Name: Light Therapy: Strong for Seasonal Depression, Promising Beyond It
Creator: Realised
Published: 2026-06-01

Summary

Bright-light therapy is a genuinely drug-comparable, underused treatment for seasonal affective disorder (Strong Evidence), but the same well-earned signal is being stretched into a "light box for everything" market where the non-seasonal, bipolar, and perinatal data are only Moderate, the dose details (morning, ~10,000 lux measured at your actual eye-to-lamp distance, ~30 min) are load-bearing and routinely mis-sold, and bipolar disorder is a hard do-not-self-start contraindication. The 10,000-lux lamp is a SUBSTITUTE for real morning sunlight, reached for when genuine outdoor light is not av

Why Strong

Strong Evidence because the entry's load-bearing claim is the seasonal one, and that claim is anchored by an APA-commissioned meta-analysis (effect size 0.84), an active-comparator RCT showing non-inferiority to fluoxetine, and a guideline body naming it first-line for seasonal depression. That is meta-analytic, replicated, drug-comparable evidence from sources without a device conflict.

NOT Foundational because this is a clinical intervention with a two-sided commercial controversy, an unsolved blinding problem, and a genuine safety contraindication, not a single undisputed axiom.

NOT Moderate overall because the seasonal case clears the bar for Strong on its own. The entry carries a split internally: the non-seasonal, bipolar, and perinatal extensions sit at Moderate (small, largely unblindable trials, low-certainty pooling), and prevention sits at Emerging (one Cochrane-eligible trial). The headline reflects the strongest, load-bearing indication while the body marks each extension down rather than letting the SAD halo inflate them.

Practical takeaway

The framing to hold: strong and underused for seasonal depression; promising but unproven-strong for everything else; and dose-and-timing dependent in a way that cheap marketing ignores.

The hierarchy to hold first: get outdoor morning light before reaching for a lamp. Real sunlight is full-spectrum and far more intense, and it is the first-line move (defer to morning_sunlight_exposure). Use the 10,000-lux lamp only when you genuinely cannot get outside (high latitude, deep winter, a schedule that keeps you indoors at dawn). Treat it as a substitute for the sun, not an equal: artificial light boxes are spectrally truncated (a narrow, blue-weighted slice, missing the UV and near-infrared the sun carries) and many units flicker. For the artificial-versus-natural deep dive, route to artificial_light_vs_sunlight.

For seasonal affective disorder (Strong Evidence):
• 10,000 lux, ~30 minutes, within the first hour or two of waking. Morning is the studied window and the one the circadian logic supports.
• Measure at your actual eye-to-lamp distance. The rated lux holds only at the manufacturer's stated close distance. Position the lamp at that distance, ~30 degrees below the line of gaze (you do not stare into it). A lamp used at arm's length is under-dosing.
• Expect onset within roughly 1–2 weeks. SAD response is often quicker than an antidepressant (CANMAT 2016 notes usually within 1 to 3 weeks), which is part of what makes it attractive, but it is not the 3–7 days some marketing implies.
• A genuinely effective, cheap, non-pharmacological, self-administered option that is underused. Say so plainly.

For non-seasonal depression, perinatal, or general low mood (Moderate Evidence):
• Reasonable as an adjunct, especially alongside an SSRI where the strongest single trial showed the combination beating either alone, but present it as promising, not as a proven equal to the SAD case.
• Same dose logic (morning, ~10,000 lux at correct distance, ~30 min). The details that make it work in SAD are the details to carry over.

For bipolar depression (Moderate Evidence, gated):
• Do not self-start morning bright light. The only studied safe path is midday timing under clinical monitoring, ideally with mood-stabiliser cover. This is a clinician conversation, not a lamp purchase.

Device discipline:
• Treat "10,000 lux" as conditional on distance and duration. Favour units with published specs, and be sceptical of "happy lamps" sold on a generic mood/sleep/focus/energy claim, which is the marketing tell.
• Watch for flicker and check that the rated lux is stated at a usable distance.

