Choosing a Future for Epidemiology:II. From Black Box to Chinese Boxesand Eco-Epidemiology

Mervyn Susser, MB, Bch, FRCP(E), DPH, and Ezra Susser, MD, DrPH

Whether the provocative factors are inherentin the population through their genes, theircells and their parasites; whether they arefailures to meet environmental influences ofsocial, physical or biological nature theyconstitute the concem of the epidemiolo-gist... Consequently, epidemiology mustconstantly seek imaginative and ingeniousteachers and scholars to create a new genre ofmedical ecologists who ... can interpret theinterplay offorces which result in disease.

Thomas Francis'9

IntroductionIn this paper on the choices before

epidemiology, we advocate a paradigm foran emergent era of eco-epidemiology. Toconnote the inclusion of systems at differ-ent levels, we term the paradigm Chineseboxes. This paradigm stems from a particu-lar distinction between the "universalism"of the physical sciences and the "ecolo-gism" of the biological sciences. It placesepidemiology on the track of ecologism, aperspective we aim to explain and justifybelow.

The practical implication of a localiz-ing ecological paradigm for the design ofepidemiological research is that an exclu-sive focus on risk factors at the individuallevel within populations-even given thelargest numbers-will not serve. We need

..- to be equally concerned with causalpathways at the societal level and withpathogenesis and causality at the molecu-lar level. Here we note that investigationsat all these levels are found in the historyof medicine and epidemiology since earlytimes. Hippocrates was concerned withthe effects of broad environmental condi-tions on health.' Later Galen, who empha-sized the individual host in the form of thefour humors, did not neglect the interac-tion of susceptibility with lifestyle. Paracel-sus, in the 16th century, aimed to graspmultiple levels.2 He tried to apply chemis-

try to medicine, and he also studied theinfluence of the stars on physiology.

The Needfor a New ParadigmThe necessity and the potential of a

new paradigm can be illustrated for theinfectious disease of human immunodefi-ciency virus (HIV) and the chronic dis-ease of peptic ulcer. While these twodiseases were selected to represent infec-tious and chronic diseases of our time,each of them also shows a blurring of thedistinction between infectious and chronicdisease. This is itself a hallmark of thenew era.

To understand and contain the glo-bal epidemic of HIV requires causalthinking at several levels of analysis. Atthe molecular level, the precision ofmolecular biology is required to deter-mine the means and the timing of trans-mission and to find a way to interrupt it.At an intermediate level, specific socialbehavior of individuals fosters sexual andother forms of transmission of the virus.At the population level, the dynamics ofthe epidemic are governed by the preva-lence of the infection itself as well as byother characteristics of the population-

Mervyn Susser is the Editor of the Journal.Ezra Susser is with the School of Public Health,Columbia University, and the New York StatePsychiatric Institute, New York, NY.

Requests for reprints should be sent toMervyn Susser, MB, Bch, FRCP(E), DPH,Columbia University, Gertrude H. SergievskyCenter, 630W 168th St, 19th Floor, New York,NY 10032.

This paper was accepted January 17,1996.

Editor's Note. Dr George Silver was theeditor responsible for this paper. As is ourpractice, Dr Mervyn Susser had no part in thereview and decision process.

See related editorial by Winkelstein(p 621) and comment by Koopman (p 630) inthis issue.

May 1996, Vol. 86, No. 5

Epidemiology's Future, II.

for instance, patterns of sexual relation-ships and of breast-feeding, prevalence ofother sexually transmitted diseases, andnutritional factors such as maternal vita-min A levels. At the global level, theinterconnections between societies deter-mine the path of the infection. Asinvestigators, we are naturally constrainedby our capabilities and by the necessaryreductionism entailed in firmly establish-ing the connection between one thing andanother and, more especially, their causalrelationships. Yet the best hopes forcontaining the epidemic rest upon acoherent strategy that can address allthese levels.

Peptic ulcer similarly illustrates thelimitations of a narrow frame of referencefor a chronic disease.3 The causal frame-work of the gastrophysiologist is likely tofocus on the wall of the stomach; that ofthe neurophysiologist, on the autonomicnervous system. The psychosomaticistexpands the framework to include inter-nal and environmental stressors, the hu-man geneticist considers familiality inblood groups and secretor status, and themicrobiologist brings the recent discover-ies about Helicobacterpylon to bear. Theepidemiologist includes all the above andadds smoking as an individual risk factor.

