PRESIDENT’S INVITED COLUMN
Too Quick to Impugn
JAMA Editorialists Doubt Integrity of
Statisticians Working in For-Profit World
Peter A. Lachenbruch, Donald Stablein, and Janet Wittes
C
atherine D. DeAngelis and Phil B. Fontanarosa, in their the experiment, gathering and validating the data, analyzing the
editorial “Impugning the Integrity of Medical Science: The data, and communicating the results. All are important aspects of
Adverse Effects of Industry Influence” that appeared in the the statistician’s role, and the properly prepared statistician will
April 16, 2008, issue of the Journal of the American Medical want to participate in each of them. In our experience, however,
Association, aim to influence the way statisticians function in med- the first three phases generally require the most thought and often
ical research studies. Their proposals would at one fell swoop take the most time. The fifth, communication, is often the most
remove many experienced biostatisticians from performing work difficult. So we are surprised and disappointed that the editorial-
they have done for decades and reshape the face of academic ists appear to separate the statisticians who determine the question
departments of biostatistics. and design the experiment from those who gather, validate, and
The editorial discusses two articles that appear in the same issue. analyze the data and prepare the manuscript. In the editorialists’
One article, “Guest Authorship and Ghostwriting in Publications own words:
Related to Rofecoxib: A Case Study in Industry Documents from
Rofecoxib Litigation,” deals with people who accepted authorship, For-profit companies that sponsor biomedical research
even first authorship, on a publication for which they had lim- studies should not be solely or primarily involved in
ited roles. The issues raised relate more generally to ghost-written collecting and monitoring of data, in conducting the
articles—those written by technical writing subcontractors without data analysis, and in preparing the manuscript reporting
attribution or acknowledgement. study results. These responsibilities should be primarily
The second article, “Reporting Mortality Findings in Trials of or solely be performed by academic investigators who are
Rofecoxib for Alzheimer Disease of Cognitive Impairment: A Case not employed by the company sponsoring the research.
Study in Documents from Rofecoxib Litigation,” discusses selective
reporting of data on safety and efficacy. The authors comment that This suggestion is in two parts. The first attacks statisticians
the reported results do not agree with reports from the company’s from the sponsoring institution; the second attacks those working
internal documents. Further, the rofecoxib report used a per proto- outside of academe. Specifically, the first part of the recommenda-
col analysis for safety, rather than an intent-to-treat analysis. While tion removes data collection and analysis from the sponsor; the
many agree with the authors about preferring intent-to-treat analy- second puts those responsibilities only in the hands of academics.
ses for safety, we recognize many other thoughtful people prefer per These two parts are quite different. The first appears to arise from
protocol analyses for these data. the editorialists’ loss of trust in the intellectual honesty of spon-
De Angelis and Fontanarosa propose several solutions to the issues sors; the second implies the editorialists believe that only in the
raised in the articles. We agree with many of those solutions. We, halls of academe can one find honest statisticians.
similar to the editorialists, are opposed to unacknowledged ghost
writing, to adding people to papers as authors when those people
Okay, but I’m Confused
have had little to do with the study or the writing, to misrepresent- We recognize that sponsors of research have the greatest interest in
ing data, to failing to acknowledge sources of support or conflicts of showing a product is effective and that such interest can adverse-
interest, and to other egregious activities. But the editorial goes fur- ly influence the integrity of their analyses; however, if the study
ther than addressing intellectual dishonesty; the editorial impugns involves a new drug, the sponsor must satisfy the FDA, among
the integrity of all statisticians who work for any for-profit organiza- others, of the validity and integrity of their work. Moreover, stat-
tion. We are frankly baffled by their attack. The cited problems and isticians in industry are seriously interested in learning the truth of
papers hardly lead to the conclusion that statisticians in for-profit the effect of their products. The sponsor has the most at stake in
companies cannot be trusted In fact, it doesn’t make sense. presenting data honestly and completely. As Kenneth J. Rothman
and Stephan Evans in their 2005 British Medical Journal article
Surprised and Disappointed
note, “[Scientific] review ought to rest on the content of a submis-
Before reviewing the editorialists’ recommendation for statistical sion, rather than solely on the basis of presumptions inferred from
analysis of biomedical research, consider Donald Bentley’s steps in group affiliation, such as [employer].”
statistical analysis. Citing Lincoln Moses in a discussion published While we understand the impetus for the first part of the recom-
in Volume 2(1) 2008 of Annals of Applied Statistics, Bentley men- mendation, we are truly confused by the second. The editorial would
tions five phases of a research project in which an applied statistician send all analyses to academic institutions. Nowhere do the editori-
should be intimately involved: determining the question, designing alists provide evidence that statisticians in for-profit organizations
2 AMSTAT NEWS JULY 2008
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68 |
Page 69 |
Page 70 |
Page 71 |
Page 72 |
Page 73 |
Page 74 |
Page 75 |
Page 76 |
Page 77 |
Page 78 |
Page 79 |
Page 80