Cancer risk perceptions in an urban Mediterranean population

Montse Garc�ıa1,2*, Esteve Fern�andez1,3, Josep Maria Borras1,3, F. Javier Nieto

4, Anna Schiaffino

1,

Merce Peris1, Gloria P�erez5 and Carlo La Vecchia6 for the Cornella Health Interview SurveyFollow-Up (CHIS.FU) Study Group1Cancer Prevention and Control Unit, Institut d’ Investigaci�o Biom�edica de Bellvitge (IDIBELL), Catalan Institute of Oncology,L’Hospitalet de Llobregat, Spain2Department of Methodology, University of Barcelona, Barcelona, Spain3Department of Public Health, University of Barcelona, Barcelona, Spain4Department of Population Health Sciences, University of Wisconsin, Madison, WI, USA5Health Information Service, Agency of Public Health, Barcelona, Spain6Laboratory of Epidemiology, ‘‘Mario Negri’’ Institute, Milan, Italy

The objective of our study was to analyze the perceived (belief) oradopted (behavior) measures to reduce cancer risk in a Spanishpopulation. We used cross-sectional data from the Cornella HealthInterview Survey Follow-up Study (CHIS.FU). We analyzed 1,438subjects who in 2002 answered questions about risk perceptionson cancer and related behavior (668 males and 770 females). Thebenefits of avoiding cigarette smoking (95.8%), sunlight exposure(94.9%) and alcohol (81.0%) were widely recognized. On the otherhand, electromagnetic fields (92.1%), food coloring and other foodadditives (78.4%) or pesticides (69.4%), whose role in canceroccurrence, if any, remain unproven, were clearly considered ascancer risk factors in this population. Compared to men, womenmore frequently reported healthy behaviors, and the role of exoge-nous factors (i.e., environmental risk factors) were widely popular.There was a socioeconomic gradient on cancer risk perceptionwith respect to several lifestyle or dietary factors. Individuals withhigher educational level scored lower in several risk factors thanthose with primary or less than primary school education. Smokersreported adopting fewer healthy behaviors than former or neversmokers. How people perceive health issues and risk or make choicesabout their own behavior does not always follow a predictable orrational pattern.' 2005 Wiley-Liss, Inc.

Key words: risk perception; cancer prevention; health education

Individual cancer risk may be influenced by exogenous factors(i.e., environmental risk factors), by genetic factors (genetic sus-ceptibility) or by the interplay between them.1 It has been statedthat more than 50% of cancer could be prevented if our currentknowledge of risk factors were successfully implemented toreduce risk factor prevalence.2 Most cancers in a population arealso attributable to potentially modifiable environmental risk fac-tors.3,4 How these risks are perceived by the individuals could berelated to the actual behavior, although evidence for such a rela-tionship is weak.5 Cognitive psychology and neuroscience theoriesindicate there are 2 fundamental ways in which human beingscomprehend risk: the ‘‘analytic system’’ uses algorithms and nor-mative rules such as formal logic and risk assessment, while the‘‘experiential system’’ is intuitive, mostly automatic and not veryaccessible to conscious awareness. Both systems operate in paral-lel and mutually depend on the other for guidance and, hence,rational decision making requires proper integration of both modesof thought.6

Research on public’s perception of cancer can help to improverisk communication and health promotion strategies from the pub-lic health system. Thus, studies on cancer risk perception havefocused in areas related to the value of screening procedures,7–9 inpeople who either have the disease or have a relative affected bythe disease10,11 or in ad hoc population samples.12 There is scarceinformation in Spain about the knowledge or concerns of the gen-eral public on the relative importance of various cancer risk fac-tors.13 Thus, the aim of this study was to analyze the perceived(belief) or adopted (behavior) measures to reduce cancer risk in aSpanish population.

Material and methods

We used cross-sectional data from the Cornella Health Inter-view Survey Follow-Up (CHIS.FU) Study. The CHIS.FU study isa population-based cohort focusing on lifestyles and their conse-quences in health status. The cohort was set up with 2,500 subjects(1,263 women and 1,237 men) randomly selected from the generalpopulation of the city of Cornella de Llobregat, located on theMetropolitan area of Barcelona, in Catalonia, Spain (http://www.cornellaweb.com). Cornella de Llobregat is an industrialtown of approximately 85,000 inhabitants, mainly working- andmiddle class, with an important migrant population (during the1960s and 1970s) from other Spanish regions (mainly from thesouth). Subjects were initially interviewed in person in 1994.9,10

