This is causing a problem for me as well. I was relying on "In Bed" samples to group sleep samples into different sessions; without the "In Bed" samples, there is no easy way to determine when one sleep session ends and another begins. It seems like a conscious decision by Apple but I hope it is a bug, or that they reconsider this change. In the meantime, does anyone have a suggestion of another way to group sleep samples into separate sleep events?

@DTS Engineer that is very helpful, thank you so much. There is a walk workout that overlaps with these step samples, so I believe you are correct. I have a related question for you if you don't mind. I am querying HealthKit step data using an HKAnchoredObjectQuery. In the initial query, I pass a predicate to only get step samples that happened in the past couple days. Once I get back a valid anchor, I tear down that initial query and set up a new HKAnchoredObjectQuery with the anchor and no predicate - that way I can receive any new step samples that get created, no matter the start and end dates. For some reason, the anchor I get back in the initial query doesn't contain these "condensed" samples. So when I do the new query using the anchor, the resultsHandler receives these condensed samples, even though they are older than the anchor. I would expect that the condensed samples would be included in the samples the initial anchor covers, so that they wouldn't be returned in a query that uses that initial anchor. Is this a bug, or am I misunderstanding how anchors work?