Testing a Breast Cancer Prevention and a Multiple Disease Prevention Weight Loss Programme Among Women Within The UK NHS Breast Screening Programme-A Randomised Feasibility Study

Michelle Harvie (  michelle.harvie@manchester.ac.uk ) Manchester University NHS Foundation Trust https://orcid.org/0000-0001-9761-3089 David P French University of Manchester Mary Pegington University of Manchester Grace Cooper Manchester University NHS Foundation Trust Anthony Howell University of Manchester Sarah McDiarmid Manchester University NHS Foundation Trust Cheryl Lombardelli Manchester University NHS Foundation Trust Louise Donnelly Manchester University NHS Foundation Trust Helen Ruane Manchester University NHS Foundation Trust Katharine Sellers Manchester University NHS Foundation Trust Emma Barrett Manchester University NHS Foundation Trust Christopher J. Armitage University of Manchester D Gareth Evans Manchester University Hospitals Foundation Trust

medications (13). Offering CVD and T2D risk assessment and a behaviour change programme at breast screening could provide systematic access to CVD and T2D care pathways due to high NHSBSP coverage and thus potentially better outcomes.
This feasibility study included women with overweight/obesity who had been previously informed of their personal risk of BC as part of the PROCAS study (1). The PROCAS study involved women aged between 46 and 73 years attending the Manchester NHSBSP (49% of those offered study entry). This included women in the age extension trial of the NHSBSP(14) 3-years either side of the usual 50-70 age group. Women were provided with a personalised estimate of their risk of BC derived using the Tyrer-Cuzick model (version-8 which includes family history, hormonal risk factors, BMI and visually assessed mammographic density (1). ThePROCAS lifestyle study reported here aimed to explore the feasibility of BCPP and MDPP for women following receipt of BC risk estimates as part of risk strati ed screening. A sample of women participating in the PROCAS study were randomised to receive an invitation to join a standard web and phone Breast Cancer Prevention Programme (BCPP) which reminded women of their personal risk of BC or a Multiple Disease Prevention Programme (MDPP) which additionally included an NHS Health Check to assess lipids, blood pressure, Hba1c, provided personalised risk feedback for CVD, T2D, and emphasised the links between health behaviours and dementia. We previously reported that higher BC risk predicted uptake retention and weight loss success across the combined BCPP and MDPP experimental conditions (Harvie et al 2019). We now report the main trial objectives which includes the feasibility, acceptability and potential e cacy for weight loss and behaviour change of the BCPP and MDPP. We also report any associated harms of the programmes, which included whether informing women they were at different levels of BC risk only (BCPP) or CVD, T2D and BC risk (MDPP), predicted lower levels of engagement, i.e. retention and weight loss success within the groups, the delity of delivery and engagement to the programmes and patient feedback. We report the prevalence of previously unknown CVD and T2D risk identi ed in the MDPP group to inform the potential of the MDPP to identify high risk individuals attending breast screening. A further aim was to determine which of these programmes should be tested further in a de nitive large scale RCT within the NHSBSP.

Method Study design
We conducted a single centre prospective two arm randomized trial (1:2) of BCPP vs MDPP amongst women attending the NHSBSP, and who had participated in the PROCAS study. Recruitment was between November 2014 and October 2015. This period was a median (range) 55 (5 -72) months after completing BC risk assessments and 6.5 (0 -63) months after receiving personalised BC risk feedback as part of the PROCAS study. The long lag between risk assessment and feedback was because of the need to validate the Tyrer-Cuzick risk assessment model among women attending the NHSBSP. All trial procedures were undertaken at the Prevent Breast Cancer Research Unit at Wythenshawe Hospital, Manchester University NHS Foundation Trust.

