Lifestyle Changes in Women at Genetic Risk of BreastCancer: an Observational Study

Lorna McLeish & Marta M. Reis & Clare Stewart &David R. Goudie & Jonathan N. Berg & Michelle Harvie &

Kirstie A. Hanning & Helen Vysny & C. Michael Steel

# International Society of Behavioral Medicine 2012

AbstractBackground Lifestyle influences breast cancer risk. Womenat increased familial risk may benefit from modifying be-haviour, but it is not known to what extent they do so.Purpose This study aims to measure changes that UK(Scottish) women make in response to increased familialrisk of breast cancer and attitudes to a risk-reduction trial.Methods A questionnaire, completed by 140 “breast cancerfamily” clinic patients, generated data on habitual diet,alcohol consumption and exercise, changes made afterlearning of breast cancer risk and attitudes to possible fur-ther changes. Subgroups of patients were defined by criterialikely to influence changes in behaviour. Between-groupdifferences were analysed by Fisher's exact test and overallcorrelations by linear regression.

Results Thirty-six subjects (26 %) reported no behaviouralchange but, overall, around 25 % of diet, exercise andalcohol items had been changed. Women perceiving theirlifetime cancer risk to be high (>50 %) and those who wereobese (BMI >25) had made significantly more changes thanothers. Younger women (<40 years) and those with daugh-ters had made fewer changes. Almost all suggested elementsof a risk-reduction trial were strongly supported.Conclusions Scottish women at increased risk of breastcancer have scope for protective changes in lifestyle andsupport a risk-reduction trial. The needs of younger womenand of those with daughters should be addressed in itsdesign.

Keywords Breast cancer . Familial . Behaviouralintervention . Diet . Alcohol . Exercise

Introduction

In the UK and several other developed countries, clinicalservices for women with a significant family history ofbreast cancer typically include risk assessment and regularsurveillance by mammography combined, in some centres,with clinical examination. Under the UK National HealthService, genetic screening is offered to those families wherethe history suggests that there may be a mutation in BRCA1,BRCA2 or one of the other major genes implicated infamilial breast cancer and, where that is confirmed, prophy-lactic surgery will be considered [1, 2]. While most of theevidence that “lifestyle” factors influence breast cancer riskcomes from studies on the general population, there aregrounds for believing that these factors interact with geneticpredisposition to determine overall risk. This appears toapply both to rare, high-penetrance mutations [3–5] and tocommoner, low-penetrance genetic variants [6–9]. There is

L. McLeish :D. R. Goudie : J. N. BergDepartment of Clinical Genetics,Ninewells Hospital and Medical School,Dundee DD1 9SY, UK

L. McLeish :M. M. Reis : C. Stewart :D. R. Goudie : J. N. Berg :K. A. Hanning :H. Vysny :C. M. SteelTayside Familial Breast/Ovarian Cancer Service,Ninewells Hospital,Dundee DD1 9SY, UK

M. M. Reis :C. StewartDepartment of Surgery and Molecular Oncology,Ninewells Hospital and Medical School,Dundee DD1 9SY, UK

M. HarvieNightingale and Genesis Prevention Centre,Wythenshawe Hospital,Manchester M23 9LT, UK

K. A. Hanning : C. M. Steel (*)Medical School, University of St Andrews North Haugh,St Andrews KY16 9TF, UKe-mail: cms4@st-and.ac.uk

Int.J. Behav. Med.DOI 10.1007/s12529-012-9263-0

uncertainty as to which modifiable components of lifestyleare most relevant in this context but, apart from reproductivehistory, breast feeding and use of exogenous sex hormones,the most clearly implicated in adding to lifetime breastcancer risk are peri- and postmenopausal weight gain, lackof exercise and consumption of alcohol [10–12]. Manyreports have suggested that a “Mediterranean” diet, whichdiffers from the typical “Western” diet in having less meatand saturated fat, more whole grains and fresh fruit andvegetables, can offer some protection though the evidencedoes not support a very strong effect [13]. Other approaches,which tend to be based on very limited data, include avoid-ance of dairy products, attributing risk to the insulin-likegrowth factors they contain [14, 15], using foods containingrelatively high concentrations of phyto-estrogens (e.g.soy), as substitutes for milk, cheese and meat, whilestill other authorities advise increased consumption ofwholegrain cereals, high-fibre foods such as pulses and/or antioxidant-rich supplements which may be derivedfrom berries or herbs [16–18]. Vitamin supplements arecontroversial; some studies indicate an increased riskamong women who regularly take multivitamin preparationsbut others support a protective role, particularly for vitamins Cand D [19, 20].

