Anda di halaman 1dari 6

DEPRESSION AND ANXIETY 26:273–278 (2009)

Research Article
Robert E. Strong, D.O., Barrie K. Marchant, M.S., Frederick W. Reimherr, M.D., Erika Williams, M.S.W.,
Poonam Soni, M.D., and Ruth Mestas, CCRC

Background: Bright visible-spectrum light therapy has proven effective in the

treatment of seasonal affective disorder (SAD) and recent basic research suggests
that blue wavelengths 470 nm account for that effectiveness. To more
stringently test the importance of these wavelengths, bright red-light was used
for the placebo (control) condition. Methods: Thirty subjects meeting DSM-IV
criteria for SAD were randomized to narrow-band light-emitting diode panels
emitting blue- or red-light in this 3-week, parallel, double-blind trial. Twenty-
five subjects participated in an open-label blue-light follow-up. Subjects were
divided in a blinded, post hoc manner into two groups: SAD only and those
experiencing depression with seasonal intensification. The outcome was assessed
using Hamilton Depression Rating Scale–17 item version (HAMD-17) and the
Structured Interview Guide for the Hamilton Depression Rating Scale—SAD
version. Responders were defined by Clinical Global Impression—Improvement
scale. Results: HAMD-17 scores improved more under the blue-light condition
(51%) than under the red-light condition (32%) (P 5.05). Further, in the blue
arm 60% of subjects responded compared with 13% in the red arm (P 5.01).
During the open-label phase, subjects from both double-blind arms improved
over baseline. SAD alone patients responded numerically better to treatment
than those experiencing depression with seasonal intensification during both
treatment periods. Conclusions: Narrow bandwidth blue-light therapy proved
superior to red-light therapy. Blue-light therapy produced results similar to both
previous 10,000 lux visible-spectrum light studies and many medication studies.
The use of bright red panels supported claims that wavelengths of 470 nm
account for the documented effectiveness of light therapy. Depression and
Anxiety 26:273–278, 2009. r 2008 Wiley-Liss, Inc.

Key words: seasonal affective disorder; wavelength; light therapy; placebo-

controlled; randomized

Mood Disorders Clinic, Department of Psychiatry, University

of Utah Health Sciences Center, Salt Lake City, Utah
INTRODUCTION Correspondence to: Robert E. Strong, D.O., Mood Disorders

Most studies estimate that 4–6% of the general Clinic, Department of Psychiatry, University of Utah Health
Sciences Center, Salt Lake City, Utah 84132.
population suffers from seasonal affective disorder
(SAD).[1] Conversely, Blazer et al.[2] found a 1% E-mail:
prevalence of major or minor depression with a Received for publication 31 July 2008; Revised 18 September
seasonal pattern. The use of alternative diagnostic 2008; Accepted 19 September 2008
procedures may explain these significantly different DOI 10.1002/da.20538
estimates. Alternatively, pure SAD may be rare[3,4] or Published online 18 November 2008 in Wiley InterScience (www.
the overlap between SAD and other depressive

r 2008 Wiley-Liss, Inc.

