Ultradian Cycles of Mood in Normal
and Depressed Subjects*
Donald P. Hal}! Jr., M.D. , David Benede k/ M.D.
and Aud rey Chang?
Ab stract
Objective: To study ultradian cycles ofmood in four inpatien t groups.
Me thod: W e assessed the mood q/ depressed and nondepressed subjects, on an hourly basis,
between 7:00 a.m. and 10:00 p .m. for two consecutive day s using a visual analog scale of mood.
Hourly mood scores.for each group were plotted against tim e.
R esults: W e.found ultradian cycles w ith mean length s q/3-4 hours in all groups. Th e majo r
depressive epis ode and adj ustment disorder depression groups, however, had ultradian cycles of
significantly greater amplitude than w ard staff:
Conclusions: Depressed patients demonstra te greater amplitude in ultradian cycles q/ mood
than nondepressed subjects.
U lt rad ia n cycles (cycles < 24 hours in len gth ) have bee n de monstrat ed in a
va r ie ty of physiological factors a nd m ent al fun cti on s of int erest to psyc hia try. Norm al
s ubjects have be en show n to d emon strat e ultradi an cycles of cog n it ive perform an ce
e fficie ncy (1 ,2), short te rm m em ory (3), slee p-wa ke fuln ess (4,5) and mood (6 ,7). Th e
nin et y minut e cycle of rapid-eye-movem ent slee p (RE M) is probably th e m ost
in te nsively st ud ied ul tradi an cycle in th e a rea of psych iatry. T his cycle has be en
report ed to be di srupt ed in d epressive di sorders (8,9). In a rece nt ex plo ra to ry study
by on e of the a u t ho r s (D.H.) a nd oth ers , it wa s d em on stra t ed t hat d ep ressed pati ent s
a lso ex hibit a n a b no r mal ultradian cycle of m ood (7) . It was re po rt ed t hat a d epressed
grou p ex h ibite d u lt r ad ia n cycles of signi fica n t ly grea te r amplit ud e t han a nond epressed grou p . Ba sed o n our r eview of th e lit erature, this was t he first report ed
st udy of ultradi an cycles of m ood in d epression .
In the presen t study, we test th e re lia b ility o f a n exp loratory study which found
ultradi an cycles o f g rea te r a m plit ude in d epressed g ro u p t han in a non-depressed
grou p (7). In this foll ow-up st udy, we hav e ex te nd ed mood assessment over two d ays
in o rde r to investig at e day to d ay sta bility o f cycles, an d have incl ud ed non-d epressed
and a dj ust me n t di sord e r wi t h d ep r essed m ood pati ent groups to ass ess sp ecificit y of
t he find ing .
'Su bm it t ed from Co m bat St ress C ent er, Fort Bra gg, N.C . a nd 2W a lte r Reed Army Med ica l
C en te r , Wa shingt on , D.C .
*T he firs t aut hor awarde d th e Am erican Psych ia t ric Association's 1994 Dist a Prod uct s
Resid e n t Resea rch Award bas ed on t his manuscrip t .
9
J EFFERSO N J O U RNAL O F PSYCHIATR Y
10
METHOD
Subjects
This investi gat ion was pe rfo rm ed on psychiat ric inpatient s a nd wa rd staff
volun tee rs from th e in pati ent psychiatric units of Walt er Reed Army M edi cal C ent er
(WRAMC) . T wo co n t ro l g ro ups we re u tili zed: one group of ward st aff (incl udi ng
me d ica l st ude nts, psyc hia t r ic resid e nts, and nu rs ing personn el) a nd a se co nd g roup of
non-depressed psych iatric in pa tie n ts (pr imarily com pose d of patient s recovering
from th ought diso rd ers). De pressed subjects were selected according to th e followin g
incl us ion criteria: p rese nce of a DSM IIIR-d efined major d epressive e pisode or
adj us t me n t di sorde r wit h de pressed mood , con t inuous presence on th e WRAJlvlC
in pa t ie nt psych ia tric wards for a t least two co nsec u t ive days and a willingn ess to
part icip at e. Po t en ti al su bjec ts we re excl ude d from eac h group if co nc urre nt ly di agnosed wit h borderlin e o r hist rionic pe rson a lit y disorder, Ax is III cond it ions beli eved
to be re lat ed to th e ir mood disorder, recent history of ha llucina t ions or th ou ght
disorders, or evide nce of wit hd ra wa l fro m psych oacti ve substances . No subj ect s
rep ort ed a histo ry of prem en stru a l synd ro me . The four groups co nsisted of: 12 wa rd
staff, 6 non-depressed psychi at ric pat ients, 8 pati e nt s wit h adjustm ent disorder with
de pressed mood , a nd 12 pati e nt s wit h major de pressive e pisod es (two mel an cholic
a nd t hree wit h psych otic fea t ures) . M al e to female ra tios were 10:2, 3:3, 4:4, a nd 4:8
for th ese groups, resp ec tively. Mean ages a re repo r t ed in Table I . The subjects we re
not aware of th e hypoth esis be ing tes ted. All subjects gav e inform ed conse n t.
