F. Perry Wilson, MD
@fperrywilson.bsky.social
Director, Clinical and Translational Research Accelerator @Yale. Columnist @medscape.
How Medicine Works and When It Doesn't in bookstores now!
How Medicine Works and When It Doesn't in bookstores now!
Full breakdown (with the data) in my column this week: buff.ly/hEeFEgc
February 3, 2026 at 12:29 AM
Full breakdown (with the data) in my column this week: buff.ly/hEeFEgc
So in the end, whether the noise is white, blue, brown or pink, silence, it turns out, is golden.
February 3, 2026 at 12:29 AM
So in the end, whether the noise is white, blue, brown or pink, silence, it turns out, is golden.
One finding that concerns me: the authors warn against using broadband noise for newborns and toddlers. REM sleep makes up about 50% of a newborn's sleep and is essential for neurodevelopment. If pink noise suppresses REM, that might not be a risk worth taking. (11/13)
February 3, 2026 at 12:29 AM
One finding that concerns me: the authors warn against using broadband noise for newborns and toddlers. REM sleep makes up about 50% of a newborn's sleep and is essential for neurodevelopment. If pink noise suppresses REM, that might not be a risk worth taking. (11/13)
The earplugs only started failing at the highest noise level tested (65 dBA), a level that exceeds 99% of bedroom noise events measured in a prior field study. For the real world, earplugs had it covered. (10/13)
February 3, 2026 at 12:29 AM
The earplugs only started failing at the highest noise level tested (65 dBA), a level that exceeds 99% of bedroom noise events measured in a prior field study. For the real world, earplugs had it covered. (10/13)
But guess what DID work? Earplugs. Plain old foam earplugs.
On earplug nights, sleep was statistically indistinguishable from the silent control night. Same deep sleep. Same REM. They recovered 72% of the noise-induced deep sleep loss. (9/13)
On earplug nights, sleep was statistically indistinguishable from the silent control night. Same deep sleep. Same REM. They recovered 72% of the noise-induced deep sleep loss. (9/13)
February 3, 2026 at 12:29 AM
But guess what DID work? Earplugs. Plain old foam earplugs.
On earplug nights, sleep was statistically indistinguishable from the silent control night. Same deep sleep. Same REM. They recovered 72% of the noise-induced deep sleep loss. (9/13)
On earplug nights, sleep was statistically indistinguishable from the silent control night. Same deep sleep. Same REM. They recovered 72% of the noise-induced deep sleep loss. (9/13)
OK so maybe pink noise doesn't help on quiet nights. But can it at least mask environmental noise?
Nope. You get the worst of both worlds. Less deep sleep AND less REM sleep. Total sleep time dropped by about 15 minutes. The noise machine made things worse. (8/13)
Nope. You get the worst of both worlds. Less deep sleep AND less REM sleep. Total sleep time dropped by about 15 minutes. The noise machine made things worse. (8/13)
February 3, 2026 at 12:29 AM
OK so maybe pink noise doesn't help on quiet nights. But can it at least mask environmental noise?
Nope. You get the worst of both worlds. Less deep sleep AND less REM sleep. Total sleep time dropped by about 15 minutes. The noise machine made things worse. (8/13)
Nope. You get the worst of both worlds. Less deep sleep AND less REM sleep. Total sleep time dropped by about 15 minutes. The noise machine made things worse. (8/13)
REM sleep is not expendable. It's critical for memory consolidation, emotional regulation, and brain plasticity. REM abnormalities are linked to depression, PTSD, anxiety, and may be prodromal for neurodegenerative disease. (7/13)
February 3, 2026 at 12:29 AM
REM sleep is not expendable. It's critical for memory consolidation, emotional regulation, and brain plasticity. REM abnormalities are linked to depression, PTSD, anxiety, and may be prodromal for neurodegenerative disease. (7/13)
Now the surprise. Pink noise alone (no jets, no sirens, just that soft steady hum) reduced REM sleep by about 19 minutes.
