Does Nighttime Light Exposure Cause Heart Disease?

I was bitten by a vampire once. On the knee. This happened at my front door one Halloween while I was giving out candy. The vampire, I’m guessing, was around 4 years old.

I asked the parents if I should worry about rabies. I thought that was a funny line, but they just blinked and looked away. No sense of humor, I suppose.

Anyway, “real” vampires, though vulnerable to sunlight, flourish in artificially lit darkness. This distinguishes them from the rest of us, according to a study published last week. The study showed that nighttime light exposure increases the risk of cardiovascular disease (CVD).

At first glance, that’s a scary finding. Because the claim is that the mere presence of light when you sleep is dangerous.

In other words, even folks who power down their screens at a reasonable hour and get enough sleep may be at greater risk of CVD if their bedroom isn’t dark enough.

This isn’t the first study to show harmful effects of nighttime light exposure. Should we be worried? And, whether or not we trust this particular study, how dark should our bedrooms be?

I’ll start with a quick look at the new data, then address the broader issues.

The new study

This study, published last Thursday in JAMA Network Open, was led by Daniel Windred at Flinders University (Australia) and an international team, including researchers from Manchester University and Harvard Medical School.

Windred and colleagues sampled 88,095 older adults (average age 62.4) from the UK Biobank, an enormous, ongoing longitudinal project.

At the outset, the participants had no cardiovascular disease. For one week they wore light-sensing devices on their wrists 24/7. Based on nighttime light exposure, the researchers divided the sample into four groups. Here they are, from lowest to highest:

“Percentile” shows each group’s rank. For instance, 0-50% is the half of the sample with the lowest nighttime light exposure. 91-100% refers to that 10% of participants with the highest exposure. (Apparently, folks in the highest group don’t turn off their lights at night.)

“Lumens” is a measure of brightness. Here, it means brightness in the exact spot where the person is sleeping.

“Example” is something I added to put the numbers in context. This is crude. I assumed a 12’ x 12’ bedroom and a person sleeping 6 feet away from the only light source. To help with the calculations I used ChatGPT (for the first time ever in a newsletter, other than to create graphics).

The main analyses showed that after roughly a decade, people with more night light exposure had higher rates of CVD, including coronary artery disease, heart failure, heart attack, atrial fibrillation, and stroke. From the top to the bottom row, each group had higher rates of each type of CVD than the group(s) above it.

For instance, compared to the lowest half of the sample, folks in the top 10% of nighttime light exposure were around 30% more likely to develop coronary artery disease, and 50% more likely to have a heart attack.

Scary stuff, but can we trust the findings?

Too much light or too little sleep?

Strong studies are transparent. You can tell exactly what the researchers did, then you can judge for yourself whether they did it right and what the results mean.

Weak studies often hide their weaknesses. Not in an unethical way necessarily. Rather, you have to dig for key information. Maybe it’s not there – or maybe it is, but it makes the study look so bad you can imagine why it was buried in the first place.

In this study, the more nighttime light people were exposed to, the higher the rates of CVD. But is light the culprit, or were they just sleeping less?

After all, a person who’s exposed to substantial nighttime light may be sleeping in a bright room, or they may be staying up late with lights and/or screens on. We do know that sleep deprivation undermines cardiovascular health. Is the new study just a rehash?

The researchers had data on the duration of each participant’s nightly sleep. If this were a strong study, we’d see that information, and it would be a central part of data analysis.

Instead, the researchers only noted how many participants slept less than 6 hours per night. This isn’t even mentioned in the main article. Rather, it’s buried in a supplemental online file.

In keeping with the spirit of Halloween, I dug up that buried data. Here it is, organized by group:

Uh-oh.

Going from left to right, you can see that people exposed to more nighttime light were also more likely to be sleep deprived. In technical terms, we would say that nighttime light exposure is confounded with inadequate sleep. We can’t tell whether it’s the light or the lack of sleep that’s increasing CVD risk.

These data should’ve been reported in the article, and the researchers should’ve controlled for sleep duration in all analyses. In the Appendix I explain what this means. The TL; DR is that the results don’t make sense unless the effects of nighttime light exposure and sleep deprivation are disentangled.

Instead, the researchers ran a bunch of analyses, drew a bunch of conclusions, and near the end of the results section briefly noted one analysis that controlled for sleep duration. Some of the significant effects disappeared when this analysis was run. (It may not have been the right analysis in the first place, as I explain in the Appendix.)

