One very synthetic way to represent a night's sleep is by way of a hyp
nogram: a graphical representation of the sleep stages as a function o
f time. The hypnogram is generally quantified by a series of variables
that measure the durations and latencies of the various sleep stages
including wake. These variables, however, do not fully account for all
the information contained in the hypnogram, in particular information
on sleep continuity. A series of variables that measure and localize
disruption of this continuity are proposed and their utility validated
on three groups of patients presenting sleep disorders. Utility is es
tablished if the variable is capable of differentiating between patien
ts and healthy controls. Two sets of variables are examined: those tha
t use the entire sleep period as unit of measurement, and those that a
re measured within each consecutive NREM-REM sleep cycle. The results
show that the variables proposed are able to differentiate between gro
ups and, therefore, are useful measures reflecting the hypnogram more
precisely. They also show that fragmentation of REM sleep does not pre
sent a systematic trend across the night, but that fragmentation of NR
EM sleep goes up linearly.