But the mystery and the challenge ofpeptic ulcer for epidemiology lies at theecological level of major secular change.We still have to unravel the factors thatcaused the peptic ulcer complex first towax and then to wane. This condition (orcomplex of conditions) mysteriouslyreached a peak in the 1950s and then, noless mysteriously, began to decline. Thiswas a cohort phenomenon that began itsrise in cohorts born before the turn of the19th century, with a steady decline incohorts born thereafter.3 A fully adequatecausal model for public health mustexplain the disease at the ecological levelas well as at lesser and more refined levelsof organization. This remains so even ifthe best explanation turns out to be thehistorical behavior of Helicobacter micro-organisms.

Universalism vs EcologismThe road is now open for epidemiolo-

gists to work at the same time at themolecular and the societal levels. To doso, we need to be guided by appropriatecausal concepts, a matter already underdiscussion in epidemiology.410

Like all the sciences, epidemiologyseeks generalizing concepts to explain the

causes of things. In the history of science,however, one can trace not one but twoconceptual tracks. The well-describeduniversalism of the physical sciences mustbe complemented by the often unacknowl-edged ecologism of the biological sci-ences. In contrast with universalism, ecolo-gism entails localization and attention tothe bounds that limit generalizationsabout biological, human, and social sys-tems.

A concept of causality based onuniversal laws is pervasive in the sciences.Most philosophers of science have con-fined the enterprise aLmost entirely to theuniversalist framework, although, ofcourse, exceptions exist.'1 We believe thatepidemiologists among others have beenmisled by standard interpretations of thenature of science.

The search for universal laws of thematerial world must deal with a paradox.The smaller the interacting microcosmicelements that such laws explain, the morelikely those elements are to be universal.Universality implies a view of space andtime expanding outward across the bound-aries and horizons of our world andothers, unimpeded by the local accretionsand characteristics of intervening struc-tures such as planets, continents, or ourbiological world, including people.

Some laws may hold across ourplanet for species and the evolutionaryprocesses that produced them. But abovethe level of molecules, no biological entitycan conform entirely to universal lawsbecause of the overarching contexts andthe interactions between levels within abiological structure. And the banal fact isthat each society is influenced by itseconomic, political, and cultural circum-stances as well as by its mix of peoples,climate, and topography. What is mostuniversal is least biological and, most ofall, least human.

It follows that universalism is notuniversally applicable to the scientificendeavor. Thus, when we enter thephysical, biological, and social realms ofthe human world, we need a parallel set ofideas interwoven with the search forgenerality. In epidemiology, the poor fit ofuniversalism with human reality is betterreplaced by a contrasting construct ofecologism. Ecological constructs try to

deal with the true complexity of thebiological world. Such constructs must invarying degree be localized; they must bebounded if they are to encompass all ofthe biological world's less-than-universallevels and their particular interactions.

Chinese Boxes:A Paradigmfor Eco-Epidemiology

In proposing a paradigm in the veinof ecologism, we draw on and develop an

earlier formulation of agent and hostesconced in an environment that com-

prises systems at multiple levels.4 Ourconcept envisages interactive systems. Asystem is a set or assembly of factorsconnected with each other in some formof coherent relationship. Thus, a system isan abstraction that allows a set of relatedfactors to be described in terms of a

coherent structure or coherent function.We speak properly of anatomical (struc-tural) systems and physiological (func-tional) systems-circulatory, nervous, or

reproductive. The human body is in itselfa system that encompasses all these.Societies comprise much more complexsystems of persisting and ordered rela-tions. The universe is a system of vastscale, a molecule one of minuscule scale.

Each system can be described in itsown terms. Each defines the limits of a

particular level of organization and thestructure within those limits. Hence, a setof factors that make up a system can beidentified. Their coherence implies a

degree of persistence and stability. Thisstability coexists, however, with the capac-

ity for change. Because the factors con-

tained in a system relate in some fashion,change and activity in one sector impingeson and affects other sectors.

