The current RDA (Recommended Dietary Allowance) for protein has historical origins and important limitations; it was not designed to define an optimal intake for health, longevity, or athletic performance, and should be interpreted cautiously when advising individuals.

Intro summary: 'The origins and limitations for the RDA for protein and what the evidence suggests about optimal intake for health, longevity, and performance.'

Public-health approaches have thus far had limited success in solving obesity at the population level; future solutions may lean more heavily on pharmacotherapy (e.g., GLP-1 receptor agonists) or require broad societal changes rather than relying solely on current public-health measures.

Preview: 'the difficulty of tackling obesity through public health, the limits of current approaches, and whether future solutions may rely more on drugs like GLP1 agonists or broader societal changes.'

The longstanding recommended dietary allowance (RDA) for protein intake is 0.8 grams of protein per kilogram of body weight per day.

""0.8 grams per kilogram of body weight""

Speaker explains origin and ubiquity of the 0.8 g/kg recommendation in public discourse about protein.

For a 180 lb (≈82 kg) adult, the transcript cites the RDA as about 60–65 grams of protein per day (i.e., meeting the usual RDA-derived intake for nitrogen balance).

Speaker used their own body weight (180 lb ≈ 82 kg) to illustrate that the RDA corresponds to roughly 60–65 g/day for that example.

Some experts (e.g., Don Layman cited) recommend distributing protein intake across meals, aiming for about 30 g of protein per meal, eaten 3–4 times/day (implying ~90–120 g/day) to maximize anabolic stimulus compared with consuming the same total in one sitting.

""about 30""

Contrasts single large protein bolus (e.g., 60 g in one sitting) with a distributed strategy of ~30 g/meal × 3–4 meals advocated by proponents of meal-based protein distribution.

Metabolic/nitrogen-balance studies (USDA-based) showed that lean, sedentary young men (~65–70 kg) achieved nitrogen balance on approximately 0.8 g protein/kg body weight — roughly 50 g protein for a 65–70 kg person.

Referenced USDA-based nitrogen-balance research in lean inactive young men used to justify the commonly cited 0.8 g/kg protein recommendation.

Do not generalize the 0.8 g/kg nitrogen-balance finding to other populations without adjustment — older adults, pregnant people, those recovering from injury/surgery, bodybuilders, or physically active individuals likely require higher protein intake.

Speaker explicitly contrasted the study populations with other groups (older, pregnant, injured, bodybuilding, active) where protein needs differ.

When applying clinical research, explicitly compare the study population to your patient: consider body size, training status, and specific goals (e.g., survival vs performance) because these differences change applicability of results.

"You have to look at the population that is studied and ask the question, how do I differ from that population?"

Transcript emphasizes checking 'Am I bigger? Am I training? Do I have a more ambitious goal...?' before applying study findings.

Evolutionary selection optimizes reproductive fitness (gene transmission), which can favor different phenotypes than those that maximize individual longevity; clinical goals (longevity vs reproductive success or size/strength) may therefore conflict with evolutionary drives.

"which is if you want to win the evolutionary game, that's your goal, which is different than living a lot longer"

Speaker contrasts 'winning the evolutionary game' (gene transmission) with 'living a lot longer' (individual longevity).

When judging scientific claims focus on three things in order: the data, the methods used to collect the data (which determine probative value), and the logic connecting the data to the conclusions; other considerations are secondary.

"in science, three things matter: the data, the methods, and the logic connecting the data to conclusions"

Advice from speakers about how to assess trustworthiness of scientific claims and handle perceived conflicts of interest.

Disclosure of industry funding or advisory roles (e.g., paid advisor, grants from industry groups) does not by itself determine whether a person’s scientific conclusions are trustworthy; assess the underlying data, methods, and logic instead.

Speakers discussed their own industry ties (e.g., advisory role, grants from protein-related industry groups) and how listeners might judge trustworthiness.

When evaluating a study, focus on three elements: the primary data, how those data were collected (methods), and the explicit chain of logic linking data to conclusions; transparency on all three is essential for declaring findings known versus uncertain.

"the three things that matter are what are the data? How were the data collected? How were the methods used to collect them? And then what is the string of logic that connects those data to their conclusions?"

General guidance on research appraisal, stated as a central principle for assessing validity of scientific claims, especially relevant to nutrition/lifestyle studies.

Nutrition scientists often struggle less with logic and more with the difficulty and cost of collecting reliable human data, because humans cannot easily be confined or controlled for long experimental periods, which limits the feasibility of tightly controlled trials.

"the species of interest is not amenable to close quarters for long periods of time"

Explains a central methodological limitation in human nutrition research that drives reliance on observational data and complicates causal inference.

Because controlled experimentation in free-living humans is often impractical and expensive, expect nutrition literature to include a mix of observational studies, pragmatic trials, and mechanistic inference; apply stricter causal criteria and triangulation (multiple methods) before changing clinical recommendations.

Practical implication for clinicians: interpret nutrition research with an expectation of methodological limitations and seek converging evidence across study types before acting.

Nutrition debates are amplified by strong social, economic, religious and personal value influences, which cause emotional responses and lead people to 'go beyond the data' and substitute non-scientific arguments.

Speaker contrasts methodological limits with the social/emotional drivers that distort interpretation and public reception of nutrition research.

Research priority recommendation: invest in development and real-world validation of objective dietary assessment technologies (e.g., wearable sensors, biochemical biomarkers, AI-assisted logging), focusing on free-living validation and reproducibility.

Implied actionable direction from discussion that measurement advancements are necessary to improve nutrition science.

Dietary research in free‑living people is limited by inadequate measurement of actual food intake; improving objective intake measurement in free‑living conditions is a high priority because current self‑report methods are insufficient.

Highlights the need for better methods to quantify food intake outside controlled settings to improve validity of nutrition research.

Because you cannot randomize or blind every dietary exposure, researchers should rely on rigorous causal inference methods and pragmatic trial designs to estimate effects when traditional blinded RCTs are infeasible.

Methodological recommendation for nutrition researchers to use alternative designs and analytic methods when RCT blinding/randomization is impractical.

Adherence is a critical determinant of dietary intervention effectiveness—simple instructions (for example, to "drink one of these every day") cannot be assumed to be followed, so trials and clinical recommendations must measure and report adherence explicitly.

Emphasizes the need to track and report adherence in both research and clinical practice to interpret outcomes correctly.

