Sleep Studies- What You Should Know
Since I am considered a “sleep expert” by many, I usually get news articles sent to me (on a daily basis) that have something to do with sleep. Occasionally, I am sent peer-reviewed studies, regarding sleep. Every great once and a while, I will get bombarded with something in between - usually when the information seems to contradict what I (or other professionals) say.
Enter “Natural Sleep and Its Seasonal Variations in Three Pre-industrial Societies” (Yetish et al., 2015, Current Biology 25, 1-7) and myriad other “scientific publications” - like the New York Times, USA Today, Local news etc.)
First, I want to be clear that I am not criticizing the science of this article at all. Quite the opposite: I’m such a nerd that I wish I had been part of the research. In fact, I will be pointing out that this article very clearly validates what most of us “sleep experts” have been saying for decades. Of course, the media may have put a sensationalistic spin on the article, but that’s none of my business.
Secondly, I want to make it clear that I am not a sleep researcher. I am a physician - which I believe qualifies me to analyze research done on humans. In my professional life, I focus on optimizing the health of modern humans, with an eye towards how we evolved. I do this by focusing on four “pillars” of health:
- Activity (exercise)
- Stress Control (spirituality, meditation, breathing etc.)
So, I felt compelled to share my “expert” opinion on the paper and what we should take from it. Here it goes:
I only skimmed the mass media pieces that were sent to me, and instead started by reading the actual peer-reviewed research publication. In doing so, I considered many variables; the design of the study (Randomized Clinical trial? Observational? Prospective? Etc.), and their “methods” for collecting data. Oh wait, to be truthful, I almost always I first look to see who funded the study. This paper seems to have no obvious financial agenda.
Next, I look at all of their graphs and tables, and data, most of this falls under the category of “results.” Then I read their “discussion” and/or “conclusion”, and figure out if it makes sense to me. In other words; do the data support the conclusion? Usually, the research article itself is worded in a way that the conclusion can be at least defended—given the data. The headlines of popular media know no such limitation.
So, I wasn’t overly surprised to learn that the paper does NOT—in fact—claim what many news agencies report—in their headlines. Nonetheless, I was a little surprised to see a bullet point under “Highlights” (in the actual paper), that reads: ‘They do not sleep more than “modern” humans, with average durations of 5.7-7.1 hr’. Now that may be a defendable statement, but it may have to be defended by semantics—such as; “The tribes we studied are modern humans”, or “They sleep the same number of hours as a modern human that sleeps 5.7-7.1 hours”, or something a little slippery like that.
However, what their data actually shows is that the range of “sleep period” - which means from when the person first fell asleep until they woke up for the day - was 6.9 to 8.5 hours. All sleep organizations that I am aware of, estimate (through clinical and observational research) the average adult human needs approximately 7.5 hours of sleep per night - plus or minus 30 minutes - so far, so good. That 7.5 hour recommendation is also referring to “sleep period.”
Where it gets a little nebulous (in my mind) is why they chose to report 5.7-7.1 hours as the “amount of sleep” their subjects got, when their data showed 6.9-8.5 hours of what we all recognize as “sleep”?
This is where we get into methods:
We have to choose a “method” by which to collect data. This could be self-reporting, researchers just looking at them, surveys, journaling, video tracking, all the way up to the polysomnograph - which is considered the “gold standard” for measuring sleep. The study we are discussing was done using something called “actigraphy,” which literally uses movement to estimate sleep time. This is how things like Fitbit, and Jawbone (and the like) track sleep. This is a reasonable “method”, as doing polysomnography (PSG) on all of these people would be an unreasonably complex and a likely misleading method. If you’ve ever had a PSG, you know what I mean.
Okay, so lets keep reading . . . they used a brand of actigraphy called “Actiwatch-2.” The first question in anyone’s mind is likely; I wonder how accurate Actiwasch-2’s are at measuring sleep. No problem: in the summary paragraph on page one they list their references on the accuracy of the Actiwatch “(4-7).” Oops! We’ve hit a snag! If one takes the time to read references 4-7 he or she would find that:
4 = A study on how osteoporotic fractures may influence how actigraphy is classified.
5 = A study on sleep with Hepatitis C infected people
6 = “The role of actigraphy in the assessment of primary insomnia: a retrospective study” (sleep medicine 2014, Jan 15).
