AbstractAim: PRECIS (PRagmatic Explanatory Continuum Indicator Summaries 2009) is a tool with a simple wheel format that trialists can use when designing their trials to improve the applicability of results but users highlighted problems. The aim of the study was to produce an improved and validated version of PRECIS, called PRECIS-2 and test this tool out with trial teams designing primary care trials.
Methods: Brainstorming and a 2-round Delphi survey of authors who cited PRECIS plus user-testing of candidate PRECIS-2 models was followed by validity and reliability testing of the most promising PRECIS-2 candidate using a sample of 15 trials rated by 19 different trialists. The validated PRECIS-2 tool was then used to consider the risk of bias (internal validity) and estimates of treatment effect of a matched set of explanatory (ideal conditions) and pragmatic (real world) trials. The PRECIS-2 website was also created with a database of pragmatic trials and a toolkit for trial groups. This was tested out at the Pragmatic Clinical Trials Unit (PCTU) in London with trial teams designing primary care trials.
Results: Forty-two people responded to the Delphi and highlighted scoring, domain choice, and tool format as issues. An expert panel of 14 in Toronto provided the basis for a PRECIS-2 model that was then user tested by 19 other methodologists and trialists. After 13 iterations, a PRECIS-2 model with 9 domains (i.e. Eligibility, Recruitment, Setting, Organisation, Flexibility Delivery, Flexibility Adherence, Follow up, Primary Outcome, Primary Analysis) was tested for validity and reliability. Inter-rater reliability was generally good, with eight of nine domains having an ICC over 0.65. Discriminant validity was reasonable for all domains, though with wide confidence intervals. Matching trials taking pragmatic (‘real world’) and explanatory (‘ideal world’) approaches was challenging but we found no indication that a pragmatic approach compromises internal validity. We were unable to extract sufficient information for a planned analysis of estimates of treatment effect. At the PCTU, the tool highlighted differences in opinion with trial team members and demonstrated convergence of opinion following discussion. There was acknowledgment that scoring of PRECIS-2 domains assisted trials teams in considering the intended audience and creation of trials relevant to practice. Useful feedback was obtained to improve the PRECIS-2 tool software for users.
Conclusions: PRECIS was improved by the addition of scoring and additional domains after consultation with over 80 international trialists. We have a validated PRECIS-2, in the visually appealing wheel format with 9 spokes, which is being made available through an increasingly accessed website. Work at the PCTU improved the usability of the PRECIS-2 website and demonstrated that the tool increases transparency in trial design and assists trialists in considering applicability of trial results. More matching work on the impact of design approaches on effect size is needed, and further data to support the risk of bias results would be valuable.
|Date of Award||2015|
|Sponsors||Medical Research Council & Chief Scientist Office|
|Supervisor||Peter Donnan (Supervisor), Frank Sullivan (Supervisor), Shaun Treweek (Supervisor), Merrick Zwarenstein (Supervisor) & Shaun Treweek (Supervisor)|
- Trial design
- Clinical trials