What would make EPK a useful research measure?

The question

Does EPK predict functional outcomes beyond total energy intake, body composition, carbohydrate availability and activity, and could fuel switching or mitochondrial function explain any association?

This is the organising question for the revised seven-paper series. It asks whether a simple intake-to-mass measure earns its place in research, then asks whether measured physiology could explain an association. Those are two different tests. The framework remains a hypothesis: the studies discussed across this collection did not test EPK.

Define the exposure before testing it

For prospective testing, EPK is defined here as mean dietary energy intake in kcal per day divided by body mass in kg, expressed as kcal/kg/day. This intake ratio is not the sports energy-availability measure, which subtracts exercise energy expenditure from intake and divides by fat-free mass. Exercise expenditure, fat-free mass, carbohydrate intake and weight trajectory must therefore be recorded separately. No diagnostic EPK threshold is established by this revision.

Two people with the same EPK may differ substantially in activity, lean mass, recent weight change and health. The ratio therefore needs context. A useful shorthand must still survive comparison with the measurements it summarises.

The mathematical test

EPK is calculated from intake and mass and therefore contains no information absent from those inputs. Its value must be judged as a useful summary, not an independent biological variable. Compare a preregistered EPK model with a flexible model of energy intake and mass, including relevant nonlinear terms and interactions, under the same validation procedure. Adding the ratio to a simple linear baseline would not by itself establish unique physiological information. Report out-of-sample prediction error, calibration and uncertainty; favour the simpler model if performance is comparable.

The comparison should use the same participants, outcome definitions and training and validation splits. Decide the primary outcome and analysis before inspecting the results. Select sample size from an explicit precision or power calculation; a large list of outcomes cannot compensate for an underpowered study.

The physiological test

Measure exposure first, a candidate physiological change next, and function afterwards. A standardised meal challenge with indirect calorimetry could estimate whole-body fuel switching. It cannot, by itself, identify mitochondrial function in the brain, heart or immune system. The cardiac pyruvate work by Cluntun and colleagues illustrates why transport and cellular utilisation deserve their own measurements [1].

A mechanistic substudy could add tissue-specific respiration or other appropriately validated measures. A relationship between EPK and function without the proposed intermediate change would favour a different explanation. Observational mediation remains conditional on assumptions about confounding; it does not prove a dietary mechanism.

Give a null result a clear meaning

The first useful outcome might be that EPK performs about as well as a more complicated model and offers practical simplicity. It might instead fail to generalise, or its apparent benefit might disappear when energy and mass are modelled adequately. Both results answer the question.

The seven papers propose different endpoints, from later eating variability to cognitive performance and immune response. A first study should choose one primary endpoint, retain the others as clearly labelled exploratory analyses, and publish the protocol and analysis plan before data collection.

Related working papers

Read the relevant PDFs: EPK–001 to 007 are the revised version 1.2 papers; EPK–008 to 010 are version 1.0 supplements. Earlier PDFs and DOI records remain on the paper archive.