The last week of November was Evidence Week at the Royal Melbourne Hospital (my place of work). I took part in some the planning for it and participated in 3 events: the EBP Workshop, the Study Errors lecture and the grand rounds session on using evidence in practice. Some of the presentations are available by clicking here. The first session I went to was the hour long lecture about study errors with presentations from Prof Paul Glasziou, A/Prof Peter Morely and Dr Danny Liew. A small number attended – perhaps the title of the lecture could have been more encompassing? The lecture is free to listen to by clicking on this link (the squeeking is Liew writing on the whiteboard!) . Anyway, here are some of my notes from this session:
- The errors most often encountered are Type 1 (false positive) and Type 2 (false negative) errors. Type 1 errors can be demonstrated in a 2×2 table (see study error eg) with the real situation being the middle line between them. Type 2 errors occur when the study is underpowered.
- Errors can occur if a trial has a primary outcome but includes secondary outcomes that are also of interest
- A smaller effect needs a bigger sample while a bigger effect needs a smaller sample
- There is more chance of Type 1 error when there are multple analysis (hypotheses) being made using the same data. Eg: 1-0.05=.95 CI. Per analysis multiply the CI by how many analysis there are -> 2nd analysis 0.95 x 0.95, 3rd analysis 0.95 x 0.95 x 0.95 etc
- Systematic reviews and meta-analysis are retrospective. Enhance these reports with preliminary data in trial registries
- Non-inferiority trials are not that useful because really, pharma want to prove that x drug is way better than y drug so …
- The best way to get safety info is to obtain post-marketing surveillance reports. Safety is not able to be tested in trials as trials are not long enough
- In surgery, the evidence evolves slowly over time
- You can’t control for all conditions in trials as there isn’t enough known about the human body
The following day I went to the EBP Workshop led by Prof Glasziou. I was to help with searching exercises but things didn’t work out. Like the EBP Residential Workshop the week after, the day was split into large group lectures and small group activities. This arrangement works very well with these type of workshops. I was very lucky to be in Glasziou’s group and there was an HTA colleague who also ended up in my group (he is into population health but hasn’t done much appraisal – and of course we talked about HTAi2012 …). There were a lot of librarians in the audience and the searching for lit was not extensive as we would like, but then it was a 1-dayer and there were other clinicians in the group who needed this background. However, I got something out of it because I had totally forgotten about the ability to add your own filters in My NCBI. The appraisals were interesting and the main point is – practice!! The more you do the easier it will be. There was a handbook given out as well: Evidence-based Practice Workbook – Bridging the gap between health care research and practice. It covers asking an answerable question, searching (screen shots need updating – PubMed has changed it’s layout again!), appraisal and applying the evidence.
The final event was the Grand Rounds lecture. It was slightly different from Glasziou’s Cochrane Symposium lecture (July 2011) which was how GPs need evidence presented to them with instructions for integration. This was more about the amount of information available as opposed to evidence, what different specialities need in terms of research evidence, and thoughts in how to integrate new technologies/change practice.
Presentations from the EBP Workshop and the Grand Rounds lecture is available on the RMH Health Sciences Library website.