Abstract
Background
Family tracing is a method recognized to find new patients with familial hypercholesterolaemia (FH). We have implemented family tracing led by FH Nurses and have determined acceptability to patients, feasibility and costs.
Methods
Nurses were located at five National Health Service (NHS) Trusts; they identified FH patients and offered them family tracing. Responses and test results were recorded on a database and summarized on a family pedigree.
Results
The majority (∼70%) of index cases participated; the proportion was lower when patients had been discharged from the clinics and in metropolitan areas. On average, 34% (range 13–50%) of relatives lived outside the catchment area of the clinics and could not attend the nurse-led FH clinics. Of the previously untested relatives, 76% who lived in the catchment area of the clinic came forward to be tested. One-third of the relatives who came forward for testing were children ≤16 y of age. The proportion of relatives diagnosed as likely to have FH was lower than would be predicted (30% vs. 50%). This was mainly due to the uncertainty of a diagnosis based on lipid measurements. The average cost to identify and test one relative was approximately £500 but was higher in the metropolitan areas.
Conclusion
Cascade testing for FH in the UK is feasible, acceptable and likely to be cost-effective if it is a routine aspect of clinical care. However, national implementation would require an integrated infrastructure, so that all individuals have access to testing, and specialist services for the management of young people.
Introduction
Familial hypercholesterolaemia (FH) is an autosomal co-dominant disorder which has an estimated prevalence in the UK of roughly 1 in 500. 1 In the UK, diagnosis of FH is based on a criteria developed by the Simon Broome (SB) FH register, 2 which includes LDL-cholesterol (LDL-C) concentrations, clinical signs and family history. Recent data suggests that only about 15% of the predicted FH patients in the UK are currently being treated in lipid clinics, 3 with the majority of these being older individuals. 1 The genetic causes of FH are well understood 4 and lead to high concentrations of LDL-C from birth. FH patients have an increased risk of premature coronary heart disease (CHD) such that, if untreated, roughly half of men and 30% of women with FH would have developed CHD by the age of 55. 5 Treatment with lipid-lowering agents such as statins can reduce coronary morbidity and mortality, 6,7 making the identification of affected individuals crucially important.
Modelling has suggested that ‘cascade testing’, that is contacting first-degree relatives of FH probands (index cases) and identifying affected relatives by their elevated cholesterol concentrations, is a cost-effective method of finding additional patients. 8,9 This method of case-finding has been used extensively in other countries in Europe, 10–12 and in the UK research projects in Manchester 13 and Oxford 14 have demonstrated its feasibility. To examine how applicable this is to mainstream health services, the Department of Health funded a pilot study 15 to determine the efficiency of cascade testing in a lipid clinic operating within the NHS Trust.
In the first audit phase of the project, we identified that 79% of FH patients in the lipid clinics met the UK SB register criteria for the diagnosis of Definite or Possible FH and that 89% were undergoing a statin therapy. 16 The audit also indicated variability in family tracing between the sites, with significant differences in the frequency of inclusion of a family pedigree in the notes (range 1–71%, mean 35%), the general practitioner (GP) being advised that first-degree relatives should be tested (range 4–52%, mean 27%), and the proportion of relatives contacted and tested (range 6–50%, mean 32%). 16 This paper reports on the effect of the implementation of systematic recording and tracing of first-degree family members.
Methods
Resources
Five centres were chosen to run the pilot study (the selection of sites has been reported previously). 16 Each centre employed a full-time nurse for 24 months in order to carry out (i) a retrospective audit of notes of usual practice; (ii) enhanced cascade testing as recommended by the current best practice; 14 (iii) re-audit of notes to obtain estimates of the number of index cases and relatives participating, and the number of affected FH relatives identified.
The nurses were supplied with a laptop, with the pedigree drawing software (Cyrillic version 2, Cyrillic Software, AB) installed to facilitate the identification and tracing of family members, and a database for FH patient and family records. The nurses were asked to record the time spent on different activities in a separate database. Near-patient testing equipment, Cholestech LDX™ (Banson, Oxfordshire, UK) machine and cassettes to run a lipid profile were provided to measure cholesterol, triglyceride and HDL-cholesterol from finger-prick blood samples. LDL-C was calculated using the Friedewald formula.
