The Body Fat Setpoint

One pound of human fat contains about 3,500 calories. That represents roughly 40 slices of toast. So if you were to eat one extra slice of toast every day, you would gain just under a pound of fat per month. Conversely, if you were to eat one fewer slice per day, you'd lose a pound a month. Right? Not quite.

How is it that most peoples' body fat mass stays relatively stable over long periods of time, when an imbalance of as little as 5% of calories should lead to rapid changes in weight? Is it because we do complicated calculations in our heads every day, factoring in basal metabolic rate and exercise, to make sure our energy intake precisely matches expenditure? Of course not. We're gifted with a sophisticated system of hormones and brain regions that do the "calculations" for us unconsciously*.

When it's working properly, this system precisely matches energy intake to expenditure, ensuring a stable and healthy fat mass. It does this by controlling food seeking behaviors, feelings of fullness and even energy expenditure by heat production and physical movements. If you eat a little bit more than usual at a meal, a properly functioning system will say "let's eat a little bit less next time, and perhaps also burn some of it off." This is one reason why animals in their natural habitat are nearly always at an appropriate weight, barring starvation. The only time wild animals are overweight enough to significantly compromise physical performance is when it serves an important purpose, such as preparing for hibernation.

I recently came across a classic study that illustrates these principles nicely in humans, titled "Metabolic Response to Experimental Overfeeding in Lean and Overweight Healthy Volunteers", by Dr. Erik O. Diaz and colleagues (1). They overfed lean and modestly overweight volunteers 50% more calories than they naturally consume, under controlled conditions where the investigators could be confident of food intake. Macronutrient composition was 12-42-46 % protein-fat-carbohydrate.

After 6 weeks of massive overfeeding, both lean and overweight subjects gained an average of 10 lb (4.6 kg) of fat mass and 6.6 lb (3 kg) of lean mass. Consistent with what one would expect if the body were trying to burn off excess calories and return to baseline fat mass, the metabolic rate and body heat production of the subjects increased.

Following overfeeding, subjects were allowed to eat however much they wanted for 6 weeks. Both lean and overweight volunteers promptly lost 6.2 of the 10 lb they had gained in fat mass (61% of fat gained), and 1.5 of the 6.6 lb they had gained in lean mass (23%). Here is a graph showing changes in fat mass for each individual that completed the study:

We don't know if they would have lost the remaining fat mass in the following weeks because they were only followed for 6 weeks after overfeeding, although it did appear that they were reaching a plateau slightly above their original body weight. Thus, nearly all subjects "defended" their original body fat mass irrespective of their starting point. Underfeeding studies have shown the same phenomenon: whether lean or overweight, people tend to return to their original fat mass after underfeeding is over. Again, this supports the idea that the body has a body fat mass "set point" that it attempts to defend against changes in either direction. It's one of many systems in the body that attempt to maintain homeostasis.

OK, so why do we care?

We care because this has some very important implications for human obesity. With such a system in place to keep body fat mass in a narrow range, a major departure from that range implies that the system isn't functioning correctly. In other words, obesity has to involve a defect in the system that regulates body fat, because a properly functioning system would not have allowed that degree of fat gain in the first place.

So yes, we are overweight because we eat too many calories relative to energy expended. But why are we eating too many calories? There are a number of reasons, but one reason is that the system that should be defending a low fat mass is now defending a high fat mass. Therefore, the ideal solution is not simply to restrict calories, or burn more calories through exercise, but to try to work with the system that decides what fat mass to 'defend'. Restricting calories isn't necessarily a good solution because the body will attempt to defend its setpoint, whether high or low, by increasing hunger and decreasing its metabolic rate. That's why low-calorie diets, and most diets in general, typically fail in the long term. Restricting calories works for fat loss, but most people find it miserable to fight hunger every day.

This raises two questions:

1. What caused the system to defend a high fat mass?
2. Is it possible to modify the fat mass setpoint, and how would one go about it?

Given the fact that body fat mass is much higher in many affluent nations than it has ever been in human history, the increase must be due to factors that have changed in modern times. I can only speculate what these factors may be, because research has not identified them to my knowledge, at least not in humans. But I have my guesses. I'll expand on this in the next post.


* The hormone leptin and the hypothalamus are the ringleaders, although there are many other elements involved, such as several gut-derived peptides, insulin, and a number of other brain regions.

