a proposed world council on epidemiology and causation ...€¦ · web viewword count for main text...
TRANSCRIPT
A proposed World Council on Epidemiology and Causation: summary of feedback and considerations in an international workshop.
Raj Bhopal,
Professor of Public health, University of Edinburgh, Scotland UK
Details for correspondence:
R S Bhopal, Bruce and John Usher Professor of Public Health
Centre for Population Health Sciences,
Usher Institute of Population Health Sciences and medical Informatics,
University of Edinburgh,
Teviot Place,
Edinburgh EH89AG
Telephone (0)131 650 3216 (switchboard extension 1000), Fax (0)131 650 690
e-mail [email protected]
Word count for main text: about 1850 words
Word count for abstract: 157 words
1
Abstract (157 words)
Rigorous evaluation of associations in epidemiology is essential, especially given big data,
data mining and hypothesis-free analyses. There is a precedent in making judgements on
associations in the monographs of the International Agency for Research on Cancer,
however, only the carcinogenic effects of exposures are examined. The idea of a World
Council of Epidemiology and Causality (WCEC) to undertake rigourous, independent,
comprehensive examination of associations has been debated, including in a workshop at the
International Epidemiology Association’s (IEA) 20th World Congress of Epidemiology, 2014.
The objective of the workshop was both to, briefly, debate the idea and set out further
questions and next steps. The principal conclusion from feedback including from the
workshop is that the WCEC idea, notwithstanding challenges, has promise and deserves more
debate. The preferred model is for a small independent body working closely with relevant
partners with a distributed approach to tasks. Recommendations are contextualised in
contemporary approaches in causal thinking in epidemiology.
2
Key messages
Moving from association to causation in epidemiology is extremely difficult and requires
expert judgement
Currently, a formal mechanism for this process only exists for carcinogens and cancer
outcomes
There may be merit in developing formal mechanisms for other controversial risk factors and
other outcomes
One formal mechanism that has been considered is a council (interim name, World Council
of Epidemiology and Causality)
Feedback on the proposal, including an international workshop, has perceived the promise of
the idea and calls for serious deliberation and practical action
3
Introduction
The association between risk factors and disease outcomes in observational human
population studies, being a precursor to experimental research where that is possible,
underpins the causal contributions of epidemiology.1;2 This approach has provided massive
rewards in public health and medicine3-5 but has led to important errors, one example being
the premature, widespread use of hormone replacement therapy to prevent cancer and heart
disease in women.6 The detailed, thoughtful evaluation of associations is increasingly
important in the context of big data, data mining and hypothesis-free analyses as exemplified
by genome wide association studies where each gene variant is a potential risk factor.7;8 As
the number of associations examined grows, so does the risk of misinterpretation, and the
principles for avoiding this need revisited.9 The critical issue is to avoid the declaration of
non-causal associations as causal, and causal ones as non-causal. This commentary reflects
on the potential of a new institution to help epidemiologists in their causal endeavours.
A proposal for an International Council as an impartial, institution-based response to the
challenges of moving from association to causation in epidemiology was published in 2008.10
In a lecture in 2008 the proposed name was changed to the World Council of Epidemiology
and Causality (WCEC).11 The vision for this proposal in 2008 was as follows:
“Epidemiology needs to provide partners who apply research, including politicians, doctors,
and public health specialists, with a unified voice. Is it not time for a World Council in
Epidemiology and Causality that provides authoritative statements on epidemiological
evidence and makes recommendation on when and how epidemiological data on associations
are ready for application?”11 The concept was for the WCEC to promote collegiate
discussions leading to consensual statements subject to revision as new evidence emerged.