Assessment qfiHood
Mood was assessed for eac h su bject o n an hou rly bas is, between th e hours of 7:00
a .m. a nd 10:00 p.m . o n two consecutive days. The hourly mood assessme nt was
TABLE I.
Means and Standard De viations (in parenthesi s ) fo r Age, Hamilton Depres s io n Scale
(H a m -D) , Vis u a l Analog Scale of Mood (VASM) , and Intra-Daily Variability
of Mood (I DVM) Scores
Mean VASM Scor e
Gro up
Ward Sta ff
Non-de pressed
Pat ien ts
Adj ustm ent
Disord er
Maj or
Depression
Age
Ham -D
Day 1
Day 2
27 (5.0)
n = 12
29 ( 10)
n= 6
33 ( 16)
n = 8
37 (18)
n = 12
1.0 ( 1.0)
n = 12
5.0 (3.6)a
n= 6
18 (5.4)b
n= 6
19 (3.4)C
n = 12
95.7 (19)
n = 12
95.7 (32)
n = 6
67.2 (24) a
n= 8
46.4 (24) b
n = 12
10 1 (23)
n = 11
92.7 (30)
n=5
80.8 (29)
n=8
46 .4 (24)b
n = 10
IDVM
Day 1
Day 2
15.8 (9.1) 13.3 (7.8)
n = 12
n = 11
19.2 ( 14)
15.5 (9.0)
n=6
n= 5
32.3 (lO) b 28.8 (9.8)b
n= 8
n= 8
20.1 (10)
15.8 (7.5)
n = 12
n = 10
"p < 0.05.
bp < 0.0 1.
c p < 0.00 I (significan t differe nces compared to ward staff using Duncan's multiple com parisons test) .
ULTRADIAN CYC LES OF MOOD
II
performed using a sing le qu estion visual a nalog scal e of m ood (VAS M). Th e VASM
has be en demonstrat ed to be a val id and reliabl e m ethod for th e re peated ass essment
of mood (10 , II ) . Th e VASM was com pose d of a 150 mm lin e with th e le ft pol e labe led
as " wo rs t sad " and the right pol e label ed a s " be st happy." Subj ect s were instru ct ed to
place a m ark on this lin e whi ch best d escribed their cu r re n t m ood stat e . Each
subj ect 's VASM score s (16 hourly scores eac h d ay) were averaged to yie ld a daily
m ean VASM score for eac h subj ect. The a m ou n t of mo od variat io n within t he d ay, or
in t ra -d a ily variability of mood (IDVM) , wa s a sse sse d by calcu la ting th e sta ndard
d eviation ofVASM scores for each s ubject on each study day. Individu al ID VM sco res
were ave ra ge d and compared between groups. Each su bject wa s a lso eval uat ed usi ng
a 17 it em Hamilton Depression Scal e (Ham-D) ( 12) o n th e firs t mo rning of testing in
order to cha rac te r ize th e se ve r ity o f d epressiv e symptom s. No n-d ep r essed subjects
were excl ude d for Hamilton D epression Scal e scores > 10. Th e one-way analysis of
vari ance (ANOVA) , followed by the Duncan 's multipl e range co m parison test we re
us ed to a ssess differen ces betwe en groups for H am-D , a nd d ail y VASM a nd IDVM
scores.
