Sei sulla pagina 1di 3


Greater Improvement in Summer Than With Light Treatment in Winter in Patients With Seasonal Affective Disorder

Teodor T. Postolache, M.D., Todd A. Hardin, M.S., Frances S. Myers, R.N., M.S.N., Erick H. Turner, M.D., Ludy Y. Yi, Ronald L. Barnett, Ph.D., Jeffery R. Matthews, M.D., and Norman E. Rosenthal, M.D.

Objective: The authors sought to compare the degree of mood improvement after light treatment with mood improvement in the subsequent summer in patients with seasonal af- fective disorder. Method: By using the Seasonal Affective Disorder Version of the Hamilton Depression Rating Scale, the authors rated 15 patients with seasonal affective disorder on three occasions: during winter when the patients were depressed, during winter following 2 weeks of light therapy, and during the following summer. They compared the three condi- tions by using Friedman’s analysis of variance and the Wilcoxon signed ranks test. Re- sults: The patients’ scores on the depression scale were significantly higher after 2 weeks of light therapy in winter than during the following summer. Conclusions: Light treatment for 2 weeks in winter is only partially effective when compared to summer. Further studies will be necessary to assess if summer’s light or other factors are the main contributors to this difference. (Am J Psychiatry 1998; 155:1614–1616)

Light therapy is an effective antidepressant for the winter depressive symptoms in patients with seasonal affective disorder (1–3). Although clinical experience suggests that mood improvement following light treat- ment is smaller than the spontaneous improvement during summer, this difference has never been formally documented.


We recruited patients with seasonal affective disorder by news- paper and radio advertising in the metropolitan Washington, D.C., area. Patients met the criteria of Rosenthal et al. for seasonal affec- tive disorder (1) and DSM-IV criteria for a past major depressive ep- isode. The criteria of Rosenthal et al. require at least one past major depressive episode and at least 2 consecutive years in which depres- sion developed in fall or winter and remitted in spring or summer. To start light treatment, a subject had to have a typical depressive score

Received Nov. 3, 1997; revisions received April 16 and May 20,

1998; accepted July 9, 1998. From the Clinical Psychobiology Branch, NIMH. Address reprint requests to Dr. Postolache, Section


Biological Rhythms, NIMH, Bldg. 10, Rm. 3S231, Bethesda,


20892; (e-mail).

Supported by the NIMH intramural research program. The authors thank Thomas Wehr, M.D., for reviewing an early draft of the article; Holly Lowe, M.S.W., and Kathleen Dietrich, R.N., for screening and rating patients; and Karen Pettigrew, Ph.D., for statistical consultation.


of at least 14 or a minimum typical score of 12 with a total depres- sive score of at least 20 on the Hamilton Depression Rating Scale, Seasonal Affective Disorder Version (4). We excluded patients if they had comorbid axis I psychiatric conditions or medical problems or were taking psychotropic medication. Twenty-two patients were eli- gible, of whom 15 were enrolled in this study (two patients required antidepressants when they did not respond to light, one underwent surgery, and four were unavailable for follow-up). Patients signed in- formed consent forms. Their mean age was 44.2 years (SD=10.6). Eleven (73.3%) were women and four (26.7%) were men. Nine pa- tients (60%) had unipolar depression, and six (40%) had bipolar disorder. We rated mood by administering the Hamilton Depression Rating Scale, Seasonal Affective Disorder Version (4), once during winter when the patients were depressed, once during winter after 2 weeks of light therapy, and once during the following summer. Response to light was defined as at least a 50% decrease in baseline score and a total posttreatment score of 7 or lower on the Seasonal Affective Disorder Version of the Hamilton depression scale (5). Eight of 15 patients responded to 2 weeks of light treatment, and 13 of 15 pa- tients responded in summer (difference not significant by McNemar test for significance of changes, with Yates’s correction for continu- ity: 2 =2.29, df=1, 0.10<p<0.20). Light treatment consisted of a standard regimen of 10,000-lux cool-white fluorescent light therapy for 45 minutes twice daily. After 2 weeks of the standard regimen, the treatment was individually cus- tomized and gradually tapered in spring. We initially analyzed the patients’ scores on the depression scale by using the Friedman’s two-way analysis of variance with condi- tions (depressed, light-treated, and summer) as factors. We then compared typical, atypical, and total depression scores across light- treated and summer conditions by using the Wilcoxon signed ranks

Am J Psychiatry 155:11, November 1998

test. The criterion alpha (two-tailed) was set at 0.015, with a Bonfer- roni correction for multiple comparisons.


The Friedman analysis for total scores on the depres- sion scale showed a significant effect for condition (Friedman test statistic=26.3, df=2, p<0.001). Simi- larly, we found a main effect for condition when we applied the analysis to typical (Friedman test statistic= 24.4, df=2, p<0.001) and to atypical (Friedman test statistic=26.23, df=2, p<0.001) scores. Further Wilcoxon signed ranks test comparisons of light-treated and summer conditions showed that de- pressive scores were significantly lower in the summer than after light therapy (figure 1). The median total de- pression score (30 in depressed state) fell from 7 in the light-treated condition to 3 in the summer (T=5, N=14, p<0.01) The median typical depression score (16 in de- pressed state) fell from 5 in light-treated condition to 2 in the summer (T=16, N=15, p<0.01). The median atypical depression score (10 in depressed state) de- creased from 4 in light-treated condition to 1 in the summer (T=4, N=14, p<0.01).


