A Short Introduction To Epidemiology - Cornell University

Perhaps the most commonly quotedepidemiologic legend is that of Snow whostudied the causes of cholera in Londonin the mid-19th century (Winkelstein,1995). Snow was able to establish thatthe cholera death rate was much higherin areas supplied by the Southwark andVauxhall Company which took waterfrom the Thames downstream fromLondon (i.e. after it had beencontaminated with sewerage) than inareas supplied by the Lambeth Companywhich took water from upstream, withthe death rates being intermediate inareas served by both companies.Subsequently, Snow (1936) studied thearea supplied by both companies, andwithin this area walked the streets todetermine for each house in which acholera death had occurred, whichcompany supplied the water. The deathrate was almost ten times as high inhouses supplied with water containingsewerage (table 1.2).1983; Loomis and Wing, 1991; Samet,2000; Vandenbroucke, 1994), it is clearthat Snow was able to discover, andestablish convincing proof for, the modeof transmission of cholera, and to takepreventive action several decades beforethe biological basis of his observationswas understood. Thus, it was not untilseveral decades after the work of Snowthat Pasteur and others established therole of the transmission of specificpathogens in what became known as the“infectious diseases”, and it was anothercentury, in most instances, beforeeffective vaccines or antibiotictreatments became available.Nevertheless, a dramatic decline inmortality from these diseases occurredfrom the mid-nineteenth century longbefore the development of modernpharmaceuticals. This has beenattributed to improvements in nutrition,sanitation, and general living conditions(McKeown, 1979) although it has beenargued that specific public healthinterventions on factors such as urbancongestion actually played the major role(Szreter, 1988).Although epidemiologists and otherresearchers continue to battle overSnow’s legacy and its implications forepidemiology today (Cameron and Jones,1.2 Risk Factor EpidemiologyThis decline in the importance ofcommunicable disease was accompaniedby an increase in morbidity and mortalityfrom non-communicable diseases suchas heart disease, cancer, diabetes, andrespiratory disease. This led to majordevelopments in the theory and practiceof epidemiology, particularly in thesecond half of the 20th century. Therehas been a particular emphasis onaspects of individual lifestyle (diet,exercise, etc) and in the last decade thehuman genome project has seen anaccelerated interest in the role of geneticfactors (Beaty and Khoury, 2000).Thus, epidemiology became widelyrecognized with the establishment of thelink between tobacco smoking as a causeof lung cancer in the early 1950's (Dolland Hill, 1950; Wynder and Graham,1950), although this association hadalready been established in Germany inthe 1930s (Schairer and Schöninger,11

2001). Subsequent decades have seenmajor discoveries relating to othercauses of chronic disease such asasbestos, ionizing radiation, viruses,diet, outdoor air pollution, indoor airpollution, water pollution, and geneticfactors. These epidemiologic successeshave in some cases led to successfulpreventive interventions without theneed for major social or political change.For example, occupational carcinogenscan, with some difficulty, be controlledthrough regulatory measures, andexposures to known occupationalcarcinogens have been reduced inindustrialized countries in recentdecades. Another example is thesuccessful World Health Organisation(WHO) campaign against smallpox. Morerecently, some countries have passedlegislation to restrict advertising oftobacco and smoking in public placesand have adopted health promotionprogrammes aimed at changes in"lifestyle".for the ethical and practical constraints,epidemiologic theory and practice has,quite appropriately, been based on thetheory and practice of randomised trials.Thus, the aim of an epidemiologic studyinvestigating the effect of a specific riskfactor (e.g. smoking) on a particulardisease (e.g. lung cancer) is intended toobtain the same findings that would havebeen obtained from a randomisedcontrolled trial. Of course, anepidemiologic study will usuallyexperience more problems of bias than arandomised controlled trial, but therandomised trial is the “gold standard”.This approach has led to majordevelopments in epidemiologic theory(presented most elegantly andcomprehensively in Rothman andGreenland, 1998). In particular, therehave been major developments in thetheory of cohort studies (which mimic arandomised trial, but without therandomisation) and case-control studies(which attempt to obtain the samefindings as a full cohort study, but in amore efficient manner). It is these basicmethods, which follow a randomisedcontrolled trial “paradigm”, which receivemost of the attention in this shortintroductory text. However, whilepresenting these basic methods, it isimportant to also recognise theirlimitations, and to also consider differentor more complex methods that may bemore appropriate when epidemiology isused in the public health context.Individual lifestyle factors would ideallybe investigated using a randomisedcontrolled trial, but this is often unethicalor impractical (e.g. tobacco smoking).Thus, it is necessary to do observationalstudies and epidemiology has mademajor contributions to the understandingof the role of individual lifestyle factorsand health. Because such factors wouldideally be investigated in randomisedcontrolled trials, and in fact would beideally suited to such trials if it were not1.3 Epidemiology in the 21st CenturyIn particular, in the last decade therehas been increasing concern expressedabout the limitations of the risk factorapproach, and considerable debate aboutthe future direction of epidemiology(Saracci, 1999). In particular, it hasbeen argued that there has been anoveremphasis on aspects of individual12

