Content uploaded by Anna Wirz-Justice
Author content
All content in this area was uploaded by Anna Wirz-Justice on Nov 05, 2018
Content may be subject to copyright.
IOURNAL OF
ELSEVIER Journal of Affective Disorders 37 ( 19%) IO9- 120
AFFECTIVE
DISORDERS
Research report
‘Natural’ light treatment of seasonal affective disorder
Anna Wirz-Justice a9 * , Peter Graw a, Kurt Krkchi a, Asita Sarrafzadeh a,
Judie English b, Josephine Arendt b, Lothar Sand a
a Psychiatric University Clinic, 4025 Basel, Switzerland
b Department of Biological Sciences, University of Surrey. Guildford, UK
Received 22 March 1995; revised 26 September 1995; accepted 26 September 1995
Abstract
Patients with seasonal affective disorder (SAD) were treated for 1 week either with a daily l-h morning walk outdoors
(natural light) or low-dose artificial light (0.5 h @ 2800 11.1x). The latter treatment (given under double-blind conditions) can
be considered mainly placebo and did not improve any of the depression self-ratings, whereas natural light exposure
improved all self-ratings. According to the Hamilton depression score, 25% remitted after low-dose artificial light and 50%
after the walk. Sleep duration or timing were not crucial for the therapeutic response. The morning walk phase-advanced the
onset and/or offset of salivary melatonin secretion, but individual clinical improvement could not be correlated with specific
phase-shifts. Morning cortisol was decreased. Low-dose artificial light did not modify melatonin or cortisol patterns. This is
the first study to provide evidence for the use of outdoor light exposure as a potential alternative or adjuvant to conventional
artificial light therapy in SAD.
Keywords: Seasonal affective disorder; Placebo light therapy; Natural light exposure; Salivary melatonin; cortisol
1. Introduction
Seasonal affective disorder (SAD) is characterised
by recurrent major depressive episodes in autumn
and winter with spontaneous remission in spring and
summer (Rosenthal et al., 1984). Studies at different
latitudes and hemispheres over the past 10 years
have led to a consensus that bright artificial light is
the treatment of choice for this group of depressive
patients (Blehar and Lewy, 1990; Rosenthal and
Blehar, 1989; Society for Light Treatment and Bio-
logical Rhythms, 1990; Terman and Terman, 1991;
* Corresponding author.
Tam and Lam, 1995). Many clinicians and patients
ask why natural light cannot be used to treat symp-
toms. Since the intensity of illumination outdoors is
orders of magnitude greater than artificial light in-
doors (SOOO- 100000 lux compared with 50-500
lux; Rem6 et al., 1991), this is an obvious question.
Indeed, illuminance levels outdoors in Switzerland
(latitude 47”N) reach > 1000 lux in the early morn-
ing in winter, and remain above this intensity for
> 6 h even when overcast and rainy.
The difficulties in carrying out a blind, controlled
trial of artificial light therapy in SAD are legion, and
have led to a great deal of controversy with respect
to placebo effects (Eastman, 1990b; Eastman et al.,
1993). The difficulties of carrying out a blind, con-
0165-0327/%/$15.00 0 1996 Elsevier Science B.V. All rights reserved
SSDI 0165-0327(95)0008 1 -X
110 A. Wirz-Justice et al./Jourml ofAffective Disorders 37 (1996) 109-120
trolled trial of natural light are even greater, in
particular since being outdoors is combined with
different degrees of movement and temperature
change. ,Yet, the importance of this apparently sim-
ple treatment as a potential adjuvant or alternative to
artificial light therapy led us to carry out a controlled
trial of the clinical efficacy of a ‘morning walk’ in
winter and compare it with artificial light.
To relate clinical response to putative alterations
of circadian rhythm phase that have been postulated
to underlie the pathophysiology of SAD (Lewy et al.,
19871, salivary melatonin and cortisol were mea-
sured at dusk (onset) and dawn (offset).
2. Methods
2.1. Patient selection and study design
SAD patients were recruited by information in the
media or referred by their doctors. After screening to
establish symptom history (Seasonal Pattern Assess-
ment Questionnaire; Rosenthal et al., 1989), 95 per-
sons were interviewed by the psychiatrist (L.S.),
mostly in summer and early autumn. 74 fulfilled
DSM-III-R criteria for SAD (American Psychiatric
Association, 1987) and 34 entered the study. 5 pa-
tients on medication completed the trial but were not
included in the analysis; 1 patient did not remain
outdoors long or regularly enough (noncompliance
and nonresponse). Complete data were available for
20 subjects with natural light (19 women, 1 man,
aged 45.2 f 17.0 years) and 8 with artificial light (7
women, 1 man, aged 40.3 & 10.1 years); with the
exception of one set of self-ratings and saliva sam-
ples stolen from the car as the subject went for his
morning walk.
The SAD patients were informed that the study
aimed to test the antidepressant efficacy of natural
outdoor light, and that further treatment with artifi-
cial light would be freely available, should they so
desire. None of the SAD patients had had previous
experience with light therapy. Because of the diffi-
culty in creating a credible placebo treatment for
such a naturalistic study, a parallel group of SAD
patients was treated with low-dose artificial light.
The ‘natural light therapy’ 3-week protocol began
at the onset of depression in winter, when SAD
patients attained a Hamilton Depression Rating Score
(HAM-D; Hamilton, 1967) 2 12. When this depth
of depression had been maintained for a week (base-
line, BL), they were instructed to go outdoors regu-
larly, for 1 h, as early in the morning as possible,
and in the sun, if present (treatment, TR). Activity
was not proscribed, but nearly all walked most of the
time (none were joggers). Important was to go out-
doors every day of that week, irrespective of in-
clemencies in the weather (treatment week). The last
week was treatment withdrawal (WD; ‘return to your
previous habits’).
