Choosing a Future for Epidemiology:I. Eras and Paradigms

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

IntroductionThe present era of epidemiology is

coming to a close. The focus on riskfactors at the individual level-the hall-mark of this era-will no longer serve. Weneed to be concerned equally with causalpathways at the societal level and withpathogenesis and causality at the molecu-lar level.'

This paper prepares the groundworkfor the argument that choices have to bemade about the future of epidemiology.To look forward, we do well to lookbackward for guidance. Part I of thisarticle sketches in brief outline the evolu-tion of modern epidemiology in threesuccessive eras. Following Kuhn,2 we setthe bounds of these eras in terms ofdominant paradigms.3 In Part II of thisarticle, we advocate a paradigm for a fourthemergent era of "eco-epidemiology."

The Evohltion ofModernEpidemiology

The underlying idea that marked thebeginnings of quantitative epidemiologyin the 17th century was concern for thepublic health and disparities in mortalityacross society. John Graunt the haber-dasher, in his book Natural and PoliticalObservations Made upon the Bills ofMortal-ity (1662),4 reported on the social distribu-tion of death in London and especially onthe mortal consequences of plague. In hisbook Political Arithmetick (1667),5 thephysician William Petty, Graunt's friendand sponsor in the Royal Society, was thefirst to provide a method to quantify thecosts of mortality.

The utilitarian approach that Graunt,Petty, and others adopted was entirely inaccord with the justifications prevailingover the beginnings of modern science inthe 15th and 16th centuries. Driven by the

twin forces of capitalism and the Protes-tant ethic, science was sanctioned (inRobert Merton's words)6 by "economicutility" and "the glorification of God."This ideology fostered discoveries withimmediate technical application in as-tronomy, navigation, firearms, optics, andmany other fields.

With the accelerating flow of discov-ery over the centuries, science as aninstitution abandoned its utilitarian rootsto become an end in itself. For some time,however, that was not true for epidemiol-ogy. That field retained a central concernwith the public health and its socialdistribution.

Thus, in the face of the miseries of19th-century England-the advance guardof industrialization and rapid urbaniza-tion-modern epidemiology graduallytook shape and then burst into activitywith the Sanitary Movement.7-9 Thereaf-ter, one can discern at least three eras inepidemiology, each with its own dominantparadigm: (1) the era of sanitary statisticswith its paradigm, miasma; (2) the era ofinfectious disease epidemiology with itsparadigm, the germ theory; and (3) the eraof chronic disease epidemiology with its

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 our

practice, 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, L.

TABLE 1-Three Eras in the Evolution of Modem Epidemiology

Era Paradigm Analytic Approach Preventive Approach

Sanitary statistics (first half of19th century)

Infectious disease epidemiology(ate 19th century through firsthalf of 20th century)

Chronic disease epidemiology(latter half of 20th century)

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 necessity forintervening factors or patho-genesis

Demonstrate clustering ofmorbidity and mortality

Laboratory isolation and cul-ture from disease sites, ex-perimental transmission,and reproduction of lesions

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

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.) oragent (guns, food, etc.) or envi-ronment (pollution, passivesmoking, etc.)

paradigm, the black box. Each of theseeras is described in historical contextbelow (Table 1).

Sanitary Statistics andMiasmaMiasma was the prevailing theory of

the Sanitarians for the greater part of the19th century. Sanitary statistics madeplain the toll of sickness and death in thecity slums of England, France, Germany,Scandinavia, and the United States (fore-runners of the squatter camps, favelas, orbarrios in today's less developed world).For the conditions in these slums, theSanitarian hypothesis of miasma im-pugned poisoning by foul emanationsfrom the soil, water, and environing air.The environmental causes were thoughtto have broad and multiple manifestationsin morbidity and mortality, and the sani-tary statistics that were collected asevidence were largely undifferentiated;that is, they were related more to overallmorbidity and mortality than to specificdiseases. Only in 1839 in England didWilliam Farr begin to use specific diagnos-tic classifications for national mortalitystatistics.10

Closed drainage and sewage systems,supplemented by garbage collection, pub-lic baths, and housing, were the remediesthat would disperse miasma, reduce mor-tality and morbidity (as indeed they did),and dispel the poverty of the new urbanpoor (as indeed they did not). A foremostproponent-and in some cases, the origi-nator-of these innovations was EdwinChadwick.11 Chadwick was a reformistwho argued that disease engendered bythe physical environment caused poverty.Friedrich Engels, his contemporary, was a

revolutionary who, in documenting the illsof Manchester factory workers, under-stood poverty to be the cause rather thanthe consequence of their ills.'2 But both

agreed that the issues were societal andthat the appropriate measures thus had tobe applied across society.

