Skip navigation

Statins


Bandolier has had many requests to do something more on statins since we originally reported on the 4S study ( Bandolier 15 ). An overview of randomised trials from JAMA [1], which gives more information than that published in the original trials, makes that possible. It is also timely to think of where we are with statins when advertisements in major medical journals tells us that they are so good that trialists have had to stop the trials. The evidence for these claims will be presented at some time, so assessing where we are now is probably a good idea.

The studies

The systematic review looked for randomised trials which involved:

  • statin drugs alone used to reduce lipid levels rather than multifactorial interventions including another type of cholesterol-lowering drug
  • inclusion of data on deaths and/or strokes
They found 16 such studies. The researchers wrote to authors of the original studies and are able to give information on fatal and nonfatal strokes and myocardial infarction not always given in the original reports. But some of the studies were small and short. So Bandolier has examined the 13 studies which had over 100 patients in any treatment group, and which were of at least 6 months duration.

Outcomes

As you might imagine, there are lots of different outcomes (death from any cause, fatal and nonfatal strokes and MI to name the most important). There are also lots of analyses using all sorts of complicated statistics which give Bandolier a headache. So to make things simpler, Bandolier has invented a new kind of outcome - the "all bad things that could happen" outcome.

This consists of deaths from any cause plus nonfatal stroke plus nonfatal MI. This may seem a bit simplistic, but has the distinct advantage that it gives lots of events and can give an overall NNT for "all bad things".

Results

Primary prevention

Three trials examined statins for primary prevention, two using pravastatin and one lovastatin. They involved 7,961 patients treated for an average of 4.6 years, and heavily weighted by the WOSCOPS trial. The NNTs are shown in the Table, and the "all bad things" in the L'Abbé plot.

Statins in primary prevention (7961 patients, mean 4.6 years)
Event NNT (95% CI)
All cause death 107 (58 to 617)
All strokes 330 (123 to no benefit)
All CHD 48 (32 to 95)
All strokes plus all CHD 42 (28 to 80)
All death plus nonfatal stroke plus nonfatal MI 35 (24 to 63)
The three trials were all in the segment of the plot where statin was effective. The "all bad things" NNT was 35 (24 to 63). This means that 35 people have to be treated with a statin for 4.6 years to prevent a bad thing (death, stroke or heart attack) happening in one of them.

Secondary prevention

Ten trials examined statins for secondary prevention, two using simvastatin, four pravastatin and four lovastatin. They involved 20,589 patients treated for an average of 2.9 years, and heavily weighted by the 4S, CARE and EXCEL trials. The NNTs are shown in the Table, and the "all bad things" in the L'Abbé plot.

Statins in secondary prevention (20589 patients, mean 2.9 years)
Event NNT (95% CI)
All cause death 33 (28 to 42)
All strokes 71 (55 to 98)
All CHD 14 (13 to 16)
All strokes plus all CHD 12 (11 to 14)
All death plus nonfatal stroke plus nonfatal MI 11 (10 to 13)
The trials were predominantly in the segment of the plot where statin was effective. The "all bad things" NNT was 11 (10 to 13). This means that 11 people have to be treated with a statin for 2.9 years to prevent a bad thing (death, stroke or heart attack) happening in one of them.

Of course, this is a summation across all trials, with three different drugs, with different rates of events with controls reflecting different populations and durations of the trials. It is possible to do an analysis by drug, and this is also presented. Direct comparability is not possible, though.

Statins compared in secondary prevention
Statin Number Duration (years) Total patient years Events with controls (%) NNT (95%CI)
Lovastatin 9,250 1 9,250 2 187 (86 - no benefit)
Pravastatin 6,514 3.7 24,102 15 27 (19 - 51)
Simvastatin 4,825 5.3 25,573 33 10 (9 - 15)
With lovastatin, although over 9,000 people have been in trials, the trials were only for one year on average, and with low numbers of events. So the NNT is high at 187.

With pravastatin over 6,500 people were involved with trials going on for nearly four years on average. The number of events with controls was 15% and the NNT was 27.

With simvastatin, just under 5,000 people were studied, but for over five years. there was a large number of events with controls (33%), and the NNT was low at 10.

Adverse effects

The systematic review found no evidence for any increased risk in non-stroke mortality, nor any significant risk of cancer over control.

Comment

Many groups are seeking to produce guidance for clinicians on the use of statins. That guidance is best founded on solid evidence, and the best evidence comes from quality systematic reviews. There is an excellent example of using systematic reviews in practice guidelines which is worth reading [2].

The evidence is conclusive: statins work, and work well. Key questions being asked include which patient and which statin. many answers to these and other questions will be driven by cost influences. Perhaps they ought to be driven by the weight of evidence as well - evidence which will change as more studies report, and as other evidence from audit and elsewhere becomes available. When it does, and if it changes the picture, then guidance can change with it.

Cost effectiveness

This always seem to be the bogeyman of cholesterol-lowering, whatever the intervention used. A systematic review of the cost-effectiveness literature is a good place to start looking for advice [3].

The problem is, as the authors tell us, is that while there is general agreement among the studies, there is high sensitivity to the assumptions used, especially in screening strategies. The major conclusions were:

  • Cost effectiveness of primary prevention with cholesterol-lowering drugs is extremely variable, depending on age at start of treatment and risk profile.
  • Pharmacological intervention is least cost-effective in the young and the elderly.
  • Cost-effectiveness improves when treatment is targeted at high risk individuals.
  • Statins are more cost-effective at reducing cholesterol-related coronary events than other interventions.

References:

  1. PR Hebert, JM Gaziano, KS Chan, CH Hennekens. Cholesterol lowering with statin drugs, risk of stroke, and total mortality. JAMA 1997 278: 313-21.
  2. DJ Cook, NL Greengold, AG Ellrodt, SR Weingarten. The relationship between systematic reviews and practice guidelines. Annals of Internal medicine 1997 127: 210-6.
  3. S Morris, A McGuire, J Caro, D Pettitt. Strategies for the management of hypercholesterolaemia: a systematic review of the cost-effectiveness literature. Journal of Health Service Research and Policy 1997 2: 231-50.



previous or next story in this issue