The target side of the protein question for the over-65 stratum: how much, and distributed how, to maintain and regain lean mass and function. The lever exists because of Anabolic Resistance — old muscle needs a bigger per-meal protein stimulus for the same building response — so the general-adult RDA is a floor set for a different question and undershoots this stratum. The evidence is expert consensus on surrogates (Bauer PROT-AGE, endorsed by four geriatric societies), backed by small metabolic studies — not a hard-outcome (fracture / disability / mortality) trial.
The target ladder — by health and activity status
PROT-AGE recommends, for adults >65:
| Stratum | Daily protein target | Locus |
|---|---|---|
| Healthy older adult | 1.0-1.2 g/kg BW/d | «consume an average daily intake in the range of 1.0 to 1.2 g/kg BW/d» |
| Exercising / otherwise active | 1.2 g/kg BW/d | «higher protein intake (ie, 1.2 g/kg body weight/d) is advised for those who are exercising and otherwise active» |
| Acute or chronic illness | 1.2-1.5 g/kg BW/d | «who have acute or chronic diseases need even more dietary protein (ie, 1.2e1.5 g/kg body weight/d)» |
| Severe kidney disease (eGFR <30, not on dialysis) | LIMIT — the exception | «Older people with severe kidney disease … are an exception to this rule; these individuals may need to limit protein intake» |
The ladder rises with catabolic load (illness > active > healthy) and inverts only at severe renal impairment. All figures are g protein per kg body weight per day.
The number this replaces — and why the RDA is a floor, not a target
(Bauer et al., 2013) The general-adult RDA is 0.8 g/kg BW/d, defined as the minimum daily protein to prevent deficiency in 97% of the population, derived from nitrogen-balance studies. PROT-AGE holds this «too low for older people»: N-balance likely underestimates requirements (it cannot see muscle redistribution and uses short adaptation windows), and in one 14-week study «consuming the RDA for protein resulted in the loss of mid-thigh muscle area» in healthy older adults even while whole-body composition looked stable. So 0.8 answers deficiency prevention; it is a floor deployed as a 100% Daily Value target, and for maintaining muscle in this stratum it is the start line, not the finish -> The Descriptive-Normative Category Error, Deficiency Repletion vs Enhancement.
The parameter table — daily and per-meal are DIFFERENT quantities (do not collapse them)
(Bauer et al., 2013) The single most common error with these numbers is treating a per-day target, a per-meal dose, and a hypertrophy target as one quantity. They sit on different denominators; only the same-denominator rows are directly comparable.
| Quantity | Value + locus | Denominator | Directly comparable to… |
|---|---|---|---|
| RDA (general adult) | 0.8 g/kg BW/d [EXTRACTED (Bauer …) chunk 01] | protein g / kg BW / day | the PROT-AGE daily rows (same denominator) |
| PROT-AGE healthy older | «1.0 to 1.2 g/kg BW/d» [EXTRACTED (Bauer …) chunk 01] | protein g / kg BW / day | RDA — Bauer raises the same quantity |
| PROT-AGE illness | «1.2e1.5 g/kg body weight/d» [EXTRACTED (Bauer …) chunk 01] | protein g / kg BW / day | rows above |
| Hypertrophy target (healthy adults, RT) | ~1.6 g/kg/d, CI 1.03-2.20 (Morton) -> Protein and Resistance Training for Muscle and Strength | protein g / kg BW / day | SAME UNIT, DIFFERENT objective + population — a ladder by goal, not a contradiction |
| Moore per-meal breakpoint (older) | «0.40 ± 0.19 … g/kg body mass» (0.60/kg LBM) [EXTRACTED (Moore …) chunk 01] | protein g / kg BM / meal | NOT comparable to any per-day row |
| Bauer per-meal threshold | «25 to 30 g protein per meal» [EXTRACTED (Bauer …) chunk 01] | absolute protein g / meal | NOT comparable — different denominator again |
| Bauer per-meal leucine | «about 2.5 to 2.8 g leucine» [EXTRACTED (Bauer …) chunk 01] | leucine g / meal | NOT comparable — a within-meal amino-acid quantity |
- The daily rows are a coherent ladder (RDA 0.8 -> older 1.0-1.2 -> ill 1.2-1.5 -> hypertrophy ~1.6), same unit throughout, ordered by goal and stratum — the higher hypertrophy number does not conflict with the lower maintenance number; they answer maximize muscle during training vs maintain muscle.