4
Institutional responses to scientific problems are bound to be scrutinised warily by
researchers and scholars. There are, however, successful precedents for this kind of work that
have strengthened rather than oppressed research. We might achieve for the assessment of
associations what the Cochrane collaboration (http://www.cochrane.org; accessed 25/2016)
does for the effectiveness of interventions. The Cochrane Collaboration was inspired by
Cochrane’s plea in 1972 that health services should provide effective interventions, and that
this required evidence.12 The (UK) Cochrane Centre formed in 1993. It is now a global, not-
for-profit, independent network of more than 20,000 authors in 53 review groups in 120
countries. Its goal is to provide evidence, principally through systematic reviews and
metaanalysis, of heath care interventions. Could something similar be achieved for
epidemiological associations?
A precedent in epidemiology is the monograph series of the International Agency for
Research on Cancer (IARC) which judges the causal basis of cancer-related risk factors.13
However, only the carcinogenic effects of exposures are examined but that is insufficient
from a public health perspective. The recent controversy about red meat and processed meat
and cancers, the subject of recent IARC monograph, illustrates the point: these exposures
have important effects beyond cancer e.g. in cardiovascular disease.14 We need a broader
perspective which considers the causal basis of the range of benefits and harms associated
with each exposure.
The feedback on the proposal for a WCEC has been positive, with important cautions on
potential harms. This commentary aims to summarise feedback on the idea of a WCEC, distil
discussions at a workshop 2014, and conclude with some potential next steps.
5
Summary of scholarly feedback on the WCEC concept
The feedback could be summarised in the following question (paraphrased): The WCEC
concept is an interesting idea, but will it work and if so how, and who will fund it? Some
illustrative feedback is given anonymously unless it has been published. Unpublished
feedback showed a high level of engagement and raised questions and challenges as reflected
in the following examples: we need examples of problems in public policy that could be moved
forward with the WCEC; the aims of WCEC need clarifying, for example, will it be purely a
scholarly organisation focusing on causality or will it have other roles supporting
epidemiologists?; we need a consensus amongst epidemiologists on the WCEC idea and this
could be done at major epidemiological congresses; we need to delineate the roles of WCEC
in relation to other relevant organisations e.g. WHO , and we need their backing; we need
examples of success in producing statements on causality ; and, two approaches for
producing causality statements include thinking of causality as a classification problem with
numerical assessments or thinking of it as a consensus by panels based on evidence.
Some feedback has been published. Two letters in response to the paper proposing a WCEC11
were supportive and suggested additional challenges for it
(http://www.ete-online.com/content/6/1/6/comments; accessed 28/10/15). The IEA
(International Epidemiological Association) Bulletin Board received comments from Rhutty
(29/10/2010) saying that the “WCEC may be an idea whose time has come, however, there
don’t seem to be many models to follow” and from Birpal (29/03/2010) saying “Move on
with your idea of a WCEC”. (The bulletin board is no longer accessible online.)
Vandenbroucke’s blog hosted by the journal Epidemiology considered the WCEC idea and
proposed developing guidelines to help think about the credibility of epidemiologic findings
6
(http://journals.lww.com/epidem/blog/watching/pages/post.aspx?PostID=4; accessed
28/10/15).
The recommendation that the WCEC be discussed at epidemiological congresses was
implemented in my lecture at the Annual Scientific Meeting of the Society for Social
Medicine in 2013 (http://socsocmed.org.uk/meetings/past-lectures; accessed 28/10/2015).
The audience supported the concept and recommended more detailed consideration, which
took place in 2014 as discussed below.
Considerations at a workshop of epidemiologists
At the 20th World Congress of Epidemiology in 2014 I and four senior colleagues (see
acknowledgements) led a 90-minute workshop introducing the WCEC idea, including a
summary of the above feedback, after which about 80 participants debated the proposal in
five subgroups and discussed actions that would be required to develop a WCEC. The
workshop concluded with feedback from each subgroup. Notes were made by an independent
person or the facilitator, edited and summarised as in the table. (The summaries of these
discussions are available on request from the author.) Text box 1 lists some of the potential
partners for WCEC identified in these discussions. The table distills and Text box 2
summarises the main points raised.