To emphasize the underlying publichealth and social values of the sanitarypioneers, it is worth noting that statisticsbegan literally as the study of the state andof the pertinent data. The newly formedLondon Statistical Society was chieflyconcerned with assembling that data.13Louis Rene Villerme in France andWilliam Farr in England,10"4 foundingfigures of epidemiology as we know ittoday, are only two among many whoworked to advance the public health inthis fashion.

Epidemiologists, largely autodidacts,were often medical heroes in this era.15,16Young physicians were excited by thechallenge of emergent patterns of diseasethat seemed rooted in a horrendousenvironment of urban misery. Beginningin 1858 John Simon, as chief medicalofficer of the national Board of Health inEngland, was able to draw around himover a few years a brilliant team-17 inall, no fewer than 8 of whom gainedelection to the Royal Society on thestrength of their work. These epidemiolo-gists mapped excess mortality across thecountry by district and in relation tohousing, infant care, and specific diseases;studied a wide range of industries andoccupations; detected many hazards fromdusts, heavy metals, and general workingconditions; and conducted national sur-veys of diet, parasite-infested meat, andfood contamination.

Unfortunately, these high points ofthe era closely preceded its culmination.Unmodified, the miasma paradigm couldnot survive advances in microbiology, andits demise brought an end to the SanitaryEra. The tenacity of some of the brilliantfigures of the movement, such as Edwin

Chadwick and Florence Nightingale, inresisting revision of their theory ratherthan subsume the new biology invitedridicule from medical scientists that hashardly yet been dispelled, and the broadperspective for which they stood graduallyfaded. The drama of the new microbiol-ogy was not to be gainsaid.

An irony of the history of publichealth is that, while the sanitarians weremistaken in their causal theory of foulemanations, they nonetheless demon-strated how and where to conduct thesearch for causes in terms of the cluster-ing of morbidity and mortality. Thereforms they helped to achieve in drain-age, sewage, water supplies, and sanita-tion generally brought major improve-ments in health. Their mistake lay in thespecifics of biology rather than in thebroad attribution of cause to environ-ment.'7Infectious Disease Epidemiologyand the Germ Theory

In 1840, Jakob Henle published atightly argued treatise that hypothesized(as a few beginning with Fracastorius andothers had done before him) that infec-tion by minute organisms was a majorcause of disease.1'" Despite John Snow'sfounding work of 1849 to 1854 in analyticepidemiology on the organismic cause ofcholera21 and his advances on Henle'sformulation,22 25 years passed beforeHenle was vindicated. Louis Pasteur'sdemonstration of a living organism as theagent in an epidemic afflicting silkwormsculminated in 1865.23 Studies of infectionand contagion in human disease-forinstance, tuberculosis, anthrax, and lep-rosy24-26-followed. Finally, in 1882, Hen-le's one-time student Robert Koch estab-lished a mycobacterium as "the cause" oftuberculosis.27 Henle, Snow, Pasteur, and

American Journal of Public Health 669May 1996, Vol. 86, No. 5

Susser and Susser

Koch can well stand as symbolic foundingfigures of the new era.

Snow and Koch faced directly themost acute public health problems of thetime. Although Henle had no means ofintervention at hand and Pasteur workedfirst on the commercial problems ofdiseases that threatened the silk industryand viticulture, they too declared andshared a public health perspective on theprevention of disease. Despite these ori-gins, the new paradigm of disease thatfollowed from their work, the germ theory,led in the end to the narrow laboratoryperspective of a specific cause modeP-namely, single agents relating one to oneto specific diseases.