Nutrition research is fundamentally limited by measurement error in dietary intake—researchers often cannot verify what participants actually ate or drank, making exposure classification uncertain.

General discussion of challenges in nutrition research: measurement validity of intake data and exposure ascertainment.

Adherence is a major practical limitation: asking participants to consume 'one of these every day' does not guarantee they actually do so, and distinguishing exact dosing (e.g., 'one and only one') is difficult in free-living studies.

Example used by speaker about instructing daily beverage consumption to highlight adherence uncertainty.

Duration constraints: short-term mechanistic effects can be measured quickly (speaker used 15 minutes after ingestion as an example), but long-term outcomes like longevity would require impractically long adherence (example cited: 'every day for the next 20 years'), making direct human longevity trials infeasible for timely answers.

Contrast between short-term measurable effects and impracticality of decades-long randomized nutrition trials for longevity outcomes.

Crossover trial designs in nutrition are vulnerable to carryover effects—participants receiving diet A then diet B (and vice versa) can have residual effects from the first period that confound comparisons; parallel-group designs avoid this Achilles heel.

Speaker referenced an ongoing methodological debate (Kevin Hall vs David Ludwig) and credited Ludwig for highlighting carryover issues in crossover nutrition trials.

Carryover effects in crossover trials can persist despite a washout period; a washout is not an absolute guarantee that prior treatment effects are eliminated.

"There's almost no way around it. Even with a washout between the crossover. Not an absolute way."

Discussion about methodological limits of crossover designs in human trials, especially nutrition studies where interventions may induce lasting biological changes.

Crossover designs provide increased statistical power by enabling within-subject comparisons (e.g., paired Student t-test), allowing fewer subjects and lower cost for studies where measurements are expensive or logistically limited.

Rationale for choosing crossover designs in resource-intensive settings such as metabolic chamber studies.

Because the validity of crossover results 'depends on your argument' about carryover rather than providing an a priori proof, trial reports and protocols should explicitly prespecify carryover assessments, sensitivity analyses, and justify washout durations.

Practical implication for trial design, analysis plans and reporting to mitigate interpretive uncertainty introduced by potential carryover.

When participant availability or throughput is limited (e.g., rare populations where recruiting 1,000 participants is impossible), a crossover design is strongly favored because it yields much greater statistical power and precision with far fewer subjects.

""I can't get 1,000 people even if I had the money because they don't exist. I can only get 10 or 20 people.""

Speaker contrasts feasibility constraints (patient availability, limited chambers) with sample size needs and favors crossover designs in low-availability settings.

Crossover trials are "much more statistically powerful" than parallel-group trials in almost all circumstances, meaning you can obtain the same precision with substantially fewer subjects.

Speaker emphasizes the statistical efficiency gain of crossover designs when assumptions are met.

Carryover from the first treatment into the second period can confound period-2 comparisons: observed differences in the second period may reflect residual effects of period-1 treatment rather than the true A vs B contrast.

Speaker explains the core threat to internal validity in crossover designs—residual/ carryover effects that bias treatment comparisons.

A sufficiently long washout can mitigate carryover for interventions with reversible, short-lived (e.g., molecular) effects, but you can "never say, 'Absolutely rule it out'"—washout reduces but does not eliminate residual-risk.

""you can never say, 'Absolutely rule it out.'""

Speaker states washout is useful for reversible/molecular effects but cannot guarantee absolute elimination of carryover.

Crossover designs are inappropriate or limited for interventions that cause lasting or permanent changes—examples include bariatric surgery, anatomical resections, learning/psychosocial interventions, and allergens that may permanently sensitize the body.

Speaker lists intervention types where washout is impossible or impractical due to permanent or long-lasting effects.

For rare-population studies where recruiting large parallel-group samples is impossible, crossover designs allow use of limited available subjects (e.g., 10–20 people) to obtain meaningful comparisons that would otherwise require much larger numbers.

""I can only get 10 or 20 people.""

Speaker contrasts feasible recruitment numbers (10–20) with ideal large samples and promotes crossover for small-N feasibility.

Crossover trials can produce persistent (non-washout) effects for some interventions (e.g., vaccines or any exposure that causes lasting biological change), which limits the validity of simple crossover comparisons unless carryover is explicitly assessed and ruled out.

Speakers contrasted interventions with lasting biological effects (like measles vaccine) to typical crossover assumptions about washout.

Observational epidemiologic studies are not inherently useless; they allow weaker causal inference and leave open alternative explanations (measurement bias, sampling bias, reporting bias, confounding) but can still provide valuable evidence if limitations are acknowledged.

Speakers argued that observational studies are weak inference but not zero value and should be interpreted with their specific biases in mind.

When interpreting nutrition studies, be explicit about the scope of generalization — e.g., effects found for 'cheddar cheese made in Wisconsin' cannot automatically be generalized to all cheese types, all manufacturing locations, or to cheese consumed with different dietary contexts.

Illustrative example used to emphasize specificity and limited generalizability of food-based intervention studies.

Researchers and clinicians should explicitly list alternative explanations for observed effects (e.g., measurement bias, sampling bias, reporting bias, confounding) rather than claiming definitive causation from single-study designs.

""the study shows what its study shows. It's weak inference, but it's not nothing.""

Speakers emphasized acknowledging limitations to avoid overstating causation from observational or limited trial designs.

For goals beyond basic survival (avoiding sarcopenia, improving physical performance), experts in the transcript recommend a protein intake 'in the ballpark of 1.2 to 1.6 grams per kilogram' as a minimum effective dose, with values 'easily up to 2' g/kg for those pursuing higher optimization.

"in the ballpark of 1.2 to 1.6 grams per kilogram ... but easily up to 2"

Recommendation contrasted with RDA; aimed at optimization (sarcopenia prevention, peak/near-peak performance) rather than mere survival.

The RDA for protein is described as intended for basic adequacy and survival; several experts argue it is insufficient when the objective is health optimization (e.g., preventing sarcopenia or maximizing performance).

"the RDA is insufficient if you're actually trying to optimize health"

Transcript references Don Lehman and others who challenge the RDA for optimization purposes.

Many biological variables follow nonmonotonic/U-shaped relationships where both deficiency and excess cause harm; examples include thyroid hormone and total calories—'too little will kill you, too much will kill you.'

""too little will kill you, too much will kill you""

Used to caution against assuming more is always better; applies to hormones, caloric intake, and other physiological regulators.