Rabbits on a High-Saturated Fat Diet Without Added Cholesterol

I just saw another study that supports my previous post Animal Models of Atherosclerosis: LDL. The hypothesis is that in the absence of excessive added dietary cholesterol, saturated fat does not influence LDL or atherosclerosis in animal models, relative to other fats (although omega-6 polyunsaturated oils do lower LDL in some animal models). This appears to be consistent with what we see in humans.

In this study, they fed four groups of rabbits different diets:

1. Regular low-fat rabbit chow
2. Regular low-fat rabbit chow plus 0.5 g cholesterol per day
3. High-fat diet with 30% calories as coconut oil (saturated) and no added cholesterol
4. High-fat diet with 30% calories as sunflower oil (polyunsaturated) and no added cholesterol

LDL at 6 months was the same in groups 1, 3 and 4, but was increased more than 20-fold in group 2. It's not the fat, it's the fact that they're overloading herbivores with dietary cholesterol!

Total cholesterol was also the same between all groups except the cholesterol-fed group. TBARS, a measure of lipid oxidation in the blood, was elevated in the cholesterol and sunflower oil groups but not in the chow or coconut groups. Oxidation of blood lipids is one of the major factors in atherosclerosis, the vascular disease that narrows arteries and increases the risk of having a heart attack. Serum vitamin C was lower in the cholesterol-fed groups but not the others.

This supports the idea that saturated fat in the absence of excess dietary cholesterol does not necessarily increase LDL, and in fact in most animals it does not.

Merry Christmas!

Growth hormone: The fountain of youth

Growth hormone, also known as human growth hormone, seems to be implicated in a number of metabolic conditions associated with aging, and, more generally, poor health.

In adults, growth hormone deficiency is associated with: decreased calcium retention and osteoporosis, loss of muscle mass, increased fat deposition, decreased protein synthesis, and immunodeficiency. In children, growth hormone deficiency is associated with stunted growth.

Levels of growth hormones decline with age, and their decrease is believed to contribute to the aging process. Abdominal obesity is associated with low levels of growth hormone, and is also associated with the onset of the metabolic syndrome, a precursor of diabetes and cardiovascular disease.

While there are many treatments in the market that include exogenous administration of growth hormones (e.g., through injection), there are several natural ways in which growth hormone levels can be increased. These natural ways can often lead to more effective and sustainable results than prescription drugs.

For example, fasting stimulates the natural production of growth hormone. So does vigorous exercise, particularly resistance exercise with a strong anaerobic component (not cardio though). And, to the surprise of many people, deep sleep stimulates the natural production of growth hormone, perhaps more than anything else. (Although only once every 24 hours; sleeping all day does not seem to work.)

In fact, during a 24-hour period, growth hormone typically varies in pulses, or cycles. The pulses are somewhat uniformly distributed during the day, with a peak occurring at night. The graph below (source: Fleck & Kraemer, 2004) plots the typical variation of growth hormone during a 12-hour period, including the deep sleep period.

As you can see, growth hormone peaks during deep sleep; which is achieved a few hours after one goes to bed, and not too long before one wakes up.

By the way, if you want to know more about human physiology and metabolism, forget about popular diet and exercise books. Next to peer-reviewed academic articles (which are often hard to read), the best sources are college textbooks used in courses on physical education, nutrition, endocrinology, and related topics. The book from which the graph above was taken (Fleck & Kraemer, 2004), is a superb example of that.

Reference:

Fleck, S.J., & Kraemer, W.J. (2004). Designing resistance training programs. Champaign, IL: Human Kinetics.

Half-hearted Atkins diet and cardiovascular disease

I would like to comment on a recent article comparing the Atkins, Ornish and South Beach diets (Miller et al., 2009; full reference at the end of this posting), which has been causing quite a lot of commotion among bloggers recently. Especially low carb. bloggers.

An excellent post by Michael Eades clarifies a number of issues with the study, including what one could argue is the study's main flaw. Apparently the study compared a half-hearted Atkins diet, with probably equally half-hearted Ornish and South Beach diets.

I refer to the study's Atkins diet as half-hearted because it seems to rely on a daily consumption of between 120 and 180 grams of carbohydrates. This is unlikely to lead to ketosis, the cornerstone of the Atkins diet, where the body uses ketone bodies (made from dietary as well as body fat) as a source of energy.

