Preventing non-communicable diseasesthrough structural changes in urbanenvironmentsManuel Franco,1,2 Usama Bilal,1,2 Ana V Diez-Roux3

The primary determinants of disease aremainly economic and social, and there-fore its remedies must also be economicand social. Medicine and politics cannotand should not be kept apart. Rose1

To achieve [a reduction in overweightand obesity] is perhaps the major publichealth and societal challenge of thecentury. Potential strategies include [….]redesign of built environments topromote physical activity, changes infood systems, restrictions on aggressivepromotion of unhealthy drinks andfoods to children and economic strat-egies such as taxation. Willet2

Non-communicable diseases (NCDs)—mainly cancers, cardiovascular diseases(CVDs), diabetes and chronic respiratorydiseases—are the main causes of deathand morbidity worldwide.3 NCDs arenow annually responsible for more than35 million deaths in the world with morethan 80% of this disease burden occurringin low-income and middle-income coun-tries.4 At the same time, NCDs are highlypreventable by means of effective prevent-ive interventions tackling shared behav-ioural risk factors such as unhealthy diets,harmful use of alcohol, tobacco use andphysical inactivity.5

Efforts to prevent NCDs have historicallyincluded strategies to target high-risk indivi-duals, which have shown, especially in thecase of obesity and diabetes, poor results.6 7

To advance the prevention of NCDs,population-wide understanding of theseshared risk factors and morbidity remainscrucial. The population approach toprevent NCDs, articulated by Rose1 in hisarticle Sick individuals and sick popula-tions, aims at shifting the distribution of itsrisk factors for the whole population, there-fore affecting everyone regardless of their

risk. Rose highlighted the need to measureand understand factors related to interpo-pulation differences in the distribution ofrisk factors (social phenomena and socialdeterminants or environmental factors),instead of focusing on factors related tointerindividual differences within a popula-tion (classic behavioural risk factors andgenetics). The population strategy is radicalin the sense that it would affect the funda-mental causes of the distribution of riskfactors in the whole population of interestby promoting large structural, social andenvironmental changes.Population preventive strategies try to

shift the entire distribution of risk factors.Even small shifts in the full distributionmay have a larger health impact than strat-egies that focus on high-risk individualswithin a population.1 Small changes in riskfactor distribution at the population levelresulting from large political or economicchange8 and whole population campaigns9

have led to substantial health impacts.Analyses of the health consequences of a

tragic historical period in Cuba during thepast three decades8 have shown population-wide loss of 4–5 kg in weight in a relativelyhealthy population was accompanied by a50% reduction in diabetes mortality and a30% mortality reduction from coronaryheart disease. Furthermore, a rebound inbody weight was associated with an increaseddiabetes incidence and mortality, and ahalting in the decline in mortality from cor-onary heart disease8 (see figure 1 for popula-tion body weight changes and diabetesburden over three decades of the study).Population-wide body weight changes overtime occurred due to large economic andsocial changes directly related to the availabil-ity of food and fuel. Food was rationed andtransportation networks had limited activity,forcing the population to walk or cycle towork. This, along with the government pro-duction and importation of more than 1.5million cycles, led to a population-wide lossof body weight with the aforementionedconsequences in terms of NCDs.Another example of large structural

changes includes the North Karelia Project.People in this area of Finland presented thehighest rates of coronary heart disease in theworld during the 60s.10 The determinants

of these incidence rates involved, as Rose1

previously stated, factors acting as massinfluences on the entire population. Thequestion shifted from “Why did this individ-ual develop CVD?” to “Why do populationrates of CVD vary so much between EastFinland and other parts of the world?”.9

Based on this concept, the North KareliaProject (which included consultations fromGeoffrey Rose himself9) designed alarge-scale intervention that included part-nership with a previously reluctant foodindustry, subsidies for the production ofhealthier foods (produce) and large-builtenvironment changes.11 The results of thisproject were so encouraging that it wasexpanded to the entire country of Finland in5 years and led to large reductions in cardio-vascular mortality of around 80% from1970 to 2006.9

These two examples provide evidenceon the potential for prevention of NCDsof the population strategy. Nonetheless,and as pointed out by Frohlich,12 there is“a common misinterpretation of thepopulation approach, which considers itsimply to mean programmes or policieshaving an impact on a large number ofpeople.” Rose’s definition of populationapproach relies on ‘upstream’ factors ascontextual determinants or policy-leveldeterminants. The same line of thought isexpressed by Willet when referring to theCuba study by Franco et al,8 highlightingthe need for structural changes directlyrelated to the levels of physical activityand healthy eating of the population as awhole.2 As detailed by Rose13 in anearlier paper, individual strategies (such asmedication) adopted in a grand scale arenot part of the population approach,because “to influence mass behaviour wemust look to its mass determinants, whichare largely economic and social.”

