Deep sleep tracking changes if I add more sleep

[rockpool]

Several times I have noted that the sleep curve for the whole night changes if I add more sleep later in the morning.

Last night I slept particularly badly as I was trying out a new device to minimize my snoring. 

Around four thirty I decided to remove the annoying device and get out of bed for a while. During the time I was up I checked my sleep tracking. It turned out I had actually slept for between 4 and 5 hours but had zero (!) minutes of deep sleep. Typically I get all my deep sleep, which is normally between 16 and 50 minutes, during the first hours of the night. After a while I went back to bed and managed to get all in all an extra three hours sleep (without the device this time). 

When I checked my sleep tracking next time it claimed I had 16 minutes deep sleep around 1 o'clock. My question: How could the fact that I added three hours sleep towards the end of the night increase my deep sleep several hours earlier? 

Moderator edit: format. 

[SunsetRunner]

I hope you get a good answer to this, since I have noticed the very same thing happening.

[rockpool]

The same thing happened again this morning. I woke up around 5 o'clock after 7 hours sleep. Got out of bed. Tracker said 16 minutes deep sleep during the first hours of sleep. I then went back to bed and slept another hour. The tracker now says I got 57 minutes (!) deep sleep during the first part of the night. 

I hope someone at Fitbit can explain this. 

[SunsetRunner]

It happened again, It took away REM this time. I woke up from a strange dream around 4:30, I took a look at the app to see how long the dream might have been. I noted two short REM cycles shortly before 4:30. I didn't pay attention to deep sleep time. After a bathroom trip I went back to sleep. Now, waking at 6:30, I see those two REM periods gone completely!  I'm going to have to start taking screen shots to email to fb support.  

[rockpool]

It would be interesting if some more people could check out how this works on their own trackers, i.e. if the details of their sleep tracking for earlier parts of the night change when they add more sleep later in the night. 

At any rate I think someone at Fitbit has some explaining to do. How and why does this happen???

How are we supposed to have any trust at all in the detailed sleep tracking if new data can totally change old data?     

[WavyDavey]

I find by trying to check sleep stages during the night usually results in odd readings.  In order to determine sleep stages, it has to run calculations across the sleep period. When you alter the sleep period, it can throw the calculations off. Letting the calculation happen when sleep is over instead of forcing it during the night gave me better results.

[rockpool]

But what if I had got out of bed at 5 o'clock that morning without sleeping another hour? In that case I would have believed that I had only had 16 minutes of deep sleep that night. Would I have been right or wrong?

Sometimes I get up around 6 o'clock thinking that this is it. No more sleep tonight. I have breakfast, read a bit, and then, if I feel sleepy and have the time, I go back to bed and sleep for another 45 minutes or so between 8 and 9.  Then when I add this sleep manually the same thing happens. 

Does this mean I should not trust the reading if I add sleep?

[Rich_Laue]

I and others have found that a forced sync at night gets interpreted as being awake and messes up the sleep schedule. 

My thoughts are not to look at the app until the wearer has been awake for at least 5 minutes. This gives enough data so that Fitbit sees the user is up and moving 

[rockpool]

Ok... I'll try that. Thanks for the tip, I'll try it out and see if it changes things. 

[rockpool · Accepted solution]

I think you may have cracked it Rich_Laue. I followed your advice twice, and the first reading from around 6 o'clock is now stable, even after I added some more sleep later in the morning. The second sleep now comes up as a different sleep without details, but on the dashboard it is counted as part of the night's sleep. I can live with that 🙂

Hope this works for you too, Sue.

Thanks for your help!

[Alclaib]

I average 1.5 hours of REM sleep and 1.4 hours of deep sleep every night.  This was only achieved after I received a sleep study and corrective sleep apnea therapy.  

[akrep_nalan]

I landed on this topic because something similar has been happening on my device--deep sleep appearing where it hadn't been when I first checked sleep stages. I've been noticing one morning later, when I check that night's sleep, that previous readings have changed (always adding more deep sleep--so 37 minutes might become 50-something). For me it's not a matter of running the analysis and then getting more sleep (except in the sense that I'm getting sleep 14 hours later). Will try Rich_Laue's "wait five minutes before syncing" suggestion for a week and see if that seems to lead to more stable readings (and will either add to this thread or, if it doesn't, open a new one). Thanks!

[rockpool]

Actually, lately I have noted that whenever I edit sleep (falling asleep or wake up time to reflect my own recollection), the whole record for the whole night changes entirely. Not just deep sleep, but everything. Rem sleep, deep sleep, light sleep, awake time, everything. I have made enquiries about this in a couple of threads, but no one has offered any  sort of explanation or even acknowledged a problem so far. 

I always follow the advice to wait for a while (often a couple of hours) before I check the sleep tracking for the night, so it is not that I confuse the tracking by being too quick.

[topannuity]

Hi Rockpool- I am new to FB. I notice that if I edit wake up time to cut off the added few minutes between actual wakeup and when I click the "I'm awake" button, that action alters the analysis too.

[rockpool]

I am glad you posted this, toppannuity. Other people have experienced the same. To my knowledge Fitbit has not offered any  explanation or even acknowledged the problem. My theory is that it is some kind of bug. I have long since stopped using the edit sleep function.

[Rich_Laue]

I'm patiently waiting for before and after screen shots.

[rockpool]

Rich-Laue, thanks for looking into this. Just to see the result, I edited today's sleep (even though there was really no point, since I was quite happy with the start and end times). This time the difference after editing was not dramatic, but there is still a difference in minutes in deep sleep (from 21 to 12 minutes).

On a side note, the  sleep recording that I had done at a hospital a few months ago showed a considerably different result from Fitbit all over (especially with regards to deep sleep, where the hospital result said I had a lot more). But the issue in this topic was the fact that the Fitbit sleep pattern invariably changes when I edit the falling asleep or wake up time. These days I tend to ignore Fitbit's recording of the different kinds of sleep altogether and look more at it as a record of the numbers of hours I have been asleep. I do find that Fitbit is usually really good at identifying when I am asleep and not. 

[jrmnr]

The sleep lab indicated that I spent far more time in deep sleep than the Blaze.

Also, there is a considerable amount of time I lie quietly in bed listening to the radio and Blaze mistakes it for light sleep.

I do not trust Fitbit sleep tracking.

[Scott-GT]

Did this ever receive a response? 

I have found that editing the back end of my sleep time in the morning causes the app to change my entire night. Even if I’m only clipping off a few minutes of being awake and walking around the house. It causes a serious lack of confidence in the accuracy of the tracking.  

[rockpool]

Scott-GT, no I don´t believe this issue ever received a response. I was asked to produce screen shots, which I did, but I believe I messed them up and sent the same shot twice. Personally I have stopped using the change falling asleep or waking up function (I haven´t felt the need, since the recorded times have seemed correct), so I cannot say for sure that the same thing still happens, but it certainly did when I tried to send the screen shots. Perhaps they will answer if you send them proper screen shots of an instance when the change occurs.