The honest caveat to share: because bright light cannot be fully blinded, some of the benefit may be expectancy. The SAD effect survives sham and drug comparison, so the core verdict holds, but a user deserves that asterisk rather than a clean "it definitely works because light."

This entry is the clinical-device intervention. For the free circadian anchor of getting outdoor daylight on waking, route to morning_sunlight_exposure; for the UV-and-vitamin-D question, route to light_exposure_vitamin_d.

Evidence detail

Why This Entry Exists

Light therapy is the rare wellness device with a serious evidence base. For seasonal affective disorder the effect size rivals an antidepressant, the mechanism is biologically coherent, side effects are minor, and it is cheap and self-administered. That earned reputation is precisely what is now being monetised: a "10,000 lux" lamp marketed as a generic fix for mood, sleep, focus, and energy, sold to people who have neither SAD nor the dose that made the trials work.

So this entry exists to hold a split that the marketing deliberately blurs. It is Strong Evidence for seasonal depression, and we should say so plainly because the intervention is underused. It is Moderate Evidence for everything else (non-seasonal depression as an add-on, bipolar depression with caution, perinatal, general "energy"), resting on small and largely unblindable trials. And it carries a real safety gate: for someone with bipolar disorder, unsupervised morning bright light can trigger a manic or mixed-state switch. This entry owns the clinical-device intervention (visible-light intensity and timing). It is not the sun-and-vitamin-D entry, and it is not the free morning-daylight anchor.

What bad advice this protects against, in all directions:
• Treating the lamp as a generic mood/focus cure → you buy the SAD evidence and apply it to a non-seasonal complaint where the data are Moderate at best and largely unblindable, then call it a failure when a sub-therapeutic dose does nothing.
• Dismissing light therapy as wellness gimmickry → you wave off a treatment that is non-inferior to fluoxetine in SAD and recommended first-line by a guideline body, sending a treatable winter depression untreated.
• Conflating treatment with prevention → light TREATS seasonal depression well, but the trial base for PREVENTING it is almost empty (Cochrane found one eligible trial), so a "use it from October to stop SAD" claim outruns the evidence.
• Self-starting morning light with bipolar disorder → the one place this entry refuses to defer: morning bright light can flip mania, and the only studied safe path is midday timing under clinical monitoring.
• Buying a "happy lamp" and assuming it equals the studied dose → "10,000 lux" holds only at a stated close distance and duration; used at arm's length, a cheap under-spec lamp is not the intervention the trials ran.

Evidence

1. For seasonal affective disorder, bright light has a drug-comparable effect size (Strong Evidence, the load-bearing claim). The APA-commissioned meta-analysis (Golden, Gaynes, Ekstrom et al., Am J Psychiatry 2005; 8 bright-light studies) found an effect size of 0.84 (95% CI 0.60–1.08) for bright light in SAD, with dawn simulation at 0.73 (95% CI 0.37–1.08). This 0.84 is the SAD (seasonal) subgroup figure, distinct from the 0.53 non-seasonal subgroup of the same meta-analysis (the number the light_exposure_vitamin_d entry quotes), so a reader seeing both is not looking at a contradiction. That magnitude is on par with antidepressant effect sizes. Because the review was commissioned by a professional society rather than a device maker, and the authors carried no obvious device conflict, this is the clean anchor for the seasonal verdict.

2. Light therapy is non-inferior to a first-line antidepressant for SAD, head-to-head (Strong Evidence corroboration). The Can-SAD RCT (Lam et al., Am J Psychiatry 2006; n=96) randomised SAD patients to light plus placebo pill versus fluoxetine plus placebo light. Response and remission were comparable, with light showing faster onset and fewer side effects. This is an active-comparator trial, not just sham-controlled, which is what lets us say "comparable to a drug" rather than merely "beats a dim light." Note the investigators also champion the intervention, a mild pro-light tilt worth flagging.