The germ theory and its attendantview of specific cause dominated medicaland public health sciences from the lastquarter of the 19th century through atleast the mid-20th century. Single agentsof disease were sought by the isolationand culture of microorganisms from dis-ease sites, the experimental transmissionof these microorganisms, and the repro-duction of lesions. The appropriate re-sponses were to limit transmission byvaccines, to isolate those affected, and,ultimately, to cure with chemotherapyand antibiotics. Laboratory-based diagno-sis, immunization, and treatment gainedprecision with every new advance. Themiasma theory was relegated to the sameoblivion as phlogiston.

At the same time, the epidemiologyof populations and environmental expo-sures, and the social dynamics of diseasethat had flowed from the miasma theory,went into decline, replaced by a focus oncontrol of infectious agents. Epidemiol-ogy was often a derivative pursuit ratherthan a creative science in its own right.The new era scarcely maintained, letalone matched, the epidemiological ad-vances of the 19th century in the designand conduct of field surveys, the construc-tion of national statistical systems for vitaldata, and the statistical analyses of largenumbers.'7 The adherents of the tradi-tional philosophy of public health lostprestige and power in the medical hierar-chy and, indeed, were disparaged in waysthat in many places continue in thepresent.

The search for other than microbio-logical causes of disease in the environ-ment stumbled if it did not altogethercease. Thus, in the United States, JosephGoldberger, in his work on pellagra from1914 through the 1920s,29,30 ran againstthe tide of belief when he established.nutritional deficiency as a cause of infec-

tion. This was even more the case when, inthe rural South, he and Edgar Syden-stricker showed the dietary deficiency tobe consequent on the poverty ofsharecrop-pers and other workers trapped by theeconomic structure of the cotton fields.

In the same period, the search for aviral cause for the growing scourge ofpoliomyelitis was of course ultimatelyjustified. But the concentration of re-sources in the laboratory search for anorganism led to the neglect of key epide-miological findings and rendered futilethe preventive approaches attempted. Asearly as 1905, Ivar Wickman in Sweden3land, a decade later, Wade Hampton Frostin the United States32 had concluded fromepidemiological data that widespreadtransmission of silent infection by someunknown agent was the underlying factorin the summer epidemics that weredevastating the children of the better-offclasses in particular.

The irony of the Sanitary Era washere reversed. While, within their limitedframe of reference, the germ theoristswere accurate in their causal attributionsfor many diseases, their narrow focusretarded the creative use of bacterialdiscoveries to advance the science ofepidemiology. Some have argued that thedecline of infectious diseases in thedeveloped countries in the first part of thiscentury, at the height of the germ theoryparadigm, owed very little to science(including the use ofvaccines and antibiot-ics) and much to nutrition or improvedliving standards.33,M While closer analysisdoes not sustain the argument against therole of science,35 the primary role ofeconomic development and social changeis not in doubt.

Whatever the causes, the greatscourges of communicable disease didcome under control in the developedcountries. Once the major infectiousagents seemed all to have been identifiedand communicable disease no longeroverwhelmed all other mortal disorders,the force of the germ theory paradigmfaded. With notable exceptions such asRene Dubos,36 few anticipated the recru-descence ofcommunicable disease or newglobal epidemics. With the emerging pre-dominance of chronic disease of unknowncause, under any credible causal paradigmthe social and physical environment hadnow to be reckoned with once more.

Chronic Disease Epidemiologyand the Black Box

World War II serves as a convenientwatershed for the beginning of the Chronic

Disease Era and the black box paradigm.Shortly after the war ended in 1945, it wasclear that, in the developed world, risingchronic disease mortality had overtakenmortality from infectious disease. The risewas not owed to the aging of populationsalone. In middle-aged men specifically,the rises in peptic ulcer disease, coronaryheart disease, and lung cancer were ineach case fast and frightening enough toearn place and title as epidemics.37

By this time, also, chemotherapy andantibiotics had been added to the medicalarmamentarium. Their overwhelmingtherapeutic effects seemed to give tan-gible evidence that the major infectiousdiseases had been conquered. Only laterwas it discerned that these treatmentswere neither the only nor the primaryfactor in the steady decline of thesediseases in the first half of the 20thcentury.33

The prevailing epidemiology of ourday expressed the effort to understandand control the new chronic diseaseepidemics. Again the era was, at theoutset, driven by public health concepts.The problems selected for investigationwere the chronic diseases that most visiblythreatened the public health, and thegroups studied were those at manifestrisk-namely, middle-aged men.

