Saving Lives, Saving Families: The health, social and economic advantages of detecting and treating familial hypercholesterolaemia (FH)

Transcription

1 1 Saving Lives, Saving Families: The health, social and economic advantages of detecting and treating familial hypercholesterolaemia (FH) Economics Chapter: Estimating the benefits from treatment and increasing the implementation of cascading screening This part of the HEART UK FH report was commissioned by HEART UK and produced by Leo M Nherera. This document is the full version of the economics chapter of the FH Report. Contact details mashizha@yahoo.co.uk or lnherera@bmjgroup.com Phone: AstraZeneca, Genzyme Therapeutics Limited and MSD (the sponsors) were asked by HEART UK to provide funding for the commission and production of this report. The sponsors have reviewed the content for compliance purposes only and are not responsible for data verification. The authors have had editorial control of the materials and the content of the material may not necessarily reflect the views of the sponsors. 1

2 Glossary of terms Term Definition Costeffectiveness measured using a single outcome, usually in natural units (for example, life-years An economic study design in which consequences of different interventions are analysis gained, deaths avoided, heart attacks avoided, or cases detected). Alternative interventions are then compared in terms of cost per unit of effectiveness. Discounting Costs and benefits incurred today are usually valued more highly than costs and benefits occurring in the future. Discounting health benefits reflects society s preference for benefits to be experienced in the present rather than the future. Discounting costs reflects society s preference for costs to be experienced in the future rather than the present. Dominance An intervention is dominated if it has higher costs and worse outcomes than an alternative intervention. Incremental The ratio of the difference in the mean costs of a technology compared with the next costeffectiveness best alternative to the differences in the mean outcomes. ratio (ICER) Life-years Average years of life gained per person as a result of the intervention. gained (LYG) Qualityadjusted life this time. QALYs have the advantage of incorporating changes in both quantity An index of survival that is adjusted to account for the patient s quality of life during year (QALY) (longevity/mortality) and quality (morbidity, psychological, functional, social, and other factors) of life. Used to measure benefits in cost utility analysis. Sensitivity analysis A way of representing uncertainty in the results of economic evaluations. Uncertainty may arise from missing data, imprecise estimates or methodological controversy. Sensitivity analysis also allows for exploring the generalisability of results to other settings. The analysis is repeated using different assumptions to examine the effect on the results. Time horizon The time span used in the NICE appraisal that reflects the period over which the main differences in health effects and use of healthcare resources between interventions are expected to be experienced. Utility A measure of the strength of a person s preference for a specific health state in relation to alternative health states. The utility scale assigns numerical values on a scale from 0 (death) to 1 (optimal or perfect health). Health states can be considered worse than death and thus have a negative value. 2

3 3 Executive summary The NICE FH Guideline (CG71) 1 recommends that people with FH should be treated with high intensity statins to prevent early deaths from coronary heart disease (CHD). Currently, it is estimated that 15% of the expected 120,000 UK FH patients have been identified and treated in primary and secondary care. If the family cascade testing strategy recommended by the NICE Guideline is implemented this number is expected to increase to 50%. In this analysis we assumed for every index case there will be five relatives with FH. Thus, out of the 120,000 FH cases in UK, 24,000 are assumed to be index cases and 96,000 are the relatives. In addition to implementing this cascade screening, there is an urgent need for an FH national register, to facilitate follow-up of relatives of affected persons. We conducted a cost-utility analysis to quantify the benefits of optimal, high intensity statin treatment or sub-optimal statin treatment compared to no treatment. High intensity statin treatment or aggressive lipid modifying therapy may be defined as any lipid lowering drug or combination which is more potent than simvastatin 40mg, including atorvastatin (20mg or greater) or rosuvastatin (10mg or greater), either alone or in combination with bile acid sequestrants, ezetimibe, nicotinic acid or fibric acid derivative. All of the above drugs and combinations are recommended by NICE CG71 for use as required if optimal LDL cholesterol lowering of greater than 50% from baseline is not achieved. For the purpose of this analysis we used the example of high intensity statins for which long-term outcome data were available however other drug treatment regimes achieving similar LDL-C reductions should be considered equivalent. We show the benefits of treatment in terms of QALY (Quality Adjusted Life Year) gains and cardiovascular (CV) events avoided. We are also able to show the benefits of cascade screening and associated savings from various levels of implementation. People with FH will incur costs to the NHS over their lifetime, whether they are treated with statins or not treated. These costs include the cost of the intervention; in this case statins, the cost of CV events such as treating MI (myocardial infarction), strokes, revascularisations, and heart failure. The costs of health care professionals such as GPs and specialists are also included. More effective treatments will reduce the incidence of CV events and in the long run saves money to the NHS, compared to less effective interventions. When comparing interventions - in this case, high intensity statins with no 3