- The per-meal rows are a different axis — the distribution constraint, not the daily total. They reconcile with the daily target only through Moore’s own within-programme arithmetic (three meals × ~0.40 g/kg = ~1.20 g/kg/d), which is a consistency check, not independent confirmation -> Anabolic Resistance.
- The young RET breakpoint is a WEAK, non-transporting anchor — do not import it downward as a ceiling. Morton’s ~1.6 g/kg-total knee is statistically non-significant (p=0.079) and the supplement effect it rests on declines with age in a largely-young pooled sample — both extracted on -> Protein and Resistance Training for Muscle and Strength. Combined with Anabolic Resistance (more protein needed per anabolic stimulus with age), the transport direction runs up, not down: an older adult training for hypertrophy should read ~1.6 as a floor, not a ceiling — the non-significant young plateau is no reason to cap intake — while the maintenance floor stays 1.0-1.2 from this page’s own (different) evidence base. This is a stratum distinction, not a contradiction: the two numbers answer maximize muscle in the young vs maintain muscle in the old, and the honest object for the older stratum is a direction + a floor, not a point optimum -> The Underivable Optimum.
Per-meal distribution — clear the threshold at each meal
(Bauer et al., 2013) Because anabolic resistance raises the per-meal trigger, hitting the daily total is not enough if it is unevenly distributed. PROT-AGE: «The per-meal anabolic threshold … is higher in older individuals (ie, 25 to 30 g protein per meal, containing about 2.5 to 2.8 g leucine)». Moore’s breakpoint (0.40 g/kg body mass per meal in older vs 0.24 in younger men — «~68%» greater) is the metabolic basis for the same advice: spread protein across breakfast, lunch and dinner so each meal clears the breakpoint, rather than back-loading it at dinner (the typical Western pattern that wastes the sub-threshold meals). Bauer notes an unsettled alternative — pulse feeding (most protein at midday) also showed benefit — so even distribution is the default, not a proven optimum.
Renal safety — the healthy-kidney vs impaired-kidney split
(Bauer et al., 2013) The routine worry — that higher protein harms aging kidneys — is not supported for healthy kidneys: «reviews of research studies reveal little or no evidence that high-protein diets cause kidney damage in healthy individuals, including those who are older», and 1.0-1.2 g/kg/d «is expected to help maintain nitrogen balance without affecting renal function». The contraindication is a genuine stratum flip, not a general caution: severe kidney disease (eGFR <30, not on dialysis) is where protein should be limited (route (c) contraindication).
Bauer’s statement is a consensus-review direction; the pooled RCT evidence now backs it -> Protein Intake and Kidney Function (Devries 2018 SR-MA, 28 RCTs): in adults without CKD, the change in GFR does not differ between higher- and lower-protein diets, and the GFR rise HP does produce is benign kidney functional reserve, not damage (Devries et al., 2018). Two caveats carry over here: the evidence is low-GRADE surrogate-level (GFR, not hard kidney outcomes — the dedicated protein-to-kidney-outcome trial is still the named gap), and Devries shares Phillips with PROT-AGE, so this is one lineage’s coherent position, not two independent confirmations. The stratum flip (severe CKD -> restrict) is unchanged.