Chronic disease epidemiology tookfirm hold with the first undeniable suc-cesses in this endeavor. British epidemiolo-gists Richard Doll, Austin Bradford Hill,Jeremy Morris, Thomas McKeown, andothers were key figures. The case-controland cohort studies on smoking and lungcancer, and the early cohort studies oncoronary heart disease that establishedserum cholesterol and smoking as riskfactors, demonstrated the power of theobservational method and established itscredentials.2

These studies carried the invisibleimprimatur of the black box paradigm("black box" being the general metaphorfor a self-contained unit whose innerprocesses are hidden from the viewer).This paradigm related exposure to out-come without any necessary obligation tointerpolate either intervening factors oreven pathogenesis (although not all ne-glected such interpolation). Epidemiolo-gists were faced once more, as in theSanitary Era, with major mortal diseasesof completely unknown origin. At theoutset, of necessity they resorted to

straightforward descriptive studies of dis-ease distribution and exploratory sweepsfor possible factors that enhanced risk.37As they moved on to test the emergent

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

Epidemiology's Future, I.

observations, these epidemiologists reliedupon ingenuity in design and the seizingof opportune circumstance to reach theirconclusions. They seldom resorted tocomplex statistical analysis.

The studies of lung cancer wereparticularly influential in giving the newparadigm credibility. Pathogenesis wasby-passed. Thus, the best biological sup-port to be found for the smoking-lungcancer relationship was quite indirect,residing in the demonstration by theKennaways and their colleagues that tarsapplied to the skin of mice were carcino-genic.38 Indeed, for another 4 decades, nodirect analogy with the epidemiologicalstudies of smoking existed in animalexperiments.

Step by step, the complexities ofchronic disease epidemiology emerged,first in matters of design and causalinference and, in parallel a little later, inmatters of statistical analysis.2 The incipi-ent thinking on design of previous de-cades was developed and systematized.39A40The structure of designs was clarified, thenecessity for statistical power and theadvantages of large numbers under-stood.41

Epidemiologists were obliged to de-part from the specific-cause model of thegerm theory. The metaphor of a "web ofcausation"40 characterized the multi-causal nature of public health problems,particularly those of chronic disease.After this beginning, one of us tried for hisown part to give systematic form to theproblems of inference that arose in thenascent epidemiology of a multivariateworld.17

Later, analytic issues and statisticalrefinement became a driving force. Thesharpening of technique led to a cycle ofcontinual refinement. Epidemiologists be-gan to explore in depth the subtleties ofconfounding, misclassification, survivor-ship, and other such issues. This labor isrepresented in the elegant and unifyingconcept of the fourfold table and thecase-control and cohort designs as alterna-tive methods of sampling the populationdisease experience to estimate risk ratiosor odds ratios.1741-44

The black box paradigm remains theprevailing model, and virtually all contem-porary epidemiologists including our-

selves haye made use of it. It can still yieldfindings of public health significance.Neural tube defects provide a recentexample. Typical black box approacheseventually led to the major discovery ofthe role of folate deficiency in neural tubedefects. Early studies found variations

with social class, geography and ethnicity,and economic cycles.45 Further studiesfound exposure to famine early in preg-nancy to be associated with an increasedrisk of congenital neural defects andprenatal vitamin intake to be associatedwith a decreased risk.4647 Finally, goingbeyond the black box, animal studiesfollowed by clinical trials of supplementa-tion established that periconceptionalfolic acid can prevent a large proportionof neural tube defects."4849

Momentumfor a New EraThe climax and, in all likelihood, the

culmination of the black box as dominantparadigm is already upon us. Two forces,characteristic of our time and muchwritten about, are blunting the black boxparadigm: (1) a transformation in globalhealth patterns and (2) new technology.

Health PattemsWith regard to health patterns, none

has had more impact than the humanimmunodeficiency virus (HIV) epidemic.Although epidemiology has made somenotable contributions to understandingthe epidemic, black box epidemiology is illequipped to address epidemic control.The causative organism as well as thecritical risk factors are known, so preven-tion is theoretically possible. Yet the HIVepidemic has demonstrated that bothdeveloping and developed countries re-main vulnerable to devastation by infec-tious disease.

