REF-1310 · Reljic et al., 2019 · Scandinavian Journal of Medicine & Science in Sports · Paper · PubMed

Will intervals make a beginner quit?

In 1318 sedentary adults across 67 HIIT programmes about 18% dropped out; how hard the intervals were did not predict quitting, longer sessions did. Between-study data.

Main caveat: Single-arm pooling. The 17.6% is a pooled proportion across HIIT arms; it is not a comparison with continuous training in the same studies.
fairly strong evidenceMeasured on non-athletes

Reljic and colleagues asked how many sedentary people drop out of HIIT-based exercise programmes and what predicts it.

They searched MEDLINE/PubMed, SPORTDiscus and Web of Science to June 2018 for English-language studies in sedentary healthy adults that used HIIT without any other intervention, lasted at least 4 weeks, described the HIIT protocol in detail and allowed a dropout rate to be calculated. Fifty-five studies with 67 HIIT interventions and 1318 participants qualified.

The trim-and-fill adjusted pooled dropout rate was 17.6% (95% CI 14.2-21.5%). Dropout was significantly lower in cycling-based than in running- or walking-based interventions (P < 0.001).

Longer session time (β = 0.02), more time per week (β = 0.005) and more overall time per intervention (β = 0.0003) predicted greater dropout (each P < 0.05); exercise intensity was not related to dropout.

The authors conclude that HIIT is tolerable and acceptable for previously sedentary people, with generally lower dropout than commonly reported for traditional programmes, and that the minimal effective dose deserves study.

Our reading: the time a programme demands, not the hardness of its intervals, went with quitting in these studies — an association across studies, in supervised-style research settings; no head-to-head comparison with continuous training was pooled here.

Protocol

Search: MEDLINE/PubMed, SPORTDiscus, Web of Science, systematically, until 06/2018.
Inclusion: (a) English; (b) sedentary healthy adults; (c) some form of HIIT without any complementary intervention; (d) at least 4 weeks; (e) a detailed description of the HIIT protocol; (f) data that allow a dropout rate to be calculated.
Included: 55 studies, 67 HIIT interventions, n = 1318 participants.
Analysis: pooled dropout rate with trim-and-fill adjustment; comparison by exercise modality; meta-regression of dropout on session time, time effort per week, overall time effort per intervention and exercise intensity. The units of the β coefficients, the definition of dropout and the intensity metric are not given in the abstract.
Funding and interests: not given in the abstract (full text not read); the first affiliation is the Hector-Center for Nutrition, Exercise and Sports, University Hospital Erlangen.

Key Results

OutcomeResult
Pooled dropout rate, all 67 interventions (trim-and-fill adjusted)17.6% (95% CI 14.2-21.5%)
Modalitydropout significantly lower with cycling than with running/walking (P < 0.001)
Session timelonger sessions predicted greater dropout (β = 0.02, P < 0.05)
Time effort per weekmore predicted greater dropout (β = 0.005, P < 0.05)
Overall time effort per interventionmore predicted greater dropout (β = 0.0003, P < 0.05)
Exercise intensitynot related to dropout
Comparison with traditional programmes: the abstract says HIIT dropout was "generally lower" than commonly reported for traditional exercise programmes; that comparison is with other literature, not pooled or tested here.
Dropout rates by modality (the percentages) are not given in the abstract.

Mechanisms

Measured: between-study associations of dropout with programme time, modality and intensity.
Authors' interpretation: HIIT volume, not intensity, drives dropout; finding the minimally effective dose of practical HIIT would aid implementation and public-health impact.
Inference: "intensity not related" is a between-study result across protocols that all counted as HIIT; it does not show that any intensity is tolerable for any beginner, and studies with harder protocols may have recruited keener volunteers.
Inference: the running/walking vs cycling difference may partly reflect who was studied with each mode and the impact load of running for a sedentary body; the abstract does not separate these.

Practical Application

Will intervals make a beginner quit? In 1318 sedentary adults across 67 HIIT programmes about 18% dropped out; how hard the intervals were did not predict quitting, longer sessions did. Between-study data.
Keep a beginner's hard session short (A- for the dropout pool; the predictor is a between-study association): longer sessions and more weekly HIIT time went with more dropout, and the VO2max gain from intervals in inactive adults flattens at short sessions anyway (REF-749, A).Not from this paper; the base's own practice: at most one short interval session a week, optional, in weeks 5-12, and for runners only once they run 20-30 min continuously without symptoms (SYNTH-75).
Intensity early is not what makes beginners quit in these studies, so the base does not need a ban on intervals for a beginner; what it guards is the dose of time and, for runners, the running load itself (SYNTH-75; tolerance and injury evidence in REF-1307, REF-1312).
For a sedentary beginner, put intervals on the bike before the run (inference): running- or walking-based HIIT lost more people than cycling here, and new running load carries its own tissue cost (REF-1315).
Real-world adherence is likely lower than in research programmes (inference): these are volunteers in studies; novice runners in a community programme stopped at about 29% within six months, half of them because of injury (REF-1311).

Limitations

Single-arm pooling. The 17.6% is a pooled proportion across HIIT arms; it is not a comparison with continuous training in the same studies.
Predictors are ecological. Session length, modality and intensity were compared across studies, not randomised within them; confounding by population and setting is likely.
Read from the abstract. Heterogeneity, study quality, the dropout definition, supervision, the intensity metric and the units of the regression coefficients are not available.
Population and setting. Sedentary healthy adults in research programmes of at least 4 weeks; not athletes, not unsupervised everyday training.
Funding and interests: not in the abstract.
How this entry was checked and changed
Why this level

(from the abstract) — a systematic review with meta-analysis (rubric A) held at A- because the studies are mostly small, the pooled figure is a single-arm proportion from research programmes rather than a comparison with another kind of training, and the predictors (intensity, session length, modality) come from between-study meta-regression, which is observational and carries about B weight within this entry; heterogeneity and study quality are not reported in the abstract (the full text is not open)