The Can-SAD Study: A Randomized Controlled Trial of the Effectiveness of Light Therapy and Fluoxetine in Patients With Winter Seasonal Affective Disorder

The researchers directly compared two common treatments for winter depression (Seasonal Affective Disorder, or SAD): bright light therapy and the antidepressant medication fluoxetine (Prozac). They wanted to see which worked better, how fast each worked, and whether one had fewer side effects.

The three groups were:

**Active light therapy + placebo pill:** Participants used a 10,000-lux fluorescent white-light box for 30 minutes every morning and took a placebo capsule daily.

**Placebo light + active fluoxetine:** Participants used a dim 100-lux light box (which looks identical but delivers much less light) for 30 minutes every morning and took 20 mg of fluoxetine daily.

**No double-placebo group:** There was no group that received both a placebo light and a placebo pill, which is a limitation the authors acknowledge.

The primary outcome was the change in depression severity measured by the 24-item Hamilton Depression Rating Scale (HAM-D) over 8 weeks. Secondary outcomes included response rates (defined as at least 50% reduction in HAM-D score), remission rates (defined as a final HAM-D score of 8 or less), and adverse events.

**Sample size:** 96 patients were randomly assigned to treatment (46 to light therapy + placebo, 50 to fluoxetine + placebo light). 78 patients completed the full 8-week study.

**Population:** Adults aged 18–65 who met DSM-IV criteria for major depressive disorder with a seasonal (winter) pattern. This means they had recurrent major depressive episodes that began in fall/winter and remitted in spring/summer for at least two consecutive years.

**Setting:** Four university-affiliated outpatient clinics in Canada (Vancouver, Calgary, Winnipeg, and Montreal). The study ran over three consecutive winter seasons (typically October through March).

**Severity:** All participants had a score of 23 or higher on the 24-item Hamilton Depression Rating Scale at baseline, indicating moderate-to-severe depression.

**Exclusion criteria:** People were excluded if they had bipolar disorder, psychosis, substance abuse, active suicidal ideation, or if they had already failed to respond to either light therapy or fluoxetine in the past. They also excluded people who had used antidepressants or light therapy within the previous 2 weeks (4 weeks for fluoxetine specifically).

The primary measurement tool was the **24-item Hamilton Depression Rating Scale (HAM-D)** . This is a clinician-administered interview that rates depression severity on a scale from 0 to 74 (higher scores = more severe depression). A score of 23 or higher was required for entry. The scale covers mood, guilt, suicidal thoughts, sleep, appetite, energy, and other symptoms.

Secondary measures included:

**Clinical Global Impression (CGI) scales:** A clinician-rated measure of overall improvement and severity.

**Structured Interview Guide for the Hamilton Depression Rating Scale – Seasonal Affective Disorder version (SIGH-SAD):** This is a modified version of the HAM-D that includes additional items specific to atypical depressive symptoms common in SAD, such as hypersomnia (sleeping too much), hyperphagia (eating too much), and carbohydrate craving.

**Adverse events checklist:** A systematic list of potential side effects, collected at each visit.

**Treatment-emergent adverse events:** Any new or worsening symptoms that appeared after starting treatment.

All assessments were done at baseline and then at weeks 1, 2, 4, 6, and 8 of treatment.

**Study design:** This was a double-blind, randomized, controlled trial (RCT). It was conducted at four Canadian centers over three winter seasons.

**Randomisation:** After a 1-week baseline observation period (to confirm they were still depressed and to wash out any previous treatments), eligible patients were randomly assigned to one of the two treatment groups. Randomisation was stratified by center to ensure each site had a balanced number of patients in each group.

**Blinding:** This was a double-blind study, meaning neither the patients nor the clinicians assessing them knew which treatment they were receiving. This is crucial because expectations can powerfully influence mood outcomes. The blinding was achieved through a "double-dummy" design:

Patients in the light therapy group received an active 10,000-lux light box and a placebo capsule that looked identical to fluoxetine.

Patients in the fluoxetine group received a 100-lux light box (which looks identical to the active box but delivers only 1% of the light intensity) and an active 20 mg fluoxetine capsule.

All light boxes were identical in appearance. The placebo light boxes used neutral density filters to reduce output to 100 lux without changing the color or appearance of the light.

**Duration:** The active treatment phase lasted 8 weeks. There was a 1-week baseline observation period before randomization. No follow-up after the 8 weeks was reported.