To our knowledge, this is the first report suggesting that mood improvement with conventional light treat- ment in depressed patients with seasonal affective dis- order is not as complete as the spontaneous remission occurring in summertime. In order to verify the sea- sonal pattern of seasonal affective disorder in partici- pants in a light-treatment study, Magnusson and Krist- bjarnarson (6) rated nine patients with seasonal affective disorder in the summer following light treat- ment but found no statistically significant difference in the depressive scores between postlight and summer conditions, possibly because of insufficient statistical power. Our finding cannot be explained in terms of the de- velopment of summer hypomania in bipolar patients. First, hypomanic symptoms would tend to increase rather than decrease the typical depressive symptoms. Second, we examined post hoc, separately in unipolar and bipolar patients, the changes in mood between the light-treated and summer conditions; the results in the two subgroups were very similar. The small sizes of these subgroups do not, however, allow for a proper statistical analysis. Such an analysis might be accom- plished by future studies on larger groups. We recognize that there is an ordering effect that may have confounded our result, given that all subjects received light therapy before the arrival of summer. It is, therefore, possible that our finding is due to an overall trend toward improvement with time. It is also possible that a longer trial of light therapy may have yielded effects comparable to those of summer as it has

Am J Psychiatry 155:11, November 1998


FIGURE 1. Clinical Ratings on the Seasonal Affective Disorder Version of the Hamilton Depression Rating Scale (Scores for Typical, Atypical, and Total Depression) in Patients With Sea- sonal Affective Disorder During Winter Depression, After 2 Weeks of Light Treatment, and in the Following Summer

2 Weeks of Light Treatment, and in the Following Summer a =Median depression score. Significant condition

a =Median depression score. Significant condition effect (Fried- man test statistic=24.4, df=2, p<0.001). Significant difference be- tween light-treated and summer (Wilcoxon T=16, N=15, p<0.01). b Significant condition effect (Friedman test statistic=26.23, df=2, p<0.001). Significant difference between light-treated and sum- mer (Wilcoxon T=4, N=14, p<0.01). c Significant condition effect (Friedman test statistic=26.3, df=2, p<0.001). Significant difference between light-treated and sum- mer (Wilcoxon T=5, N=14, p<0.01).



been suggested that further gains are seen beyond 2 weeks’ treatment (7). Nevertheless, the amount of light treatment given in this study is equivalent to, or greater than, that given in most clinical trials (5, 8). Moreover, in our study, the rate of response in summer according to the Terman et al. criteria (5) was 86%, which is su- perior to the 67% response rate after 4 weeks of light therapy reported by Bauer et al. (7). A number of factors—for example, the summer-win- ter differences in photoperiod, light intensity, and amount of ultraviolet radiation reaching the eye and skin—might have contributed to the superior anti- depressant effects of summer as compared with light therapy. Consistent with this last suggestion is a recent report on the possible influence of light exposure to skin on circadian rhythms (9). Besides light, other physical factors, such as temperature, as well as chem- ical, biological, and socioeconomic variables, differ be- tween winter and summer and may have an effect on mood and behavior. The influence of these factors on patients with seasonal affective disorder may well be worth studying in their own right.


1. Rosenthal NE, Sack DA, Gillin CJ, Lewy AJ, Goodwin FK, Davenport Y, Mueller PS, Newsome, DA, Wehr TA: Seasonal


affective disorder: a description of the syndrome and prelimi- nary findings with light therapy. Arch Gen Psychiatry 1984; 41:


2. Terman M, Terman JS: A multi-year controlled trial of bright light and negative ions, in Abstracts of the 8th Annual Meeting of the Society for Light Treatment and Biological Rhythms. Wheat Ridge, Colo, Society for Light Treatment and Biological Rhythms, 1996, p 1

3. Eastman CI, Young MA, Fogg LF, Lui L: Light therapy for win- ter depression is more than a placebo. Ibid, p 5

4. Williams JBW, Link MJ, Rosenthal NE, Terman M: Structured Interview Guide for the Hamilton Depression Rating Scale, Seasonal Affective Disorder Version (SIGH-SAD). New York, New York State Psychiatric Institute, 1988

5. Terman M, Terman JS, Quitkin F, McGrath P, Stewart J, Raf- ferty B: Light treatment for seasonal affective disorder: a re- view of efficacy. Neuropsychopharmacology 1989; 2:1–22

6. Magnusson A, Kristbjarnarson H: Treatment of seasonal af- fective disorder with high-intensity light: a phototherapy study with an Icelandic group of patients. J Affect Disord 1991; 21:


7. Bauer MS, Kurtz JW, Rubin LB, Marcus JG: Mood and behav- ioral effects of four-week light treatment in winter depressives and controls. J Psychiatr Res 1994; 28:135–145

8. Terman M, Amira L, Terman JS, Ross DC: Predictors of re- sponse and nonresponse to light treatment for winter depres- sion. Am J Psychiatry 1996; 153:1423–1429

9. Campbell SS, Murphy PJ: Extraocular circadian phototrans- duction in humans. Science 1998; 279:396–399

Am J Psychiatry 155:11, November 1998