Data storage and analysis
Data storage and retrieval conformed to the Data Protection Act and local Caldicott guardian rules. Data on laptop computers were stored in an encrypted drive that was secured biometrically, and only anonymized data were transferred to the coordinating centre for analysis. Data cleaning and analysis were performed as described previously. 16
Nurse training
Nurses at each site were familiarized with lipid metabolism and its disorders, particularly FH. Training was also provided in lipid clinic methods, genetics and the use of pedigree drawing software.
Audit
Medical records of known FH patients were accessed to obtain information on diagnosis, treatment and family tracing, and the results of the pre-service development audit previously reported; these include complete details of pre- and post-treatment lipid concentrations for the ‘Not-FH’ by SB, Definite and Possible FH patients. 16 At the end of the service development, the audit process was repeated to determine the impact of providing FH nurses to the clinics.
Family tracing
The Manual of Operations and associated documentation (letters, standard operating procedures, etc.) are available at
Index patients
The telephone numbers of the FH nurses were provided to the patients and relatives for enquiries and consultation. Reasons for non-participation were collected. Index patients who met the SB criteria for Definite or Possible FH were contacted by letter, sent from their consultant, or were approached at their next clinic appointment. Those contacted by letter were re-approached after three weeks and if they did not respond after three attempts it was assumed that they did not wish to participate. All communications were documented. Index patients consenting to have their information stored on an electronic database and to family tracing were asked to provide the names and relationships of all first-degree relatives. The nurse offered to contact the relatives on the patient's behalf (Direct Contact) or provided the patient with stamped addressed envelopes to be posted to their relatives (Family Contact). The contact method was recorded.
Relatives of index patients
Relatives received a personalized letter explaining the reason for suspecting that they might have increased cholesterol and suggesting that they arrange to be tested at the project site or visit their GP (to avoid undue distress, FH was not mentioned until the relative spoke to a nurse). A letter for the GP was provided should the relative prefer this option, or live outside the catchment area of the project site. The GP was asked to advise the project site of any investigations undertaken by them and to refer any with cholesterol above the cut-off to a lipid clinic. The GP pack contained a letter explaining why their patient should be tested, a leaflet on FH and colour coded, age- and gender-specific cut-off charts for LDL-C and total cholesterol 17 and a feedback form for the coordinating centre.
Relatives were re-approached after three weeks and if they did not respond after three attempts it was assumed that they did not wish to participate. Relatives attending the clinic were provided with more detailed information about FH, the effect of cholesterol and lifestyle advice. Home visits were offered to families unable to get to the clinic or if several family members (particularly if children) were to be tested. No information on relatives other than first name, date of birth and relationship to the index case was stored electronically until they consented to being involved in the project. Relatives consenting to being tested were offered a point-of-care (finger-prick) test or asked to return for a fasting lipid profile. They were also asked to consent to having their information stored on an electronic database. Those electing for the point-of-care test were given a preliminary result, either ‘unlikely FH’ or ‘likely FH’ based on their LDL-C or total cholesterol concentrations. All relatives were given lifestyle advice. The ‘likely FH’ relatives were asked to return for a fasting lipid profile. Results of all the tests were sent to the GP who was asked to refer high-risk individuals to the clinic. Relatives diagnosed as FH were offered lifestyle advice and therapeutic options for the management of their hyperlipidaemia, as is the norm in these clinics. Newly diagnosed FH patients became index cases for further cascade testing.
Age- and gender-specific LDL-C diagnostic cut-off tables were used to assess risk in the relatives. These have previously been described
17
and are available on
Diagnostic total cholesterol* and LDL-cholesterol cut-offs (mmol/L) for first-degree relatives of familial hypercholesterolaemia patients
This table gives the boundaries of the grey zone, above the upper limit is the red zone and the individual is likely to have FH; below the lower limit is the green zone and the individual is unlikely to have FH
*Cut-offs for total cholesterol are indicated in bold
Costs
Data were collected on the time spent by the nurse on different activities, these were defined as the amount of time spent (h:min) on; making contact (letters and telephone calls); appointments at the clinic, home visits, general administration (ordering consumables, preparing packs for GPs, etc.), training and sick leave. For any activities falling outside these categories, the nurse was asked to record these as ‘other’ and define them (in a free text field). The costs of consumables and other expenses were collated for each site. The average cost of contacting and testing a relative was calculated for each site from the ‘total cost’ of the service development divided by the number of relatives tested at the site, plus any out-of-catchment relatives tested as a result of the project. The ‘total cost’ was defined as the cost of employing the nurse plus equipment, consumables and expenses for the service development period.
