Questions and Answers about the design and conduct of HPS

This material is also available as an MS-Word document. To view/download it in this format, please click on the appropriate flag to select the nearest server to you:
UK	USA

A glossary containing many of the medical terms is also available.

Q	What is the Heart Protection Study (HPS)?
A	It is the world's largest ever trial of cholesterol-lowering drugs and of antioxidant vitamins, and the first cholesterol-lowering trial to include a substantial number of women, elderly people and those with diabetes and with below-average cholesterol.
Q	Why was it undertaken?
A	Throughout the whole range, blood cholesterol levels are an important cause of coronary heart disease (CHD). Prolonged lower blood cholesterol levels are associated with lower risks of CHD. Cholesterol- lowering therapy may therefore be worthwhile for individuals at high risk of coronary heart disease events irrespective of their cholesterol levels. Observational studies also suggest that increased dietary intake of antioxidant vitamins may be associated with lower risks of CHD. Heart disease is Britain's biggest killer, the biggest killer throughout the developed world, and an increasing killer in the developing world. Even a small reduction in the numbers who suffer from it each year could save far more lives than a much larger reduction in the numbers who suffer from rare diseases.
Q	What were the study objectives?
A	Primary objectives: to assess in a wide range of people at increased risk of CHD, the effects of: prolonged cholesterol-lowering with a statin on total and cause specific mortality; and supplementation with antioxidant vitamins E and C and beta-carotene on total CHD and fatal CHD. Secondary objectives: to assess: the effects of cholesterol-lowering with a statin on deaths from coronary disease, from other vascular causes, and from various non-vascular causes, including cancer. the effects of cholesterol-lowering with a statin, and of antioxidant vitamin supplementation, on total CHD in the first 2 years and in the later years of treatment (to see if the effect increases with time); the effects on cause-specific mortality during the treatment period (and in the longer term); and the effects on total stroke and presumed ischaemic stroke during the treatment period. the effects on total CHD and on major vascular events in lots of different sub-groups of patient: for example, those in different prior disease categories, men and women, young and old, and those with different blood cholesterol levels at presentation.
Q	Who carried out the study?
A	The Clinical Trial Service Unit (CTSU) at Oxford University designed, coordinated and analysed the study, which involved 69 hospitals throughout Britain. CTSU's work chiefly involves studies of the causes and treatment of major diseases such as heart attack, stroke and cancer.
Q	How much has the study cost?
A	About £21 million (GBP) over eight years - nearly $32 million (US).
Q	Who paid for it?
A	Funding was from 4 sources - the UK's Medical Research Council (MRC), the UK's British Heart Foundation (BHF) and the international pharmaceutical companies Merck & Co Inc [manufacturer of the statin] and Roche Vitamins Ltd [manufacturer of the vitamins]. The funding was provided as grants to Oxford University, so that the pharmaceutical sponsors had no say in how the money was spent, in the day-to-day running of the study, the analysis of the data, or the way the results are presented, published or publicised.
Q	Is the study independent of the sponsors?
A	Yes. The idea for this study came from the principal investigators in CTSU, and it was designed, has been run, and is being analysed and interpreted by them, entirely independently of all the sources of funding. Indeed, quite unusually, the sponsors will not even have direct access to the full data file (except to audit the quality of the data) so that all analyses are conducted free from any potential conflicts of interest. Data will, of course, be made available to government regulatory authorities so that they can confirm the study analyses, but this will be done directly between these authorities and the investigators.
Q	What other steps are taken to ensure the independence of the research?
A	A data monitoring committee, which does not include anyone involved in the trial, acts as watchdog, examines interim analyses, and responds to any concerns.
Q	When did the study start?
A	The first patients were recruited to a pilot study in 1987. Planning and fund-seeking for the main study started in 1990. The first patient was recruited in May 1994. The 20,000th volunteer was recruited on 8 April 1997 and recruitment finished that year.
Q	What type of study was it?
A	A randomised double-blind clinical trial; i.e. one in which volunteers randomly receive either the active study treatment or a placebo (dummy tablet or capsule), and in which neither volunteers nor doctors know who is receiving active study treatment and who is taking a placebo.
Q	How many patients have taken part?
