Avoiding the sun resulted in 3,000 deaths, but avoided 40 skin cancer deaths

Risk-benefit balance of habitual ultraviolet exposure for cardiovascular, cancer, and skin cancer mortality: UK Biobank cohort study.

medRxiv preprint doi: https://doi.org/10.64898/2026.01.08.26343592; - Jan, 2026

Jiayue. Gu1 *, Andrew C. Stevenson2, Annie R. Brady1, Graeme J.M. Cowan3, Chris Dibben2, Richard B. Weller1

Objective: To examine how habitual ultraviolet (UV) exposure relates to cause-specific mortality and incidence, to quantify trade-offs between non-skin disease and skin cancer, and to explore potential circulating mediators.

Design: A population-based prospective cohort study with epidemiological and proteomic mediation analyses.

Setting: UK Biobank, recruited from 22 assessment centres across England, Scotland, and Wales.

Participants: 419 007 adults of White European ancestry with data on habitual UV exposure and follow-up for mortality and incident cardiovascular disease and cancer. A proteomic subcohort of 44 712 participants had plasma profiling.

Main outcome measures: Habitual ultraviolet exposure was summarised using Sun-BEEM (Sun-Behavioural and Environmental Exposure Model), a multidimensional score integrating environmental and behavioural indicators, categorised as low, medium, or high. Primary outcomes were all-cause, cardiovascular, and cancer mortality and incidence, and associations with Sun-BEEM categories were estimated using multivariable Cox models. Two extensions were implemented: an epidemiological extension using parametric g-computation to estimate deaths under counterfactual low and high UV scenarios; and a biological extension using proteomic mediation analyses to identify circulating proteins potentially linking UV exposure to cardiovascular and cancer mortality.

Results: Compared with low Sun-BEEM, medium and high exposure were associated with lower all-cause mortality (hazard ratio 0.89, 95% confidence interval 0.87 to 0.91; and 0.84, 0.82 to 0.87), with similar inverse associations for cardiovascular and non-skin cancer mortality. Skin cancer mortality showed no clear dose-response relationship with UV exposure, although incident keratinocyte cancers increased across Sun-BEEM categories. Counterfactual modelling suggested that, if associations are causal, a uniformly high UV pattern would prevent many more cardiovascular and other cancer deaths than the additional melanoma and keratinocyte cancer deaths. Proteomic mediation analyses implicated UV-downregulated immunoregulatory, mucosal-barrier, and cardiorenal-neuroendocrine pathways.

Conclusions: Higher habitual UV exposure, measured using a multidimensional score, was associated with lower cardiovascular and non-skin cancer mortality without clear increases in skin cancer mortality, supporting a more balanced view of sunlight and health.

What is already known on this topic

Public health advice in temperate countries mainly treats sunlight as a skin cancer hazard.
Few studies have explicitly quantified the trade-off between the potential benefits of habitual ultraviolet exposure for major non-skin diseases and its harms for skin cancer.
Mechanistic research on how ultraviolet exposure affects health outcomes has focused largely on vitamin D, with only limited work on non-vitamin D pathways.

PDF

GrassrootHealth review of the study

GRH

Low UV exposure was estimated to prevent:
39 melanoma deaths + 1 keratinocyte cancer death = 40 skin cancer deaths prevented

But it was also estimated to result in: 2,982 additional all-cause deaths

This equates to approximately 75 additional deaths for every 1 skin cancer death prevented.

MedFrame review of the study

Youtube 17 minutes

(00:01) Study Overview: A new preprint study by dermatologist Dr. Richard Weller analyzed a UK cohort of roughly 419,000 adults over 15 years to compare the overall health costs and benefits of UV light (sunlight) exposure versus avoidance. (01:30) The "Sunbeam" Score: Participants were categorized into low, medium, and high UV exposure groups based on a 4-point scoring system that factored in daily time spent outdoors, residential UV levels, solarium/sunlamp use, and sun protection habits.
(04:00) All-Cause Mortality Benefits: The data revealed a biological dose-response curve, meaning that as UV exposure increased, participants saw significant, compounding reductions in all-cause mortality, cardiovascular mortality, and non-skin cancer mortality.
(05:40) Melanoma Mortality: While the medium UV group showed a borderline increase in melanoma deaths, the high UV group showed no statistically significant increase in melanoma mortality. The researchers suggest this may be due to adaptive biological responses, such as increased skin pigmentation and epidermal thickening.
(07:22) Disease Incidence: Higher UV exposure was linked to a significant decrease in the incidence (occurrence) of non-skin cancers. However, it did significantly increase the incidence of highly treatable, non-melanoma skin cancers like basal and squamous cell carcinomas.
(08:14) Specific Mortality Reductions: In terms of exact percentages, the high UV exposure group experienced a 16% reduction in all-cause mortality, a 23% reduction in cardiovascular mortality, and an 11% reduction in non-skin cancer mortality compared to the low UV group.
(12:34) Low UV Scenario (Avoidance): The researchers calculated that if everyone in the 419,000-person cohort completely avoided UV light, it would prevent 39 melanoma deaths, but at the severe cost of 2,982 additional all-cause deaths (primarily from cardiovascular disease and other cancers).
(12:58) High UV Scenario (Exposure): Conversely, if everyone in the cohort had high UV exposure, there would be 4,736 fewer all-cause deaths, offset by a comparatively tiny increase of just 23 melanoma deaths and 12 other skin cancer deaths.
(14:00) Potential Mechanisms & Caveats: The massive epidemiological benefits of sunlight likely go beyond Vitamin D, potentially involving infrared light penetrating the body and UV-induced immunomodulation. However, the host notes
(15:15) that this data applies to habitual midlife/adult exposure; severe sunburns during childhood remain a strong predictor of elevated melanoma risk.