Time lines
A composite, average length of time taken to complete the cascade process for a family was calculated from the dates entered in the family tracing database. The dates used were: index case contacted, index case seen, relatives contacted, first relative seen, last relative seen. From these dates, the length of time was calculated for each index case from: (i) contacting an index case to seeing an index case; (ii) seeing the index case to contacting the relatives; (iii) contacting the relatives to seeing the first relative; (iv) seeing the first relative to seeing the last relative.
Results
An overview of the results of the family tracing is shown in Figure 1.
Overview of the family tracing results. FDR, First degree relative
Index cases
Characteristics of 733 FH patients and index cases for the cascade testing have been reported previously, 16 but briefly they had a mean age of 51 y, 45% were male and by the SB criteria 27% had a diagnosis of Definite FH, 55% of Possible FH and 18% were Not FH, with all groups having significantly elevated pretreatment mean total cholesterol ± standard deviation (SD) (10.0 ± 1.2 mmol/L, 9.0 ± 1.2 mmol/L and 9.0 ± 1.2 mmol/L, respectively). The first audit was limited to patients diagnosed before 31 December 2003, but for family tracing, this cohort was extended to include all FH patients known to the clinic. Of the 931 index cases contacted, 643 (69%) participated and 545 (59%) provided information for family tracing. The response rate was different by location and as shown in Table 2 was lowest in Manchester and Birmingham, where the participation rate was 56% (P < 0.01). Overall, as shown in Table 3, the response rate was not different according to the ethnicity or gender of the index case. Those aged 45 y or older were more likely (P = 0.001) to participate than those <45 y of age (Figure 2). Index cases discharged to GP care were less likely to attend than those under the care of the lipid clinic (56% vs. 76%, P < 0.001). The overall proportion of index cases discharged from the lipid clinics was 28%, but as shown in Table 2 this varied with location from 18% in Nottingham and Manchester to 41% in Bournemouth (P < 0.01).
Response from index cases by age. Significant difference in response by age <45 vs. >45 and P = 0.001
Key statistics by location
Characteristics of respondents
Relatives
On average, each index case had four living first-degree relatives (this did not differ significantly with location, Table 2) and on average 2.5 were contacted. Of the relatives, 35% were known to have been previously tested, and this proportion was varied with location, thus being 13% in Manchester and 51% in Birmingham (P < 0.01). A significant proportion of the relatives who resided outside the catchment area of the clinic attended by the index case, and so were not able to visit the nurse. The overall proportion was 34%, but as shown in Table 2 this varied from only 13% in Birmingham to 50% in Guildford (P < 0.01). The 504 out-of-catchment relatives were asked to go to their GP for a cholesterol test, and of these, results were received for 178 (35%).
Of the 990 first-degree relatives living in the catchment area, 21% reported that they had already been tested (this varied with clinic from 5% to 34%, P < 0.01, Table 2). The remaining 591 came forward for testing. Of the local relatives, 38% (of 952) of them responded to the first letter they received, 25% (of 587) to the second letter and 13% (of 442) to the third and final letter. The proportion of invited relatives who came forward for testing was not influenced by ethnicity or gender but was influenced by contact method and age (Table 3). A greater proportion of relatives came forward for testing when they were contacted by the nurse (Direct Contact), than when they were contacted by the index case (Family Contact) (73% vs. 62%, P = 0.006). A greater proportion of younger relatives lived in the catchment area and one-third of the relatives who came forward for testing for the first time were children ≤16 y of age (126/443) (Figure 3).
Proportion of relatives living in the catchment area of the clinic and coming forward for testing, by age
Overall, 41% of relatives <24 y of age had LDL-C concentrations suggestive of FH (red and grey zone), while a significantly smaller proportion (36%) of relatives >24 y of age had such concentrations. In part, this is because, a greater proportion of the previously tested relatives were older (average age 45 vs. 32). Testing at younger ages (under 24 y) gave an unambiguous result (red or green zone), in significantly more subjects than in older relatives, with fewer (10% vs. 16%, P = 0.007) having indeterminate lipid concentrations for which repeat testing was required (Figure 4).
LDL-cholesterol results on relatives: in (i) young people, under 24 y of age (n = 215) and (ii) adults, 24 y and over (n = 287). Red zone – likely FH; grey zone – an indeterminate result; green zone – unlikely to have FH
The total cost to employ a nurse, provide equipments and consumables for the family tracing period was divided by the total number of relatives tested to calculate an average cost to find and test a relative. It was approximately £500, with a range of £330–2470 between clinics (Table 2).