A	20,536 finally, but this involved postal invites to 131,000 and screening 63,603, of whom 32,145 agreed to go into the 2-month run in period prior to study entry.
Q	How was the study designed to assess the various combinations of treatment?
A	It is known as a factorial (2 x 2) design. 5,000 were allocated active statin and active vitamins 5,000 were allocated active statin and placebo vitamins 5,000 were allocated placebo statin and active vitamins 5,000 allocated placebo statin and placebo vitamins Assessment of cholesterol-lowering involves comparisons of the 10,000 allocated active statin versus the 10,000 allocated placebo statin. Likewise, the assessment of antioxidant vitamin supplementation involves comparisons of the 10,000 allocated active vitamins versus the 10,000 allocated placebo vitamins.
Q	Was it difficult to recruit volunteers?
A	We needed to invite 2 to get 1 to turn up for screening. About half of those screened were both eligible and willing. Of the half who did not enter, one third would have been eligible but refused, one third would have had difficulty attending clinics regularly, and one third were not eligible. So about 16% of those approached by initial letter entered the study. This compares favourably with other studies.
Q	What age range were the volunteers?
A	Between 40 and 80 at the time of screening.
Q	How many men and how many women?
A	15,454 men and 5,082 women.
Q	Why were there three times as many men as women?
A	Men suffer from heart disease at a younger age than women do, and so there were more men within the study age range available to approach. But it was also more difficult to recruit women - for each 100 invitations sent to men about 17 were willing and eligible to join. For each 100 invitations sent to women only about 9 agreed. We do not know why there was this difference. Nevertheless, this is still the largest study to look at cholesterol-lowering in women, because we made special efforts to recruit women. In fact, it is the first to include a substantial number of women. Up to now the effects of cholesterol-lowering in women have been extrapolated from evidence in men.
Q	What other categories of volunteers were specially needed?
A	Groups who were at particularly high risk of heart disease for whom there had been too little evidence, including: People with diabetes - they are at high risk of developing heart disease. People over 70 - a group often overlooked by previous smaller studies, and in which statin use is controversial. People with non-coronary vascular disease - for example, those that had suffered strokes, mini- strokes or other circulatory problems. People with pre-existing vascular disease or diabetes who had average or below-average cholesterol levels (since it had been suggested that there might be a 'threshold' below which lowering cholesterol would not produce worthwhile benefits).
Q	Who was eligible to enter the trial?
A	Anyone between the ages of 40 and 80 who was considered to be at substantial risk of CHD within 5 years because of evidence of vascular disease anywhere in the body, or previously diagnosed with diabetes. This meant: any evidence of CHD (heart attack, angina, previous coronary revascularisation). any other vascular disease (other arterial revascularisation, stroke or mini-stroke, or symptoms suggestive of blockages in the leg arteries). any history of diabetes (either type 1 - early onset insulin dependent diabetes; or type 2 - late onset non-insulin dependent).
Q	What was the breakdown in participants?
A	There was overlap between the groups (i.e. some people had more than one condition putting them at increased risk): history of heart attack - 8,510 other history of CHD without heart attack (e.g. angina, heart bypass surgery or coronary angioplasty) - 4,876 history of a stroke, mini-stroke or surgery to the neck arteries - 3,280 disease of other arteries - 6,748 diabetes mellitus - 5,963 treated hypertension (high blood pressure) - 8,457 Other categories: below average cholesterol at baseline (i.e. LDL cholesterol under 3.0 mmol/l: this would be approx. 120mg/dl in US measurements) - 6,793 total cholesterol less than 5.0 mmol/l (approx. 200mg/dl in US measurements) - 4,072 age 65 and above - 10,697 age 65 to 69 - 4,892 age 70 and above - 5,805
Q	What criteria excluded participation in the trial?
A	The main exclusion criterion was a history of any other severe disease that might limit compliance or the ability to attend clinic visits regularly over 5 years, or that might cause death within the next few years. This included severe heart failure, severely disabling stroke, severe chronic airway disease or a history of cancer (other than non-melanoma skin cancer). Because of potential side-effects and drug interactions with statins, we also excluded volunteers with severe liver or kidney disease, muscle disease or any condition likely to lead to organ transplantation. We also excluded people if they had had a heart attack, stroke or had been in hospital for angina in the last 6 months, although they were eligible for recruitment later.