Full transcript

(00:01) medcra.com. Welcome to another MedCram video. I just posted on X in a bombshell preprint just released by dermatologist Richard Weller from the University of Edinburgh in Scotland. If everybody in the 419,000 UK cohort avoided UV light and were followed for 15 years, it would save 39 deaths from melanoma, but at the expense of 2,982 deaths from everything else.
(00:29) Folks, this is a preprint just dropped now that I want to review that actually looks at this cohort and measures out what the costs are for avoiding UV light and for getting UV light as a surrogate for sunlight. So, this study has not yet been peer- reviewviewed, but is on the medical archive website, which we'll put a link to in the description below.
(00:52) Before we get into that though, please visit us at medcram.com for more continuing medical education videos. And if sunlight is your thing that you want to learn about, we have a whole section on health optimization, specifically with sunlight, visit us at medcram.com. This is what the study did.
(01:10) Let's talk about it. There's about 419,000 adults in the UK. And that's important to understand that it's adults that we're talking about in terms of sunlight. So, we really can't make any implications here about children. That's important because sun exposure to the point of burning or skin damage in childhood can actually lead to more melanoma.
(01:30) That's actually been seen and we'll talk more about that. But here's an interesting score that they looked at known as sunbeam be. And it asked a yes no question in four areas. Number one, time outdoors. You got a one point if you were outdoors greater than or equal to 4 hours a day. You got a zero if it was less than 4 hours a day.
(01:52) What about residential UV light? They could actually look at it in terms of kilogjles per square meter. And if you were above 9,516 in terms of the area where you lived, you got a point. If you didn't, you got zero. And they also asked about solarium use. Cariums in the UK are these portions of the homes which are sort of like glass houses that you can go into where there's a lot of light that comes in or sunlamp use.
(02:18) And if it was greater than one time per year or equal to one time per year, you got a point. And then the last question was sun and UV protection. If you sometimes used it or usually used it, you got a point. Or if you never used it or specifically never went out into the sun, then you got a zero. You put those together and you could be divided into zero to one points is low UV, two points is medium UV and three to four points would be high UV.
(02:48) So then what they did was they followed them for about 15 years. So this was done about 15 years ago and it was released some of this data in March of 2025. This data was gathered from 2007 to about 2010. So let's take a look here at specific mortality. Of course, they did a bunch of confounders and tried to control for things, but generally speaking here, what we're talking about is mortality, the number of cases, and this is a forest plot.
(03:17) So, how do you read this? This is the center line, which is one. That's a hazard ratio of one, and these are the bars. And if these bars are completely on one side or the other, then that means there is statistical significance. If the bar includes the number one, then it's not statistically significant. There's actually one here that just reached it right at that area.
(03:37) And we'll talk a little bit more about that. So, in terms of all cause mortality, what you can see here is you've got a square and you've got a circle. The circle is actually medium versus low in terms of our here medium versus low. So, how do they stack up? And then the square here is high versus low. So, again, high versus low.
(04:00) So, how did they actually stack up? The first pair is always the circle. The second pair is the square. And so you can see here that there's a dose response curve. This is really important to understand biological gradient. There is some suggestion of causality when you talk about Bradford Hill criteria when there's a biological gradient and a dose response curve.
(04:19) And so what we're seeing here is that as we go from medium versus low to high versus low that we're moving away from the number one line here. And that is indicative of the fact that there is a biological gradient which can suggest that in fact this is the thing that is actually causing it. I know this is a prospective study. It's observational.
(04:40) You really can't say that there is causation. But the fact that the magnitude of the effect gets bigger when you increase the magnitude of the thing that you're studying that does suggest but is not guaranteed of there being causation. So let's take a look here. all-c causeed mortality there was both medium and high a reduction in all-c cause mortality cardiovascular mortality same thing even cancer mortality if we look at cancer excluding skin cancer same sort of thing Dr.
(05:11) Richard Weller is a dermatologist. He's interested in melanoma and other skin cancers like squamous cell and basilc cell cancer. Let's take a look here. Interestingly, when we look at medium versus low, this actually hits one. So, technically not statistically significant, but certainly almost all complete on one side of that line, which would seem to indicate that there is some increased melanoma death in the medium to low.
