HOME
How Statistics in Medicine Can Be Misleading
When newspapers and scientists want to create the maximum impact for their (usually modest) findings in the field of prevention of heart disease or cancer, they publish the "relative risk" statistics, rather than "absolute risk" statistics.
To explain the two terms and their significance, let me give an example: If the chance of a thing happening is reduced from two per thousand to one per thousand, that is a "relative risk" reduction of 50%, which sounds a lot. But in reality, the "absolute risk" has gone down from 0.2% to 0.1%, or to put that another way, your chance of surviving has been increased from 99.98% to 99.99% which is really very small and nothing like as headline grabbing.
The following letter from Dr Uffe Ravnskov to the British Medical Journal concerns the misleading reporting of heart disease risk statistics on which decisions about aggressive intervention with statins are based.
Researchers claim clinical trials are reported with misleading statistics
BMJ, 18 June 2002Prevention is better than
cure, but the effect of most preventive measures is
trivial. The
relative risk reduction is therefore highly misleading, as
pointed out by Nuovo et al. (1), but even the absolute risk
reduction, or number needed to treat, means almost nothing to
most patients. For this group it is much more 'honest' and
effective to provide the figures that represent their chances
of surviving with and without treatment.
To
give an impression of the small improvement that is achieved
by commonly used preventive measures I have calculated the
chance of surviving with and without treatment of
hypertension, and with and without treatment of
hypercholesterolaemia (table).
For hypertension I have used the figures from a
meta-analysis of seventeen controlled, randomised trials (2).
For hypercholesterolaemia I have used the data from the 4S (3)
, as it demonstrated the most favourable outcomes in patients
with established cardiovascular disease, and the data from the
WOSCOPS trial (4) as this trial demonstrated the most
favourable outcome in healthy individuals with high
cholesterol. In both trials I have chosen the figures for
total mortality, because I assume that this is the most
relevant figure for most patients, and because no bias is
associated with that outcome. For the hypertension trials
total cardiovascular mortality is the measurement used,
because a meta-analytic calculation of total mortality was not
performed.
| Table. Benefits from treatment of high blood pressure and of high cholesterol | BP lowering |
4S | WOSCOPS |
| Relative risk reduction; % | -20 | -29 | -21 |
| Absolute risk reduction; % | -0.8 | -3.3 | -0.9 |
| Chance of surviving without treatment; % | 96 | 88.5 | 90.6 |
| Chance of surviving with treatment; % | 96.8 | 91.8 | 91.4 |
As mentioned, these figures are the most optimistic available. For instance, in many of the studies that were included in the meta-analysis on hypertension, total mortality was not reduced significantly. Further, the results from the most recent secondary preventive statin trial HPS were only half as good as those from 4S (5); and total mortality in the first primary preventive statin trial EXCEL was increased in the treatment group (6). If patients were given these odds, and then clearly informed about possible side effects from treatment, I guess that most of them might choose to spend their money on horse racing.
- Nuovo J, Melnikow J, Chang D. Reporting Number Needed to Treat and Absolute Risk Reduction in Randomized Controlled Trials. JAMA 2002;287:2813-4
- Hebert PR, Moser, M, Mayer J, Hennekens CH. Recent evidence on drug therapy of mild to moderate hypertension an decreased risk of coronary heart disease. Arch Int Med 1993;153:578-81.
- Randomised trial of cholesterol lowering in 4444 patients with coronary heart disease: the Scandinavian Simvastatin Survival Study (4S) Lancet 1994;344:1383-9.
- Shepherd J, Cobbe SM, Ford I, Isles CG, Lorimer AR, Macfarlane PW, McKillop JH, Packard CJ, for the West of Scotland Coronary Prevention Study Group. Prevention of coronary heart disease with pravastatine in men with hypercholesterolemia. N Engl J Med 1987;333:1301-7.
- Ravnskov U. Statins as the new Aspirin. Conclusions from the heart protection study were premature. BMJ 2002;324:789.
- Bradford RH, Shear CL, Chremos AN, Dujovne C, Downton M, Franklin FA, Gould AL, Hesney M, Higgins J, Hurley DP, et al. Expanded Clinical Evaluation of Lovastatin (EXCEL) study results. I. Efficacy in modifying plasma lipoproteins and adverse event profile in 8245 patients with moderate hypercholesterolemia. Arch Intern Med 1991;151:43-9.
Last updated 26 January 2003
Related Articles