Systems also relate to one another;they do not exist in isolation. A metaphormay serve to illuminate this ecologicalperspective. We liken it to Chineseboxes-a conjurer's nest of boxes, eachcontaining a succession of smaller ones.

Thus, within localized structures, we

envisage successive levels of organization,each of which encompasses the next andsimpler level, all with intimate linksbetween them.

Within each level, a relativelybounded structure such as a nation or

society or community may be character-ized by lawful relations that are localizedto that structure and can be discovered.At any given level within the hierarchy ofscale and complexity, these lawful rela-tions are generalizable, but only to theextent that they hold for other similar

structures, whether they are societies,cities, local communities, or individuals.

The paradigm represented by the

metaphor of Chinese boxes could be

suited to a new eco-epidemiology (Table1). This paradigm treats relations within

American Journal of Public Health 675May 1996, Vol. 86, No. 5

Susser and Susser

TABLE 1-Eras In the Evolution of Modern Epidemiology and an Emergent Era

Era Paradigm Analytic Approach Preventive Approach

Sanitary statistics (first halfof 19th century)

Infectious disease (late 19thcentury through first halfof 20th century)

Chronic disease epidemiol-ogy (latter half of 20thcentury)

Eco-epidemiology (emerg-ing)

Miasma: poisoning by foul ema-nations from soil, air, and water

Germ theory: single agents relateone to one to specific diseases

Black box: exposure related tooutcome, without necessityfor intervening factors orpathogenesis

Chinese boxes: relations withinand between localized struc-tures organized in a hierarchyof levels

Demonstrate clustering ofmorbidity and mortality

Laboratory isolation and cult-ure from disease sites, experi-mental transmission andreproduction of lesions

Risk ratio of exposure to out-come at individual level inpopulations

Analysis of determinants andoutcomes at different levelsof organization: within andacross contexts (using newinformation systems) and indepth (using new biomedi-cal techniques)

Introduce drainage, sewage,sanitation

Interrupt transmission (vaccines,isolation of the affected throughquarantine and fever hospitals,and ultimately antibiotics)

Control risk factors by modifyinglifestyle (diet, exercise, etc),agent (guns, food, etc), orenvironment (pollution,passive smoking, etc)

Apply both information and bio-medical technology to findleverage at efficacious levels,from contextual to molecular

and between localized structures that arebounded socially, biologically, or topo-graphically. The appropriate epidemiologi-cal approach is to analyze determinantsand outcomes at different levels of organi-zation. Such contextual analysis woulddraw on new information systems bothwithin and across levels to achieve breadth.It would draw on new biomedical tech-niques to achieve depth. The action thatfollows would need to find leverage at themost efficacious levels, whether contex-tual or molecular or both together.

The metaphor of Chinese boxes isperhaps not apt in every dimension, inthat levels exist in a hierarchy not only ofscale but also of complexity, with multipleinteractions between and within lev-els.14'15 The outer box might be theoverarching physical environment which,in turn, contains societies and populations(the epidemiological terrain), single indi-viduals, and individual physiological sys-tems, tissues and cells, and finally (inbiology) molecules.

To study even ecological systems indepth, we still have to use the basicmethodological procedures of science andlimit the fields of observation. Epidemiol-ogy can never aspire to the reductionismthat Freeman Dyson defines as the "effortto reduce the world of physical phenom-ena to a finite set of fundamental equa-tions."112 Steven Weinberg calls this "grandreductionism," which entails a view ofnature. However, epidemiologists must ofnecessity live with and use what Weinbergcalls "petty reductionism," which entailsonly a research procedure or program.13

But these approaches should not beallowed to obscure the contextual struc-ture of enveloping systems.14"15 To dealwith such a hierarchy of envelopingsystems, the need for a new paradigm ispatent.

Choosing the FutureAlthough we hear stirrings, we have

yet to adopt, develop, and apply this typeof paradigm in epidemiology. What wepresent here is no more than a skeletalframework. As this embryonic paradigmis tested in the field, no doubt itssimplifications and inadequacies willemerge, and some of its deficiencies willbe repaired.

The paradigm is bound to evolve andchange as the constraints of existingthought are broken, and one can expectit to confer new power on epidemi-ology. Such a paradigm will require a slewof sophisticated methods-borrowed,adapted, and created anew-that enableepidemiologists to test models at levelsfrom the molecular to the social.

