Impact of clinical practice guidelines on clinicians' behaviour: tonsillectomy in children

This study was conducted to assess the extent to which developing and implementing clinical practice guidelines for listing children for tonsillectomy (with or without adenoidectomy) influenced the behaviour of participating ENT surgeons.

A before and after study in which the intervention (the development and dissemination of local practice guidelines) was introduced sequentially into different hospitals and surgical practice. The study was conducted in four ear, nose, and throat surgical services in the North of England, with 16 consultant ENT surgeons and their junior staff on 1190 children aged 0 to 14 years who were judged, prior to consultation with an ENT surgeon, to have been referred for throat-related problems for which tonsillectomy was one possible treatment option were included in the study. Decision reached by surgeons and proportion of decisions that complied with new guidelines.

Of the clinical decisions to list children for tonsillectomy taken before introduction of locally agreed guidelines, 73% (486/660) conformed to the criteria in the subsequent guidelines, 15% (97/660) did not, and in 12% (77/660), it was impossible to judge. After the intervention, the corresponding figures were 73% (386/530), 14% (73/53), and 13% (71/530), respectively. When decisions were taken to break the guidelines, this was more often to list for tonsillectomy when it was not indicated--83% (141/170)--than to withhold tonsillectomy when it was indicated--17% (29/170). The aspects of guidelines that were breached in decisions to carry out tonsillectomy were: the age of the child was younger than the guidelines recommended--54% (75/141); there had been fever attacks of tonsillitis than the guidelines recommended--22% (32/141); and there were "significant" symptoms not included in the guidelines--20% (29/141).

In spite of strong evidence to the contrary, local guidelines were formulated at a level that the majority of surgeons already attained. Guideline development and implementation, therefore, had very little impact on clinical practice. The process of local formulation of guidelines was not sufficient to achieve change toward evidence-based practice; clinical preference proved to be quite intractable. There is a need to enhance the ability of clinicians in the assessment and interpretation of research evidence. Previous work has emphasized the need to explore factors that influence clinical behaviour toward evidence-based practice. Our study suggests the need for more research into why clinicians continue to follow clinical preference even when invited to base agreed local clinical policies on evidence.