As I see it, the main findings of the study were that the participants in the half-hearted Atkins diet, after a period of 4 weeks on the diet, and when compared with the participants in the other diets, had: (a) greater levels of total cholesterol and LDL cholesterol, with only a small improvement in their HDL cholesterol and triglycerides levels; and (b) greater levels of markers for inflammation (e.g., C-reactive protein).

The participants were young and healthy. Their average age was 30.6 years, and their average body mass index was 22.6. On average, their total cholesterol was 184.9 mg/dL, triglycerides were 78.1 mg/dL, LDL cholesterol was 107.2 mg/dL, and HDL cholesterol was 62.2 mg/dL. These are arguably fairly healthy numbers; although quite a few doctors might want to put most of these folks preventively on statins because of their LDL being greater than 100.

What I find interesting about this study, and consistent with both my own experience and also a theory that I have, is that it suggests that a low carb. diet has to really be low carb. in order to bring about the benefits that one normally sees as a result of a diet that induces ketosis. A diet with, say, > 150 g of refined grains per day, is not really a low carb. diet.

Again, in my experience, and that of many other people, a truly low carb. diet (very low in, if not devoid of, refined carbs and sugars), will lead to an impressive increase in HDL cholesterol (especially for those who have low HDL to start with), an equally impressive decrease in triglycerides, increased insulin sensitivity, and possibly a decrease in LDL.

However, a half-hearted Atkins diet may actually lead to elevated LDL (of the small-dense type), and more inflammation, just like this study suggests it does, without the benefits regarding HDL and trigs. The reason is that the still relatively high level of carbohydrate intake, especially if it comes in the form of refined carbs. and sugars, will lead to higher levels of insulin being secreted into the bloodstream. This will promote increased body fat deposition. The extra saturated fat being consumed will be turned into body fat, and not used as energy, starving the cells and leading to increased hunger.

A diet rich in saturated fat may indeed be bad when it is also a diet even moderately rich in insulin-boosting, easily digestible carbs. This may be one of the main reasons why there have been so many studies in the past showing a correlation between saturated fat consumption and heart disease; studies that typically did not control for carbohydrate consumption.

In a recent interview on the Livin' La Vida Low-Carb Blog, Dr. John Salerno goes into more detail regarding this issue, recommending a much more rigid adoption of the Atkins diet than many think is okay. (In fact, I often talk to people who think that if they cut a very high carb. intake in half - e.g., from 400 to 200 grams per day - replacing the carbs with fat, they will be halfway into a full blown Atkins diet.) Dr. Salerno has worked in the past with Dr. Atkins. He calls his diet the Silver Cloud Diet. I am not sure I agree with all that Dr. Salerno had to say, but his argument in favor of a diet very low in carbs. does make sense to me.

Finally, I think that it is dangerous to extrapolate the results of any study, no matter how comprehensive, to the population in general. Each individual is unique in terms of his or her genetic makeup and life history; the latter also influences metabolic patterns. (Even identical twins raised together may display different metabolic patterns, because of their different life histories.)  So, while a low carb. diet may work well for a lot of people, it may have very negative effects on a few. Increases in inflammation markers and adverse effects on LDL cholesterol (especially when LDL is measured directly, accounting for particle numbers and sizes) are warning signs that any low carb. dieter should pay attention to.

Reference:

Miller, M. et al. (2009). Comparative effects of three popular diets on lipids, endothelial function, and c-reactive protein during weight maintenance. Journal of the American Dietetic Association, 109, 713-717.

What's the Ideal Fasting Insulin Level?

[2013 update.  I'm leaving this post up for informational purposes, but I think it's difficult to determine the "ideal" insulin level because it depends on a variety of factors including diet composition.  Also, insulin assays are not always comparable to one another, particularly the older assays, so it's difficult to compare between studies]

Insulin is an important hormone. Its canonical function is to signal cells to absorb glucose from the bloodstream, but it has many other effects. Chronically elevated insulin is a marker of metabolic dysfunction, and typically accompanies high fat mass, poor glucose tolerance (prediabetes) and blood lipid abnormalities. Measuring insulin first thing in the morning, before eating a meal, reflects fasting insulin. High fasting insulin is a marker of metabolic problems and may contribute to some of them as well.

