Over the past 15 years, there has been considerable debate concerning
the extent to which insomnia patients can be classified into diagnosti
c subtypes. Despite this debate, relatively little research has been c
onducted to empirically determine whether naturally occurring insomnia
subtypes might be identified within populations of sleep clinic patie
nts. In the current study we used a hierarchical cluster analysis to e
mpirically identify subtypes among a mixed group of normal sleepers an
d the insomnia outpatients who presented to our sleep center over the
past decade. Using factor-analytically derived composite variables tha
t summarized data obtained from sleep history questionnaires and polys
omnographic monitoring, this clustering procedure resulted in the iden
tification of 14 subgroups that varied between four and 34 patients/su
bjects in size. Subsequently, subgroup mean scores for the composite v
ariables used in the clustering procedure were used to construct profi
les for each of the 14 clusters. A multivariate profile analysis, empl
oyed to elucidate subgroup differences, showed that these cluster prof
iles differed in terms of their configural shapes, average elevations,
and degrees of interscale differences. Furthermore, both DSM-III-R (A
merican Psychiatric Association) and International Classification of S
leep Disorders (ICSD) insomnia diagnoses, assigned independent of clus
ter Endings, suggested that these subtypes differed significantly in r
egard to their diagnostic compositions. Nevertheless, a far-from-perfe
ct concordance was observed between such clinically assigned diagnoses
and cluster group membership. In fact, many of the empirically identi
fied groups were composed of various DSM-III-R and/or ICSD diagnostic
subtypes. These results provided only partial support for current DSM
and ICSD insomnia categories. However, our results support the existen
ce of multiple, clinically discrete insomnia subtypes and provide info
rmation that may be useful in future revisions of current insomnia nos
ologies.