王飞跃的个人博客分享 http://blog.sciencenet.cn/u/王飞跃

博文

[转载]Taking the Powerball Approach to Funding Medical Research

已有 3977 次阅读 2014-8-19 16:57 |系统分类:海外观察|文章来源:转载

Takingthe Powerball Approach to Funding Medical Research

Winninga government grant is already a crapshoot. Making it official by running alottery would be an improvement.

Fromhttp://online.wsj.com/news/articles/SB10001424052702303532704579477530153771424

By

Ferric C. Fang And

Arturo Casadevall

April 14, 2014 7:08 p.m. ET

Participants anxiously await the announcement of the winning numbers, thinking tothemselves, "Someone has to win, why not me? Just think of what I could dowith a million dollars!" But alas, better luck next time. Powerball? MegaMillions? Unfortunately, no—we are talking about research funding from theNational Institutes of Health.

 

Scientists who seek NIH funding submit research proposals that are reviewed by panelscalled study sections. Each application gets a score, and funding decisions aremade on the basis of these scores and available funds. Thanks to the budgetsequester, the rate at which proposals receive funding is at a historically lowlevel. The primary mechanism for funding biomedical research in the UnitedStates has become a crapshoot.

 

And NIH's peer review has been criticized. The latest critique is a landmark studyjust published in January online and in the Feb. 14 issue of CirculationResearch by Michael Lauer and colleagues at the National Heart, Lung and BloodInstitute. It analyzed nearly 1,500 successful grants and found no correlationbetween the productivity of a project, as measured by the citation ofgrant-supported publications, and its score. Although study sections may bereasonably good at recognizing low-quality proposals, they are unable toaccurately rank good ones. NIH peer reviewers fare no better than random chancewhen it comes to predicting how well grant recipients will perform.

 

Other studies have reported that NIH peer review lacks statistical rigor. In 2008,for example, David Kaplan and colleagues at Case Western Reserve Universityreported that the number of reviewers is too low to provide an acceptable levelof precision. The same year, Valen Johnson, then at M.D. Anderson CancerCenter, applied a Bayesian statistical model to NIH peer review and foundsubstantial evidence of reviewer bias that was estimated to influence a quarterof funding decisions.

 

Forced to decide between seemingly equally meritorious grants, reviewers tend toprioritize applications on the basis of "grantsmanship," the skillwith which an application is assembled, which has little to do with futureproductivity or innovation. In fact, Virginia Tech and Stanford researchersJosh Nicholson and John Ioannidis pointed out in a 2012 commentary in Naturethat the structure of the NIH peer-review system encourages conformity anddiscourages innovation.

 

The NIH spends a lot of money on grant peer review. The annual budget of the NIHCenter for Scientific Review is $110 million. Individual institutes and centersalso spend a lot on peer review. The costs are not only financial. Writing andreviewing grants is extremely time consuming and diverts the efforts ofscientists away from doing science itself. If an application receives a goodscore but is not funded, researchers must spend months revising their proposalsand generating additional data to try again. If they fail a second time, theycannot resubmit for three years.

 

A number of scientists have suggested alternatives, including having scientistsvote on which grants get funded, basing decisions on an applicant's trackrecord, or awarding small amounts of money to all applicants.

 

We suggest another approach. Study sections could simply be asked to determinewhether applications are meritorious or not. Meritorious applications wouldthen be randomized by computer, and funding awarded to as many projects as canbe accommodated by the research budget. Applications not chosen would becomeeligible for the next drawing, with individual researchers permitted to enter onlyone application per drawing.

 

An NIH lottery would have many immediate advantages. First, it would convert thecurrent biased and arbitrary system to a more transparent process. Second, itwould relieve reviewers from having to identify the top applications since itis increasingly obvious that this is not possible. Third, applicants withmeritorious but unfunded proposals could continue to reapply. Fourth, it wouldlessen the blow of grant rejection. Fifth, funds currently used for the futileexercise of ranking proposals could be devoted to supporting scientificresearch. Sixth, the realization that many meritorious projects remain unfundedmay promote more serious efforts to improve research funding and to studyalternative approaches to peer review.

 

There are ample precedents for using lotteries to allocate scarce resources fairly.They have been used, for example, to determine military service when there wasconscription and for school placement to assign students to highly desirableschools. College student and low-income housing is often awarded by lottery.

 

The chronic underfunding of science gambles with society's future. The NIH claimsthat it funds the best science by the best scientists and takes pride in therigor of its peer review. But tens of thousands of meritorious grantapplications now go unfunded each year, and according to multiple outsideanalyses, the ones funded are not necessarily the best. Of course we wouldprefer to see research funding restored to levels that allow all deserving researchprojects to proceed. Until that time comes, a lottery would be an efficient waywith which to award grants.

 

 

Dr. Fang is professor of laboratorymedicine and microbiology at the University Of Washington School Of Medicine.Dr.Casadevall is professor and chairman of microbiology and immunology at theAlbert Einstein College of Medicine of Yeshiva University.



http://blog.sciencenet.cn/blog-2374-820675.html

上一篇:[转载]参考消息
下一篇:[转载]数据众包:Paying Farmers to Welcome Birds

0

该博文允许注册用户评论 请点击登录 评论 (0 个评论)

数据加载中...

Archiver|手机版|科学网 ( 京ICP备14006957 )

GMT+8, 2018-9-25 15:40

Powered by ScienceNet.cn

Copyright © 2007- 中国科学报社

返回顶部