3. Guidelines codify the exact split this entry holds (Strong for seasonal, Moderate for non-seasonal). CANMAT's 2016 clinical guidelines (Ravindran, Lam et al.) name light therapy first-line monotherapy for seasonal MDD, and second-line monotherapy or adjunct for mild-to-moderate non-seasonal MDD. A guideline body drawing the asymmetry independently is strong support for carrying a split tier internally rather than a uniform verdict. The caveat: CANMAT is partly authored by the same Lam group that runs the trials, transparent but self-referential. No device-industry funding in the guideline.

4. Bright light works for non-seasonal depression as an add-on to an SSRI (Moderate Evidence, strongest single non-seasonal receipt). Lam et al. (JAMA Psychiatry 2016; double-blind RCT, n=122) found MADRS improvement of 16.9 for light plus fluoxetine versus 8.8 for fluoxetine alone (p=0.02); the combination (effect size ~1.11) beat light-alone (~0.80), fluoxetine-alone (~0.24), and sham. Publicly funded (Canadian Institutes of Health Research); the comparator was a generic SSRI, not a competing device, which keeps it relatively clean. It is one well-run trial, which is why it sits at Moderate pending the meta-analytic context below.

5. Taken as a whole, the non-seasonal base is weak and at high risk of bias (Moderate ceiling, the both-ways counterweight). Perera et al. (BJPsych Open 2016) found a beneficial effect across 881 participants in 20 RCTs (SMD −0.41, 95% CI −0.64 to −0.18), but only a minority of studies were low risk of bias and the pooled certainty for the primary depression-score outcome was rated GRADE "low." Al-Karawi & Jubair (J Affect Disord 2016; 9 trials) reported an SMD of −0.62 (95% CI −0.88 to −0.35) while explicitly flagging the non-blinding problem. A significant pooled signal sitting on low-certainty evidence is the definition of a Moderate ceiling, and the honesty about bias risk comes from pro-light researchers, which is reassuring rather than damning.

6. Bright light reduces bipolar depression with high remission, but the trial deliberately used MIDDAY timing to dodge the manic switch (Moderate Evidence). Sit et al. (Am J Psychiatry 2018; double-blind RCT, n=46) compared 7,000 lux white against 50 lux dim red at midday over 6 weeks: remission 68.2% versus 22.2% (adjusted OR 12.6), with no mood-polarity switches observed. The protocol design itself encodes the safety caveat: the investigators chose midday precisely because morning light is known to be riskier in bipolar disorder. NIMH-funded. Small n and an intentionally altered protocol keep this at Moderate.

7. The manic-switch risk is real but low in protected settings, and is the entry's hard safety gate (Moderate efficacy plus incidence data). Tseng et al. (2016 meta-analysis; bipolar, 9 studies, n=489) reported an SMD of −0.69. A historical review (Sit group, J Affect Disord 2018; 41 studies, 799 bipolar patients) found 0.9% switched to mania and 1.4% to hypomania, far below the 15–40% switch rate seen with antidepressants, but non-zero and timing-sensitive (women appear especially sensitive to morning light and mixed states). The load-bearing caveat: those low rates came from trials that excluded high-risk patients and used antimanic cover. Unsupervised self-use is less protected, so 0.9% is not a green light.

8. Bright light helps antepartum depression in a blinded trial (Moderate Evidence, the perinatal extension). Wirz-Justice et al. (J Clin Psychiatry 2011; double-blind placebo-controlled RCT, n=27) found response 81% for bright white versus 46% for dim placebo, and remission 69% versus 36% (p<0.05). Promising and from a foundational circadian group (Wirz-Justice, pro-light but not device-commercial), but a single trial with a very small n, so it stays at Moderate.