Elevated fasting insulin is a hallmark of the metabolic syndrome, the quintessential modern metabolic disorder that affects 24% of Americans (NHANES III). The average insulin level in the U.S., according to the NHANES III survey, is 8.8 uIU/mL for men and 8.4 for women (2). Given the degree of metabolic dysfunction in this country, I think it's safe to say that the ideal level of fasting insulin is probably below 8.4 uIU/mL.

Let's dig deeper. What we really need is a healthy, non-industrial "negative control" group. Fortunately, Dr. Staffan Lindeberg and his team made detailed measurements of fasting insulin while they were visiting the isolated Melanesian island of Kitava (3). He compared his measurements to age-matched Swedish volunteers. In male and female Swedes, the average fasting insulin ranges from 4-11 uIU/mL, and increases with age. From age 60-74, the average insulin level is 7.3 uIU/mL.

In contrast, the range on Kitava is 3-6 uIU/mL, which does not increase with age. In the 60-74 age group, in both men and women, the average fasting insulin on Kitava is 3.5 uIU/mL. That's less than half the average level in Sweden and the U.S. Keep in mind that the Kitavans are lean and have an undetectable rate of heart attack and stroke.

Another example from the literature are the Shuar hunter-gatherers of the Amazon rainforest. Women in this group have an average fasting insulin concentration of 5.1 uIU/mL (4; no data was given for men).

I found a couple of studies from the early 1970s as well, indicating that African pygmies and San bushmen have rather high fasting insulin. Glucose tolerance was excellent in the pygmies and poor in the bushmen (5, 6, free full text). This may reflect differences in carbohydrate intake. San bushmen consume very little carbohydrate during certain seasons, and thus would likely have glucose intolerance during that period. There are three facts that make me doubt the insulin measurements in these older studies:

1. It's hard to be sure that they didn't eat anything prior to the blood draw.
2. From what I understand, insulin assays were variable and not standardized back then.
3. In the San study, their fasting insulin was 1/3 lower than the Caucasian control group (10 vs. 15 uIU/mL). I doubt these active Caucasian researchers really had an average fasting insulin level of 15 uIU/mL. Both sets of measurements are probably too high.

Now you know the conflicting evidence, so you're free to be skeptical if you'd like.

We also have data from a controlled trial in healthy urban people eating a "paleolithic"-type diet. On a paleolithic diet designed to maintain body weight (calorie intake had to be increased substantially to prevent fat loss during the diet), fasting insulin dropped from an average of 7.2 to 2.9 uIU/mL in just 10 days. This is despite a substantial intake of carbohydrate, including fruit and vegetable sugars.  The variation in insulin level between individuals decreased 9-fold, and by the end, all participants were close to the average value of 2.9 uIU/mL. This shows that high fasting insulin is correctable in people who haven't yet been permanently damaged by the industrial diet and lifestyle. The study included men and women of European, African and Asian descent (7).

One final data point. My own fasting insulin, earlier this year, was 2.3 uIU/mL. I believe it reflects a good diet, regular exercise, sufficient sleep, and a relatively healthy diet growing up. It does not reflect: carbohydrate restriction, fat restriction, or saturated fat restriction.

So what's the ideal fasting insulin level? My current feeling is that we can consider anything between 2 and 6 uIU/mL within our evolutionary template.

Total cholesterol and cardiovascular disease: A U-curve relationship

The hypothesis that blood cholesterol levels are positively correlated with heart disease (the lipid hypothesis) dates back to Rudolph Virchow in the mid-1800s.

One famous study that supported this hypothesis was Ancel Keys's Seven Countries Study, conducted between the 1950s and 1970s. This study eventually served as the foundation on which much of the advice that we receive today from doctors is based, even though several other studies have been published since that provide little support for the lipid hypothesis.

The graph below (source: canibaisereis.com, with many thanks to O Primitivo) shows the results of one study, involving many more countries than Key's Seven Countries Study, that actually suggests a NEGATIVE linear correlation between total cholesterol and cardiovascular disease.

Now, most relationships in nature are nonlinear, with quite a few following a pattern that looks like a U-curve (plain or inverted); sometimes called a J-curve pattern. The graph below (source also: canibaisereis.com) shows the U-curve relationship between total cholesterol and mortality, with cardiovascular disease mortality indicated through a dotted red line at the bottom.