In 2002, we attempted to contact again and interview by telephonethe cohort members. A detailed description of the subject recruit-ment and procedures is provided elsewhere.15,16 Briefly, weobtained a 64.3% response in the total of the cohort; thus, at fol-low-up we gathered information from 1,608 subjects. Of theremaining 35.7%, 147 individuals were deceased, 425 had emi-grated, 123 refused the interview and 197 could not be located.Subjects aged <15 years (n 5 93) and those with disabilities didnot respond to the cancer risk assessment (n 5 77) and hence weanalyzed 1,438 subjects who answered in 2002 the questions aboutrisk perceptions on cancer from the direct follow-up questionnaire(668 males and 770 females).

The questionnaire included a general section on perceived riskof cancer. Information was specifically collected on perception ofrisk (‘‘Do you consider that . . . can prevent cancer?’’) and relatedbehavior (‘‘How do you behave in relation to . . . ?’’) for 9 majorrecognized or potential risk factors for cancer, as investigated in aprevious European survey.13,17 For each factor, 3 replies wereincluded in the questionnaire for belief (‘‘yes,’’ ‘‘no,’’ ‘‘I do notknow’’) and 2 for behavior’s adoption (‘‘yes,’’ ‘‘no’’). To evaluatepotential obsequiousness bias,18 we also included a tricky item on

Grant sponsor: Fondo de Investigaci�on Sanitaria; Grant number: PI02/0261; Grant sponsor: Cornella de Llobregat City Council; Grant sponsor:Instituto de Salud Carlos III, Network for Research in Epidemiology andPublic Health (RCESP); Grant number: C03/09; Grant sponsor: Institutode Salud Carlos III, Network for Research in Cancer (RTICC); Grant num-ber: C03/10.The Cornella Health Interview Survey Follow-up (CHIS.FU) Study

Group is composed of Esteve Fern�andez (principal investigator), AnnaSchiaffino and Montse Garcia (study coordinators) and Merce Mart�ı,Esteve Salt�o, Gloria P�erez, Merce Peris, Jorge Twose, Carme Borrell,F. Javier Nieto and Josep Maria Borras (associate researchers).*Correspondence to: Cancer Prevention and Control Unit, Catalan

Institute of Oncology, Gran Via s/n km 2.7, 08907 L’Hospitalet de Llobre-gat, Spain. Fax:134 93 260 79 56. E-mail: mgarciam@iconcologia.netReceived 21 October 2004; Accepted after revision 5 January 2005DOI 10.1002/ijc.21091Published online 4 May 2005 in Wiley InterScience (www.interscience.

wiley.com).

Int. J. Cancer: 117, 132–136 (2005)' 2005 Wiley-Liss, Inc.

Publication of the International Union Against Cancer

ultraviolet radiation (UV) exposure (‘‘Do you consider thatincreasing UV exposure can prevent cancer?’’: ‘‘yes/no’’).

Because of changes in the composition of the cohort since 1994due to attrition during follow-up,19 the sample of respondents in2002 overrepresented middle-aged men and women of the incep-tion cohort. Consequently, comparison of baseline and follow-updata was carried out after sex- and age-standardization by thedirect method, using the 2001 Cornella Census as the referent pop-ulation. Specific analyses according to sex, educational level andsmoking behavior were also performed.

Results

Table I gives the percent population perceiving (belief) oradopting (behavior) selected measures to reduce cancer risk,ranked according to belief. Avoiding smoking ranked 1st accord-ing to belief (95.8%) but was only the 5th most reported behavior(70.6%). Overall, 94.9% of the respondents (ranked 2nd accordingto belief) considered that limiting exposure to sunlight was protec-tive to prevent cancer but just 77.9% of them (ranked 2nd accord-ing to behavior) reported adopting such behavior.

The proportion of individuals who considered that avoiding expo-sure to electromagnetic fields (92.1%) and avoiding genetically modi-fied food (82.3%) could prevent cancer was higher than those whodeclared that reducing alcohol drinking was a protective factor forcancer (81.0%). A smaller proportion believed in the preventivepotential of selected nutritional and dietary factors, such as avoidingoverweight (72.6%) or avoiding excess calorie intake (72.6%), rank-ing them 7th and 8th, respectively. However, the latter ranked 3rdand 4th according to behavior adopted by the participants. Most par-ticipants (78.4%) believed that limiting consumption of food coloringand other food additives was protective for neoplasms, which rankedhigher than avoiding pesticide-treated fruit and vegetables (69.4%).