4 treatment for instance - high intensity statins result in less lifetime costs overall when all cost have been accounted for. High intensity statins also result in better health outcomes such as less CV events and mortality (which will translate to more life years gained and QALYs). This is said to dominate the comparator. Sometimes high intensity statins will result in overall more costs compared with the comparator, but more health benefits. This is the case with high intensity statins when compared to low intensity statins, and a judgment then has to be made whether the additional benefits are worth paying for. The NHS is usually willing to pay 20,000-30,000 for an additional unity of health benefit measured in QALYs. The lifetime treatment costs and health benefits have been estimated in line with NICE guidance, which states that both costs and benefits should be discounted at 3.5% and that health benefits should be estimated in QALYs 3. As mentioned earlier, the overall costs include the cost of drugs and CV events avoided over a lifetime. The lifetime costs for the at-risk relatives of FH index cases who are followed from the age of 30 to 85 years and are not offered statin treatment was estimated to be 18,008. For those people offered low intensity (sub-optimal) statin treatment the lifetime cost was estimated to be 17,664, and for those offered high intensity optimal statin treatment the lifetime cost is 17,959. The estimated lifetime health benefits in QALY gains from treatment was estimated to be for those not offered statins, while those offered sub-optimal and optimal statin treatment gained and respectively. People who are not offered statin treatment still gain some QALYs. This is because they will be eligible for other treatments such as antihypertensive medication and aspirin, except that in this instance they are assumed not to receive statins. The analysis shows that both optimal and sub-optimal statin treatment results in greater health benefits at lower overall cost when compared to no treatment. Statin treatment dominates no treatment and therefore, statin treatment is preferable to no treatment. Furthermore, the results show that optimal statin treatment with high intensity statins is more beneficial than sub-optimal treatment, resulting in more costs and greater health benefits. The additional cost and benefits yields an estimated incremental cost-effectiveness ratio of about 400/QALY, which is highly cost-effective. 4

5 5 Our analysis was able to demonstrate how the benefits in QALY translate into CV events avoided. For instance, we have demonstrated that by offering affected relatives optimal statin treatment, 101 CV deaths per 1,000 FH patients treated will be avoided compared to no treatment. A further 69 additional CV deaths will be avoided per 1,000 FH patients treated if 1,000 people affected with FH are offered optimal treatment rather than sub-optimal treatment. We also calculated the cost of screening or diagnosis. Our analysis demonstrated that the cost of diagnosis per patient using the method of LDL-cholesterol measurement only is 202 for the index cases and 116 for relatives. For DNA diagnosis the cost of for each index case is 125 and 75 for each relative 4. Currently, the cost of DNA testing in England varies and may be higher than that reported in this analysis. 1 Lower costs are available through laboratories using the latest technology (such as in Northern Ireland). Using the DNA costs obtained in Northern Ireland the overall cost per patient for cascading relatives using the recommended strategy (combined LDL-cholesterol and DNA based diagnosis) is 160 per relative when 1,000 index cases are screened. Greater benefits and savings for the NHS will accrue if all eligible FH patients are identified and treated. We have shown that using cascade screening at every level of implementation will yield savings in the form of CV events avoided, and that this translates into QALYs and LYG gained. Whilst it remains aspirational to achieve 100% identification of the predicted UK FH cases, there is considerable room to improve from the current estimated 15% of cases that have been identified, and we believe a realistic goal is that 50% of the UK FH patients could be found within five years by the cascade testing strategies recommended by NICE. This analysis has demonstrated that, even if cascade testing were carried out only from the 15% of eligible FH patients currently identified, the CV events avoided would result in savings of 591,647 per 1,000 screened index cases, if identified relatives are offered optimal treatment. This would translate to 8.5 million nationally for the UK. If, more realistically, 50% of the 96,000 predicted relatives of UK FH index cases were identified by the recommended cascade screening methods, then 1.97 million per 1,000 screened index cases could be saved with optimal treatment. This would translate to 94.7 million nationally for the UK over a 55 year period, or roughly 1.7m per year. By not fully implementing cascade screening as recommended in the NICE guideline (identifying 50% of potential relative s cases) the NHS is losing approximately 1.4m per year. 1 Personal communication 5

6 To achieve maximum benefit from currently recommended family cascade screening methods, a national disease register for FH will be essential to facilitate efficient use of resources. This does not currently exist in the UK. This analysis has shown that the costs of implementing a national FH register are very small when compared to the expected benefits of identifying relatives of affected index cases. We have estimated that the FH register will cost the whole of the NHS 149,275 per year, or 101,000 if this were used to fund two staff members or persons per PCT in England respectively to operate the register at the PCT level. Funding two people per PCT will currently cost about per person identified per year and would fall to 3.11 per person per year if screening identifies up to 50% of eligible patients. If the PCT decides to fund one person per year, the costs will be 6.10 and 1.83 per person in the database if 15% or 50% of patients are identified respectively. 6

7 7 Methods, modelling benefits of statin treatment A previous economic analysis using the Markov model 5,6 was used to estimate the benefits of treatment with statins, (high and low intensity). An additional analysis was done to estimate benefits of no treatment. The high intensity statin regime used in the analysis was atorvastatin 80mg daily, in comparison with a low intensity statin (simvastatin 40mg daily) and a placebo (considered equivalent to no treatment). Health outcomes were reported as quality adjusted life years (QALYs), life years gained (LYG) and cardiovascular events avoided. The model was populated with a hypothetical cohort of 1000 people with FH. The age at diagnosis was included as a variable. Costs and QALYs were calculated until patients either died or reached the age of 85 years. Model structure and assumptions The rate at which people progress through the model was regulated by transition probabilities, which describe the likelihood of moving between health states over each model cycle. For patients on high intensity and lower intensity statins the transition probabilities were adjusted to reflect the expected reduction in cardiovascular events from the clinical trial data. All patients are assumed to have no CVD at the start of the model. However during each annual cycle, there is a risk that they could experience a coronary event (myocardial infarction, unstable angina, or coronary death), a cerebrovascular event (stroke, or cerebrovascular death) or that they could die from other causes. Individuals can experience more than one event in subsequent cycles, but the model only applies the risks, costs and health effects of the most recent event (the Markov assumption). The probability of a person experiencing the different events depends on their age, CVD risk, current health state and treatment. Baseline risks were taken from a NICE Health Technology Assessment (HTA) on statins for the prevention of coronary events, 7 which shows the incidence of CHD in the general population but, for age groups younger than 60 years, this gives a much lower level of risk than that experienced by people with FH. We therefore applied the age-adjusted relative risk of cardiovascular disease reported from the Simon Broome Register, which is a long term follow-up study of a cohort of people with FH 8. Thus for the age group the risk of developing cardiovascular disease was increased by a factor of 84.3, for those aged a factor of 5.76 was used and those over 60 a factor of 1.2. For the base case model we considered patients diagnosed between the age of years (average age 50 years), for relatives we modeled an average of 30 years. The risk of stroke was assumed to be the same as seen in the general population, because data from the Simon Broome Register indicated that these risks were not significantly higher in the FH population 8. Death from other causes was assumed to be the same as that the general population and 7