RDA for protein is approximately 0.4 g per pound (≈0.8 g per kilogram); increasing intake to roughly double the RDA (≈0.8 g/lb or ≈1.6 g/kg) — or slightly more — shows no evidence of harm in virtually all groups except very rare individuals.

Speaker framed this as a practical safe upper range relative to the RDA for most people when calories are otherwise appropriate.

There is little debate between the RDA 'base level' and a higher, 'superior' level near double the RDA—most experts/speaker view the higher intake as at least not worse and in the majority of cases better than the RDA.

""base level like economy rental""

Speaker uses metaphor comparing RDA to an 'economy rental' and the higher level to a 'good rental' to convey relative benefit and broad consensus.

Phenylketonuria (PKU) and specific protein allergies (e.g., whey allergy) are explicit exceptions: people with PKU must avoid phenylalanine and those with a whey allergy should avoid whey, but these are restrictions on specific proteins rather than on total protein intake.

"phenylketonuroic can't have phenylalanine, okay, fine. Someone who's got allergy to whey protein, can't have whey protein"

Speaker distinguishes between specific amino-acid or protein-type contraindications and general protein recommendations.

Higher protein intake is described as beneficial for medium-term measurable outcomes including body weight regulation, appetite control, bone strength, and muscle mass/strength.

Speaker lists multiple observable benefits of consuming more protein compared with baseline/RDA intake.

The speaker states they 'know of no evidence for harm' from higher protein intake even in people with chronic kidney disease, while acknowledging uncertainty for 'the very rarest folks.'

"I know of no evidence for harm, even in people with chronic kidney disease"

Speaker is taking a permissive stance toward higher protein in CKD based on their understanding, but frames it as personal knowledge rather than systematic review.

The speaker argues that for most adults the RDA for protein (~0.8 g/kg, e.g., ~60 g/day for the speaker) is likely too low and that many people are better served by a higher intake around 1.6 g/kg (described as “2X the RDA”), which for the speaker equates to ~150–160 g/day.

Transcript discussion contrasting RDA-level protein (~60 g/day for the speaker) versus ~1.6 g/kg (~150–160 g/day) as a target for robustness/muscle preservation and growth.

Protocol: When asserting that higher protein intake causes harm, demand controlled human intervention studies (randomized preferred; nonrandomized controlled acceptable) that manipulate separable levels of protein and demonstrate deleterious effects on clinically meaningful endpoints (e.g., myocardial infarction, stroke, mortality, strength, function, appearance).

"show me the data"

Speaker issued an open call to experts for intervention studies proving harms of differing protein intakes in humans and specified criteria for acceptable evidence.

Warning/Observation: The speaker contacted leaders (including those skeptical of higher protein intake) and reported receiving no controlled human intervention studies that demonstrate clinically meaningful harms from differing protein intakes—indicating that current claims of protein-related harm may rest on observational or surrogate-data rather than controlled trials.

Open call over ~12+ months to top experts asking for intervention evidence of harms from protein; none provided.

Explanation: Surrogate changes (e.g., shifts in a molecule level or gut microbiota composition) are not intrinsically meaningful to patients and should not be taken as proof of harm unless they are shown to lead to hard clinical outcomes (e.g., increased mortality, cardiovascular events, or loss of strength).

"We don't intrinsically care about whether this molecule in our body is higher than that molecule or this gut microbe."

Speaker emphasized prioritizing outcomes that matter to patients rather than molecular or microbiome changes alone.

Total parenteral nutrition (TPN) is used when patients cannot receive enteral feeding (e.g., ventilated ICU patients with nonfunctional guts); TPN is delivered via a central venous catheter/central line because enteral access and gastrointestinal feeding are not possible.

"they can't consume enteral nutrition, which means they can't eat because they're probably ventilated and their guts aren't even working"

Speaker described TPN context in ICU patients too sick for enteral nutrition (likely ventilated, gut not working).

Randomized trial evidence in critically ill, catabolic ICU patients showed no statistically significant mortality benefit from a higher-protein feeding strategy; the higher-protein group also did not clearly fare better on other major clinical endpoints in that trial.

"there was no statistically significant effect on lifespan."

Speaker recalling an ICU RCT comparing higher vs lower protein intake in very catabolic, critically ill patients (possibly receiving TPN); exact study not specified.

Findings from ICU settings (e.g., TPN-fed or unconscious patients) should not be extrapolated directly to free-living individuals choosing foods at the grocery store; the clinical context, metabolic state, routes of feeding, adherence, and goals differ substantially.

"Most of us are saying, when I go to the grocery store and decide what I want to bring home for dinner tonight, then what?"

Emphasis on external validity limitations when applying critical-care nutrition data to general population dietary advice.

What would be most informative are large, free-living randomized trials with high adherence in the target population (tens of thousands to allow subgroup analyses); such trials are uncommon, so reliance on imperfect epidemiologic or small RCT data is necessary but limits causal certainty.

Speaker describing the ideal evidence needed for definitive dietary guidance and current evidence gaps.

When synthesizing evidence for lifestyle or nutritional recommendations, explicitly weigh the type and generalizability of each study (animal, cell, short-term metabolic, epidemiology, clinical trials) because causal inference is frequently fraught when studies don't align; recommendations are strongest when multiple independent lines of evidence converge.

"If they all line up great, then it's easy."

General guidance for evidence appraisal in lifestyle medicine; speaker contrasts ideal aligned evidence (e.g., smoking) with common misalignment across study types.

Mouse studies, cell studies, and short-term human feeding studies — particularly those conducted in overfed conditions — often do not generalize to long-term real-world human eating patterns, so be cautious extrapolating these results to routine dietary recommendations.

Speaker warns about common misinterpretation when translating preclinical or short-term metabolic studies into long-term dietary advice.

Acute mechanistic studies of muscle protein synthesis (MPS) have examined protein doses across roughly 0.8, 1.0, 1.2, 1.4 and 1.6 g protein/kg body weight to define a dose–response and identify a plateau beyond which MPS no longer increases.

Speaker references a specific dose–response study that measured MPS at incremental per‑kg protein intakes.

Training status modifies the acute protein dose required to maximize MPS: less-trained (untrained) individuals achieve higher MPS responses at lower amounts of amino acids compared with more trained individuals, implying protein dosing for maximal acute MPS should account for training/adaptation status.

"be careful what patient population you're looking at in the study and make sure it applies to you."

Interpreting the MPS dose–response study — speaker emphasizes population (training) differences.