Table II describes the cancer risk beliefs and behavior accordingto sex. Males and females ranked perceived risks in the same orderand the rank order of behaviors varied only slightly by gender.

However, more women than men believed on the protective role ofavoiding electromagnetic fields, food coloring and other food addi-tives or pesticides; women also more frequently reported certainhealthy behaviors (reducing alcohol drinking, limiting sunshineexposure, avoiding smoking and avoiding overweight) than men.

Table III shows the public’s cancer risk perception and relatedbehaviors according to educational level. A socioeconomic gra-dient on cancer risk perception is evident for several lifestyle ordietary factors. The proportion of individuals with college or uni-versity studies who considered that avoiding genetically modifiedfood (77.7%) and limiting consumption of food coloring and otheradditives (72.3%) could prevent cancer was lower than amongsubjects with less than primary studies (91.3% and 85.7%, respec-tively). Individuals with higher educational level declared less fre-quently that avoiding overweight (70% vs. 84.8%) and avoidingexcessive calorie intake (63.7% vs. 85.0%) could be effective inpreventing cancer than subjects with less than primary studies. Asimilar pattern was apparent in both sexes (data not shown).

Table IV shows the proportion of population perceiving oradopting selected measures to reduce cancer risk according totheir smoking status. A lower proportion of smokers declaredadopting potentially preventive behaviors than former and neversmokers. Nevertheless, 95.7% of current smokers believed thatavoiding smoking can prevent cancer. A lower proportion ofsmokers declared to avoiding exposure to sunlight (69.7%) thanformer (82.2%) and never smokers (81.0%). Reducing alcoholdrinking was considered beneficial to prevent cancer by 77.4% ofthe smokers, in comparison with former (81.0%) and never smok-ers (83.0%). Moreover, fewer smokers considered the importanceof avoiding overweight (73.6%) or avoiding excessive calorieintake (68.6%) to prevent cancer than former and never smokers.

Discussion

While avoiding smoking was reported as the major protectivefactor for cancer, it was not the most frequently adopted behavior.

TABLE I – CANCER RISK PERCEPTION AND BEHAVIOR CONCERNING LIFE-STYLE AND DIETARY FACTORSAMONG 668 MEN AND 770 WOMEN, CORNELLA DE LLOBREGAT, SPAIN, 20021

Belief Behavior

Rank %2 Rank %2

Avoiding smoking 1 95.8 5 70.6Limiting exposure to sunshine 2 94.9 2 77.9Avoiding exposure to electromagnetic fields 3 92.1 7 64.0Avoiding genetically modified food 4 82.3 6 65.4Reducing alcohol drinking 5 81.0 1 95.4Limiting consumption of food coloring and other additives 6 78.4 8 61.4Avoiding overweight 7 76.9 3 74.8Avoiding excessive calorie intake 8 72.6 4 71.4Avoiding pesticide-treated fruit and vegetables 9 69.4 9 35.5Increasing exposure to UV 10 7.9 — —

1Age and sex standardizations.–2Percentage answering yes to the belief or behavior.

TABLE II – CANCER RISK PERCEPTION AND BEHAVIOR CONCERNING LIFE-STYLE AND DIETARY FACTORS ACCORDING TO SEX1

Males n5 668 Females n5 770

Belief Behavior Belief Behavior

Rank %2 Rank %2 Rank %2 Rank %2

Avoiding smoking 1 94.5 4 65.4 1 97.1 5 75.9Limiting exposure to sunshine 2 93.9 3 71.0 2 95.9 2 84.6Avoiding exposure to electromagnetic fields 3 89.1 7 57.9 3 95.0 7 69.8Avoiding genetically modified food 4 78.5 6 60.4 4 85.8 8 69.7Reducing alcohol drinking 5 76.6 1 92.0 5 85.2 1 98.6Limiting consumption of food coloring and other additives 6 73.7 8 52.0 6 82.7 6 69.9Avoiding overweight 7 71.2 2 71.5 7 82.3 4 78.2Avoiding excessive caloric intake 8 68.3 5 62.6 8 76.9 3 79.8Avoiding pesticide-treated fruit and vegetables 9 66.3 9 31.6 9 72.8 9 39.2Increasing exposure to UV 10 9.1 — — 10 6.5 — —

1Age standardization.–2Percentage answering yes to the belief or behavior.

