A recently published Japanese study has reached a conclusion that practically screams for a popular science headline about longevity:
The finding in a nutshell: People who eat more fish—or, more specifically, EPA and DHA—especially for breakfast, are more likely to be early chronotypes. That could easily be turned into the following message:
Fish for breakfast makes you an early riser.
But here’s the thing: that’s exactly what the study doesn’t show.
And upon closer inspection, the study is still interesting—though for a slightly different reason. And before experts start sharing fish recipes on social media, here’s my analysis of the study.
What exactly was studied?
The researchers analyzed data from 5,975 adults in Japan. The dietary data came from the Japanese nutrition app Asken. In particular, the study examined how much seafood, as well as EPA and DHA, the participants consumed at various times of the day.
To determine the chronotype , the Munich ChronoType Questionnaire (MCTQ) was used as a starting point , albeit in a modified form. We’ll come back to that later. The key factor in the MCTQ is the so-called MSFsc—Mid-Sleep on Free Days, sleep-corrected. Simply put, this involves first determining the mid-sleep time on days off. The MCTQ assumes a biological sleep need of 8 hours. For example, someone who falls asleep at 11 p.m. and wakes up at 7 a.m. has a mid-sleep time of 3 a.m.
The MCTQ then takes into account any “catch-up sleep” on days off and adjusts the sleep midpoint accordingly. The MSFsc thus serves as a continuous measure of chronotype. The MCTQ is therefore not based on a subjective question such as “Are you more of a morning person or an evening person?”, but rather on actual—albeit self-reported— sleep times over the past 6 weeks.
And this is exactly where the first important caveat comes in.
No objectively measured chronotype
The chronotype classification in this study is not based on data from wearable devices, which is strange in principle, because this would be more objective than a subjectively completed questionnaire. At best, the two would have been compared to improve validity. The sleep durations used to calculate the MSFsc were reported by the participants in 30-minute increments, which can also be problematic because this results in nearly 60 minutes of inaccuracy when comparing the data.
However, it is still important to make a clear distinction between the terms: MSFsc is a well-established chronotype proxy, not a direct measurement of the internal circadian clock. Even less so should MSFsc be equated with a direct measurement of the circadian phase, such as DLMO. While studies show correlations between MSFsc and DLMO, they do not show a perfect match.
Of particular interest: The study does not use the traditional classification:
- Early riser
- Standard Type
- Late-type
Instead, the sample was divided into four equal (!) quartiles based on the MSFsc:
Early – Moderately early – Moderately late – Late.
So there was essentially no “typical case” in the sense of a Gaussian distribution, as is the case with the MCTQ itself. The first quarter of the sample was classified as the “Early” type, and the last quarter as the “Late” type.
The median MSFsc values were:
- Early: 1:47 a.m. (Means bedtime at 9:47 p.m.)
- Moderately early: 2:04 a.m. (Means bedtime at 10:04 p.m.)
- Moderately late: 3:36 a.m. (Means bedtime at 11:36 p.m.)
- Late: 4:51 a.m. (Means bedtime was 12:51 a.m.)
That is a crucial point for interpretation.
This is because “Early” does not mean here, “This person is an early chronotype according to a generally accepted definition.” It means, first and foremost, that this person is among the 25 percent with the earliest sleep midpoint in this Japanese sample.
The classification is therefore relative to the population under study. By way of comparison: The latest chronotype in the population of the Japanese study still corresponds to a normal chronotype in Germany. Another interesting finding: In our projects, the range between the earliest and latest chronotypes (measured using the RNA test) spans 13.5 hours. In this study, involving nearly 6,000 participants, the range between the respective earliest and latest median values was just 3 hours.
And now for the ever-present chicken-and-egg problem
The study shows a link between fish/EPA/DHA and an earlier onset of MSFsc. But what is the cause and what is the effect? Let’s consider two possible scenarios.
Scenario 1: Fish Affects Chronotype
EPA and DHA could influence circadian processes through various biological mechanisms. In that case, the following sequence would be conceivable: EPA/DHA → circadian regulation → an earlier sleep-wake cycle. That would be extremely interesting.
Scenario 2: Chronotype Influences Eating Behavior
However, the following is just as plausible: an early chronotype → waking up earlier → an early breakfast → different food choices → more fish/EPA/DHA. In that case, fish would not be the cause of the early chronotype. Rather, it would be part of a lifestyle pattern associated with an early chronotype.
And there’s a third scenario:
Scenario 3: Another factor influences both
For example: Lifestyle → early chronotype + earlier meals + different diet. In this case, the observed association would be real, but the fish itself would not necessarily be the cause. A cross-sectional study cannot distinguish between these three possibilities. The authors themselves therefore point out that no causality can be inferred from their data.
Particularly problematic: “Breakfast” is not a biological time reference
From a chronobiological perspective, it is also interesting that the diet was categorized by meals.
