drug company funding of clinical psychiatric research related to outcomes robert kelly, md beth...
TRANSCRIPT
Drug Company Funding of Clinical Psychiatric Research
Related to Outcomes
Drug Company Funding of Clinical Psychiatric Research
Related to Outcomes
Robert Kelly, MD
Beth Israel Medical Center
New York, New York
Co-authorsCo-authors
Lisa J. Cohen, PhD
Randye J. Semple, PhD
Philip Bialer, MD
Alison Bodenheimer, BA
Elana Neustadter, BA
Arkady Barenboim, MD, PhD
Igor Galynker, MD, PhD
AcknowledgementsAcknowledgements
Ramin Mojtabai, MD, PhDTheresa Perlis, PhDAdam Cohen, MDKatherine DuHamel, PhDDaniel Eisenberg, MDMatthew Steinfeld, BA
Financial InterestsFinancial Interests
I have no significant financial or other relationship with the manufacturer of any product or service I intend to discuss
--Robert Kelly
Focus on Conflicts of InterestFocus on Conflicts of Interest Increasing Media & Public Attention
“Spitzer Sues a Drug Maker, Saying It Hid Negative Data”
– The New York Times, June 3, 2004
“What does the eight-hundred-pound gorilla do? Anything it wants to.”
– Angell, New York Review of Books, July 4, 2004
“Worrisome Ailment In Medicine: Misleading Journal Articles”
– Wall Street Journal, May 5, 2005
Focus on Conflicts of InterestFocus on Conflicts of Interest
Pressure on Policy-makersPublic reaction to media coverage
Political pressure to fix problems
Individual incidents raise public outcry
Resulting in need for action
Evidence LimitedEvidence Limited
Anecdotes
Speculation
Studies of Funding vs Outcomes
Analyses of Methods/Reporting
Previous Work of Interest-Bekelman et al. 2003
Previous Work of Interest-Bekelman et al. 2003
Reviewed studies of sponsorship-outcome association
Compared new medication to placebo or medication in use.
Four studies used blinded outcome raters Outcome: New drug favorable? Sponsorship assessed as industry vs non-
industry rather than related to drug studied.
Previous Work of Interest-Heres et al. 2006
Previous Work of Interest-Heres et al. 2006
Head-to-head comparisons of second-generation antipsychotic medications (N = 21)
Blinded outcome ratersOutcomes favored study sponsorAnalysis of Methods/ReportingRecommendations
GoalsGoals Examine sponsorship-outcome
associationFocus: Sponsorship-drug relationshipPublished clinical psychiatric studiesBroad range of original/regular articlesBlinded outcome ratersControl for potential confounds
Measure extent of pharmaceutical company sponsorship
Assess how these phenomena have changed over time
Selection of ArticlesSelection of Articles Original/regular articles Four leading psychiatric journals
AGP = Archives of General Psychiatry
AJP = American Journal of Psychiatry
JCP = Journal of Clinical Psychiatry
JCPP = Journal of Clinical Psychopharmacology Two years (N = 850)
1992
2002
Selection of ArticlesSelection of Articles Excluded articles
abstracts did not name any drug used in treatment (n=528)not dealing with issues directly related to treatment (n=19)whose abstracts mention a drug company or brand name (n=1)previously seen by any of the reviewers (n=0)whose sponsors could not be properly categorized (n=1)
Study sample: 301 articles mentioning 542 drugs
Outcome Rating ConsiderationsOutcome Rating Considerations
Goal: Measure influenceBehaviorOpinionTrained raters + guidelines
ChoicesMeasure (e.g., Favorable or not)Operationalized vs. subjectiveBlinded vs. unblindedSelection criteria
Outcome RatingOutcome Rating
Two raters Drug outcomes: Favorable vs. not
favorable Detailed rating guidelines Subjective approach Blind with respect to knowledge of sponsor Only abstracts viewed Results compared and disagreements
resolved by discussion.