Given the plethora of findings in the literature and thepaucity of incontrovertible evidence that any given behav-ioural change will make a substantial difference to an indi-vidual woman's risk of breast cancer, it is understandablethat most family history clinics offer only general advice,recommending a combination of diet and exercise thatshould contribute overall to good health and avoiding over-confident claims of anticancer efficacy [21]. The TaysideClinic (Dundee, Scotland), in which this study was con-ducted, issues a personal letter to every woman enrolled inthe familial breast cancer surveillance programme thatincludes advice to take regular exercise, to maintain ahealthy body weight, to eat a balanced diet with plenty offresh fruit and vegetables and to drink alcohol only inmoderation. These precepts are reinforced verbally duringclinic visits. The current familial cancer guidelines from theUK National Institute for Clinical Excellence (NICE) rec-ommend only that women attending clinics are made awareof the link between overweight/obesity and developingbreast cancer after the menopause [2].

Nevertheless, women enrolled in surveillance pro-grammes for familial breast cancer are often anxious to takeaction to modify the risk, both for themselves and for otherfamily members, notably their daughters [22, 23]. There istherefore a case for evaluating a behavioural interventionprogramme, probably concentrating on weight controlthrough diet and exercise, coupled with limitation of alcoholconsumption. As a first step, we have investigated changesin lifestyle already adopted by women in this situation,

further changes that might be acceptable to them within anintervention trial and factors that may encourage or inhibitcompliance.

Survey Methods

The study was approved by the Tayside committee forMedical Research Ethics REC ref no. 07/S1402/61. Over aperiod of 18 months from early 2008, women who had beenenrolled for at least 1 year in the Tayside Familial Breast/Ovarian Cancer Clinical Service, based at Ninewells Hospi-tal Dundee, were provided, before their annual appointment,with a written explanation of the purpose of the study andinvited to participate. We excluded any who had alreadybeen diagnosed with breast cancer, known carriers ofBRCA1 or BRCA2 mutations (since they had receivedparticularly intensive counselling) and those who had un-dergone or had decided to undergo prophylactic total mas-tectomy, for whom behavioural modification of breastcancer risk is no longer relevant.

Those consenting were then asked to complete, at theclinic, an anonymised four-section structured questionnaire.

Section 1 comprised limited personal information; agerange (under 35 or in 5-year spans from 35 to 59), maritalstatus, numbers of sons and daughters, self-reported heightand weight (used to calculate body mass index (BMI)), whatthey estimated to be their own lifetime risk of breast cancer(between 0 and 100 %) and the number of years elapsedsince they had learned of the possible familial breast cancerrisk. A coded individual risk category, as assessed by theprofessional clinic staff, was attached to each questionnaireonly after it had been completed by the subject and returned.Briefly, a woman having more than three close relatives (atleast one first degree) diagnosed with breast cancer at anyage or three such relatives diagnosed at average age lessthan 60 years is at “high” risk, while those with one affectedfirst degree relative diagnosed before age 40 or with twoclose relatives (at least one first degree) diagnosed at anyage or with three close relatives with later onset disease is at“moderate” risk. These definitions are modified by instancesof bilateral breast cancer, male breast cancer and/or ovariancancer in the family, as set out in the relevant NICE andScottish clinical guidelines. Broadly, moderate implies alifetime risk of 17–29 %, while the corresponding figurefor high risk is 30 % or more [1, 2].

Section 2 (see Table 1) enquired about habitual diet andactivity before enrolment in the clinic (i.e. before beinggiven an assessment of their familial risk of breast cancer).Seventeen food items were listed: brief explanations andexamples of the categories were given and each responseinvolved ticking one of five boxes, indicating how frequent-ly that item had been eaten. Further questions asked about

Int.J. Behav. Med.

alcohol consumption (specifying one unit as “1 glass ofwine or beer or 1 small measure of spirits”), daily physicalactivity, examples being “running, jogging, gardening, ac-tive walking, swimming, playing tennis, doing a physicallyactive job” and practice of yoga, meditation, Tai-Chi or thelike. Free text space was provided for further details ofvitamins, health foods or yoga etc.

Section 3 asked respondents to list changes they had madesince becoming aware of their possible familial risk, using thesame list of categories (see Table 2). Free text space allowedindividual comments on changes that had been made, includ-ing changes that had been tried but abandoned.

The final section suggested 13 interventions that might beconsidered for inclusion in a risk-reduction trial (see Table 3).