Breakfast, lunch, dinner, and snacks. This may be practical for a nutritional study, but it is by no means valid for a chronobiological assessment. After all, a breakfast at 5:30 a.m. is biologically different from a breakfast at 10:30 a.m. If we want to know whether food acts as a timing cue, what we’re really interested in is: When exactly, relative to the individual’s circadian phase, was the meal eaten? The category “breakfast” does not answer this question. As a result, an important dimension of chrononutrition is left unaccounted for.
Japan is not just a minor detail here
And here’s another factor that’s often underestimated when interpreting such studies: country and culture. The study was conducted in Japan. This is not simply a random sample of “people.” Japan has a distinct food culture. Fish and seafood are traditionally much more deeply rooted in the diet than they are in Germany, for example. At the same time, breakfast customs, meal patterns, work schedules, social rhythms, and lifestyle habits differ between Japan and Central Europe.
If a link is found between early chronotype and fish consumption, we must therefore at least ask: Are we measuring a biological effect of EPA/DHA here—or, in part, a culturally influenced daily pattern that also varies completely between rural coastal areas, where fish makes up a larger portion of the diet, and large cities, which have different eating habits and, moreover, entirely different levels of sunlight exposure?
That does not mean the findings are therefore worthless—on the contrary. They simply do not provide a valid basis for assessing the pattern of sleep-wake cycles.
The Time Zone Effect
Last but not least, Japan exhibits an interesting peculiarity, for example, when compared to Germany. Japan spans three time zones, but there is only one official time in the country: JST (UTC+9). This means that at the same time in the east, when the sun rises, it takes about 128 minutes for it to rise in the west as well. This effect is also not accounted for in the study. With a total span of 3 hours between the earliest and latest groups, these 128 minutes are no small matter when assessing a “natural” sleep midpoint.
As previously noted, the sample in this Japanese study also shows a strikingly early MSFsc distribution. Whether this is a characteristic specific to the Japanese population, the region studied, the season, the sample selection, the time zone/time-of-day factor, or the app user base cannot be determined from the study.
It would be interesting if an East-West trend were also evident here, as that would significantly enhance the study’s significance.
A Look at Germany
The MCTQ was originally developed in Germany and therefore has an exceptionally large database for comparison. In a German MCTQ population of more than 51,000 people, the average MSFsc was approximately 4:27 a.m. The chronotype distribution is not simply divided into three clearly distinct groups. Rather, it forms a broad Gaussian distribution across the entire day, with a few extremely early and extremely late types and a large number in between. The large MCTQ database shows a distribution that is approximately normally distributed, with a slight shift toward later types.
This is relevant to the Japanese study, since its four groups correspond to quartiles of the Japanese sample. So one cannot simply say, “The Japanese ‘Early’ group corresponds to the German ‘early chronotype.’” To do so, one would have to directly compare the actual MSFsc distributions of both populations and, as far as possible, take into account age, gender, working hours, geographic location, and other influencing factors.
So what’s actually left?
In my opinion, the scientifically soundest statement is far less sensational than the eye-catching headline:
In this Japanese sample, higher consumption of seafood—specifically EPA/DHA—particularly at early meals, was associated with an earlier bedtime (as determined by MCTQ) and, in some cases, with less social jet lag. However, “early” was not defined in terms of specific time and could therefore also be defined relatively in comparison to the general population. However, there is no clarity on this point.
While this is an interesting finding, it does not mean that “fish turns people into early risers, ” and certainly not that“if you want to get up earlier, you should eat fish for breakfast.”
To do that, we would need an intervention study. For example, we would have to randomly assign people with similar chronotypes to different dietary conditions and then monitor them over an extended period to see whether their circadian phase actually changes. Ideally, we would not only record sleep times but also collect objective sleep data and a biological marker of the circadian phase. Only then could we answer the crucial question:
Do EPA and DHA alter chronotype, or do people with different chronotypes simply eat differently?
My Conclusion
The headline “Fish = Makes Your Chronotype Earlier” would be too simplistic from a scientific standpoint . The study shows something different—and, in my opinion, more interesting: Chronotype and diet appear to be temporally linked, though this is nothing new in principle. Whether the chronotype influences eating behavior, eating behavior influences circadian rhythms, or both are shaped by shared environmental and lifestyle factors remains an open question—which, again, is nothing new in principle.
Given the study design and the lack of information, there is only extremely limited evidence of a possible causal relationship, especially as it pertains to the average Central European, or even Germans.
And that is precisely why I would not interpret the study as evidence of a “Fisch effect” on the body’s internal clock, but merely as further indication that chronotype, sleep, diet, and the timing of our meals cannot be considered in isolation from one another.
So… Good news for all vegans: According to the study, fish isn’t a lifeline for early chronotypes.
Source:
https://www.jstage.jst.go.jp/article/jnsv/72/4/72_272/_pdf/-char/en
