Transcript 1.3 MB - Grady College of Journalism and Mass Communication

Knight Health & Medical
Journalism
April 18, 2008
Grady College, The University of Georgia
Shannon Brownlee
[email protected]
Jeanne Lenzer April 18, 2008
Between 1980 and 2001,
support for research by
pharmaceutical companies
grew from $1.5 billion
to $22 billion.
Warner TD, Weiss Roberts L: Scientific Integrity, Fidelity and
Conflicts of Interest. Curr Opin Psychiatry 17(5):381-385, 2004
There are significant benefits of Pharm
Industry funding of research:







Money for research into improving
treatment
Some support for basic research
Emphasis on applied research
Competition to produce best product
Support for academic medicine
Support for individual researchers
Jobs and careers
A conflict of interest
exists when a primary
ethical or professional
interest clashes with
(financial) self-interest
TWO FORMS of CONFLICT
with ACADEMIC
RESEARCHERS


Honoraria, speaking fees, consulting
contracts – the soft money
Research funding
BOTH AFFECT CLINICAL
PRACTICE
*Industry uses $$ to bias studies to
sell products
*exert control over published
research – which guides practice
*influence FDA decisions and clinical
guideline setting
*influence continuing medical
education
C. Everett Koop
America’s Most Trusted Doctor:

“Maybe some people would
be affected by a million
dollars. But not me. I have a
reputation to uphold.”
INDUSTRY KNOWS WHAT IT
IS DOING
“If you want to prove a point, fund
a study…”
Pharmaceutical representative
APA Industry sponsored symposia (Fall
1997)
P.R. or Education

““The broad
distinction between
healthcare p.r. and
medical education
is becoming
obsolete.” Neil Kendle,
Lowe Fusion Healthcare

“They don’t know
where p.r. ends
and medical
education begins”
Communique
“EXPERIMERCIALS”
INDUSTRY NOW FUNDS 80
PERCENT OF CLINICAL TRIALS
How do conflicts of interest and
funding affect research
outcomes?
In general, published results of
research sponsored by pharmaceutical
companies favor new drug
treatments over traditional
therapy and patented products over
generic.
(e.g., Davidson RA. Source of funding and outcome of clinical trials.
J Gen Intern Med. 1986;1:155-158.)
He Who Pays the Piper


“Strong and consistent evidence
shows that industry-sponsored
research tends to draw pro-industry
conclusions.”
“Industry preferentially supports trial
designs that favor positive results.”
Bekelman, J. E. et al. JAMA 2003;289:454-465.
Relation Between Industry Sponsorship and Study Outcome in Original Research Studies*
Bekelman, J. E. et al. JAMA 2003;289:454-465.
*1,140 studies
Copyright restrictions may apply.
How much does funding
matter?
Type of Conflict




Support to attend
symposium
Honorarium
Research funding
Consultant
Supportive
Critical
67%
27%
75%
87%
25%
17%
20%
17%
Stelfox, H. T. et al. N Engl J Med 1998;338:101-106
“Conflict of Interest in the
Debate over Calcium-Channel
Antagonists”
Stelfox, H. T. et al., New England
Journal of Medicine,
1998;338:101-106
Stelfox Results

Supportive authors were much more
likely than critical authors to have
financial conflicts of interest – with
ANY manufacturer . . . Critical
authors were much less likely to be
financially associated with
manufacturers . . .
HOW DOES IT HAPPEN?





Recruiting patients
Randomization to treatment groups,
or treatment vs placebo
Blinding of patients and researchers
Analysis
Publication
HOW TO DISTORT YOUR
RESULTS WITHOUT REALLY
TRYING
Hide data in plain sight
 Suppress data

Hiding in plain sight










Randomization bias
Subgroup analysis, data dredging, posthoc analysis
Straw man comparators
Combination endpoints
Shifting goalposts
Publication bias
Underpowering
Segregation of data
Surrogate endpoints
Failure to look at all-cause mortalilty
Hiding in plain sight

Randomization bias
Healy, D. Journal of Psychiatry &
Neuroscience, vol 28, p 331
Randomization bias AKA fooling with data
Occurrence of suicidality in an SSRI
screening
Run-in/wash out
randomization
drug
pbo
Start
treatment
Stop
treatment
SSRI Clinical Trial Design
Analysis of suicidality
screening
Run-in/wash out
randomization
drug
pbo
Start
treatment
Stop
treatment
Healy, D. Journal of Psychiatry & Neuroscience, vol 28, p 331
THAT’S
CHEATING
Hiding in Plain Sight
(also known as lying)
Keller, M. “Efficacy of paroxetine in the
treatment of adolescent major
depression: a randomized, controlled
trial” J Am Acad Child Adolesc
Psychiatry. 2001 Jul;40(7):762-72
FAIRY TALE
"REMARKABLE efficacy
and safety in the
treatment of adolescent
depression."
Pfizer/Pharmacia Sales Memo
REALITY