What does the new study really tell us?

The study shows that some people are at greater risk of cardiovascular disease than others.

Which people? We can’t tell whether it’s folks with a lot of nighttime light exposure, folks who are sleep deprived, or some combination of the two.

Mainstream news and social media buzzed a bit about the new study. But if the results can’t be trusted, we shouldn’t be scaring people about ambient light in their bedrooms.

For guidance, I looked at older research on the topic.

What we know

Briefly, a number of small but well-controlled lab studies have shown that exposure to nighttime light causes unfavorable cardiovascular/cardiometabolic changes.

For instance, the best of these studies, in my opinion, published in 2022 by Northwestern researchers, looked at the effects of a single night of 100-lumen exposure. (Again, this is roughly what you’d experience in a well-lit office or kitchen.)

After just one night, participants showed increased insulin resistance, a key contributor to cardiometabolic problems ranging from hypertension to type 2 diabetes and obesity. This is consistent with findings from other lab studies as well as research on the effects of urban light pollution (which I will skip over here because the data are messy, albeit persuasive).

Interestingly, in the 2022 study, melatonin levels weren’t affected by the 100-lumen exposure. Rather, heart rate increased overall, while variability in heart rate declined. These and other changes hint that excessive nighttime light exposure creates a stress response. (Evidence for the effects of light exposure on melatonin while asleep is inconsistent.)

Here are three conclusions, followed by some practical suggestions:

1. If a person is exposed to sufficiently bright light while they sleep, the physiological effects include a stress response plus impaired sleep owing to the suppression of melatonin.

2. At relatively low levels of nighttime light exposure, the stress response may still be problematic. (Melatonin effects are unlikely, in part because only about 5 to 9% of light passes through closed eyelids and, ultimately, signals the region of the hypothalamus that controls melatonin production.)

3. Regardless of whether nighttime light exposure is low or high, the more a person wakes up at night and then opens their eyes, the more we can expect both the stress response and impaired sleep arising from melatonin suppression to undermine their health.

Practical advice

I think we should view sleep holistically. Better sleep may depend on eating well, staying active, managing stress, getting sunlight, practicing good sleep hygiene (see here), and making adjustments if we feel awful in the morning or lethargic much of the day.

Since Daylight Saving Time ends on Sunday, I also suggest that we advocate for staying off it permanently (see here for a persuasive argument and here if you’re interested in advocacy).

I would stop here, except that in the 2022 study, and in at least one other like it, nighttime light exposure diminished cardiometabolic functioning even when people reported good-quality sleep.

This is a reminder that even though you should listen to your body, science also has something useful to say.

Here’s one way to reconcile what they tell us: Don’t stress about ambient lighting when you sleep, but do keep it in mind. Prioritize sleeping enough and sleeping well. Then, keep the lights as low as you can without creating anxiety or disturbing your sleep.

If you need some light at night – or if light is unavoidable – be reassured that the studies all show variable outcomes. In other words, sleeping with lights on doesn’t guarantee more insulin resistance the next morning or a heart attack 5 years later, even if it increases the risk. Really, you should just try to sleep well. That’s the main goal.

Thanks for reading!

---

Appendix: Statistical control

Readers ask sometimes what it means to “control” for some variable, or “adjust for the influence of a covariate”. I’ll discuss this here, in plain English, then touch on what the researchers did in the new study.

Imagine that person A and person B run a 100-meter race, but person B gets a 10-meter head start and wins the race. Was B truly the faster runner? Maybe, maybe not. It’s easy to find out though. You can divide 100 meters by runner A’s time and 90 meters by runner B’s time, then compare the values. This is one way to “control for” runner B’s head start.

In the new study, folks with more nighttime light exposure also had shorter sleep. You might say that sleep deprivation gave them a “head start” on cardiovascular disease. Using mathematical procedures a bit more complicated than in the runner example, we can control for the effects of that head start, in order to disentangle the impact of nighttime light exposure from the effects of sleep deprivation.

In the new study, the researchers controlled for a number of variables in their main analyses but not sleep duration, which is unfortunate, for the reasons I discussed in the main text. In the one supplementary analysis that included sleep duration as a covariate, it’s not clear how the variable was treated. The details matter. You could penalize runner B by adding 1.5 seconds to their time. That’s one way of “controlling for” the head start, but it’s a terrible approach if you want to know who truly ran more quickly.