In the first paragraph of his commentary, David Sackett noted that the force of law has been used to mandate seat belt use.1 In the third paragraph, he asserted that no personal health intervention should be undertaken in the absence of positive randomized trials.
With these 2 statements, he apparently advocates that seat belt laws be repealed until a trial is conducted in which those who are about to be involved in motor vehicle accidents are randomly assigned to either seat belt use or no seat belt use, with the group using seat belts experiencing a lower mortality rate. Similarly, I can only assume that Sackett is also opposed to discouraging children from smoking until randomized trials in which large numbers of teenagers are assigned to smoking or nonsmoking groups showing that those who smoke experience greater long-term mortality rates. To extend this train of thought even further, we should presumably not ban drunk driving until randomized trials have demonstrated that it is dangerous.
Many of the preventive medical manoeuvres currently in use will never be supported by data from randomized trials. In the 3 examples outlined above, randomized trials would be unethical even if they were possible. The accumulated evidence from nonrandomized studies for the benefits of seat belts, the harmful effects of smoking and the dangers of drunk driving is so vast that further study would be in no one's best interest (except perhaps the tobacco industry).
The wearing of seat belts and the avoidance of smoking and drunk driving are measures that cannot conceivably be harmful. The nature of the evidence we require before advocating a preventive medical intervention depends on the nature of the intervention. A pharmacological intervention is vastly different from a lifestyle intervention, and the quality of the evidence we require may also be vastly different.
Reference
- 1.