Conflicts of interest in medical
journalism
B
Bernard
d Lo,
L M.D.
MD
The Greenwall Foundation
University of California San Francisco
May 9,
9 2013
1
My conflicts of interest
 Ethics
Ethi Advisory
Ad i
Committee,
C
itt
R
Roche
h
2
Page 1
Questions for audience
D
Does
ffunder
d off RCT iinfluence
fl
how
h
critically you read it?
1. More critical if funded by industry
2. More critical if funded by
y NIH
3. Equally critical regardless of funder
3
Questions for audience
D
Does
ffunder
d off RCT iinfluence
fl
how
h
critically your colleagues read it?
1. More critical if funded by industry
2. More critical if funded by
y NIH
3. Equally critical regardless of funder
4
Page 2
Outline of talk
 Are
A conflicts
fli t off interest
i t
t important?
i
t t?
 Does disclosure help us review or read
critically?
 Should we focus on bias rather than
conflicts of interest?
 What are the benefits and burdens of
conflict of interests policies?
5
Are conflicts of interest important?
6
Page 3
Page 4
Concerns about these pivotal
industry-sponsored trials
 Biased
Bi
d design
d i
 Under-reporting of serious adverse
events
 Delayed
y p
publication of negative
g
findings
 Paper drafted before PI recruited
9
Concerns about these pivotal
industry-sponsored trials
 Undermine
U d
i evidence
id
base
b
for
f clinical
li i l
decisions
 Harm to patients
 Should these be never events?
10
Page 5
Association between funding and
conclusions
 Investigator
I
ti t ties
ti to
t manufacturer
f t

3.6 times more likely to find drug effective
 Trial sponsored by manufacturer

4.0 times more likely to find drug effective
JAMA 2003; 289: 454
BMJ 2003; 327: 1167
11
Conflict of Interest
in Medical Research, Education,
and Practice
www.nap.edu/catalog.php?record_id=12598
/
Page 6
What is a conflict of interest?
 Primary
Pi
interest
i t
t

Well-being of patients
 Integrity of research and education
 Secondary personal interests

Financial gain
 Professional, intellectual commitment
13
What is a conflict of interest?
 Unacceptable
U
t bl risk
i k that
th t primary
i
interest
i t
t
unduly influenced by secondary
interest
 Conclusions may be biased
 “Potential” and “perceived” not helpful
terms
14
Page 7
Conflicts of interest
Bias
What are the assumptions
underlying COI policies?
M
More
di
disclosure
l
= transparency
t
16
Page 8
What are the assumptions
underlying COI policies?
M
More
di
disclosure
l
= transparency
t

Less undue influence and bias
Greater accountability + trust
17
Does disclosure help us review or
read clinical trials critically?
18
Page 9
Special Article
A Randomized Study of How Physicians Interpret
Research Funding Disclosures
Aaron S. Kesselheim,, M.D.,, J.D.,, M.P.H.,, Christopher
p
T. Robertson,, Ph.D.,, J.D.,,
Jessica A. Myers, Ph.D., Susannah L. Rose, Ph.D., Victoria Gillet, B.A., Kathryn M.
Ross, M.B.E., Robert J. Glynn, Ph.D., Steven Joffe, M.D., and Jerry Avorn, M.D.
N Engl J Med
Volume 367(12):1119-1127
September 20, 2012
How do MDs use disclosure?
 MDs
MD read
d abstracts
b t
t for
f hypothetical
h
th ti l
trial that varied in rigor and funding
source

Random assignment
 How wiling are you to

Believe findings?
 Prescribe drug?
20
Page 10
.
Kesselheim AS et al. N Engl
J Med 2012;367:1119-1127
Kesselheim AS et al. N Engl
J Med 2012;367:1119-1127
Page 11
Conclusions
• … Industry sponsorship … reduces [MD] their
willingness to believe and act on trial findings,
i d
independently
d tl off th
the ttrial's
i l' [methodologic]
[ th d l i ] quality.
lit
• These effects may influence the translation of clinical
research into practice.
What are the challenges and
limitations of disclosure?
24
Page 12
Challenges with disclosure
1 Proportionality:
1.
P
ti
lit focus
f
on
relationships that present the
greatest risk of undue influence