274 Strong et al.

disorders[5,6] may explain these differences. This MATERIALS AND METHODS

complication in diagnostic assessment may affect
The 3-week, parallel, double-blind phase compared blue-light
research into SAD and light therapy by not separating (active treatment) with red-light (placebo condition) using weekly
those patients experiencing incomplete summer remis- visits. Subjects were instructed to sit about 50 cm in front of the light
sion.[7] Finally, the DSM-IV Criterion for Seasonal panel and to glance at the light for a few seconds every minute for
Pattern Specifier does not preclude nonseasonal 45 min daily between 6:00 and 8:00 AM. Subjects were instructed not
depressive symptoms such as dysthymia. to reveal the color of their panel. These instructions were provided by
Light therapy has been relegated to the edge of the staff not associated with assessments. Subjects asking about specific
treatment paradigm[8] despite its documented efficacy. colors were told that the importance of color remained uncertain and
A meta-analysis on its efficacy in treating SAD[9] that this study was designed to compare different wavelengths.
produced an effect size of .84. This lack of interest Diagnosis of SAD and assessing of outcome measures was carried out
by clinicians (R.E.S., F.W.R., and P.S.), with extensive research
may be partly attributable to limited training and
education. Subjects who completed the double-blind phase could participate
Light therapy was intertwined with the original in a 4-week open-label trial using blue-light with biweekly visits.
description of SAD. Rosenthal et al.[10] reported on Instructions in this period were identical to those in the double-blind
patients with recurrent depression that developed in period, except that subjects could deduce that the red-light had been
fall or winter and remitted the following spring or the placebo.
summer. He included preliminary findings indicating The University of Utah Institutional Review Board reviewed and
that bright artificial light was effective in treating this approved the study. The study was conducted in accordance with the
disorder and a hypothesis about the depressive effects Declaration of Helsinki 1975 as revised in 2000.
of melatonin.
Subsequently, intrinsically photosensitive retinal gan- SUBJECT SELECTION CRITERIA
glion cells that project from the retina to the suprachias- Subjects with recurrent major depression according to the DSM-
matic nucleus, intergeniculate leaflet and olivary pretectal IV with a seasonal pattern and a Structured Interview Guide for the
nucleus,[11] areas associated with circadian entrainment, Hamilton Depression Rating Scale—SAD version (SIGH-SAD)
negative masking (suppressive effect of light on physio- score Z20 were recruited. Subjects were required to have routine
logical processes, which have a circadian rhythm), sleeping patterns suitable for the treatment procedures, which
regulation of sleep–wake states, the pupillary light reflex, remained stable for the duration of the study. Subjects were required
to be in general good health, without any photosensitizing conditions
and melatonin suppression were identified. It is likely
or the use of photosensitizing pharmacological agents. Subjects were
that these intrinsically photosensitive retinal ganglion excluded if they had
cells give rise to the therapeutic effects of light. Although
it is clear that melanopsin-expressing ganglion cells in the (a) recently used light therapy;
retina are intrinsically photosensitive, (especially by (b) failed previous treatment with light therapy;
wavelengths of 446–477 nm) Dacey et al.[12] found that (c) abnormal thyroid-stimulating hormone (TSH) values;
in rodents these cells can also be activated by rods and (d) a co-occurring psychiatric disorder or medical condition that
could affect their mental status;
cones. Thus, further evidence is needed to clarify
(e) ocular or dermatological health problems that might be affected
whether blue-wavelengths are necessary and/or sufficient by light therapy.
for the treatment effects of light therapy.
Glickman et al.[13] reported on a placebo-controlled
parallel trial of SAD comparing blue-light (468 nm at ASSESSMENT MATERIALS
398 lux) with dim red-light (654 nm at 23 lux). This Hamilton Depression Rating Scale—17 item version (HAMD-17):
3-week study found a significant (P 5.02) positive This depression scale was used at each visit.
treatment effect with a medium effect size (d 5 .48). Structured Interview Guide for the Hamilton Depression Rating Scale—
However, a categorical measure of remission was not SAD version (SIGH-SAD): This structured interview version of the
significant. Response rates in the blue-light condition HAMD-17 was developed for SAD and was used at each visit.
Clinical Global Impression—Improvement (CGI-I): This scale was
were typical for visible-spectrum light therapy and also
used at each visit following the screening interview.
similar to acute trials of antidepressants, except for a
quicker response time.
This protocol was designed to address three primary LIGHT TREATMENT DEVICES
questions. The light treatment units were designed and produced by the
sponsor of this study. The active unit was a 470 nm blue light-
* Was blue-light associated with a significant improve- emitting diode (LED) unit and the control unit was a 650 nm red
LED unit. Lux readings were blue panel 5 176 and red panel 5 201.
ment compared with red-light at visually similar
Photon density/cm2/s were blue panel 5 5.45 E14 and red pa-
nel 5 3.17 E14. The panels measured 4.500 by 300 . Although the
* Did nonseasonal depressive symptoms (SAD only photon densities of the two panels differed, their visual intensities
versus depression with seasonal intensification) were very similar.
impact the treatment effect? Safety of the blue LED unit used in a previous study was
* Was blue-light associated with significant adverse demonstrated by Glickman et al.[13], and an identical optical hazard
events (AEs)? analysis[14] was performed on the blue panels used in this study.