Table 4 summarizes the result of the enhanced service on the family information contained in the notes, showing the pre and post audit data. Most the of notes (55%; range 36–74%) now contained a pedigree, but most importantly, the number of relatives originally noted was 2.5/index case but had increased to 4.0 after the cascade testing, representing a considerable increase in the pool available for testing. Of these 62% had evidence in the notes of being tested and the results recorded, compared with 32% pre audit.
Summary of audit of family tracing preservice and postservice development
*Results for index cases participating in cascade testing
Discussion
Overall, the results demonstrate that cascade testing by a trained nurse based in a lipid clinic is an acceptable and efficient method to identify new FH patients in the UK. Of the invited index cases, 70% participated in the project, illustrating good acceptability for family tracing, although for some (10%) this was simply to report that all their family members had been tested. Patients discharged to GP care were less likely to return to the clinic to discuss family tracing (76% vs. 56%, P < 0.001) and we would recommend that FH patients remain under the care of a lipid clinic, at least until family tracing has been completed. The number of discharged patients varied with location, in some cases this was driven by financial constraints, with hypercholesterolaemia considered to be a condition that could be adequately addressed in primary care. In other places, the consultants work in partnership with their local GPs in the follow-up of FH patients. It was notable that older patients were more likely to come forward. Although the reasons for this were not explored further, it is possible that they have more time, being free of work or child care commitments, or are more likely to have experienced, or witnessed, the effects of CHD. We have no reason to suspect that ethnicity of the patients was a determinant of participation and, although the numbers of non-White British patients and relatives were relatively small (138), this was confirmed by the results. It has been reported that women are more likely to access programmes aimed at improving health, 18 but in this pilot study equal numbers of men and women index cases and relatives participated.
Cascade testing was undertaken from 545 patients, of whom only 26% had a clinical diagnosis of Definite FH. Originally, the project had intended to cascade only from Definite FH patients, but the initial audit had shown that the majority of FH patients attending the clinic were of Possible FH category, so the project remit was extended to include this group, while accepting that this would be likely to contribute to a reduction in the efficiency of the cascade testing.
If relatives are likely to respond, most do so after the first or second letter. The cost-benefit of sending a third letter is lower but is still valuable. Relatives were not phoned by the nurse as this was considered to be too intrusive. There are concerns about the ethically appropriate way to carry out cascade testing for FH. Traditionally, clinical geneticists have relied on asking probands to contact their at-risk relatives (family contact) to advise them that they should be screened. It has been argued, 19 however, that in the case of a treatable disorder such as FH, it is equally acceptable for a health-care worker to contact relatives on the proband's behalf (direct contact). In the UK it is not possible to contact relatives or obtain their names and addresses without the proband's consent. The completeness and accuracy of this information will vary, and will influence the efficiency of cascade testing.
Direct contact is likely to be more efficient as it does not rely on the index case to act on the information given by the nurse, it has the advantages that the information passed to relatives about FH will be more accurate, the nurse can follow-up any non-responders and the patient is not burdened with this responsibility. However, this has to be balanced against preserving confidentiality for patients and their relatives, the possible harm that an unsolicited letter may cause to a relative and the right of an individual ‘not to know’. Overall, it has been argued that in the case of FH, a treatable disorder, direct contact is appropriate and the evidence from this work and others 20 shows an increased uptake of testing. Care was taken in implementing this method; index cases were given a choice as to which method they preferred; if they choose direct contact, then wherever possible the index cases were asked to let their relatives know that they would be approached; the letter to the relative did not mention FH and no information on relatives was stored unless they consented to testing.
Overall, 35% of relatives tested at the pilot sites were diagnosed as FH, which is considerably lower than the predicted 50% for an autosomal co-dominant disorder. There are several contributing reasons for this. First, in order for an index case to receive a clinical diagnosis of Possible FH they must already have a diagnosed relative, thus reducing the number of possible newly diagnosed relatives by at least one. Secondly, this is also because a high proportion of relatives had already been tested, but cascading from patients with clinically diagnosed FH but who were not ‘true’ FH also contributed. There was however no marked difference in the proportion of relatives from ‘SB Definite’ and ‘SB Possible’ index cases that were diagnosed with FH (40% vs. 41%). Interestingly, the proportion of new cases in relatives from index cases categorized by the clinicians as FH or Definite FH was significantly higher than Possible FH index cases (52% vs. 36%, P = 0.004): here Possible FH equated to ‘it is possible that this person has FH but there isn't sufficient evidence to make a diagnosis’ while the clinical diagnosis was considered to be more certain for FH and Definite FH.