The ‘artificial light therapy’ protocol was given as
treatment to a parallel group of SAD patients who
did not wish to or could not participate in the walk
study (not randomly assigned). Their demographic
and clinical characteristics were similar to those of
the walk group. The 3-week protocol was identical
with that above, apart from the treatment week. The
patients used a light box for 0.5 h/day at a time of
their choice between 6-9 h. The light intensity aver-
aged 2800 lux at 80 cm distance, although both
patients and doctor were informed that it had been
designed to deliver 10000 lux. Since all participants
were unaware that the received light intensity was
low until the end of the study - and our previous
dose-response study had shown that 2500 lux for
0.5 h is insufficient for an adequate therapeutic effect
(Wirz-Justice et al., 1987) - the artificial light study
was thus carried out under double-blind conditions.
This then permitted an estimate of the response rate
to low-dose artificial light, which would include the
placebo effect as an important component.
The study was carried out over two winters be-
tween 1988- 1990; there was no difference between
the 2 years’ results in any variables, so data are
combined. The patients were interviewed again in
summer, together with a week of self-ratings.
2.2. Instrumentarium
SAD patients were interviewed weekly (HAM-D
together with Atypical Items; Rosenthal and Heffer-
nan, 1986), Clinical Global Impression (CGI; Na-
tional Institute of Mental Health 1976); self-ratings
were carried out weekly with a validated depression
scale (van Zerssen and Koeller, 1976). Sleep logs
and food and drink frequency questionnaires (FDFQ;
A. Wiry-Justice et al./Journal of Affective Disorders 37 (1996) 109-120 111
Kr&tchi et al., 1990) were completed daily. An
estimate of the timing and duration of outdoor light
exposure was obtained by daily ‘light logs’. These
required the patient to document each time he or she
went outdoors (accuracy of 15’ intervals). Seg-
mented visual analogue scales (Eastwood et al., 1984)
were used by the doctor and patients to rate expecta-
tions for the treatments before the trial, and the
patients retrospectively judged efficacy.
2.3. Melatonin and cortisol
Saliva was collected half-hourly from 20:00 h
until bedtime (until at least 22:30 h, and longer if
still awake) on the evening of the 5th day of each
week; and from awakening (beginning at least at
07:30 h, and earlier if awake) until 10:00 h on the
morning of the 6th day. These times were chosen to
cover both onset and offset of melatonin secretion in
the majority of subjects (though not the peak time at
night). The collection was made under naturalistic
conditions. 9 SAD patients collected samples inside
at home throughout the study. 10 SAD patients
collected samples during the walk outside as well as
at home). Saliva was centrifuged and frozen until
assay of melatonin and cortisol in Guildford. A
direct RIA for melatonin in saliva was developed
and validated with a GCMS melatonin assay (Dr.
A.J. Lewy; English et al., 1993). The RIA for corti-
sol (Read et al., 1977) has been validated with a
GCMS cortisol assay (Read, 1987). The sensitivity
of the assays was 5 pg/ml and 1 nmol/l for mela-
tonin and cortisol, respectively. The coefficient of
variation for melatonin was 15.5% (for low concen-
trations, 11 pg/ml), 8.9% (for medium concentra-
tions, 26 pg/ml) and 4.2% (for high concentrations,
57 pg/ml); similarly for cortisol 12.5% (1.6 nmol/l),
7.3% (4.1 nmol/l) and 5.6% (12.5 nmol/l). Onset
and offset of melatonin were defined as when the
mean of duplicate determinations was > or I 5
pg/ml, respectively, for two or more consecutive
time points, when data available.
Table 1
Effect of a I-h morning walk
Variables Baseline Walk Withdrawal ANOVA P
F-test(df 2.34)
Weekly ratings
HAM-D
Atypical items
CGI global score
CGI global improvement
von Zerssen
Sleep logs
Sleep onset
Wake-up time
Sleep midpoint
Time in bed (h)
Sleep duration (h)
Sleep latency (min)
No. awakenings
Time awake (min)
Sleep quality
(Lickert scale from + 2/ - 2)
- quietude
- depth
- recuperative
Sleep need (yes = I /no = 2)
Naps (min)
18.1 * 4.5
8.6 * 4.4
4.0 f 0.8
0.1 f 0.5
23.6 f 9.3
23: 11 f 74’
751 f 69
3:31 f 56’
8.7 f 1.5
7.9 f 1.6
26.2 f 15.6
1.5 f 0.9
43.8 f 35.8
0.3 f 0.5
0.2 f 0.7
0.1 kO.6
1.2f0.2
16.9 * 27.3
9.4 k 5.6
5.4k4.1
2.5 f 1.2
1.8 i I.1
17.1 * 11.0
22:58 + 60’
7:ll +44’
3:os * 43’
8.2 zt 1.1
7.7 f I.0
25.0 f 16.6
1.5 + 0.7
33.1 + 28.5
0.2 * 0.5
0.2 + 0.6
0.1 f 0.6
1.2 f 0.2
22.3 f 31.7
10.1 * 5.9 31.4
5.9 * 3.7 5.6
2.4 * 1.2 24.7
1.2 * 1.5 13.1
18.0 * 11.1 9.9 *
23:02 f 72’ 1.4
7:20 + 53’ 8.2
3:11 *54’ 7.1
8.3 & 1.1 3.3
7.8 f I.1 0.7
22.2 * 15.8 1.5
1.250.9 2.2
31.4+42.7 3.7
0.2 ?c 0.6
0.2 * 0.7
0.1 * 0.5
1.2 f 0.3
10.1 * 15.5
0.1
0.1 *
0.1 ’
0.7
1.9
< 0.01
<O.Ol
< 0.01
< 0.01
< 0.01
NS
< 0.01
< 0.01
= 0.05
NS
NS
NS
= 0.04
NS
NS
NS
NS
NS
Mean & SD; * df (2.32); NS, not significant.
