the pharmaceutical promotional literature a users guide
TRANSCRIPT
The Pharmaceutical Promotional Literature
A User’s Guide
Case
It is a busy day in your practice and you are sitting at your desk, legs up, leafing through a recent issue of Diversion, The Magazine for Physicians at Leisure. You come across an ad for Plavix,TM which states that this medication reduces the risk of cardiovascular events by 9% compared to aspirin. You wonder if you should be switching all your patients to Plavix. TM
24.9
17.8
21
15.713.9
12.511
9
0
5
10
15
20
25
30
1996 1997 1998 1999 2000 2001 2002 2003
Pro
mot
iona
l exp
endi
ture
s ($
bil
lion
s)
Promotional spending on prescription drugs, l996-2003
Source: IMS Health
DTC ads 12.85%$3.2billion
Samples 66% billion $16.4
Detailing to doctors 21.3%$5.3 billion
Promotional spending on prescription drugs, 2003
Total spending: $24.9 billionSource: IMS Health
But isn’t all this advertising and promotion a good thing? Isn’t it an important way for doctors to learn about new products?
Scientific versus commercial sources of influence
• Telephone questionnaire of 85 randomly selected internists in Boston area
• Questioned about two classes of drugs: – Propoxyphene analgesics
– Cerebral and peripheral vasodilators.
Am J Med 1982;273:4
Scientific versus Commercial Sources of Influence
Am J Med 1982;273:4
62%
4%4%
68%
0%
20%
40%
60%
80%
100%
Scientific papers Drug ads
Very important Minimally important
Scientific versus Commercial Sources of Influence
Am J Med 1982;273:4
71%
32%
49%
0%
20%
40%
60%
80%
100%
Impaired cerebralblood flow majorcause of dementia
Vasodilators usefulin managing
"confused geriatricpatients"
Propoxyphene morepotent than aspirin
% o
f ph
ysic
ians
Pharmaceutical Advertisements in Leading Medical Journals: Experts’ Assessments
• “Peer review” of all ads from 10 journals during January, 1990.
• 109 advertisements were analyzed by 113 experienced physician peer reviewers and 54 clinical pharmacists.
• 71% of reviewers had received money from the drug industry within the past 2 years; 53% had received more than $5000.
Ann Int Med 1992;116:912
Pharmaceutical Advertisements in Leading Medical Journals: Experts’ Assessments
FDA regulations specify that ads are false, lacking in fair balance, or otherwise misleading if:
• They make claims about relative safety and efficacy or about the populations in which the drug is useful that are not supported by the current literature.
• Use literature or references inappropriately to support claims in the advertisement.
• Use statistics erroneously.
• Use headlines, sub-headlines, or pictorial or other graphic material in way that is misleading.
Ann Int Med 1992;116:912
Pharmaceutical Advertisements in Leading Medical Journals: Experts’ Assessments
Ann Int Med 1992;116:912
3044
57
92
0
20
40
60
80
100
Disagreedwith DOC
claim
Ad wouldlead toproper
prescribing
Little or noeducational
value
Not incompliancewith 1 or
more FDAcriteria
The Quantity and Quality of Scientific Graphs in Pharmaceutical Advertisements
• Review of all pharmaceutical ads in from 10 leading American journals in 1999.
• 498 unique advertisements (3,185 total).
• 74 unique graphs
JGIM 2003;18:294-297
The Quantity and Quality of Scientific Graphs in Pharmaceutical Advertisements
• 36% of graphs contained “numeric distortion.”
• 66% of graphs contained “chart junk.”
• 54% reported intermediate outcomes.
JGIM 2003;18:294-297
Are the risk reductions relative or absolute?
Dead Alive
Therapy 8 92 100
Placebo 12 88 100
Dead Alive
Therapy 8 92 100
Placebo 12 88 100
Risk (Rx) = 8/100 = 8%Risk (Pl) = 12/100 =12%
Dead Alive
Therapy 8 92 100
Placebo 12 88 100
Relative Risk(RR) = Risk (Rx)/ Risk (Pl) = .08/.12 = .67
Relative Risk Reduction (RRR) = 1 - RR = 1- .67 = .33 or 33%
Dead Alive
Therapy 8 92 100
Placebo 12 88 100
Absolute Risk Reduction (ARR) = Risk (Pl) - Risk (Rx) = .12 - .08 = .04 or 4%
Number Needed to Treat (NNT):
NNT = 1/ARR
Number of patients needed to treat to prevent one outcome
NNT = 1/ARR
ARR = 4%NNT = 1/.04 = 25
Completeness of reporting trial results: effect on physicians’ willingness to prescribe
• Questionnaire concerning Helsinki Heart Study.• 148 Italian physicians completed questionnaire.
Results of HHS:• Cardiac events in treatment group: 2.73%• Cardiac events in placebo group: 4.14%
Lancet 1994. 343; 1209
Completeness of reporting trial results: effect on physicians’ willingness to prescribe
• ARR = 1.41 %• RRR = 34%• NNT = 71• Difference in event free rates (97.3% vs 95.9%)• RR of cardiac events - RI deaths = 6%
Lancet 1994. 343; 1209
Completeness of reporting trial results: effect on physicians’ willingness to prescribe
“You are in doubt whether to start drug treatment to reduce serum cholesterol of one of your patients. We will gave you 5 statements derived from 5 different randomized trials recently published in leading medical journals. On the basis of each statement you should indicate how likely you are to prescribe each drug for your patient. Assume that the dosage is the same for each treatment.”