FWIW, people didn't notice. Their subjective sleep ratings were no different from the silent night. (6/13)
FWIW, people didn't notice. Their subjective sleep ratings were no different from the silent night. (6/13)
February 3, 2026 at 12:28 AM
Now the surprise. Pink noise alone (no jets, no sirens, just that soft steady hum) reduced REM sleep by about 19 minutes.
FWIW, people didn't notice. Their subjective sleep ratings were no different from the silent night. (6/13)
FWIW, people didn't notice. Their subjective sleep ratings were no different from the silent night. (6/13)
First, the obvious: environmental noise hurts your sleep. It reduced deep N3 sleep by about 23 minutes compared to silence. (5/13)
February 3, 2026 at 12:28 AM
First, the obvious: environmental noise hurts your sleep. It reduced deep N3 sleep by about 23 minutes compared to silence. (5/13)
The conditions:
Silent control night
Environmental noise (jets, trains, drones, a baby crying, every 4-6 min)
Pink noise alone
Environmental noise + pink noise
Environmental noise + earplugs
Nice crossover design. (4/13)
Silent control night
Environmental noise (jets, trains, drones, a baby crying, every 4-6 min)
Pink noise alone
Environmental noise + pink noise
Environmental noise + earplugs
Nice crossover design. (4/13)
February 3, 2026 at 12:28 AM
The conditions:
Silent control night
Environmental noise (jets, trains, drones, a baby crying, every 4-6 min)
Pink noise alone
Environmental noise + pink noise
Environmental noise + earplugs
Nice crossover design. (4/13)
Silent control night
Environmental noise (jets, trains, drones, a baby crying, every 4-6 min)
Pink noise alone
Environmental noise + pink noise
Environmental noise + earplugs
Nice crossover design. (4/13)
A rigorous new study from Mathias Basner and colleagues at UPenn put this to the test. 25 volunteers spent a full week in a sleep lab, wired up with polysomnography, experiencing a different soundscape each night. (3/13)
buff.ly/mhbYgSv
buff.ly/mhbYgSv
February 3, 2026 at 12:28 AM
A rigorous new study from Mathias Basner and colleagues at UPenn put this to the test. 25 volunteers spent a full week in a sleep lab, wired up with polysomnography, experiencing a different soundscape each night. (3/13)
buff.ly/mhbYgSv
buff.ly/mhbYgSv
Millions of people sleep with some kind of background noise. White noise machines, pink noise apps, fans, AC units.
The top 5 "white noise" videos on YouTube have been watched more than 700 million times.
But the evidence that any of this helps? Pretty thin. (2/13)
The top 5 "white noise" videos on YouTube have been watched more than 700 million times.
But the evidence that any of this helps? Pretty thin. (2/13)
February 3, 2026 at 12:28 AM
Millions of people sleep with some kind of background noise. White noise machines, pink noise apps, fans, AC units.
The top 5 "white noise" videos on YouTube have been watched more than 700 million times.
But the evidence that any of this helps? Pretty thin. (2/13)
The top 5 "white noise" videos on YouTube have been watched more than 700 million times.
But the evidence that any of this helps? Pretty thin. (2/13)
Bottom line: Barring specific risk factors, moderate alcohol intake has very little impact on health.
How much should you drink? Just enough to find enjoyment without regret.
Probably a drink or two.
Full breakdown here: buff.ly/EUlU5Nb (14/14)
How much should you drink? Just enough to find enjoyment without regret.
Probably a drink or two.
Full breakdown here: buff.ly/EUlU5Nb (14/14)
January 28, 2026 at 7:46 PM
Bottom line: Barring specific risk factors, moderate alcohol intake has very little impact on health.
How much should you drink? Just enough to find enjoyment without regret.
Probably a drink or two.
Full breakdown here: buff.ly/EUlU5Nb (14/14)
How much should you drink? Just enough to find enjoyment without regret.
Probably a drink or two.
Full breakdown here: buff.ly/EUlU5Nb (14/14)
Interestingly, a recent Sports Medicine study of ~25,000 adults found alcohol was associated with higher mortality—but exercise ablated that risk entirely.
