“Randomised control trials (RCTs) have played a role in the assessment of surgical innovations and there is scope and need for greater use”. While common in other areas of medical research, RCTs are often under used in evaluating surgical interventions due to the practical and methodological issues they present researchers.
By far one of the most vexing issues that RCTs throw up is the so called ‘clustering effect’. The experience, training and level of practice possessed by a surgeon means multiple patients operated on by the same surgeon often experience similar outcomes to their procedures. This phenomenon, known as ‘clustering’, can lead to a loss of precision when evaluating the results of RCTs and means that researchers must be extremely careful when choosing the sample size of their studies.
A new study published in Trials describes the creation of a database of previous surgical trials to quantify clustering effects at both institution and surgeon levels. Calculating the intracluster correlation coefficiencies (ICCs) in 10 multicenter surgical trials for a possible 108 outcomes, the authors found evidence for a clustering effect in a large number of possible outcomes. Lead investigator and Editorial Board member of Trials Jonathan Cook noted, “Our data on clustering effect for multicentre trials of surgical interventions suggests it is more of an issue than has previously been acknowledged.”
At the moment researchers have a shortage of data on which to assess the impact of clustering. This inability to judge the level of clustering present in RTCs makes it difficult for researchers to adjust their trial designs to compensate for any loss of precision. The authors hope that by adding datasets from future surgical trials to the database, researchers will one day be able to access a valuable resource that will help to inform and improve the design of surgical trials.