In October 2015 the Council of the International Epidemiological Association considered the
workshop report. It thought that the WCEC’s work would be complex and challenging and
would best be done by an independent organisation.
7
Conclusions and recommendations in relation to a WCEC in the context of current
debates on causality
The idea of establishing a new institution to help establish causality from epidemiological
and related population health sciences’ data, clearly, has been judged interesting and worthy
of more detailed deliberation. The feedback has not provided unequivocal endorsement of
the idea, and has pinpointed both the potential strengths and limitations of the idea.
Viewpoints and the recommendations arising are considered below in the context of current
debates on causality.
A WCEC would need to embrace and spur developments in methods and concepts of causal
reasoning. There have been many recent advances in causal inference, especially using causal
diagrams and structural equation models,15 and discussions of ways of moving from
traditional reasoning as developed by the US Surgeon General’s report on smoking and
health and Bradford Hill (among others), which are still evolving,16-18 to quantifying the
probability of causation.19;20 Indeed, Pearl has claimed that causality has been mathematised.15
In epidemiology and public health, at least, mathematics alone does not move us from
association to causation, though it provides concepts and tools to make our causal reasoning
explicit and to test, and refine, our causal models. There have been controversies on whether
causal claims regarding non-manipulable variables are meaningful, with contradictory views
exemplified in recent discussions on the variable race.21;22;22;23 (Annals of Epidemiology will
be debating this issue separately.)
The approach of IARC in synthesising epidemiological and non-epidemiological evidence
using expert groups and consensual methods has served well within the field of cancer.13 The
judgments do not always gain public or even professional support for implementation. For
8
example, as mentioned in the introduction, the latest work on red meat, processed meat and
cancer14 is being met in the UK press with scepticism, including amongst epidemiologists,
(http://www.bbc.co.uk/news/health-34615621; accessed 28/10/15). Moreover, similar
observations and recommendations were widely publicised following a report in 2007 but
they had little effect.
(http://www.nhs.uk/news/2007/November/Pages/Cancerrecommendations.aspx; accessed
28/10/15). One of the criticisms of these reports is that there are benefits of red meat that
have not been considered. There is still much work to do in moving from association to
causation in an independent, authoritative, and convincing way, that leads to acceptance and
action.
The WCEC concept could provide a forum for spurring both methodological work and
judgements on the causal basis of associations in a collegiate, distributed and international
way. Among the debates around causality that a WCEC could help with are whether, when
and how epidemiologists should bridge the gap between science and policy implementation,
the latter requiring clarity on the causal significance of empirical evidence24; when human
experimentation (including randomized trials) are justified to enhance causal knowledge; the
role, evaluation and development of classical causal guidelines, in the context of
mathematical tools and approaches to causal analysis; and whether it is futile to seek causal
knowledge for exposures that cannot be changed e.g. race and ethnicity. In summary, a
WCEC could be a hub and a catalyst both for compiling evidence on causality and for
developing methods for deriving causal knowledge from associations. A WCEC could help
epidemiologists by succeeding in three goals i.e. catalysing and coordinating discussions on
causality, undertaking or stimulating reviews of evidence on potentially causal relationships,
and helping develop and disseminate concepts and methods for causal reasoning. The
9
accurate evaluation of the causal content of association is the greatest challenge in
epidemiology, and this proposal is a means of potentially strengthening our armamentarium.
Acknowledgements
Professors Jan Vandenbroucke, Alex Broadbent, Laurence Gruer, and Dr Roger Bernier,
helped plan, deliver and summarise the workshop and offered comments on this manuscript.
Doctor Lade Ayodele summarized discussions of subgroup 1 and Dr Margie Walling of
subgroup 5. Doctor Fredrik Norstrom offered comments on the manuscript. Mrs Anne
Houghton provided other secretarial support. Many people assisted in the implementation of
the workshop, including Professor Cesar Victora and Doctor Thomas Hennessy. Thanks to
the Council of IEA for discussing a previous draft of this paper and providing feedback.