Recruitment and randomisation
The study included women with overweight/obesity (BMI ≥25kg/m 2 ) aged 47-73 years who had been identi ed and informed they were at high (10-year risk ≥8%), above average (5-7.9%), average (2-4.9%) or low-risk (<2%) of BC as described previously (1). Women were randomised 1:2 to receive a mailed invitation to either the BCPP or MDPP programmes. Randomisation to the two different invitations allowed us to assess uptake separately to the two programmes (Fig 1). Interested women were asked to check their eligibility on the trial website, or to ring the trial o ce if they did not have immediate access to the internet. Women were excluded if they did not have access to a phone or the internet, had a previous diagnosis of cancer, T2D or CVD, were currently prescribed statins, had a major physical or psychiatric condition which made them unsuitable for a home based diet and physical activity (PA) programme, or were current users of HRT since weight only affects risk of BC amongst non-HRT users (15). The invitation letter included an opt-out slip to indicate reasons they were not eligible or not interested.
The sample size of 120 (40 BCPP and 80 MDPP) was pragmatic and included 10 women from each BC risk group to the BCPP and 20 from each BC risk group to the MDPP to allow us to assess uptake, retention and weight loss success across the BC risk groups. We included a larger MDPP group to assess issues that are likely to arise when patients are given this novel programme Randomisation was undertaken independently from the research team in the Department of Statistics at Manchester University NHS Foundation Trust using nQuery Advisor 7.0. The rst batch used a 1:2 randomisation of BCPP to MDPP across the four BC risk groups. Subsequent randomisations were adaptive to the response rate of each risk category in the two groups until 45 women were randomised to the BCPP and 81 to the MDPP groups.
The BCPP and MDPP programmes Both programmes were delivered by research dietitians who had undergone training on communicating personalised risk of BC, CVD, T2D and dementia and the lifestyle prevention of these conditions. The BCPP reminded women of their personal BC risk. The MDPP provided personal BC risk and also their personalised risk of developing CVD (10-year and lifetime risk and heart age from QRISK2 (16) and T2D (10 year risk from QDiabetes (17) and measured HbA1c).
Both programmes promoted weight loss of ≥5% and communicated how weight loss reduces disease risk as a gain-framed message to increase response e cacy (18). The BCPP group received generic advice that weight loss of ≥5% and adherence to PA and alcohol recommendations could lead to signi cant reductions in risk of BC (25%) (19;20), T2D (60%) (21) and CVD (30%) (22). The MDPP group were provided personalised estimates of changes in CVD risk from predicted reductions in blood pressure and total cholesterol from published literature (i.e. a 1 mm/Hg reduction in systolic blood pressure per 1% weight loss up to a 10% weight loss(23) and a 1% reduction in total cholesterol for every 1% weight loss up to a 15% weight loss) (24). Changes in T2D risk were estimated by entering the target reduced weight in the QDiabetes tool. Results of the NHS Health Check were sent to the patient's family doctor to allow appropriate follow up and clinical management, e.g. checking abnormal results and consideration of medications for raised cholesterol, blood pressure and HbA1c.
Both programmes included individual face-to-face diet and PA advice for women to follow either a daily or intermittent (5:2) energy restricted Mediterranean diet which we have found to be effective in our previous weight loss studies (25). Women were also advised to meet PA recommendations (i.e. at least 150 minutes of moderate intensity or the equivalent amount of vigorous activity as well as 40 minutes of resistance exercise per week) (26). Suitability to follow a home based PA programme was con rmed using the adult pre-physical activity screening system tool (27), with family doctor clearance where necessary. PA advice was tailored to each patient's preferences, abilities and co-morbidities, with referral to local services where appropriate. Women were asked to limit alcohol intake to <10 units/week and smoking cessation was discussed where appropriate. Those reporting high-risk alcohol intakes with Alcohol Use Disorders Identi cation Test (AUDIT) scores >8 (28) or who were currently smoking were encouraged to access local NHS alcohol and smoking cessation services and their family doctor was informed.
Both programmes were supported by a trial website which encouraged women to self-monitor and record their weight, diet (completion of restricted days in the 5:2 diet and actual food and drink intake) and cardiovascular and resistance exercise. It hosted weekly menu plans, recipes, information about BC, CVD, T2D and dementia, tips for planning and managing emotional eating, online videos of the recommended resistance and stretching exercises (Physiotec, Canada) and a monthly newsletter to maintain engagement. Women received tailored feedback on their self-reported behaviours from their allocated trial dietitian in scheduled phone calls (weeks 1,4,8) and weekly personalised e-mails for 6-months. Between 6 and 12-months women received automated monthly emails in response to website entries. These e-mails gave positive feedback for records showing weight loss or weight maintenance, or encouraged re-engagement with the programmes if weight had increased by ≥1kg or if no website entries had been recorded.
Communicating personalised risk information without increasing self-e cacy or encouraging self-regulation is not likely to produce changes in behaviour (29) . Given this, both programmes included key behaviour change techniques as recommended in the UK NICE behaviour change guidance, including goal setting, planning, relapse prevention, self-monitoring, and encouraging individuals to identify sources of social support for changing behaviours (30).