Depression and Anxiety

Research Article: Blue-Light Treatment of SAD 275

SEPARATION OF SAD DIAGNOSTIC TABLE 1. Baseline demographic measures and

CATEGORIES assessment of depressive symptoms
Although all subjects met criteria for the seasonal pattern, some Red-light Blue-light Totala
subjects had additional nonwinter symptoms. The diagnostic determi-
nation regarding SAD alone or a depressive illness with seasonal N 15 15 35
intensification of their depressive symptoms was made post hoc. A Ageb 39.579.9 51.1712.3 44.3712.60
clinician, who was blind to treatment, reviewed all patient charts. Items Male/femalec 2/13 5/10 9/26
suggesting that there were nonseasonal components to the SAD HAMD-17 19.775.1 20.373.7 20.074.4
diagnosis included: current, year-round use of antidepressant medica- SIGH-SAD total 34.175.9 34.175.6 34.175.6
tion; history of past antidepressant treatment in nonwinter seasons; or SAD alone 11 (73%) 8 (53%) 20 (61%)d
depressive symptoms during summer months. Depression with seasonal 4 (27%) 7 (47%) 13 (39%)
All randomized and nonrandomized subjects meeting admission
PROCEDURES criteria.
The primary objective of comparing the efficacy of blue- versus t 5 2.31, df 5 28, P 5.03.
c 2
red-light was accomplished by comparing improvement in HAMD- X 5 1.70, df 5 1, P 5.19.
17 scores using a last observation carried forward (LOCF) format and Two nonrandomized patients had insufficient data to be categorized
using a repeated measures ANOVA procedure with treatment (red accurately regarding the two SAD subgroups.
versus blue) as the primary predictor variable and controlling for
three other between-subjects variables: gender, age, and SAD
diagnostic category. Improvement across time was a within-subjects TABLE 2. Improvement in depressive symptoms during
factor. A similar model was used to assess improvement in the SIGH- the double-blind phase as a function of treatment
SAD scores. As gender and age proved to be nonsignificant, both
analyses were repeated and are reported without them. Categorical Red-light Blue-light P-valuea
improvement was analyzed using Fisher’s exact test.
Improvement in continuous measures of depression at the open- N 15 15
label termination visit was accomplished (LOCF) using a repeated Categorical improvement: N (%) improved
measures ANOVA procedure comparing the final open-label evalua- CGI-Ir2 2 (13%) 9 (60%) .01
tion with the baseline assessment including the two SAD diagnostic HAMD-17 (50% improvement) 2 (13%) 8 (15%) .025
categories as a between-subjects variable. Categorical improvement SIGH-SAD (50% improvement) 2 (13%) 6 (40%) .107
was analyzed using Fisher’s exact test. The two treatment groups Decline in HAM-D scores: mean7SD (percent improvement over
(previously blue- or red-light) were assessed separately for change baseline)
from double-blind endpoint to open-label endpoint using paired SIGH-SAD total 10.2710.7 15.779.6 .15b
t-tests for continuous variables and the Fisher exact test for (21%) (40%)
categorical variables. HAMD-17 6.073.9 10.175.3 .05c
Side effects were assessed for all subjects following randomization. (32%) (51%)
Both specific side effects and open-ended questioning were used. The SAD alone 5.974.2 11.675.5 .08
safety analysis set was reported based on the treatment received. (32%) (61%)
Analysis was completed using the SPSS version 14 statistical Seasonal intensification of 6.373.4 8.375.0 .70
package. depression (30%) (40%)
P-values were calculated (LOCF) using a repeated measures
RESULTS ANOVA procedure for continuous variables and the Fisher exact
test for categorical variables.
Thirty-five subjects met admission criteria for the b
Treatment had a nonsignificant effect on SIGH-SAD scores
study. Of these, 5 were not randomized and provided (F1,26 5 2.19, P 5.15).
no outcome data and 30 furnished outcome data. The Treatment had a significant main effect on HAMD-17 scores
patients in the blue-light group were significantly older (F1,26 5 4.52, P 5.043). Diagnosis (SAD alone versus seasonal
than the red-light group (P 5.03) (see Table 1). intensification of depression) interacted with treatment at a
nonsignificant level (P 5.321).
HAMD-17 and SIGH-SAD scores were not statisti-
cally different. Although the blue-light sample had
more males, this difference was not statistically
significant. A substantial minority of subjects (39%)
had seasonal intensification of major depressive dis- shown in Table 2 and Figure 1, on the HAMD-17,
order. More of these were randomized to the blue-light subjects improved an average of 6.073.9 (32%) points
(n 5 7) than the red-light (n 5 4) treatment arm. in the red-light condition compared with 10.175.3
(51%) in the blue-light condition (F1,26 5 4.52,
P 5.043). As shown in Table 2 and Figure 2, on the
DOUBLE-BLIND OUTCOME SIGH-SAD total, subjects improved an average of
Table 2 shows improvement for the 30 randomized 10.2710.7 (21%) points in the red-light condition
subjects. The treatment effect was statistically signifi- and 15.779.6 (40%) in the blue-light condition
cant for the HAMD-17, but not the SIGH-SAD. As (F1,26 5 2.19, P 5.15).
Depression and Anxiety
276 Strong et al.