8 was taken from the life tables of England and Wales 9. The model assumes that risk of cardiovascular disease (CVD) increases with age for both males and females. The rate of increase used in the model was 0.02%, which is the average between males and females of the rate of increase previously reported by Ward et al Statins were categorized as high intensity if they produce greater LDL-cholesterol reductions than simvastatin 40mg 6 ; these high intensity preparations included once daily simvastatin 80mg, atorvastatin 40mg to 80mg and rosuvastatin 20mg to 40mg. All other statins were considered to be low intensity. Atorvastatin 80mg was the only high intensity statin for which clinical outcome data was available in a patient group with a comparable prognosis to patients with FH, and was therefore chosen to use as the high intensity preparation in the model. The low intensity statin used was simvastatin 40mg because this was considered the appropriate initial treatment in the non FH population at greater than 20% 10 year risk of a cardiovascular event. Outcomes and treatment effects The outcomes included in the model were death from cardiovascular disease, death from other causes, MI, unstable angina, revascularisation, heart failure and stroke. For the main analysis we assumed that the benefit derived from statins by FH patients did not differ from that experienced by patients using statins after myocardial infarction or with acute coronary syndrome for the high versus low dose comparisons. We thus performed a meta-analysis of the four trials that compared high intensity statins with low intensity statins after myocardial infarction 10,11,12,13. For treatment comparisons against placebo, we used data reported in the HTA by Ward et al. 7. Table 1, Treatment effect used in the model, optimal statin treatment (high doses) versus low suboptimal statin treatment (low doses) in CHD secondary prevention Outcome Point estimate Lower 95% CI Upper 95% CI MI stroke Heart Failure Rev Unstable angina CVD death Death other Source: Nherera 2011, meta-analysis of high intensity vs. low intensity statin trials Table 2, Treatment effect used in the model, optimal statin treatment versus no statin treatment in CHD secondary prevention 8

9 9 Outcome Point estimate Lower 95% CI Upper 95% CI MI stroke Heart Failure* Rev Unstable angina CVD death Death other Source: Ward et al 2007 *Assumed to be same as the effect seen in high versus low doses Cost data Estimates of cardiovascular events costs and the sources of the data are shown in table 3 Costs were at prices. In accordance with NICE Methods guidance 10 an annual discount rate of 3.5% was used for both costs and health benefits. Drug costs were taken from the BNF September 2011 no. 62, 14 while those of cardiovascular events were taken from the NHS reference cost 2009/ It is anticipated that Atorvastatin will come off patent before the end of 2011; hence cost of the drug has been revised in anticipation of this. Thus, the total costs of interventions included both the direct cost of drug treatment and also the potential savings from treatments avoided due to reduced incidence of cardiovascular disease. Health related quality of life The value of health outcomes was estimated using a QALY approach, with estimates of survival and health related quality of life associated with each health state included in the model. Quality of life data were obtained from a previously published study 5 see table 3 and was adjusted to reflect the decreases in health-related quality of life with age by multiplying the disease utility weight by age utility weight, derived from the 1996 Health Survey for England 16. There is no evidence of an excess of adverse effects from statin trials 17,18,19,20 although observational evidence points to some unpleasant or potentially dangerous effects 21. The model therefore assumed that patients who experience an adverse effect would discontinue treatment, with no significant cost or lasting health impact and there is no difference in treatment related quality of life between high intensity statins and low intensity statins or placebo. 9

10 Table 3, Health state/event costs and health state utilities used in the model Health state Cost used in the model Health state utility multipliers No event MI (first year) 1, MI (subsequent) Stroke (first year) 10, Stroke (subsequent) 2, Heart failure 1, Heart failure (subsequent) Revascularisation 10, Revascularisation (subsequent) Unstable angina (first year) 1, Unstable angina (subsequent) CVD Death other Source: Nherera 2011 Table 4, Age specific quality of life estimates Age group Mean upto 35 years Source, Health survey of England 1996 Methods for estimating the numbers screened by different cascading methods The method described below is adopted from a recent publication 5. We assumed that each index case has five first degree relatives available for testing 22 and each of these five has two first degree relatives (i.e. second degree relatives of the index case), and each of these has two first degree relatives (i.e. third degree relatives of the index case). We also assumed that 65% of the first degree relatives and 60% of the second degree relatives will agree to be tested. These are high estimates for take up in population screening, but are based on data from the UK FH Cascade Audit Project FHCAP study 22 where these values were 85% and 80% respectively. Finally, to simplify the model we assumed that 50% of the tested relatives would be the children of the probands and would be in 10