7 = Sleep-wake cycles in people with a history of mood disorders.
So now is as good as of time as ever to explain the concept of “generalization.” When we do a study, regardless of how conclusive it is, we need to know how “generalizable” the data/information/conclusion is. I’ll give a playful example; if we were studying the average amount of muscle mass in 25 year-old men, data would be significantly different if the men were Navy SEALs or grad students at MIT. Furthermore, it would be a stretch to make any assumptions about the amount of muscle mass we should aim for in a 55 year-old woman - based on our research example. This is why things like population size, randomization, demographics etc. are always taken into consideration when designing a study - and make good science really hard to do.
I’m sure you can see where I’m going with this; “osteoporotic fractures”? Hepatitis C? Mood disorders? But, reference number 6 seems relevant: “The role of actigraphy in the assessment of primary insomnia: a retrospective study.” Okay, so retrospective isn’t ideal, and we aren’t studying insomniacs (in fact the populations studied did not even have a word in their vocabulary for insomnia). But . . . I read the article - because it seemed (to me) the most generalizable and relevant to this paper.
While reading this reference paper, I was surprised by what they list in their “Results” section; wake after sleeping, number of times people woke up, how much people moved around, and something called the “fragmented index” looked very different when comparing people with insomnia and without insomnia. However, their data did NOT show a difference in total time asleep or any difference between the two group’s waking time. In other words, the “Actiwatch” was not able to show if there was any difference in how much people with insomnia vs. non-insomniacs slept, or even when they woke up for good.
This led the researchers of said article to conclude that the device they were using “Actiwatch” “did not reproduce the information found with other devices, in other clinical trials studying insomnia.” Leading to their suggesting, “the need to adopt shared technical solutions for actigraphy.” Fair enough.
But, surely Yetish et al. considered that conclusion when choosing their method? So, I had to read the other “reference” papers too. The one that I scoffed at originally “Osteoporotic Fractures in Men (MrOS) Study Group (2011) [from the Journal of Clinical Sleep Medicine 7, 357-367] (reference #4), was actually a pretty damn good paper (yes, I know how geeky that sounds). Anyways, this study also used the “Actiwatch.” They found that actigraphy vs. PSG “differed significantly”, and conclude that one possible mode using the Actiwatch “corresponded moderately” to PSG—but, that the Actiwatch both over, and underestimated total time asleep. In other words; “actigraphy isn’t THAT bad.” Yetish et al. mention that the Actiwatch can over estimate sleep, but did not mention that it could also under-estimate sleep.
So why am I beating this actigraphy data to death? Because some forms of media are singling out that bullet that I mentioned above: “They do not sleep more than “modern” humans, with average duration of 5.7-7.1 hr,” which is kind of tied to their actigraphy data.
So, where does that 5.7-7.1 number come from? To be honest, I can’t definitively say. I think I know. But, “sleep” itself is not a straightforward concept. What does it mean to be asleep? Most people can’t really say. In sleep research, we use brain wave patterns to say that somebody is asleep or not asleep, and we combine those brain wave recordings with other variables like respiratory rate, pulse, oxygenation of red blood cells, etc. and try to figure out if people are having “normal sleep architecture”- a phrase that means you are going in and out of various sleep stages in a predictable pattern. Actigraphy cannot show this. As I said earlier, it measures movement, and extrapolates that into sleep/no-sleep algorithms. I have read the study several times, and from what I can tell, they are saying that the researchers are seeing something on actigraphy data that suggest the subjects are awake for a significant time between when they go to sleep, and when they get up for the day (sleep period).
However, in the first paragraph of page four they note that “Nocturnal awakenings were also infrequent”- meaning the subjects didn’t wake up very often (at least significantly wake up) during their sleep period. Again, “sleep period” means from the time they first fell asleep, until the time they got up for the day.
In the summary on page two, they note “Sleep periods . . . averaged 6.9-8.5 hr, with sleep durations of 5.7-7.1 hr, . . .” They do refer to some graphics (Fig. 1B) but there are no labels on the X or Y axis, so I’m not clear on the data.
In the supplemental material they give a graph of temperature and time (Fig. S1) that shows the sleep onset occurring around 10 PM and sleep offset (waking up for the day) occurring around 6 AM.
