REF-176 · Iannetta et al., 2021 · Sports Medicine, · Paper

Evaluating the Accuracy of Using Fixed Ranges of METs to Categorize Exertional Intensity in a Heterogeneous Group of Healthy Individuals: Implications for Cardiorespiratory Fitness and Health Outcomes

B+moderate evidenceMeasured on recreational peopleWritten before October 2026, being re-checked

Iannetta et al. compared the common fixed-MET intensity classification (light / moderate / vigorous / near-maximal) with an individualised schema based on the exercise intensity domains (moderate / heavy / severe) in a large heterogeneous sample of healthy adults.

Each participant did a ramp-incremental test from which the gas exchange threshold (GET), respiratory compensation point (RCP) and VO2max were determined; the domains were then expressed in METs for each person (moderate = METs ≤ GET; heavy = METs > GET but ≤ RCP; severe = METs > RCP). Result: a critical discrepancy between domain boundaries and fixed MET categories.

Across individuals the upper limit of the moderate domain ranged from 2 to 13 METs and of the heavy domain from 3 to 18 METs, while the severe domain began anywhere from 4 METs upward. Assigning fixed MET values to intensity categories therefore risks misclassifying the physiological stress of exercise and misreading dose-response relationships.

Participants

n=565 healthy individuals (165 females, 400 males), age range 18-83 years, heterogeneous fitness
Ramp-incremental exercise test; GET, RCP and VO2max determined per person

Protocol

Cross-sectional study: 565 healthy individuals (165 females, 400 males; age range 18-83 years) (from the abstract).
Each performed a ramp-incremental exercise test; gas exchange threshold (GET), respiratory compensation point (RCP) and VO2max were used to build an individual domain schema in METs (moderate ≤ GET; heavy > GET but ≤ RCP; severe > RCP).
This schema was compared with the fixed-MET classification (light, moderate, vigorous, near-maximal) by Pearson's chi-square tests over contingency tables at each MET value; a multi-level regression model looked for predictors of domain amplitude.

Key Findings (key figures)

Upper limit of the moderate domain: 2 to 13 METs across individuals
Upper limit of the heavy domain: 3 to 18 METs across individuals
Severe domain: from 4 METs onward in some individuals
Statistics: chi-square over contingency tables per MET value; multi-level regression for predictors of domain amplitude
The abstract reports no percentage misclassification rate and no "error" figure for individualised classification.

Practical Application

In 565 healthy adults aged 18-83, the same MET value fell in different intensity domains for different people: the top of the moderate domain ranged from 2 to 13 METs and the top of the heavy domain from 3 to 18 METs, so a fixed MET band (e.g. "vigorous") cannot say which domain an individual is in (cross-sectional, Iannetta 2021).
Do not build training zones from MET tables or generic activity-intensity categories; the domain an athlete sits in is fixed by their own GET/VT1 and RCP/VT2, not by an absolute metabolic rate.
Inference, not tested here: for a fit athlete a "vigorous" MET value may still be moderate-domain work, and for an untrained person a "moderate" MET value may already be heavy or severe — this is the direction the reported ranges imply, though the abstract does not give per-fitness-group figures.
Not from this paper; the base's own practice: public-health prescriptions in minutes of "moderate" activity are treated as health floors, not as training prescription for a target event.

Limitations

From the design as the abstract reports it: cross-sectional, one ramp test per person — no intervention and no training or health outcome was measured.
Healthy adults with more males than females (400 vs 165); the abstract gives no figures by fitness level, sex or age group.
The individual domains rest on GET and RCP taken from a single incremental test.
How this entry was checked and changed
Why this level

regraded 2026-09-27 from A: cross-sectional measurement study with ≥100 participants, no intervention

Content correction 2026-09-27
"Misclassification rate 30-60% depending on fitness" and "individualized classification <5% error" deleted: neither figure is in the abstract.
Added the abstract's actual ranges (moderate upper limit 2-13 METs, heavy 3-18 METs, severe from 4 METs) and the domain definitions used.
Trained/untrained direction statements relabelled as inference; "150 min" guideline figure removed.
Level A -> B+ (cross-sectional measurement study, n=565).
2026-09-28: added Protocol and Limitations from the abstract (Europe PMC PMID 33900580).