lifestyle, and little attention paid to thepopulation-level determinants of health(Susser and Susser, 1996a, 1996b;Pearce, 1996; McMichael, 1999).Furthermore, the success of risk factorepidemiology has been more temporaryand more limited than might have beenexpected. For example, the limitedsuccess of legislative measures inindustrialised countries has led thetobacco industry to shift its promotionalactivities to developing countries so thatmore people are exposed to tobaccosmoke than ever before (Barry, 1991;Tominaga, 1986). Similar shifts haveoccurred for some occupationalcarcinogens (Pearce et al, 1994). Thus,on a global basis the "achievement" ofthe public health movement has oftenbeen to move public health problemsfrom rich countries to poor countries andfrom rich to poor populations within theindustrialized countries.studies. Even if one is focusing onindividual “lifestyle” risk factors, there isgood reason to conduct studies at thepopulation level (Rose, 1992). Moreover,every population has its own history,culture, and economic and socialdivisions which influence how and whypeople are exposed to specific riskfactors, and how they respond to suchexposures. For example, New Zealand(Aotearoa) was colonised by GreatBritain more than 150 years ago,resulting in major loss of life by theindigenous people (the Māori). It iscommonly assumed that this loss of lifeoccurred primarily due to the arrival ofinfectious diseases to which Māori had nonatural immunity. However, a morecareful analysis of the history ofcolonisation throughout the Pacificreveals that the indigenous peoplemainly suffered major mortality fromimported infectious diseases when theirland was taken (Kunitz, 1994), thusdisrupting their economic base, foodsupply and social networks. Thisexample is not merely of historicalinterest, since it these same infectiousdiseases that have returned in strengthin Eastern Europe in the last decade,after lying dormant for nearly a century(Bobak and Marmot, 1996). Similarly,the effects of occupational carcinogensmay be greater in developing countrieswhere workers may be relatively youngor may be affected by malnutrition orother diseases (Pearce et al, 1994).It should be acknowledged that not allepidemiologists share these concerns(e.g. Savitz, 1994; Rothman et al, 1998;Poole and Rothman, 1998), and somehave regarded these discussions as anattack on the field itself, rather than asan attempt to broaden its vision.Nevertheless, the debate has progressedand there is an increasing recognition ofthe importance of taking a more globalapproach to epidemiologic research andof the importance of maintaining anappropriate balance and interactionbetween macro-level (population),individual-level (e.g. lifestyle), andmicro-level (e.g. genetic) research(Pearce, 2004).These issues are likely to become moreimportant because, not only isepidemiology changing, but the worldthat epidemiologists study is also rapidlychanging. We are seeing the effects ofeconomic globalization, structuraladjustment (Pearce et al, 1994) andclimate change (McMichael, 1993, 1995),and the last few decades have seen theoccurrence of the “informationalrevolution” which is having effects asgreat as the previous agricultural andindustrial revolutions (Castells, 1996).There are three crucial concepts whichhave received increasing attention in thisregard.The Importance of ContextThe first, and most important issue, isthe need to consider the populationcontext when conducting epidemiologic13

