Over 8% of chartevents have charttime > storetime

How do we interpret events where the observation/measurement occurs after it has been manually input/validated?

From the docs (https://mimic.mit.edu/docs/iv/modules/icu/chartevents/):

charttime records the time at which an observation was made, and is usually the closest proxy to the time the data was actually measured. storetime records the time at which an observation was manually input or manually validated by a member of the clinical staff.

Histogram of differences between charttime and storetime where charttime > storetime:

Start with the ICU chartevents documentation linked in the issue and compare its definitions of charttime and storetime with the reported histogram. Investigate why charttime can exceed storetime and document a clear interpretation for these records, including what users should understand about the timestamps.