🥚Eggs and CVD🥚
Could there BE a more classic single food nutritional epi example? I submit that there is not!
When another egg paper recently popped up in
@CircAHA
I jumped at the chance to walk through some of my favorite fundamental issues with single food analyses.
“Obesity is genetic”
If you were just triggered, then this thread is for you.
If “but the human genome can’t evolve in a generation!” has passed your lips in the past few months, then this thread is for you.
🧬This is why obesity is genetic. And the environment is to blame.
Meta-analysis of meta-analyses of RCTs:
dietary interventions to modify macronutrient composition still irrelevant for weight loss in patients with type 2 diabetes.
No more low-carb vs. low-fat weight loss trials. It’s over. No difference. Move on.
Randomized my 1st subject for my 1st trial today.
I’ve been an epidemiologist for 10yrs and never met a participant behind the data. Totally perception-changing. TY to anyone who’s ever participated in a study in any way. Public health research cannot exist w/o you.
#epitwitter
Santa Claus will be coming to town this year, Dr. Anthony Fauci says.
“I took care of that for you,” he says. “…I took a trip up there to the North Pole; I went there and I vaccinated Santa Claus myself. I measured his level of immunity, and he is good to go.”
#CNNSesameStreet
Damn. The
@CDCgov
@BRFSS
just released its annual state-level obesity prevalences.
US obesity worsened again in 2022, with dramatic increases in many states in just 1 year.
The US is completely off the rails.
All the data, maps, subgroups here:
Iodine is from the sea. Land animals are not naturally iodine rich. So why is dairy a good source of iodine? Because standard dairy cow feed is fortified with iodized salt. The cow takes the supplement for you.
Read more about iodine here:
Vegan diets don't supply sufficient iodine. Another sign that this is not your natural human diet. If it were meant for humans, it would supply all the nutrients you need, naturally.
As the obesity prevalence climbed in the US and globally, epidemiologists kept an eye on Japan.
Despite an initial upward trend like the rest of the world, obesity in Japan quickly leveled off and came back down to <5%.
Here's why...
📉Cholesterol-lowering therapies like statins are a cornerstone of preventing cardiovascular disease since the 1990s.
So with that, please join me in an epidemiologic adventure as we address an emerging alternative viewpoint that refutes the role of LDL (bad cholesterol) in CVD.
📉Cholesterol-lowering therapies like statins are a cornerstone of preventing cardiovascular disease since the 1990s.
So with that, please join me in an epidemiologic adventure as we address an emerging alternative viewpoint that refutes the role of LDL (bad cholesterol) in CVD.
Just published in
@NEJM
and presented
@AHAScience
Aggressive lowering of triglycerides in pts with T2D had no reduction in CVD. LDL remains king for causal lipid hypotheses.
Lots of comments about whether BMI, self-reported or otherwise, is useful...
I give a lecture on this, and can share a few slides.
Monitoring population level trends is *exactly* what BMI does best.
Here's why...
Damn. The
@CDCgov
@BRFSS
just released its annual state-level obesity prevalences.
US obesity worsened again in 2022, with dramatic increases in many states in just 1 year.
The US is completely off the rails.
All the data, maps, subgroups here:
🎉We’re excited to announce the members of the 2025 Dietary Guidelines Advisory Committee! The Committee looks at the relationship between diet and health — and informs nutrition guidance everyone can use. Learn more:
#DietaryGuidelines
The
@usda
MyPlate serves as the primary educational tool for the Dietary Guidelines for Americans.
📊New CDC report: how many Americans have even heard of MyPlate? And how many tried to follow its guidance?
💣Are you ready for this?
🍽️1
Ahhhh! It's official! Ahhhh! I just received my first ever R01! I'm actually going to be the PI of my dream substitution weight loss trial! This feels foreign...what do I do now? How do I not have this creepy grin for the rest of my zoom calls today??
🥹 Floored, honored, SO grateful!
Colleagues and I
@MassGenBrigham
@HarvardChanSPH
were just awarded a U01 by
@theNCI
to investigate mechanisms of obesity & cancer.
The best part of this program is the collaborations it builds across the funded centers. I cannot wait to begin!
📜This comprehensive thread by lead author
@KevinH_PhD
on our new paper of what the modern Energy Balance Model (EBM) of obesity *is* and *is not* is a *must read*.
However - I want to emphasize a few points from the epi/population health P.O.V. that underscore the biology:
/1
Walmart knows ozempic Rx fillers purchase less food…but how? How isn’t there a firewall to pharmacy data? Or was this an IRB approved research endeavor? For internal sales meetings, let alone public announcements, it is unsettling that patient data can be mined like this…wut?