When evidence from different domains (cell/animal/epidemiology/clinical trials) does not align, clinicians should evaluate both the intrinsic strength of each study and its external generalizability to the target patient population rather than relying on a single study type.

Practical decision-making rule for weighing conflicting or incomplete evidence in lifestyle interventions.

When multiple lines of evidence (cell, animal, epidemiology, clinical trials) consistently point in the same direction, causal inference is strengthened; the speaker used smoking as an example where all evidence types aligned to support the recommendation to stop smoking.

""Smoke is really bad. Don't smoke.""

Methodological principle for evaluating causality in lifestyle medicine evidence.

When applying protein-dose MPS studies to individual patients, match the study population to the patient: training status matters (less-trained people may require lower per-kg protein to maximize acute MPS), so tailor per-kg protein prescriptions rather than applying a single threshold to all.

""be careful what patient population you're looking at in the study and make sure it applies to you.""

Practical protocol recommendation for clinicians using MPS literature to guide protein dosing.

In previously sedentary adults (the majority of people in the U.S.), a practical minimal resistance training prescription — whole-body sessions 30 minutes long, three times per week, performed at moderate intensity (not to failure and allowing recovery the next day) — produces substantial, clinically meaningful training benefits.

Speaker contrasts untrained (sedentary) individuals with already-trained people to illustrate magnitude of benefit from a low-dose program.

Trained individuals who are already doing high volumes of training (e.g., an hour per day) will derive little to no additional benefit from the same low-dose program that substantially helps sedentary people; the incremental gains are minimal once a person is near the training 'asymptote.'

Used as a rationale for differing responses to identical interventions across baseline fitness levels.

When interpreting or applying study results, always match the study's baseline population and training dose to your patient — e.g., you cannot reasonably compare outcomes from someone training an hour per day to someone who went from sitting to training 90 minutes per week; external validity matters.

"be careful what patient population you're looking at"

General guidance about study-to-patient translation; numerical examples used to illustrate mismatch in baseline activity.

Small, initial interventions produce large effects in deficient or untrained systems but little change in systems already near normal or optimized — illustrated by analogies (leptin-deficient mice respond strongly to small leptin dose; students naive to algebra gain from small tutoring doses).

Conceptual explanation used to explain differential response to interventions based on baseline.

For previously sedentary adults, a pragmatic initial resistance-training protocol is three whole-body workouts per week of 30 minutes each (90 minutes/week total), performed without reaching failure or profound exhaustion (intended so the person can get out of bed the next day); this regimen produces large, clinically meaningful training benefits in untrained individuals.

"unbelievable, unbelievable benefit"

Speaker contrasted effects in completely sedentary people starting a modest program versus already-trained individuals; emphasized non-exhaustive sessions and clear numeric regimen (3 × 30 min/wk = 90 min/wk).

Always match the study population to your patient: do not generalize benefits observed in previously sedentary subjects to already well-trained patients (and vice versa); interpretation of interventions (exercise or nutrition) must consider baseline training/status.

"be careful what patient population you're looking at in the study and make sure it applies to you"

Speaker warned to 'be careful what patient population you're looking at in the study and make sure it applies to you.'

The common claim that 'you don't need much more protein than 0.8' is criticized as an over-simplification based on limited evidence; the speaker likens relying on that single low estimate to telling a child they only need 10 minutes of algebra practice per day to master the subject, arguing the claim underestimates how much protein many people may need.

""That's like telling a kid they only need to study algebra at 10 minutes a day if they want to master it.""

Speaker responds to frequent citation of a study supporting a 0.8 (unit unstated) protein threshold and uses an analogy to emphasize insufficiency of that interpretation.

Nutrition randomized controlled trials (especially those on interventions like protein intake) typically have much smaller sample sizes than pharmaceutical trials (examples given: small nutrition studies with as few as ~6 participants per group versus pharmaceutical trials with ~60,000 participants), differing by "multiple orders of magnitude."

""different by multiple orders of magnitude""

Speaker contrasts sample sizes across fields (nutrition vs. statins, GLP-1 agonists, vaccines) to explain why nutrition results are often underpowered.

Because many nutrition trials are small and underpowered, the resulting evidence base is weak, which helps explain why nutrition studies often fail to demonstrate large, clear effects even when real effects may exist.

This links the methodological problem (small sample sizes) to practical outcomes (inconclusive/lack of large effect findings).

A central explanation for small, underpowered nutrition studies is economic: it is harder to fund large-scale nutrition RCTs, so investigators often cobble together funding from government and industry rather than receiving single large funder support.

""This is really an economic challenge, not an intrinsic challenge.""

Speaker emphasizes funding structure as a root cause rather than an intrinsic scientific limitation of nutrition research.

The commonly cited figure of “0.8” (g/kg/day) for protein is often presented as if it represents the optimal intake, but that recommendation is a minimal requirement and may be insufficient for many goals (e.g., muscle maintenance, aging populations); the speaker compared treating 0.8 as ‘enough’ to telling a child to study algebra 10 minutes/day and expect mastery.

"you don't need much more protein than 0. 8 because of that study... like telling a kid they only need to study algebra at 10 minutes a day if they want to master it."

Critique of interpreting the 0.8 g/kg/day protein reference value as an optimal target rather than a minimal RDA-level requirement.

Nutrition randomized trials are routinely much smaller than pharmaceutical trials, with examples cited such as nutrition subgroup comparisons having ‘six in each group’ or studies of ~600 people versus pharmaceutical trials enrolling ~60,000, producing much lower statistical power and limited ability to detect subgroup effects.

"60,000 over there in that pharma study... six in each group."

Comparison of sample sizes between nutrition studies and large industry-funded pharmaceutical trials and the impact on inferential strength.

Because of small sample sizes and heterogeneous nutrition exposures, many nutrition studies fail to show large effects even when clinically meaningful differences may exist in subpopulations; thus null results in small trials should be interpreted cautiously.

Implication of limited power and heterogeneity in nutrition research causing potential false-negative findings or inability to detect subgroup effects.

Pharmaceutical companies can feasibly fund very large randomized controlled trials because drugs are patentable and the companies can recoup multi-billion-dollar development costs (examples cited: single programs costing 'hundreds of millions' up to '2–3 billion', taking ~10 years), whereas nutrition studies lack that economic return and so rarely secure comparable funding; the speaker contrasted '60,000 people' (pharma) vs '600 people' (nutrition) as illustrative sample-size differences driven by economics.