Uptake and retention
For each programme we recorded the number of women who were invited and consented to join the study, and who were not interested or not eligible to take part and the reasons for this. Also retention to the programmes at 3, 6 and 12 months.

Health-related behaviour change and weight loss
We assessed change in weight and body fat (Tanita MC-180MA,Tanita Europe, Amsterdam, The Netherlands), self-reported PA (IPAQ)(31), alcohol harms and intake (AUDIT (28) and 7-day food diary), saturated fat intake (7-day food diary) (32) and current smoking in both groups.
Potential harms of the programmes Potential harms of both programmes were assessed in terms of changes in perceived self-rated anxiety (state trait anxiety) (33) and health status (EQ-5D-5L) (34). Within the MDPP we assessed whether being identi ed at different levels of CVD and T2D disease risk was associated with disengagement from behaviour change, evidenced by any differences in withdrawal and degree of weight loss.
Previously unidenti ed CVD and T2D risk in the MDPP group We assessed the prevalence of previously unknown raised CVD and T2D risk identi ed in the MDPP programme. Also the numbers referred to NHS behaviour change services (i.e. exercise on referral, alcohol and smoking cessation services) or commenced on statins or blood pressure medications.

Fidelity of delivery and engagement to the programmes
We assessed the numbers of scheduled calls and e-mails received, engagement with the web site and the amount of dietitian time that was used to deliver the programmes.

Patient feedback
We collated anonymised 12-month patient questionnaires which rated satisfaction with the overall programme, study visits and the trial website. This included free text suggestions for how the programmes could be improved. A speci c website feedback questionnaire was completed at 3 and 6 months to include feedback from women who might leave the programme before 12 months.

Analysis
Descriptive statistics of uptake and retention and changes in weight and health behaviours in the MDPP and BCPP groups were undertaken. Baseline observation carried forward (BOCF) values are reported for changes in weight (% and kg), body fat (kg), alcohol intake and PA, and a completers only analysis for state anxiety and self-rated health status. We also undertook exploratory formal statistics to assess comparisons of changes in these parameters relative to baseline between the BCPP and MDPP conditions. The ndings of these are reported with the caveat there is multiple testing of secondary outcomes and the study is not necessarily powered to show differences.
Within the MDPP group, multivariable logistic regression was used to assess the association between estimated level of CVD and T2D risk and the likelihood of withdrawal at 12 months. This was assessed in terms of 10-year CVD QRISK2, heart age compared to actual age or having a rst degree relative with CVD, 10-year QDiabetes and having a rst degree relative with diabetes. A priori confounding variables included; BC risk (high, above average, average and low), index of multiple deprivation (IMD quintiles ; 1, 2 and 3-5)(35), smoking status (never, ex-smoker and current smoker), age and BMI at baseline.
Within the MDPP group linear mixed-effects models were used to assess the percentage of weight loss from baseline over time in low and high CVD and T2D risk categories using all available baseline 3, 6 and 12-month weight data. The best-tting variancecovariance structure was based on information criterion, and nested models were compared using likelihood ratio test. Each speci c CVD and T2D risk category (low vs high) was treated as a xed-effect, and time as a continuous covariate. Time was also included as a random-effect, allowing the rate of weight loss to differ between patients. Polynomial factors of time were tested as both xed and random effects to see if their inclusion improved model t. The confounding variables used in the withdrawal analysis were included as xed effects, and comparisons made between the coe cients of the risk variables before and after adjustment. P<0.05 was considered statistically signi cant for all analysis. All statistical analyses were conducted using SPSS version 23 (IBM Corp, Armonk NY, USA), R version 3.5.1 and Stata Statistical Software: Release 16.)