Maybe the J-curve just reflects that moderate drinkers tend to have other healthy habits.
buff.ly/ddmFEr9 (13/14)
Maybe the J-curve just reflects that moderate drinkers tend to have other healthy habits.
buff.ly/ddmFEr9 (13/14)
January 28, 2026 at 7:46 PM
Interestingly, a recent Sports Medicine study of ~25,000 adults found alcohol was associated with higher mortality—but exercise ablated that risk entirely.
Maybe the J-curve just reflects that moderate drinkers tend to have other healthy habits.
buff.ly/ddmFEr9 (13/14)
Maybe the J-curve just reflects that moderate drinkers tend to have other healthy habits.
buff.ly/ddmFEr9 (13/14)
What about cardiovascular "benefits"?
Observationally, you see J-curves suggesting 1-2 drinks/day is protective.
Mendelian randomization tells a different story: genetic predisposition to alcohol INCREASES risk of coronary disease, AFib, and AAA.
buff.ly/btjqteT (12/14)
Observationally, you see J-curves suggesting 1-2 drinks/day is protective.
Mendelian randomization tells a different story: genetic predisposition to alcohol INCREASES risk of coronary disease, AFib, and AAA.
buff.ly/btjqteT (12/14)
January 28, 2026 at 7:46 PM
What about cardiovascular "benefits"?
Observationally, you see J-curves suggesting 1-2 drinks/day is protective.
Mendelian randomization tells a different story: genetic predisposition to alcohol INCREASES risk of coronary disease, AFib, and AAA.
buff.ly/btjqteT (12/14)
Observationally, you see J-curves suggesting 1-2 drinks/day is protective.
Mendelian randomization tells a different story: genetic predisposition to alcohol INCREASES risk of coronary disease, AFib, and AAA.
buff.ly/btjqteT (12/14)
Genetic predisposition to MODERATE drinking didn't increase breast cancer risk. But predisposition to PROBLEMATIC drinking did.
Suggests moderate drinking is a proxy for risk, not a direct cause. Heavy drinking likely IS a cause.
buff.ly/bW9Ch7V (11/14)
Suggests moderate drinking is a proxy for risk, not a direct cause. Heavy drinking likely IS a cause.
buff.ly/bW9Ch7V (11/14)
January 28, 2026 at 7:46 PM
Genetic predisposition to MODERATE drinking didn't increase breast cancer risk. But predisposition to PROBLEMATIC drinking did.
Suggests moderate drinking is a proxy for risk, not a direct cause. Heavy drinking likely IS a cause.
buff.ly/bW9Ch7V (11/14)
Suggests moderate drinking is a proxy for risk, not a direct cause. Heavy drinking likely IS a cause.
buff.ly/bW9Ch7V (11/14)
But all these are observational studies. Correlation, not causation.
Enter Mendelian randomization using genetic variants that predispose people to drink more as "quasi-random" variables.
If alcohol causes cancer, having those genes should increase risk. (10/14)
Enter Mendelian randomization using genetic variants that predispose people to drink more as "quasi-random" variables.
If alcohol causes cancer, having those genes should increase risk. (10/14)
January 28, 2026 at 7:46 PM
But all these are observational studies. Correlation, not causation.
Enter Mendelian randomization using genetic variants that predispose people to drink more as "quasi-random" variables.
If alcohol causes cancer, having those genes should increase risk. (10/14)
Enter Mendelian randomization using genetic variants that predispose people to drink more as "quasi-random" variables.
If alcohol causes cancer, having those genes should increase risk. (10/14)
The exception: breast cancer.
A UK study of 1+ million women found virtually no increased cancer risk in non-smokers who drank modestly, EXCEPT breast cancer, which showed a linear increase regardless of smoking.
(Alcohol affects estrogen levels).
buff.ly/pVEI2uj (9/14)
A UK study of 1+ million women found virtually no increased cancer risk in non-smokers who drank modestly, EXCEPT breast cancer, which showed a linear increase regardless of smoking.
(Alcohol affects estrogen levels).
buff.ly/pVEI2uj (9/14)
January 28, 2026 at 7:46 PM
The exception: breast cancer.
A UK study of 1+ million women found virtually no increased cancer risk in non-smokers who drank modestly, EXCEPT breast cancer, which showed a linear increase regardless of smoking.