Explanation of why pharma runs very large RCTs while nutrition research remains small.

The FDA's statutory mandate requires a 'reasonable basis' that benefits outweigh harms for a proposed use, and in practice the agency expects large randomized controlled trials (among other evidence) in order to grant marketing approval for pharmaceuticals.

Explains the regulatory bar that drives the design and scale of drug trials.

Pharmaceutical randomized trials are among the most rigorous human health studies available because companies invest enormous resources and time—single development programs can cost in the low billions and span roughly a decade—making them able to meet high evidentiary standards.

Claims about trial rigor tied to investment and scale (supports why pharma evidence often appears stronger than nutrition evidence).

Food products and whole-food nutrition interventions face structural barriers to large trials because they are difficult to patent (the speaker used the example 'grapefruit') and food-industry margins are lower, so companies have less incentive and capacity to fund large, long, expensive randomized trials.

"it's hard to patent the grapefruit, right?"

Rationale for why nutrition science is underfunded relative to pharmaceuticals; explains patentability and margin issues.

The practical consequence of these economic and regulatory dynamics is that absence of large-scale, high-cost randomized trials in nutrition often reflects funding and patentability constraints rather than proof that nutritional interventions are ineffective.

Interpretive implication for clinicians and patients when evaluating the evidence base for nutrition interventions.

FDA regulatory pathways generally require a reasonable basis that benefits outweigh harms for a proposed drug use, which is commonly established by large randomized controlled trials and other supporting data — this regulatory bar drives the need for very large, expensive RCTs.

"they must have a reasonable basis for concluding that the benefits outweigh the harms under proposed conditions of use."

Explains how statutory/regulatory requirements create demand for large, rigorous RCTs in drug development.

The financial scale of modern drug development can be enormous (speaker estimated '2 billion today, 3 billion, whatever the number is, 10 years and a couple billion dollars'), which is economically justifiable for drugs that can be patented and recoup costs, but not for foods or many nutritional interventions.

"let's say it's 2 billion today, 3 billion, whatever the number is, 10 years and a couple billion dollars"

Provides approximate development cost and timeframes used to explain why pharmaceutical companies invest in large trials while food/nutrition does not.

Foods and basic nutritional exposures are harder to monetize and patent (example: 'it's hard to patent the grapefruit'), so food industry margins and incentives typically can't justify funding large RCTs comparable to drug trials.

"it's hard to patent the grapefruit"

Explains patentability and margin differences as structural reasons for fewer large-scale nutrition trials.

Practical implication: absence of large, high-cost RCTs in nutrition often reflects economic and patentability constraints rather than proof that nutritional interventions lack effect — interpret null or low-quality nutrition evidence in light of these structural limits.

Advises clinicians and guideline developers to consider funding/structural biases when assessing the evidence base for dietary interventions.

A randomized controlled trial (commissioned by Frito‑Lay) compared snacks fried in corn oil (higher PUFA), low‑fat snacks (higher carbohydrate), and traditional snacks higher in saturated/trans fat; when calories were controlled, the corn‑oil (full‑fat) chips produced more favorable cardiometabolic biomarker outcomes than the low‑fat/high‑carb snacks or the higher‑saturated‑fat snacks.

"you're better off eating the full-fat corn oil chips"

Study was industry‑commissioned and measured biomarkers/CV risk surrogates rather than clinical cardiovascular events; speaker references publication in a major nutrition journal.

In that trial the exception was triglycerides: the low‑fat/high‑carbohydrate arm produced worse triglyceride levels, while the high‑fat corn‑oil arm improved triglycerides relative to the low‑fat/high‑carb diet; by contrast, traditional trans‑fat containing products were the worst overall for biomarkers.

Speaker framed this as a recollection of the trial results (biomarkers focussed); triglycerides responded differently than other biomarkers.

The trial primarily measured biomarkers / CV‑type surrogate endpoints rather than clinical cardiovascular events, so results should be interpreted as effects on risk markers not hard outcomes.

Speaker explicitly asked and answered that the outcomes were biomarkers.

A randomized controlled trial commissioned by Frito‑Lay compared chips fried in corn oil (polyunsaturated fat) vs low‑fat/high‑carbohydrate chips vs traditional chips/cookies/crackers higher in saturated and trans fats and found that, when calories were controlled, the corn‑oil (full‑fat) option produced better cardiovascular biomarkers overall; trans‑fat versions were worst and the low‑fat/high‑carb arm produced the highest triglycerides.

Speaker describes a 20+ year old industry‑commissioned RCT published in 'A. J. C. M.' with biomarker (CV risk) outcomes; calories were controlled across arms.

Randomized nutrition trials are expensive—the speaker estimates their own trial would cost roughly on the order of a million dollars in today's costs— which limits how many such trials are performed and who can fund them.

Speaker referencing the cost of a trial they conducted and how cost constrains research.

NIH and other public funders have historically treated many nutrition intervention studies as the sort of research 'the industry should fund'—a perspective that contributes to underfunding of trials by public sources and shapes what research gets prioritized.

Explaining institutional assumptions about who should pay for applied nutrition research.

Large observational epidemiology in nutrition can be very expensive but often produces limited 'new information yield'—even studies with millions of subjects using self-reported intake and a few biomarkers may not resolve causal questions (e.g., relationships between protein intake and longevity) or substantially change prior uncertainty.

Critique of the value and limitations of large observational nutrition studies.

The speaker supports repurposing research funding away from low–information-yield observational studies toward other priorities (summarized as 'less here, more here') and notes that NIH is actively addressing funding priorities, though he agrees with some NIH changes and disagrees with others.

Policy-level opinion on research funding allocation.

When evaluating new large epidemiologic nutrition studies, clinicians should be skeptical about claims of definitive answers when exposure measurement is primarily self-report with only a couple biomarkers—such studies may at best generate hypotheses rather than settle causal questions.

Interpretive guidance derived from critique of observational nutrition literature.

When evaluating new large nutrition studies, clinicians should note measurement method: self-report dietary assessment with only a 'couple of biomarkers' limits the study's ability to answer causal questions, even at very large sample sizes.

Speaker emphasizes measurement limitations (self-report + few biomarkers) as a key limiter of interpretability regardless of sample size.

The speaker argues the single biggest limitation of nutrition epidemiology (for questions like protein intake and longevity) is opportunity cost: large, expensive epidemiologic cohort studies consume funds that could instead support one large randomized controlled trial or multiple medium-sized RCTs that would yield stronger causal evidence.