Sponsorship RatingSponsorship Rating Four categories:
Same company sponsorship (S)Competing company sponsorship (C)Mixed sponsorship (M)Non-pharmaceutical sponsorship (N)
Sponsorship type
Sponsors olanzapine risperidone
None N N
Janssen C S
Lilly S C
Janssen + Lilly M M
Research QuestionsResearch Questions1. Drug company sponsored studies
increased?2. Favorable outcomes increased?3. Same company favorable outcomes >
non-pharmaceutical?4. Competing company favorable
outcomes < non-pharmaceutical?5. Mixed sponsorship outcomes different
than non-pharmaceutical?
Drug Company SponsorshipDrug Company Sponsorship
25
57***
0
20
40
60
80
100
Percent
Year
1992
2002
* = p < 0.05** = p < 0.01*** = p < 0.001
Favorable OutcomesFavorable Outcomes
52 50
0
20
40
60
80
100
Percent
Year
1992
2002
* = p < 0.05** = p < 0.01*** = p < 0.001
Favorable Outcomes byType of Sponsorship
Favorable Outcomes byType of Sponsorship
78***
48***
28***
0
20
40
60
80
100
Percent
Sponsorship
Same Co
Non-pharm
Comp Co
* = p < 0.05** = p < 0.01*** = p < 0.001
Favorable Outcomes byType of Sponsorship
Favorable Outcomes byType of Sponsorship
78*
50
30*
78***
46**
28***
0
10
20
30
40
50
60
70
80
90
100
Percent
1992 2002
Same Co
Non-pharm
Comp Co
* = p < 0.05** = p < 0.01*** = p < 0.001
Favorable Outcomes byType of Sponsorship
Favorable Outcomes byType of Sponsorship
91*
54
33*
75*
42
29*
87***
58***
22***
64**
3835*
0102030405060708090
100
Percent
AGP AJP JCP JCPP
Same Co
Non-pharm
Comp Co
* = p < 0.05** = p < 0.01*** = p < 0.001
Control for ConfoundsControl for Confounds Example: Ice cream consumption +
death by drowningPositively correlatedIce cream eating precedes drowning=> Ice cream risk factor for drowning!
Possible confoundsTemperatureWeatherSeasonProximity to water
No significant correlation after control
Possible ConfoundsPossible Confounds Year Journal Drug studied Time since FDA drug approval Diagnosis Sample size Study design variables
Placebo comparisonDouble-blind, single-blind, open-label, or case series/chart reviewUse of comparison drugUse of non-drug comparison treatment
Logistic RegressionLogistic Regression N = 205 (1 drug/article; missing values) TFDA related to outcome (p=0.01) Sponsorship related to outcome
(p=0.001) Favorable outcomes for same company
sponsorship> competing company sponsorship (OR=0.07, p<0.001)> mixed sponsorship (OR=0.14, p=0.02)> non-pharmaceutical sponsorship (OR=0.19, p=0.004)
LimitationsLimitations
Potential confounds not consideredSelective fundingSelective publication–Easterbrook, Lancet 1991
Warranted vs. unwarranted influence
Influence warranted when drug companies encourage doctors to make the right decisions.
Why Sponsorship-Outcomes Relationship?Why Sponsorship-Outcomes Relationship? Safer, 2002; Heres, 2006 Method Modifictions
Dosing schedulesStudy endpointsStudy time framesMeasurement scalesStatistical proceduresInclusion/exclusion criteria
Reporting ModificationsHighlighting findings favorable to sponsorEditorializing in the abstract
Restrictive Approach DrawbacksRestrictive Approach Drawbacks
“We desperately need new medications for substance abuse treatment, so we should encourage the pharmaceutical companies to invest.”—Petros Levounis, MD, Director of The Addiction Institute of New York
“Although scandals, real or perceived, have a short lifetime, unmet health needs persist.”—Thomas P. Stossel, MD, NEJM 2005
Treatment Without Side-EffectsTreatment Without Side-Effects Registration of clinical trials Conflict of interest transparency Dual sponsorship encouragement Study design/reporting modifications education
Articles– Safer, 2002– Heres, 2006– Lewis & Warlow, 2004– Montori, 2004
Newsletters– Carlat Psychiatry Report
Most important lesson: Outcomes depend on research question asked
Courtroom AnalogyCourtroom Analogy Plaintiff & Defendant = Competing
Pharmaceutical Companies Jury = General Practitioners and
Psychiatrists in Clinical Practice Judge = Leading Experts