As a trialist, the pressure of working on a trial is much bigger than being in a small group educational session. Challenges of implementing a trial are multiple, mainly influenced by the values of outcomes for different people. Whose question is the trial answering? If you’ve ever found yourself puzzled by these issues, you may find some solace in reading my notes from a courseon clinical trials.
7 instructors and 21 participants – all from Northern Ireland (except 2 Dubliners), 2 medics, three 1-st year PhD students and some professors – talked about clinical trials for two days last week at Queens University Belfast. The aim of the course wasn’t to learn everything, but to think laterally about trials. Professor Clarkecovered the basics of starting trials: formulating a clear research question, deciding on comparisons and placebos and dealing with confounding factors. The 7 main ways of dealing with confounding are:
- Matching
- Exclusion
- Stratified sampling
- Standardisation
- Multivariate modelling
- Randomisation
The pleasures and terrors of trial recruitment were described by Dr Maguire. Everybody struggling with meeting the recruitment targets should read the top 10 tipsfor recruiting into trials at the All-Ireland Hub for Trials Methodology Research website. Trialists should plan for what they’re going to do if things don’t go the way they planned. Recruiters can also become tired and it’s good to think ahead about what would possibly prevent them from recruiting. Even small rewards to recruiters, such as cream eggs, can increase their satisfaction. Satisfaction=Retention. Research networks for General Practitioners can facilitate recruitment.
Dr McAneney introduced us to the role of social networks in clinical trials. We are all connected. All the users of Facebook can be linked by 3.74 steps. Networks make the trials work or crash. Networks allow diffusion of innovation. Decisions of participants and researchers are influenced by networks.
Prof McAuleyhelped the participants to write the protocol and funding application for their first trial. Publishing a trial protocol sets the bar pretty high for researchers – transparency and accountability are key. If it’s not possible to publish the protocol in a peer-reviewed journal, then post it online. Every protocol is changed over time and they should be listed on the first page. The CONSORT diagram is an essential part of a protocol. It’s the only slide that’s projected during meetings of grant reviewers.
Dr Shorterand Prof Buntingcontinued with tips for analysing outcomes. The essence of any research is control. Although power calculations for trials seem difficult, they involve only a short sequence of basic steps. Categorical outcomes require more data and more participants than continuous outcomes. Analysis of clinical trials assumes that our participants are all from the same population. The classical assumption of trials analysis was that individual differences do not matter, they were ignored. Another assumption that things are measured perfectly never holds.
Finally, Dr Dunlopfinished the course presentations with ethics and data storage.