11 11 the age range and the remainder would be the siblings of the probands, with an age range of years. We assumed that a monogenic cause underlies a diagnosis of FH, that is, a true FH patient is someone with a relevant gene mutation. To estimate the proportion of true FH patients in the hypothetical 1000 patients referred from general practice, we first used data from Taylor et al 23 in which 30% of the patients currently being treated in lipid clinics had clinically Definite Familial Hypercholesterolaemia (DFH) and 62% had clinically Possible Familial Hypercholesterolaemia (PFH) according to the Simon Broome criteria, while 8% failed to meet the criteria for either DFH or PFH. This last group was designated 75% of them were classified as true negatives and not considered for cascade testing while 25% are assumed to have a mutation positive. Based on reported mutation detection rates, we then assumed that 90% of those identified as DFH had a relevant gene mutation (were true FH), as well as 35% of those identified as PFH; meaning that 51% of our hypothetical cohort would be true FH patients ((0.9 x 0.3) + (0.35 x 0.62)) + (0.08*0.25). To estimate the proportion of FH and non-fh relatives that would be identified from true FH index cases with LDL-cholesterol based diagnosis and cascade screening (Strategy 1) we used data from the UK and the Netherlands 22, which gave an age-averaged estimate of 32% true positives (i.e. had FH and were identified), 8% false positives (did not have FH but were identified as having FH), 42% true negatives, (did not have FH and were not identified as FH) and 18% false negatives (had FH but were not identified as FH). From false-positive index cases cascade testing will identify no true-positive relatives, but a proportion will be identified as affected (i.e. false-positives) because they have LDL-C levels above the diagnostic cut-offs. Conversely, there will be no cascade screening from false negatives, so some true FH relatives will not be identified. For the DNA based diagnosis and cascade testing strategy (Strategy 2), the mutation detection rate was taken to be 80% in the DFH group and 30% in the PFH group 24. Cascade testing only takes place from mutation-positive index cases and results in a 50% detection rate (since FH is a monogenic autosomal co-dominant disorder). However, since current mutation detection methods are not 100% sensitive, a proportion of the mutation-negative index cases will be false negatives. For the DFH group we assumed that this would be true of half of the 20% negatives in the DFH group, meaning that overall = 90% of the DFH patients are true positives. For the PFH cases, it was assumed that a similar proportion of mutations would not be detected as in the DFH group (i.e. for every 8 mutations detected in the PFH group one would be missed so the false negative rate in the PFH patients with no detected mutation = 30% x = 3.8 mutations per 100 patients), and an upper estimate of 7% of the PFH mutation-negative index cases as false negatives was used. 11

12 For the Strategy 3 DNA based diagnosis and cascade testing is used for mutation positive index cases, but cascade testing is additionally undertaken using cholesterol (LDL-C) diagnostic cut-offs in the 20% of DFH patients in whom no mutation has been detected. The proportions of true- and falsepositive diagnoses from this mutation negative group were estimated as in the LDL-cholesterol method. Finally, for Strategy 4 the DNA based diagnosis and cascade testing is used for mutation positive index cases but additionally cascade testing is undertaken using cholesterol (LDL-C) diagnostic cut-offs in both DFH and PFH mutation negative index cases. Costs of cascade screening The FH guideline CG 71 1 established that a FH clinic would need the following personnel: a consultant and a FH specialist dietician. Other personnel needed are assumed to be already available within the hospital setting such as dietician, administrative clerical and phlebotomy support. The table below shows unit costs and estimated time needed for both index and relatives to estimate the overall cost of LDL-cholesterol screening per patient. All costs were taken from the standard Unit Costs of Health and Social Care 27. Estimates of time taken by each health care professional and costs of full fasting and non-fasting lipid profiles were taken from literature1. Cost of DNA testing was 125 and 75 for probands and relatives respectively were taken from the business case report of West Midlands 4. Table 5, Cost of running a clinic Health care Unit Time Time Cost Cost Source professional cost/hr Hrs(IC) Hrs(Rel) Index Relatives Specialist PSSRU 27 dietician/nurse Clerk ibid Phlebotomy ibid Consultant ibid Non-fasting TC Nherera Full, fasting TC ibid letter for ibid relatives Total Calculated Genetic testing Bradshaw

13 13 Results Familial hypercholesterolemia is a chronic disease with the risk of events over a prolonged period of time. As the potential events are associated with costs and health outcomes, we constructed a Markov model to estimate the long-term costs and health outcomes (numbers of cardiovascular events that will be observed) in FH patients who are a) Not treated with statins (because they are not identified) b) Sub-optimally treated with generic/low intensity statins c) Optimally treated with high intensity statins It is important to point out that no treatment implies that people are not offered statins in this analysis; they may be offered treatments other than statins. The estimation of benefit from statin treatment was necessary in order to show what is gained by treating FH patients who are eligible for statin treatment and conversely what untreated (because they are not identified) FH patients will lose in terms of health benefits and the subsequent costs to the NHS. Our analysis showed that if relatives of FH index cases are identified and treated optimally they will gain discounted 2 and undiscounted QALYs over their lifetime. If relatives are treated sub-optimally they will gain discounted and undiscounted QALYs over their life time. If relatives are not screened and identified, they will not be treated with statins, however, they will still accrue some benefits from other interventions other than statins and we estimate that they will gain discounted and undiscounted QALYs respectively. The QALY is estimated by weighting the time spent in each health state by the health-related quality of life (utilities) for that health state obtained from the literature. The lifetime QALY gain from statin treatment is comparable with treatment gain from other therapy areas. For instance, an economic analysis of screening for pre-diabetes among overweight and obese adults in the US by Hoerger et al estimated treatment benefit in QALYs to be 9.01 for screening compared to 8.9 for no screening 28 Another economic analysis by Freeman et al estimated the QALY 2 Discounting: Costs and benefits incurred today are usually valued more highly than costs and benefits occurring in the future. Discounting health benefits reflects society s preference for benefits to be experienced in the present rather than the future. 13