(05:40) However, as you go up on the dose, what happens is melanoma deaths become statistically insignificant. In other words, the confidence interval is just so large. Now, you would expect that it should have gone up. You have a more narrow confidence interval, but that's not what we see here. And when we look at other skin cancer, the same sort of situation.
(05:59) There's just nothing here. I mean, the confidence interval is so wide. And in fact, when we go up to a higher dose here, it's actually protective in a sense, although we would say that if it was actually statistically significant, but it's not. So, we really cannot tell. Again, this is all cause and cause specific mortality.
(06:18) Here, we're not looking at mortality, but we're actually looking at the incidence of disease. So, incident cardiovascular disease and incident for cancer, same outcomes. Again, the circle is medium versus low and the square is high versus low. What do we see here? We see that for cardiovascular disease, it's right on the line like we saw before.
(06:42) So, we can't really say that there is statistical significance in terms of reduction in incidence, but it is pretty close to that cancer. In terms of incidence, there was a risk that was statistically significant, but very low in terms of just general cancer. However, when the dose response went up, it became insignificant.
(06:59) When we look at cancer excluding skin cancers, however, notice that there was a statistically significant reduction in non-skin cancers and that dose response curve actually was even higher amplitude when that dose response went up to high versus low. What about melanoma in terms of incidence? This is not mortality. We're just talking about incidence here.
(07:22) Again notice here that in terms of incidence there was a statistically significant increase in incidence of melanoma in the medium versus low but a not statistical significant increase. In other words the confidence interval was beyond one and the high versus the low. And then when we talk about other skin cancers like basil cells and squamous cells clearly what we're seeing here is that there is an increase in the incidence of other skin cancers other than melanoma.
(07:51) So like squamous cell, basil cell, things of that nature that are generally not as severe or as deadly and can be treated locally for the most part. There are exceptions but clearly a statistical significant increase that with a dose response curve actually goes up. So that's interesting in terms of validation of this population seeing things like that that we would expect to see. Let's talk about results in words.
(08:14) All cause mortality. What did we see in the medium versus low? There was an 11% reduction in all-c causeed mortality in the high versus low. There was a 16% reduction in all cause mortality in cardiovascular mortality. In the medium versus low, there was a 18% reduction in cardiovascular mortality.
(08:36) And in the high versus low, there was a 23% reduction in cardiovascular mortality. For non-skin cancer mortality in terms of cancer, again, 8% reduction. and 11% reduction respectively. Melanoma mortality, there was about 440 deaths in this cohort over those 15 years and there was a borderline increase at medium as we talked about a 22% increase but not convincing at high.
(09:03) Why is that the case? Even though it's slightly elevated, the confidence interval includes the number one. So not statistically significant. Other skin cancer mortality, it was very imprecise and there was wide confidence intervals. Let's go to incidents again. Looking at that graph that we looked at in terms of cardiovascular disease, you see some modest reductions.
(09:23) However, the confidence intervals are near the null point which is one. So unclear whether or not that's actually happening. In terms of non-skin cancers, modest reduction here, although there's a dose response curve. In terms of incident melanoma, again it was 1.12 which is a medium increase. However, at the higher it crosses one.
(09:42) So we cannot say that there is statistical significance. However, other skin cancers, mostly keratinocy, we can see here a 14% increase in the medium and a 19% increase in the high. Both of these are statistically significant and there is a dose response curve. So, what about the validation of this? And there's a paragraph in the paper which states this.
(10:05) They say landmark analyses excluding deaths within the first two and 5 years of follow-up produce effect estimates of similar magnitude to the main analyses. That's important because you want to make sure that there isn't a reverse causation type of situation going on. And this actually suggests that reverse causation is unlikely to fully explain the observed associations.
(10:24) Stratified analyses by sex, age group, and sunburn history also showed broadly consistent patterns with no strong evidence of effect modification. Essentially, what we're saying here is we're not seeing an artifact of reversal causation. So this is an interesting analysis. What they wanted to see is everybody in the population, the cohort of 419,000, some of them had low UV, some of them had medium, some of them had high.
(10:54) They wanted to know what would happen to the deaths if everybody in that cohort had low UV exposure. If everybody followed the recommendations of not going outside into the sun, covering up, in other words, they were to answer in a negative way to all of those ways. What would happen here? we have more of a fraction in terms of percentages and then here we have actual numbers of deaths.