9. The placebo problem is structural and unsolved (a confidence-capping caveat, not a refutation). You cannot truly blind a patient to whether a light is bright or dim, so expectancy contaminates every trial. The field's own control-condition methodology reviews (including the 2024 framework paper on selecting and evaluating light-therapy controls) note that dim-light controls give obvious cues. Several trials do report non-significant between-group expectancy ratings and an antidepressant effect emerging only after ~3 weeks (a delay you would not expect from pure belief), and the SAD effect survives both sham and active-comparator (fluoxetine) designs. So the effect stands, but with an asterisk, and the critique comes from inside the light-research community rather than from contrarians.

10. Dose, timing, and device spec are load-bearing and frequently mis-sold (a practice caveat that produces false negatives). Consumer-lamp specs themselves state that "10,000 lux" holds only at a close stated distance (often ≤24 inches) for a stated duration. Independent spectrometer bench-tests of dozens of lamps have found many units fail to hit rated lux at normal viewing distance, and some flicker. The studied clinical protocol is 10,000 lux, ~30 minutes, early morning, with the lamp ~30 degrees below the line of gaze. A cheap lamp used at arm's length is not the intervention the trials ran, and prescribing "10,000 lux" without the distance-and-duration asterisk delivers a sub-therapeutic dose and a false-negative experience.

11. For PREVENTING SAD, the trial base is almost empty (Emerging Evidence for prevention specifically, much weaker than treatment). Nussbaumer-Streit et al. (Cochrane, 2019; CD011269.pub3) found only one eligible RCT (Netherlands, n=46) for light-therapy prevention of SAD, rated very-low-quality and high risk of bias. Cochrane could not conclude that light prevents winter depression. Treatment and prevention are different claims, and the contrast is itself the honest both-ways point: strong for treating, near-silent for preventing.

Mechanism

Why morning timing for SAD. The leading model is circadian: seasonal depression is associated with a delayed circadian phase in winter, and morning bright light advances that phase back toward alignment. The signal travels via the retinohypothalamic tract from melanopsin-containing intrinsically photosensitive retinal ganglion cells (ipRGCs) to the suprachiasmatic nucleus, the master clock. This is why timing is not cosmetic: morning light pulls the phase one way, evening light the other, following a phase-response logic. The same logic is why morning light is riskier in bipolar disorder, where a strong phase advance can tip toward mania.

Why intensity and distance matter. ipRGCs respond to high illuminance, and lux falls off steeply with distance from the source. A lamp rated at 10,000 lux at 24 inches can deliver a fraction of that at arm's length, dropping the retinal dose below the level the trials used. The intervention is a dose of light to the eye, not the presence of a lamp in the room, which is exactly the distinction consumer marketing erases.

Why the bipolar switch is timing-dependent. If morning light works by advancing a delayed clock, then in a system already prone to mood instability that same phase shift can overshoot into hypomania or a mixed state. Midday light produces a weaker phase signal, which is the mechanistic reason the Sit trial deliberately moved the dose to midday and saw no switches.

Why the placebo cannot be fully removed. Bright light is perceptible. A patient knows whether their light is bright or dim, so the expectancy component cannot be blinded away the way a sugar pill blinds a drug trial. The effect survives because it appears in active-comparator designs and emerges on a biological timescale (~3 weeks), but the mechanism of belief genuinely rides alongside the mechanism of photons, and honest framing holds both.