👇This supp. fig waterfall plot from DIETFITS. It represents so much untapped potential to understanding long-term weight loss. Can we back-burner carbs/fat for a sec and focus on drivers of adherence? Also, if u have past trial data, can you plz send me a waterfall plot, I❤️them
This is huge.
You know how epi is ugh and short-term interventions are meh?
May I present to you, the *UNICORN OF NUTRITION SCIENCE*
🌠RCT of cocoa-flavs and multi-vits
🌠Placebo-controlled
🌠Blinded
🌠21,000 adults
🌠3.6 yrs follow-up
🌠INCIDENT CVD!
🌠ALL CAUSE MORTALITY!!!
In
@AJCNutrition
:
@HowardSesso
, Dr. JoAnn Manson of
@BWHDPM
@BrighamWomens
& colleagues report the main findings of the first ever randomized trial of a cocoa flavanol supplement on cardiovascular disease endpoints
and (1 of 5)
Congrats to Dr. Willett on 40 years (!) teaching the Nutritional Epidemiology course at HSPH. Honored to serve as co-instructor, I still learn something new every year.
🧵 When it comes to diet/disease studies how well do the bodies of evidence: -RCTs- and -observational epi- agree?
@lschwinshakl
et al tackled this oft-raised question with an impressively expansive systematic review
@BMJ
.
Here's a thread about *all of that*:
Country-level correlation studies like the "lettuce will kill you" preprint provide epidemiologists with a good laugh-cry, & I want to be sure everyone understands why.
So why are diet/disease correlations (AKA ecologic studies) less than useful for deriving causality? 1/
We're hosting a symposium Mar 6 (in-person
@BrighamWomens
& virtual) to bring you a line-up of brilliant nutrition colleagues & epidemiologists to discuss methods and works in progress addressing some of diet research's biggest challenges.
To register:
Do you have a cohort with *baseline diet* and 20-30 yrs of observational follow-up?
Are you pondering an analysis of hi vs. lo (insert food/nutrient) intake to see whether its related to (insert dx) decades later?
Don't do it!
Follow me for more quality nutritional epi tips!
Can you imagine having your research study followed by a Netflix documentary crew?! 😱🤯
Dr. Gardner is unparalleled at making nutrition science look so much fun -- I highly recommend bingeing this for the pop-sci (and then check out the real manuscript for all the gritty deets)
Omnivore vs Vegan Diet Study Among Identical Twins
Study was fun to design & run - thrilled it was published in
@JAMANetworkOpen
👇
Certainly a new & unique experience to have the conduct & findings featured on Netflix!
#3
since released on 1/1/24 🎥🤣
🇺🇸It is an absolute honor to serve on the US
#2025DGAC
. I am blown away by the federal coordination, dedicated staff, and 19 brilliant colleagues.
Learn more about our daunting but critical road ahead and follow along:
If the main source of monounsaturated fats in the US diet is beef, and commercial cattle feed is processed grains soy corn and sunflower, does this mean steak is our main source of seed oils?
Our latest publication📰is out with
@rik_epi
and co-authors:
We derived a novel metabolomics score so strongly related to future risk of T2D, that honestly it is an embarrassment of odds ratio riches...
The DPP: $Huge NIH-funded trial (N=3200) in the 1990s. Conclusion: randomized diet/lifestyle significantly reduced T2D more than metformin or placebo. Big 58% reduction.
The diet? Reduce total fat as much as possible. Be hypocaloric and lose weight. Exercise.
Everyone! Gather around! I am SO excited for this
@AJCNutrition
paper that just came out.
But first, Nutrition Science, I want to introduce you to my good friend, Causal Inference. CI is this niche of epidemiologic methods (paradoxical, I know).
/1
This RCT presents a fantastic deep dive into modulators of weight loss
Factors w/ *NO* DIFFERENTIAL EFFECTS on 12wk WL:
- High carb vs. high fat (yet again)
- Predicted "carb" or "fat" responder genotype (a la DIETFITS)
- Insulin & IR
- Insulin on high-carb
- Insulin on high-fat
Interested in personalized nutrition? Our new pub examining whether weight loss differs between genotype-concordant and genotype-discordant diets was published in
@NatureComms
and is featured as an Editors’ Highlight.
A thread on findings below.
Link:
The hypocrisy of the mud slinging from the low-carb community every time a meat/vegan study makes headlines is truly the most bizarre form of non-scientific discord.
