Nutrition Reference

Dietary Assessment

Doubly Labelled Water

Also known as: DLW, doubly labeled water

The reference method for measuring total energy expenditure in free-living humans, using the differential elimination of two stable isotopes over one to two weeks.

By Dr. Helena Weiss · RD, PhD (Nutritional Sciences) ·

Key takeaways

  • Subjects drink water enriched with deuterium (²H) and oxygen-18 (¹⁸O); the two isotopes leave the body at different rates, and the difference gives carbon dioxide production and therefore energy expenditure.
  • It is the criterion method against which every self-report instrument — food records, recalls, questionnaires, and app-based logging — is validated.
  • Its precision for total energy expenditure over a measurement period is generally cited in the range of 2-8%, which sets a floor on how well any self-report method can ever be shown to agree.
  • Because it measures expenditure rather than intake, it can only validate reported intake when body weight is stable across the measurement period.
  • Cost and isotope supply make it impractical for routine use, which is why the dietary-assessment literature relies on a small number of DLW studies to calibrate everything else.

Doubly labelled water (DLW) is the reference method for quantifying total energy expenditure in people going about ordinary life. A subject drinks a measured dose of water enriched with two stable isotopes — deuterium (2H) and oxygen-18 (18O) — and provides urine or saliva samples over the following one to two weeks.

The method rests on a single asymmetry. Deuterium leaves the body only as water. Oxygen-18 leaves as both water and carbon dioxide, because oxygen exchanges between body water and the bicarbonate pool. The 18O label therefore disappears faster, and the difference between the two elimination rates is proportional to carbon dioxide production. Combined with an assumed or measured respiratory quotient, that yields total energy expenditure.

Why it anchors the entire field

Every self-report instrument in nutrition — the weighed food record, the 24-hour recall, the food frequency questionnaire, and by extension the smartphone food log — is validated by comparison against a criterion measure. DLW is that criterion.

Its central finding has been remarkably stable across four decades: self-reported energy intake is systematically lower than DLW-measured expenditure in weight-stable adults, and the gap is not small. Under-reporting on the order of 20% of intake is a routine result, and it is larger in participants with higher body mass. Any discussion of dietary-assessment accuracy that does not account for this is discussing a smaller problem than the real one.

The intake-expenditure substitution

DLW measures expenditure, not intake. Using it to evaluate reported intake requires the assumption that the two are equal, which holds only when body weight and composition are stable across the measurement window. Studies that validate intake reporting against DLW therefore restrict enrolment to weight-stable participants and verify stability by weighing at both ends.

Where weight is changing, the energy balance term must be estimated and added, and the added uncertainty degrades the comparison. This is a frequent source of confusion when DLW results are cited outside their original design.

Precision, and what it implies for app validation

DLW's own precision for total energy expenditure across a measurement period is generally given as roughly 2-8%, depending on dose, sampling schedule, isotope-ratio mass spectrometry performance, and assumptions about the dilution spaces.

This has a consequence that is often missed in consumer-facing accuracy claims: the criterion method has error too. An instrument cannot be demonstrated to agree with truth more closely than the reference can establish truth. Where a validation study reports agreement substantially tighter than the reference method's own precision, the appropriate reading is that the comparison was made against a different reference — typically weighed portions scored against a composition table, which is a per-meal reference rather than a free-living energy-balance reference.

Those two designs answer different questions. Weighed-portion comparison asks whether an instrument can estimate the contents of a plate placed in front of it. DLW asks whether a person's reported intake matches what their body actually expended over two weeks of normal life, including the meals they forgot. The second is the harder question and the larger source of error, and it is not a property of the software.

Practical limits

  • Cost. Isotope dose plus mass spectrometry puts per-subject cost in the range where studies are typically dozens rather than hundreds of participants.
  • Isotope supply. 18O availability has constrained study design at various points.
  • Time resolution. DLW gives an average over one to two weeks. It cannot resolve a single day, and therefore cannot evaluate day-level or meal-level accuracy at all.

The last point is the one that matters most for interpreting technology claims. A method that averages across fourteen days is the right tool for detecting systematic under-reporting and the wrong tool for asking whether tonight's dinner estimate was correct.

Frequently asked

Why is doubly labelled water called the gold standard?

Because it measures total energy expenditure in people living normally, without requiring them to report anything. Every self-report instrument depends on the participant remembering and describing what they ate; DLW depends only on isotope elimination kinetics, which the participant cannot bias. That independence is what makes it the reference against which reporting methods are calibrated, and it is why the systematic under-reporting of energy intake was discovered at all.

Can doubly labelled water validate a food-logging app?

It can validate the app as an intake-reporting instrument over one to two weeks in weight-stable subjects, which is a meaningful and rarely-performed test. It cannot validate per-meal estimation accuracy, because it produces a single average across the whole measurement period and has no time resolution below that. Per-meal accuracy requires a different design — weighed reference portions scored against a composition table — and the two results are not interchangeable.

How accurate is doubly labelled water itself?

Generally cited at roughly 2-8% for total energy expenditure across a measurement period, depending on dose, sampling schedule, and mass spectrometry performance. This is the often-overlooked ceiling on validation: no instrument can be shown to agree with truth more precisely than the reference can establish truth. Claims of agreement much tighter than that were measured against a different reference, not against DLW.

References

  1. Schoeller DA. "Measurement of energy expenditure in free-living humans by using doubly labeled water". The Journal of Nutrition , 1988 .
  2. International Atomic Energy Agency. "Assessment of Body Composition and Total Energy Expenditure in Humans Using Stable Isotope Techniques". IAEA Human Health Series No. 3 , 2009 .
  3. Subar AF, Freedman LS, Tooze JA, et al.. "Addressing Current Criticism Regarding the Value of Self-Report Dietary Data". The Journal of Nutrition , 2015 .
  4. "FoodData Central — reference composition values". U.S. Department of Agriculture .

Related terms