07.05.2015 Views

IMS Magazine - Summer 2012 edition in PDF format - Institute of ...

IMS Magazine - Summer 2012 edition in PDF format - Institute of ...

IMS Magazine - Summer 2012 edition in PDF format - Institute of ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

VIEWPOINT<br />

survival <strong>of</strong> the fittest author, and away from<br />

dissem<strong>in</strong>at<strong>in</strong>g authentic results. Scientists<br />

are <strong>in</strong>creas<strong>in</strong>gly be<strong>in</strong>g evaluated us<strong>in</strong>g metrics<br />

such as the h-<strong>in</strong>dex, which is based on<br />

number <strong>of</strong> publications and number <strong>of</strong> citations<br />

per publication 5 . Assess<strong>in</strong>g a scientist’s<br />

performance us<strong>in</strong>g such metrics creates undeniable<br />

desire, need, and pressure to publish<br />

<strong>of</strong>ten, and <strong>in</strong> highly cited journals. Countries<br />

such as Ch<strong>in</strong>a, South Korea, and Turkey <strong>of</strong>fer<br />

cash <strong>in</strong>centives to encourage local scientists<br />

to submit their manuscripts to high-impact<br />

journals 6 . Despite the low probability <strong>of</strong> success,<br />

the high volume <strong>of</strong> submissions overwhelms<br />

reviewers and congests the publish<strong>in</strong>g<br />

pipel<strong>in</strong>e. Given these circumstances, how<br />

can the literature be trusted, and how can<br />

scientific progress be made?<br />

The soar<strong>in</strong>g retraction rate observed<br />

<strong>in</strong> high-impact journals may<br />

be an accelerated manifestation <strong>of</strong><br />

the decl<strong>in</strong>e effect. Because highimpact<br />

journals publish studies<br />

with novel and dramatic results,<br />

those results are more likely to be<br />

overstated, and are less likely to be<br />

reproducible. The lack <strong>of</strong> reproducibility<br />

is unsurpris<strong>in</strong>g consider<strong>in</strong>g<br />

the tweak<strong>in</strong>g, select<strong>in</strong>g, and beautify<strong>in</strong>g<br />

<strong>of</strong> data that <strong>of</strong>tentimes precedes<br />

publication.<br />

First reported <strong>in</strong> the 1930s, an established<br />

barrier to scientific advancement is the decl<strong>in</strong>e<br />

effect, which is observed <strong>in</strong> the literature<br />

as scientifically discovered effects that<br />

dim<strong>in</strong>ish over time. Schooler suggests that<br />

“if early results are more likely to be reported<br />

when errors comb<strong>in</strong>e to magnify the apparent<br />

effect, then published studies will show<br />

systematic bias towards <strong>in</strong>itially exaggerated<br />

f<strong>in</strong>d<strong>in</strong>gs, which are subsequently statistically<br />

self-corrected” 7 . The soar<strong>in</strong>g retraction rate<br />

observed <strong>in</strong> high-impact journals may be an<br />

accelerated manifestation <strong>of</strong> the decl<strong>in</strong>e effect.<br />

Because high-impact journals publish<br />

studies with novel and dramatic results, those<br />

results are more likely to be overstated, and<br />

are less likely to be reproducible. The lack <strong>of</strong><br />

reproducibility is unsurpris<strong>in</strong>g consider<strong>in</strong>g<br />

the tweak<strong>in</strong>g, select<strong>in</strong>g, and beautify<strong>in</strong>g <strong>of</strong><br />

data that <strong>of</strong>tentimes precedes publication.<br />

Detrimentally, if the studies <strong>in</strong> high-impact<br />

journals are reach<strong>in</strong>g the widest audiences<br />

and are convey<strong>in</strong>g less-than-accurate results,<br />

the scientific field is be<strong>in</strong>g misguided.<br />

The argument can be made that science is an<br />

imperfect field, with enormous variability. In<br />

biomedical research, there are <strong>in</strong>numerable<br />

known and unknown variables that <strong>in</strong>fluence<br />

experimental outcomes, some as significant<br />

as the stra<strong>in</strong> <strong>of</strong> mouse used, some as <strong>in</strong>nocuous<br />

as the time <strong>of</strong> day an experiment is conducted.<br />

Every experiment conta<strong>in</strong>s outliers,<br />

<strong>in</strong>consistent replicates, and unexpected f<strong>in</strong>d<strong>in</strong>gs,<br />

but these are rarely reported, and perhaps<br />

that is the bigger issue. As Dr. Karen Davis,<br />

Associate Director <strong>of</strong> the <strong>IMS</strong> and Editor<br />

<strong>of</strong> Pa<strong>in</strong>, previously communicated to the <strong>IMS</strong><br />

<strong>Magaz<strong>in</strong>e</strong> (“Positive Pressure,” Fall 2011), an<br />