Ultradian Cycles
Tim e se r ies pl ot s were d eveloped by plo ttin g t ime of d ay (X -ax is) versu s VASM
sco re (Y-axis ) for eac h su bject. Group tim e se r ies pl ot s were a lso d eveloped by
ave r a ging hourly mood scores for a ll su bjects in each g ro u p, for bo t h study d ays and
pl otting values in a similar m ann er . For exa m ple , th e 7:00 AM point on th e m aj or
d epression g ro u p plot r epres ents the ave ra ge of 22 assessme n ts: 12 subj ects on d ay I
a nd 10 su bjects on day 2. A co m pa ra ble m ethod o f ave raging hourly g ro u p m ood
sco res has be en used in th e study of circad ia n r hy t h ms of mood ( 13) . In this st udy,
cycle lengths (p eriods) were d et ermin ed by m ea suring th e peak -to-peak or t rou gh -t otrough di st ance a long th e X-axis, beginn in g wit h th e fir st com ple te cycle . Amplitud e
wa s d et e rmin ed by m easuring peak- to-trou gh a nd t ro ug h-to -peak di sta nces a long th e
Y-axis. D et ail ed d escriptions of this a nd o t he r m ethod s of ti m e se ries data a nalysis
have been review ed else whe re ( 14) . Sp ectral a na lysis ( 15) wa s used to co m pa re d a t a
d erived from visual in sp ecti on to a u to ma te d com p u te r-d e r ive d result s. A Hammi ng
window of 3 wa s us ed for lim it ed sm oot hing of tim e se ries plo ts. Th e M a nn-W h it ney
two sa m ple r ank su m test was used to co m pa r e the m ean ultra di an cycle period and
a m plit ud e of pati ent gro u ps to ward staff (16) .
RESULTS
Mood S cales
G ro u p H am-D score m eans a nd sta nda rd d eviation s are list ed in T abl e I .
ANOVA showe d sig nifica n t g roup differen ces in H amilt on D epression Scal e sco res
[F( 3,32) = 74.72 ; P < 0.001]. Duncan 's multiple com pa r iso n test s r eveal ed that
adj ust me n t di sord er and maj or d epression groups had m ark edl y g re at er H am-D
scores th an ward sta ff (p < 0.00 I) . A small ye t sig n ifica n tly g rea te r H am-D sco re wa s
12
JEFFER SON JOURNAL OF PSYC HI ATR Y
obs erved in th e non-depressed pati ent group versu s ward staff (p < 0.05) . Th ere was
no significant differen ce betwe en H am-D scores of adj us t me nt di so rd e r a nd m ajor
d epression groups. Approxim at ely 1100 VASM assess me n ts were ob ta in ed ove r 70
su bject-days . Group differen ces in dail y m ean VASM score wer e sim ilar to H am-D
findings (Table I) . C o m pa r ison of eac h group 's m ean intra-d ail y va riabilit y o f moo d
(IDVM), using Duncan 's multiple co m pa r iso n test , reveal ed th a t th e a dj us t m ent
dis ord er group had sig nifica n t ly g rea te r variation o f mood sco res th a n both con t ro l
g ro u ps on both d ays (p < 0.0 I) . IDVM in th e major d epressive e pisod e su bjects wa s
not significantly different from co n t ro ls.
V I/radian Cycles
Visual inspection of tim e se r ies plot s rev ealed clearly a p pa re n t ultradi a n cycles
in approximat ely 50 % of individual su bject days. Th es e plots, how ever, were not
cha ra cte r ist ic of all subj ects. Subjects with clear ult radian cycles on day I d id not
d emonstrat e the sam e cycl ica l patt ern on day 2. Th e mood plot s of a few su b jec ts
(approxim at ely 20 %) d emonstrat ed fr equ ent cha nges in a m plit ud e a nd pe riod , a nd
a p peare d to vary in a random m ann er. Averaging of hourl y g ro u p VASM scores,
however, yield ed tim e se r ies plots with less vari ation in cycle length a nd a m plit ud e
(Figure I). Sp ectral a nalysis o f g ro u p tim e se r ies d at a co nfir me d t hat th e dom in ant
cycle len gth was 3 hours for st aff a nd 4 hours for th e rem aining groups. Th e re wa s no
sig nifica n t differen ce between g ro u ps in ultradian cycle length . Amplitud e , however,
was significa n tly grea te r (based on M ann-Whitney two sa m ple r ank sum test ) in th e
m ajor d epressive e pisode g ro u p (u = 4.5,25 ; P = 0.026) a nd adj ustm ent di sord e r
(u = 8,34; p = 0.367 ) g ro u ps , wh en co m pare d to ward s taff. Th e mean amplitud e of
th e non-depress ed pati ent g ro u p wa s not significantly differe nt from wa rd staff.