Q	How many hospitals took part in the study?
A	69 UK hospital-based clinics
Q	How many doctors took part?
A	About 85: 51 hospitals had 1 doctor involved, but other hospitals had 2 or 3.
Q	Was it difficult to persuade doctors to participate?
A	No. Doctors from a wide range of disciplines were interested.
Q	What specialities did these doctors represent, and were there any general practitioners (i.e. family doctors)?
A	All the collaborating doctors were consultants and all were hospital-based. There were nearly 30 cardiologists (many of whom were also general physicians), 18 lipidologists or chemical pathologists, 12 diabetologists (who were mostly also general physicians), 21 other general physicians, 3 neurologists and 2 vascular surgeons. But even though general practitioners were not directly involved, about 10,000 general/family practices helped by providing further information about medical conditions (such as heart attacks and other vascular disease) reported by participating patients during the study.
Q	How many nurses took part?
A	Currently about 70, but over the course of the study there have been almost 200.
Q	How many clinic visits did the volunteers make in total?
A	There have been 308,000 consultations, of which 286,000 (93%) were face to face - i.e. in clinic (more than 99%) or in hospital or at home. The other 7% were telephone contacts.
Q	How many visits did each volunteer normally need to make?
A	It varied, but anything up to 20, with most volunteers having 12-15 visits.
Q	How many blood samples were collected and analysed?
A	Blood was taken for analysis during a total of 258,000 volunteer consultations.
Q	How many forms did you collect?
A	292,000 forms on paper, and about 17,000 electronically at the final follow-up visit.
Q	How was such a huge trial organised and run from one unit?
A	Teamwork and streamlining. About 20 computing, administrative and clerical staff at the CTSU in Oxford coordinated all 69 clinics. Potentially suitable participants were identified from local hospital databases. Letters were sent to all those that might be eligible, in the name of their local hospital doctor. Responses were either returned to the CTSU by post or by using a central Freefone number manned by CTSU staff. The local hospital doctors employed senior nurses to set up and run HPS clinics in their hospitals. The nurses liaised closely with the CTSU and were sent lists each week of the people who had been invited and when they were expected to attend. All appointments were managed through the coordinating centre computer, with participants and staff having access to the system via the 24-hour Freefone service. So, at any time during the study, the coordinating office was aware of the status of each patient in each clinic in the UK. All data and blood collected by the clinic nurses were forwarded immediately by courier or mail to the CTSU, and all blood analyses were done by the coordinating centre's laboratory staff. Volunteers were seen initially by a nurse for an interview of up to an hour. If eligible and willing to participate, a blood sample was taken and they were given a treatment pack and an appointment to return in two months. If participants did attend for the second visit and were still eligible and willing, they were then randomised into the study with the agreement of their own doctors. The nurse telephoned the central randomisation service on the Freefone number and was told to issue a particular drug pack to the patient. The treatment combination in that pack would be the appropriate one for the treatment group to which that individual had been allocated at random.
Q	This trial called for a great degree of co-operation and commitment from volunteers - what did CTSU offer in return?
A	It would not have been possible to undertake this trial without our volunteers. We are enormously in their debt and immensely grateful to them all. We were very aware that taking part demanded commitment. We were determined that the participants should be kept fully informed of anything relevant to the study (with the agreement of their own doctors). We provided regular newsletters for every volunteer, with information on the aims and progress of the trial, new findings from similar trials, articles about the experience of participants and tips on healthy living. In addition, it was important for advice to be available for medical staff and volunteers who had any questions or concerns about the study. So, the clinicians at CTSU worked a rota to provide a 24-hour on- call service. The preliminary results of this trial, which are being made public on 13 November 2001, are being posted to all volunteers and their general practitioners at the same time as they are announced by the investigators at the American Heart Association meeting in Los Angeles. Full details will also be available from 15.30GMT 13 November 2001 on a special website set up by the CTSU at: http://www.hpsinfo.org/
Q	How did you cope with volunteers who became ill on the trial (e.g. had a heart attack)?
