Quality assessment of clinical practice guidelines: who watches the watchmen?

Guidance was once such a simple thing: provided by a valued few to those who sought to learn.Knowledge, experience and wisdom were rare commodities, the topics to be discussed limited.The world has turned.Information is now abundant, healthcare data are widespread.Gradually, it has become impossible to read, appraise and interpret all published material, even within subspecialty fields.In an attempt to deal with this issue, guidelines have increasingly found their way into clinical practice and regulatory decision-making.As defined by the Institute of Medicine, clinical guidelines are 'systematically developed statements to assist practitioner and patient decisions about appropriate health care for specific clinical circumstances'. 1 In other words, guidelines should safeguard users and patients from wrong decisions.Although research supports their ability to do just that, 2 in reality, there have been many that were contradictory, of poor quality, impractical or simply not evidence based. 3It has caused clinicians to be wary of them.Indeed, they have brought forth a new dilemma: if the guideline declares who is right and who is wrong, how then do we assure that the guideline itself is right?In this month's Journal, Hazlehurst et al. provide a comparative assessment of guidelines produced by five major professional societies in endocrinology.The authors assessed a basic aspect of guideline quality: the methodology by which individual guideline groups rated the level of evidence underpinning their recommendations.They found that of the 58 guidelines published between 1995 and now, only 29 explicitly mentioned that a grading system was used at all.Overall, only 6Á5% of recommendations were based on 'high level' evidence.In addition, different organizations used different systems, with different criteria for ratings of quality of evidence, leading to potential confusion and misunderstanding in the clinicians using them.Clearly, a valid, standardized system that is applied consistently would be the ideal.