Table 1 Reported frequenciesof pre-enrolment consumption/activity (percentage giving eachresponse, N0140)

Daily >3 times/week 1–3 times/week Rarely Never

Red meat 0.5 19.5 52.5 23.5 3.5

Poultry 1.5 29 63.5 3.5 3

Fish 1.5 8 60.5 27.5 3

Cruciferous vegetables 18 21 46.5 27.5 3

Other vegetables 35.5 30 30 5 0

Salads 16 22 47.5 13 1

Fruit 42.5 17.5 27 12 1.5

Whole milk 15.5 1.5 2 22.5 58.5

Fat-reduced milk 71 3 7 10 9

Cheese/yoghurt 34 27.5 30.5 9.5 0

Soy products 3 1.5 2 13 80.5

High-fibre foods 14 19 46 21 0

Wholegrain foods 41.5 23.5 18 15.5 2

Vegetable oil 22 25.5 31.5 22 1.5

Confectionary 12.5 30.5 29 27 1

Vitamin supplements 14 1 2 15 66.5

Alcohol (>2 units/day) 6 (1–2 units/day) 17.5 – (<1 unit/day) 59 (None) 17.5

Physical activity (<1/2 h/day) 32 (1/2–1 h/day) 38 – (>1 h/day) 30 –

Yoga/Tai-Chi etc. (Daily) 2 (Weekly) 10 – (Rarely) 19 (Never) 68.5

Table 2 Percentage of respond-ents (N0140) reporting changesin diet etc. since learning offamilial breast cancer risk

Much less Less Same More Much more

Red meat 10.7 22.1 67.1 0 0

Poultry 1.4 3.7 80.7 14.2 0.7

Fish 0 0 69.3 28.6 2.1

Cruciferous vegetables 0 1.4 67.1 27.9 3.6

Salads 0 0.7 68.7 25.7 5.0

Fruit 0 1.4 68.6 25.7 5.0

Whole milk 9.4 2.2 82.1 7.9 0

Fat-reduced milk 0 2.2 82.1 7.9 0

Cheese/yoghurt 3.6 10.7 73.8 9.5 2.1

Soy products 0 0 92.9 4.5 2.1

High-fibre foods 0 2.1 68.6 23.3 5.7

Wholegrain foods 0 3.7 70.7 20.0 5.7

Vegetable oil 2.2 4.3 78.3 10.0 5.0

Confectionary 12.9 17.9 69.3 0 0

Vitamin supplements 0.7 1.4 85.3 8.8 3.3

Health foods 0 0 95.7 2.9 1.4

Alcohol 5.0 10.0 77.1 7.9 0

Physical activity 0.7 7.1 57.1 30.0 5.0

Yoga/Tai-Chi etc. 0.7 3.7 82.9 10.9 1.4

Int.J. Behav. Med.

For each option respondents were asked to indicate their will-ingness to adopt that change in lifestyle by ticking one of fiveboxes. Free text space invited suggestions for other changesthat individual respondents might like to see included.

For ease of presentation, response numbers have beenexpressed as percentages (with rounding) in all tables.

Analysis and Statistical Methods

As discussed below, we considered four patient characteristicsthat might plausibly influence lifestyle change: individualperception of breast cancer risk, obesity, age and having oneor more daughters. Subgroups of respondents were identifiedby choosing a single “cutoff” value for each of these criteria.Numbers of ticks in each box were recorded. Since 78 % ofreported changes in diet or exercise were “more” or “less”rather than “much more” or “much less”, we combined theseresponses to create just three categories; “no change”, “in-crease” or “decrease”. The totals for each category of responsewere compared between defined subgroups by Fisher's exacttest (two-tailed), using the online statistical package Graphpad(www.graphpad.com).

As the above process supported the prediction thatthe four specified patient characteristics could encourageor inhibit lifestyle change, we created an individualindex of “healthy change” by awarding a positive scoreof 1 for each potentially beneficial change reported (e.g.less red meat, more fruit, more exercise, less alcohol)and subtracting 1 for each change in the “unhealthy”direction (e.g. less salad, more confectionary, lesswholegrain foods, more alcohol, etc.). Individual valuesfor this index ranged from −3 to +14. At the same time,we derived a composite “incentive score” from 0 to 4for each subject, one point being given for each of thefour characteristics that appeared to encourage lifestyle

change (risk perception >50 %; BMI>25; age >39 years;having no daughters). The relationship between the index ofhealthy change and the incentive score was tested by simplelinear regression using the Vassarstats online statistical pack-age (http://vassarstats.net/corr_stats.html).