Finally however, the largest effect was likely because of the diagnostic ambiguity caused by the overlap in LDL-C concentrations between FH and non-FH relatives. 17 The organization managing the implementation of FH cascade testing using genetic diagnosis in the Netherlands (StOEH) provided the LDL-C concentrations of the relatives of mutation-positive probands. 10 These molecularly defined cohorts of mutation-positive and mutation-negative relatives demonstrated the expected high degree of overlap in LDL-C concentrations. 17 Applying a Bayesian model to the data to develop LDL-C cut-offs showed that 20% of the children of an FH index case (under the age of 14 y) were misdiagnosed (15% false-negatives and 5% false-positives), while 58% of the brothers and sisters of the patient (45–55 y) were misdiagnosed (42% false-negatives and 16% false-positives). As shown in Table 1, LDL-C cut-offs were developed to give age- and gender-specific concentrations for ‘Likely FH’ (red) and ‘Unlikely FH’ (green). The charts also incorporated an ‘indeterminate’ zone (grey), where subjects required monitoring to determine a clinical diagnosis. This has led us to conclude that DNA testing would be required to underpin cascade testing for FH. 8
Of the 591 tested relatives, 244 (41%) had LDL-C concentrations in the ‘red’ or ‘grey’ zone indicative of FH, but proportions falling into the red, grey and green zones in younger compared with older relatives was significantly different (P = 0.007). In the younger age group, there were more red zone diagnoses – 31% vs. 20%, and fewer in the grey zone – 10% vs. 16%, resulting in a clearer diagnostic picture at younger ages and a reduction in the numbers requiring re-tests or long-term monitoring. People with LDL-C concentrations unlikely to indicate FH (below the cut-off) were advised to have their cholesterol concentrations monitored every 2–5 y.
Because of the uncertainty of a diagnosis of FH using LDL-C measures, the more relevant figure of the costs incurred in finding and testing one relative was calculated, rather than the costs to identify a new FH patient. The average cost to identify, contact and test one relative, using lipid measures, was £500 (range across clinics £300–£2450). The costs of finding and testing one relative were much higher in the two metropolitan locations (Birmingham and Manchester) averaging at £2200 (compared with an average of £350 at the other three sites). While the following factors are not necessarily unique to these two sites, in combination they may have influenced the efficiency of cascade testing. The multisite locations involved spending considerable time travelling between sites (this was exacerbated by a shortage of parking at hospitals in these metropolitan locations), nurses were required to liaise with several consultants and to gain an understanding of how each of the different trusts operated. Due to the high cost of living, patients and relatives living in these areas are more likely to be employed and therefore reluctant to attend appointments, as getting time off work can be difficult and travelling to clinics is very time-consuming. It is possible that the ‘North/South divide’ in access to healthcare played a role; 21 however, the results from Nottingham, where costs were the lowest, would not support this. At the two metropolitan sites, the nurses used reply slips in their letters to the index cases, and sent letters out in ‘mass mailings’, which reduced the opportunity for timely (within two weeks) follow-up telephone calls, and many may have refused without fully understanding the reasons for cascade testing. It is possible that index cases are less likely to reply if they know that all their relatives have been contacted and tested. In Birmingham there was evidence that earlier cascade testing had resulted in over half of the relatives having been previously tested, and while the same evidence was not apparent in the Manchester data, this is an area where family tracing has previously been implemented. 13 However, the difference should not be underestimated, as if it is a reflection of the deprivation indices seen in less prosperous areas of the country, it would have a profound effect on cascade testing.