Analysis of mass data at the indi-vidual level of organization alone, asimplied by the black box paradigm, doesnot allow us to weigh at which points inthe hierarchy of levels intervention islikely to be successful.50'51 No vaccine nowin prospect seems likely to achieve theefficacy level that could also achieveepidemic control. Absent such efficacy,the failure to control the disease resides inour lack of understanding of transmissionand illness in the social context. We knowwhich social behaviors need to change,but we know little about how to changethem, even when entire societies are at

stake.In retrospect, our confidence during

the Chronic Disease Era about the con-

trol of infectious diseases seems naive andalso blind to the less developed world. Forthe majority of the world's population,chronic infections-tuberculosis, syphilis,malaria, and many others-were never

under control. As with HIV infection, theimmediate causes and the risk factors

were known, but this knowledge could notbe translated into protection of the publichealth.

Similarly, our confidence in our abil-ity to control chronic noncommunicablediseases themselves by modifying behav-ior that carries risk has been shaken.Again, knowledge of risk factors andinterventions directed solely at changingthe behavior of individuals, even acrossseveral communities, have proven insuffi-cient.52'53

Health problems driven by societalproblems point to the location of theunderlying difficulties. The black boxparadign alone does not elucidate soci-etal forces or their relation to health. Thefocus on populations is generally directedat the individuals within them. Preventionat the societal level, conceptualized asintervening with individuals en masse, isoften nullified when the target is a socialentity with its own laws and dynamics.

TechnologyWith regard to technology, the devel-

opments that will drive research and thatcan lead epidemiology to a new paradigmreside primarily in biology and biomedicaltechniques on the one hand, and ininformation systems on the other. Theseadvances have begun to reshape all healthdisciplines.

Biological techniques such as geneticrecombination and imaging have trans-formed the ability of epidemiologists tocomprehend human disease at the mi-crolevel. For example, the methods ofrecombinant DNA have led to recogni-tion of both viral and genetic componentsin insulin-dependent diabetes54; to thedefinitive tracking from person to personof HIV, tuberculosis, and other infectionsthrough the molecular specificity of theorganisms55; to the discovery of a herpesvirus as almost certainly the agent inKaposi's sarcoma56; and to the drama ofthe familial tracking and marking of thefirst breast cancer gene.57 Imaging hasundermined the notion of schizophreniaas functional psychosis and given backingto the existence of environmental fac-tors.58 It has also allowed us to discover a

frequency of brain lesions in the prema-ture newborn that was unsuspectedly highoverall and concentrated in the earliesthours of life.59 Learning from the new

technology has only begun. Once unimag-inable possibilities follow from the map-ping of the human genome for specifyingthe role of heredity in disease, and no lessfrom the visualization of physiologica:

American Journal of Public Health 671May 1996, Vol. 86, No.5

Susser and Susser

processes for interpreting human func-tion.

The potential contribution of theseadvances to epidemiology is an exquisiterefinement of the definition and measure-ment of susceptibility, exposure, andoutcome. Such refinement also clarifiesthe intervening pathways and so eluci-dates with precision causal processes andnot merely causal factors. We can beconfident that new techniques, properlyapplied, can help dig epidemiology out ofthe slough of marginally significant riskestimates.60,61

In parallel, technology at the societallevel in the form of the global communica-tion network has opened new possibilitiesfor understanding and controlling disease.Information networks can provide instantaccess to-and enable the continuousassemblage of-existing stores of vitalstatistics and other relevant health andsocial data62 across the world. Such datahave myriad uses for newly empoweringpublic health. They promise a capacity fordevising and testing well-directed interven-tions at a societal level. Stores of data canbe mined to describe distributions acrosssocieties, to make comparisons of strataand groups nationally and internationally,to generate and test hypotheses, and toserve as sampling frames. Continuousaccumulation of data over time can servefor overall surveillance of health states,the detection of nascent epidemics andnew diseases, the response to disasters,and the evaluation of interventions. Thistechnology thus brings comprehension oflarge-scale phenomena and even systemswithin our grasp; it places at our com-mand the ability and the necessity torecognize broad dynamic patterns and,not least, disease in its social context.

ConclusionWhen research under the current

black box paradigm in its pure form relieson risk ratios that relate exposure tooutcome with no elaboration of interven-ing pathways, it forfeits the depth offeredby our new biological knowledge. Inaddition, because of an implicit andsometimes explicit commitment to analyz-ing disease solely at the individual level,research under this paradigm also dis-penses with the potential breadth offeredby new information systems in placingexposure, outcome, and risk in societalcontext.

