REF-1124 · Novotny, Gebauer & Baer, 2012 · The American Journal of Clinical Nutrition, · Paper · PubMed
Discrepancy between the Atwater factor predicted and empirically measured energy values of almonds in human diets
Novotny et al. fed 18 healthy adults (10 men, 8 women) a fully controlled American diet with 0, 42 or 84 g/d of whole almonds, 18 days per phase, in random order, and collected all faeces and urine for the last 9 days of each phase. Bomb calorimetry of diet, faeces and urine gave the metabolizable energy of the almonds themselves: 4.6 ± 0.8 kcal/g, or 129 kcal per 28-g serving.
The Atwater factors used for labels and databases predict 6.0–6.1 kcal/g, or 168–170 kcal per serving. In the authors' words the Atwater factors overestimated the measured energy of almonds by 32%. The reason measured here is fat: faecal fat rose from 1.7 to 6.3 and 10.8 g/d with the almond dose.
A label calorie is an average-digestibility estimate, and for whole almonds that average is wrong in one direction.
Protocol
Randomised crossover controlled feeding; three 18-day phases with 0 (control), 42 or 84 g/d of natural whole almonds; the base diet was cut to keep each phase isocaloric. Twelve participants received all three doses; three repeated the 0-g dose and three the 84-g dose to estimate within-person variability (n = 36 almond observations).
Participants: 10 men and 8 women, aged 56.0 ± 8.6 y (range 32–67), body weight 79.5 ± 14.5 kg, BMI 27.4 ± 4.2 (reported as mean ± SEM).
Weight-maintenance diet on a 7-day menu; portions adjusted in 200-kcal steps during the first 9 days if weight drifted; no adjustments in the 9-day balance period. Almonds eaten at breakfast and dinner under observation. Chewing not instructed.
Balance period: all faeces (Brilliant Blue marker) and urine collected for 9 days. Energy by adiabatic bomb calorimetry; protein by Dumas nitrogen; fat by ether extraction.
Metabolizable energy (ME) of almonds calculated from the difference in gross and metabolizable energy intake between the almond and control diets, per participant.
Key Results
| Outcome | Result (full text) |
|---|---|
| Measured ME of almonds | 4.6 ± 0.8 kcal/g = 129 kcal per 28-g serving |
| Atwater prediction | 6.0 kcal/g by the specific factors (168 kcal/serving, used by the USDA database); 6.1 kcal/g by the general 4/9/4 factors minus insoluble fibre (170 kcal/serving, label method); 181 kcal/serving by the general factors without the fibre subtraction |
| Size of the error | Atwater factors overestimated measured energy by 32% (129 vs 168–170 kcal per serving) |
| Dose | 4.7 ± 0.7 kcal/g at 42 g/d vs 4.3 ± 1.0 kcal/g at 84 g/d; no dose effect (P ≥ 0.2) |
| Repeatability | first vs second measure 4.6 ± 0.7 vs 4.6 ± 1.0 kcal/g (P ≥ 0.8); mean intraindividual CV 6% |
| Individual estimates | 2.2 to 6.0 kcal/g; all but 2 of 36 observations below the Atwater value |
| Faecal fat (0 / 42 / 84 g/d) | 1.7 / 6.3 / 10.8 g/d (P < 0.0001) |
| Faecal energy (0 / 42 / 84 g/d) | 132.2 / 217.7 / 282.3 kcal/d (P < 0.0001) |
| Energy digestibility of the whole diet | 90.5% / 87.5% / 85.5% |
| Age, BMI | no effect on almond energy density |
Mechanisms
Measured: faecal fat, faecal energy and faecal dry weight rose with almond dose; fat digestibility of the whole diet fell from 97.8% (control) to 93.1% (42 g/d) and 89.9% (84 g/d).
Authors' interpretation (from other studies they cite, not measured here): fat in whole nuts sits inside intact plant cell walls that survive chewing and digestion, so part of it never becomes available; more chewing reduced faecal fat in one cited study.
Authors' interpretation: almond butter is likely closer to the label than whole almonds (based on peanut work); not tested here.
Inference, not tested: part of the 2.2–6.0 kcal/g spread between individual estimates is measurement noise, because each estimate is a difference between two whole-diet balances; the within-person CV of 6% shows repeated estimates in one person agreed closely.
Practical Application
For whole almonds the label overstates what the body gets: 129 kcal measured vs 168–170 kcal predicted per 28-g serving, an Atwater overestimate of 32%, in 18 adults on a controlled diet (B+). Calories are not imaginary; the label is an average-digestibility estimate, and for whole nuts that average is too high.Logged from the label, a handful of almonds counts more than was absorbed, so the error leads to eating slightly less than intended, almost never more (34 of 36 individual estimates were below the Atwater value).
The error is small in a whole day: arithmetic on the numbers above, 42 g/d (1.5 servings) is counted roughly 60 kcal higher than measured. The authors calculate that 84 g/d eaten in place of highly digestible foods leaves 100–150 kcal/d less available energy at intakes of 2000–3000 kcal/d.
Do not transfer it to almond butter, flour, milk or oil: only whole natural almonds were tested, and the authors expect butter to be closer to the label.
Do not re-calculate every nut by hand. The error runs the same way in 34 of 36 measurements, and the 2-3-week body-mass trend shows whether intake is right whatever the logging error (house practice, SYNTH-87).
Limitations
n = 18 healthy, mostly overweight, middle-aged adults (BMI 27.4); no athletes and no high energy intakes.
Whole natural almonds only, in a mixed American diet; roasted, chopped, ground or butter forms were not tested; chewing was not controlled.
The almond ME is calculated by difference between two whole-diet balances; individual estimates spread from 2.2 to 6.0 kcal/g.
Funded by the USDA and the Almond Board of California; a commodity board has an interest in a lower calorie figure, although the method is standard and the lead group is a government laboratory. The authors declare no conflict of interest ("None of the authors had a conflict of interest."), and the paper states: "The Almond Board of California played no role in the design, implementation, or analysis of the study or in interpretation of the data."
Body weight and body composition were not outcomes; weight-loss consequences are the authors' extrapolation.
How this entry was checked and changed
Why this level
controlled crossover, n = 18 (rubric: crossover ≥15 = B+); the outcome, energy the body gets from almonds, is the question itself rather than a surrogate; read from the full text
Full text (checked 2026-10-13)
Overall mean energy density 4.6 ± 0.8 kcal/g, or 129 kcal for a 28-g serving; Atwater 6.0 and 6.1 kcal/g, 168–170 kcal; the Atwater factors "overestimate the energy content of almonds by 32%" (full text).
Measured values for individual volunteers ranged from 2.2 to 6.0 kcal/g; all but 2 of 36 observations were below the Atwater value (full text).
Mean intraindividual CV 6%; 4.7 ± 0.7 kcal/g at 42 g/d vs 4.3 ± 1.0 kcal/g at 84 g/d, no dose effect (P ≥ 0.2) (full text).
Faecal fat 1.7, 6.3 and 10.8 g/d; whole-diet energy digestibility 90.5%, 87.5% and 85.5% (full text).
84 g almonds daily in exchange for highly digestible foods: 100–150 kcal/d less available energy at intakes between 2000 and 3000 kcal/d (authors' calculation) (full text).
Supported by the USDA and the Almond Board of California; "The Almond Board of California played no role in the design, implementation, or analysis of the study or in interpretation of the data"; "None of the authors had a conflict of interest" (full text).