Results

Among women invited to take part, over 95 % did so. Onehundred and forty completed the questionnaire. Of these,one quarter were under age 40; 62 % were between 40 and49 and the remainder were 50 or over.

Four belonged to families with known BRCA1 or BRCA2mutations but had not yet decided whether to undergo testingto establish their own mutation status. For subgroup defini-tion, these four were combined with a further 46 (33 %) whohad been assessed as high risk. The remaining 90 (64 %) wereat moderate risk. One woman assessed as high risk and two atmoderate risk gave no estimate of their own lifetime risk ofbreast cancer, 34 quoted a figure <50 %, 45 judged it to be50 % and 58 believed they were at greater than 50 % risk.There was no correlation between professionally assessed andperceived risk; the mean risk estimate for those at high riskbeing 55.98 % (SEM 2.68), while for those at moderate riskthe corresponding figure was 50.88 (SEM 2.11). Mean differ-ence was 5.10, (95% confidence intervals −1.66 to 11.87): p00.14 by Student's t test.

Seventy-three women (52 %) had at least one daughter;33 (24 %) had no children and 32 (23 %) had sons only. Twoprovided no information about children. Sixty-five women(46 %) were judged to be obese as the BMI exceeded 25(calculated from reported height and weight).

“Habitual” diet (before enrolment in the familial breast can-cer surveillance programme) appeared largely consistent with

Table 3 Percentage of respond-ents (N0140) reporting willing-ness or otherwise to adoptspecific lifestylerecommendations

Definite“no”

Probable“no”

Uncertain Probable“yes”

Definite“yes”

Less red meat 1.5 0 6 48.5 44

More fish 3 1.5 2 44 49

More cruciferous vegetables 1.5 2 1 39 56

More fruit 1 1 5.5 36 57

Less whole milk 1 8.5 13 38 39

Less cheese etc. 1 5.5 12 42 40

Use soy milk in cooking 5.5 19 23.5 26.5 26

More wholegrain foods 2 1.5 6 30 61

More vitamin supplements 3 1.5 11 42 42

More health foods 2 6 16.5 35 40

Less alcohol 1.5 3 10 42 43

More exercise 1 1 8 44.5 46

Attend support group 4.5 8.5 28 35 24

Int.J. Behav. Med.

well-documented Scottish population records [24]. As shown inTable 1, there were only four strict vegetarians; overall con-sumption of vegetables was modest. Only 36 % of respondentsrecorded eating any vegetables every day, while thecorresponding figure for fruit was 43 %. Sweets (candies) and/or cakes were eaten more than three times per week by 45 %,but only 13 and 4 % were regular users of vitamin and healthfood supplements, respectively. Most respondents drank lessthan one unit of alcohol per day (59 %) or none at all (18 %),but 4.5 % reported drinking more than two units per day. Thestudy group divided almost equally into the three categories ofdaily exercise (“<1/2 h per day”; “1/2–1 h per day”; and “>1 hper day”). Only 12.5 % practised daily or weekly yoga, medi-tation or comparable activities, among which “Christian prayer”and “keep-fit” classes were mentioned specifically.

As recorded in Table 2, 36 subjects (26 %) had made nochanges at all to their diet, alcohol consumption or exercisepatterns since first attendance at the breast cancer family clinicfor risk assessment and enrolment in the surveillanceprogramme. Overall, however, for each of the listed items,around 25 % of women had made a change, and, in 89 % ofinstances, the changes were in the direction that would beconsidered “healthy” (less red meat, alcohol and confectionary,more fruit, vegetables, salads, fish, wholegrain cereals fibre andexercise). The distribution of healthy changes was not uniform:30–35 % of respondents had reduced consumption of red meatand sweets, increased consumption of fish, fresh fruit, saladsand other vegetables and were taking more exercise, but only15 % had reduced their alcohol intake, fewer than 7 % had(knowingly) increased their use of soy products and only 12 %of our cohort had taken up, or increased their participation in,yoga and comparable activities.

There was no significant difference in frequency of “life-style” changes between the 50 women assessed as being at highrisk (including the four members of BRCA1/2 mutation fami-lies) and the 90 assessed as at moderate risk. By contrast, the 34women who believed that they were at less than 50 % lifetimerisk for breast cancer had, on average, changed 21 % of thelisted categories, while the 103 who perceived their risk to be50% or greater had changed 26%of the listed items (p00.011).