PAPER: “Most adverse events were
not serious.”
Reality: Seven children had to be
hospitalized for side effects.
REALITY
PAPER: “Paroxetine is generally well
tolerated and effective for major
depression in adolescents.”
Reality: Paroxetine failed on the
protocol’s two primary outcome
measures
Hiding in Plain Sight:
straw man comparators


Compare to a placebo, not an
existing drug – NOT required by FDA
Compare to low dose for efficacy,
high dose for side effects (COX-2 vs
non-selective NSAIDS)
Am J Psychiatry 2006; 163:185–194
Am J Psychiatry 2006; 163:185–194
Fairy Godfather Dr. Alan F.
Schatzberg on the atypical
antipsychotics:
"A potential breakthrough of
tremendous magnitude.”
-- New York Times
FAIRY TALE
ABSTRACT: After the start of risperidone
treatment, days in acute care inpatient
facilities were reduced by 26 percent,
and days in residential treatment were
reduced by 57 percent.
Impact of risperidone on the use of mental health care
resources. Psychiatric Services
1997 (September) 48:1153-1159
REALITY
The “26%” reduction in
inpatient days referred to a
change from 5.0 to 3.7 days,
a non-significant finding.
REALITY

3 of the 6 authors were in the
department of outcomes research at
the Janssen Research Foundation
in New Jersey. The statistician was a
consultant.
Hiding the data in plain
sight
Study: 68% of 88 journal articles had
deficient abstracts. The average
proportion of deficient abstracts
was 44% for 5 of the top journals
Pitkin RM, Branagan MA, Burmeister LF: Accuracy of Data in
Abstracts of Published Research Articles. JAMA.
1999;281:1110-1111
Two ways to hide data
Hide data in plain sight
 Suppress data

Suppress the Data
CELEBREX IS EASIER ON THE
STOMACH THAN IBUPROFEN – JAMA
2000
Silverstein, F. et al. “Gastrointestinal Toxicity With Celecoxib vs
Nonsteroidal Anti-inflammatory Drugs for Osteoarthritis and
Rheumatoid Arthritis The CLASS Study: A Randomized Controlled Trial”
JAMA. 2000;284:1247-1255
Publication bias

Nix negative trials

Selectively report only positive results

Republish positive results
Tramer, MR BMJ 1997
"Journals have devolved
into informationlaundering operations for
the pharmaceutical
industry."
Richard Horton, editor of The Lancet,
in “The Dawn of McScience,” The New York Review of Books, 2004
IMPACT of MARKET DRIVEN RESEARCH
more me-too drugs
more polypharmacy
longer term use of drugs
more inappropriate use
lower treatment adherence
more risk of side effects
IMPACT
Private industry now
leverages the direction of
80 percent of clinical
research.
FIDUCIARY RESPONSIBILITY





Physician - Patient
Press - Public
Government - Public
Lawyer - Client
Pharmaceutical CEO - Stockholders
Questions for Sources
1. Who funded the study?
2. Who was in charge of the data?
3. Who wrote the paper?
4. Do you have any financial conflicts of
interest with the manufacturer?
5. Do you have any conflicts of interest with
any drug or biotech company, and if so,
which ones and what type?
6. Who might have a different interpretation
of your results?
THE LIST


Go to www.overtreated.com to the
section “For doctors and the media”
for access to our list of sources with
no COI
Contact us as
[email protected]
IMPACT of MARKET DRIVEN RESEARCH

COX-2s are less effective and more
dangerous – 60,000 excess deaths
associated with Vioxx


Antidepressants have more risks and
fewer benefits than previously thought,
and many have significant withdrawal
syndromes when discontinued.


The govt. funded CATIE study is showing
that new, expensive antipsychotic drugs
are not necessarily more effective or
better tolerated than older, cheaper drugs.
Follow up study of 45 highly cited and
clinically influential articles in key journals
(1990-2003) that claimed a drug was effective:



76% were challenged by subsequent
studies (that were at least as well
designed or better designed than the
original).
Of those, in 41% subsequent studies
showed no effect or less effect than
originally claimed.
Six of the original studies were not
randomized controlled studies, and 83%
(5/6) were subsequently contradicted.
Ioannidis JP: Contradicted and Initially Stronger Effects in
Highly Cited Clinical Research. JAMA. 2005;294:218-228.