But behavioral economics: effects of small
monetary amounts
25
Challenges with disclosure
2 Accuracy:
2.
A
di
distinguish
ti
i h off
relationships with different risk

Scientific consulting vs. speakers bureaus
 Grant to support research expenses vs.
stock options for researcher
26
Page 13
Challenges with disclosure
3 Inconsistent
3.
I
i t t disclosures
di l

Multiple disclosures: grant, abstract or
manuscript, CME talk, committee
 Many accessible to public
• Sunshine law website

Trigger suspicion, allegations
• But disclosure requirements inconsistent

Proposals for standard format
27
Challenges with disclosure
4 Unintended
4.
U i t d d adverse
d
consequences

Some disclosures ineffective or
counterproductive
 Deter desirable academia-industry
collaborations
 Undue administrative burdens and
opportunity costs
28
Page 14
COI policies:
Beyond disclosure …
 Disclosure
Di l
a necessary step,
t
but
b t just
j ta
first step

Does not remove bias
 May even be counterproductive
 Management or prohibition of certain
relationships
29
Should we focus on bias in clinical
trials rather than on conflicts of
interest?
30
Page 15
Perhaps not …
1 Limits
1.
Li it on critical
iti l reading
di
2. Enhance readers’ ability to evaluate
validity
3. Strengthen
g
evidence base from
clinical trials
31
1. Limits on critical reading
 Is
I it possible
ibl to
t read
d all
ll RCTs
RCT equally
ll
critically?
32
Page 16
1. Limits on critical reading
 Human
H
attention,
tt ti
concentration,
t ti
mental effort have limits, lapses
 Use of heuristics
 Overconfident appraisal
pp
of own
decision-making
34
Page 17
2. Enhance readers’ ability to
evaluate validity
 Explicit
E li it validity
lidit criteria
it i for
f reviewers
i

Appropriate comparator, endpoints?
 Summary of reviews and responses?
 Commentary after publication?

Use Internet to pool expert judgments
35
3. Strengthen evidence base
 Access
A
tto clinical
li i l trial
t i l results
lt for
f other
th
researchers, public

Limits of clinicaltrials.gov
 For secondary analyses and meta-analyses,
need clinical study reports or anonymized
participant-level outcomes
 Recent policy of European Medicines
Agency
36
Page 18
3. Strengthen evidence base
 Access
A
tto clinical
li i l trial
t i l reports
t

IRB should require sponsor and PI to state
whether access will be granted
 IRB should require disclosure about access
in consent form
• Participants may value access to contribute to
generalizable knowledge, help future patients
37
3. Strengthen evidence base
 Access
A
t clinical
to
li i l trial
t i l reports
t

Journals should require access as condition
of acceptance
• BMJ policy
38
Page 19
3. Strengthen evidence base
 Specifics
S
ifi regarding
di
access to
t be
b
resolved

Time lag before disclosure
 Protection of trade secrets
 Interventions not approved for marketing
 Confidential access for scientific review
panels, IRBs, DSMBs for other trials
 Procedures for access
• Submit protocol?
P
i
bli h
39
l ?
Criticisms of COI policies
 Responses
R
outt off proportion
ti
to
t
problems

Overreaction to egregious cases
 Undue burdens on MDs and
institutions
40
Page 20
Criticisms of COI policies
 Weak
W k and
d inconsistent
i
i t t
 Not enforced
41
Evaluation of COI policies
 Effective
Eff ti iin achieving
hi i
goals?
l ?
 Unintended adverse consequences?
 Administrative and opportunity costs?
 Need empirical research to base COI
policies on evidence rather than
assertions
42
Page 21
Take home message
 COI policies
li i mustt balance
b l
countert
vailing goals, values, considerations
 Focus on bias as well as COIs

Professional societies and journals should
take
k lleadd
43
Page 22