At this time, the task will seemdaunting, even hopeless, to many of us.Few epidemiologists are equipped toundertake it. At the beginning of thiscentury, however, Ronald Ross pioneeredan analogous approach.'6 In 1902 he wonthe Nobel prize for establishing (in the1890s by painstaking microscopy) thatmosquitoes transmit malaria. He thereaf-ter took an epidemiological approach toeradicating the disease. Epidemiologyand a mathematical bent led him to

multivariate modeling to determine whatthe efficacy might be of interventions ofdifferent sorts.

We draw further on historical prece-dent to justify optimism. A study of theliterature of the early Chronic Diseaseera17 confirms direct experience of theelementary design and analytic tools inuse at the opening of the era. Designprinciples were only just taking form, andmultivariate analysis was almost inacces-sible. The contrast with the powerfuldesigns and sophisticated analyses of thelatter years of the era could scarcely begreater. Many such precedents give usreason to believe that the requisite ana-lytic tools are within reach, provided thatthe attention of epidemiologists is focusedon their development and use.

Here one must recognize that amolecular paradigm taken on its own ishugely attractive because of its explana-tory power. Without conscious countervail-ing effort, that paradigm will very likelycome to dominate epidemiology no lessthan did the germ theory in its time. Inthat event, with the sacrifice of conceptualand analytic breadth, epidemiology couldagain be reduced to a derivative pursuit oflaboratory science, and the mainstream ofour subject could be lost to creativescience. A countervailing force, which atthe same time restores public health toepidemiology, resides in a developedversion of the Chinese boxes paradigm.

One must also take heed of anotheremergent paradigm. Information systemscombined with systems analyses mightwell lead into a systems paradigm, with its

676 American Journal of Public Health May 1996, Vol. 86, No. 5

Epidemiology's Future, II.

own attractions for mathematically mindedepidemiologists. Standing alone, this para-digm would sacrifice biological depth andthe direct address to health disorders. Toavoid constriction, both the emergentthemes of biology and information as wellas the black box of our era need to besubsumed into a broader paradigtm suchas the Chinese boxes proposed here foreco-epidemiology.

A cogent scientific paradigm alone isnot enough to anchor epidemiologists topublic health, however. So what more isneeded to accomplish the linkage, onemay ask, beyond simple evangelism for anepidemiology inviolably tied to the publichealth?

SocializationAt the least, a practical program

must be devised to ensure that, in thecourse of their education, epidemiologistsare socialized in a manner that keeps alivethe idea of improving the public health asa primary value. Epidemiologists must bescientific but also in some degree profes-sional in the sense traditional to medicine,the law, and the clergy. That is, societyaccords them a privileged and autono-mous function founded on special train-ing. That autonomy carries reciprocal andprimary ethical obligations for service toindividuals or society.

To maintain such an ethic, we shallhave to choose and act accordingly. Thepower of the socialization process toimbue values is well documented in thework in medical education pioneered byRobert Merton and his colleagues18 andin much that followed.

In this respect, epidemiology andpublic health face ambiguities of role andstatus. As emphasized above, the publichealth function has been to serve popula-tions and, informed by notions of socialequity, to prevent and control disease inthose populations. Yet the historic originsof epidemiology are predominantly if notexclusively in medicine. And for millen-nia, the medical function, enshrined inethics and teaching, has been primarily toserve sick individuals.

In this century, epidemiology andpublic health have often withered in amedical environment that almost inevita-bly must give primacy to the individualcare of sick persons who solicit care. It

follows that autonomous schools of publichealth among others can have a crucialrole in socialization.

The diversification of public healthprofessions has resulted in further roleambiguities. In addition to the doctorsand sanitarians who were its originalmainstay, the public health corps nowcomprises statisticians, economists, socialscientists, professional administrators, or-ganization and other specialists and epide-miologists without medical training. Thisdiversification has centrifugal force. Toimbue these diverse groups with thevalues of public health, schools of publichealth will have to give due weight to theprocess of socializing their students tocommon values.

