Correspondence (reply): In Reply

A number of different factors will affect how patients assess any risk associated with a possible treatment. This risk is being discussed with the doctor first of all. As Dr Meyer observes, communication with the patient can be expedient only if the treating physician himself/herself is able to assess the risk correctly. Risk communication should be the subject of intensive training as early as during undergraduate medical training. The crucial element is the time available for the discussion between doctor and patient, since risk communication is a difficult subject and is also time consuming.

In addition to this, a fundamental problem presents itself when using ordinary language to describe probability terms, in that such terms are used differently in different contexts. The description that deaths from a disease are “rare” is interpreted differently when used in ordinary language than the product information, that headache as a side effect is “rare” in the context of the administration of a medical drug. This contextual dependency is very difficult to circumvent.

For the purposes of our article, we intentionally restricted ourselves to investigating whether persons with regularly responsibility for risk communication assign the correct probabilities to the terms used on package inserts. In this context, Dr Niederhofer comments that “This raises the question of how—in view of the fact that the information on the package inserts and its interpretation by experts threw up substantial differences—these results would have looked if patients had also be included in the survey.” This aspect has been the subject of comprehensive study in the literature. Several studies showed that the verbal descriptions were not assessed correctly. Furthermore, the risk of persons in the general population is being overestimated; in addition to the list of references that accompanies our article (1) (references 11–17) we wish to draw attention in particular to the publication by Fischer and Jungermann (2), which explicitly refers to German persons in the general population.

In our article we dealt exclusively with the problem of risk estimates in a simple medical context. Of course there is a whole series of legal and other requirements for package inserts. And it is obviously almost impossible to present package inserts neatly and make them short, and comprehensible while also including every single characteristic of the medication (3).

We wholeheartedly support Dr Niederhofer in that intensive efforts are required to ensure that package inserts for patients are conceived in a precise way but are still comprehensible and clearly presented. If the use of new media were to be permitted to support package inserts, then this might become possible.