14 gains of high dose dabigatran to be QALYs and that of warfarin to be over a 30 year time horizon. 29 If health related quality of life is not considered, i.e. combination of a person s physical, mental and social wellbeing; not merely the absence of disease, the health outcome of the economic model will be measured in terms of life year gained i.e. average years of life gained per person as a result of the intervention. For those patients that are offered optimal statin treatment they will gain discounted life years and undiscounted life years, while those offered sub-optimal treatment will gain discounted life years and undiscounted life years. We estimate that those not offered statins will gain discounted and undiscounted life years. Healthcare costs are estimated by weighting the time spent in each health state by the expected treatment costs for that health state. If patients were offered statins we then add the annual cost of statins. Overall, cost of treatment for the relatives will be 17,959 with optimal statin treatment. These costs as mentioned earlier consist of the cost of events such MI, strokes, revascularizations and then annual cost of statins. The lifetime costs for sub-optimal statin treatment are estimated to be 17,664. Lastly the costs of no treatment are estimated to be 18,015 per person; these costs exclude the cost of statins since we assume they are not treated with statins. Statin treatment either optimally or suboptimally is a dominant strategy compared to no treatment i.e. it results in more health benefits at less costs overall. This is said to be cost-saving. Table 6, lifetime treatment costs and benefits (QALY and LYG) for relatives who are followed for 55 years (from 30 years to 85 years) Discounted Undiscounted Intervention Cost QALYs LYG Cost QALYs LYG No treatment 18, , Low intensity treatment 17, , High intensity treatment 17, , The observed QALY and LYG gained in table 6 above will translate to the following modeled cardiovascular events avoided when no treatment is compared to both sub-optimal and optimal treatment. We have also shown that events can also be avoided by treating FH patients optimally by reducing baseline LDL-C by more than 50%, with recommended high intensity statins as recommended in the NICE FH Guideline. Table 7 and Figure 1 visually illustrate the observed events avoided below. 14

15 15 Table 7, Base case relative s lifetime events avoided per 1000 treated patients (Followed up from 30 years to 85 years) Events Events avoided, suboptimal treatment with low dose vs. No treatment Events avoided, optimal treatment with high dose vs. No treatment Events avoided, optimal treatment with high dose vs. suboptimal treatment with low dose MI Stroke Heart failure Revascularisation Unstable angina CV death Other death* Total alive** MI: Myocardial Infarction CV death: Cardiovascular death *Negative numbers for other deaths are results of less people dying of cardiovascular deaths. Statins have no effect on other causes of deaths - for example, deaths due to car accidents or suicide ** Total people alive is a positive outcome, the more people alive, the better the statin treatment will be. 15

16 Lifetime events per for relatives followed upto age 85 years Figure 1, The lifetime observed events avoided for 1000 FH relatives assuming they are optimally or sub-optimally treated with either high or low doses of statins or they are not cascaded (hence not treated) No treatment Sub-optimal treatment with low dose Optimal treatment with high dose Cardiovascular events including mortality Table 7 and Figure 1 show that there are more benefits to be derived from treating FH patients optimally with high intensity statins to achieve a reduction in LDL-C of over 50% from baseline, compared to no treatment or sub-optimally with low-intensity statins. Less CV events including CV death - will be observed in those patients offered high intensity statins, compared to no treatment. The figure also shows that people treated optimally can expect to live longer and are more likely to die of non-cardiovascular causes. Furthermore, it is clear that by treating FH people even with sub-optimal doses of statins, 20 lives will be saved per 1,000 cases, and if FH people are treated with optimal statin doses, 69 lives will be saved per 1,000 people with FH. Table 8, Estimated lifetime cost savings from avoided cardiovascular events when true FH relatives of 1000 FH index cases are offered treatment 16

17 17 Events Cost savings, suboptimal treatment with low dose vs. No treatment Costs savings, optimal treatment with high dose vs. No treatment **Costs savings, optimal treatment with high dose vs. sub-optimal treatment with low dose MI 160, ,937 79,633 Stroke 2,087,066 2,828, ,493 Heart failure 247, ,083 7,179 Revascularisation 1,304, , ,547 Unstable angina 63,550 46,910-16,641 Cardiovascular death 149, , ,467 Death other causes - 2,828-7,861-5,033 Total lifetime savings per 3,820,657 3,944,312-35, treated people ** There are no cost savings between optimal treatments compared to sub-optimal treatment as shown by a negative cost. However, the additional payment on optimal treatment has been shown to be worth paying for. We have estimated the overall savings from treating 1,000 people with FH over a lifetime. We multiplied the events avoided by the cost of treating each event. For those who are alive, we calculated the cost of treating CVD to be a weighted cost of the CVD events included in the model, weighted by the distribution of these events at baseline without treatment. The cost of treating a CVD event was estimated to be 1,500 including treatment. Overall, treating 1,000 relatives of FH cases will save 3,820,657 when people are offered sub-optimal low intensity statin treatment and 3,944,312 when they are offered high intensity statin treatment If we consider all relatives of FH index cases in the UK (96,000), the total savings will be estimated to be almost 378 million or 6.9 million per year (96,000 multiplied by the cost-saving per patient of about 3,944).This estimate assumes all FH people are identified and offered optimal treatment. Modeling cascade screening outputs using different methods Testing the families of known cases of FH (cascade testing) can identify those with FH. In brief we evaluated four cascade screening strategies and the four strategies are described below. FH diagnosis for the index cases is made using the Simon Broome criteria. The cases are then classified as Definite Familial Hypercholesterolemia (DFH), Possible Familial Hypercholesterolemia (PFH) or No Familial Hypercholesterolemia (NFH). Relatives are cascaded from the DFH and PFH index cases while family cascade testing is not undertaken in NFH cases. The four cascading screening methods are Strategy 1, LDL-cholesterol Based diagnosis and cascade testing: Relatives will be cascaded from DFH and PFH by LDL-cholesterol testing only Strategy 2, DNA Based diagnosis and family cascade testing: 17