The chronic-outcome evidence — mass yes (in the deficit stratum), strength no (Komar)
(Komar et al., 2015) PROT-AGE and the per-meal breakpoint rest on consensus + acute surrogates. Komar 2015 (gold SR+MA, 16 RCTs, 999 subjects aged >=65, leucine-rich protein 2-7.8 g/d for >=10 days — a different research lineage from the Phillips/Wolfe programme) is the closest chronic clinical-outcome test held for this stratum, and it both supports and bounds the target:
- Supports the mass rationale — where there is a deficit. Leucine-rich protein raised lean body mass +0.99 kg [95% CI 0.43, 1.55; p=0.0005] and body weight +1.02 kg [0.19, 1.85], but subgroup analysis confines the effect to the sarcopenic stratum (LBM +1.14 kg [0.55, 1.74] in sarcopenia vs null, -0.05 kg [-1.55, 1.46], in healthy elderly). So raise protein to protect muscle is cashed at the mass endpoint mainly for those already losing it — a repletion reading, not enhancement in the replete Deficiency Repletion vs Enhancement.
- Bounds it — strength did not move. Grip WMD +0.23 [-0.26, 0.73; p=0.36; I2=65%] and knee
extension +0.07 Nm/kg [-0.26, 0.40; p=0.68] were null — the mass gain did not read out as measured
strength (Surrogate Outcomes). Komar itself flags this as likely under-powered («a small number
of trials potentially insufficient to yield significant results»; broader-inclusion SRs did find grip
gains), so it is insufficient evidence on strength, not proof of no effect — but it withholds any
strength claim from this stratum’s protein target. A different lineage reaches the same grip null:
Coelho-Junior’s observational MA (below) also found no grip/chair-rise advantage for higher intake
[E-independent]— an RCT-pool and an observational-cohort route agreeing that raising protein does not move strength in this stratum. `Corroborated by (Coelho-Júnior et al., 2018) - The exposure is protein+energy, not isolated leucine or protein-per-se. Komar cannot separate leucine from the co-ingested protein/AA matrix, and «an increase in total energy consumption might be a prerequisite for the success of these supplementations with respect to parameters such as LBM» — a reminder that in frail/undernourished elderly the energy the supplement carries may be doing much of the work, so the lever is enough protein and energy, not a leucine trick.
- Pool-overlap caveat (no independence lift). Komar’s elderly-protein RCTs (Tieland, Chalé, Leenders, …) plausibly overlap Tagawa’s protein-supplement pool below, so their agreement that protein raises LBM in older adults is partly shared-evidence, not a second independent route.
The direct-function outcome test — observational, outcome-specific, no knee (Coelho-Junior)
(Coelho-Júnior et al., 2018) Every source above sets the target against a surrogate (nitrogen balance, LBM, acute MPS) or, in Komar, a chronic mass endpoint. Coelho-Junior 2018 (gold SR+MA, 7 observational studies, 8754 community-dwelling adults >=60 across 6 countries) is the closest thing held to a direct physical-function test of the intake bins — one level up the outcome ladder, but bought at the cost of design: it pooled cohorts/cross-sectional/case-control studies and excluded all RCTs and interventions, so it estimates association, not effect. It both partly supports the ladder and bounds it hard.
- Supports raising above the RDA — but only at the widest contrast. Very-high (>=1.2 g/kg/d) vs low (<0.8) reached a small significant SMD 0.18 [95% CI 0.01 to 0.35], p=0.04 on pooled lower-limb function, and high (>=1.0) vs low reached significance on walking speed (SMD ~0.06 [0.02, 0.11]). The authors read this as «findings … indicate that a very high (>=1.2 g/kg/day) and high protein intake (>=1.0 g/kg/day) are associated with better lower-limb physical performance when compared to low protein (<0.80 g/kg/day) intake in community-dwelling older adults».
- Outcome-specific — strength did not move. «relative high protein intake does not seem to propitiate a better performance on isometric handgrip (IHG) and chair rise in comparison to relative low protein intake» — high-vs-low was null on grip (SMD -0.36 [-1.15, 0.44]) and lower-limb strength (-0.09 [-0.26, 0.08]). The association is confined to mobility/walking, not force-generation — the same grip/chair-rise null Komar found by a different route (Surrogate Outcomes).