133CANCER RISK PERCEPTIONS IN SPAIN

How people perceive health issues and risk and how they makechoices about their own behavior do not always follow a rationalpattern.20 However, we have to take into account that people maymake choices that appear irrational based on strict statisticalassessment. If a community has reason to distrust the public healthand medical community, or government authorities who announcerisk levels, members of the public may choose to ignore thoseannouncements and may perceive risks in an apparently ‘‘irra-tional manner.’’ This risk perception, however, may be quiterational given the political history and social context. In essence,people do not know who or what to trust so they default to anapparently ‘‘irrational’’ high level of risk perception. The current‘‘irrational’’ perception of risk associated with genetically modi-fied foods, pesticides, food colorings and electromagnetic wavesmay be affected by mistrust of official pronouncements regardingtheir risk because of past experiences with changing official riskevaluations. In Spain, we have good examples such as the ToxicOil Syndrome,21 the Prestige disaster22 and the accident at the oilrefinery that have increased concern in the population about healtheffects of chemical substances.23

According to most theories regarding the adoption of healthprotective behaviors,24–26 perceptions of susceptibility to illnessare necessary prerequisites. Thus, perceived cancer risk motivatesthe acceptance of screening for breast, cervical, and colorectalcancer.27–29 In relation to skin cancer, public campaigns to alerton the harmful effects of high and prolonged exposure to sunlighthave been conducted in Spain during the last few years.

While reducing alcohol consumption was considered a protectivefactor by 81.0% (ranking in the 5th place) of the participants, ahigher percentage was obtained when considered the behavioradopted (95.1%). Similar results regarding alcohol consumption but

with a higher belief of the harm it causes have been described inother Mediterranean countries.13 This is likely related to other shortterm effects of alcohol drinking such as accidents and violence.

We detected several apparent misconceptions about cancer riskbased on current scientific knowledge and standard statistical riskassessment. The importance of avoiding electromagnetic fields(whose role in cancer occurrence, if any, remains unproven),30,31 waswidely reported (92.1%). The role of food coloring and other addi-tives or pesticides were also perceived as risk factors by the largemajority of the individuals. Perhaps respondents are expressing theirperception of hypothetical risk vs. actual risk. Moreover, mediareports about health risk associated with electromagnetic fields ordietary additives may have contributed to raising fears in the public.

It appears that the characteristics of risk factors that trigger morealarm are the following: involuntary exposure, inescapable damageperceived as dreadful, as well as the availability bias (events areperceived to be more frequent if we can easily recall examples ofthem).32 Regarding risks associated with food, optimistic biases(people tend to believe they are less at risk from a given hazardcompared to someone else with similar demographic characteris-tics) are much greater for lifestyle hazards than for those associatedwith the technologies involved in food production.33

Whereas overweight was a well known risk factor for a majorityof the public (76.9%), it ranked below other risk factors such aselectromagnetic fields or food coloring and other additives. This isprobably because, even though most people are aware of theimportance of overweight as a cardiovascular risk factor, its rolein cancer is not as well known, despite its inclusion in the Euro-pean Code Against Cancer.34

There were differences in the percentage of beliefs and behav-iors according to sex, with females showing higher percentages

TABLE IV – CANCER RISK PERCEPTION AND BEHAVIOR CONCERNING LIFE-STYLE AND DIETARY FACTORS ACCORDING TO SMOKING STATUS1

Never smoked2

n 5 796Current smokers

n 5 379Former smokers

n5 262

Belief Behavior Belief Behavior Belief Behavior

Rank %3 Rank %3 Rank %3 Rank %3 Rank %3 Rank %3

Avoiding smoking 1 96.1 1 100 1 95.7 — — 1 95.2 1 100Limiting exposure to sunshine 2 95.7 3 81.0 2 94.2 3 69.7 2 93.8 3 82.2Avoiding exposure to electromagnetic fields 3 92.3 7 64.1 3 91.7 6 61.3 3 92.0 7 67.9Avoiding genetically modified food 4 83.5 6 66.0 5 76.8 5 61.9 4 88.6 8 69.5Reducing alcohol drinking 5 83.0 2 96.7 4 77.4 1 93.6 6 81.0 2 94.2Limiting consumption of food coloring and

other additives6 80.5 8 63.4 7 71.1 7 51.9 5 84.0 6 71.0

Avoiding overweight 7 77.5 4 76.0 6 73.6 2 72.1 7 80.2 4 75.9Avoiding excessive caloric intake 8 74.0 5 75.1 8 68.6 4 64.4 8 75.4 5 71.9Avoiding pesticide-treated fruit and vegetables 9 72.9 9 33.2 9 64.1 8 36.5 9 68.0 9 40.5Increasing exposure to UV 10 7.6 — — 10 9.3 — — 10 6.3 — —

1Age and sex direct standardization.–2We had one missing value at smoking status variable.–3Percentage answering yes to the belief orbehavior.

