Optimal exercise dose for glycemic control in prediabetes across different exercise types
Question
Which exercise modality and weekly exercise dose best reduces HbA1c in adults with prediabetes, and how do baseline BMI and exercise type modify minimum effective and optimal doses?
Summary
This systematic review and Bayesian network and dose-response meta-analysis of 39 randomized trials in adults with prediabetes found that exercise interventions reduced HbA1c compared with control conditions. HIIT had the largest estimated HbA1c reduction and reached a clinically meaningful dose threshold at 350 METs.min/week, but the certainty for HIIT was very low, so the true effect remains uncertain.
Methodology
- Adults with prediabetes included in 39 randomized controlled trials.
- 3421 participants.
- Multiple exercise modalities across included RCTs.
- Intensity: Weekly dose quantified as METs.min/week using 2024 Adult Compendium values; HIIT category minimum clinically meaningful dose 350 METs.min/week and optimal dose 850 METs.min/week.
- At least 4 weeks in included RCTs.
- Systematic review with pairwise meta-analysis, Bayesian network meta-analysis, Bayesian dose-response meta-analysis, meta-regression, sensitivity analyses, risk-of-bias assessment, and CINeMA certainty assessment.
- HbA1c, Adherence, and Adverse events were tracked.
Outcomes
HbA1c, network meta-analysis
All exercise modalities reduced HbA1c compared with usual care; HIIT had the largest estimated effect but very low certainty.
HIIT MD -0.50%, 95% CrI -0.62 to -0.36, SUCRA 99%; CT MD -0.34%, AT MD -0.28%, RT MD -0.28%, MBT MD -0.21%.
HbA1c, pairwise meta-analysis
HIIT significantly reduced HbA1c compared with controls in direct pairwise analysis.
HIIT MD -0.78%, 95% CI -1.16 to -0.40; CT -0.36%, AT -0.28%, RT -0.30%, MBT -0.19%.
Dose-response relationship
A nonlinear J-shaped dose-response was observed, with estimated overall optimal dose of 850 METs.min/week.
After BMI adjustment at 850 METs.min/week: obesity MD -0.339%, 95% CrI -0.440 to -0.238; normal weight and overweight statistically significant but not clinically meaningful by MCID.
HIIT dose-response
HIIT produced clinically meaningful reductions at lower dose than other modalities in unadjusted modality analysis.
HIIT minimum dose 350 METs.min/week, MCFB -0.306%, 95% CrI -0.371 to -0.240; optimal dose 850 METs.min/week, MCFB -0.52%, 95% CrI -0.598 to -0.443.
Adherence
No statistically significant adherence differences were found between exercise interventions and control.
HIIT OR 0.91, 95% CrI 0.15 to 5.1; CT OR 2.5, 95% CrI 0.49 to 13; RT OR 2.2, AT OR 1.6, MBT OR 1.1.
Adverse events
Reported events were low-incidence and mostly mild to moderate, but HIIT had seven reported events among 94 participants in two studies.
AT: 4 events/135 participants; HIIT: 7 events/94 participants; RT: 3 events/42 participants among studies reporting events.
Insights
- HIIT may be promising for HbA1c reduction in prediabetes, but the certainty is very low and should be communicated cautiously.
- For glycemic-control programming, weekly dose may need to account for MET intensity, duration, frequency, and BMI rather than minutes alone.
- The review supports content about an overall estimated optimum around 850 METs.min/week, while noting that exact workout structures are not specified.
- Safety messaging should reflect limited adverse-event reporting and the clinical nature of prediabetes populations.
Limitations
- 79.5% of included studies were conducted in Asian populations, limiting global generalizability.
- Dose was estimated using standardized MET compendium values rather than individual objective energy expenditure.
- Limited evidence for BMI >=32 kg/m2.
- HIIT evidence had very low certainty and local inconsistency in some comparisons.
- Protocol-level work-rest details for HIIT were not available in the main article.
Safety
- AT reported four adverse events across three studies among 135 participants: sprain, cardiovascular event, fall, and hypoglycaemia.
- HIIT reported seven adverse events across two studies among 94 participants: six soft tissue injuries, one cardiovascular event, and one knee pain case.
- RT reported three adverse events in one study among 42 participants: two muscle strains and one hypoglycaemia case.
- Authors characterized reported adverse events as primarily mild to moderate with low incidence.