Lancet 1994. 343; 1209
Completeness of reporting trial results: effect on physicians’ willingness to prescribe
Likelihood of prescribing:• Drug “A” (RRR) = 77%• Drug “B” (ARR) = 24%• Drug “C” (% event free) = 37%• Drug “D” (NNT) = 34%• Drug “E” (complete) = 28%
Lancet 1994. 343; 1209
P < 0.001 for RRR vs other measures
Are the results statistically significant?
Are they clinically significant?
Are the graphs telling the truth?
Are the graphs telling the truth?
• Does the size of the effect shown equal the size of the effect in the data?
Tufte’s Lie Factor:
Size of effect shown in graphic
Size of effect in data
Are the graphs telling the truth?
• Does the size of the effect shown equal the size of the effect in the data?
• Is only a small percentage of the possible event rate displayed?
Are the graphs telling the truth?
• Does the size of the effect shown equal the size of the effect in the data?
• Is only a small percentage of the possible event rate displayed?
• Does the y-axis start at zero?
Are the graphs telling the truth?
• Does the size of the effect shown equal the size of the effect in the data?
• Is only a small percentage of the possible event rate displayed?
• Does the y-axis start at zero?
• Is the survival curve longer than the study?
Are the references “real?”
Is Cal Ripken in the ad?
(Appeal to celebrity)
Logical Fallacies in Pharmaceutical Promotion
Argumentum ad populum
Appeal to popularity
J Gen Intern Med 1994;9:563-7
Logical Fallacies in Pharmaceutical Promotion
Argumentum ad verecundiam
Appeal to authority
J Gen Intern Med 1994;9:563-7
Logical Fallacies in Pharmaceutical Promotion
Argumentum ad celebritam
Appeal to celebrity
Logical Fallacies in Pharmaceutical Promotion
Fallacy of ignoratio elenchi
(or fallacy of irrelevant conclusions,
or fallacy of ignoring the issue
or the non-sequitur)
J Gen Intern Med 1994;9:563-7
Logical Fallacies in Pharmaceutical Promotion
Appeal to emotion
Check-list
• Are the risks relative or absolute?
Check-list
• Are the risks relative or absolute?
Relative.
Absolute = 0.9%
Check-list
• Are the risks relative or absolute? Relative
• Is the result statistically significant?
Check-list
• Are the risks relative or absolute?
• Is the result statistically significant?
Yes, marginally.
P = .045 95% CI (0.3% to 16.5%)
Check-list
• Are the risks relative or absolute? Relative
• Is the result statistically significant? Yes
• Is the result clinically significant?
Check-list
• Are the risks relative or absolute? Relative
• Is the result statistically significant? Yes
• Is the result clinically significant?
NoNNT = 1/ARR =1/.009 =111 95%CI (57 - 2500)
Check-list
• Are the risks relative or absolute? Relative
• Is the result statistically significant? Yes
• Is the result clinically significant? No
• Does the size of the effect shown equal the size of the effect in the data?
Check-list
• Are the risks relative or absolute? Relative
• Is the result statistically significant? Yes
• Is the result clinically significant? No
• Does the size of the effect shown equal the size of the effect in the data? No
Check-list
• Are the risks relative or absolute? Relative
• Is the result statistically significant? Yes
• Is the result clinically significant? No
• Does the size of the effect shown equal the size of the effect in the data? No
• Are the references "real?”
Check-list
• Are the risks relative or absolute? Relative• Is the result statistically significant? Yes• Is the result clinically significant? No• Does the size of the effect shown equal the
size of the effect in the data? No• Are the references "real?”Yes, the “CAPRIE” study, The Lancet, Vol. 348,
November 16,1996.
Check-list
• Are the risks relative or absolute? Relative
• Is the result statistically significant? Yes
• Is the result clinically significant? No
• Does the size of the effect shown equal the size of the effect in the data? No
• Are the references "real?” Yes
• Is Cal Ripken in the ad?
Check-list
• Are the risks relative or absolute? Relative
• Is the result statistically significant? Yes
• Is the result clinically significant? No
• Does the size of the effect shown equal the size of the effect in the data? No
• Are the references "real?” Yes
• Is Cal Ripken in the ad? No, thankfully.
Conclusions
• Pharmaceutical ads are often inaccurate, biased, and misleading.
• They misuse statistics and graphics, over-state results, and employ fallacious reasoning.
• They should not be used to guide clinical decisions.
• Keep your patients on aspirin!
A few sources of prescribing information
• Medical Letter (http://www.medicalletter.com)
• Prescriber’s Letter (http://www.prescribersletter.com)
• Therapeutics Initiative (http://www.ti.ubc.ca)
• Drug and Therapeutics Bulletin (UK)
(http://www.dtb.org.uk/idtb)