Results
Recruitment and retention to the programmes Uptake to the study was comparable with invitations to both the BCPP and MDPP (45/508 9%; vs. 81/848, 10% p=0.7) (Figure 2). An additional 57 women invited to the BCPP (11%) and 135 invited to the MDPP (16%) were interested in join but were not eligible. This was mainly due to existing health problems such as previous CVD, cancer and raised cholesterol (7.1% BCPP, 12.0% MDPP). Other reasons for declining included lack of internet access (1.4% BCPP, 0.8% MDPP) and current use of HRT (1.0 % BCPP, 1.2% MDPP). A further 8 women invited to the BCPP (2%) and 7 invited to the MDPP (1%) wished to enter the study but responded after recruitment was complete. An additional 130 women (10%) invited to both programmes returned a response slip to say they were not interested to take part. The most common reasons cited were lack of time (28%), issues travelling to the centre for trial appointments (18%), family illness or caring duties (12%). Uptake to both programmes was greater in women at high and above average BC risk compared to average and low risk as reported previously (Table 1) (36).
Baseline characteristics of participants are reported in Table 2. The BCPP and MDPP groups were comparable. Women were mainly from the three least deprived quintiles of the index of multiple deprivation and were white British. Approximately one third of the cohort had a family history of BC (BCPP 26.7% MDPP 34.6 %,). Sixty-two percent of the BCPP and 60.5% MDPP groups had previously undertaken a commercial weight loss programme, whilst there were few smokers (BCPP 6.7% vs MDPP 7.4%).
The proportion reporting at risk drinking (AUDIT score ≥8) reduced from 8 (17.5%) in the BCPP and 22 (27.2%) in the MDPP at baseline to respectively 4 (8.9%) and 14 (17.3%) at 6 months. Likewise the proportion of women reporting saturated fat intakes above the recommended intake of 20g/day reduced from 29 (87.9%) in the BCPP and 40 (72.7%) in the MDPP at baseline to respectively 12 (36.4%) and 19 (34.5%) at 6 months. Both groups also reported modest increases in moderate intensity PA at 6 months. One woman from each group had stopped smoking by 6 months and were stillnot smoking at the 12-month follow up.
Does level of disease risk predict retention and weight loss success in the BCPP & MDPP? Retention and weight loss success appeared to be greater amongst women at higher risk BC risk in the BCPP group, but were more variable across BC risk categories for the MDPP group (Table 1).
Within the MDPP, women informed they had a higher CVD QRISK2 score at baseline were more likely to leave the programme.
Adjusted odds ratio for leaving (95% CI) 1.36 (1.08, 1.80) for each unit increase in 10 year CVD QRISK2 score (p=0.02) and 1.22 (1.03,1.46) for each year of predicted heart age greater than their actual age (p =0.02). However, neither weight loss success nor rate of weight loss differed according to level of estimated CVD or T2D risk (Table 5 and Figure 3).
Graphs for the predicted probabilities of percentage weight loss over time from the mixed-effects models. These contained the baseline level of CVD and T2D risk predictor, time, time 2 and predictor x time interactions as xed effects, as well as patient-level random effects adjusting for the confounding variables and confounder x time interactions which had been found to be signi cant at P<0.1 (Table 5).

NHS Health Check results in the Multiple Disease Prevention Programme
Ten women (12.3%) in the MDPP group had previously had an NHS Health Check that had been undertaken more than 6 months prior to joining the study. Relatively high numbers of women in the MDPP group had previously unknown increased risks for CVD (14.8% with ≥10% 10 year risks of CVD and 14.8% with total cholesterol ≥7.5 mmol/L) and for T2D (56% with ≥10% 10-year risk of T2D, and 6.2% with HbA1c ( Table 3). Some of these women were subsequently commenced on medications for blood pressure (n=2) and raised cholesterol (n=1) by their family doctor.

