Andrew@reliabilityoxford.co.uk
Risk Ratios are usually accompanied by an estimate of precision measured by the 95% confidence interval (CI).
What this means is that the true risk ratio (for that particular study) has a 95% probability of falling within the stated range and that its most likely value is the one reported. For example, a RR of 2.1 (95% CI 1.8 to 3.1) means that the ‘true’ risk ratio has a 95% probability of being between 1.8 and 3.1, the most likely value being 2.1.
The wider the 95% CI the less probable the stated RR.
If the 95% CI includes 1.0 the result is usually considered not to be statistically valid. E.g. RR = 2.1 (95% CI = 0.8 to 5.6) and e.g. RR = 0.75 (95% CI = 0.4 to 1.5).
It is debatable whether or not the 95% confidence interval is the appropriate one to consider in the context of the civil law. The usual test of evidence at civil law is that of the balance of probabilities, or >50%.
It is technically possible to quote the results of epidemiological experiments using a 50% confidence interval. However, this might give the impression of over confidence in the result, most epidemiologists would agree that their experiments are subject to influences beyond their control and would doubt the value of reporting a 50% CI.