[Obligation of guideline recommendations: Perception survey among guideline developers]

Specific recommendations form the centerpiece of medical guidelines. The intended strength of a recommendation is usually expressed by the use of different wordings. Recent investigations showed that guideline users (recipients) interpret the binding character of wordings in different ways. The perception of different wording and their binding character of the strength of guideline recommendations among guideline developers (sender) have so far not been investigated in the German-speaking area.

German-speaking guideline developers were invited online and at specialist meetings to participate in a survey investigating 13 different terms used in guideline recommendations. The aim was to measure their perceived binding character on a visual analog scale (VAS: 0-100). Additional demographic and occupational data were collected. The results were compared with data from a previous survey among guideline users.

Overall, 136 guideline developers with an average of 4.2 (median 3) guideline (co-) authorships participated in the survey. While guideline developers interpreted the two imperative recommendations "must" and "must not" with a similarly high level of obligation, the level of obligation was not rated homogenously for strong and weak recommendations like "shall" or "should". Two out of five negative formulations were perceived as more binding than their corresponding positive formulations. In comparisons with the ratings of the guideline users the terms "darf nicht" (must not), (-5.8 VAS, p≤0.0001) and "muss" (must), (-2.9 VAS, p≤0.0006) were perceived as less binding by the guideline developers, as was "sollte nicht"(should not) (-6.6 VAS, p≤0.0001) and "kann nicht empfohlen werden" (cannot be recommended) (-9.4 VAS, p≤0.0001). Only "soll" (shall) was perceived as more binding (+8.7 VAS, p≤0.0001).

The current survey demonstrates that guideline developers perceive the binding character of terms used in guideline recommendations in a similar way as it was found for guideline users in a previous survey. Negative recommendations are more often rated as more binding by guideline developers as the corresponding positive terms. The discordant ratings by the guideline developers can be explained by their methodological knowledge regarding formulations of recommendations. It would be desirable that efforts be made to identify wordings in guidelines which better discriminate the intended level of obligation of guideline recommendations with regard to these results.