A	Throughout the trial, the volunteers' own doctors remained responsible for their patients' care. If a participant had a heart attack their management was up to the local doctor. If the doctor felt that cholesterol-lowering treatment was indicated they were free to prescribe whatever they thought was most appropriate for that particular patient. Until spring 1998, such patients would have stopped their study statin or placebo tablets. After that date, when it had become clear that higher doses of statins were well tolerated and apparently safe, prescribed statin could be added to study treatment if the dose of the non- study statin did not exceed the equivalent of about 40mg daily simvastatin (the study statin).
Q	Would the participant be "unblinded" so the doctor knew what drug the patient was taking?
A	If a doctor specifically needed to be unblinded that would have been done, but normally they were asked to make a decision about cholesterol-lowering treatment based on their usual criteria. As the trial has progressed, the proportion of participants prescribed non-study statin has gradually increased. By final follow-up in 2001, about a fifth of volunteers were on additional statin, over half of whom were also taking the study statin or placebo.
Q	Have you had to "unblind" any volunteers because of major problems?
A	Remarkably few. We are obliged to report to the regulatory agencies all "serious adverse events" thought with reasonable probability to be due to study treatment. There have been only 16 such reports among the 20,536 randomised volunteers during an average of 5-6 years of treatment, all of whom were unblinded. However, in the light of the unblinding and other information, some of these problems were later not thought to be due to study treatment.
Q	Statins have been associated with muscular problems. What precautions did you take?
A	The known rare side effect of myopathy (muscle problems) due to statins dictated some of our trial procedures. People were excluded at the start if they had a history of inflammatory muscle disease or if the blood test for the muscle enzyme "creatine kinase" (CK) done at the first visit showed a value 3 times the upper limit of normal. All participants were warned at the screening interview about the possibility of muscle pain or weakness, and this was also stated explicitly in the patient information leaflet. Nurses advised volunteers that if they experienced unusual or unexplained muscle pain or weakness they should telephone the study nurse or the 24 hour Freefone number where a doctor was always available to talk to any such volunteer. If muscle pain was thought to be significant, or the doctor considered it prudent, an extra visit was arranged and a blood test done to check the muscle enzyme levels. At all routine visits, nurses asked participants to report any new or unexplained muscle pain. If this was reported a blood sample would be used to check the muscle enzyme CK levels. This is a reliable and specific test for muscle damage.
Q	What was the drop out rate from the trial?
A	Relatively low. On average during the trial, about 80% of participants took their allocated study statin or placebo tablets regularly, and 85% took their vitamins or placebo regularly. Because the analysis of results will be done on an 'intention-to-treat' basis, everyone who was randomised is counted even if they never took the study treatment. If people were unwilling to continue the study treatments they were still encouraged to continue attending the clinics or, at least, to keep in phone contact. Over 95% were being actively followed in this way, with the remainder followed by contact with their GP.
Q	How did you decide which statin to use?
A	Simvastatin (Zocor) was the first statin to become available in the UK. CTSU had already run a pilot study among over 600 Oxford volunteers looking at its safety and efficacy at lowering cholesterol. Merck & Co.Inc., the manufacturers, were prepared to help fund a large-scale trial jointly with other funders (UK Medical Research Council, UK British Heart Foundation, and Roche Vitamins Ltd which manufactures the vitamins studied).
Q	How did you decide what dosage to use?
A	Based on our pilot study, which randomised about 200 volunteers to 40mg of simvastatin daily, 200 to 20mg daily and 200 to placebo, the 40mg dose of simvastatin produced somewhat greater cholesterol reductions than 20mg, without any obvious hazards or safety concerns over the first 3 years. Given the continuous relationship between lower cholesterol and lower CHD risk in observational studies, it was felt that the maximum dose then considered safe and available should be used.
Q	What is the normal statin dosage range in treatment outside of this trial?
A	For this statin, the most commonly used doses are 10mg and 20mg daily. Different brands vary slightly in their potency. But the other statins would have similar cholesterol-lowering potency when prescribed at their most common dosage.
Q	What percentage of people who could benefit from statins actually gets them?