The 65womenwith BMI >25 hadmade 27% of the possiblechanges, while the remaining 75 had changed 23 % (p00.015).

Seventy-three women had at least one daughter, and theirresponses were compared with those of 65 with no daughters.Unexpectedly, women with daughters had changed, on aver-age, 23 % of the items listed while the others had changed27 % (p00.03). One possible explanation would be thatwomen with daughters were already concerned about familialbreast cancer risk before contacting the clinic and so hadalready adopted a healthy lifestyle for the benefit of theirdaughters. In that case, they would have less scope for furtherchange following enrolment in the surveillance programme.Analysis of those items for which a healthy behaviour pattern

could be most clearly identified (red meat, fish, fruit, vegeta-bles, salads, soy products, wholegrain cereals, fibre, vegetableoil and confectionary) confirmed that women with daughtershad a significantly better diet profile than those withoutdaughters, before attending the breast cancer family clinic (p<0.02). However, the shift towards healthier consumption ofthese items, after enrolment in the surveillance programme,was much greater for those without daughters (p00.0001),equivalent to twice the initial difference between the twosubgroups. It was also apparent that mothers of daughtershad made more changes in alcohol consumption and exercisethan the rest of the cohort, although there had been no pre-enrolment difference between the subgroups for these items.

On comparing data for the 32women (23%) aged under 40and the 108 (77 %) aged 40 or over, the younger groupreported changing only 16.5 % of the items listed comparedto 27.2 % for the older subset (p<0.0001). When analysis wasrestricted to the dietary changes specified above that wouldgenerally be considered healthy, plus increasing exercise andreducing alcohol consumption, the younger women had made19 % of the possible changes as against 29.4 % for their oldercounterparts (p<0.0002). It was also apparent that on most ofthe items in section 2 of the questionnaire (pre-enrolment),including the amount of physical activity, the younger womenreported a less healthy lifestyle than the older ones, the excep-tions being consumption of alcohol and of confectionary.

Combining, for each subject, the four factors that appearto influence lifestyle change into a single incentive scoreand creating an individual “index of healthy change”allowed us to test their relationship by simple linear regres-sion. The result confirmed a strong correlation (slope00.97,t03.343 with 137 degrees of freedom, p00.0004). However,the fit was only moderate (r200.079).

There was strong support for almost all the options offeredfor possible inclusion within an intervention trial (Table 3). Onaverage, 83 % of respondents said that they “probably” or“definitely” would adopt almost all of the changes suggested.Notable exceptions were use of soy milk in cooking andattending a support group once every 2 months (53 and59 % in favour, respectively). No significant differences werefound between respondents categorised by assessed or per-ceived risk level, but the younger women (under 40 years, n032) were significantly more likely than the rest (n0108) togive positive responses (89 vs 81.2 %, p00.0003). Womenwho reported that they had made no changes in diet or exer-cise appeared just as willing to try the suggested measures asthose who had already implemented at least some changes.

Discussion

The data recorded here are derived from self-reports ratherthan direct measurement, and it is recognised that much

Int.J. Behav. Med.

more rigorous approaches are available to assess diet andother aspects of lifestyle [25]. This, combined with the lackof any measure of the relative strengths of the four compo-nents of our incentive score, probably accounts for much ofthe scatter on linear regression analysis. However, inaccu-racies in the raw data are unlikely to have introduced sys-tematic bias with respect to the subgroups analysed.Furthermore, in planning an intervention trial, the beliefsand perceptions of potential participants are arguably asimportant as the details of their actual behaviour. The find-ings are reassuring with regard to a possible interventiontrial. Many of the women attending the Tayside BreastCancer Family Clinic are following a lifestyle which,though not grossly unsatisfactory, could readily be im-proved and the great majority express willingness (evenenthusiasm) for health-enhancing changes.

Women who are already overweight reported significantlymore lifestyle changes than the remainder of the cohort. Giventhe recognised willingness of obese women to report healthybehaviour [26] and despite the high level of significance in ourstatistical analysis, the findings must be viewed with caution.If correct, they may indicate that information, providedthrough the clinic and from public sources, about risk factorsfor breast cancer is having an impact and that further educa-tional efforts are likely to be worthwhile.

