The decision-critical fact: “sun exposure” is not one exposure for melanoma. The pattern matters
more than the dose. Intermittent, intense, burning exposure raises risk; regular chronic (occupational)
exposure does not — and the single biggest measurable signal is a history of sunburn. Facet of the
sun-uv-exposure cluster (nucleus: Sun Exposure UV and Health); this page holds the harm arm.
The pattern distinction (Gandini 2005, gold MA — 57 studies, 38,671 cases)
type-B — one term (“sun exposure”) resolves into distinct objects with opposite risk directions (Gandini et al., 2005):
| Exposure pattern | Pooled RR (highest vs lowest) | 95% CI | Direction |
|---|---|---|---|
| Sunburn history | 2.03 | 1.73-2.37 | strong risk |
| Intermittent (recreational/vacation) | 1.61 | 1.31-1.99 | risk |
| Total (all kinds) | 1.34 | 1.02-1.77 | weak risk |
| Chronic / occupational | 0.95 | 0.87-1.04 | null (slight inverse) |
- On the 19 studies reporting both patterns, intermittent (RR 1.46) exceeded chronic (RR 1.09), P=0.015 — the split is not an artifact of comparing different study sets (Gandini et al., 2005).
- Sunburn risk climbs with latitude (>=50 deg RR 2.54) and fair-skin prevalence; childhood sunburn RR 2.24. Funnel asymmetry suggests the sunburn estimate is inflated (trim-and-fill RR 1.73) — still a clear positive (Gandini et al., 2005).
Do NOT read the chronic-exposure inverse as occupational sun protects. The reference category is low continuous exposure, which itself contains high-intermittent people — a confounded contrast, not a protective effect (Gandini et al., 2005):
«It is important to stress that it is unlikely that the inverse association with chronic sun exposure means that occupational sun exposure protects against melanoma … The reference category for occupational sun exposure is low continuous pattern sun exposure, which will include people with high intermittent pattern sun exposure, as well as people with low sun exposure of any kind.»
Mechanism [@gandini2005sun]. Intermittent bursts hit unprotected skin -> high UV transmission to melanocytes; regular exposure on tanned, thickened skin is blocked at the epidermis. This is why pattern beats total dose — and why sunburn (intense UV reaching melanocytes) is the sharpest marker. Caveat: the estimates are observational (recall bias, control selection), and naevi were treated as an intermediate, not adjusted for.
The intervention leg (Green 2011, the only sunscreen->melanoma RCT)
Nambour Skin Cancer Prevention Trial, N=1,621 fair-skinned Australian adults, daily SPF-16 vs discretionary sunscreen 1992-96, followed to 2006 (~15 y) (Green et al., 2011):
- All melanoma: HR 0.50 (0.24-1.02, P=.051) — borderline. Absolute 11/812 (1.3%) vs 22/809 (2.7%).
- Invasive melanoma: HR 0.27 (0.08-0.97), a 73% reduction — significant but the authors label the invasive subgroup exploratory.
- In situ: HR 0.73 (0.29-1.81), NS.
«Invasive melanoma was reduced by 73% in the daily sunscreen group (HR, 0.27; 95% CI, 0.08 to 0.97; P .045) …» (Green et al., 2011)
Outdoor time was similar between arms, so the effect is not from the intervention group spending less time in the sun. Only-trial-of-its-kind; secondary endpoint; small event counts; A. C. Green disclosed L’Oreal research funding.
How the two sources relate — complementary, NOT an independence-convergence
Parameter table (BLOCKING before any cross-source claim):
| Parameter | Gandini 2005 | Green 2011 | Same quantity? |
|---|---|---|---|
| Quantity | observational risk-factor RR (highest vs lowest exposure) | RCT intervention HR (daily vs discretionary sunscreen) | NO — etiologic association vs intervention effect |
| Exposure contrast | intermittent 1.61 / sunburn 2.03 / chronic 0.95 | sunscreen use vs discretionary | NO |
| Outcome | histologically-confirmed cutaneous melanoma incidence | primary cutaneous melanoma incidence | outcome yes; effect measure no |
- Because the fourth column is NO on the effect measure, the RR and the HR are not poolable and must not be equated — Gandini establishes UV pattern causes melanoma (etiology); Green shows protection reduces it (intervention). Jointly they make UV’s causal, modifiable role hard to dismiss, but this is not type-E independent backing: Adèle Green authored constituent case-control studies inside Gandini’s pool (Green 1985), so the RCT and the MA share an author and are not independent routes (inferred from Gandini et al., 2005; Green et al., 2011).
Decision relevance
- The big lever for melanoma is avoiding intermittent burning, not minimizing total sun. Sunburn history carries the largest RR and is the most actionable target.
- Sunscreen is the one randomized protective intervention, plausibly reducing invasive melanoma — but the trial is small/borderline and the invasive result exploratory. It reduced melanoma without cutting outdoor time.
- Stratum: risk concentrates in fair skin (Fitzpatrick I-II) and at high latitude / high ambient UV; early-life exposure matters (migrant data). The net decision — how this harm trades against the observational sun-exposure mortality benefit — lives on Sun Exposure UV and Health.