Risks And Contraindications

• Bipolar disorder is the hard gate. Unsupervised morning bright light can trigger a manic or mixed-state switch. The trial-setting switch rate (0.9% mania, 1.4% hypomania) came from protected populations with antimanic cover and high-risk exclusions, so real-world unsupervised self-use is less protected. The only studied safe path is midday light under clinical monitoring. Women appear especially sensitive to morning light and mixed states.
• Treatment is not prevention. Light treats seasonal depression well, but Cochrane found essentially no trial base for preventing it (one eligible RCT). Do not let a "use it prophylactically" claim borrow the treatment evidence.
• The SAD-evidence halo must not bleed onto general use. Non-seasonal, perinatal, and bipolar data are Moderate, resting on small, largely unblindable trials. Hold the split tier; do not present a uniform "strong" verdict.
• Under-spec devices produce false negatives. Telling a user "10,000 lux" without "measured at your actual eye-to-lamp distance, ~30 min, morning" delivers a sub-therapeutic dose and an unfair "it didn't work."
• Minor adverse effects are real. Headache, eye strain, and nausea occur, usually transient and dose-related. People with retinal disease or on photosensitising medication should check with a clinician first.
• The expectancy caveat. Because bright light cannot be fully blinded, some benefit is plausibly expectancy. The effect survives sham and active-comparator trials, so this caps confidence rather than overturning it, but honest framing keeps the asterisk.

Controversy

Nature of the controversy: a well-earned signal for one indication being stretched across the board, with error at both poles. Over-extending the SAD evidence into a generic "light box for everything" claim on one side; dismissing a drug-comparable treatment as wellness gimmickry on the other. The unblindable placebo sits underneath both, keeping every claim honest.

Position A — "Effective and underused." Bright light is a genuinely effective, drug-comparable, well-evidenced treatment for SAD that clinicians under-prescribe. Golden 2005 puts the effect size at 0.84; the Can-SAD RCT shows non-inferiority to fluoxetine; CANMAT names it first-line for seasonal depression; the circadian mechanism (morning light advancing a delayed phase via melanopsin ipRGCs) is coherent; side effects are minor; onset is fast; and it is cheap and self-administered. The honest complaint here is that a strong, safe, accessible intervention is left on the shelf.

Position B — "Over-generalised and confounded." The strong-for-SAD verdict is being over-extended into a generic mood/sleep/focus market. The non-seasonal base is real but weak (few low-risk trials, GRADE "low" on the primary outcome). The core methodological problem is that you cannot blind bright versus dim light, so expectancy contaminates everything. Bipolar use carries a genuine, timing-dependent switch risk. And consumer "happy lamps" are frequently under-spec, advertising peak lux without the distance-and-duration asterisk and sometimes flickering or under-delivering at realistic distances. The device industry profits from blurring the strong-SAD signal into a generic claim.

The funding picture. Cui bono runs both ways. On the pro side, the light-therapy device market is roughly $1.0–1.7B and growing, with home self-use the largest segment (~44%); makers (Philips, Verilux, Beurer, Northern Light) profit from stretching the SAD evidence into a generic "mood/sleep/focus/energy" claim, and affiliate "best SAD lamp" sites share the incentive. The pro-research base is also somewhat self-referential, with the Lam/CANMAT/Sit clusters both running the trials and writing the guidelines (transparent and publicly funded, but not independent). On the contrarian side, anti-pharma and biohacker influencers profit from positioning a lamp as a drug-free antidepressant "cure," over-selling it the other way and ignoring the switch and dose caveats. The cleanest signals come from the bodies with the least to sell: the APA-commissioned meta-analysis and Cochrane.

Realised Position: Treat bright-light therapy as a Strong Evidence, drug-comparable intervention for seasonal affective disorder, and say so plainly because it is underused. For everything else (non-seasonal depression as an add-on, bipolar depression, perinatal, general "energy"), present it as promising but Moderate Evidence, gated on the same details that make it work: morning timing for SAD, ~10,000 lux measured at the user's actual eye-to-lamp distance, ~30 minutes, within the first hour or two of waking, following a phase-response logic. Flag the unblindable-placebo caveat honestly rather than hiding it. Carry a hard safety note that anyone with bipolar disorder should not self-start morning light (manic-switch risk; midday plus clinical monitoring is the only studied path). Refuse the prevention claim (Cochrane found one trial). And resist both gravitational pulls: the device-marketing stretch and the drug-free-miracle stretch.