"the greatest limit or problem with the nutrition epidemiology ... is the opportunity cost."

Applied to research funding decisions and study design priorities for nutrition/protein and long-term health outcomes.

Classic epidemiologic limitations remain central: confounding (including healthy user bias) makes observed associations between protein intake and health outcomes difficult to interpret — correlation is not causation, as illustrated by the ice cream–murder example (heat is the true confounder).

"correlation is not necessarily causation."

General caution when interpreting associations from observational nutrition studies.

Measurement error in dietary assessment is a major and often underappreciated problem: measurements are frequently non-random (systematic), and because they are not random and are often not accounted for statistically, they can introduce substantial bias into epidemiologic estimates of diet–health relationships.

"measurement is not random and even if it was random, it's usually not taken into account."

Pertains to dietary measurement methods (food frequency questionnaires, recalls, etc.) and their statistical handling in cohort studies.

The speaker argues that the single greatest limitation of nutrition epidemiology in studying protein intake and long-term health is the opportunity cost: large, expensive observational cohorts divert funds that could instead run one high-quality randomized controlled trial (or several medium-size RCTs) to answer causal questions about protein and longevity.

"the greatest limit or problem with the nutrition epidemiology ... is the opportunity cost"

Commentary about research prioritization in the field of protein consumption and long-term outcomes (longevity, major health events).

Confounding and healthy-user bias substantially limit inferences from observational studies of protein intake and health; associations (for example between protein intake and mortality or disease risk) can reflect unmeasured lifestyle, socioeconomic, or behavioral factors rather than causal effects of protein per se.

"correlation is not necessarily causation"

General limitations of nutritional epidemiology applied specifically to protein–health outcome associations; refers to commonly discussed biases including 'healthy user' effects.

Dietary measurement error is a major problem: intake assessment is often systematically biased (not random) and such non-random measurement error is usually not fully accounted for in analyses, which can distort associations between protein intake and long-term outcomes.

"measurement is not random and even if it was random, it's usually not taken into account"

Addresses the measurement validity challenges in nutritional epidemiology as applied to protein consumption studies.

Overall, the speaker warns that continuing to rely heavily on large observational nutrition cohorts for questions about protein and longevity risks producing interesting but ultimately inconclusive findings, and may delay definitive, practice-changing evidence obtainable from well-designed randomized trials.

Synthesis of the prior criticisms emphasizing the consequences for evidence quality and clinical guidance on protein intake.

Non-random measurement error is a major source of bias in epidemiologic studies: when reporting error is correlated with exposure (e.g., people who eat more of X systematically underreport compared with those who eat less), statistical adjustments that assume random error will not correct the bias and can leave or create spurious associations.

Applied particularly to dietary and self-reported exposure data in observational studies.

Confounding—especially by culture, socioeconomic status, social class, and education—is one of the three principal biases that distort epidemiologic questions and must be considered and controlled for when interpreting observational associations.

Speaker lists confounding as the top bias when evaluating epidemiologic evidence.

Selection bias (including collider bias) arising from who enters a study, who stays in it, and how exposures are classified over time can produce misleading associations; controlling for a collider can create an inadvertent association between exposure and outcome.

Includes enrollment, retention, and timing-of-exposure decisions as sources of selection-related bias.

The speaker identifies the three biggest intrinsic issues for epidemiologic inference as (1) confounding (notably by culture and socioeconomic status), (2) measurement error (particularly non-random), and (3) selection biases—these three should be explicitly examined when interpreting observational findings.

A concise prioritized list offered by the speaker about biases most likely to affect observational results.

Investigators commonly engage in selective emphasis or de-emphasis of results—often not explicit lying but both intentional and unintentional distortion—so reader skepticism and transparency (e.g., pre-registration, full reporting) are necessary to mitigate this threat to validity.

"I don't think many investigators are lying in an explicit sense, but I do think there are both intentional and unintentional efforts at distorting that as people want to tell a story and they emphasize some things and de-emphasize others."

Speaker frames this as a non-intrinsic but large problem affecting credibility of published findings.

Confounding by cultural factors, socioeconomic status, social class and education is a primary source of bias in observational lifestyle research and can produce spurious associations if not adequately controlled.

Speaker lists confounding — particularly by culture, socioeconomic status, social class and education — as the top bias affecting epidemiological questions about lifestyle exposures.

Non-random measurement error—for example, systematic under-reporting of intake by people who eat more of a food—can bias results in ways that simple statistical adjustment for random error cannot correct.

Speaker emphasizes measurement error is often non-random and correlated with exposure level, leading to biased estimates.

Selection biases, including collider bias (arising when you control for or select on a variable influenced by both exposure and outcome), are major threats—examples include who chooses to join a study, when people start a study, and how exposure timing is defined.

Speaker names selection bias and collider bias and cites issues of study entry timing and exposure classification as important sources of bias.

Practical approach for clinicians interpreting lifestyle epidemiology: explicitly assess (1) potential confounding by socioeconomic/cultural variables, (2) whether measurement error is likely non-random (and how it might bias results), and (3) selection/timing mechanisms (including possible collider bias); prioritize studies that address these issues with design or causal methods.

Synthesis of speaker's prioritized biases (confounding, non-random measurement error, selection biases) into actionable steps for critical appraisal.

Many journal editors lack the ability, resources, or willingness to perform deep forensic checks on submitted manuscripts (accessing raw data, reproducing analyses), so routine peer review often cannot detect fabricated or manipulated data; only a minority of top journals (e.g., NEJM, Science, JAMA) have relatively greater capacity but still cannot catch everything.

Describing limitations of editorial and peer-review processes in biomedical journals.

Peer reviewers function like restaurant critics — they evaluate presentation, interest, and perceived quality of the manuscript — but they are not health-inspector equivalents who can perform spot checks, surprise inspections, or lab-level audits of raw data and methods.

"peer reviewers are like restaurant critics."

Analogy used to explain the scope and limits of peer review.

Journals sometimes request and obtain raw data when reviewers spot anomalies, and in many such cases the raw data reveal clear problems; authors who resist sharing raw data often provoke suspicion and conflict with reviewers/editors.

"we will get the raw data from people. And then we'll often see things that are quite funny."

Speaker recounts observations from editorial practice about handling suspicious manuscripts.

