Focus on Statistics and Epidemiology
Marc Bonten speaks about his own mistakes and how he realized that one of his published studies was underpowered. He also talks about a study to demonstrate the efficacy of selective antibiotic decontamination in the ICU.
![Author profile image](https://pbs.twimg.com/profile_images/1389840482046410754/9BcxGjhZ_normal.jpg)
Elisabeth Bik
Science integrity consultant, PhD. #ImageForensics Currently without cat or blue check, but soccer-mom of godson. She/her. https://t.co/V4MR7WVHiT π³π±πΊπΈ
![Header media for article](https://pbs.twimg.com/media/Ft5t0jpXoAEJf8l.jpg)
-
The next speaker is @MarcBonten with 'Focus on statistics and epidemiology'
— Elisabeth Bik is at the Thomas More campus in Geel (@MicrobiomDigest) April 17, 2023
He starts by saying that he will only talk about his own mistakes, so he will not to be insulting anyone else :-)#ECCMID2023 pic.twitter.com/J7hV4W7x8N -
MB: One of my older published studies was underpowered, and it took me several years to realize that it was not very good or conclusive. #ECCMID2023
— Elisabeth Bik is at the Thomas More campus in Geel (@MicrobiomDigest) April 17, 2023 -
MB: We did a study to demonstrate the efficacy of selective antibiotic decontamination in the ICU, with three interventions, 13 ICUs, cross-over design. Around 6000 patients. Initial results showed no effect of decontamination. #ECCMID2023
— Elisabeth Bik is at the Thomas More campus in Geel (@MicrobiomDigest) April 17, 2023 -
MB: But some clinical parameters of the three groups were slightly different. We used Cox regression, but should have been random effects logistic regression. We had used the wrong statistical test.
— Elisabeth Bik is at the Thomas More campus in Geel (@MicrobiomDigest) April 17, 2023
Now, results were statistically significant.https://t.co/ktvRq9xPsc#ECCMID2023 -
MB: In another study, we found - after publication - that one of the participating ICUs had switched intervention codes (0 or 1). After reanalyzing the result, the outcome was very different. We had to retract and republish that study.
— Elisabeth Bik is at the Thomas More campus in Geel (@MicrobiomDigest) April 17, 2023
Read: https://t.co/LsCbbovsh8#ECCMID2023 -
MB: As medical doctors, we need to work with statistical experts to avoid making mistakes. #ECCMID2023
— Elisabeth Bik is at the Thomas More campus in Geel (@MicrobiomDigest) April 17, 2023