112 A. WirzJustice et al./Journal of Affective Disorders 37 (1996) 109-120
OJ BL-7 BL WALK WD
Fig. 1. Time course of depression ratings (HAM-D score): at the beginning (BL-7) and end (BL) of the baseline week, after 1 week ‘natural
light’ treatment (WALK), and after 1 week withdrawal (WD). n = 20 SAD (n = 18 completed the withdrawal week). Statistics in Table 1.
Table 2
Effect of 0.5-h 2800 Lux artificial light
Variables Baseline Light Withdrawal ANOVA P
F test(df 2.12)
Weekly ratings
HAM-D
Atypical items
CGI global score
CGI global improvement
von Zerssen
Sleep logs
Sleep onset
Wake-up time
Sleep midpoint
Time in bed (h)
Sleep duration (h)
Sleep latency (mm)
No. awakenings
Time awake (min)
Sleep quality
(Lickert scale from + 2/ - 2)
- Quietude
- Depth
- Recuperative
Sleep need (yes = 1 /no = 2)
Naps (min)
20.1 f 5.2
7.6 f 3.3
4.5 f 1.2
-0.7 f 1.0
18.6 * 8.9
23:33 rt 62’
7:23 f 58’
3:28 f 54’
751 * 52’
7:18 * 45’
36.5 f 29.3
1.9f0.9
32.1 * 15.2
- 0.2 f 0.6
- 0.5 f 0.3
- 0.3 f 0.6
1.1 f 0.1
14.1 f 15.2
12.1 f 6.6
5.3 f 3.9
4.0 * 1.7
0.5 + 1.8
19.1 f 13.5
23:09 f 58’
6~34 f 45’
2152 f 44’
7125 rt 55’
7:00 f 65’
30.9 + 21.9
2.1 f 0.7
26.9 f 13.4
0.6 f 0.9
- 0.4 f 0.8
-0.4* 1.1
1.1 * 0.1
22.5 f 50.1
15.6 f 7.5
6.1 f 3.5
4.2 f 2.1
- 1.3 f 0.8
19.4 * 10.9
22159 f 47’
7~03 f 72’
3:Ol f 47’
8:03 f 79’
7:30 f 89’
26.8 f 19.1
1.8 f 1.1
32.8 + 22.0
0.0 + 1.0
- 0.4 f 0.9
- 0.4 f 0.9
1.2 f 0.2
6.0 f IO.9
9.0
0.7
0.3 *
3.2 ’
0.5
1.6 ’
3.2 *
5.4 *
0.8 *
0.6 *
1.7 l
0.3 *
0.3 *
0.6 *
0.2 * *
0.1 l *
0.8 *
0.6
< 0.01
NS
NS
NS
NS
NS
NS
= 0.03
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
Mean f SD; * df (2.10); * * df (2,8); NS, not significant.
A. WirzJustice et al./Journal of Affective Disorders 37 (1996) 109-120 113
3. Results
3. I. Clinical jindings
A l-h morning walk for 1 week significantly
reduced depressive symptomatology in all rating
scales (Table 1). There was no significant relapse
after withdrawal. Fig. 1 shows the time course of
clinical response to natural light in each individual
patient. Responders (R) were defined as those whose
HAM-D score declined by 2 50% (attained by 65%);
50% showed remission of symptoms (HAM-D score
< 8) (Terman et al., 1989). The effect size for this
treatment (d = 2.17) was considerable (values > d =
0.80 are considered large; Cohen, 1988, p. 40).
Clinical response (calculated in terms of scores BL-
TR/BL) was not correlated with the initial depth of
depression (BL) for any of the rating scales
(HRS/Atypical Items/van Zerssen, r =
0.1/0.2/0.07; NS).
50% of the 8 SAD patients treated with artificial
light responded and 25% showed remission of symp-
toms. The effect size was lower than that found after
natural light (d = 1.62). The significant improve-
ment found in the HAM-D ratings was not present in
Table 3
Light logs of time spent outdoors *
Time of day Walk n Light therapy n
Before noon ( 8 8- 12 h)
Winter BL 20.7 f 11.3 19 8.8 f 11.6 8
Winter TR 61.2 f 23.4 19 1.7 f 3.6 8
Winter WD 16.4 f 13.7 18 1.8 f 2.4 7
Summer 42.6 f 45.7 20 31.5 + 28.7 4
Afternoon(Z12-18h)
Winter BL 51.1 f 25.9 19 34.7 f 25.8 8
Winter TR 33.0 f 23.6 19 27.0 * 21.3 8
Winter WD 33.4 f 26.8 18 30.3 * 37.0 7
Summer 133.9 l 66.1 20 153.3 f 61.9 4
’ In minutes, average of 7 days/subject; mean + SD of n subjects.
ANOVA of the time course (BL, TR, WD, Summer):
‘Walk’ group
(1) Before noon, n = 18: F(3,5 I) = 26.9, P = 0.0001; all pairwise comparisons were significant, except BL-WD.
(2) After noon, n = 18: Ff3.51) = 30.8, P = O.COOl: all pairwise comparisons with summer were significant.
Light-therapy group
(3) Before noon, n = 3: F(3,6) = 4.0, P = 0.07; all paitwise comparisons with summer were significant.