C ONC LUSIO S
Th e result s of thi s inv estigation r eplicat e those of a n ex plora tory st ud y com parin g s ubjects with m ajor d epressiv e synd ro mes to non-d epressed su bjects (7). This
follow-up st udy utilized a different m edi cal ce n te r population a nd a different m e thod
of tim e series a na lysis. As in the ex plora tory st udy, we found th a t t he m ajor
d epression group d emonstrat ed uItradian cycles of g re a te r a m plit ude tha n co n t rols.
Again, we found no sig nifica n t difference between groups in uItradi an cycle pe riod.
Cycl e lengths were five hours for both groups in th e original investigation a nd th ree
to four hours for th e groups in this follow-up study. These cycle len gths a r e more
rapid than th e four to six hour cycle s o f mood r eport ed in normal m al e su bjects (6) .
In this study we found that th e three hour periodicit y wa s m ost evide n t between
2:00 p.m . and 8:00 p.m . As evide nce d by Figure I a nd T abl e 2, a concu r re n t six hou r
cycle harmonic may have a sig nifica n t e ffect o n m orning m ood. Th e three to four
hour mood cycle , however, appears to acco u n t for the g reat es t a mou n t of int ra-d a ily
m ood vari ability ba sed o n visu al in sp ecti on a nd s pe ct ra l a nalysis o f tim e seri es da ta .
The a bse nce of a morning mo od peak in th e d epressed grou p m ay be a cha ra cte r is tic
13
ULT RADIAN CYCLES OF MOOD
140
•
Ward staff (n= 12)
o
Non-deprelllled Patients (n= 8)
[J
Adjustment Disorder (n =8 )
li.
KaJor Deprellllion (n = 12)
120
100
VASM
SCORE
80
/
60
8:00 10:00 12:00 2:00
4:00
8:00
8:00
10 :00
TIME OF DAY
FIGURE 1. Time series data plots of visual analog scal e o f mood (VAS M) scores ve rsus
time of day for depres sed and non-depressed groups of s ubj ec ts .
JEFFERSO N JO URNAL OF PSYCHI AT RY
14
TABLE 2.
Time of Ultradian Cycle Mood Peaks for Depres sed and Non-d epr e s sed Groups
Group
Ward Staff
Non -depressed Pati ents
Adjustment Diso rd e r
Major Depression
\
Tim e of Mood Peaks
Barn
12pm
9am
9a m
2pm
3pm
2pm
2pm
5pm
5pm
5pm
5pm
8pm
Bprn
9pm
7pm
sp ecifi c to this st ud y populat ion or may r eflect a ph en om en ological di st inct ion of
major d epressive episode s. Th e presen ce of a 12:00 p.m. mood peak is s pecific to the
ward staff group, and may reflect th e difference in mid-day m eal expe rie nces for this
grou p.
In t his fo llow-up st udy, we have a lso explored th e sp ecificity of ultradi an m ood
cycle cha nges to d e p ression usin g smaller groups of non-depressed a nd adj us t me nt
di sord e r d epression pat ie n ts. The finding of great er ultradian cycle a m plit ude in th e
adjustment disorder group sugg est s th at great er cycle a m plit ude m ay be a characteristic of o t he r d epressive disorders. Th e absen ce of a m plitude cha nges in t he nond epres sed psych iatric patient group sug ges ts that this findin g is not ge ne ralized to a ll
psychi atric dis ord ers. Furth er study is needed to co nfir m that g rea te r ultradi an cycle
amplitude is sp ecific to d epressive synd ro me s, a nd not gen erali zed to o t he r psyc hiatric diagnosis (eg, anxiety disorders) or st re ssful cond it ions.