Socialization of students to publichealth will require conscious inductionthrough learning about its traditions andits history. They will need exposure tofaculty and others who understand andembody public health values. They willneed learning experiences in communitysituations as vivid and telling as thoseprovided for medical students by clini-cians at the bedside. They will need tocomprehend the hurt and waste of de-prived or disordered communities. Theywill need to recognize the true scale of theeffects that a few percentage points in acogent indicator can have on a nation'shealth.

Without intense socialization andlearning, we may well find-because ofthe natural momentum and narrow focusthat specialization generates-that thelinks between the values of public healthand its specialized disciplines dissolve aswe watch. In this respect, epidemiology ismost certainly at risk. O

AcknowledgmentsA first version of this paper and its companionserved as the keynote (given by M. S.) for thePan-American Epidemiology Congress, Salva-dor, Brazil, April 28, 1995. Thanks are due toMauricio Barreto and his organizing commit-tee for posing the challenge. The currentversion was the basis for the Thomas FrancisMemorial Lecture (given by M. S.) at theUniversity of Michigan, Ann Arbor, February23, 1996.

For their careful reading and commentson earlier versions, we also owe thanks toRodney Ehrlich, Elizabeth Fee, Nancy Krieger,Bruce Link, John McKinlay, Gerald Oppenhei-mer, Nigel Paneth, Charles Poole, and ZenaStein.

References1. Hippocrates; Chadwick J, Mann WN,

trans. Epidemics, BooksI& IHI: The MedicalWorks of Hippocrates. Oxford, England:Blackwell Scientific Publications; 1950:29-80.

2. King LS. The Growth of Medical Thought.Chicago, Ill: University of Chicago Press;1963.

3. Susser M, Stein Z. Civilization and pepticulcer.Lancet. 1962;1:115-119.

4. Susser M. Causal Thinking in the HealthSciences: Concepts and Strategies ofEpidemi-ology. New York, NY: Oxford UniversityPress; 1973.

5. Smith A. The epidemiological basis ofcommunity medicine. In: Smith A, ed.Recent Advances in Community Medicine.Vol. 3. Edinburgh, Scotland: Churchill-Livingston; 1985:1-10.

6. Krieger N. Epidemiology and the web ofcausation: has anyone seen the spider? SocSci Med. 1994;39(7):887-903.

7. Wing S. Limits of epidemiology. PhysiciansSoc Respons Q. 1994; no. 1:74-86.

8. Evans RG, Barer ML, Marmor TR, eds.Why Are Some People Healthy and OthersNot? The Detenninants ofHealth ofPopula-tions. New York, NY; Aldine de Gruyter;1994.

9. Pearce N. Traditional epidemiology, mod-em epidemiology, and public health.Am JPublic Health. 1996;86:678-683.

10. Link BG, Phelan JC. Editorial: understand-ing sociodemographic differences inhealth-the role of fundamental socialcauses. Am J Public Health. 1996;86:471-473.

11. Mayr E. The Growth ofBiological Thought:Diversity, Evolution, and Inhentance. Cam-bridge, Mass: Belknap Press of HarvardUniversity Press; 1982.

12. Dyson F. The scientist as rebel. NY RevBooks. May 25, 1995;42:31-33.

13. Weinberg S. Reductionism redux. NY RevBooks. October 5, 1995;42:39-42.

14. Susser M. The logic in ecological: I. thelogic of analysis. Am J Public Health.1994;84:825-829.

15. Susser M. The logic in ecological: II. thelogic ofdesign.Am JPublic Health. 1994;84:830-835.

16. Ross Sir R. The Prevention ofMalaria. 2nded. New York, NY: EP Dutton; 1910.

17. Susser M. Epidemiology in the UnitedStates after World War II: the evolution oftechnique. Epidemiol Rev. 1985;7:147-177.Reprinted in: Susser M, ed. Epidemiology,Health, and Society: Selected Papers. NewYork, NY: Oxford University Press; 1987:22-49.

18. Merton RK, Reader GC, Kendall PL. TheStudent Physician: Introductory Studies onthe Sociology of Medical Education. Cam-bridge, Mass: Harvard University Press;1957.

19. Francis T. Quoted by: Wegman M. Tho-mas Francis, Jr.: an appreciation. ArchEnviron Health. 1970;21:230-233.

May 1996, Vol. 86, No. 5 American Journal of Public Health 677