TABLE III – CANCER RISK PERCEPTION AND BEHAVIOUR CONCERNING LIFE-STYLE AND DIETARY FACTORS ACCORDING TO EDUCATIONAL LEVEL1

Less than primary2 Primary studies Secondary1 Universityn5 337 n 5 695 n5 400

Belief Behavior Belief Behavior Belief Behavior

Rank %3 Rank %3 Rank %3 Rank %3 Rank %3 Rank %3

Avoiding smoking 1 94.3 3 82.2 1 94.9 6 68.1 1 98.2 5 67.3Limiting exposure to sunshine 2 93.3 2 82.8 2 94.0 2 75.5 2 97.0 3 78.5Avoiding exposure to electromagnetic fields 3 93.2 5 75.4 3 92.2 7 67.4 3 91.3 7 54.0Avoiding genetically modified food 4 91.3 6 74.5 4 82.9 5 68.1 5 77.7 6 57.1Reducing alcohol drinking 7 84.9 1 96.4 5 80.8 1 95.1 4 79.1 1 95.3Limiting consumption of food colouring and other additives 5 85.7 4 76.8 6 80.3 8 63.5 6 72.3 8 50.2Avoiding overweight 8 84.8 8 72.4 7 79.1 3 73.2 8 70.0 2 78.8Avoiding excessive calorie intake 6 85.0 7 73.1 8 74.9 4 70.6 9 63.7 4 71.8Avoiding pesticide-treated fruit and vegetables 9 71.4 9 42.3 9 67.2 9 39.0 7 71.2 9 27.1Increasing exposure to UV 10 9.1 — — 10 7.2 — — 10 8.0 — —

1Age and sex direct standardisation.–2We had six missing values at educational level variable.–3Percentage answering yes to the belief orbehavior.

134 GARC�IA ET AL.

than males in all cases. Women might be more concerned aboutpotential risks due to the health care role they adopt to care theirfamilies.35 In our study, 30.4% of women were housewives andmost of them belonged to the low-middle socioeconomic class.Moreover, women between 50 and 64 years old are almostuniversally screened for breast cancer. The coverage in Cornellade Llobregat is almost completed.36 This periodical contact withthe Health Care Services may facilitate more awareness of otherpotential cancer risk factors.

We have observed a socioeconomic gradient in several lifestyleand dietary factors on cancer risk perception. Social class structureis an important determinant of population health status. In Spain,socioeconomic inequalities in mortality and morbidity have beenfound, with the less privileged classes consistently showing higherrates.37 Education influences health through its relation with higherincome and better living conditions, since well educated people areless likely to be unemployed and more likely to have jobs withhigher salaries.38 Furthermore, the well educated have certain psy-chological resources, such as a strong sense of personal control andsocial support, in addition to economic resources, that are associ-ated with higher health status.39 Moreover, living in a relativelydeprived area can have a detrimental effect as an individual level ofdeprivation has been taken into account. Area level of deprivationmay influence individual conduct directly through psychosocialmechanisms, determining convictions and attitudes, as well as limit-ing opportunities and resources for changing behavior.40

Smokers reported lower perceptions of risk about cancer andadopted less frequently health behaviors compared to former andnever smokers. Since cigarette smoking causes more preventabledeaths from cardiovascular disease and cancer than any othermodifiable risk factor, physicians and public health professionalsshould educate smokers about their personal health risks as part ofcomprehensive efforts to promote smoking cessation.41 Moreover,it is important that smoking education campaigns and materialsmust present clear and accurate quantitative information about thehealth risks of smoking.42

We observe a low degree of agreement between certain beliefsand behaviors. This might be partially explained by the fact thatsome factors are addictive (i.e., smoking and alcohol). This couldexplain why people continue smoking and/or drinking despiteknowing their harmful effects. Other factors, due to unknown expo-sures (i.e., electromagnetic fields or pesticide-treated fruits and veg-etables) cannot be avoided. Finally, for factors that are more easilymodifiable (i.e., exposure to sunlight or excessive calorie intake) wecan act and protect our health if we are better informed.