TABLE 3. Open-label outcomes compared to baseline

and double-blind scores for 25 subjects entering the
open-label phase. Data are presented using an LOCF
format within each phase

Double- Open-
Baseline blind label P-valuea

Red-light group N 5 13
HAMD-17 19.475.3 13.275.7 6.577.9 .022
SIGH-SAD total 33.676.1 22.879.3 10.2710.5 .003
CGI-Ir2 na 2 (15%) 11 (85%) .001
Blue-light group N 5 12
Figure 1. Average HAMD-17 scores at each visit for both HAMD-17 20.674.0 9.876.7 8.077.1 ns
treatment groups. [Color figure can be viewed in the online SIGH-SAD total 34.276.1 17.5710.2 12.379.9 ns
issue, which is available at] CGI-Ir2 na 8 (67%) 10 (83%) ns
P-values compare open-label with double-blind scores (LOCF).
Continuous data were assessed using paired t-tests and categorical
data were assessed using the Fisher exact test.

17 scores (F1,21 5 120.6, Po.001) and 73% on SIGH-

SAD scores (F1,21 5 112.7, Po.001).
As shown in Table 3, subjects from the red-light
group showed statistically significant improvement
during the open-label period in all measures. In
contrast, subjects from the blue-light group showed
only nonsignificant additional gains during the open-
label period. Thus, the two groups had equivalent
levels of depression at the end of the open-label phase.
Figure 2. Average SIGH-SAD scores at each visit for both (The five subjects who did not enter the open-label
treatment groups. [Color figure can be viewed in the online period were excluded from this table.)
issue, which is available at]
Differences between SAD alone and depression with
seasonal intensification. Both subjects with SAD
alone and depression with seasonal intensification
Only two (13%) subjects were ‘‘much’’ or ‘‘very showed statistically significant improvement by the
much’’ improved on the CGI-I under the red-light end of the open-label phase. On the HAMD-17 the 16
condition versus nine (60%) subjects under the blue- subjects with SAD alone improved 71% over baseline
light condition (Fisher’s exact test 5 .01). (P 5.001), whereas the 9 subjects experiencing depres-
Differences between SAD alone and depression sion with seasonal intensification improved 56%
with seasonal intensification. There were 19 subjects (P 5.005) (LOCF). Similarly, 88% of subjects with
with pure SAD. The 8 subjects in the blue-light SAD alone met CGI-I criteria for improvement
arm improved 61% on the HAMD-17 compared compared with 67% of subjects experiencing depres-
with a 32% improvement for the 11 subjects sion with seasonal intensification for a nonstatistical
in the red-light arm (P 5.08). There were 11 difference between groups.
subjects experiencing depression with seasonal intensi-
fication. The seven subjects in the blue-light arm
improved 40% on the HAMD-17 compared with a
30% improvement for the four subjects in the red-light Blue-light therapy was essentially equivalent to red-
arm (P 5.70). light therapy in the incidence of AEs. As shown in
Table 4, AEs were limited in number and similar to
10,000 lux light studies (fluorescent lamps that appear
OPEN-LABEL OUTCOME white). All AEs were in the mild-to-moderate range
Of the 15 subjects randomized to red-light, 2 and only one subject left the study as a result.
dropped out during the double-blind period. Thus,
13 entered the open-label phase. Of the 15 subjects
randomized to blue-light, 1 dropped out during the TIME OF YEAR
double-blind phase and 2 more chose not to partici- The average date for the screening visit was January
pate. Thus, 12 entered the open-label phase. 25, 2006. The average date for ending the double-blind
Change from baseline to endpoint was significant for phase was February 22, 2008, and the average date for
these 25 subjects. Subjects improved 64% on HAMD- ending the open-label phase was March 22, 2008.
Depression and Anxiety
Research Article: Blue-Light Treatment of SAD 277