This graph has been obtained through a nonlinear analysis, and I think it provides a better picture of the relationship between total cholesterol (TC) and mortality. Based on this graph, the best range of TC that one can be at is somewhere between 210, where cardiovascular disease mortality is minimized; and 220, where total mortality is minimized.

The total mortality curve is the one indicated through the full blue line at the top. In fact, it suggests that mortality increases sharply as TC decreases below 200.

Now, these graphs relate TC with disease and mortality, and say nothing about LDL cholesterol (LDL). In my own experience, and that of many people I know, a TC of about 200 will typically be associated with a slightly elevated LDL (e.g., 110 to 150), even if one has a high HDL cholesterol (i.e., greater than 60).

Yet, most people who have a LDL greater than 100 will be told by their doctors, usually with the best of the intentions, to take statins, so that they can "keep their LDL under control". (LDL levels are usually calculated, not measured directly, which itself creates a whole new set of problems.)

Alas, reducing LDL to 100 or less will typically reduce TC below 200. If we go by the graphs above, especially the one showing the U-curves, these folks' risk for cardiovascular disease and mortality will go up - exactly the opposite effect that they and their doctors expected. And that will cost them financially as well, as statin drugs are expensive, in part to pay for all those TV ads.

The Dirty Little Secret of the Diet-Heart Hypothesis

The diet-heart hypothesis is the idea that saturated fat, and in some versions cholesterol, raises blood cholesterol and contributes to the risk of having a heart attack. To test this hypothesis, scientists have been studying the relationship between saturated fat consumption and heart attack risk for more than half a century. What have these studies found?

The large majority of observational studies have found no connection between habitualsaturated fat consumption and heart attack risk. The scientific literature contains dozens of these studies, so let's narrow the field to prospective studies only, because they are considered the most reliable. In this study design, investigators find a group of initially healthy people, record information about them (in this case what they eat), and watch who gets sick over the years.

A Sampling of Unsupportive Studies

Here are references to ten high-impact prospective studies, spanning half a century, showing no association between saturated fat consumption and heart attack risk. Ignore the saturated-to-polyunsaturated ratios, Keys/Hegsted scores, etc. What we're concerned with is the straightforward question: do people who eat more saturated fat have more heart attacks? Many of these papers allow free access to the full text, so have a look for yourselves if you want:

A Longitudinal Study of Coronary Heart Disease. Circulation. 1963.

Diet and Heart: a Postscript. British Medical Journal. 1977. Saturated fat was unrelated to heart attack risk, but fiber was protective.

Dietary Intake and the Risk of Coronary Heart Disease in Japanese Men Living in Hawaii. American Journal of Clinical Nutrition. 1978.

Relationship of Dietary Intake to Subsequent Coronary Heart Disease Incidence: the Puerto Rico Heart Health Program. American Journal of Clinical Nutrition. 1980.

Diet, Serum Cholesterol, and Death From Coronary Heart Disease: The Western Electric Study. New England Journal of Medicine. 1981.

Diet and 20-year Mortality in Two Rural Population Groups of Middle-Aged Men in Italy. American Journal of Clinical Nutrition. 1989. Men who died of CHD ate significantly less saturated fat than men who didn't.

Diet and Incident Ischaemic Heart Disease: the Caerphilly Study. British Journal of Nutrition. 1993. They measured animal fat intake rather than saturated fat in this study.

Dietary Fat and Risk of Coronary Heart Disease in Men: Cohort Follow-up Study in the United States. British Medical Journal. 1996. This is the massive Physicians Health Study. Scroll down to table 2 and see for yourself that the association between saturated fat intake and heart attack risk disappears after adjustment for several factors including family history of heart attack, smoking and fiber intake. That's because, as in most modern studies, people who eat steak are also more likely to smoke, avoid vegetables, eat fast food, etc.

Dietary Fat Intake and the Risk of Coronary Heart Disease in Women. New England Journal of Medicine. 1997. From the massive Nurse's Health study. The abstract claims that saturated fat was associated with heart attack risk. However, the association disappeared when they adjusted for monounsaturated and polyunsaturated fat intake. Have a look at table 3.

Dietary Fat Intake and Early Mortality Patterns-- Data from the Malmo Diet and Cancer Study. Journal of Internal Medicine. 2005.