(4) After noon n = 3: F(3,6) = 5.4, I’ = 0.04; all pairwise comparisons with summer were significant.
the Atypical Items score, CGI, or von Zerssen self-
ratings (Table 2).
Both patients and doctor had high expectations for
‘natural light’ (segmented VAS, where 9 = extreme
positive effect, 5 = no effect, 1 = extreme negative
effect: 7.3 k 0.8 (SD) and 6.6 k 0.5, respectively).
However, there was no correlation between these
expectations and clinical response (for the rating
scales HRS/Atypical Items/van Zerssen: patient,
r = 0.32/0.32/0.28, NS; doctor, r =
0.22/0.26/0.43, NS). The patients retrospectively
judged the morning walk to be as efficacious as
initial expectations (7.3 &- 1.8); this correlated signif-
icantly with improvement according to the Atypical
Items score (r = 0.47, P = 0.04) and von Zerssen
self-rating (r = 0.51, P = 0.021, though not with the
HAM-D score ( r = 0.33, NS). The global improve-
ment score of the CGI (Table 1) was taken as a
measure of the doctor’s retrospective judgement; this
correlated significantly with improvement according
to the HAM-D (r = 0.53, P = 0.02) and Atypical
Items score (r = 0.47, P = 0.04) though not with the
von Zerssen self-rating (r = 0.42, NS). Patients ret-
rospectively judged artificial light to have had no
effect (5.7 + 2.3), even though their expectations had
been identical with those for the morning walk (7.1
114 A. Wirz-Justice et al./Journul ofAffective Disorders 37 (1996) 109-120
f 0.9). Additionally, although the doctor’s expecta-
tions for artificial light were similar to those for the
morning walk (6.4 f 0.51, his retrospective judge-
ment (global improvement score of the CGI) showed
that artificial light was significantly less effective
than the morning walk (F(1,25) = 9.9, P < 0.01).
A
I L 0 BASELINE I
0 WALK
w 0
$4. @ WITHDRAWAL
F
8 2. f$!&@@:@
ooq. - - - - -. - -. - -. z
6 7 8 9 10 12 15 18 20 21 22 23 24
B
0 1 .
6 7 8 9 10 12 15 18 20 21 22 23
C
1” ‘.
T - BASELINE
s WALK
.: “““” WITHDRAWAL
: . .
‘..,; .,
p
d
2 15 -
0 1 ‘, . . .,
6 9 12 15 18 21
TIME OF DAY (h)
A. WirzJustice et ul./Joumal of Affective Disorders 37 (1996) 109-120 115
3.2. Sleep and food logs
The morning walk advanced the mean wake-up
time by 40’, advanced sleep midpoint by 26, and the
patients spent 30’ less time in bed, reporting 10’ less
intermittent time awake during the night (Table 1).
The only significant effect of ‘low-dose’ artificial
light in the morning was a 36’ advance of sleep
midpoint (Table 2).
The morning walk, but not artificial light, selec-
tively suppressed sweet carbohydrate intake in the
second half of the day (data not shown; analysis
using previously validated FDFQ methods; Krauchi
et al., 1990).
3.3. Light logs
During the winter months when this investigation
was carried out (14 October-15 March) dawn oc-
curred between 06:23 and 07:45 h, dusk between
16:ll and 1805 h. The time estimated spent out-
doors (as logged in 15’ intervals) from 8-12 h and
from 12- 18 h was averaged over each week for each
individual and then averaged over each group (Table
31.
In winter, a daily average of 72’ spent outdoors
during baseline increased to 94’/day during the walk
week, and sharply declined in the withdrawal week
to 50’/day. This provides evidence for compliance
with respect to time spent outdoors as well as for
lack of continued outdoor exposure during with-
drawal. In contrast, the SAD patients receiving artifi-
cial light treatment spent on the average 44’/day
outdoors during baseline and this decreased during
the treatment week to 29’, and remained low during
withdrawal (32’). Both treatment groups spent a
great deal more time outdoors in summer (177’ and
185’/day, respectively). There was no correlation
between time spent outdoors, or early or late timing
of the morning walk, and clinical response, either for
responders or nonresponders (data not shown).
Finally, the time each patient spent outdoors dur-
ing a given week was added up for each 15’ interval
between 6-21 h (0’ = was never outdoors at that
particular time of day, 105’ = was always outdoors at
that time of day). The group mean reveals how long,
on average, SAD patients spent outdoors at each
time of day. These data are presented with the
hormonal patterns (Figs. 2 and 3). For the walk
study, identical low exposure occurred during the
morning of the baseline and withdrawal week. After-
noon light exposure was the same in all 3 weeks.
The only marked increase was from - 8-10 h dur-
ing the morning walk. The artificial light treatment
group spent less time outdoors during the baseline
week than the walk group (Table 3, Fig. 3): the
pattern was characterised by practically no outdoor
light exposure before noon, but a similar time spent
outdoors in the afternoon as the walk group.