Many individua l tim es seri es plots showed a wide rang e o f cycle len gth and
amplitude. Subjects d emonst rat ing clear ultradian cycles on d ay I rarely s howe d the
sam e patt ern on d ay 2. We do not be lieve that this im pl ies th e a bse nce of an
endogenous r hythm, rath er, that e nviro n me n ta l vari abl es m ay distort th e ultrad ia n
cycle cha rac te r ist ics in a different mann er on su bse q ue n t d a ys. The high vari a bil ity of
ultradian cycle cha rac te r ist ics betwe en a nd within su bjects has be en not ed , re peated ly, in th e st udy of ultradi an cycles of EEG bands (17 ,18). St udi es of ult rad ia n
rh ythm s in m et abolic factors have a lso not ed po or sta bility o f ult rad ian cycles and
diffi cu lty in con t rolling e nviro n me n ta l co nd itio ns ( 19). Avera gin g of g rou p hourly
TABLE 3.
Means and Standard Deviations (in parenthesi s) for Ultr ad ia n Cycl e Chara cte r ist ic s
of Depressed and Non-depre s sed Groups
Group
Ward St aff
Non-depressed Pati ents
Adjustment Dis order
Maj or Depression
Cycle Len gth
(ho urs)
Cycle Ampl itu de
(VASl\! uni ts)"
3.3 (0.6)
3.7(2. 1)
3.7 ( 1.1)
4.0 (1.4)
4.0 (3.1)
5.2 (2.3)
8.7 (4 .6)b
H.4 (4 . 1)b
' VASM = Visual Ana log Sca le of Mood .
b p < 0.05, significant differen ce from ward sta ff usin g Mann-Whitney co m pa r ison.
U LTRA1)IAN CYC LES OF M O OD
15
m ea surem ent s m ay sig n ifica n t ly reduce th e int e rferen ce of t hes e variables, yie ldi ng a
co m mo n und erlyin g cycle.
Observation of peak a nd trou gh tim es in th e p resent st udy revea ls that pea ks
ge ne ra lly occur near m eal tim es. So me inves t iga to rs have not ed that ca rbohyd ra tes
incre ase in sulin levels , leadin g to cha nges in brain trypt opha n ratios and in cr ea sed
se ro to nin production , wh ich may eleva te m ood (20) . O t he rs have reported a bi-ph asic
patt ern of bet a- endorph in elevat ion afte r m eal s, wit h t he first pea k occurring 5
m in u tes afte r th e m eal a nd the second occu rring a pproximat ely 90 minut es lat er
(2 1). Bioch emical a na lyses a re ne ed ed to determ ine if level s of these fact ors co rrespo nd to ult ra di an cycles of m ood. Such a nalyses may lead to a g re a te r und erst and in g
of m ood co n trol.
Affectiv e di sorders m ay a lso involve a dys regu lation of u ltradian a nd circad ian
co rt iso l cycles. On e st udy ha s rep ort ed co rt iso l secre t ion cycle s of three a nd fo ur
hours for normal a nd d epressed g ro u ps , respecti vel y (22). Alt houg h leng t h of cycle
wa s not sign ifica n t ly different betwe en g ro u ps, th e m a gnit ud e of cort iso l secr etions
were significa n tly great e r in the d epressed grou p. O ne may sp ecu la t e that the three
to four hour cycles o f co rt isol secre t io ns , not ed in t ha t st ud y, may co r res po nd to t he
three to four hour cycle s of m ood no ted in t he p resen t st ud y, a nd th at th e g rea t e r
magn it ud e of co r t isol se cre to ry episode s may cor re s pond to g rea te r m ood cycle
amplit ud e . Th is hypo thesis is su pported by a study of diurnal cycles o f m ood wh ich
fo u nd th at cor t isol level s co r re la te d wit h m ood cha nge s in d ep r essed pati ents but was
unrel at ed to m ood curves of no n-d e p ress ed s ubj ects (23) .