Fidelity of programme delivery
The initial face-to-face consultation with the research dietitian included personalised risk for just BC (BCPP) or BC, CVD and T2D (MDPP) and diet and PA advice for both groups. On average these took 40 minutes for the BCPP group and 60 minutes for the MDPP group.
Across both groups, women received 90% of their scheduled week 1, 4 and 8 calls and 95% of their scheduled e-mails. Mean (range) duration of calls was 21 (6 to 50) minutes and mean ( range) time taken for staff to prepare and send e-mails was 15 ( 5 to 30) minutes . There was good engagement with the self-monitoring website. Ninety-seven percent of women used the website at least once in the rst 3 months, 87% 3-6 months, 83% 6-9 months and 79% 9-12 months. Twenty-seven percent of women received an automated e-mail between 6 and 12 months to alert them they had recorded weight gain of ≥1kg. Mean (95% CI) number of website entries over the 12-month period was 89 (76-102) entries per woman. Ninety-nine percent of women opted to follow the 5:2 diet at baseline with respectively 82.5% and 58% still following this diet at 3 and 6 months. A number of women in the BCPP and MDPP groups were referred to NHS behaviour change services; PA referral services 8 (18%) BCPP and 23 (28.4%) MDPP, smoking cessation 1 out of 3 smokers BCPP and 5 out of 6 smokers MDPP, and alcohol services 0 BCPP and 4 (4.9%) MDPP by their allocated dietitian.
Website evaluation was completed by 109/126 (87%) of women. Ninety-two (73%) patients agreed or strongly agreed the website was easy to use, and 92 (73%) felt con dent to use the website. Thus, the phone and e-mail feedback was well received. Women understood and positively evaluated the CVD and T2D risk feedback and dietary elements of the programme. However, they were less con dent to follow the PA plans.
Free text feedback showed that women felt their self-e cacy to undertake PA would increase with a demonstration class or DVD.
They emphasised the need for review calls to be scheduled, and that feedback e-mail should be personalised and not generic. They valued the advice from the dietitian who was considered a credible source of information. A number of changes to the website were suggested including a forum to facilitate peer-to-peer communication and support, an 'ask the expert' function, and adding an average weight loss line to their weight tracker so they could compare their weight loss with the average of the other participants.

Main ndings
The BCPP and MDPP appear feasible with good uptake, retention weight loss success and delity. Both programmes produced mean weight loss at 12 months of over 6kg (7%). The MDPP highlighted previous unknown risk of CVD and T2D but does not appear to increase engagement with behaviour change beyond a standard BCPP amongst women attending breast screening. There was no evidence of harms in terms of self-rated anxiety or poorer health status with either programme.
Overall uptakes to the mailed invitation to both programmes was around 10%. This is comparable to previously reported uptakes to behaviour change programmes when written invitations are given to women when they attend the NHSBSP, or when postal invitations are sent to women at increased risk of BC (37). This feasibility study highlights potential strategies to increase recruitment to weight loss programmes within the NHSBSP. Firstly, focussing on women at high or above average BC risk who had uptakes of 14-20% compared to 4-10% for women at average or low BC risk reported herein and previously within this cohort (36). Higher risk women will also derive a greater absolute risk reduction from changes in health behaviours compared to low risk women. (38;39) Future programmes could include women with existing health conditions. Signi cant proportions of women who were interested to join the programmes were ineligible due to pre-existing weight related health conditions (7.1 BDPP and 12% of the MDPP group). We presume these weight related health conditions were not currently being addressed within existing NHS weight loss/behaviour change services. Inclusion of these women would have given uptakes of and 18.1to the BCPP and23.0% to the MDPP. Lastly, recruiting women at their mammogram appointments and using remote collection of outcome measures such as weight would overcome identi ed barriers of time and travel around additional appointments reported by women who declined to join this study.
Both programmes appear highly effective for weight loss and behaviour change. The mean weight loss at 12 months over 6 kg using baseline carried forward imputation compares favourably with outcomes from group-based commercial (2-4kg), internet (1.9 kg), and primary care programmes (0.7 -3.6 kg). (40) There was good engagement with the website and phone supported programmes. This hybrid face-to-face and remotely delivered programme allowed women to receive initial individual face-to-face advice from a health care professional combined with the website that allows self-monitoring and prompt regular feedback from the health care professional. Such programmes have been found to be both more and less effective compared to standard face-to-face programmes, dependent on the level of user engagement achieved (41).
Most previous website programmes have been tested amongst younger cohorts. The few studies amongst older women concur with our ndings of good engagement (42)(43)(44), and the potential for their utility amongst women of breast screening age. The Further patient involvement and engagement (PPIE) work can help re ne the invitation and recruitment strategy to the programmes. This could include a personalised letter of endorsement of the trial from their family doctor (46). PPIE work is speci cally required to promote uptake amongst women from black, Asian, minority ethnic and socially deprived groups who had low uptakes to this study, as seen with previous breast screening and cancer trials (47). Other potential strategies, such as face-toface introduction to the trial at the time of screening and telephone follow up after the mailed invite (48) are unlikely to be feasible due to the size of the target population.
There has been minimal research on the feasibility or e cacy of multiple disease prevention programmes. It was thought that multiple disease risk information could be personally relevant to a larger group of women than just providing risk information on a single disease such as BC, increasing the likelihood of having a positive effect on behaviours (49). Alternatively, it was possible multiple disease risk information could reduce motivation and self-e cacy and threaten self-integrity to change health behaviours in some individuals. We found no difference in engagement and behaviour change with a supervised MDPP compared to the BCPP.
There was no evidence of harm with the MDPP such as increased anxiety, or lower self-rated health assessed with the EQ-5D as a result of being provided with personalised risk of CVD and T2D. This is an important nding which is in line with other studies of providing risk information, despite concerns about adverse effects(50) (51). The higher drop out in women at higher baseline risk of CVD in the MDPP group may be an incidental nding. CVD risk information by itself has minimal effect on health behaviours (52). In the recent INFORM trial neither genotypic or phenotypic coronary heart disease risk information in uenced the e cacy of a web based lifestyle programme for changes in PA, diet or weight (53). The mixed effects models in our current study showed comparable weight loss success according to level of CVD; however, these models were based on small numbers and assumed that data was missing at random. Future studies should assess whether CVD risk information impacts on engagement and the e cacy of weight loss programmes.