The apogee of the black box para-digm is heralded by epidemiology texts ofthe 1980s.63,M4 These mark two trends.

They move away from the public healthorientation of the pioneers of the ChronicDisease Era. At the same time, analysisedges out design as the central focus. Atthe extreme we find an epidemiologyuntrammeled by the call to addressdisease in social groups, communities, andother formations of the social structure.Thus, a widely used modem text endorsesa pithy definition of epidemiology as thestudy of disease occurrence,64(Pl7) implic-itly setting aside public health ends.Epidemiology in this view is akin to thephysical sciences in sharing a search forthe highest levels of abstraction in univer-sal laws.65 Research in this universalistvein cannot take advantage of the extraor-dinary shifts and opportunities opened bynew dynamics of disease and new technol-ogy.

In the evolution of modem epidemi-ology, dominant paradigms have beendisplaced by new ones as health patternsand technologies have shifted. As hap-pened with previous paradigms, the blackbox, strained beyond its limits, is soonlikely to be subsumed if not supersededentirely by another paradigm. This para-digm reflects a particular era in ourdevelopment as a discipline. In our view,we stand at the verge of a new era. 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 toElizabeth Fee, Nancy Krieger, Bruce Link,John McKinlay, Gerald Oppenheimer, NigelPaneth, Charles Poole, and Zena Stein.

References1. Kuhn TS. The Stucture ofScientific Revolu-

tions. 2nd ed. Chicago, Ill: University ofChicago Press; 1970.

2. 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.

3. Susser M. Epidemiology today: "a thought-tormented world." IntJEpidemioL 1989;18:481-488.

4. Graunt J. Natural and Political ObservationsMade upon the Bills of Mortality. London,England: T. Roycraft; 1662. Reprinted:

Baltimore, Md: The John Hopkins Press;1939.

5. Petty W; Hull CH, ed. The EconomicWritings ofSir Wdliam Petty, togetherwith theObservations upon the Bills ofMortality MoreProbably by Captain John Graunt. 2 vols.Cambridge, England: Cambridge Univer-sity Press; 1899.

6. Merton RK The Sociology of Science:Theoretical and Empirical Investigations.Chicago, Ill: University of Chicago Press;1973.

7. Simon J. English Sanitary Institutions. 2nded. London, England: John Murray; 1887.

8. Rosen G. A History of Public Health.Reprinted with an introduction by Fee E.Baltimore, Md: The John Hopkins Univer-sity Press; 1993.

9. Fee E, Acheson RM, eds. A History ofEducation in Public Health. New York, NY:Oxford University Press; 1991.

10. Farr W; Humphreys NA, ed. Vital Statistics:A Memorial Volume of Selections from theReports and Writings of Wlliam Farr, 1885.Reprinted with an introduction by SusserM and Adelstein A. Metuchen, NJ: Scare-crow Press for the New York Academy ofMedicine; 1975.

11. Chadwick E. 1842 Report on the SanitaryCondition of the Labouring Population ofGreat Britain. Reprinted: Edinburgh, Scot-land: Edinburgh University Press; 1965.

12. Susser M. Ethical components in thedefinition of health. Int J Health Serv.1974;4:539-548. Reprinted in: Susser M,ed. Epidemiology, Health, and Society: Se-lected Papers. New York, NY: OxfordUniversity Press; 1987:186-193.

13. Cullen MF. The Statistical Movement inEarly Vctorian Britain. Sussex, England:Harvester Press; 1975.

14. Villerme LR. Rapport fait par M. Vil-lerme, et lu a l'Academie de medecine, aunom de la Commission de statistique, surune serie de tableau relatifs au mouvementde la population dans les douze arrondisse-ments municapaux de la ville de Parispendant les cinq annees 1817, 1818, 1819,1890, et 1821. Archives generales de mede-cine. 10:216-245.