There is a systemic need for 'health-inspector' style oversight in scientific publishing — authorized spot-checks, surprise audits, and access to investigational equipment/personnel — to detect misconduct that conventional peer review cannot catch.

Recommendation for structural changes to improve research integrity enforcement beyond peer review.

There is a need for a formal 'inspector' function (analogous to health inspectors) with authority to do surprise checks, spot testing, and equipment-based verification beyond what peer review typically performs.

Speaker suggests editorial peer review is insufficiently empowered to perform the equivalent of compliance inspections and recommends authoritative oversight with spot checks.

Detecting 'tortured phrases' or 'word salad' via automated screening can serve as a heuristic that suggests plagiarism, automated text generation, or fabrication and warrants further investigation of the manuscript and underlying data.

""tortured phrases""

Speaker specifies that certain awkward or unnatural phrasings are used as screening flags indicating potential misconduct or fabrication.

Many journal editors lack the resources, technical ability, or willingness to deeply verify underlying study data or detect fabrication; this limits the capacity of editorial review to ensure published research integrity.

Speaker contrasts majority of editors with a minority who have more resources, noting resource, skill, and courage deficits as barriers to in-depth checks.

Peer reviewers function like 'restaurant critics'—they evaluate presentation, interest, and plausibility of manuscripts, but typically do not and cannot perform forensic checks of raw data or laboratory practices.

""peer reviewers are like restaurant critics""

Analogy used to distinguish the roles of peer reviewers versus regulatory inspectors who perform spot checks and have equipment/authority.

Use logical-statistical consistency checks (the so-called 'grim test' family of checks): if you know the scale and sample size, the reported mean can only take certain values, so a reported mean outside those feasible values indicates a likely error or fabrication.

Applies to psychometric and other bounded scales where minimum/maximum and sample size constrain possible means.

Training AI 'peer-review' or fraud-detection agents on corpora of known fraudulent manuscripts to teach pattern recognition is a reasonable strategy to improve detection, but current efforts are early-stage and not mature.

Proposal for improving automated detection of fabricated or plagiarized research.

Statcheck is an automated tool that verifies reported statistics when authors follow the American Psychological Association (APA) reporting format by checking internal consistency of reported test statistics and p-values.

Useful for journals or reviewers working with APA-style reported results to catch transcription or calculation errors.

There are no compelling observational epidemiologic data arguing against exceeding the Recommended Dietary Allowance (RDA); large studies with hard endpoints usually analyze protein intake continuously rather than testing discrete threshold effects for exceeding the RDA.

Speaker framed this in response to a question about evidence against exceeding the RDA (discussion context referenced protein intake specifically).

Do not rely solely on AI-based peer-review or manuscript-screening agents for fraud detection: these tools are in early stages and should be used as adjuncts alongside human expertise and traditional checks.

Speaker emphasizes the infancy of AI peer-review agents and implies current limitations in reliability.

In the speaker's view, 'I don't think there are any compelling observational epidemiologic data' demonstrating harm from exceeding the Recommended Dietary Allowance (RDA); most large epidemiologic studies examine protein intake as a continuous variable rather than testing hard thresholds above the RDA.

"I don't think there are any compelling observational epidemiologic data."

Responding to a question about epidemiologic evidence against exceeding the RDA (context implies protein intake), the speaker states a lack of compelling observational evidence and notes study designs favor continuous analyses.

Randomized and observational studies examining protein intake vs hard clinical endpoints (cancer, heart disease) are mixed and not dispositive; many studies analyze protein continuously rather than using hard intake thresholds, so both directions (beneficial and harmful effects of higher protein) have been reported.

Speaker emphasizing heterogeneity of the literature and limitations in how protein intake is analyzed.

Protein intake is highly confounded by social class and by the type/source of protein consumed (e.g., plant vs animal), which complicates causal interpretation of observational associations between protein amount and disease outcomes.

Speaker identifies major confounders that may bias observational protein–outcome relationships.

A pragmatic minimum protein target of about 1.0 g per kilogram body weight is presented as reliably safe (example: ~85 g/day for an 85 kg person), but the speaker states uncertainty about harms if intake is doubled (~2 g/kg), including potential cancer risk.

""I don't know that if you double that, that you're not going to get cancer.""

This is offered as a practical safety floor rather than a definitive optimal dose; speaker explicitly notes uncertainty about higher intakes.

A practical lower-bound for protein intake cited is approximately 1 gram per kilogram of body weight (example: ~85 g/day for an 85 kg person) as a level that is 'safe' against under-nutrition in adults.

"let's just round up and call it one gram per kilogram of body weight"

Speaker argues that at ~1 g/kg body weight people are not going to 'starve to death' and uses 85 g as an explicit example.

The relationship between higher protein intake and risks of cancer or heart disease is uncertain — published studies show associations in both directions and are not dispositive.

Speaker notes that literature can be shown to support either protective or harmful effects of higher protein; absolute knowledge is lacking.

The speaker reports that they are aware of no compelling human evidence showing that eating dishes like Fettuccine Alfredo (high in saturated fat and sodium) causes increased cancer or definitive clinical harm, and therefore cautions against asserting such harms without human data.

Responding to claims that specific high-fat restaurant foods directly cause heart attacks or cancer.

The speaker advises skepticism toward extrapolating from mouse studies and routine epidemiologic studies to assert human clinical harms from specific dietary items, implying limitations in external validity and causal inference for those study types.

General methodological criticism of animal models and observational epidemiology when used to claim human dietary harms.

Regarding protein intake, the speaker highlights that current evidence is not clear enough to demonstrate that the RDA for protein is too low, but also not clear enough to definitively prove it is adequate, leaving the clinical needlepoint uncertain and open to reasonable disagreement.

Discussion about whether recommended dietary allowances for protein should be increased.

Be skeptical when extrapolating from mouse studies or certain epidemiologic studies to human clinical harm — the speaker specifically cautions against overinterpreting animal models and some observational data.

The speaker called for skepticism of mouse studies and certain epidemiologic studies as evidence of harm from dietary components like saturated fat.

The evidence base is currently insufficient to definitively conclude whether the recommended dietary allowance (RDA) for protein is too low or adequate; both claims (RDA too low vs. adequate) lack definitive data according to the speaker.

Speaker discussed debates about higher protein intake and noted that evidence neither clearly shows the RDA is too low nor that it is adequate beyond doubt.

The current evidence base is insufficiently clear to definitively state that the RDA for protein is too low or that it is adequate; the speaker frames this as genuine uncertainty rather than a settled conclusion.