(11:16) What we're seeing here is the green is actually how many fewer we would see and the red is how many more we would see. So let's take a look here at actual deaths. If everybody was in the low UV, we would have about 2,000 more allcaused deaths. We would have this many more deaths in cardiovascular disease, this many more deaths in cancer, and this many more deaths excluding skin cancer.
(11:41) But how many people would have less deaths of melanoma? Well, you can see there, it's hard to see, but it's about 39. And how many people would have less non-melanoma skin cancer deaths? It would be one. What would happen if everybody was high UV? In other words, they were all able to answer the questions in a way that would put them at three or four points, which is in the high category.
(12:04) You would have about 4,000 fewer deaths, all cause. And then you can see the breakdown. How many more excess deaths would you see there in terms of melanoma? It's such a small number, it's not even showing up. So, let me actually show you in fact the numbers here. What if everyone had low UV? What if everyone had high UV scenarios? Let's say again if everybody were low in the sunbeam category, there would be 39 fewer melanoma deaths and just one fewer keratinocy cancer deaths.
(12:34) But to pay for that, you would have an additional 2,982 allcaused deaths. And you can see the breakdown there where those would be. However, again, if we said that if everybody had high sunbeam numbers here in this cohort, so in other words, they're getting out into the sun, they've got a salarium that they're using or tanning bed more than once per year.
(12:58) Notice that there would be 4736 fewer all-c causeed deaths, but at the cost of 23 additional melanoma deaths and 12 additional keratinocy cancer deaths. That's really quite phenomenal. I don't think we've really had an analysis where we've actually been able to calculate that number. When we talk about sunlight, it seems as though we hear about the skin effects of ultraviolet, but not many people are discussing the epidemiological benefits that we can actually see from sunlight in terms of deaths that are far more prevalent in terms of epidemiological
(13:33) cardiovascular deaths, cancer deaths, things of that nature. And again I would stress that this is a epidemiological study. So we really cannot purely describe that this is causal. However again Bradford Hill criteria and the dose response curve is suggestive of causation here. What's the mechanism? Those of you who have been following MedCram know that we've talked about infrared light which obviously comes with ultraviolet light when you're talking about sunlight.
(14:00) And of course ultraviolet light can make vitamin D. But another part of this study which is really interesting was the proteolytics where we looked for the actual types of biomarkers that we're seeing. So they looked at IL22 and a number of these other biomarkers which basically are other ways of mitigating a lot of this that need to be further investigated.
(14:23) But the point is that there may be much more than just vitamin D going on. And I want to leave you with what they say in the actual paper, which we'll leave a link to. Although UV radiation is a recognized cause of skin cancers, we did not observe a strong positive association between higher habitual UV exposure in middle age and skin cancer mortality.
(14:46) Melanoma mortality was modestly higher in the intermediate sunbeam category but not in the highest category and estimates were imprecise because of the small number of deaths and could be compatible with chance. This nonlinear pattern may reflect threshold or adaptive responses to chronic UV exposure such as epidermal thickening, increased pigmentation, and UV induced imunom modulation so that risk does not increase in a simple dose response fashion with cumulative dose.
(15:15) The timing of exposure is also likely to be critical and we alluded to this before, right? children and early adult intermittent holiday sun exposure promotes melanocytic navvice development on intermittently exposed sites and neas burden is one of the strongest phenotypical predictors of melanoma risk whereas UK bioank recruited participants aged 40 and older and thus captures habitual midlife exposure rather than this early life susceptibility window against this background our data suggests that within the expos exposure
(15:50) range and age span captured in this cohort. Increased midlife UV exposure in this UK population is unlikely to confer a large increase in fatal melanoma risk. Although modest effects in either direction cannot be excluded. So I thought that this study was really interesting. Yes, there is seemingly a bias.
(16:15) We think about the effects of the sun because of the direct effects on the covering of our body, which is the skin. But as we've talked about before, we now have very good evidence that long wavelengths from the sun, specifically in the infrared spectrum, are able to penetrate through the body and affect all sorts of changes on just about every single [music] cell type in our body.
(16:35) If you found this interesting, please join us for more videos at medcram.com where we have continuing medical education videos.

Avoiding the sun resulted in 3,000 deaths, but avoided 40 skin cancer deaths

Risk-benefit balance of habitual ultraviolet exposure for cardiovascular, cancer, and skin cancer mortality: UK Biobank cohort study.

GrassrootHealth review of the study

MedFrame review of the study

Full transcript

Related in VitaminDwiki