Women perceiving their own cancer risk to be high(>50 %) reported more changes in behaviour. In fact, theyare greatly overestimating actual risk since even a closerelative of a BRCA mutation carrier will rarely be at morethan 40 % lifetime risk if she has not herself undergonemolecular testing [2]. Despite efforts to explain the level ofrisk in realistic terms, it is a near-universal experience thatwomen enrolled in surveillance programmes for familialbreast cancer tend to believe that their risk is much higherthan the professional estimate [21, 27–31]. Stronger moti-vation towards lifestyle change cannot be a justification forallowing or encouraging an inappropriately high perceptionof risk. It must be preferable to promote the idea that riskcan be modified by manageable changes in lifestyle such asincreased exercise and reduced alcohol consumption.

The finding that mothers of daughters make fewer dietarychanges than other women is disconcerting in view of theevidence that dietary contributions to breast cancer risk mayoperate most strongly in the early years of life [32]. It isdesirable that the whole family—specifically the youngdaughters of women currently attending breast cancer familyclinics—should be included in a healthy lifestyle approach tomodifying breast cancer risk but this cannot be taken forgranted. The observation from our study on the limited life-style changes achieved by mothers of daughters only justachieves statistical significance. It needs to be explored furtherthrough in-depth interviews, but, if confirmed, it may beappropriate to invite the active participation of daughters (at

least those of an age capable of giving informed consent) inany planned intervention trial. Given that maintaining ahealthy weight and reducing alcohol intake are among thebest-established protective measures against breast cancer, itis reassuring that our clinic patients who have daughtersappear to be taking more exercise and drinking less even iftheir freedom to modify the family diet may appear to berestricted.

The other unexpected finding is that younger women notonly report less healthy lifestyles (on most of the measuresapplied) than their older counterparts but have evidentlymade significantly fewer beneficial changes. This may sim-ply reflect a specific application of the “inverse care law”[33], in that women under age 40 are likely to carry morepressing family responsibilities and, hence, look after them-selves less well than their older “sisters”. It is encouragingthat they do express very strong enthusiasm for measuresthat might contribute to reducing their breast cancer risksbut special attention will have to be paid to them if theirgood intentions are to be translated into action, and this is animportant consideration because the earlier the age at whichlifestyle change is implemented, the more effective it islikely to prove.

There have been very few comparable studies from otherbreast cancer family clinics. To date, the only published UKstudy, from the West Midlands of England, assessed a co-hort of predominantly postmenopausal women, of whom77 % were overweight. While their diets appeared to berather better than those recorded in the present survey, only15 % achieved the recommended 4 h of physical activity perweek. No enquiry was made regarding changes in lifestylethat may have been made in response to breast cancer risk[34]. A small, mainly qualitative survey from the USAfound that perceived risk correlated positively with lifestylechange [35], while a larger one noted that, as in the presentanalysis, healthy lifestyle change was related to age, youn-ger women being least likely to change [36]. A third, NorthAmerican study, restricted to Ashkenazi Jewish subjectswith BRCA1 or BRCA2 mutations observed no significantbehavioural changes after counselling [37]. Two investiga-tions from Spain and the USA found that in the absence ofcounselling, women with a family history of breast cancermade few, if any, healthy lifestyle changes [38, 39]. Twofurther studies from North America, both concentrating onwomen at high familial risk, particularly those likely to carryBRCA1/2 mutations, after risk assessment and counselling,found a very much higher uptake of “complementary” ther-apies, including yoga, meditation and prayer, than recordedhere [40, 41].

With few exceptions, responses to the prospect of a risk-reduction trial based on changes in diet, alcohol consump-tion and physical activity were very favourable. However,this cannot be taken as a guarantee of compliance. It is

Int.J. Behav. Med.

noteworthy that the two groups for whom lifestyle changemight be most important (younger women and those withdaughters) appear to have made the fewest healthy adjust-ments and may require targeted help if their good intentionsare to be translated into practice. Yet regular participation ina support group was among the least popular of the optionspresented. In other health-promoting initiatives, text mes-saging and call-centre support have proved effective [42,43]. Particularly, if exercise were to be a major componentof an intervention programme, interactive regular reportingof progress and recording of weight change might prove asubstitute for the kind of motivation that would otherwisecome from a support group.

Acknowledgments The authors are grateful to all the patients whocontributed to this study, which was funded by the Leverhulme Trust.

Conflict of interest The authors have no conflict of interest todisclose.

References

1. Haites N, Hodgson SV and the Scottish Office Working Group.Guidelines for the development of cancer genetics services. In: P JMorrison, SV Hodgson and NE Haites (eds) Familial breast andovarian cancer: genetics, screening and management. CambridgeUniversity Press 2002, pp 166-203.