On page six of the paper they record the nights as lasting 11-13 hours (summer-winter).
On page one they list the average sleep onset time as being 3.3 hours after sundown and average sleep offset being “usually before sunrise.”
Later (page 4 paragraph 4) they say that average waking was approximately 1 hour before sunrise, but approximately 1 hour after sunrise, in one group studied during the shorter summer nights.
By my math that leaves the average night in the study at 12 hours, going to sleep 3.3 (12-3.3) hours after sundown, waking from 1 hour before dawn to 1 hour after dawn (-1 +1=0) meaning an average of 8.7 hours of sleep per night—throughout the year.
Not very sophisticated math, but I think it’s enough to dispel the misconception that these people are all sleeping (based on actigraphy) 5.7-7.1 hours.
Now, I realize that this is all getting tedious and confusing at this point. So, let’s just say that my conclusion is slightly different (in at least one metric), and I am open to being wrong here. If there is something about the data that I have missed, I welcome the opportunity to be corrected and learn something new.
On page six they report “They do not regularly sleep more than most individuals in industrial societies” and then reference a genetic study, the osteoporosis fracture study, the hepatitis C study, the actigraphy study, and the mood disorder studies mentioned earlier. They add one additional study—that I did not read—“Habitual sleep/wake patterns in the Old Order Amish”. It seems to me that their actigraphy data is being confused with the fact that the average American is only in bed for 6.5 hours per night. These people are in their version of a bed (an animal hide on the ground) for long enough to go to sleep, and get up 7-8.5 hours later.
The reason I titled this: “Move on: Nothing to see here,” is because of the overwhelming amount of metrics in which this article supports decades worth of clinical, and observational data.
If any of you have been unlucky enough to view one of my lectures or be a client or patient of mine you will have heard me quote the same science over and over again, with regards to when and how much to sleep. Let’s compare conclusions; I have listed myself as “DP” (Doc Parsley), and the data from this article “NS” (Natural Sleep).
DP: “Eat how you evolved to eat (Paleo-ish template).”
NS: “The San eat meat from local game, baobab fruit, fish and berries. The Hadza eat game meat, baobab fruit, honey, tubers, and berries. The Tsimane eat a mix of hunted game, fish, and cultivated plant foods, primarily rice, manioc and plantains”. (supplementary experimental procedures page one)—seems pretty “Paleo-ish.”
**Note the lack of grains, breads, cereals, flours, sugar, chemicals**
DP: “Block the blue light entering your eyes for 3-4 hours before bed, use gaming glasses, computer programs like f.lux, dim your lights, get special light bulbs w/o blue light (Philips) . . .”
NS: “Average sleep onset was 3.3 hours after sunset” pg.2, “. . . Actiwatch measured light levels [from small fires] remained below 5.0 lux . . .” pg.4
DP: “The average adult needs 7.5 hours (+/- 30 minutes per night) to be optimal”
NS: “Sleep periods . . . averaged 6.9-8.5 hours” pg.2 summary.
DP: “Keep electronics out of your bedroom” & “shift work is bad for you.”
NS: “We show here that sleep occurs almost entirely during the dark periods in these traditional societies . . . Striking difference in the sleep onset and offset times as a function to light exposure . . .” pg. 7, paragraph 1—they even reference a supportive study  “Access to Electric light is associated with shorter sleep duration in traditionally hunter-gatherer community. J. Biol. Rhythms 30, 342-350.”
DP: “Sleep in a cool, dark room. Ideal sleeping temperature is between 64-68 degrees F.”
NS: “We found that the nocturnal sleep period in the Hadza was always initiated during a period of falling ambient temperature (Fig S1).” Pg.5 paragraph 2.
**Fig S1 shows “sleep period” initiating at approximately 65 degrees Fahrenheit**
DP: “Our ancestors would have slept more during longer nights of the winter, and less during summer nights.”
NS: “Sleep is strongly modulated by the seasons, averaging 53-56 min longer in the winter . . .” pg.6 paragraph 3 **and these subjects live much closer to the equator than we do**
DP: “Our ancestors—once adjusted for traumatic death—did not live shorter life-spans than us.”
NS: “[this population] . . . adults have lower levels of blood pressure and atherosclerosis and higher levels of physical fitness than industrial populations. Many live into their 60s, 70s, 80s and beyond.” Pg.2, paragraph 3.