10
Text box 1 Some potential partners for WCEC
11
WHO World Health Organisation
IARC International Agency for Research on Cancer (of the WHO)
IEA International Epidemiological Association
ISEE International Society of Environmental Epidemiology
EUPHA European Public Health Association
APHA American Public Health Association
WFPHA World Federation of Public Health Associations
NIHR National Institute of Health Research
CDC Centres for Communicable Disease Control and Prevention
INSERM Institut National de la Santé et de la Recherche Médicale;
French Institute of Health and Medical Research
US Preventive Services Task Force
Journals/ Journal Editors associations e.g. ICMJE (International
Committee of Medical Journal Editors)
Association of Medical Informatics
International Institute of publishing sciences
Specialist associations, e.g. IDF (International Diabetes Federation),
occupational health, statistics, computing
Philosophical associations
Text box 2
Eight main points, actions and recommendations arising from workshop and other feedback
1. The idea of a WCEC is potentially valuable and further discussion is
desirable. (This is not to imply there is universal support for the idea.).
2. The discussions should involve a wide range of partners including those
listed in the text box 2. WCEC should seek to absorb the best practice
from organisations working on causality (usually in specialist areas).
3. Following more detailed discussions, including the development of a
costed strategic plan, a small, central, independent office should be
established to develop a work agenda and the necessary collaborations.
The long term aim would be to have a distributed model of working.
4. One or a few important causal topics should be chosen to develop the
work, as exemplars. Long term priorities will need to be decided in
engagement with potential users of outputs/recommendations arising.
5. Funding needs to be identified for this developmental phase, preferably
from one major and several supporting sources.
6. Guidelines on causation, including on publication standards, should be
developed by WCEC and its partners. These guidelines may need to be
specialised to meet the needs of different kinds of epidemiology. These
may be incorporated into existing relevant websites.
7. While the focus of the WCEC will be on epidemiological and related
type of evidence, the causal concepts and approaches will need to draw
on a wide range of contributing disciplines.
12
8. A committee of 8-12 people should be formed to move the above ideas
forward. The committee should include representatives of
organisations that have developed causal statements.
13
Table. Ideas and key points relating to the concept of a WCEC from subgroup discussions at the workshop in 2014*
Idea/key point Sub-group(s) explicitly expressing idea/point
A. General points on concept of a WCEC and its core workConcept is worth pursuing 1, 2, 3Interdisciplinary and international approach needed 1, 3Name may need broadening 1The work would need to be distributed, i.e. a small central group with larger network
2
Goals need further discussion, especially in relation to those of existing organisations
3, 5
Causality is a historical challenge in all sciences 4Resolving causality issues leads to rapid advances and uncertainty impedes progress
5
B. Communications and EngagementPartnership with many organisations is needed, including those outside epidemiology and public health
1, 4
Need to deal with media 4Encourage counter-arguments 1Need to identify the key audiences, as meeting the needs of everyone would be difficult
5
C. Early actionsLegal standing needs clarification 4Identify potential funders (one major and supplementary funders) 2, 3Produce causal guidelines for writers of papers e.g. as a STROBE appendix or a tool-kit, or website
1, 3, 4
Identify controversial and confusing areas to work on. Could start with one topic
1, 5
Develop mechanism for identification of priority topics 1, 2Develop detailed discussion paper including examples of how WCEC could help to resolve controversies
1, 4
Disseminate and discuss ideas via conferences 1Accrue and consider different approaches to causality 4
D. Structure and organisationSelf-standing, independent body is preferred 1, 2Set up committee of 8-12 people to identify and involve external partners
1
Estimate workload and resources needed in tackling a specific controversial topic in relation to causality
4
*The summaries of the five subgroup discussions are available to readers on request from the authors.
14
Reference List
(1) Bradford Hill A. The environment and disease: association or causation? Occupational Medicine 1965;295-300.
