j.h.p.m.van_der_velde@lumc.nl
26 november 2020 bij 18:08- Antwoorden
Hi, thanks for this clear presentation. I have some questions/comments :
– did you adjust for total activity, maybe people in certain clusters have become more active over the entire day?
– were the associations for the outcome HOMA IR adjusted for BMI or body fat%?
– did you also look at activities at different intensities instead of total activity?
Sorry for the many questions, but since I’m working on something very similar in the NEO study and actually found the opposite regarding HOMA IR, it would be great to exchange some thoughts.
G. Albalak (Leiden)
26 november 2020 bij 18:30- Antwoorden
Hi Jeroen, thank you for your questions!
– We did not (yet) adjust for total activity, that is however something that we have been thinking about for a while. The problems conserning this are that we are not sure if it’s even possible to adjust for total PA after doing a principal component analysis. Secondly, we were also struggling with which value we should use for this adjustment since using hourly data on acceleration causes absolute values of activity to be lost.
– We did not adjust HOMA-IR for BMI. However, we did adjust de associations with fasting insulin and this significant association remained the same after adjustment.
– Our approach was to step back from using summary data (light, moderate and vigorous activity) since we feel that you might loose some information with that. Instead we chose to use raw acceleration data that, in the end, can be translated back to these summary denominators by looking at the threshholds for acceleration.
I hope that this answers your questions! If you are interested, we can have an offline discussion about our findings.
j.h.p.m.van_der_velde@lumc.nl
26 november 2020 bij 18:34- Antwoorden
Hi Gali, Thanks for the response. It would be nice to continue discussing at a different medium. I’ll send you an email later
JF. Harmsen (Maastricht)
26 november 2020 bij 19:16- Antwoorden
Hi Gali, thanks for the clear presentation.
Jeroen already knows my question which I also had for his data 🙂
Do you have any information on chronotype of your subjects? I assume not, right?
On average, based on your figures on slide 7, your population seems to have an early to intermediate chronotype.
Subjects that were active at PCA2 (0:00-4:00), who tend to have a higher BMI and higher HbA1c, are most likely late chronotypes, right? Subjects that are most active at PCA3 with lowest HOMA-IR are instead most likely early chronotypes.
I think there would be merit in getting some chronotype information of these people. If you have accelerometer data during their sleep phase, you could take the midpoint (clock time) of the sleeping period on weekend days, this would give you a good indication of their true chronotype.
G. Albalak (Leiden)
26 november 2020 bij 19:35- Antwoorden
Excellent point! This is something that we were thinking about ourselves as well. I think it would be great to connect our findings to certain chronotypes. We are definitely going to consider your input.
Thank you.
5 Reacties
Hi, thanks for this clear presentation. I have some questions/comments :
– did you adjust for total activity, maybe people in certain clusters have become more active over the entire day?
– were the associations for the outcome HOMA IR adjusted for BMI or body fat%?
– did you also look at activities at different intensities instead of total activity?
Sorry for the many questions, but since I’m working on something very similar in the NEO study and actually found the opposite regarding HOMA IR, it would be great to exchange some thoughts.
Hi Jeroen, thank you for your questions!
– We did not (yet) adjust for total activity, that is however something that we have been thinking about for a while. The problems conserning this are that we are not sure if it’s even possible to adjust for total PA after doing a principal component analysis. Secondly, we were also struggling with which value we should use for this adjustment since using hourly data on acceleration causes absolute values of activity to be lost.
– We did not adjust HOMA-IR for BMI. However, we did adjust de associations with fasting insulin and this significant association remained the same after adjustment.
– Our approach was to step back from using summary data (light, moderate and vigorous activity) since we feel that you might loose some information with that. Instead we chose to use raw acceleration data that, in the end, can be translated back to these summary denominators by looking at the threshholds for acceleration.
I hope that this answers your questions! If you are interested, we can have an offline discussion about our findings.
Hi Gali, Thanks for the response. It would be nice to continue discussing at a different medium. I’ll send you an email later
Hi Gali, thanks for the clear presentation.
Jeroen already knows my question which I also had for his data 🙂
Do you have any information on chronotype of your subjects? I assume not, right?
On average, based on your figures on slide 7, your population seems to have an early to intermediate chronotype.
Subjects that were active at PCA2 (0:00-4:00), who tend to have a higher BMI and higher HbA1c, are most likely late chronotypes, right? Subjects that are most active at PCA3 with lowest HOMA-IR are instead most likely early chronotypes.
I think there would be merit in getting some chronotype information of these people. If you have accelerometer data during their sleep phase, you could take the midpoint (clock time) of the sleeping period on weekend days, this would give you a good indication of their true chronotype.
Excellent point! This is something that we were thinking about ourselves as well. I think it would be great to connect our findings to certain chronotypes. We are definitely going to consider your input.
Thank you.