In industrialized countries, this is likelyto prolong life expectancy for some,but not all, sections of the population.In developing countries, the benefitshave been even more mixed (Pearce etal, 1994), while the countries ofEastern Europe are experiencing thelargest sudden drop in life expectancythat has been observed in peacetimein recorded human history (Bobackand Marmot, 1996) with a major risein alcoholism and “forgotten” diseasessuch as tuberculosis and cholera.theories and identifies the major publichealth problems which new theoriesmust be able to explain. A fruitfulresearch process can then begenerated with positive interactionbetween epidemiologists and otherresearchers. Studying real publichealth problems in their historical andsocial context does not excludelearning about sophisticated methodsof study design and data analysis (infact, it necessitates it), but it may helpto ensure that the appropriatequestions are asked (Pearce, 1999).This increased interest in populationlevel determinants of health has beenparticularly marked by increasedinterest in techniques such asmultilevel modelling which allowindividual lifestyle risk factors to beconsidered “in context” and in parallelwith macro-level determinants ofhealth (Greenland, 2000). Such a shiftin approach is important, not onlybecause of the need to emphasize therole of diversity and local knowledge(Kunitz, 1994), but also because of themore general moves within science toconsider macro-level systems andprocesses (Cohen and Stewart, 1994)rather than taking a solely reductionistapproach (Pearce, 1996).Appropriate TechnologyA related issue is the need to use“appropriate technology” to addressthe most important public healthresearch questions. In particular, asattention moves “upstream” to thepopulation level (McKinlay, 1993) newmethods will need to be developed(McMichael, 1995). One example ofthis, noted above, is the recent rise ininterest in multilevel modelling(Blakely and Woodward, 2000; Pearce,2000), although it is important tostress that it is an increase in“multilevel thinking” in thedevelopment of epidemiologichypotheses and the design of studiesthat is required, rather than just theuse of new statistical techniques ofdata analysis. The appropriateness ofany research methodology depends onthe phenomenon under study: itsmagnitude, the setting, the currentstate of theory and knowledge, theavailability of valid measurement tools,and the proposed uses of theinformation to be gathered, as well asthe community resources and skillsavailable and the prevailing norms andvalues at the national, regional or locallevel (Pearce and McKinlay, 1998).Thus, there has been increasedinterest in the interface betweenepidemiology and social science(Krieger, 2000), and in theProblem-Based EpidemiologyA second issue is that a problem-basedapproach may be particularly valuablein encouraging epidemiologists tofocus on the major public healthproblems and to take the populationcontext into account (Pearce, 2001;Thacker and Buffington, 2001). Aproblem-based approach to teachingclinical medicine has been increasinglyadopted in medical schools around theworld. The value of this approach isthat theories and methods are taughtin the context of solving real-lifeproblems. Starting with “the problem”at the population level provides a“reality check” on existing etiological14

development of theoretical andmethodological frameworksappropriate for epidemiologic studiesin developing countries (Barreto et al,2001; Barreto, 2004; Loewenson,2004), and in indigenous people in“Western “ countries (Durie, 2004). Asnoted above, this short introductorytext focuses on the most basicepidemiologic methods, but I attemptto refer to more complex issues, andthe potential use of more complexmethods, where this is appropriate.SummaryPublic health is primarily concerned withthe prevention of disease in humanpopulations, and epidemiology is thebranch of public health which attemptsto discover the causes of disease inorder to make disease preventionpossible. It thus differs from clinicalmedicine both in its emphasis onprevention (rather than treatment) andin its focus on populations (rather thanindividual patients). Thus, theepidemiological approach to a particulardisease is intended to identify high-risksubgroups within the population, todetermine the causes of such excessrisks, and to determine the effectivenessof subsequent preventive measures.Although the epidemiological approachhas been used for more than a centuryfor the study of communicable diseases,epidemiology has considerably grown inscope and sophistication in the last fewdecades as it has been increasinglyapplied to the study of noncommunicable diseases. At the beginningof the 21st century, the field ofepidemiology is changing rapidly, notonly with regards to its basic methods,but also with regards to the hypotheseswhich these methods are used toinvestigate. In particular, in recent yearsthere has been a revival in public healthapplications of epidemiology, not only atthe national level, but also at theinternational level, as epidemiologiststackle global problems such as climatechange. This text does not attempt toreview the more complex methods usedto study such issues. However, it doesprovide a coherent and systematicsummary of the basic methods in thefield, which can be used as a logical basefor the teaching and development ofresearch into these more complexissues.15