- No knee is located; the 1.0 cutpoint is imposed, not found. The bins (0.8 / 1.0 / 1.2) were set a priori from the RDA debate, so a categorical contrast cannot show a curve feature — and «there were no significant differences in the physical functioning of high and middle protein intake groups» (0.8-0.99 vs >=1.0 came up empty). Only the widest contrast separated. So the direct-function data cannot distinguish the RDA boundary from the 1.0 target and gives no warrant for a knee at 1.0 — consistent with this page’s direction + a floor, not a point optimum reading The Underivable Optimum. «there is a lack of direct evidence testing the proposed cut-off points for protein consumption».
(inferred from Coelho-Júnior et al., 2018)
- Relative protein (g/kg) is confounded by body weight. The low-intake bin had the highest mean BMI (29.1 vs 27.1 in very-high): a heavier or more adipose person mechanically shows lower g/kg at the same absolute grams, so “low relative protein” partly indexes higher adiposity/lower lean mass, not lower eating. And in raw Table 2 the very-high bin had the lowest absolute grip (19.1 kg) and slowest chair-rise — the classic reverse-causation / frailty pattern, where lighter frailer people land in the high-g/kg bin -> The U-Shaped Association Artifact.
- Self-reported intake (24-h recall, 3-4 day records, FFQ) carries the field’s binding measurement error -> Measurement Error in Dietary Assessment, which attenuates a real gradient toward null — so the flat middle-vs-high result is weak evidence of no gradient, not proof of one. Heterogeneity was «considerable» (I2 up to 96% on mobility); the authors themselves conclude «our results should be taken with caution and should be confirmed with further studies», and call for RCTs.
- One effect-modification hint, held loosely (route (b), the false-positive generator). A single included cohort found protein-function association «in non-sarcopenic, but not in sarcopenic older women» — a subgroup signal running against the expectation that illness raises the need, though the authors also invoke the hypothesis that illness/frailty «may require higher protein levels (1.2-1.5 g/kg)». A single-study subgroup is a candidate to test, not a stratum rule.
The bone-outcome facet — higher protein tracks higher BMD and lower vertebral fracture (one cohort, observational)
(Weaver et al., 2021) Every outcome above is muscle or function; bone is a separate channel of the same exposure. The closest direct test held is Weaver 2021 — a single prospective observational cohort (Health ABC, n=2160 community-dwelling White and Black adults, mean 73.5 y, 51.5% women, 5-y follow-up), not an SR/MA. Protein was measured as % of total energy intake (108-item FFQ; tertiles <13% / 13-15% / >=15% TEI, mapping to ~0.8 / 0.9 / 1.1 g/kg/d), so the upper-vs-lower contrast (~1.1 vs ~0.8 g/kg) is almost exactly the PROT-AGE-target vs RDA contrast this page centres. It gives a directional bone signal at that contrast, but a weak one.
- Cross-sectional BMD — positive, small. Upper vs lower tertile: «mean baseline BMD in participants in the upper protein tertile was 1.8% higher for the total hip, and 6.0% (trabecular), 2.7% (cortical), and 3.3% (integral) higher for the L3 vertebra» (full-model p<.05, positive linear trends). Middle vs lower did not differ — only the widest contrast separated, so no knee at the 13-15% band is located -> The Underivable Optimum.
- BMD change — null. «Protein intake did not affect change in BMD at any site over the follow-up period» (4-y DXA, 5-y QCT). The cross-sectional advantage did not read out as slower bone loss: it is a level difference (lifetime accumulation, or the healthy-eater profile), not a demonstrated effect on the trajectory. This weakens the causal reading of the fracture signal below.
- Fracture — vertebral only, overall-null. «Total protein intake was not associated with 5-year overall fragility fracture risk (Table 4), although reduced risk of clinical vertebral fracture was observed among participants in the upper protein tertile compared to those in the lower protein tertile (hazard ratio: 0.36 [95% confidence interval (CI): 0.14, 0.97], p = .04)». The HR 0.36 rests on 8 vs 14 vertebral events, CI upper bound 0.97 — barely significant. Hip fracture was NOT reduced — «Protein intake did not decrease hip fracture risk in our cohort … the <1% incidence of hip fracture in our cohort limited the study’s power» (insufficient evidence on hip, not no effect).