15. Brockington CF. Public Health in theNineteenth Century. Edinburgh, Scotland: E& S Livingstone; 1965.

16. Lambert R. Sir John Simon (1816-1904)and English SocialAdministration. London,England: MacGibbon and Kee; 1963.

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

18. Rosen G. Social aspects of Jacob Henle'smedical thought. Bull Inst Hist Med. 1937;5:509-537.

19. Henle J; George Rosen, trans. On Mias-mata and Contagia. Baltimore, Md: TheJohns Hopkins Press; 1938.

20. Shryock RH. Germ theories in medicineprior to 1870: further comments on continu-ity in science. Clio Med. 1972;7:81-109.

21. Snow J. On the Mode of Communication ofCholera. 2nd ed, much enlarged. London,England: J. Churchill; 1855. Reprinted as:Snowon Cholera. New York, NY: Common-wealth Fund; 1936.

22. Winkelstein W Jr. A new prospective onJohn Snow's comTmunicable disease theory.AmJEpidemiol. 1995;142(suppl):S3-9.

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

Epidemiology's Future, I.

23. Vallery-Radot R; Devonshire RL, trans.The Life of Pasteur. London, England:Constable; 1901.

24. Villemin JA. Cause et nature de la tubercu-lose: son inoculation del l'homme au lapin.Compt Rend Acad Sci. 1865;61:1012-1015.Cited by Dubos R, Dubos J. The WhitePlague, Tuberculosis, Man and Society.Boston, Mass: Little Brown; 1952:98-99.

25. Hansen GA, Looft C, eds. Leprosy: In ItsClinical and Pathological Aspects. Bristol,England: John Wright & Co; 1895.

26. Irgens LM, Bjerkedal T. Epidemiology ofleprosy in Norway: the history of theNational Leprosy Registry of Norway from1856 until today. Int J Epidemiol. 1973;2:81-89.

27. Koch R. Die aetiologie der Tuberkulose.1882. Reprinted in: Schwalbe J, ed. Gesam-melte Werke von Robert Koch. Leipzig,Germany: Georg Thieme, Verlag; 1912;1:428-455.

28. Evans AS. Causation and disease: theHeule Koch Postulatis revisited. Yale JBiolMed. 1976;49: 175-195.

29. Goldberger J, Wheeler GA, SydenstrickerE. A study of the diet of nonpellagrous andpellagrous households in textile mill com-munities in South Carolina in 1916. JAMA.1918;71:944-949.

30. Terris M, ed. Goldberger on Pellagra.Baton Rouge, La: Louisiana State Univer-sity Press; 1964.

31. Wickman 1. Weitere Studine uber Poliomy-elimitis acute; ein Beitrag zur Kenntnis derNeurophagen und Kornchenzellen. DtscheZ. Nervenheik. 1909/ 10;38:396-437. Bei-trage zur kenntnis der Heine-MedinschenKrankheit (Poliomyelitits acuta und ver-wandter Erkrankungen). Berlin, Germany:Karger; 1907. Both cited in: Paul JR.History of Poliomyelitis. New Haven, Conn:Yale University Press; 1971; chap 10.

32. Frost WH. Poliomyelitis (infantile paraly-sis): what is known of its cause and modesof transmission. Public Health Rep. 1916;51:1817-1833.

33. McKeown T. The Modem Rise of Popula-tion. London, England: Edward Arnold;1976.

34. McKinlay JB. From "promising report" to"standard procedure": seven stages in thecareer of a medical innovation. MilbankMem Fund Q. 1981;59(3):374-411.

35. Susser M. Health as a human right: anepidemiologist's perspective on the publichealth. Am J Public Health. 1993;83:418-426.

36. Dubos R. The Mirage of Health: Utopias:Progress and Biological Change. New York,NY: Harper & Row; 1959.

37. Morris JN. Uses of Epidemiology. London,England: Churchill Livingstone; 1957.

38. Burrows H, Heiger I, Kennaway EL. Theexperimental production of tumors ofconnective tissue. Am J Cancer. 1932;16:57-67.

39. Witts L, ed. Medical Surveys and ClinicalTrials. London, England: Oxford Univer-sity Press; 1959:64-90.

40. MacMahon B, Pugh TF, Ipsen J. Epidemio-logical Methods. Boston, Mass: Little, Brown& Co, 1960.

41. Fleiss JL. Statistical Methods for Rates andProportions. 1st ed. New York, NY: JohnWiley & Sons; 1974.

42. Miettinen 0. Estimability and estimationin case-referent studies. Am J Epidemiol.1976;103:226-235.

43. Breslow NE, Day NE, eds. StatisticalMethods in Cancer Research. Vol 1. TheAnalysis of Case-Control Studies. Lyon,France: International Agency for Researchon Cancer; 1980.