18 Relatives will be cascaded from DFH and PFH for those mutation positive (Mut+ve) by DNA testing Strategy 3, DNA diagnosis and combined cascade testing strategy for DFH: Strategy 2 with additional LDL-cholesterol based family cascade testing in DFH mutation negative (DFH Mut-ve). Relatives will be cascaded from DFH and PFH mutation positive (Mut+ve) by DNA testing while those who have DFH mutation negative (Mut-ve) will be cascaded using LDLcholesterol testing Strategy 4, DNA diagnosis and combined cascade testing strategy for DFH and PFH: Strategy 3 with additional LDL-cholesterol based family cascade testing in PFH mutation negative (DFH Mut-ve). Relatives will be cascaded from DFH and PFH mutation positive (Mut+ve) by DNA testing while all DFH and PFH mutation negative (Mut-ve) will be cascaded using LDL-cholesterol testing. The cost-effectiveness of different cascading methods has been established and DNA based method strategy 4, has been found to be cost-effective 1,5. It has also been established that any cascading method is better than not cascading at all 5. Cascade testing from 1000 suspected FH index patients will result in people being classified as true or false positives and true of false negatives. Testing for relatives will be performed from DFH and PFH and if an index case is deemed true negatives (TN), it means they do not have FH and no cascading will be performed in these patients. The table below shows the numbers of index cases who are true positives (TP), false positives (FP), false negatives (FN) and true negatives (TN) according to the assumptions outlined in the methods section from 1000 suspected index cases. Cascading will only be undertaken from those with definite FH (true positives-tp), those incorrectly identified as having FH (false positives-fp) and those incorrectly identified as having no FH when indeed they have FH (false negatives-fn). Table 9, Total numbers of index cases identified as TP, FP, FN and TN from 1000 suspected FH cases. Index cases Strategy 1 Strategy 2 Strategy 3 Strategy 4 True positives FH False positives - not FH cascaded False negatives FH not detected *True negatives - no FH not cascaded Total numbers tested *No cascading is done in true negative cases i.e. from those with no FH 18

19 19 From the table above, cascading will take place from 940 FH cases since there are 60 true negatives from each strategy that will not be cascaded from. The tables below shows the benefits of cascade screening in terms of the numbers, life year and QALY gained when relatives are followed for a lifetime. We calculated the costs per index case diagnosed or screened which were divided into clinical confirmation and genetic confirmation costs. All screening methods incurred a clinical confirmation costs i.e. the costs of cholesterol testing. Cholesterol confirmation costs were derived from the unit costs of healthcare professionals (physician, dietician and phlebotomy) multiplied by the time that each healthcare professional spent with the patient 1,27. The costs of lipid profiles for each patient were then added. Genetic confirmation costs were calculated by multiplying the cost of DNA testing obtained from 4 by numbers of cases who had DNA testing. The total cost per index case was calculated by adding the total cost both cholesterol and genetic confirmation for each method divided by 1000 index cases that were screened, see the table 10 below for further details. Table 10, Total costs and cost per index case diagnosed by method of screening Index cases Strategy 1 Strategy 2 Strategy 3 Strategy 4 Clinical confirmation 201, , , ,630 Genetic confirmation - 117, , ,500 Total cost 201, , , ,130 Cost per patient Results for relatives per 1000 screened index cases In this analysis, we assume a PCT has 1000 index cases that are being seen. Benefits shown in the analysis for cascading assumes the effect of high dose statins compared with no treatment, since if no cascading takes place we assume the relatives will not be identified and will not benefit from treatment. We also assume different levels of implementation for cascade testing to illustrate the benefits of cascade screening. For ease of interpretation of results we assumed that the level of implementation is the same as the effectiveness of implementation, e.g. 15% increase in cascade screening is the same as the 15% increase in effectiveness of cascade screening. Table 11, number of relatives FH cases tested and positive cases identified per 1000 index cases screened up to second degree relatives Relatives Strategy 1 Strategy 2 Strategy 3 Strategy 4 19