In a se pa ra te a nalysis of this da ta , focus ing o n di agnostic iss ues of adjustm e nt
di sorder d epression s, we have r ep ort ed th at in tra-d a ily va r ia bili ty of mood (IDVM) is
signifi cantly great er in adjustm ent di sord er g rou p th an in th e o t her three groups
(ma nuscr ip t und er r eview ) . Based on thi s findin g , one may infe r t hat th e mo od
patt erns of pati ent s with adj us t me n t di sord er d epression do no t ad he re as well to t he
underlyin g ultradian cycle patt ern . This is co ns is t e nt wit h th e obse rva t ion th at
a dj ustm e n t di sord ers a re associa te d wit h st ro ng rea ctivity to e nviron me n tal s tim uli.
Althou gh the presen ce o f ult radi an cycles was clearly d e m onst ra ted in a ll
g ro u ps, so me limit ation s to this st udy a re not ed. Sex and age differ en ces were no t
well m at ch ed betwe en grou ps . The use of me dicat ions, nico t in e , ca ffe ine a nd o t he r
food it ems wa s not co n t ro lle d . D ue to the lon g hal f-lives of a n t id e pr essant m edi ca tions, however, we d o not beli eve th at th is factor sig n ifica n tly e ffec ted findin gs.
Furth er st ud ies may ben efit from g re a te r con t ro l of th ese a nd o t he r m ood-a lt e r ing
variables . The number o f patient s in the non-d epressed a nd adj us t men t d iso rd e r
pati ent g ro u ps wa s relatively small a nd findin gs from these gro ups should be
co ns ide re d prelimina ry. A non-param etric sta t is t ica l test was used to e ns u re a
co nse rva t ive probability sta te me n t. A la rge r numbe r of pat ien ts a nd further restriction of th e popul ation s st ud ies m ay pe r mit a clearer im pression of u nd erlying mood
cycles . Spe cificity of ultradian cycle cha ng es in d epressive di sord e rs m ay be fur t her
assessed by exam in ing non -ps ych iatric inpatient s a nd psych iat r ic ou tpat ient s, who
would not be su bject to th e sam e synch ro niz ing effec ts of zie tg ebe rs (environm en tal
cues) o n a n inpati ent psychi at ry ward .
16
J EFFERSO N JOURNAL OF PSYCHI ATRY
In co ncl usion, se ve ral obse rva tions are m ad e. First , mood va riation ove r th e
co u rse of th e day do es not a p pear to be chaot ic or co m pletely d ependent upon
e nviro n me n ta l cha nges. Rath er, we hav e found that underlyin g mood rhythms of
mood may form th e basis for th es e cha nge s. Th e und erlyin g ultradi an cycl e of mood
may be disrupt ed by e nviro n me n tal st im uli, resulting in m ood cycle arrhythmias .
However, the underlying cycle is reveal ed by ave raging hourly m ood sco res of groups .
At this point it is not possibl e to d et ermine if ultradian cycles of mood are regu lat ed
by e nd oge nous clock s, e nv iro n me n tal st im uli, or a n int eract ion of the two factors.
Second , we have found th at ultradi an cycle a m plit ude was grea ter in a maj or
d epressive e pisode a nd adj us t me n t di sord er with d epressed mood groups th an in a
ward staff g ro u p. It sho u ld al so be not ed that th e ultrad ia n cycles repres ent subtle
cha nges in ba selin e m ood st at e , a nd that th ese cycles m ay be mo re promin ent in
a fte rnoo n a nd eve ning hours. Repeat ed as sess me n t of m ood ove r t he course of th e
d ay in th e inpatient se tt ing and vari ed appointm ent tim es in th e outpatien t setting
m ay be useful to obtain an accurat e asses sm ent of mood .
REFERENC ES
I. Kle in R, Arm ita ge R: R hyth ms in h u ma n performance : one a nd on e-h alf hou r of
osci lla t ions in cog nit ive sty le. Sci ence 1979; 204:1326- 1327.
2. Love tt -D ovstj\V, Payn e W'D, Pod nie KS: An ultradian rhythm in rea ction tim e in m an .
Neu ro psychob iology 1978; 4:93-98.
3. Bossomj , Nate lson BH , Levin BE, Sto kes PE: U lt ra d ian r hythms in cognit ive fun ctions a nd
th ei r rel ati on ship to visce ra l pro cesses. Physiol a nd Beh av 1983; 3 1: 11 9-1 23.