<strong>in</strong>complete or <strong>in</strong>correct scientific record can<br />

lead to propagation <strong>of</strong> unfounded ideas, unnecessary<br />

replication <strong>of</strong> experiments, cl<strong>in</strong>ical<br />

translation <strong>of</strong> harmful therapies, or delayed<br />

development <strong>of</strong> alternate hypotheses. Once<br />

flawed ideas are published, they can never<br />

truly be retracted. Budd et al. report that<br />

even after an article is retracted, it cont<strong>in</strong>ues<br />

to be cited, without acknowledgement <strong>of</strong> the<br />

retraction 3 . Understandably, once digital versions<br />

<strong>of</strong> articles are downloaded, researchers<br />

are unlikely to consult the orig<strong>in</strong>al source<br />

aga<strong>in</strong>, and are therefore unlikely to be aware<br />

<strong>of</strong> corrections.<br />

What is needed is a more complete account<br />

<strong>of</strong> scientific f<strong>in</strong>d<strong>in</strong>gs, be they negative or<br />

positive, consistent or <strong>in</strong>consistent, surpris<strong>in</strong>g<br />

or expected. Dur<strong>in</strong>g the <strong>IMS</strong> Scientific<br />

Day <strong>2012</strong> Bernard Langer Lecture <strong>in</strong> Health<br />

Sciences, Dr. Thomas R. Insel, MD, Director<br />

<strong>of</strong> National <strong>Institute</strong> <strong>of</strong> Mental Health, posited<br />

that if experiments are soundly designed<br />

and flawlessly executed, then those results<br />

should be dissem<strong>in</strong>ated, regardless <strong>of</strong> what<br />

they are. Furthermore, Schooler believes “we<br />

need a better record <strong>of</strong> unpublished research<br />

before we can know how well the current<br />

scientific process, based on peer review and<br />

experimental replication, succeeds <strong>in</strong> dist<strong>in</strong>guish<strong>in</strong>g<br />

grounded truth from unwarranted<br />

fallacy” 7 . He recognizes the difficulties <strong>in</strong><br />

implementation, but suggests an open-access<br />

database <strong>of</strong> research methods, which allows<br />

scientists to log their hypotheses and methodologies<br />

prior to experiments, and their<br />

published and unpublished results afterwards<br />

7 .<br />

Schooler believes “we need a better<br />

record <strong>of</strong> unpublished research<br />

before we can know how well the<br />

current scientific process, based on<br />

peer review and experimental replication,<br />

succeeds <strong>in</strong> dist<strong>in</strong>guish<strong>in</strong>g<br />

grounded truth from unwarranted<br />

fallacy.”<br />

To purify the publication pool, the current<br />

publish<strong>in</strong>g paradigm should be restructured<br />

to diffuse the pressure experienced by researchers.<br />

Among other strategies, graduate<br />

students should be given avenues to publish<br />

their negative results, and senior scientists<br />

should be evaluated on metrics other than<br />

their publish<strong>in</strong>g records. To purify the publication<br />

pool, more focus should be placed<br />

on honest account<strong>in</strong>g <strong>of</strong> data <strong>in</strong> its entirety,<br />

and less focus should be placed on polish<strong>in</strong>g<br />

<strong>of</strong> data for high-impact journals. Researchers<br />

should be m<strong>in</strong>dful <strong>of</strong> the permanence <strong>of</strong><br />

published results, and wary that embellished<br />

or modified data can significantly h<strong>in</strong>der<br />

scientific progress. To alleviate current publish<strong>in</strong>g<br />

pressures is to purify the publication<br />

pool, and move science forward.<br />

Disclaimer: The op<strong>in</strong>ions expressed by the<br />

author(s) are <strong>in</strong> no way affiliated with the <strong>Institute</strong><br />

<strong>of</strong> Medical Science or the University <strong>of</strong> Toronto.<br />

Comments are welcome at theimsmagaz<strong>in</strong>e@<br />

gmail.com.<br />

References<br />

1. Fang FC, Casadevall A. Retracted science and the retraction<br />

<strong>in</strong>dex. Infect Immun. 2011;79(10):3855-3859.<br />

2. Zimmer C. A Sharp Rise <strong>in</strong> Retractions Prompts Calls<br />

for Reform. The New York Times. <strong>2012</strong> Apr 17;Sect. D-1.<br />

3. Van Noorden R. Science publish<strong>in</strong>g: The trouble with<br />

retractions. Nature. 2011;478(7367):26-28.<br />

4. Fanelli D. “Positive” results <strong>in</strong>crease down the Hierarchy<br />

<strong>of</strong> the Sciences. PLoS One. 2010;5(4):e10068.<br />

5. Hirsch JE. An <strong>in</strong>dex to quantify an <strong>in</strong>dividual’s scientific<br />

research output. Proc Natl Acad Sci U S A.<br />

2005;102(46):16569-16572.<br />

6. Stephan P. Research efficiency: Perverse <strong>in</strong>centives.<br />

Nature. <strong>2012</strong>;484(7392):29-31.<br />

7. Schooler J. Unpublished results hide the decl<strong>in</strong>e effect.<br />

Nature. 2011;470(7335):437.<br />

<strong>IMS</strong> MAGAZINE SUMMER <strong>2012</strong> GENOMIC MEDICINE | 32

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!