ReferencesBarreto ML (2004). The globalization ofepidemiology: critical thoughts fromLatin America. Int J Epidemiol 33:1132-7.between science and indigenousknowledge. Int J Epidemiol 33: 113843.Greenland S (2000). Principles ofmultilevel modelling. Int J Epidemiol29: 158-67.Barreto ML, Almeida-Filho N, Breihl J(2001). Epidemiology is more thandiscourse: critical thoughts from LatinAmerica. J Epidemiol Comm Health55: 158-9.Hennekens CH, Buring JE (1987).Epidemiology in medicine. Boston:Little, Brown.Barry M (1991). The influence of theU.S. tobacco industry on the health,economy, and environment ofdeveloping countries. New Engl J Med324: 917-20.Hippocrates (1938). On airs, waters andplaces. Med Classics 3: 19.Krieger N (2000). Epidemiology andsocial sciences: towards a criticalreengagement in the 21st century.Epidemiologic Reviews 22: 155-63.Beaty TH, Khoury MJ (2000). Interface ofgenetics and epidemiology.Epidemiologic Reviews 22: 120-5.Kunitz S (1994). Disease and socialdiversity. New York: Oxford UniversityPress.Blakeley T, Woodward AJ (2000).Ecological effects in multi-levelstudies. J Epidemiol Comm Health 54:367-74.Last JM (ed) (1988). A dictionary ofepidemiology. New York: OxfordUniversity Press.Bobak M, Marmot M (1996). East-Westmortality divide and its potentialexplanations: proposed researchagenda. Br Med J 312: 421-5.Loewenson R (2004). Epidemiology inthe era of globalization: skills transferor new skills? Int J Epidemiol 33:1144-50.Cameron D, Jones IG (1983). JohnSnow, the Broad Street pump andmodern epidemiology. Int J Epidemiol12: 393-6.Loomis D, Wing S (1991). Is molecularepidemiology a germ theory for theend of the twentieth century? Int JEpidemiol 19: 1-3.Castells M (1996). The information age:Economy, society and culture. Vol 1.The rise of the network society.Oxford: Blackwell.McMichael AJ (1993). Planetaryoverload: global environmentalchange and the health of the humanspecies. Cambridge: CambridgeUniversity Press.Cohen J, Stewart I (1994). The collapseof chaos: discovering simplicity in acomplex world. London: Penguin.McMichael AJ (1995). The health ofpersons, populations, and planets:epidemiology comes full circle.Epidemiol 6: 633-5.Doll R, Hill AB (1950). Smoking andcarcinoma of the lung. Br Med J 2:739-48.McMichael AJ (1999). Prisoners of theproximate: loosening the constraintsDurie M (2004). Understanding healthand illness: research at the interface16

on epidemiology in an age of change.Am J Epidemiol 149: 887-97.epidemiology include the readicationof poverty? Lancet 352: 810-3.McKeown T (1979). The role of medicine.Princeton, NJ: Princeton UniversityPress.Samet JM (2000). Epidemiology andpolicy: the pump handle meets thenew millennium. EpidemiologicReviews 22: 145-54.McKinlay JB (1993). The promotion ofhealth through planned sociopoliticalchange: challenges for research andpolicy. Soc Sci Med 36: 109-17.Saracci R (1999). Epidemiology inprogress: thoughts, tensions andtargets. Int J Epidemiol 28: S997-9.Pearce N (1996). Traditionale

Risk factor epidemiology 11 - Epidemiology in the 21. st. century 12 . PART 1: STUDY DESIGN OPTIONS. 2. Incidence studies 21 - Incidence studies 22 - Incidence case-control studies 28 . 3. Prevalence studies 33 - Prevalence studies 33 - Prevalence case-control studies 38 . 4. More complex study designs 41 - Other axes of .