3.4. Cortisol and melatonin
Evening onset of cortisol secretion occurs during
sleep. The timing of our samples thus precluded
Fig. 2. Temporal relationships between salivary cortisol (A), salivary melatonin (B) and outdoor light exposure (0. Documentation in
winter during a baseline week, during the ‘morning walk’ week and during the withdrawal week. Hormones were assayed on the evening of
day 5 and the morning of day 6 of the weekly protocol. All saliva collections were complete from 7:30- IO:00 h and from 20:0&22:30 h;
because of individual variability in bedtimes, the inconsistent number of early morning ( < 7 h) and late night samples (> 23 h) were not
included in the ANOVA, but visually inspected and used for defmition of onset and offset in the x2 analysis. (A) Decline in morning
cortisol after a morning walk (n = 17; ANOVA: F(l.16) = 5.0, P = 0.04). Comparison of the withdrawal week data with their paired
baseline values revealed that cortiso1 had returned to pretreatment values (n = 15; ANOVA: F(l,l4) = 0.1, P = 0.74, NS). The evening
cortisol values were all low and did not change across study weeks. (B) Morning melatonin offset before and after a morning walk (n = 18;
ANOVA: F(l.17) = 5.9, P = 0.03; interaction F(5,85) = 2.8, P = 0.02). Comparison of the withdrawal week data with paired baseline
values revealed that melatonin remained low (n = 16; ANOVA: F(l.15) = 4.4, P = 0.05; interaction F(5.75) = 4.2, P = 0.002). Evening
melatonin onset was similar in all groups (ANOVA: F(2.30) = 0.7, P = 0.50, NS). (C) Patterns of the average duration of time spent
outdoors before and after a morning walk. Bach 15’ interval between 6:00-2l:oO h logged as spent outdoors was summed up for each
patient over each week and then averaged over the entire group. This yielded the curves for the duration of time spent outdoors during
daytime for each week.
116 A. Wirz-Justice et nl./Jourml ofAffective Disorders 37 (1996) 109-120
measurement of this phase-marker. Morning cortisol fied after low-dose artificial light treatment (Fig.
in the entire group was slightly, but significantly 3A).
lower after the morning walk, and returned to base- Evening melatonin onset in the group as a whole
line values in the withdrawal week (all statistics in
legends to figures) (Fig. 2A). Cortisol was not modi- was not significantly modified by the morning walk.
However, the morning offset occurred earlier, and
A
” 10 -
k 0 BASELINE
0 LIGHT
8 WITHDRAWAL
4
6 7 8 9 10 12 15 18 20 21 22 23 24
B
0 6 7 8 9 10 12 15 18 20 21 22 23
C
45 -
It - BASELINE
LIGHT
z “““” WITHDRAWAL
p-
p
9 12 15 18 21 24
TIME OF DAY (h)
A. WirzJusrice ef a/./Journal of Aflectiue Disorders 37 (1996) l@-120 117
remained earlier in the withdrawal week (Fig. 2B).
Most melatonin samples were collected under natu-
ralistic conditions at home. Thus, direct effects of
bright light (‘masking’) were absent. In the evening,
the sampling time began after dusk and in addition,
inspection of individual and average (Fig. 2C) light
logs showed that they were never exposed to outdoor
light after 17:00 h. Thus, melatonin onset was mea-
sured under relatively low light conditions with min-
imal masking. Similarly, morning melatonin offset
was measured under relatively low indoor light con-
ditions, except for 10 subjects who collected samples
during the morning walk (9 collected samples at
home before they went outdoors). Examination of
individual light logs and individual melatonin pro-
files showed that all except 2 in each group had low
morning melatonin already by the fiist two samples,
independent of whether they collected samples in-
doors or outdoors. The long-lasting effect of earlier
morning melatonin offset in the withdrawal week is
unlikely to be a result of masking, since patients
received practically no outdoor light exposure before
09:OO h (Fig. 2C).
Further analysis of individual melatonin patterns
used all available data (i.e. samples earlier than
07:30 h and later than 22:30 h). 2 subjects of 19 had
no detectable melutonin onset during baseline, and,
therefore, 17 were compared before and after the
morning walk. 12 SAD patients showed a phase-ad-
vance, 2 no change, and 3 a phase-delay of onset
timing (x2 = 8.59, P < 0.05, df 2). This was re-
versed during withdrawal: 12 SAD patients phase-
delayed, 2 showed no change, and 2 phase-advanced
(x2 = 10.14, P < 0.01, df 2). In 1 patient, no data
were available for withdrawal.
4 subjects had no detectable melatonin secretion
in the morning, resulting in n = 15 for analysis of
melatonin offset times. 12 SAD patients showed a
phase-advance after the walk, 1 no change, and 2 a
phase-delay of offset timing (x2 = 12.15, P < 0.01,
df 2). Again, after withdrawal, the majority (9) tended
to phase-delay, with 5 showing a phase-advance
( x2 = 4.15, NS, df 2). In 1 patient, no data were
available for withdrawal. This analysis indicates that
a significant phase-advance of melatonin onset
and/or offset was found in more than 3/4 of the
SAD patients after a morning walk. On withdrawal,
the majority delayed onset and/or offset times.
In SAD patients given artificial light therapy, the
group mean melatonin onset and offset were not
significantly modified (Fig. 3B). In comparison with
the ‘walk’ group, this smaller group of SAD patients
went, by chance, later to bed (Tables 1 and 21, and
also showed a later melatonin onset (22:30 h) during
baseline. It may be relevant that this group also spent
less time outdoors during the morning of the baseline
week (thus lowering the effective Zeitgeber strength)
(Fig. 30. However, it is important to note that there
was no difference in depression ratings between the
two groups at baseline (Tables 1 and 2), providing
further evidence for inhomogeneity with respect to
circadian phase-position within the SAD diagnosis
(Wirz-Justice et al., 1993).