(Alcohol affects estrogen levels).
buff.ly/pVEI2uj (9/14)
A UK study of 1+ million women found virtually no increased cancer risk in non-smokers who drank modestly, EXCEPT breast cancer, which showed a linear increase regardless of smoking.
(Alcohol affects estrogen levels).
buff.ly/pVEI2uj (9/14)
Cancer risk is more complicated.
Modest alcohol is associated with slightly increased GI/breast/liver cancer risk, but at moderate levels, the risk clusters among smokers.
Basically no signal in non-smoking men.
buff.ly/nSsoQVV (8/14)
Modest alcohol is associated with slightly increased GI/breast/liver cancer risk, but at moderate levels, the risk clusters among smokers.
Basically no signal in non-smoking men.
buff.ly/nSsoQVV (8/14)
January 28, 2026 at 7:46 PM
Cancer risk is more complicated.
Modest alcohol is associated with slightly increased GI/breast/liver cancer risk, but at moderate levels, the risk clusters among smokers.
Basically no signal in non-smoking men.
buff.ly/nSsoQVV (8/14)
Modest alcohol is associated with slightly increased GI/breast/liver cancer risk, but at moderate levels, the risk clusters among smokers.
Basically no signal in non-smoking men.
buff.ly/nSsoQVV (8/14)
Now for the rest of us. What are the actual harms of 1-2 drinks?
Blood pressure: Pretty linear relationship. More alcohol = higher BP. Not huge (~10 mmHg systolic at the high end), but matters if you're already struggling with it.
buff.ly/oIKeYbQ (7/14)
Blood pressure: Pretty linear relationship. More alcohol = higher BP. Not huge (~10 mmHg systolic at the high end), but matters if you're already struggling with it.
buff.ly/oIKeYbQ (7/14)
January 28, 2026 at 7:46 PM
Now for the rest of us. What are the actual harms of 1-2 drinks?
Blood pressure: Pretty linear relationship. More alcohol = higher BP. Not huge (~10 mmHg systolic at the high end), but matters if you're already struggling with it.
buff.ly/oIKeYbQ (7/14)
Blood pressure: Pretty linear relationship. More alcohol = higher BP. Not huge (~10 mmHg systolic at the high end), but matters if you're already struggling with it.
buff.ly/oIKeYbQ (7/14)
The "sick quitter effect."
Many non-drinkers used to drink... and stopped because of health problems.
A 2024 review found that of 25 studies on alcohol and mortality, only 5 properly accounted for this. Those 5? No J-shaped curve at all.
buff.ly/jrS0Xe6 (6/14)
Many non-drinkers used to drink... and stopped because of health problems.
A 2024 review found that of 25 studies on alcohol and mortality, only 5 properly accounted for this. Those 5? No J-shaped curve at all.
buff.ly/jrS0Xe6 (6/14)
January 28, 2026 at 7:46 PM
The "sick quitter effect."
Many non-drinkers used to drink... and stopped because of health problems.
A 2024 review found that of 25 studies on alcohol and mortality, only 5 properly accounted for this. Those 5? No J-shaped curve at all.
buff.ly/jrS0Xe6 (6/14)
Many non-drinkers used to drink... and stopped because of health problems.
A 2024 review found that of 25 studies on alcohol and mortality, only 5 properly accounted for this. Those 5? No J-shaped curve at all.
buff.ly/jrS0Xe6 (6/14)
And if you don't currently drink? Don't start just because you've heard a drink or two is "healthy."
You've probably seen those J-shaped curves showing non-drinkers have HIGHER mortality than moderate drinkers.
But there's a problem. (5/14)
You've probably seen those J-shaped curves showing non-drinkers have HIGHER mortality than moderate drinkers.
But there's a problem. (5/14)
January 28, 2026 at 7:46 PM
And if you don't currently drink? Don't start just because you've heard a drink or two is "healthy."
You've probably seen those J-shaped curves showing non-drinkers have HIGHER mortality than moderate drinkers.
But there's a problem. (5/14)
You've probably seen those J-shaped curves showing non-drinkers have HIGHER mortality than moderate drinkers.
But there's a problem. (5/14)