Discussion of protein RDA and whether status quo should be accepted or re-evaluated.

A practical policy approach is to define a consumption bracket that would cover ~90% of the population — i.e., an intake range that is safe and appropriate for day-to-day grocery-store decisions — rather than waiting for definitive toxicity trials.

Speaker suggests focusing on a pragmatic range for most people to simplify clinical/patient guidance.

Clinicians should translate the complex scientific debate into 'straightforward' guidance for patients who are primarily asking practical questions like 'How much protein should I be eating?' or 'What should my family be eating?', recognizing patients want concise, actionable advice.

Emphasis on patient-facing communication and simplifying nuance into practical recommendations.

Patient counseling on protein intake should be preference-sensitive: for example, lifelong vegetarians who genuinely dislike the taste/texture of meat should be managed according to their preferences rather than assuming avoidance is ideology-based.

Speaker describes patient subgroups and reasons for their dietary patterns to inform respectful, individualized counseling.

The speaker argues that maintaining the status quo indefinitely is not always the most prudent path; in the context of protein guidance, they believe it's appropriate to re-examine recommendations after extensive review.

Policy stance favoring active reassessment of nutrient recommendations rather than passive maintenance.

Because most listeners/patients 'don't care about most of what we've said' and want simple answers, clinicians should distill scientific debates into clear, personalized recommendations rather than presenting all uncertainty and nuance at the first encounter.

Practical communication strategy for clinician-patient interactions regarding nutrition uncertainty.

Clinician counseling should be tailored to patient preferences and practicalities—e.g., for lifelong vegetarians who 'can't stand the feel of meat' the guidance should respect preference rather than insist on meat-based protein.

"who can't stand the feel of meat"

Speaker emphasizes individualized dietary counseling and respecting dietary preferences when recommending protein sources.

For people aiming to 'thrive' (goals like longer life, greater strength, overall health), aim for roughly 2.0 grams of protein per kilogram of body weight per day, with that daily intake spaced out across meals throughout the day.

Speaker frames this as the recommended target for people who want more than mere survival (i.e., better strength, health, longevity).

A practical, easy-to-remember heuristic is to target 1.6–2.0 g/kg/day of protein, which the speaker summarizes as 'almost a gram per pound of body weight.'

"almost a gram per pound of body weight"

Presented as a simple rule-of-thumb for everyday use.

If the goal is mere survival, the RDA (recommended dietary allowance) for protein is probably adequate for most people, but the RDA is not the speaker's target for those who want to optimize strength, health, or longevity.

Contrast between minimal sufficiency (RDA) versus targets for optimization.

For people aiming to 'thrive' (goals such as greater strength, healthspan, longevity), target approximately 2.0 g protein per kg body weight per day, with intake spaced throughout the day.

Speaker contrasts RDA (survival) with higher intakes for thriving and explicitly recommends ~2 g/kg/day, spaced across meals.

For people who only want to 'survive', the RDA is probably adequate, but the speaker distinguishes 'survive' (RDA) from 'thrive' (higher protein targets).

"if you just want to survive, the RDA is probably okay for most people"

Speaker explicitly contrasts minimal RDA-level protein (survival) with higher intakes recommended for additional health/performance goals.

Patients who are lifelong vegetarians or who avoid all animal products commonly cannot tolerate or choose not to eat meat and will likely have difficulty reaching the upper protein targets; in practice you may only be able to nudge them up to about 1.2 g/kg/day.

Speaker differentiates between patients who dislike meat and those who avoid all animal products; states an upper practical limit (~1.2 g/kg) for these patients despite efforts to increase intake.

Warning: people who will not eat any animal products will have a harder time reaching higher protein intakes and therefore may need targeted strategies (e.g., more frequent plant-based protein meals, fortified foods, or supplementation) to approach recommended goals.

Speaker notes the practical difficulty of reaching upper protein targets on strict plant-based diets and implies the need for pragmatic adjustments.

The RDA for protein is framed as a minimal threshold ('just enough to pass') rather than an optimal intake; clinicians and patients should view the RDA as the baseline, not necessarily the target for optimal function or performance.

"the RDA for protein intake is like what my uncle saw the American schools as, it's a yes, it's just enough to pass high school and do something."

Speaker uses analogy comparing RDA to minimal schooling; no numeric RDA value provided.

Example product: a drink containing ~20 grams of protein and ~90 calories (described as 'almost pure protein'); using these as a sole food source likely would not, in the speaker's view, directly cause kidney failure in a healthy person, but would risk inadequate vitamins/minerals, reduced dietary pleasure, and insufficient carbohydrates to support high-intensity exercise.

Speaker describes a personal favorite protein product and hypothetical exclusive use as a diet.

Harm from an extreme single-focused behavior (e.g., studying 12 hours/day or living solely on protein drinks) typically arises from the substitution effect — i.e., displacing other necessary behaviors or nutrients (exercise, socializing, sleep, vitamins/minerals), not from the focal activity itself.

"It's not the direct effect of the studying. It's the substitution effect."

Speaker contrasts direct physiologic harm with harm caused by what the behavior replaces.

When recommending or evaluating concentrated single-nutrient products (e.g., protein shakes), clinicians should assess whether the patient's overall diet provides essential micronutrients and dietary variety and whether macronutrient balance supports their functional goals (e.g., performance), rather than focusing solely on the safety of the concentrated product.

Derived from speaker's approach of questioning vitamins/minerals, pleasure, and carbohydrate availability when considering an all-protein diet.

There is no example cited in the transcript showing that high-intensity cognitive activity (e.g., studying algebra 12 hours/day vs 2 hours/day) directly causes harm; harms are more likely to arise from substitution effects (reduced exercise, socializing, or sleep) rather than the studying itself.

"Did anybody ever say or show that if you study algebra or anything else, 12 hours a day, instead of two hours a day, it directly causes harm. And I don't know of anything like that."

Speaker contrasts direct effects of intense studying with harms from substituting away from other health behaviors.

The speaker used a concrete example of a ready-to-drink product that contains 20 grams of protein and 90 calories (described as 'almost pure protein') and cautioned that living primarily on such products could leave a person deficient in vitamins/minerals and reduce pleasure from eating.

"This is a drink that has 20 grams of protein and 90 calories, almost pure protein."

Used as an example to illustrate substitution risks when consuming concentrated protein products as the main food source.