of the American Medical Association,
> the Lancet, and the New England Journal of Medicine. They then determined the
> number of citations per publication per year, a metric used to ascertain how
> the medical community was responding to the findings.
>
> Investigators observed that industry-funded studies had more citations per
> publication per year than NIH- or other federally funded studies. This was
> true in all settings except for one: industry-funded studies were not cited
> as frequently when the studies failed to show a benefit.
>
>
Prof. Luis Paulo
C.P. 2386
20001-970 Rio de Janeiro, RJ
outros e-mails
lpbraga@gbl.com.br
bragaprof@gmail.com
nosso BLOG
http://observatoriodauniversidade.blog.br
|