Literally *everything* they chuck in attempt to discredit research(ers), applies to their own low-carb evidence.
If your research was important enough to be done in the first place then it literally doesn't matter whether your results are null or significant.
Report results with honesty & integrity. Don't marry your hypothesis. An important well-powered null is a beautiful thing.
@deirdre_tobias
@marionnestle
You are my new favorite scientist.
I disagree about a lot of this, and I'll give it a closer look tomorrow (I just got in and saw it all), but I cannot tell you how much I appreciate your jumping in here and talking about all this in public.
Thank you.
One of the first things taught in introductory *epidemiology* textbooks is that correlation *could be* causation...
...and that it's your job to know how likely that is for a given set of experiments & observations.
@CharlesMBrenner
@deirdre_tobias
“One of the first things taught in introductory statistics textbooks is that correlation is not causation. It is also one of the first things forgotten.”
― Professor Thomas Sowell
I thoroughly enjoyed this episode of
@PlantChompers
sitting down with first author of the HSPH red meat/diabetes paper, Xiao Gu. The nutritional epi methods discussion does not disappoint. Their dismantling of sloppy blogger and YouTube hot takes is 🏆A+.
Precision Nutrition will be a fascinating scientific journey. No doubt we’ll learn a TON about internal mechanisms.
And, if $150 mil later, it leads us to finally acknowledge that the *external* food environment is actually the bigger issue, it will have been worth every cent.
NIH is investing $150M in a study to figure out the "best diet" for people.
Meanwhile, the food industry will continue to churn out cheap ultraprocessed junk food that we'll all eat instead.
We're hosting a symposium Mar 6 (in-person
@BrighamWomens
& virtual) to bring you a line-up of brilliant nutrition colleagues & epidemiologists to discuss methods and works in progress addressing some of diet research's biggest challenges.
To register:
I am hiring a postdoc for nutritional epidemiology! Come research with me in Boston
@bwh_postdocs
!
*please share with your friends and colleagues*
Details and to Apply Online: .
This whole article from last month lit a fire in me, I'm still enraged.
This hot mic quote from the
@KelloggsUS
CEO, he's so cavalier about the US obesity epidemic -- food industry monitors and will "mitigate" successful weight loss strategies because they threaten sales. Gross.
Walmart says it’s already seeing an impact on shopping demand from people taking the diabetes drug Ozempic, Wegovy and other appetite-suppressing medications.
Anyone claiming obesity is genetic 🧬isn’t wrong. It explains between-person differences in BMI and why weight gain propensity is similar to our parents.
Anyone claiming obesity is environment🍩 isn’t wrong. It explains where our society has ended up in a short few decades.
“Nutritional epi is garbage, self-reported diet is useless”
is my absolute favorite hot take❤️
Thanks
@NutritionDanny
for providing a safe space to discuss evidence and this poorly informed sentiment.
Take a listen! But⚠️ it gets deep, you can't be any geek off the street
New episode with
@deirdre_tobias
, discussing the criticisms of, and misconceptions about, the food frequency questionnaire (FFQ).
Find on all podcast apps or via the link:
New perspective piece
@AmJEpi
!
We lay out what makes dietary interventions/exposures inherently complex, and why all researchers (yes, YOU) *REALLY* need to stop ignoring this reality…
“Improving nutrition science begins with asking better questions”
Revealed! My personal daily Protocol:
1️⃣Tumble out of bed and stumble to the kitchen
2️⃣Pour myself a cup of ambition
3️⃣Yawn and stretch and try to come to life
4️⃣Jump in the shower & the blood starts pumpin
5️⃣Out on the street, the traffic starts jumpin
6️⃣On the job from 9 to 5
I’m excited to share some big news.
I've been working on a book for quite some time — more than five years, in fact — though it's really based on more than three decades of research and experience. Today, I am thrilled to share that my new and first-ever book, Protocols: An
This week's coverage of the latest alcohol/mortality paper was disappointing. Most missed the point of this issue entirely.
In cased you missed it, the hot takes were all like:
"THINK THAT NIGHTLY PINOT IS YOUR FOUNTAIN OF YOUTH? NEW SCIENCE SAYS THINK AGAIN, SUCKERS☠️"
@DrNadolsky
@SBakerMD
I thought the take-away was that the person's keto diet was so horrendously bad for their LDLc that even cookies were healthier in contrast. 🤷♀️
15/
I think this is the main take-away.