DP: “We have effectively taken ourselves off of the planet that we evolved on. We make it dark in the day, light in the night, cold when it’s hot, hot when it’s cold, and eat all sorts of stuff that we have invented.”
NS: In these societies, electricity and its associated lighting and entertainment distractions are absent, as are cooling and heating systems. . . . pg.3, paragraph 1.
DP: “BMI is useless.”
NS: “BMI is useless.” They didn’t actually say this, but it did nothing for them. Pg.3
DP: “Naps are a useful tool to mitigate against sleep deprivation.”
NS: Although they did not have a validated way to measure this, they found naps to be 3 times as frequent in the summer (when nights were shorter and subjects were sleeping less). 7% vs. 22% pg.4 paragraph 2.
DP: “Bright light therapy has been shown to enhance mood, and accelerate waking after sleep, it can be used to adjust circadian rhythms.”
NS: “Maximal light exposure was at approximately 9AM . . .” pg.5 paragraph 1
DP: “Going to bed earlier, and getting up earlier is better—if total time asleep is the same.”
NS: “Decreased sleep times [were] entirely the result of later sleep onset” (e.g.- going to bed later means you’ll sleep less). This study showed that when left to awaken by natural cues, morning wake times did not vary much at all. Pg. 6
And on, and on, and on and on.
So, in my “expert” opinion—there is not much to talk about here. There does seem to be some confusion about “sleep period” vs. “sleep duration” and how the latter was calculated.
However, from a practical standpoint, I don’t really care. If you were a client of mine, and you told me that you spent 6.9-8.5 hours per night sleeping, that you didn’t feel sleepy during the day, that you need a nap 7-22% of the time, that you are eating the diet that these subjects were, that you are as active as these subjects are, that your vital signs are normal, and you have no signs of obesity or Neolithic disease. I’d say: “Perfect. How’s your stress level? Why are you here to see me?”
I think we are looking at the same ideas here, and arguing semantics. If I’m wrong here, and there is some actigraphy data base that I am unaware of—that shows industrialized people are staying in bed 7-8.5 hours and only sleeping 5.7-7.1 hours per day, then that data base would need to be compared to the dozens and dozens of CLINICAL studies that have used PSG to show that hormonal regulation, cognitive skills, reaction times, problem solving ability, athleticism, injury rates, accident rates, memory, emotionality, communication, disease, and chronic illness risk all suffer when people sleep too little. That is “generalizable” data. That research has been done with more sophisticated, more controlled and more validated accuracy, across many demographics, ethnicities, geographical locations, age groups, etc.
So, here is my conclusion, and summary:
This study showed that 10 small groups of highly active hunter-gathers—aged 20-55, that have never drank coffee or alcohol, eating an ideal diet, living at a latitude between -6 and -20 (San Diego is 32, New York is 40), that slept in clusters of 2-6 people, on an animal hide (laid on the ground), without; electricity, bug spray, protection from animals, heating or cooling, slept from about 10:30 PM to about 6:30 AM most days of the year. They tended to sleep when it was cool, but woke up as they got too cold. They tended to fall asleep about 3 hours after the sun went down and wake up just before or just after the sun came up. They napped more when nights were shorter. They weren’t obese, they didn’t have chronic disease, and they are relatively fit, and long-lived. They wore a watch-like device that showed they weren’t completely asleep for all of that time that they looked like they were sleeping, which is to say that they weren’t completely asleep for the time period that they would report as their “sleep.” This nuance is what confuses me about this study, as we know that micro-wakings are a part of “normal” sleep.
I’m not sure how generalizable that is to industrialized people? But, it actually sounds pretty plausible, and frankly, ideal.
The goal of evolutionary health is to use our knowledge and understanding of our ancestors, combined with our sophisticated understanding of new science, to adjust our life-styles in a way that leads to the best possible quality of life and performance. Nothing in this study should challenge existing beliefs about what constitutes health and performance.
After reading this paper, I would not change a single recommendation that I usually make to my clients.
So, keep your diet clean, exercise, control your stress, and regulate your sleep around the sun. If you cannot get enough sleep, or sleep at the ideal time of day, consider using a nutritional supplement like ours to see if it might help improve your sleep quality, and sleep onset.
Sleep well my friends!