(2) Evans AS. Causation and Disease: A Chronological Journey. American Journal of Epidemiology 1978; 108:1-238.
(3) Gilbert R, Salanti G, Harden M, See S. Infant sleeping position and the sudden infant death syndrome: systematic review of observational studies and historical review of recommendations from 1940 to 2002. Int J Epidemiol 2005; 34(4):874-887.
(4) Doll R. Uncovering the effects of smoking: historical perspective. Statistical Methods in Medical Research 1998; 7:87-117.
(5) Buck C, Llopis A, Najera E, Terris M. The Challenge of Epidemiology: Issues and Selected Readings. Washington DC: Pan American Health Organization; 1988.
(6) Krieger N, Lowy I, Aronowitz R, Bigby J, Dickersin K, Garner E et al. Hormone replacement therapy, cancer, controversies, and women's health: historical, epidemiological, biological, clinical, and advocacy perspectives. J Epidemiol Community Health 2005; 59(9):740-748.
(7) Khoury MJ. Planning for the Future of Epidemiology in the Era of Big Data and Precision Medicine. Am J Epidemiol 2015; 182(12):977-979.
(8) Khoury MJ, Little J, Gwinn M, Ioannidis JP. On the synthesis and interpretation of consistent but weak gene-disease associations in the era of genome-wide association studies. Int J Epidemiol 2007; 36(2):439-445.
(9) Ioannidis JP. How to make more published research true. PLoS Med 2014; 11(10):e1001747.
(10) Bhopal R. Concepts of Epidemiology: Integrating the Ideas, Theories, Principles and Methods of Epidemiology. 2nd ed. Oxford: Oxford University Press; 2008.
(11) Bhopal R. Seven mistakes and potential solutions in epidemiology, including a call for a World Council of Epidemiology and Causality. Emerg Themes Epidemiol 2009; 6(6).
(12) Cochrane AL. Effectiveness and Efficiency. London: The Nuffield Provincial Hospitals Trust; 1972.
(13) WHO International Agency for Research on Cancer. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. 1-25. 2006. Lyon, France, WHO.
Ref Type: Generic
(14) IARC Monograph WorkingGroup. Carcinogenicity of consumption of red and processed meat. Lancet [ 2015
(15) Pearl J. CAUSALITY. 2 ed. Cambridge: Cambridge University Press; 2009.
(16) Bradford Hill A. The environment and disease: association or causation? Occupational Medicine 1965;295-300.
15
(17) Geneletti S, Gallo V, Porta M, Khoury MJ, Vineis P. Assessing causal relationships in genomics: From Bradford-Hill criteria to complex gene-environment interactions and directed acyclic graphs. Emerg Themes Epidemiol 2011; 8(1):5.
(18) Fedak KM, Bernal A, Capshaw ZA, Gross S. Applying the Bradford Hill criteria in the 21st century: how data integration has changed causal inference in molecular epidemiology. Emerg Themes Epidemiol 2015; 12:14.
(19) VanderWeele TJ, Hernán MA. Causal effects and natural laws: Towards a conceptualization of causal counterfactuals for nonmanipulable exposures, with application to the effects of race and sex. In: Berzuini C, Dawid P, Bernardinelli L, editors. Causality: Statistical Perspexctives and Applications. John Wiley & Sons, Ltd.; 2012. 101-113.
(20) Swaen G, van AL. A weight of evidence approach to causal inference. J Clin Epidemiol 2009; 62(3):270-277.
(21) Hernan MA. A definition of causal effect for epidemiological research. J Epidemiol Community Health 2004; 58(4):265-271.
(22) Glymour C, Glymour MR. Commentary: race and sex are causes. Epidemiology 2014; 25(4):488-490.
(23) VanderWeele TJ, Robinson WR. On the causal interpretation of race. Epidemiology 2014; 25(6):937-938.
(24) Rothman K, Adami H, Trichopoulos D. Should the mission of epidemiology include the eradication of poverty? Lancet 1998; 352:810-813.
16