Decision-Making under Uncertainty in the Setting of Environmental Health Regulations Author(s): James M. Robins, Philip J. Landrigan, Thomas G. Robins, Lawrence J. Fine Source: Journal of Public Health Policy, Vol. 6, No. 3 (Sep., 1985), pp. 322-328 Published by: Palgrave Macmillan Journals Stable URL: http://www.jstor.org/stable/3342398 Accessed: 23/10/2008 14:39 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=pal. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that promotes the discovery and use of these resources. For more information about JSTOR, please contact
[email protected].
Palgrave Macmillan Journals is collaborating with JSTOR to digitize, preserve and extend access to Journal of Public Health Policy.
http://www.jstor.org
Decision-MakingUnder Uncertaintyin the Setting of EnvironmentalHealth Regulations JAMES M. ROBINS, PHILIP J. LANDRIGAN, THOMAS G. ROBINS, and LAWRENCE J. FINE
INTRODUCTION 4Th
ca~(a
i
i1
IR Austin Bradford Hill promulgated guidelines (1) for 2
S
_
:r O cf'*)
determiningthe likelihoodthat an associationobserved in a non-experimentalstudy is causal.These guidelines
areoftenusedto weigh epidemiologic evidencein deciding whether or not to regulate exposures to chemical
hazards.We argue that excessiverelianceon his guidelines is inappropriatefor many decisionsbearing on the regulationof environmentalexposures.We consideran alternative,explicitlyBayesian approachto regulatorydecision-making,and we discussthe advantages and pitfallsof this approach.We concludethat a majoradvantageof the Bayesianapproachis thatit helpsmakeexplicitthe non-scientificsocialand politicaldeterminantsof public healthdecisions. BRADFORD
HILL S GUIDELINES:
STRENGTHS
AND WEAKNESSES
In regulatorydecision-making,substantialweight is given to evidence obtainedfromepidemiologicstudiesof humanpopulations.Unfortunately, statisticalassociationsobserved between exposuresand diseasesin such studiesmay be either causalor non-causal.An observedassociationmay reflectcausality,or it may be due to the confoundingeffectsof unmeasured risk factors, to various forms of bias, or to samplingvariability. Conversely,failureto find an associationmay reflecteither true lack of causality,or may resultfrom confounding,bias or chance.In an attempt to systematizeassessmentof theseissues,SirAustinBradfordHill promulgatedguidelinesfor determiningthelikelihoodthatan associationobserved in an epidemiologicstudy is causalin nature. In BradfordHill's guidelines,the observedstrengthof an association between exposureand diseaseis the cardinalcriterionof causality.When 322
ROBINS
ET AL. * ENVIRONMENTAL
HEALTH
REGULATIONS
323
is measuredas the relaassessingcausality,the strengthof the association tive risk(theratioof the rateof diseasein an exposedgroupto thatin an unexposedgroup).If a relativeriskof 10 or greateris foundin an epiriskfactorsor demiologicstudy,it is difficultto imaginethatunmeasured unconsidered biasescouldaccountforsuchanextremeassociation. In such a case,scientificconsensus on causalitywill almostcertainlybe reached.On theotherhand,if a relativeriskis foundto be only 1.5,it is easyto suppose that uncontrolledbiasesor unmeasuredrisk factorscould explainthe observedassociation, andexpertscanbe expectedto disagreein theirinterof a such result. Thislattersituationwould pertaineven if the pretation studypopulationwere of sufficientsize to producea narrowconfidence intervalfor therelativerisk[e.g., (1.2, 1.4)]andan extremep-value(e.g., p
