Last week I set off for my first ever academic conference. The ‘first’ wasn’t just for me – this was the inaugural MRC Population Health Methods and Challenges conference, so newness all round. In the course of 48 hours, over 100 speakers imparted their insights into the methodological issues they grapple with, and the solutions on which they work.
Two things I was particularly heartened by: firstly the number of people who, like me, came under the sub-theme ‘Making best use of routine data’. There are a lot of us about, hacking away at large databases to search for the gems among the rubble. There are plenty of discoveries to be had, but patient and systematic sifting is required. The second thing is a related matter – the amount of care and attention being put into trying to make sound inferences from observational data. Trial enthusiasts sometimes look down their noses at observational data, but not everything is amenable to an RCT approach, and we risk missing out on opportunities for early detection of signals from the increasing large quantities of routine data being amassed if we neglect observational approaches. They have their limitations, but when thoughtfully applied they can still yield potentially useful insights.
I don’t know if I’m allowed to say this but I actually got more out of the parallel sessions than the plenaries. Possibly because of the nature of the conference – the parallel sessions were where the really nitty-gritty details came out, imparted by people who were clearly up to their elbows in data cleaning and validation on a regular basis. That’s not to say that broad overviews of bigger areas aren’t valuable, but on this occasion I found that the bite-size talks delivered by early career researchers gave me more food for thought with respect to things I might like to try with my own data.
Talks which particularly piqued my interest: James Fagg (Institute of Child Health, UCL) and Catherine Welch (UCL) both spoke about multiple imputation in complex data structures (hierarchical and longitudinal, respectively). Working with large data sets myself I’m very aware of the potential for structural variation in the data and the need to account for this in the data handling and analysis. There was also an interesting series of talks by Linda Wijlaars (UCL), Lisa Szatkowski and Tessa Langley (both University of Nottingham) on trying to model the impact of changes to policy and clinical guidelines on a range of health outcomes. Given that policy decisions are rarely implemented via controlled trials, evaluating the real world impact is important and should be done more frequently.
Any criticisms? There was some discussion about whether the 10 minute slots in the parallel sessions were too short for speakers to convey much detail, but most handled it very well, and it was long enough to showcase approaches and prompt further discussion in the coffee breaks. Some commented that there was a shortage of qualitative research – perhaps the call for abstracts wasn’t heard in all the right places. That should be fixable in the future. My only major gripe is that despite more than half the attendees being women, all five plenary sessions were delivered by men. I find it hard to believe there weren’t any female senior academics available to share their perspectives and experiences.
Overall, it was an interesting and enjoyable conference. I’d recommend it, particularly to early career researchers looking to dip their toe in conference waters, or keen to meet other people grappling with methodological issues. I hope it returns in future years, as I think it definitely fills a niche and provides a useful platform for sharing ideas.