4. Discussion
Our hypothesis was that 1 h of natural light
exposure in the morning in winter would be as
effective an antidepressant in SAD patients as our
previous trials of 1 h of 2500 lux intensity artificial
light. This proved to be the case: the response rate of
65% was in the same range as artificial light in the
morning (67%, n = 18) (Wirz-Justice et al., 1993). It
Fig. 3. Temporal relationships between salivary cortisol (A), salivary melatonin (B) and outdoor light exposure (C). Documentation in
winter during a baseline week, during the ‘low-dose artificial light therapy’ week and during the withdrawal week. Details as in legend to
Fig. 2. (A) Decline in morning cortisol after morning artificial light treatment (n = 6) and withdrawal (n = 5). There were no significant
differences between the groups (ANOVA: F(2,8) = 0.7, P = 0.53, NS; interaction F(22.88) = 1.1, P = 0.39, NS). ‘lie evening cortisol
values were all low and did not change across study weeks. (B) Morning melatonin offset before. and after morning artificial light treatment
attd withdrawal. There were no significant differences between the groups (ANOVA: F(2.10) = 0.42, P = 0.67, NS; interaction
F(22,llO) = 0.6, P = 0.56, NS). Evening melatonin onset was also similar in all groups (ANOVA: F(2,12) = 2.5, P = 0.13, NS; interaction
F(10,60) = 0.64, P = 0.77, NS). Melatonm onset in the baseline week occurred later than in the ‘walk’ group of SAD patients (Fig. 2B).
(C) Patterns of the duration of time spent outdoors during daytime for each week (see legend to Fig. 2C and text for details).
118 A. Wire-Justice et al./Journal ofAffective Disorders 37 (19%) 109-120
is also similar to the 66% overall response rate in a
pooled analysis involving 332 SAD patients treated
with 2 h of 2500 lux morning light (Terman et al.,
1989). Comparison with historical controls has of
course limitations; however, it is noteworthy that 7
of our SAD patients returned the same winter to be
treated with conventional light boxes (2500 lux for 1
h for 1 week) and 5 were responders (71%).
Improvement was registered in all observer and
self-rating scales after this short treatment, albeit not
to such a great extent as seen during spontaneous
remission in summer. Although 1 week of light
treatment is generally used for research purposes,
and is clearly efficacious (see Terman et al., 1989), 1
week is certainly not optimal for reducing all depres-
sive symptoms. All our previous l-week trials of
artificial light find this ‘residual’ depressivity com-
pared with self-ratings in summer (Wirz-Justice et
al., 1989). Recent studies document that response
rates increase with longer duration (2-4 weeks) of
light treatment (Bauer et al., 1994; Eastman et al.,
1993).
With the widespread publicity now attendant on
light therapy, it is practically impossible to create a
convincing placebo or maintain blindness of both
rater and patient. The double-blind study with 0.5 h
2800 lux light was, therefore, important in that it
provided a controlled situation to estimate the extent
of placebo response in our SAD sample as well as
rater bias (Eastman, 1990b; Eastman et al., 1993).
No significant improvement was found in any rating
scales other than the HAM-D, where 25% fulfilled
the dual criteria for remission (Terman et al., 1989).
This replicates previous findings of 25% remission
with 0.5 h 3000 lux (Terman et al., 1990) and 31%
remission with 0.5 h 2500 lux light (Wirz-Justice et
al., 1987) but this time with the additional important
information resulting from being carried out under
double-blind conditions. Thus, our SAD patients do
not appear to be high placebo responders (see also
discussion in Wirz-Justice et al., 1993).
Also, our SAD patients, although subjectively
classifying themselves as hypersomnic, showed nor-
mal timing and duration of sleep during their winter
depression. Thus, sleep disturbances per se are un-
likely to be etiopathogenic for SAD. A recent, com-
prehensive sleep EEG study documents this lack of
sleep disturbances in our SAD population, and the
minor effects of clinically efficacious midday light
treatment on the EEG (Brunner et al., 1996). Clinical
improvement after the morning walk might be con-
sidered a consequence of partial sleep deprivation in
the second half of the night (although the 40’ earlier
wakeup time is clearly insufficient for such an inter-
vention to be therapeutic). Analysis of the daily
individual sleep logs did not reveal any marked sleep
loss in responders that could have been responsible
for the antidepressant effect. In fact, the day-to-day
consistency of wake-up time during the walk was
quite high in all subjects (mean deviation from base-
line 20 f 6’).
Evidence for a neurobiological effect of the mom-
ing walk came from the selective suppression of
sweet carbohydrate intake in the second half of the
day. This is the third independent replication of this
specific effect of light therapy linked with mood
improvement (Krauchi and Wit-z-Justice, 1992;
K&rchi et al., 1990).
The abnormal circadian phase hypothesis of SAD
was addressed by analysis of melatonin rhythms.
Evening onset and morning offset of melatonin se-
cretion appear to be separately regulated by two
oscillatory mechanisms (IllnerovB et al., 1989), also
in humans (Wehr et al., 1993). The morning walk
provided sufficiently high light intensity early enough
to phase-advance morning melatonin offset and/or
evening melatonin onset in the majority of subjects.
The independent and not always parallel shifts in the
onset and offset of melatonin secretion emphasise
the importance of measuring both markers before
making an assumption as to changed phase-position.
No correlation could be established between an indi-
vidual’s decrease in depression ratings and a phase-
advance in meIatonin onset and/or offset. Nocturnal
melatonin duration, as far as could be estimated from
these phase-markers, was not changed after the
morning walk.
The nadir of the cortisol rhythm and the timing of
secretion onset, as validated markers of circadian
phase (Linkowski et al., 1993), were not measurable
during the evening collection times. Peak cortisol,
and the morning decline appear to be mainly influ-
enced by environmental factors (Linkowski et al.,
1993). Thus, the decline in morning cortisol found
after the walk may have been a consequence of
behavioural aspects of the protocol rather than a
A. Wirz-Justice et al./Journai ofAffective Disorders 37 (1996) 109-120 119
modification of circadian phase. There was no corre-
lation between an individual’s decrease in depression
ratings and an earlier decline in morning cortisol.