It's striking how consistent the relationship between LDL and all-cause mortality is. Women with higher baseline LDL were more likely to die during follow-up. Regardless of their HDL and TG (or age, BMI, smoking, etc.).
I ❤️THIS:
“Unlike politics, when it comes to science there is no ‘two sides,’ “Scientific evidence is a process of critically evaluating data in a systematic and rigorous way and interpreting that data,” she said. Unlike politics, it’s based on facts, not feelings.
-
@DBelardoMD
Spoke to the
@LosAngelesBlade
this week about how
@DrOz
&
@drmarkhyman
are both a symptom of broader & more systemic problems in our healthcare system.
We need to hold physicians accountable who are profiting off of pseudoscience & misinformation
Sweat equity and persistence paid off 🥹 long-time passion project is finally funded! Thank you
@AmDiabetesAssn
!
So excited to work with Drs.
@LajousMartin
& Dalia Stern on cutting-edge causal inference methods for nutritional epidemiology and T2D prevention.
More to come...
This podcast on nutritional epidemiology with Danny and Alan was so much fun to record. If I sound way too excited geeking out on methods, it's because I was. We didn't talk about keto once and I even got to drop "triangulation of methods" ❤ My happy place.
It was an honor to talk to
@deirdre_tobias
about all things nutrition epidemiology.
If you like nerding out on nutrition science, you'll enjoy this.
Link:
Or find Sigma Nutrition Radio on all podcast apps.
.
@MichelleObama
announced yesterday the launch of her food-and-drinks company Plezi.
Packaged in colorful bottles that look like typical sugary energy drinks, PLEZi drinks retail for $3.89 for a pack of four and are intended to be less sweet than other children’s beverages.
Can we use observational data to emulate randomized dietary intervention trials?
Come find out!
#ASN2023
Monday session for the nutrition research methods nerds out there…
📔 New clinical nutrition med school textbook is out soon & available for pre-order!
"Essentials of Clinical Nutrition in Healthcare"
I'm proud to have co-authored the chapter on *Popular Diets* with the one and only
@KCKlatt
Pre-order:
If a car leaves Boston going 60mph at 7:00am to retrieve child's shoes left in Rhode Island 60 miles away, and camp starts at 8:00am, and at 7:30am you realize the only pair of back-up shoes are in the same car on its way to Rhode Island, then how much coffee do you need?
It’s important clinically to understand RR *and* absolute risk reduction.
But for the 👏LOVE 👏OF 👏SCROD the minuscule cumulative incidence of CVD from a 3-yr pharma trial is 👏NOT the same thing as your 👏PATIENTS’ 👏LIFETIME 👏RISK.
HEART DISEASE IS THE
#1
CAUSE OF DEATH.
You can tell the TwitterDiet zealots have never been involved in any clinical nutrition research bc they're constantly in mentions about drug trials with disease endpoints whining about absolute vs relative risks, things you can't even calculate for their fav diet alternative..
It was an absolute honor to contribute to the development of this consensus report, years in the making, and to work alongside such impressive colleagues and legends.
Where does the evidence stand for clinical uptake of precision medicine in diabetes?
@AJCNutrition
Editorial by me, Frank Hu, and Clemens Wittenbecher comments on mis-suse of systematic reviews/meta-analyses in nutrition research and where improvements can and must be made. 1/
I enjoyed writing a For Debate perspective for
@diabetologia
in defense of *Systematic Reviews and Meta-Analyses*. Their place atop evidence hierarchies is questionable and the method is far from perfect.
So why should we continue to publish them?
Why anecdotes are not data:
When you say "my patients did amazingly well with this diet"
All I hear is "8 dentists reccomend Crest toothpaste "
Denominators matter. A lot.
A few words of caution for those who debate the causes of obesity using food/nutrient time trend data...
Firstly, total population level availability data, like sugar in the 20th century, say nothing about how sugar is being consumed.
Commentary I wrote w/
@Riudecanyenc
just published in
@NatureMedicine
on some methodological issues to consider in causal interpretation of Mendelian randomization & dietary exposures.
Thread later this week but please read and share in the meantime ✌️
Below are the familiar overweight and obesity prevalence trends for US adults. See the inflection point at mid-1970s?
Accepted hypothesis: Obesity epidemic not due to change in genetics.
My wild hypothesis: ALSO NOT DUE TO A SUDDEN COLLECTIVE ABANDONMENT OF WILLPOWER.
17/
I came into this project genuinely unattached to any outcome.
These data are very clear. Women with higher LDL, regardless of so many other metabolic traits and factors, were more likely to die during follow-up.