Our findings were consistent with a study conducted in Spain in1998 for the main risk factors (smoking, exposure to sunlight,alcohol).13 However, the risk perception of electromagnetic fieldsand genetically modified food is higher in our study, which couldstem from its recent impact in the media. Avoiding overweightand an excessive calorie intake were perceived more frequently asprotective factors in our study than in the previous one. It can beargued that the public in 2002 had more information than 5 yearsearlier, due to public debates regarding the Mediterranean diet andits changes in the preferences of people.43,44

We assessed the risk of cancer in general rather than specifictypes of cancer, so a certain degree of misclassification mayaccount for some of the findings. The study sample was largeenough to provide reliable estimates, but was not completelyrepresentative of the study base. Regarding reporting bias in theform of obsequiousness bias,18 so participants giving answers inthe direction they perceive are of interest to the researchers, mostof respondents (92.1%) correctly answered that UV exposure wasa risk for cancer. Since Cornella de Llobregat is an industrial cityin the metropolitan area of Barcelona, with a low-middle socioe-conomic status, these characteristics have to be considered whenjudging the external validity of these estimates.

As a limitation of this study, we do not know the perceivedlevel of risk implied by a ‘‘yes’’ response. Several factors canaffect the response on cancer risk perception using questionnaires,such as the use of single items or numerical scores, or the itemorder itself.45,46 The widespread popularity of some beliefs andbehaviors do not necessarily translate into ‘‘overestimates’’ of riskbecause the questions were categorical: ‘‘yes/no.’’ Thus, withoutknowing the perceived level of risk, it is not possible to state theseare overestimates. Some respondents may have correctly per-ceived that some risks were probably low but because they areunknown, they may still be present and hence a ‘‘yes’’ responsethat they should be avoided. In essence, it is not possible to trans-late a high frequency of yes response into a high magnitude of per-ceived risk.

In Spain, it would be a more efficient use of public healthresources to advocate broad changes in behavior to prevent can-cer47 rather than to address the disease outcomes.

Influential organizations should develop methods of risk com-munication that address these apparent ‘‘irrational’’ perceptionswith cogent explanations of actual statistical risk (and the limita-tions of current science). Moreover, the implications for usingrisk-communication approaches based on communicating scien-tific uncertainty should be widely discussed and analyzed.

In addition to this specific type of surveys, further research toadvance our knowledge of cancer risk perception is needed. Forinstance, investigating the best way to inform or communicate thelevel of cancer risk associated with particular exposures to thepublic,20 taking into account the context of health behaviors in ourcountry; investigating the characteristics associated with‘‘rational’’ risk perception according to current scientific knowl-edge of cancer causes; or investigating the profile of subjects withdissonant behaviors and beliefs on cancer risk.

Contributors

E.F., M.G. and A.S. conceived the CHIS.FU study. All theresearchers of the CHIS.FU Study Group designed the final studyprotocol. M.G. and A.S. coordinated the field work, created data-bases and checked all data. M.G. performed statistical analysis forthis article. M.G. and E.F. drafted the manuscript. C.L.V., J.M.B.,F.J.N., G.P. and M.P. gave expert advice and made written contri-butions to subsequent versions of the manuscript. All the authorsapproved the final version of the paper. E.F. is the guarantor of thestudy.

References

1. Doll R, Peto R. The causes of cancer: quantitative estimates, of avoid-able risk of cancer in the United States today. J Natl Cancer Inst1981;66:1193–265.

2. Colditz GA, Atwood KA, Emmons K, Monson RR, Willet WC, Tricho-poulus D, Hunter DJ. Harvard Report on cancer prevention, vol. 4.Harvard Cancer Risk Index. Cancer Causes Control 2000;11:477–88.

3. Baghurst KI, Baghurst PA, Record SJ. Public perceptions ofthe role of dietary and other environmental factors in cancer causation orprevention. J Epidemiol Community Health 1992;46: 120–6.

4. Peto R. Cancer epidemiology in the last century and the next decade.Nature 2001;411:390–5.

5. Leventhal H, Kelly K, Leventhal EA. Population risk, actual risk, per-ceived risk, and cancer control: a discussion. Natl Cancer Inst Monogr1999;25:81–5.