The daily light logs provide detailed documenta-
tion of how little outdoor light SAD patients receive
in winter. During their depressive phase, the group
as a whole were on the average outdoors for 63’/day
(n = 27). Their light log data were remarkably simi-
lar to a questionnaire estimate we had obtained in
another group of SAD patients (64’/day, n = 38;
unpubl. data). Our light logs also provide very analo-
gous results to those found by Eastman in the USA:
12 SAD patients averaged 73’/day in winter (East-
man, 199Oa). However, we found no correlation
between the amount of time spent outdoors per se
and the individual SAD patient’s clinical response to
a morning walk. Using a light monitor, another
group has found a negative correlation between
severity of depression and amount of light exposure,
and a trend between greater severity of depression
and later timing of morning light exposure (Oren et
al., 1994). In our study, neither correlation was
significant. There does not appear to be a simple
relationship between the amount of bright natural
light received by an individual and concomitant clin-
ical state. However, a regular morning walk was
clearly of clinical use.
Recent studies have documented that exercise is a
potential Zeitgeber in humans as well as animals
(Van Reeth et al., 1994). The intensity and duration
of exercise required to induce a phase-shift is rather
high, and only phase-delays have been documented
so far. Although it is clear that a morning walk
involves factors other than just light exposure (in-
creased motor behaviour as well as decreased tem-
perature), our subjects certainly did not exert them-
selves physically in their promenade in the sense of
high exercise.
In conclusion, we have studied an apparently
simple question often asked of light therapy re-
searchers, why not treat winter depression with natu-
ral light? The results can be set next to artificial light
protocols of similar duration. Since going outdoors is
not a usual winter behaviour, SAD patients need to
be informed that this is a practicable, simple, and
cheap therapeutic alternative (which can be com-
bined with artificial light, as required). At temperate
latitudes, outdoor light in winter is sufficiently in-
tense independent of weather conditions. The study
also provided further evidence that sleep duration or
phase are not crucial for the symptoms of SAD
(Wire-Justice and Anderson, 1990). Efforts to corre-
late clinical improvement with specific phase-shifts
in melatonin were unsuccessful.
Acknowledgements
This study was made possible by support from the
Swiss National Science Foundation (Grant 3.900-
0.88). We thank Professors H.-U. Fisch (Psychiatric
University Outpatient Clinic Beme) and C. Budde-
berg (Psychiatric University Outpatient Clinic Zlirich)
for generous provision of interview facilities.
References
American Psychiatric Association ( 1987) Diagnostic and Statistic
Manual of Mental Disorders (DSM-III-R). 3rd Rev. Ed. Wash-
ington, DC.
Bauer, M.S., Kurtz, J.W., Rubin, L.B. and Marcus, J.G. (1994)
Mood and behavioral effects of four-week light treatment in
winter depressives and controls. J. Psychiatr. Res. 28, 135- 145.
Blehar, MC. and Lewy, A.J. (1990) Seasonal mood disorders:
consensus and controversy. Psychopharmacol. Bull. 26, 465-
530.
Brunner. D.P., Kriuchi, K., Dijk, D.-J., Leonhardt. G., Haug,
H.-J. and Wirz-Justice, A. (1996) The sleep EEG in seasonal
affective disorder and in control women: effects of midday
light treatment and sleep deprivation. Biol. Psychiatry, in
press.
Cohen, J. (1988) Statistical Power Analysis for the Behavioral
Sciences. Erlbaum, Hillsdale, NJ.
Eastman, C.I. (199Oa) Natural summer and winter sunlight expo-
sure patterns in seasonal affective disorder. Physiol. Behav.
48. 61 I-616.
Eastman, C.I. (199Ob) What the placebo literature can tell us
about light therapy for SAD. Psychopharmacol. Bull. 26.
495-504.
Eastman, CL, Young, M.A. and Fogg, L.F. (1993) A comparison
of two different placebo-controlled SAD light treatment stud-
ies. In: L. Wetterberg (Ed.), Light and Biological Rhythms in
Man. Pergamon Press, Oxford, UK. pp. 371-383.
Eastwood, M.R., Whitton, J.L. and Kramer, P.M. (1984) A brief
instrument for longitudinal monitoring of mood states. Psychi-
atry Res. 11, 119-125.
English, J., Middleton, B.A., Arendt, J. and Wire-Justice, A.
( 1993) Rapid direct measurement of melatonin in saliva using
an icdinated tracer and solid phase second antibody. Ann.
Clin. Biochem. 30, 415-416.
120 A. Wire-Justice et al./Jour~l ofAffective Disorders 37 (1996) 109-120
Hamilton, M. (1%7) Development of a rating scale for primary
depressive illness. Br. J. Sot. Clin. Psychol. 6, 278-296.
Illnerova, H., Vanecek, J. and Hoffmann, K. (1989) Different
mechanisms of phase delays and phase advances of the circa-
dian rhythm in rat pineal N-acetyltransferase activity. J. Biol.
Rhythms 4, 187-200.
Krliuchi, K. and Wirz-Justice, A. (1992) Seasonal patterns of
nutrient intake in relation to mood. In: G.H. Anderson and
S.H. Kennedy @ds.), The Biology of Feast and Famine:
Relevance to Eating Disorders. Academic Press, Orlando, FL.
pp. 157-182.
Kriiuchi, K., Wirz-Justice, A. and Graw, P. (1990) The relation-
ship of affective state to dietary preference: winter depression
and light therapy as a model. J. Affect. Disord. 20, 43-53.
Lewy, A.J., Sack, R.L., Singer, C.M. and White, D.M. (1987) The
phase shift hypothesis for bright light’s therapeutic mechanism
of action: theoretical considerations and experimental evi-
dence. Psychopharmacol. Bull. 23, 349-353.