Urban environments present uniqueopportunities for research and policy evalu-ation of population approaches to preven-tion. By definition cities are dense, andcharacterised by substantial man-madecomponents of their environments and byfrequent social interactions. These character-istics make cities excellent candidates forpolicy interventions on social and physicalfactors affecting large numbers of people. Inaddition, cities are internally heterogeneous,with large within city variation in social andphysical environments, which have beenshown to be associated with NCD.14 15

Cities also encompass multiple contexts rele-vant to health, such as the larger metropol-itan area, the city itself and neighbourhoodswithin the city.

Cross-city comparisons may also beinformative, for example, contrasting the

1Social and Cardiovascular Epidemiology ResearchGroup, School of Medicine, University of Alcala,Madrid, Spain; 2Department of Epidemiology, JohnsHopkins Bloomberg School of Public Health, Baltimore,Maryland, USA; 3Department of Epidemiology, DrexelUniversity School of Public Health, Philadelphia,Pennsylvania, USA

Correspondence to Dr Manuel Franco, Social andCardiovascular Epidemiology Research Group, School ofMedicine, University of Alcala, Madrid 28871, Spain;mfranco@uah.es

Editorial

Franco M, et al. J Epidemiol Community Health June 2015 Vol 69 No 6 509

distribution of NCDs risk factors in com-parable populations of two different cities(eg, Copenhagen and Madrid). Studyinghow the distribution of NCDs risk factorswithin cities may change in Madrid intwo very different moments in time, 2015and 2030, may also shed light to preventNCDs.

Studying population prevention strat-egies in urban areas presents the limitationthat social phenomena such as transporta-tion or food policies may come into forceabove urban areas at the city or thenational level. Nevertheless, other featuresof the environment as healthy food avail-ability and affordability, and walkability ofthe area do actually happen differentlyacross urban areas. Nonetheless, in orderto understand why rates of disease varywithin or across cities, characteristics ofthe environment (social or physical) mustbe measured and analysed.16

The development of population strat-egies in cities requires identification of thesocial and environmental drivers of behav-ioural patterns across and also withincities.

A number of observational studies haveexamined associations between social andphysical environments of neighbourhoodsand NCDs. An example of a recent neigh-bourhood and health study can be found

in figure 2, from the Heart HealthyHoods project in Madrid.17 Studying theupstream factors that affect NCDsrequires studying the socioeconomic com-position of neighbourhoods in close rela-tion to environmental domains ofneighbourhoods such as tobacco, physicalactivity, alcohol and food environments.These four domains of the urban environ-ment can be understood and measured interms of the social norms and physical

resources that make up these environ-ments (left side of figure 2). These foururban environment domains may veryimportantly have a direct relationshipwith the well-known and well-studiedindividual NCD risk factors, namelytobacco use, physical inactivity, harmfuluse of alcohol and unhealthy diets. Theeffect of social determinants measured atthe individual level (such as individualgender roles) should also be studied,

Figure 1 Body mass index (BMI) distributions in Cienfuegos, Cuba, 1990–2010, and diabetes burden in Cuba, 1980–2010.

Figure 2 Characteristics of the urban environment and individual behavioural risk factorsrelated to non-communicable diseases (NCDs).

Editorial

510 Franco M, et al. J Epidemiol Community Health June 2015 Vol 69 No 6

especially as an effect modifier of moreupstream factors (right side of figure 2).

Although much can still be learned fromobservational studies, strengthening causalinference will require other study designs. Itwill be very useful to capitalise on naturallyoccurring changes and quasi-experiments(whenever available).16 By natural experi-mental studies, we mean the methodo-logical approaches to evaluating the impacton health or other outcomes of interven-tions or policies, which are not under thecontrol of researchers but which are amen-able to the research.18 19

Measuring neighbourhood-level deter-minants of individual behaviours can helpus to answer the population-level questionof “How would rates of NCDs change inimpoverished Madrid if healthy food wasmore affordable?” or “How would rates ofNCDs change in Madrid if transportationpolicies were similar to those ofCopenhagen?”. Bicycling as an active formof transportation has been encouraged inCopenhagen by major municipal cam-paigns and investments in a cohesivebicycle infrastructure after large protests inthe 1970s and 1980s20 by Copenhagenresidents. Answering these types of ques-tions may require the use of natural experi-ments allowing researchers to study ifurban changes (not always health related)have had a sizeable effect on health.

In order to understand and developlarge-scale structural changes in our urbansettings the input from different disciplinessuch as epidemiology, sociology, geog-raphy, urban planning, primary care andhealth systems research, and public policywill be the key.21 In addition, developingpopulation preventive strategies requires adeep understanding of how societal pat-terns of disease are created by political,economical and cultural decisions.22

Differences across areas or neighbour-hoods are not ‘natural’ but rather resultfrom specific policies (or from the absenceof policies).16 22 Understanding the relation-ship of the social and physical environmentwith NCDs, and developing adequate andefficient preventive strategies will requirethe work of multiple disciplines, often withdiverse methodological approaches includ-ing large-scale quantitative observationalstudies and qualitative studies of the ways inwhich people relate to and are affected byurban environments. Interdisciplinary workpartnering with communities and policy

experts is warranted to prevent the majorpublic health challenge of NCDs that weface in our cities.