**Statistical approach:** The primary analysis was intent-to-treat (ITT), meaning all 96 randomized patients were included in the analysis regardless of whether they completed the study. Missing data were handled using the last observation carried forward (LOCF) method, which assumes that people who drop out would have stayed at their last measured level of depression. This is a conservative approach that can underestimate treatment effects if people drop out because they are feeling worse, or overestimate effects if people drop out because they are feeling better and no longer need treatment.

**What this design can and cannot prove:**

**Can prove:** Because this is a randomized, double-blind trial, it can establish that both treatments are effective and that they are roughly equivalent in efficacy over 8 weeks. The randomization ensures that the groups are comparable at baseline, and the blinding minimizes placebo effects and observer bias.

**Cannot prove:** This study cannot prove that either treatment is better than no treatment, because there was no double-placebo group (no light + no pill). Both groups received some form of active treatment (either real light or real medication). The 67% response rate in both groups could partly reflect the natural course of winter depression (which tends to improve as spring approaches) or a placebo response. The study also cannot tell us about long-term effects beyond 8 weeks, about whether combining both treatments is better than either alone, or about how these treatments compare in people with milder depression (since only moderate-to-severe cases were included).

**Major methodological weaknesses:**

No double-placebo control group, making it impossible to estimate the absolute effect size of either treatment versus natural history or placebo.

The LOCF method for handling dropouts can bias results.

The study was relatively small (96 patients), which limits its ability to detect small differences between treatments.

The 100-lux "placebo" light may not be truly inert — some studies suggest that even dim light can have a mild antidepressant effect, which would make the fluoxetine group look less effective than it really is.

**Primary outcome — change in HAM-D scores over 8 weeks:**

Both groups showed significant improvement over time (p < 0.001 for the time effect).

There was no significant difference between the light therapy group and the fluoxetine group at any time point after week 1 (group-by-time interaction: p = 0.52).

At week 1, the light therapy group had a significantly greater reduction in HAM-D scores than the fluoxetine group (mean difference of approximately 3–4 points, p < 0.05 in post hoc testing). This difference disappeared by week 2 and did not reappear.

**Response rates (≥50% reduction in HAM-D score):**

Light therapy: 67% (31 of 46 patients)

Fluoxetine: 67% (33 of 50 patients)

No significant difference (p = 0.96)

**Remission rates (final HAM-D score ≤ 8):**

Light therapy: 50% (23 of 46 patients)

Fluoxetine: 54% (27 of 50 patients)

No significant difference (p = 0.70)

**Adverse events:**

Overall number of adverse events was similar between groups (p = 0.57).

However, fluoxetine was associated with significantly more treatment-emergent adverse events, specifically:

- Agitation: 12% in fluoxetine group vs. 2% in light group (p < 0.05)

- Sleep disturbance: 14% in fluoxetine group vs. 4% in light group (p < 0.05)

- Palpitations: 8% in fluoxetine group vs. 0% in light group (p < 0.05)

No serious adverse events were reported in either group.

**Dropout rates:**

Light therapy: 17% (8 of 46 patients dropped out)

Fluoxetine: 20% (10 of 50 patients dropped out)

No significant difference (p = 0.73)

The results show that both treatments produced a large clinical improvement. The average HAM-D score at baseline was around 28–29 (severe depression). After 8 weeks, scores dropped to approximately 10–12 in both groups — a reduction of about 60–65%. This is a clinically meaningful change: patients went from being moderately-to-severely depressed to having only mild residual symptoms or being in full remission.

The early advantage of light therapy at week 1 was modest but noticeable. On the 24-item HAM-D (range 0–74), light therapy produced about a 3–4 point greater reduction than fluoxetine in the first week. To put this in perspective, a 3-point change on the HAM-D is roughly equivalent to the difference between "moderate" and "mild" ratings on 2–3 individual symptoms (e.g., going from "moderate difficulty falling asleep" to "no difficulty falling asleep"). This is a small-to-medium effect size (Cohen's d approximately 0.3–0.4).

The response and remission rates (67% and 50–54%, respectively) are comparable to what is seen in antidepressant trials for non-seasonal depression, suggesting that both treatments are genuinely effective for SAD.

**Acknowledged by the authors:**

Lack of a double-placebo condition (no active treatment group), making it impossible to rule out natural recovery or placebo effects.

The 100-lux "placebo" light may not be completely inert — some evidence suggests that even low-intensity light can have biological effects.

The study was not large enough to detect small differences between treatments (limited statistical power).

The 8-week duration may be too short to see the full effects of fluoxetine, which can take 4–6 weeks to reach peak efficacy.