20 True positives FH False positives - not FH cascaded False negatives FH not detected True negatives - not FH not cascaded Total tested Table 12, number of FH cases tested and positive cases identified per 1000 index cases screened up to third degree relatives Relatives Strategy 1 Strategy 2 Strategy 3 Strategy 4 True positives FH False positives - not FH cascaded False negatives FH not detected True negatives - not FH not cascaded Total tested It is evident from the tables above that cascading using the recommended strategy 4 will result in 1,415 and 1,772 positive cases being identified and offered treatment per 1000 index cases when second and third degree relatives are followed respectively. The cholesterol only testing method will identify an additional 730 positive FH cases if up to second degree relatives are followed which increases to 808 if up to third degree relatives are followed. As mentioned earlier, at least some cascading is better than not cascading at all; however, it has been established that by not using the recommended screening strategy society as a whole will lose interms of QALYs and that will be an inefficient use of NHS resources. Table 13 and 14 below shows the lifetime QALYs and LYG by each screening strategy when optimal treatment is compared to no treatment or sub-optimal treatment. The benefits have been estimated from the numbers of discounted QALY and LYG gained shown in table 6, multiplied by the number of positives cases identified by each strategy. For instance, the cholesterol testing method identifies 730 true FH cases when up to second degree relatives are followed. This number is then multiplied by discounted QALYs for an optimal dose of statins, arriving at the figure of 8,696 QALYs. Table 13, QALYs and LYG per 1000 index cases cascaded up to second and third degree relatives using different methods, optimal statin treatment compared to no treatment Strategy 1 Strategy 2 Strategy 3 Strategy 4 Relatives tested upto second degree True positives - FH QALY LYG

21 21 Relatives tested upto third degree True positives - FH QALY LYG Table 14, QALYs and LYG per 1000 index cases cascaded up to second and third degree relatives using different methods, optimal statin treatment compared to sub-optimal statin treatment Strategy 1 Strategy 2 Strategy 3 Strategy 4 Relatives tested upto second degree True positives - FH QALY LYG Relatives tested upto third degree True positives - FH QALY LYG Cost of diagnosis/screening Using the costs in table 5, which shows that the cholesterol method costs 202 per index case and 116 per relative, we estimated the cost of clinical confirmation by multiplying the total number of people tested by the cost per person. For genetic confirmation we multiplied all the relatives who had DNA tests by the cost of DNA tests per person, which is 75 for relatives and 125 for index cases. The total cost of diagnosis was the sum of clinical and genetic confirmation. The cost per person was the total cost divided by the numbers of people tested. Cholesterol method strategy 1 is the cheapest followed by strategy 4, due to large numbers of people screened. The analysis shows that there is little difference in cost between screening upto second and third degree relatives, yet there is a considerable difference interms of QALYs or LYG. This implies that were possible relatives should be followed for upto third degree relatives. Table 15, the cost of diagnosis per relative using the different cascading methods per 1000 screened index cases when upto second degree relatives are screened Relatives Strategy 1 Strategy 2 Strategy 3 Strategy 4 Clinical confirmation 476, , , ,767 Genetic confirmation - 180, , ,463 Total costs 476, , , ,230 Cost per patient

22 Table 16, the cost of diagnosis per relative using the different cascading methods per 1000 screened index cases when upto third degree relatives are screened Relatives Strategy 1 Strategy 2 Strategy 3 Strategy 4 Clinical confirmation 520, , , ,776 Genetic confirmation - 227, , ,818 Total costs 520, , , ,594 Cost per patient Modelling reality of FH screening and events avoided So far, the above analysis has assumed a best case scenario where all the FH people are identified and offered treatment. However, in reality only a small proportion of people with FH are identified and are being offered treatment. The current estimates suggest that only about 15% of eligible patient are being offered treatment 2. Crucial to realising the benefits of treatment mentioned above is identifying those patients that have FH and offer them treatment. It is widely recognized that cascade testing may only identify 50% of these predicted cases; hence there is strong case for investigating new methods for identifying new FH probands. Below we will illustrate the potential savings that can accrue to the NHS and the number of cardiovascular events that can be avoided by implementing cascade screening. The analysis is done assuming that cascading is done using the recommended strategy 4 of DNA based method and cascading from both definite and probable FH using additional LDL-cholesterol testing in order to maximise the benefits. Any other cascading method will yield substantial benefits, but it will be a comparatively inefficient use of resources as more cases could be identified cost-effectively using the recommended Strategy 4. Table 17, Lifetime cardiovascular events avoided per 1000 relatives assuming different proportions of people are found by cascade screening method when FH people are optimally treated by offering them high intensity statins Events Proportions of people found by cascading method and offered treatment 15% 20% 30% 40% 50% 60% 70% 80% 90% 100% Myocardial Infarction (MI) Stroke Heart failure Revascularisation Unstable angina

23 23 Cardiovascular death Death other causes Total alive The numbers of events in each column have been estimated by multiplying the number of events that that will be observed in an ideal situation (all patients being identified) by the actual percentages of FH people being screened by cascade screening. For instance, if all FH patients are identified and treated optimally (100%) then 141 MIs will be avoided per 1000 treated cases, but if say only 15% are seen then only 21 MIs will be avoided per 1000 treated cases (15%*141). The column with 50% identification rate represents what the current cascading methods should realise if the NICE guideline is followed, according to current estimates 1. The first column shows the events avoided using the current estimated rate of identification, which suggests that only 15% of FH cases are being treated. It is clear that by increasing cascading screening from its current levels to the realistically achievable rate of 50% will save lives and save resources for the NHS. The events avoided will translate into the QALYs and LYG shown below. Table 18, Estimated QALY and LYG by different cascading methods when different proportions of FH people are identified due to implementing cascading screening per 1000 relatives optimal treatment compared to no treatment upto second degree relatives Strategy Proportions of people found by cascading method and offered treatment 15% 20% 30% 40% 50% 60% 70% 80% 90% 100% QALY gains Strategy Strategy Strategy Strategy LYG gained Strategy Strategy Strategy Strategy