A	Data are always a little out of date but treatment rates are low. Recent sources report the following: Euroaspire 2 (Lancet 2001, Vol 357, 995-1001) collected data in 1999 and 2000 on 3,000 CHD patients in hospitals in 9 European countries, excluding the UK. As they were patients being treated in cardiology centres, the figures may not reflect the general situation (see below). It found that 60% had cholesterol over 5.0mmol/l, with a quarter with cholesterol over 6.0mmol/l. Two-thirds of all people with CHD were on cholesterol-lowering therapy, but only a half reached a target of 5mmol/l or less. In the UK (BMJ 2001; 322:1463), 24,000 patients with CHD from 550 general practices were surveyed in 1997 and 1998. Half the women and one third of the men did not even have their cholesterol measurement documented. Only 18% of men and 13% of women with CHD were on statins. CTSU's own more recent data collected as part of another study shows that 30% of heart attack survivors (for whom previous evidence was strongest) are not on a statin, and 75% of these have cholesterol levels of over 5mmol/l. Of the ones on a statin, a quarter have not fallen below 5mmol/l. In the USA (Circulation 2001; 103: 38-44), among patients in 1998-1999 with a history of heart attack - the strongest indication for a statin - less than a third were discharged on a lipid-lowering drug. So, there is widespread evidence of massive under-treatment even among those groups where the evidence of benefit is most clear. There is even less use in other high-risk groups where previous evidence is less clear (such as those without a history of heart disease, the elderly, or those with average or below-average cholesterol levels).
Q	How did you decide what analyses to undertake at the start, and have extra analyses been done as a result of findings from other trials that have reported in the meantime?
A	At the beginning of the trial, the major concern was the safety of cholesterol-lowering, and its effects on total mortality and other specific causes of death and on heart attacks. As a result of reassuring results from other statin trials many people are less concerned about safety issues, but they remain important, particularly as statins are starting to be used more and more widely. Because of suggestive evidence from other trials, we are now very interested in assessing the effects of cholesterol-lowering on stroke, as well as on heart attack and other vascular events, in the various predefined subgroups (such as the elderly, women and people with diabetes, previous strokes or other vascular disease). For the antioxidant vitamins, the main hope is that there will be an effect on coronary events.
Q	Why did you include people who already had 'low' LDL cholesterol, and was there any risk of doing this (e.g. some studies have controversially pointed to links between low cholesterol and cancer)?
A	In observational studies the association between cholesterol and risk of CHD seems to be continuous, without any threshold below which having lower cholesterol is not associated with lower risk. It therefore seemed logical to include people across the whole range of cholesterol levels (including those with average and below-average levels) if those people were otherwise at increased risk of CHD. Only in this way would it be possible to see whether there would be benefit from lowering cholesterol. There have been concerns about safety with cholesterol-lowering, and this has been monitored during the study by the independent data monitoring committee. Previous concerns about cancer illustrate why cancer incidence during the trial is an important endpoint.
Q	There have been reports that low cholesterol may increase mortality in people over 70. Did this concern you?
A	These reports illustrate the importance of conducting randomised trials (such as HPS) that reliably study the effects of lowering cholesterol in substantial numbers of older people. It is only with such large-scale randomised evidence that the balance of benefits and any risks can be reliably assessed.
Q	How do you ethically justify denying some patients statins (in this trial) when it is already well established that statins lower LDL cholesterol?
A	Nobody who was thought by their own doctor to need a statin was denied it. But, there are many questions about just how widely statins should be used. At entry to the study, the lipid profile (i.e. total cholesterol, HDL, LDL and triglyceride measurement) of each participant who entered the initial run-in phase was sent to their family doctor (i.e. general practitioner). The family doctor was also sent a form to return if he or she felt that their patient was likely, immediately or in the future, to need to be prescribed statin treatment - in which case that volunteer was not randomised. During the study, the volunteers' family doctors were free to monitor cholesterol levels if they considered this to be indicated. Non-study cholesterol-lowering treatment could be added by the managing doctor at any time. We explained to participants at the start of the study that it was already known that lowering cholesterol reduced the risk of heart attacks, but that most people among the range of volunteers recruited were not then receiving any kind of cholesterol-lowering treatment. At that time (between 1994 and 1997) there remained major concerns among many doctors about the safety of cholesterol-lowering, and most thought that there was no need to lower cholesterol in people with average or below-average levels. Subsequently during the study, participants and their family doctors were reminded that non-study cholesterol-lowering therapy should be started if, perhaps in the light of emerging evidence from other studies or changes in a participant's condition, it was thought to have become indicated.