2. National Institute for Clinical Excellence (NICE). Familial breastcancer: clinical guideline no. 14. NHS London, 2004.

3. King MC, Marks JH, Mandell JB. Breast and ovarian cancer risksdue to inherited mutations in BRCA1 and BRCA2. Science.2003;302:643–6.

4. Trygvadottir L, Sigvaldason H, Olafsdottir GH, Johansson JG,Jonsson T, Tulinius H, Eyfjord JE. Population-based study ofchanging breast cancer risk in Icelandic BRCA2 mutation-carriers, 1920–2000. J Natl Cancer Inst. 2006;98:116–22.

5. Evans DG, Shenton A, Woodward E, Lalloo F, Howell A, MaherER. Penetrance estimates for BRCA1 and BRCA2 based on ge-netic testing in a clinical cancer genetics service setting: risks ofbreast/ovarian cancer quoted should reflect the cancer burden inthe family. BMC Cancer. 2008;8:155.

6. Gramling R, Lash T, Rothman KJ, Cabral HJ, Silliman R, RobertsM, Stefanick ML, Harrigan R, Bertoia ML, Eaton CB. Familyhistory of later-onset breast cancer, breast healthy behaviour andinvasive breast cancer among postmenopausal women: a cohortstudy. Breast Cancer Res. 2010;12:R82.

7. Vachon CM, Cerhan JR, Vierkant RA, Sellers TA. Investigation ofan interaction of alcohol intake and family history on breast cancerrisk in the Minnesota Breast Cancer Family study. Cancer.2001;92:240–8.

8. Carpenter CL, Ross RK, Paganini-Hill A, Bernstein L. Effect offamily history, obesity and exercise on breast cancer risk amongpostmenopausal women. Int J Cancer. 2003;106:96–102.

9. Narod SA. Modifiers of risk in hereditary breast and ovariancancer. Nature Rev Cancer. 2002;2:113–23.

10. Kushi LH, Byers T, Doyle C, Bandera EV et al., American CancerSociety guidelines on nutrition and physical activity for cancer pre-vention: reducing the risk of cancer with healthy food choices andphysical activity. (2006). http://caonline.americancancersoc.org/cgi/content/full/56/5/254.

11. World Cancer Research Fund and American Institute for CancerResearch. Food, nutrition, physical activity and the prevention ofcancer: a global perspective. Washington DC: AICR, 2007.

12. Gonzales CA, Riboli E. Diet and cancer prevention: contributionsfrom the European prospective investigation into cancer and nu-trition (EPIC) study. Eur J Cancer. 2010;46:2555–62.

13. Willett WC. Fruits, vegetables and cancer prevention: turmoil inthe produce section. J Natl Cancer Inst. 2010;102:510–1.

14. Pasani P, Berrino J, Fusconi E, Curtosi PFB. A European Case-OnlyStudy (COS) on familial breast cancer. J Nutr. 2005;135:3040S–1S.

15. Holmes MD, Pollack MN, Willett WC, Hankinson SE. Dietarycorrelates of plasma insulin-like growth factor1 and insulin-likegrowth factor protein 3 concentrations. Cancer EpidemiolBiomarkers Prev. 2002;11:852–61.

16. Brennan SF, Cantwell MM, Cardwell CR, Velentzis LS, WoodsideJV. Dietary patterns and breast cancer risk: a systematic review andmeta-analysis. Am J Clin Nutr. 2010;91:1294–302.

17. Ambrosone CB, Marshall JR, Vena JE, Laughlin R, Graham S,Hemoto T, Freudenheim JL. Interaction of family history of breastcancer and dietary anti-oxidants with breast cancer risk. CancerCauses Control. 1995;6:407–15.

18. Cade JE, Burley VJ, Greenwood C, the UKWomen's Cohort StudyGroup. Dietary fibre and risk of breast cancer in the UK women'scohort study. Int J Epidemiol. 2007;36:431–8.

19. Larsson SC, Akesson A, Bergkvist L, Wolk A. Multivitamin useand breast cancer incidence in a prospective cohort of Swedishwomen. Am J Clin Nutr. 2010;91:1268–72.

20. Cui Y, Rohan E. Vitamin D, calcium and breast cancer risk: areview. Cancer Epidemiol Biomarkers Prev. 2006;15:1427–37.

21. Chappuis PO and Foulkes WD. Management of BRCA1/2 muta-tion carriers. Ch. 15. In: PJ Morrison, SV Hodgson and NE Haites,(eds) Familial Breast and Ovarian Cancer: Genetics, Screening andManagement. Cambridge University Press 2002. Pp 237-274

22. Juilian-Reynier C, Eisenger F, Chabal F, Aurran Y, et al. Cancergenetics clinics; target populations and consultees' expectations.Eur J Cancer. 1996;32A:398–403.