(inferred from Weaver et al., 2021)
- Residual confounding is the load-bearing caveat. The upper-protein tertile is the textbook healthy-eater profile: «Participants in the upper protein tertile were more likely to be female, White, nonsmokers, nonsedentary, have lower energy intake … higher dietary calcium and vitamin D intakes, be on calcium, vitamin D, or estrogen supplementation, and have higher cognitive function». Full-model adjustment covered these plus BMI, physical activity, osteoporosis meds and diagnosis — but a single cohort cannot isolate protein from the covarying better-nutrition / higher-SES / less-frail bundle. The authors concede «the observational nature of this study does not allow for evaluation of a causal association between dietary protein intake and changes in BMD and fracture risk» -> The U-Shaped Association Artifact.
- Exposure = % TEI, self-reported FFQ. Protein-as-%-energy indexes a dietary pattern (the upper tertile ate less total energy and lower fat/carb %), and FFQ carries the field’s binding measurement error -> Measurement Error in Dietary Assessment (attenuation toward null).
- The animal-vs-source signal is likely total-protein, not source. Animal protein tracked some BMD sites while vegetable protein did not, but Weaver reads this as confounded by dose: «the positive effects on BMD with increased animal protein intake observed in our study may stem from a higher total protein intake, rather than the specific protein source» (and an RCT MA comparing equal animal-vs-vegetable protein found no BMD difference). So this adds no source-choice bone decision -> Dietary Protein and Mortality.
Decision relevance
(inferred from Bauer et al., 2013; Moore et al., 2014)
- For a healthy older adult, aim ~1.0-1.2 g/kg/d (1.2 if active), from ~25-30 g protein per meal across three meals — the move is raising and distributing, not loading. Most older adults eat less than young adults, so the common failure is undershoot, not excess.
- Bias the target UP within a stratum when the cost of undershoot is high — during illness,
weight loss, or an energy deficit (protect lean mass), the illness row (1.2-1.5) or the active row
applies; the plateau is forgiving on the high side for healthy kidneys, so a modest overshoot is
low-harm -> The Estimate-to-Action Gap. The one hard flip down is severe renal disease.
- The weight-loss case is directly evidenced in adults >=50 (Kim et al., 2016): a gold MA of 20 energy-restricted RCTs found HP (>=1.0 g/kg/d) vs normal protein spared lean mass (WMD +0.45 to +0.83 kg) and increased fat loss at the same total weight loss, at a magnitude essentially identical to younger adults -> the deficit-specific target and dose-response live on Protein Intake During Energy Restriction.
- Protein quality rides the number silently. The per-meal leucine threshold (2.5-2.8 g) is why a high-quality/animal or leucine-dense source clears the bar with fewer grams; hitting the target from low-leucine plant sources needs more grams or deliberate complementation -> Protein Quality and the DIAAS Score. This is a distinct decision from the source-substitution (mortality) question on Dietary Protein and Mortality.
- Bone is a third weak outcome channel favouring adequate protein — a benefit signal, not established benefit. (inferred from Weaver et al., 2021) One observational cohort links the PROT-AGE-target intake (>=15% TEI, ~1.1 g/kg) to higher BMD and lower vertebral fracture vs the RDA-level intake — but the bone-loss trajectory was null, hip fracture was underpowered, and residual confounding is unresolved, so grade this leg low-confidence. It corroborates the adequate protein in older adults case at a new outcome without settling it; the muscle/function rationale stays the primary warrant. Some pooled protein MAs report an 11-16% hip-fracture reduction (others null) that this single cohort was too small to see — not a joined tension, a power gap; the held dairy MAs were themselves null on hip fracture -> Dairy and Bone Health, Exercise and Bone Mineral Density. Protein’s bone effect may run partly through the muscle->bone loading pathway, so this channel is not independent of the muscle target -> Protein and Resistance Training for Muscle and Strength. The named gap: no RCT tests protein -> fracture in this stratum.