4. H a yashi M, Hori T: Basic ultradi an co m po ne nts of dayt im e slee pin ess in hu ma n subj ects.
j ap an ese Psychological Research 1990; 32: 128-1 36.
5. La vie P, She rso n A: Ultras hort slee p-wa lking sched u le, I: Eviden ce of ultradi an rh yth micity in sleepability. Electroe nc ep h C lin Neuro physio l 1981; 52: 163-1 7+.
6. Tsuji Y, Fu ku du H , O kuno H , Kobayash i T : D iurn al rhythm of a lpha wave ac tivity a nd
mood . Electroencep h C lin Neurophysiol 1985; 52:543.
7. H all DP , Sing H C , Ro ma nosk i Aj: Id e ntifica t ion and cha rac te riza tion of g rea ter muod
var ian ce in de pression . Am j Psychiat ry 1991; 148:1341-1 345.
8. Bea rsm a DGM , Daa n S, Van den Hoofdakk e r RH : Dist ribut io n of RE ~ I lat en cies a nd
ot her sleep phe nome na in de pression as expla ine d by a single ultrad ian rh ythm di sturba nce . Slee p 1984; 7: 126-1 36.
9. Buysee OJ, J arrett DB , Mi cwald jM, Kup fer ~J : G ree nho us e JB: M in ute-by-m in u te
a na lysis of REM slee p timing in majo r depression. Biul Psychi a try 1990; 28:91 1- 925.
10. Folst ein M F, Luria R: Rel iab ilit y, va lidi ty and clinical applica tion of the visual anal og mood
sca le. Psychol M ed 1973; 3:4 79-486.
I I. Luria RE : The va lidity and reliability of th e visu a l ana lug muud sca le.j Psychi at r Res 1975;
12:51-57.
12. H a m ilt on M: A rating scale for depres sion .j Neurol Neurosurg Psychi atry 1960; 23:56-62.
13. Monk T H, Buysee Dj , Reynold s III CF, jarrett D B, Ku pfe r Dj: Rh ythmic a nd ho meosta tic
influ en ces on mood , activat ion and performance in yo ung a nd old men . j G e ro nt ol 1992;
47:22 1-227.
14. Sullberger A: Biologica l Rhyt h m Resea rch. New York: Elsevier , 1965.
U LT RADI AN CYC LES OF 1\1000
17
15. SPSS/ PC a nd Trends. C hicago, SPSS In c, 1990
16. H a yes WL : Sta t ist ics 3rd cd. Ne w Yo rk : H olt , Rin eh art an d W in st on, 1981.
17. Ort ega SM, Cabre ra MC: U lt ra dian rhyt hms in th e EE C an d tas k performan ce. C hronobio i 1990; 17:183-1 94.
18. O kawa M , Ma tou sek M , Pet er se n I: Spo nta neo us vigi lance fluctuations in th e dayti me.
Psych ophysiology 1984; 2 1:20 7- 2 11.
19. Lloyd D, Stupfe l M : The occurrence and func tio n of ult radi a n rhythms. Bio i Rev 1991;
66:275-299.
20. T oornviet AC , Pijl H , Meinde rs AE: Food and moo d : t he ce nt ra l connect ion. Neth J Med
1992; 4 1:45-47.
2 1. Matsumura M, Fukuda N, Sa ito S, Mori H : Effect of a test mea l, d uod e nal acidifi cation ,
a nd tetra gastrin o n th e plasm a co nce ntra t ion of bet a en dor phin -like immunorea ctivit y in
m an . Reg ul Pe pt 1982; 4 : 173-1 8 1.
22. H a lbr e ich U, Asni s C M , Shind ledecker R, Zum off B, Na t han S: Cortisol secr etion in
e ndoge no us dep ression . Ar ch G cn Psychi atry 1985; 42 :909-9 14.
23. von Zerssen D , Do err P, Emrich HM , Lund R, Pirk e KM: Diurnal varia t ion in mood and
cor tiso l rhythm in depression a nd norm al sta tes of mind . Eur Ar ch Psych iat ry Neuro l Sci
198 7; 237 :36-45.