24 QALY gains by strategyfrom true FH relatives Figure 2, Estimated QALY by different cascading methods when different proportions of FH people are identified due to implementing cascading screening per 1000 relatives optimal treatment compared to no treatment upto second degree relatives QALY gain strategy 1 QALY gain strategy QALY gain strategy 3 QALY gain strategy % 20% 30% 40% 50% 60% 70% 80% 90% 100% Proportions of FH cases identified and offered optimal treatment Table 19, Estimated QALY and LYG by different cascading methods when different proportions of FH people are identified due to implementing cascading screening per 1000 relatives optimal treatment compared to sub-optimal treatment upto second degree relatives Strategy Proportions of people found by cascading method and offered treatment 15% 20% 30% 40% 50% 60% 70% 80% 90% 100% QALY gains Strategy Strategy Strategy Strategy LYG gained Strategy Strategy Strategy Strategy Figure 3, Estimated QALY and LYG by different cascading methods when different proportions of FH people are identified due to implementing cascading screening per 1000 relatives optimal treatment compared to sub-optimal treatment upto second degree relatives 24

25 QALY gains by strategyfrom true FH relatives QALY gain strategy 1 QALY gain strategy 2 QALY gain strategy 3 QALY gain strategy % 20% 30% 40% 50% 60% 70% 80% 90% 100% Proportions of FH cases identified and offered optimal treatment Table 20, Estimated QALY and LYG by different cascading methods when different proportions of FH people are identified due to implementing cascading screening per 1000 relatives optimal treatment compared to no treatment upto third degree relatives Strategy Proportions of people found by cascading method and offered treatment 15% 20% 30% 40% 50% 60% 70% 80% 90% 100% QALY gains Strategy Strategy Strategy Strategy LYG gained Strategy Strategy Strategy Strategy

26 QALY gains by strategyfrom true FH relatives Figure 4, Estimated QALY and LYG by different cascading methods when different proportions of FH people are identified due to implementing cascading screening per 1000 relatives optimal treatment compared to no treatment upto third degree relatives QALY gain strategy 1 QALY gain strategy QALY gain strategy 3 QALY gain strategy % 20% 30% 40% 50% 60% 70% 80% 90% 100% Proportions of FH cases identified and offered optimal treatment Table 21, Estimated QALY and LYG by different cascading methods when different proportions of FH people are identified due to implementing cascading screening per 1000 relatives optimal treatment compared to sub-optimal treatment upto third degree relatives Strategy Proportions of people found by cascading method and offered treatment 15% 20% 30% 40% 50% 60% 70% 80% 90% 100% QALY gains Strategy Strategy Strategy Strategy LYG gained Strategy Strategy Strategy Strategy

27 QALY gains by strategyfrom true FH relatives 27 Figure 5, Estimated QALY and LYG by different cascading methods when different proportions of FH people are identified due to implementing cascading screening per 1000 relatives optimal treatment compared to sub-optimal treatment upto third degree relatives QALY gain strategy 1 QALY gain strategy 2 QALY gain strategy 3 QALY gain strategy % 20% 30% 40% 50% 60% 70% 80% 90% 100% Proportions of FH cases identified and offered optimal treatment Table 22, lifetime savings for relatives due to cardiovascular events avoided assuming different proportions of FH people are found by cascade screening method when FH people are treated optimally offered low high of statins Savings from proportions of people found by cascading method and offered Event treatment Event 15% 20% 30% 40% 50% 60% 70% 80% 90% 100% Myocardial 35,9 71,98 95,97 119,9 143,96 167,95 191,95 215,9 239,93 Infarction (MI) 91 47, , 565,71 848,5 1,131, 1,414, 1,697,1 1,979,9 2,262,8 2,545, 2,828,5 Stroke ,2 76,52 102,0 127,5 153,05 178,55 204,06 229,5 255,08 Heart failure 62 51, Revascularisatio 55,3 110,6 147,5 184,4 221,36 258,25 295,14 332,0 368,93 n 40 73, ,03 14,07 18,76 23,45 42,21 Unstable angina 6 9, ,146 32,837 37, ,910 Cardiovascular 71,2 142,5 190,1 237,6 285,15 332,67 380,20 427,7 475,25 death 88 95,

28 Cost savings at different levels of cascade implementation Death other causes Total savings - 1,17 9-1,572-2,358-3,144-3,931-4,717-5,503-6,289-7,075-7, ,86 1,183, 1,577, 1,972, 2,366,5 2,761,0 3,155,4 3,549, 3,944, , 647 Figure 6, lifetime savings for relatives due to cardiovascular events avoided assuming different proportions of FH people are found by cascade screening method when FH people are treated optimally offered low high of statins Myocardial Infarction (MI) Stroke Heart failure Revascularisation Unstable angina Cardiovascular death Death other causes Total savings 3,950,000 3,700,000 3,450,000 3,200,000 2,950,000 2,700,000 2,450,000 2,200,000 1,950,000 1,700,000 1,450,000 1,200, , , , ,000-50,000 15% 20% 30% 40% 50% 60% 70% 80% 90% 100% Cascade screening implementation The analysis shows that currently by identifying 15% of FH patients, the NHS would be saving 591,647 and gaining 3,167 discounted QALYs per 1000 screened index cases when cascade testing and offering optimal treatment (Strategy 4). This will translate to a 8.5 million saving to the NHS over a 55 year period, or about 155,000 per year. If the recommended strategy is fully implemented i.e. identifying 50% of the affected relatives, potentially 10,558 QALYs will be gained and 1,972,156 will be saved per 1,000 screened index cases. This will translate to 94.7 million savings nationally over a 55 year period, or 1.7m per year. By not fully identifying eligible patients (15% currently being identified rather than the potential 28