I just listed 10 prospective studies published in top peer-reviewed journals that found no association between saturated fat and heart disease risk. This is less than half of the prospective studies that have come to the same conclusion, representing by far the majority of studies to date. If saturated fat is a dominant cause of cardiovascular disease, why are its effects essentially undetectable in the best studies we can muster?

Studies that Support the Diet-Heart Hypothesis

To be complete, some studies have found an association between saturated fat consumption and heart attack risk. Here's a list of all four that I'm aware of, with comments:

Ten-year Incidence of Coronary Heart Disease in the Honolulu Heart Program: relationship to nutrient intake. American Journal of Epidemiology. 1984. "Men who developed coronary heart disease also had a higher mean intake of percentage of calories from protein, fat, saturated fatty acids, and polyunsaturated fatty acids than men who remained free of coronary heart disease." The difference in saturated fat intake between people who had heart attacks and those who didn't, although statistically significant, was very small.

Diet and 20-Year Mortality From Coronary Heart Disease: the Ireland-Boston Diet-Heart Study. New England Journal of Medicine. 1985. "Overall, these results tend to support the hypothesis that diet is related, albeit weakly, to the development of coronary heart disease."

Relationship Between Dietary Intake and Coronary Heart Disease Mortality: Lipid Research Clinics Prevalence Follow-up Study. Journal of Clinical Epidemiology. 1996. "...increasing percentages of energy intake as total fat (RR 1.04, 95% CI = 1.01 – 1.08), saturated fat (RR 1.11, CI = 1.04 – 1.18), and monounsaturated fat (RR 1.08, CI = 1.01 – 1.16) were significant risk factors for CHD mortality among 30 to 59 year olds... None of the dietary components were significantly associated with CHD mortality among those aged 60–79 years." Note that the associations were very small, also included monounsaturated fat (like in olive oil), and only applied to the age group with the lower risk of heart attack.

The Combination of High Fruit and Vegetable and Low Saturated Fat Intakes is More Protective Against Mortality in Aging Men than is Either Alone. Journal of Nutrition. 2005. Higher saturated fat intake was associated with a higher risk of heart attack; fiber was strongly protective.

The Review Papers

Over 25 high-quality studies conducted, and only 4 support the diet-heart hypothesis. In case you're concerned that I'm cherry-picking studies, here are links to review papers on the same data that have reached the same conclusion:

Meta-analysis of prospective cohort studies evaluating the association of saturated fat with cardiovascular disease.  American Journal of Clinical Nutrition. 2010.  "A meta-analysis of prospective epidemiologic studies showed that there is no significant evidence for concluding that dietary saturated fat is associated with an increased risk of CHD or CVD."

A Systematic Review of the Evidence Supporting a Causal Link Between Dietary Factors and Coronary Heart Disease. Archives of Internal Medicine. 2009. "Insufficient evidence (less than or equal to 2 criteria) of association is present for intake of supplementary vitamin E and ascorbic acid (vitamin C); saturated and polyunsaturated fatty acids; total fat; alpha-linolenic acid; meat; eggs; and milk" They analyzed prospective studies representing over 160,000 patients from 11 studies meeting their rigorous inclusion criteria, and found no association between saturated fat consumption and heart attack risk.

The Questionable Role of Saturated and Polyunsaturated Fatty Acids in Cardiovascular Disease. Journal of Clinical Epidemiology. 1998. Dr. Uffe Ravnskov challenges the diet-heart hypothesis simply by collecting all the relevant studies and summarizing their findings.

Where's the Disconnect?

The first part of the diet-heart hypothesis states that dietary saturated fat raises the cholesterol/LDL concentration of the blood. The second part states that increased blood cholesterol/LDL increases the risk of having a heart attack. What part of this is incorrect?

There's definitely an association between blood cholesterol/LDL level and heart attack risk in certain populations, including Americans. MRFIT, among other studies, showed this definitively, although the lowest risk of all-cause mortality was at an average level of cholesterol.

So we're left with the first premise: that saturated fat increases blood cholesterol/LDL. Could this hypothesis be less well supported than it appears?  The data that are used to support it come almost exclusively from short-term feeding studies (<3 and="" association="" between="" blood="" consumption="" effect="" fat="" found="" habitual="" have="" here="" how="" information="" is="" lipids.="" little="" long="" months="" most="" nbsp="" observational="" on="" p="" persists="" saturated="" studies="" surprisingly="" this="">