44. Greenland S, Thomas DC, MorgenstemH. The rare-disease assumption revisited.A critique of "estimators of relative risk forcase-control studies." Am J Epidemiol.1986;124:869-876.

45. Elwood JM, Little J, Elwood JH. Epidemi-ology and Control of Neural Tube Defects.New York, NY: Oxford University Press;1992.

46. Stein Z, Susser M, Saenger G, Marolla F,eds. Famine and Human Development. TheDutch Hunger Winter of 1944-45. NewYork, NY: Oxford University Press; 1975.

47. Milunsky A, Jick H, Jick SS, et al.Multivitamin/folic acid supplementation inearly pregnancy reduces the prevalence ofneural tube defects.JAMA. 1989;262:2847-2852.

48. Smithells RW, Nevin NC, Seller MJ, et al.Further experience of vitamin supplemen-tation for prevention of neural tube defectrecurrences. Lancet. 1983;i:1027.

49. Medical Research Council Vitamin StudyResearch Group. Prevention of neuraltube defects: results of the medical re-search council vitamin study. Lancet. 1991;338:131-137.

50. Koopman JS, Longini IM, Jacquez JA, et

al. Assessing risk factors for transmission ofinfection. Am J Epidemiol. 1991;133:1199-1209.

51. Blower SM, McLean AR. Prophylacticvaccines, risk behavior change, and theprobability of eradicating HIV in SanFrancisco. Science. 1994;265:1451-1454.

52. COMMIT Research Group. CommunityIntervention Trial for Smoking Cessation(COMMIT): II. Changes in adult cigarettesmoking prevalence. Am J Public Health.1995;85:193-200.

53. Susser M. Editorial: the tribulations oftrials-interventions in communities. Am JPublic Health. 1995;85:156-158.

54. Solimena M, De Camilli P. Coxsackieviruses and diabetes. Nature Med. 1995; 1 (1):25-26.

55. Alland D, Kalkut GE, Moss AR, et al.Transmission of tuberculosis in New YorkCity: an analysis by DNA fingerprintingand conventional epidemiologic methods.NEnglJMed. 1994;330:1710-1716.

56. Chang Y, Cesarman E, Pessin MS, et al.Identification of herpesvirus-like DNAsequence in AIDS-associated Kaposi's sar-coma. Science. 1994;266:1865-1869.

57. Hall JM, Lee MK, Newman B, et al.Linkage of early onset familial breastcancer to chromosome 17Q21. Science.1990;250:1684-1689.

58. Andreasen NC, Arndt S, Swayze II , et al.Thalamic abnormalities in schizophreniavisualized through magnetic resonanceimage averaging. Science. 1994;266:294-298.

59. Paneth N, Pinto-Martin J, Gardiner J, etal. Incidence and timing of germinalmatrix/intraventricular hemorrhage in lowbirth weight infants. Am J Epidemiol.1993;137:1167-1175.

60. Stein Z, Hatch M. Biological markers inreproductive epidemiology: prospects andprecautions. Environ Health Perspect. 1987;74:67-75.

61. Rothman N, Stewart WF, Schulte PA.Incorporating biomarkers into cancer epi-demiology: a matrix of biomarker andstudy design categories. Cancer EpidemiolBiomarkers Prev. 1995;4:301-311.

62. Friede A, Reid JA, Ory HW. CDCWONDER: a comprehensive on-line pub-lic health information system of the Cen-ters for Disease Control and Prevention.Am J Public Health. 1993;83:1289-1294.

63. Kleinbaum DG, Kupper LL, MorgensternH, eds. Epidemiologic Research: Principlesand Quantitative Methods. Belmont, Calif:Lifetime Learning Publications, Wads-worth Inc; 1982.

64. Rothman KJ. Modem Epidemiology. Bos-ton, Mass: Little, Brown & Co; 1986.

65. Rothman KJ, ed. Causal Inference. Chest-nut Hill, Mass: Epidemiology ResourcesInc; 1988.

May 1996, Vol. 86, No. 5American Journal of Public Health 673