Although it's been said, many times, many ways:
Case series of Diet X for change in some health intermediate (weight, lipids, etc.) aren't useful to anyone but the individual.
No denominator. No comparator. No causal inference. No public health gains.
If you think epi methods are some done-deal established set of tools finalized in a text book somewhere, you would be wrong.
There is SO much methods work to be done, especially for nutrition science. It just doesn't get funded.
better methods = better science = better health
It's getter harder to blame obesity epidemic exclusively on our passive existence in an obesogenic environment. The crap food environment clearly evolved to tap into something biological, not just societal/economic.
Double-blind, randomized, placebo-controlled trial, adolescents (n=201) w
#obesity
randomized to receive weekly subcutaneous semaglutide (2.4 mg) or placebo for 68w, plus lifestyle program.
#BMI
change at wk 68 was −16.1% w semaglutide & 0.6% w placebo.
*Clinicians* and *public health* often use different tools to address a common problem.
Let's say a building was on fire 🔥
One quickly puts out fire with water.
One investigates & improves electrical wiring standards.
We want both tools. Both are good. Remember this.
And the big kicker -- only 8% of US adults report ever having attempted to follow its dietary guidance. LESS THAN 10% HAVE TRIED TO FOLLOW THE DIETARY GUIDELINES.
And if we're convinced pts over estimate their healthiness, then 8% is likely an OVERestimate.
🍽️6
Open data has many upsides, but access w/o prerequisites sparked a concerning trend in retraction-worthy analysis errors. Glad to see the original trialists point out the glaring mistakes in this mss using their data, but that doesn’t always happen and is prob too late anyway.
A paper by
@AdrianSotoMota
@nicknorwitz
@davidludwigmd
et al. reanalyzed data from our previous study, claiming to have invalidated our primary findings, & concluding that our data support the carbohydrate-insulin model (CIM) of obesity. Our response:
A handful of requests for posthoc redefinition of "metabolically healthy" -- to define a healthy hba1c as <5.5 instead of 6.5%.
Here. Stratified on "met healthy". LDL & all-cause mortality. Crude incident rates & fully adjusted HRs.
Interpret as you will. In the folder.
@DukeRamze
@deirdre_tobias
Genuinely curious. Can someone run the data on pop hBA1c below 5.6? Most doc would agree that above 5.7 is unhealthy. If I read the results as it stands, it tells me "high LDL in an unhealthy pre-diabetic population" is not a good thing.
Found the missing chamber under the food pyramid that holds all the secrets to diet and health but I'm not telling where it is even though I'm dedicating my life's work to reversing the obesity epidemic w/o personal financial gain b/c I'm an establishment researcher.
Let me first remind you of "Is everything we eat associated with cancer? A systematic cookbook review" and its scathing illustration published a decade ago now by Prof Ioannidis.
Take-away: Everything is good. Everything is bad. Nutritional epi get your act together.
Noted.
THE most impactful proponent of Mendelian randomization methods has pumped the breaks, acknowledging his concerns that MR can generate spurious relationships & overstated causal conclusions, especially in nonlinear associations.
TLDR: 🤌obesity still not protective for mortality
1/n In 2019
@BMJ_latest
published a non-linear Mendelian randomization (NLMR) paper with the conclusion that gaining weight would reduce mortality risk in many people. The NLMR method produces spurious findings MR's credibility calamity in this 🧵
Chandler’s genome didn’t change. The environment did.
So it’s environment.
But Chandler gained less than many of his peers over that same time period, despite living in the same city, education, and job status.
So it’s his genetics.
It’s both.
All ask is that your dietary exposure is a well-defined dose-response isocaloric substitution with a distribution representative of actual intake in the population ascertained via repeated diet assessments validated against objective biomarkers, why is this so hard?
I'm hiring a postdoc in 'omics & obesity epidemiology! A strong analytic skillset is desirable. Come join our
#NCI
funded research
@MeDOCResearch
investigating the role of obesity in cancer
@BWHDPM
in Boston!❤️
Link to apply:
You cannot be a nutrition scientist without knowing who Dr. Willett is -- a true pillar of nutritional epidemiology.
Definitely adding this episode to my To-Do list:
My conversation with Walter Willett, PhD is now out.
Was an honour to sit down with him in Boston a few weeks ago.
Mastering the Art of Healthy Eating: Diet Tips and Nutrition Research |... via
@YouTube
A gene-environment interaction is exactly how we can see massive population shifts in BMI over time due to environmental changes, and still explain statistical heterogeneity and heritability with genetics.