Many of these concerns would be addressed if the nurse worked alongside the consultant in ‘FH Clinics’, as family tracing would be part of standard clinical care rather than a separate project. There are many other factors that may have played a part in the costs, including the infrastructure at the NHS trusts, the level of support from consultant(s) and the level of experience, and commitment of the nurse. With our recommendations in place it is likely that costs would be lower, and be similar to the £400 estimated in The Netherlands, where a national system based on DNA testing operates. This programme has been shown to be highly cost-effective at an estimated US $8700 per life-year gained (discounted). 22 This is well below the value of £20,000 per quality-adjusted life year (QALY) used by the National Institute for Health and Clinical Excellence (NICE) as a benchmark for cost-effectiveness and is comparable with the estimates for the cost-effectiveness of genetic testing for breast cancer. 23
Not surprisingly, the availability of the FH nurse and carrying out of the cascade testing had a significant impact in the second audit note with regards to the extent of family and relative information. Since a pedigree is a vital part of the accurate diagnosis of FH such information should be contained in essentially all patient notes. Initially, only 35% of notes contained a pedigree and this had increased to 55% overall and 90% for index patients who replied to the invitation to be involved in family tracing. The cascade process also resulted in a significant, 1.6-fold increase in the number of relatives noted per index case, representing a considerable increase in the number available for testing. Of these, 62% had evidence in the notes of being tested and the results recorded. The fact that this number is not higher is partly because some relatives did not come forward for testing, but also reflects the difficulties in making contact with more distant relatives, with 60% of those within the catchment being tested and obtaining feedback on 35% of those unable to visit the FH nurse. Follow-up of relatives living outside the catchment area of the pilot sites was anticipated to be difficult and even if GPs test relatives, the cascade process breaks down as the first-degree relatives of the new FH cases are not followed up. However, we were not able to ascertain if relatives did not go to their GPs, or if GPs were reluctant to test relatives, or if tests were done but the results not returned to the coordinating centre. Anecdotally, we have reason to believe that GPs were reluctant to test young people because of concerns about the patients' subsequent ability to obtain life insurance, although the evidence is that this will not be a problem if the patients' LDL concentrations are reduced with treatment. 24
Several issues were identified during the cascade testing project that would influence the efficiency of the process when rolled-out UK-wide. The average time taken from the nurse contacting the index case to seeing the last relative was over six months (Table 2). This included an average of over two months taken from contacting an index case to seeing them, time that would not be necessary if the nurse was permanently working alongside the consultant in the lipid clinic, and therefore has the opportunity to see FH patients at the time of their usual appointment. Currently, lipid clinics receive patients with a variety of disorders, with perhaps just one or two FH patients in one session. FH nurses would be used more effectively if sessions were organized so that the nurse could see a number of FH patients in one clinic. This would allow the nurse to cover a number of trusts within a week. Providing additional administrative support for the nurses would allow them to concentrate on seeing patients and relatives, reduce the time taken (lapsed time) to send the letters out to relatives and to track any non-responders.
The implementation of cascade testing requires an infrastructure that allows for both family tracing and the increasing clinical workload as new cases are identified. A survey of lipid clinics in the UK 3 has highlighted a patchy provision of services, particularly access to paediatricians, hence additional investment will be required to coordinate family tracing across NHS Trusts, Primary Care Trusts (PCTs) and Strategic Health Authorities (SHAs), and to provide appropriate services for children and adolescents.
Using the current techniques, the causative mutation can be detected in 70–90% of patients with definite FH, 25,26 which compares well with the 25–30% detection rate in the BRCA1/2 genes in patients with familial breast cancer. 27 Currently, in the UK, DNA testing is only undertaken in a few centres, 28 mainly as a result of research projects. 25,26,29 Implementing DNA testing would ensure that resources for family tracing are concentrated where evidence of the monogenic disorder is definitive, and provide an accurate method to identify affected relatives.
NICE guidance on the identification and management of FH is due to be published in August 2008, the draft guidelines 30 support the recommendations we have prepared here for the identification of family members of known cases, and we have no reason to believe that this will be changed significantly in the final document. Finding the resources to implement the guidance is unlikely to be straightforward and may require that funding is provided from several sources, e.g. government bodies and charities.
Footnotes
Acknowledgements
The UK FH Cascade Audit Project is supported by the Department of Health and was overseen by the Steering Group: Dr Philip Adams, Di Asplin, Dr Cyril Chapman, Prof Paul Durrington, Tessa Ing, Michael Livingston, Prof Theresa Marteau, Prof Michael Modell, Prof Andrew Neil, Diana Paine, Dr Alan Rees, Dr Simon Sanderson, Allison Townend, Irena Wasilewska, Prof Peter Weissberg and Dr Ron Zimmern. SEH acknowledges support from the UK Departments of Health and of Trade and Industry for the IDEAS Genetics Knowledge Park, and from the British Heart Foundation (grants PG2005/014). We would like to thank nurses Gretta Wood, Mabella Farrer, Ruth Eatough and Fran Lloyd for their input to the project.