When people would choose treatment has wide implications

Editor—Steel's paper on the thresholds for number needed to treat at which health professionals and lay people would choose treatment for hypertension is important.¹ Only people with a particular condition can decide whether the inconvenience of treatment is worth the resulting reduction in risk. We cannot presume to know on their behalf. This view of personal autonomy is enshrined in the General Medical Council's advice on obtaining informed consent.

Normally, if patients do not want to make a decision about treatment they opt to follow a health professional's recommendation. Unless it is informed by information on lay preferences, that recommendation is likely to result in systematic overtreatment. Risk thresholds for intervention chosen by lay people are much higher than those chosen by general practitioners and specialists. Most guidelines are written by experts.

The modal five year number needed to treat to prevent a death that was chosen by lay people was 33. This corresponds to a 3% reduction in absolute risk of death, which (since 1 in 3 events results in death) in turn corresponds to a 9% reduction in absolute risk of a cardiovascular event. If lay people judge a 9% reduction in risk of death to be the point at which treatment becomes worthwhile this means that treatment should be recommended when their five year risk is 30% (treatment typically reduces cardiovascular risk by 30%, and 30%×30%=9%).

This contrasts sharply with the joint British recommendations on the prevention of cardiovascular disease in primary care. These regard intervention at a five year threshold of 15% as mandatory and at a five year threshold of 7.5% as to be considered. (These figures correspond to a five year number needed to treat to prevent a death of about 75 and 125 respectively; not surprisingly, these lie between the thresholds chosen by general practitioners and specialists as these two groups were the main authors of the guidelines).

What happens if we take the lay public's view of when treatment is appropriate? We would not recommend treatment for any non-diabetic woman under 60 or any non-diabetic man under 50. (None is at sufficiently high risk for us to believe that they might want treatment.) We should therefore perform routine blood pressure checks in these people only if we explicitly know that they would want treatment at lower risk thresholds than most of the population would. In other words, we need some form of informed consent to screen such patients routinely.

The findings of this paper need to be confirmed.