Strengths and weakness
This is one of the few studies to test the feasibility and acceptability of a BC prevention and multiple disease prevention programmes in women in the breast screening programme. Also, one of the few to test website and phone programmes amongst middle -aged women. Twelve-month data including 6 months of website-only support provides an indication of the long term effects of the programmes for weight loss maintenance.
Key limitat19ions that a future effectiveness trial should address are as follows. The study invited a selected group of women who had previously been recruited to the PROCAS study, who were recruited sometime after their attendance at breast screening.
Studies need to test uptake to the programmes in an unselected cohort at the time of breast screening. In addition, all trial procedures were conducted at a research unit. Future effectiveness studies should ideally be in a screening site or a nearby community location to inform how these programmes could be integrated within the NHSBSP or be totally remote or online.
Secondary trial end-points (e.g. weight) were assessed by members of the research team which sometimes included the research dietitians delivering the intervention who were not blinded to the intervention. The study included two intervention groups and did not include a no intervention control group which would be required in a de nitive randomised trial.

Implications and Future research
We set out to test the feasibility and acceptability of a BCPP or MDPP amongst women in the breast screening programme and whether there were any additional bene ts with the MDPP programme. Hence which programme was worthy of further investigation in a large scale RCT within the NHSBSP. Both programmes met the NICE criteria for the e cacy of weight loss programmes, i.e. at least 60% of partcipants are likely to complete, average weight loss is at least 3%, and at least 30% of partcipants lose ≥5% of their initial weight (54). The simpler lower cost BCPP would be the preferred test programme to assess long-term behaviour change and weight loss and reduced risk of BC vs standard care and other weight and health behaviour related conditions. Increased coverage and appropriate medication are current targets for the NHS Health Check programme (13). Given this, future trials could assess the impact of offering a MDPP around the time of breast screening to increase coverage of CVD and T2D risk assessment and whether a one-stop shop can increase NHS e ciency and potential uptake to the NHS BSP.
Dietitians delivered the test programmes. Future programmes could consider provision of original dietary advice by a dietitian as a trusted source of knowledge, with ongoing support from a health trainer to reduce costs. This model of delivery has been favourably evaluated in other weight loss settings (55) and likely to be more effective than programmes which are entirely delivered by non-specialists which have achieved only modest weight loss (56). Further work is required to improve the PA component of the programme which had modest effects on PA.

Conclusion
The MDPP identi ed previously unknown CVD and T2D risk factors but does not appear to increase engagement with behaviour change beyond a standard BCPP amongst women attending breast screening. The results suggest a de nitive effectiveness trial of the BCPP intervention is warranted. Weight loss achieved with a BCPP will reduce risk of BC as well as CVD and T2D and other weight related cancers and health conditions.