6. Slovic P, Finucane ML, Peters E, MacGregor DG. Risk as analysisand risk as feelings: some thoughts about affect, reason, risk, andrationality. Risk Anal 2004;24:311–22.

7. Izquierdo A, Viladiu P, Borras JM, Moreno V, Catalan G. Lapercepci�on de las causas del c�ancer y de su prevenci�on en lasmujeres de Girona. [Perception of the causes of cancer and its pre-vention in women of Girona]. Med Clin (Barc) 1996;107:14–19. [Spa]

8. Ostroff JS, Hay JL, Schantz SP, Maher MM. A survey of smoking sta-tus and cancer risk perceptions among participants attending a hospi-tal-based head and neck screening program. Psychol Health2000;14:979–90.

9. Vernon SW. Risk perception and risk communication for cancerscreening behaviours: a review. J Natl Cancer Inst Monogr1999;25:101–19.

135CANCER RISK PERCEPTIONS IN SPAIN

10. MacKillop WJ, Stewart WE, Ginsberg AD, Stewart SS. Cancerpatients’ perceptions of this disease and its treatment. Br J Cancer1988;58:355–8.

11. Braithwaite D, Emery J, Walter F, Prevost AT, Sutton S. Psychologi-cal impact of genetic counseling for familial cancer: a systematicreview and meta-analysis. J Natl Cancer Inst 2004;96:122–33.

12. Slovic P, Fischhoff B, Lichtenstein S. Rating the risks. In: GlickmanTS, Gough M, eds. Readings in risk. Washington, DC: Resources forthe Future, 1993. 61–75.

13. La Vecchia C, Anelli M, Zuccato E, Fanelli R, Sermeus G, MilazzoN. Perceived risk of cancer in population samples from 5 Europeancountries. Int J Cancer 2000;86:747–8.

14. Departament de Salut P�ublica. Ajuntament de Cornella. Enquesta deSalut. Cornella de Llobregat, 1993–1994. Cornella de Llobregat:Ajuntament de Cornella de Llobregat, 1995. p. 1–50.

15. Garcia M, Schiaffino A, Fern�andez E, Marti M, Salt�o E, Perez G,Peris M, Borrell C, Nieto FJ, Borras JM. The Cornella Health Inter-view Survey Follow-Up (CHIS.FU) Study: design, methods, andresponse rate. BMC Public Health 2003;3:12.

16. Garcia M, Fernandez E, Schiaffino A, Borrell C, Marti M, Borras JM.Determinants of attrition in a population-based cohort: The CornellaHealth Interview Survey Follow-up Study (CHIS.FU). Ann Epidemiol2005;15:98–104.

17. La Vecchia C, Anelli M, Zuccato E, Fanelli R, Fioretti F, Sermeus G,Valdicelli L, Milazzo N. Perceived risk of cancer in Italy. Eur J Can-cer Prev 1999;8:457–9.

18. Delgado M, Llorca J. Bias. J Epidemiol Community Health 2004;58:635–41.

19. Szklo M. Population-based cohort studies. Epidemiol Rev 1998;20:81–90.

20. Calman KC. Cancer: science and society and the communication ofrisk. BMJ 1996;313:799–802.

21. Doll R. The aetiology of the Spanish Toxic Syndrome: interpretationof the epidemiological evidence. Gac Sanit 2000;14(Suppl):72–88.

22. Porta M, Casal Lareo A, Castano-Vinyals G. El impacto en la saludhumana de la catastrofe del Prestige. [The impact of the Prestige dis-aster on human health] Gac Sanit 2004;18(Suppl 1):245–9. [Spa]

23. Ordonez-Iriarte JM, Gomez ME, Sanchez JI, Fernandez-Aguado C,Lopez R, Ferrer JB. El medio ambiente y su impacto en la salud: ries-gos tradicionales, nuevos riesgos. [The environment and its impacton health: old risks, new risks]. Gac Sanit 2004;18(Suppl 1):222–33.[Spa]

24. Ajzen I. Nature and operation of attitudes. Annu Rev Psychol. 2001;52:27–58.

25. Janz NK, Becker MH. The health belief model. A decade later. HealthEduc Q 1984;11:1–48.

26. Rogers RW, Mewborn CR. Fear appeals and attitude change: effect ofa threat’s noxiousness, probability and occurrence and the efficacy ofcoping responses. J Pers Soc Psychol 1976;34:54–61.

27. Katapodi MC, Lee KA, Facione NC, Dodd MJ. Predictors of per-ceived breast cancer risk and the relation between perceived risk andbreast cancer screening: a meta-analytic review. Prev Med 2004;38:388–402.