TABLE 4. The number of adverse events and side response rate for subjects enrolled during February.
effects under both light conditions This did not happen. Two of nine red-light subjects
who were enrolled in January met CGI-I criteria for
Red-light Blue-light
treatment response, whereas none of the six that were
Headache 1 5 enrolled in February responded.
Sleep disturbance 1 2 Almost 40% of the patients in this study showed a
Eye irritation 0 2 pattern of depression with seasonal intensification, a factor
Upper respiratory infection 4 1 deserving further exploration despite the limited numbers.
Lightheaded/dizzy 0 1 During the controlled phase, SAD alone subjects experi-
Sunburn 0 1 enced a treatment effect on the HAMD-17, which
Photosensitivity 0 1 approached significance, whereas those experiencing
Nausea 0 1
depression with seasonal intensification did not. Further,
Odd dreams 0 1
Memory problems 0 1
the SAD alone subjects had a numerically higher open-
Dysphoria 0 1 label treatment response rate than those experiencing
Nightmares 1 0 depression with seasonal intensification. One potential
Neck pain 1 0 explanation for these numerical differences is that those
experiencing depression with seasonal intensification are
suffering from more than one disease process (dual
DISCUSSION vulnerability hypothesis), thus requiring treatment with
antidepressant medication. This explanation would also
These data support the hypothesis that light therapy suggest that some patients diagnosed with a depressive
using narrow-band LED panels emitting wavelengths disorder, who only partly respond to treatment with
of 470 nm is effective in the treatment of SAD. This antidepressant medication, might suffer from a comorbid
trial attempted to address several concerns in previous form of SAD and require light therapy for full response.
trials. First, the dimness of past red-panels may have The number and severity of AEs and side effects was
made them recognizable as placebo. Second, therapeu- minimal and similar to earlier studies using white
tic benefits of red-light may have been limited as much fluorescent lamps.[15, 16] The severity of these symptoms
by the dimness of the light source as by the wavelength. was in the mild-to-moderate range. Most problems
Both concerns were addressed by the use of these remitted spontaneously and only one subject discontinued
brighter red-light panels. Our patients treated with treatment because of them (ocular photosensitivity).
bright-red panels had a response very similar to the
response rates reported in the past studies of dim-red-
light panels. CONCLUSION
As in previous trials, the 470 nm wavelength proved In conclusion, blue-light therapy using narrow-band
superior to the 650 nm wavelength in treating SAD. LED panels appeared to be an effective treatment for
This treatment response was observed in the double- SAD. It outperformed bright red-light therapy and
blind controlled phase of the trial and continued in the produced results similar to both 10,000 lux bright light
open-label phase. Although both groups improved studies and medication studies. The use of bright red
during the double-blind phase, the blue-light group narrow-band LED panels provided a more credible
improved significantly more than the red-light group. control condition than dim red panels. The results support
The blue-light group had a 60% response rate using biological studies implicating photosensitive retinal gang-
CGI-I criteria, whereas the placebo red-light group lion cells and melatonin suppression in the treatment of
had only a 13% response rate. seasonal affective disorder. Finally, our data suggest that:
Regardless of treatment status during the first phase, (1) nonseasonal depressive symptoms might be associated
subjects were much improved at the end of the open- with a lower response to light therapy and (2) seasonal
label period with average HAMD-17 scores of 6.577.9 factors might play a role in the incomplete recovery
for the sample previously on red-light therapy and frequently seen in patients with major depression.
8.077.1 for the sample previously on blue-light
therapy. The fact that the red-light group had Acknowledgments. The original clinical trial was
improved significantly to a point of equivalency with sponsored by Apollo Light Systems. This analysis was
the blue-light group gives some additional support for performed in cooperation with, but not funded by,
the efficacy of blue-light treatment. Apollo Light Systems. This article was presented in
Given that subjects were enrolled in January and part as a poster at the 2007 Annual Meeting of the
February improvement could be attributed to the Society of Biological Psychiatry, San Diego, CA.
change in seasons rather than treatment. However, no
evidence for this was observed. (Nineteen patients were
enrolled in January and eleven were enrolled in REFERENCES
February.) If there was improvement due to a change 1. Saeed SA, Bruce TJ. Seasonal affective disorders. Am Fam
in season, there should have been a higher red-light Physician 1998;57:1340–1346.