- The daily amount and the source are two decisions, not one. This page sets the amount and distribution for the older stratum; the animal-vs-plant source choice (mortality) and the hypertrophy target (RT) are separate -> Protein and Resistance Training for Muscle and Strength.
- Protein may raise muscle even where resistance training is NOT on the table — a plausibility pointer, not a stratum-specific proof. For the frail, the dysphagic, or those who cannot train, the PROT-AGE targets above rest on trials that mostly paired protein with exercise. Tagawa’s MA (a different research lineage — Miyachi lab, no PROT-AGE overlap) «demon- strates for the first time that protein supplementation is significantly effective without resistance training in a di- verse population», and even «less than 0.3 g/kg BW/d (0.17 g/kg BW/d, on average) was suffi- cient to significantly increase LBM» — «Daily addition of a high-protein food item such as an egg (6-8 g protein) or 1 cup (200 mL) of milk (6.8 g protein)» is Tagawa’s named lever for «the elderly or people with dys- phagia, as well as in low-income, food-insufficient populations». But the finding is in a diverse population, not specifically the anabolic-resistant old — and Anabolic Resistance predicts a blunted fed-state response at low doses in exactly this stratum, so the without-RT effect may attenuate where it is most wanted. So: protein-by-diet-alone is a reasonable move when training is infeasible (small, distributed top-ups, not loading), and it is a separate lineage from PROT-AGE (partly answering the one research school limit below) — but treat it as a directional pointer to test, not as elderly-specific corroboration; it sets no age-specific dose of its own. (Tagawa et al., 2020)
Limits — consensus on surrogates, one research school, sponsor exposure
(inferred from Bauer et al., 2013; Katsanos et al., 2006; Moore et al., 2014)
- Surrogate, not patient-important, endpoints. The targets are set against nitrogen balance, lean body mass, and acute MPS — surrogates. No fracture, disability, or mortality trial shows that hitting 1.0-1.2 vs 0.8 g/kg/d changes what the person experiences; muscle function and sarcopenia are on the outcome menu, but the causal link from raising protein to better function is asserted, not proven -> Surrogate Outcomes. The loop is open. The one patient-important endpoint now touched is vertebral fracture (Weaver, above) — but it is a single observational cohort with an underpowered, barely-significant signal and unresolved confounding, so it moves the bone leg from surrogate-only to a weak observational fracture signal, not to established benefit.
- Consensus + small studies, one programme — partly widened by Komar. Bauer is a Delphi consensus
(
recommendationtier — cite its underlying reviews for any effect/certainty claim); Moore and Katsanos are small, acute, male-only metabolic studies. Their agreement is coherence within one research school (shared authors and cross-citation), not independent corroboration -> Anabolic Resistance. Moore is male-only; Katsanos is a mixed-sex parallel-group study. Komar 2015 (gold SR+MA, Vienna group) adds a different-lineage chronic-outcome confirmation of the mass claim (in sarcopenic elderly) — partly answering the one-school limit — but it still cites Katsanos and leaves strength and hard outcomes unproven, so it upgrades the mass rationale without closing the surrogate loop. - Sponsor exposure. PROT-AGE was funded by a Nestlé Nutrition grant and most authors disclose medical-nutrition-industry ties (Nestlé, Nutricia, Abbott); Moore and Katsanos are dairy / amino-acid funded. Under symmetric standards this is a directional-bias flag on the “eat more protein / supplement / leucine” conclusion — held loosely, not dismissed, since the underlying tracer physiology is hard to confound.
- Sex, and the >65 label. Moore’s per-meal dose-response is men only (Katsanos included women, so the leucine-threshold finding is not male-restricted); the female per-meal protein breakpoint is only partly evidenced. “Older adult” is treated as >65, but anabolic resistance is a gradient, not a switch at a birthday.