28. Eaker S, Adami HO, Sparen P. Attitudes to screening for cervical can-cer: a population-based study in Sweden. Cancer Causes Control2001;12:519–28.

29. Blalock SJ, DeVellis BM, Afifi RA, Sandler RS. Risk perceptions andparticipation in colorectal cancer screening. Health Psychol 1990;9:792–806.

30. IARC. Monographs on the evaluation of carcinogenic risks tohumans. Non-ionizing radiation, part 1: static and extremely low fre-quency (ELF) electric and magnetic fields. vol. 80. Lyon: IARC,2002.

31. Doll R. The Pierre Denoix memorial lecture. Nature and nurture inthe control of cancer. Eur J Cancer 1999;13:16–23.

32. Bennet P. Understanding responses to risk: some basic findings. In:Bennet P, Calman K, eds. Risk communication and public health.New York: Oxford University Press, 1999. 3–19.

33. Frewer LJ. Public risk perceptions and risk communication. In: Ben-nett P, Calman K, eds. Risk communication and public health. NewYork: Oxford University Press, 1999. 20–31.

34. Boyle P, Autier P, Bartelink H, Baselga J, Boffetta P, Burn J, BurnsHJ, Christensen L, Denis L, Dicato M, Diehl V, Doll R, et al. Euro-pean Code Against Cancer and scientific justification: third version(2003). Ann Oncol 2003;14:973–1005.

35. Fernandez E, Schiaffino A, Marti M. Influencia del trabajo dom�esticosobre la salud y la utilizaci�on de servicios sanitarios en mujeres contrabajo remunerado y amas de casa. [Influence of housework on healthand use of health services by employed women with paid work andhomemakers]. Gac Sanit 2000;14:287–90. [Spa]

36. Masuet C, Seculi E, Brugulat P, Tresserras R. La practica de lamamografia preventiva en Catalu~na. Un paso adelante. [The practiceof preventive mammography in Catalonia [Spain]: a step forward].Gac Sanit 2004;18:321–5. [Spa]

37. Borrell C, Pasarin MI. The study of social inequalities in Spain,where are we? J Epidemiol Community Health 1999;53:388–9.

38. Ross CE, Wu C. The link between education and health. Am SociolRev 1995;60:719–45.

39. Regicor E, Gutierrez-Fisac JL, Dominguez V, Calle ME, Navarro P.Comparing social inequalities in health in Spain 1987 and 1995/97.Soc Sci Med 2002;54:1323–32.

40. Shaw M, Dorling D, Davey Smith G. Poverty, social exclusion, andminorities. In: Marmort M, Wilkinson RD, eds. Social determinantsof health. Oxford: Oxford University Press, 1999. p. 211–239.

41. Ayanian JZ, Cleary PD. Perceived risks of heart disease and canceramong cigarette smokers. JAMA 1999;281:1019–21.

42. Sutton S. How ordinary people in Great Britain perceive the healthrisks of smoking. J Epidemiol Community Health 1998;52:338–9.

43. Gonzalez CA, Argilaga S, Agudo A, Amiano P, Barricarte A, Beguir-istain JM, Chirlaque MD, Dorronsoro M, Martinez C, Navarro C,Quiros JR, Rodriguez M, et al. Diferencias sociodemogr�aficas en laadhesi�on al patr�on de dieta mediterr�anea en poblaciones de Espa~na.[Sociodemographic differences in adherence to the Mediterraneandietary pattern in Spanish populations]. Gac Sanit 2002;16:214–21.[Spa]

44. Martinez-Gonzalez MA, Sanchez-Villegas A. The emerging role ofMediterranean diets in cardiovascular epidemiology: monounsatu-rated fats, olive oil, red wine or the whole pattern? Eur J Epidemiol2004;19:9–13.

45. Taylor KL, Shelby RA, Schwartz MD, Ackerman J, LaSalle H,Gelmann EP, McGuire C. The impact of item order on ratings ofcancer risk perception. Cancer Epidemiol Biomark Prev 2002;11:654–9.

46. Weinstein ND. What does it mean to understand a risk? Evaluatingrisk comprehension. J Natl Cancer Inst Monogr 1999;25:15–20.

47. Begg CB. The search for cancer risk factors: when can we stop look-ing? Am J Public Health 2001;91:360–4.

136 GARC�IA ET AL.