23. McLeish L. Demands and needs of women attending 2 Scottishfamily history breast cancer clinics. MSc thesis, University ofManchester. 2003.

24. Report of Working Party of the Chief Medical Officer for Scotland.“Scotland's Health: a challenge to us all. The Scottish diet”. TheScottish Office, 2006. www.healthscotland.com/documents/1181.aspx.

25. Gibson RS. Principles of nutritional assessment. Oxford UniversityPress 2005.

26. Kretsch MJ, Fong AK, Green MW. Behavioral and body sizecorrelates of energy intake underreporting by obese and normal-weight women. J Am Diet Assoc. 1999;99:300–6.

27. Evans DGR, Burnell LD, Hopwood P, Howell A. Perception ofrisk in women with a family history of breast cancer. Br J Cancer.1993;67:612–4.

28. Croyle RT, Lerman C. Risk Communication in genetic testing forcancer susceptibility. J Natl Cancer Inst Monogr. 1999;25:59–66.

29. Watson M, Lloyd S, Davidson J, Meyer L, Eeles R, Ebbs S,Murday V. The impact of genetic counselling on risk perceptionand mental health in women with a family history of breast cancer.Br J Cancer. 1999;79:868–74.

30. Cull A, Anderson E, Campbell J, Mackay J, Smyth E, Steel M. Theimpact of genetic counselling about breast cancer risk on women'srisk perceptions and levels of distress. Br J Cancer. 1999;79:501–8.

31. Hopwood P, Shenton A, Lalloo F, Evans DG, Howell A. Riskperception and cancer worry: an exploratory study of the impact ofgenetic risk counselling in women with a family history of breastcancer. J Med Genet. 2001;38:139–42.

32. Ziegler RG, Hoover RN, Pike MC, Hildsheim A, et al. Migrationpatterns and breast cancer risk in Asian-American women. J NatlCancer Inst. 1993;85:1819–27.

Int.J. Behav. Med.

33. Tudor Hart J. The inverse care law. Lancet. 1971;297:405–12.34. Begum P, Richardson CE, Carmichael AR. Obesity in post-

menopausal women with a family history of breast cancer: preva-lence and risk awareness. Int Semin Oncol. 2009;6:1.

35. Spector D, Mishel M, Skinner CS, Deroo LA, van Riper M,Sandler DP. Breast cancer risk perception and lifestyle behaviorsamong white and black women with a family history. Cancer Nurs.2009;32:299–308.

36. McEligot AJ, Moutappa M, Ziogas A, Anton-Culver H. Diet andpredictors of dietary intakes in women with family history ofbreast and/or ovarian cancer. Cancer Epidemiol. 2009;33:419–23.

37. Quach J, Porter K, Leventhal H, Kelly KM. Health behaviorsamong Ashkenazi Jewish individuals receiving counselling forBRCA1 and BRCA2 mutations. Fam Cancer. 2009;8:241–50.

38. Ochoa EM, Gomez-Acebo I, Rodriguez-Cundin P, Navarro-Cordoba M, Llorca J, Diersson-Sotos T. Relationship betweenfamily history of breast cancer and health-related behaviour.Behav Med. 2010;36:123–9.

39. Spector D, DeRoo LA, Sandler DP. Lifestyle behaviors in blackand white women with a family history of breast cancer. Prev Med.2011;52:394–7.

40. DiGianni LM, Rue M, Emmons K, Garber JE. Complementarymedicine use before and one year following genetic testingfor BRCA1/2 mutations. Cancer Epidemiol Biomarkers Prev.2006;15:70–5.

41. Myers CD, Jacobsen PB, Huang Y, Frost MH, Patten CA, CerhanJR, Sellers TA. Familial and perceived risk of breast cancer inrelation to use of complementary medicine. Cancer EpidemiolBiomarkers Prev. 2008;17:1727–534.

42. Cole-Lewis H, Kershaw T. Text messaging as a tool for behaviourchange in disease prevention and management. Epidemiol Rev.2010;32:56–9.

43. Young RJ, Taylor J, Friede T, Hollis S, et al. Pro-active call centertreatment support (PACCTS) to improve glucose control in type 2diabetes: a randomised controlled trial. Diabetes Care. 2005;28:278–82.

Int.J. Behav. Med.