Cross-Pillar Connections

• Mental / Sleep (circadian_rhythm_optimization): the umbrella for the phase-response logic that makes morning light work and makes bipolar morning light risky. This entry is the clinical-device application of that circadian machinery.
• Sleep / Mental (morning_sunlight_exposure): the free circadian anchor (outdoor daylight on waking). Distinct from this entry, which owns the measured-intensity clinical lamp. Route users wanting the no-cost option there.
• Sleep (melatonin_dose_and_timing): the other side of the phase-shifting coin (light advances, evening melatonin can complement); both follow phase-response logic and can interact in circadian protocols.
• Mental (depression_lifestyle_interventions): the broader non-pharmacological depression toolkit into which light therapy slots as one evidence-graded lever.
• Mental / Physical (exercise_for_mood_dose_response): the sibling non-drug mood lever; both are reasonable adjuncts for non-seasonal low mood and often combine.
• Diet / Cross-pillar (light_exposure_vitamin_d): the UV-and-vitamin-D entry. Explicitly NOT the same intervention. That one owns sun exposure and D synthesis; this one owns visible-light intensity and timing for mood and circadian phase.

What would change our mind

Falsifiability: explicit upgrade/downgrade criteria from source

• We'd promote the non-seasonal use toward Strong if a large, well-blinded (or credibly active-comparator) non-seasonal MDD RCT, or a high-certainty meta-analysis, replicated the Lam 2016 augmentation effect.
• We'd retire the expectancy caveat and harden the SAD verdict if a validated, genuinely indistinguishable sham (a metameric or matched-photopic control) showed the SAD effect persisting undiminished. Conversely, a definitive demonstration that the effect collapses under proper blinding would force a downgrade.
• We'd say something about prevention only if adequately powered prevention RCTs existed. Right now we cannot, because Cochrane found a single eligible trial.
• We'd tighten the bipolar gate from "caution, midday, monitored" to "avoid without specialist supervision" if real-world pharmacovigilance showed the manic-switch rate materially higher than the 0.9–1.4% seen in protected trials.
• We'd soften the device-under-spec warning if independent lux-at-distance labelling standards became reliable.
• What would NOT move us: the SAD-specific Strong verdict (APA/Cochrane/CANMAT-grade and replicated), or the bipolar safety gate (a safety floor, not an evidence question).

Industry bias note

Structural incentives the evidence base may reflect

Cui bono runs both ways, which is why the independent bodies (APA, Cochrane) are the anchors.
• The pro/device end: the light-therapy device market is roughly $1.0–1.7B and growing ~4–5%/yr, with home self-use the largest segment (~44%). Makers (Philips, Verilux, Beurer, Northern Light) profit from stretching the well-earned SAD evidence into a generic "mood / sleep / focus / energy" claim (Amazon listings literally bundle all four) and from advertising peak lux without the distance-and-duration asterisk that makes it therapeutic. Affiliate "best SAD lamp" review sites share this incentive. The pro-research base is also somewhat self-referential: the Lam/CANMAT/Sit clusters both run the trials and write the guidelines (transparent and publicly funded, but not independent), and there is a mild academic incentive to publish positive findings in a field where the placebo cannot be blinded.
• The contrarian/wellness end: anti-pharma and biohacker influencers profit from positioning a lamp as a drug-free antidepressant "cure," over-selling it in the opposite direction and ignoring the manic-switch and dose caveats.
• The clean signal: the APA-commissioned meta-analysis (no device COI) and Cochrane (independent, methodologically strict) converge on a strong treatment effect for SAD and a near-empty prevention base. Realised anchors on the SAD-specific Strong evidence, explicitly down-rates non-seasonal and general use to Moderate, refuses the prevention claim, and resists both the device-marketing pull and the drug-free-miracle pull.

Light Therapy: Strong for Seasonal Depression, Promising Beyond It