**Additional limitations a critical reader would note:**

**Industry funding:** The study was supported by a grant from the Canadian Institutes of Health Research, but fluoxetine and placebo capsules were provided by Eli Lilly (the manufacturer of Prozac). This creates a potential conflict of interest, though the authors state that Eli Lilly had no role in study design, data analysis, or manuscript preparation.

**Self-report bias:** While the HAM-D is clinician-rated, it relies heavily on patient self-report of symptoms. Patients who believe they are receiving active treatment may report more improvement.

**Blinding integrity:** It is difficult to fully blind light therapy. Patients receiving the 10,000-lux light may have noticed it was brighter, and those receiving fluoxetine may have experienced side effects (like agitation or sleep disturbance) that revealed their treatment assignment. The authors did not formally test whether blinding was successful.

**Population limits:** The study only included people with moderate-to-severe winter depression. Results may not generalize to people with mild SAD, subsyndromal SAD (the "winter blues"), or non-seasonal depression.

**Geographic limits:** The study was conducted in Canada, where winter days are very short. Results may differ in locations with different latitudes and day lengths.

**No long-term follow-up:** We don't know whether the benefits persisted after the 8-week study ended, or whether people relapsed when they stopped treatment.

**Single dose of fluoxetine:** The study used a fixed dose of 20 mg/day. In clinical practice, doses are often adjusted up to 40–60 mg. A higher dose might have produced better results or more side effects.

For someone running their own n=1 experiment to test whether light therapy or an antidepressant works for their winter depression:

### What to test

**Light therapy:** Use a 10,000-lux fluorescent white-light box (not a "dawn simulator" or blue-light device). Sit 16–24 inches from the light box with eyes open (but not staring directly at the light) for 30 minutes each morning, ideally within 30 minutes of waking.

**Fluoxetine (if prescribed by a doctor):** 20 mg once daily, taken in the morning to minimize sleep disruption.

**Comparison:** You could test light therapy alone for 2–4 weeks, then add or switch to fluoxetine (under medical supervision) and compare your response.

### Minimum meaningful duration

**Light therapy:** You may see some improvement within 1 week (as the study found), but a minimum of 2–3 weeks is needed to assess the full effect. The study used 8 weeks.

**Fluoxetine:** Allow at least 4–6 weeks, as SSRIs typically take this long to reach full effect. The study used 8 weeks.

**For a self-experiment:** Run each condition for at least 4 weeks to allow for the slower onset of medication.

### What to measure (specific metrics)

**Primary metric:** Use a validated depression scale daily or weekly. The **Patient Health Questionnaire-9 (PHQ-9)** is a free, 9-item self-report scale (0–27, higher = worse). A 5-point or greater reduction is considered clinically meaningful.

**Secondary metrics:**

- **Sleep quality:** Track total sleep time, time to fall asleep, and number of awakenings using a sleep diary or wearable device.

- **Energy levels:** Rate your energy on a 1–10 scale each morning and evening.

- **Appetite and cravings:** Note any changes in carbohydrate craving or overeating.

- **Side effects:** Track agitation, sleep disturbance, palpitations, nausea, or any other unusual symptoms.

**Objective measure:** If possible, use a light meter to confirm your light box delivers 10,000 lux at the distance you're sitting.

### Key confounds to control for

**Time of year:** Winter depression naturally improves as days lengthen. Run your experiment during the same season each year, or compare the same weeks across different years.

**Sleep schedule:** Light therapy works partly by shifting your circadian rhythm. Go to bed and wake up at consistent times throughout the experiment.

**Other treatments:** Avoid starting other antidepressants, supplements (like St. John's Wort or melatonin), or psychotherapy during the experiment.

**Alcohol and caffeine:** Keep intake consistent, as both affect mood and sleep.

**Exercise and outdoor light exposure:** Track these, as they are powerful mood regulators. Try to keep them constant.

**Expectation effects:** If possible, have someone else prepare your light box or medication so you don't know which condition you're in (single-blind). At minimum, use a consistent routine and avoid reading about expected effects during the experiment.

### What a positive result would look like

**For light therapy:** A PHQ-9 score that drops by 5 or more points within 2–3 weeks, with improvement in morning energy and mood. You might notice feeling more awake and alert within 30–60 minutes of using the light.

**For fluoxetine:** A PHQ-9 score that drops by 5 or more points within 4–6 weeks, with improvement in mood, appetite, and sleep. Side effects like mild nausea or headache in the first week are common and usually resolve.

**A strong positive result