Linkowski, P., Van Gnderbergen, A., Kerkhofs, M., Bosson, D.,
Mendlewicz, J. and Van Cauter, E. (1993) A twin study of the
24-h cortisol profile: evidence for genetic control of the
human circadian clock. Am. J. Physiol. 264, E l73- 18 1.
National Institute of Mental Health (1976) Clinical Global Impres-
sions. 028 CGI. In: W. Guy (Ed.), ECDEU Assessment for
Psychopharmacology. Rockville, MD. pp. 217-222.
Oren, D.A., Maul, D.E., Schwartz, P.S., Brown, C., Yamada, E.M.
and Rosenthal, N.E. (1994) Exposure to ambient light in
patients with winter seasonal affective disorder. Am. J. Psychi-
atry 151.591-593.
Read, G.F. (1987) RIA of salivary cortisol. Hormone Res. 25,
18-22.
Read, G.F., Riad-Fahmy, D. and Walker, R.F. (1977) Determina-
tion of cords01 in human plasma by radioimmunoassay: use of
the ‘2sI-labe11ed radioligand. Ann. Clin. Biochem. 14, 343-
349.
Rem& C.E., Win-Justice, A. and Terman, M. (1991) The visual
input stage of the mammalian circadian pacemaking system: I.
Is there a clock in the mammalian eye? J. Biol. Rhythms 6,
5-29.
Rosenthal, N.E. and Blehar, MC. (1989) Seasonal Affective
Disorder and Phototherapy. Guilford, New York, NY.
Rosenthal, N.E. and Heffeman, M.M. .(I9861 Bulimia, carbohy-
drate craving and depression: a central connection? In: R.J.
Wurtman and J.J. Wurtman (Eds.), Nutrition and the Brain.
Raven, New York, NY. pp. 139-166.
Rosenthal, N.E., Sack, D.A., Gillin, J.C., Lewy, A.J., Goodwin,
F.K., Davenport, Y., Mueller, P.S., Newsome, D.A. and Wehr,
T.A. (1984) Seasonal affective disorder: a description of the
syndrome and preliminary fmdings with light therapy. Arch.
Gen. Psychiatry 41, 72-80.
Rosenthal, N.E., Kasper. S., Schulz, P.M. and Wehr, T.A. (1989)
New concepts and developments in seasonal affective disor-
der. In: C. Thompson and T. Silverstone @is.), Seasonal
Affective Disorder. CNS Clinical Neuroscience, London, UK.
pp. 97- 132.
Society for Light Treatment and Biological Rhythms (1990) Con-
sensus statement on the efficacy of light treatment for SAD.
LTBR 3.5-9.
Tam, E.M. and Lam, R.W. (1995) Treatment of seasonal affective
disorder. LTBR 7, 32-38.
Terman, M. and Terman, J.S. (1991) Seasonal affective disorder
and light therapy: report to the Depression Guidelines Panel,
P.H.S. Agency for Health Cam Policy and Research. New
York State Psychiatric Institute, New York.
Terman, M., Terman, J.S., Quitkin, F.M., McGrath, P.J., Stewart,
J.W. and Rafferty, B. (1989) Light therapy for seasonal affec-
tive disorder. A review of efficacy. Neuropsychopharmacol-
ogy 2, l-22.
Terman, J.S., Terman, M., Schlager, D.. Rafferty, B., Rosofsky,
M., Link, M.J., Gallin, P.F. and Quitkin, F.M. (1990) Efficacy
of brief intense light exposure for treatment of winter depres-
sion. Psychophatmacol. Bull. 26, 3-l 1.
Van Reeth. 0.. Sturis, J., Byme, M.M., Blackman, J.D..
L’Hermite-BalMaux, M., Leproult, R., Oliner, G., Refetoff,
S., Turek, F.W. and Van Cauter, E. (1994) Nocturnal exercise
phase delays circadian rhythms of melatonin and thyrotropin
secretion in normal men. Am. J. Physiol. 266, E964-E974.
von Zerssen, D. and Koeller, D.M. (1976) Paranoid-
Depressivit&skala. Die Klinischen Selbstbeurteilungsskalen
KSb-S. Beltz Test Gesellschaft, Weinheim, Germany.
Wehr, T.A., Moul, D.E., Barbato. G., Giesen, H.A., Seidel, J.A.,
Barker, C. and Bender, C. (1993) Conservation of photope-
riod-responsive mechanisms in human beings. Am. J. Physiol.
265, R846-R857.
Wit-z-Justice, A. and Anderson, J.L. (1990) Morning light expo
sum for the treatment of winter depression: the one true light
therapy? Psychophannacol. Bull. 26, 5 1 I-520.
Wirz-Justice, A., Schmid, A.C., Graw, P., Krtiuchi, K., Kielholz,
P., Pijldinger, W., Fisch, H.-U. and Buddeberg, C. (1987)
Dose relationships of morning bright white light in seasonal
affective disorder (SAD). Experientia 43, 574-576.
Wirz-Justice. A., Graw. P., Krlluchi, K. and Piildinger, W. (1989)
Phototherapy in Switzerland: ‘Mehr Licht’. In: C. Thompson
and T. Silverstone (Eds.). Seasonal Affective Disorder. CNS
Clinical Neuroscience, London, UK. pp. l69- 185.
Wirz-Justice, A., Graw, P., Kduchi, K., Gisin, B., Jochum, A.,
Arendt, J., Fisch, H.-U.. Buddeberg, C. and Paldinger, W.
(1993) Light therapy in seasonal affective disorder is indepen-
dent of time of day or circadian phase. Arch. Gen. Psychiatry
50, 929-937.