Depression and Anxiety

278 Strong et al.

2. Blazer DG, Kessler RC, Swartz MS. Epidemiology of recurrent 10. Rosenthal NE, Sack DA, Gillin JC, et al. Seasonal affective
major and minor depression with a seasonal pattern. The National disorder: a description of the syndrome and preliminary
Comorbidity Survey. Br J Psychiatry 1998;172:164–167. findings with light therapy. Arch Gen Psychiatry 1984;41:
3. Hardt J, Gerbershagen HU. No changes in mood with the 72–80.
seasons: observations in 3000 chronic pain patients. Acta 11. Gooley JJ, Lu J, Fisher D et al. A broad role for melanopsin in
Psychiatr Scand 1999;100:288–294. nonvisual photoreception. J Neurosci 2003;23:7093–7106.
4. Nayyar K, Cochrane R. Seasonal changes in affective state 12. Dacey DM, Laio H, Peterson BB et al. Melanopsin-expressing
measured prospectively and retrospectively. Br J Psychiatry ganglion and irradiance and project to the LGN. Nature
1996;168:627–632. 2005;433:749–754.
5. Wehr TA. Seasonal vulnerability to depression. Implications 13. Glickman G, Byrne B, Pineda C, et al. Light therapy for seasonal
for etiology and treatment. Encephale 1992;18:479–483. affective disorder with blue narrow-band light-emitting diodes
6. Kasper S, Kamo T. Seasonality in major depressed inpatients. (LEDs). Biol Psychiatry 2006;59:502–507.
J Affect Disord 1990;19:243–248. 14. Sliney DH. Blue light risk and light therapy. A presentation at the
7. Lam RW, Tam EM, Yatham LN et al. Seasonal depression: the 2nd CIE Expert Symposium Lighting and Health. Ottawa,
dual vulnerability hypothesis revisited. J Affect Disord 2001;63: Ontario, Canada, September 7–8, 2006.
123–132. 15. Labbate LA, Lafer B, Thibault A, Sachs GS. Side effects induced
8. Wirz-Justice A. Beginning to see the light. Arch Gen Psychiatry by bright light treatment for seasonal affective disorder. J Clin
1998;55:861–862. Psychiatry 1994;55:189–191.
9. Golden RN, Gaynes BN, Ekstrom RD, et al. The efficacy of light 16. Terman M, Terman JS. Bright light therapy: side effects and
therapy in the treatment of mood disorders: a review and meta- benefits across the symptom spectrum. J Clin Psychiatry 1999;60:
analysis of the evidence. Am J Psychiatry 2005;162:656–662. 799–808.

Depression and Anxiety