FUNDAMENTAL CONCEPTS AND TERMS1. NCDs: diseases of long duration and

slow progression that are not (directly)passed from person to person.Typically the main four groups includeare: CVDs, cancer, chronic respiratoryconditions and diabetes.4

2. Population strategy for NCDs preven-tion: strategy that seeks to control thedeterminants of incidence in the popula-tion as a whole through mass environ-mental interventions (large structuraland radical changes) aimed to shift theentire distribution of NCDs risk factors.1

3. Studying urban environmentsA. Social norms: social norms are

properties of societies that provideguidance for people’s attitudes andexert powerful influences over theirindividual health behaviours.23 Themeasurement of social norms relieson anthropological and sociologicalmeasurement techniques.

B. Physical resources: the materialresources available to peopleaccording to their status and loca-tion, which allow people to fullydevelop their health potential.24

The measurement of urban phys-ical resources relies on tools pro-vided by geography, sociology andeconomics.

Competing interests None.

Provenance and peer review Commissioned;internally peer reviewed.

To cite Franco M, Bilal U, Diez-Roux AV. J EpidemiolCommunity Health 2015;69:509–511.

Received 24 October 2014Accepted 25 October 2014Published Online First 13 November 2014

J Epidemiol Community Health 2015;69:509–511.doi:10.1136/jech-2014-203865

REFERENCES1 Rose G. Sick individuals and sick populations. Int J

Epidemiol 1985;14:32–8.2 Willett WC. Weight changes and health in Cuba.

BMJ 2013;346:f1777.3 Beaglehole R, Bonita R. Global public health:

a scorecard. Lancet 2008;372:1988–96.

4 World Health Organization. Global status report onnoncommunicable diseases 2010. Geneva,Switzerland: World Health Organization, 2011.

5 Ezzati M, Riboli E. Can noncommunicable diseasesbe prevented? Lessons from studies of populationsand individuals. Science 2012;337:1482–7.

6 LOOK AHEAD Research Group, Wing RR, Bolin P,et al. Cardiovascular effects of intensive lifestyleintervention in type 2 diabetes. N Engl J Med2013;369:145–54.

7 Simmons RK, Echouffo-Tcheugui JB, Sharp SJ, et al.Screening for type 2 diabetes and populationmortality over 10 years (ADDITION-Cambridge):a cluster-randomised controlled trial. Lancet2012;380:1741–8.

8 Franco M, Bilal U, Orduñez P, et al. Population-wideweight loss and regain in relation to diabetes burdenand cardiovascular mortality in Cuba 1980–2010:repeated cross sectional surveys and ecologicalcomparison of secular trends. BMJ 2013;346:f1515.

9 Oppenheimer GM, Blackburn H, Puska P. FromFramingham to North Karelia to U.S.community-based prevention programs: negotiatingresearch agenda for coronary heart disease in thesecond half of the 20th century. Public Health Rev2011;33:450–83.

10 Vartiainen E, Laatikainen T, Peltonen M, et al.Thirty-five-year trends in cardiovascular risk factors inFinland. Int J Epidemiol 2010;39:504–18.

11 McAlister A, Puska P, Salonen JT, et al. Theory andaction for health promotion illustrations from theNorth Karelia Project. Am J Public Health1982;72:43–50.

12 Frohlich KL. Commentary: what is a population-based intervention? Returning to Geoffrey Rose. Int JEpidemiol 2014;43:1292–3.

13 Rose G. Strategy of prevention: lessons fromcardiovascular disease. BMJ (Clin Res Ed)1981;282:1847–51.

14 Diez Roux AV. Residential environments andcardiovascular risk. J Urban Health 2003;80:569–89.

15 Diez Roux AV, Mair C. Neighborhoods and health.Ann N Y Acad Sci 2010;1186:125–45.

16 Diez Roux AV. Neighborhoods and health: where arewe and were do we go from here? Rev EpidemiolSante Publique 2007;55:13–21.

17 Heart Healthy Hoods Project: http://www.hhhproject.eu Starting Grant 2013 European Research Council.

18 Craig P, Cooper C, Gunnell D, et al. Using naturalexperiments to evaluate population healthinterventions: guidance for producers and users ofevidence. Medical Research Council, 2011.

19 Craig P, Cooper C, Gunnell D, et al. Using naturalexperiments to evaluate population healthinterventions: new Medical Research Councilguidance. J Epidemiol Community Health2012;66:1182–6.

20 Gehl J, Svarre B. How to study public life: IslandPress, 2013.

21 Kawachi I. Social epidemiology. Soc Sci Med2002;54:1739–41.

22 Krieger N. Proximal, distal, and the politics ofcausation: what’s level got to do with it? Am JPublic Health 2008;98:221–30.

23 Berkman LF, Glass T, Brissette I, et al. From socialintegration to health: Durkheim in the newmillennium. Soc Sci Med 2000;51:843–57.

24 Lynch J, Kaplan G. Socioeconomic position.Berkman LF, Kawachi I, eds. Social Epidemiology.Oxford: Oxford University Press, 2000:13–35.

Editorial

Franco M, et al. J Epidemiol Community Health June 2015 Vol 69 No 6 511