Q	Why did the trial also look at vitamins?
A	In the early 1990s there was enormous interest and optimism about the potential of antioxidants to reduce the risk of CHD. A large body of epidemiological data showed that people with higher intakes of antioxidant vitamins in the diet were less likely to suffer heart disease. Small-scale trials had suggested benefit from vitamin supplements but the results were uncertain. There also seemed to be a good theoretical rationale as LDL cholesterol is more likely to be deposited in the arterial wall if it is oxidised, and dosing with oral antioxidants makes it less likely that the LDL cholesterol becomes oxidised when exposed to oxidant stress. While HPS has been in progress, a few large-scale trials of antioxidants have given disappointing results, so there is less optimism than before. Large-scale trials of vitamins alone are often difficult to fund, so adding a vitamin component in a factorial design to another funded comparison is a good way of answering both questions.
Q	Why did you choose the vitamins you did choose?
A	The antioxidant vitamin of most interest at the time the study was being planned was vitamin E. There was a further scientific rationale that a combination of water soluble (vitamin C) and lipid soluble (vitamin E and beta carotene) antioxidant vitamins may be more efficacious, with added vitamin C also helping in the regeneration of vitamin E in the body. So we were happy to use a combination of antioxidants, and Roche Vitamins Ltd agreed to help fund a study using this antioxidant vitamin combination.
Q	How could you control what volunteers did in terms of taking extra vitamins, given that they are available over the counter?
A	Participants were asked to avoid taking high doses of vitamin E, but it was not forbidden. In a large study such as this, it would be extremely unlikely that a few individuals taking non-study vitamins would affect the analyses and there are likely to be similar numbers of such people in each of the treatment groups. No restrictions were put on multivitamin use, but we know the rates of use overall are low and, typically, multivitamin preparations contain less than 10mg vitamin E.
Q	Were you concerned about the recent report that antioxidants may interfere with the effects of statins on cholesterol?
A	This was an interesting observation of an apparent effect on the increase in good HDL cholesterol with the combination of statin and niacin (another treatment for lowering LDL cholesterol), but it was based on a fairly small number of individuals. In the very much larger numbers in HPS, no such effect was observed with the vitamins studied on the slight HDL-raising effect of simvastatin on its own. This again illustrates the importance of getting large-scale randomised evidence, and looking at clinical endpoints, rather than lipid measurements as surrogates.
Q	What doses or amounts of antioxidant vitamins were used?
A	A mixture of 600mg of vitamin E, 250mg vitamin C and 20mg beta-carotene daily, either alone or in combination with the cholesterol-lowering statin.
Q	Can these intakes be achieved through diet, and if so, what foods must be eaten?
A	A recommended plant-based diet that contains 5 to 9 servings per day of fruits and vegetables can provide 250mg of vitamin C. Diets recommended by the National Cancer Institute in the US and the US Department of Agriculture have been shown to provide 6-7mg of beta-carotene daily. Vitamin E is more difficult; it is present only in small amounts in food, and the richest sources (such as vegetable oils, nuts and seeds) are usually high in calories.
Q	What happens to the volunteers now - will you continue to follow them?
A	We have always planned to follow participants long-term through national registries for death and cancer. We are also examining the possibility of other forms of long-term follow-up for non-fatal events. Nothing is decided at present but it would be interesting to do so.
Q	Will the volunteers be told whether they were on active treatment or placebo?
A	We do not plan to tell them, but if they want to know they can find out by ringing the Freefone number.
Q	Why do you not plan to tell them?
A	Some people obviously do want to know and they can easily find out. However, experience tells us that not everyone wants to know. The main reason for not telling people is that it minimises the risk of bias in the collection of long-term follow-up. (Someone's knowledge of which treatment they had may affect what they later report.) These effects are subtle but important. The reasoning is the same for keeping those who collect and code reported events blind to the treatment allocation.
Q	What if the volunteer asks his doctor?
A	A patient's own doctor can request to be unblinded in the same way that a patient can.