. 2020 Jul 7;11:606. doi: 10.3389/fphys.2020.00606

Table 4.

Meta-analysed effects on homeostatic model assessment of insulin resistance (HOMA-IR) and free androgen index (FAI) expressed as population mean effects in control and exercise groups, and as modifying effects of exercise duration, baseline, and dietary co-intervention.

	HOMA-IR^b			FAI^c
	Mean (%)	90% CL (%)	Magnitude	Mean (%)	90% CL (%)	Magnitude
Population mean effects^a
Control group	32.4	1.3, 72.9	Small↑^**	−2.9	−11.1, 6.1	Trivial⁰⁰⁰
Moderate exercise	10.1	−6.7, 30.0	Small↑^*	2.2	−6.7, 12.0	Trivial⁰⁰⁰
Vigorous exercise	−15.6	−33.2, 6.7	Small↓^*	2.4	−7.4, 13.2	Trvial⁰⁰
Resistance exercise	−1.0	−14.4, 14.5	Trivial	−15.3	−28.4, 0.2	Small↓^*
Moderate—control group	−16.8	−38.8, 13.1	Small↓^*	5.2	−6.2, 18.0	Trivial⁰⁰
Vigorous—control group	−36.2	−55.3, −9.0	Moderate↓^**	5.4	−6.1, 18.4	Trivial⁰⁰
Resistance—control group	−25.2	−44.4, 0.6	Moderate↓^**	−12.3	−27.3, 4.6	Small↓^*
Modifying effects
Baseline in control group	43.9	11.7, 85.4	Moderate↑^**	7.5	−11.5, 30.6	Small↑^*
Baseline in exercise group	13.1	−25.1, 70.9	Trivial	1.1	−15.1, 20.3	Trivial
30 h of exercise duration	−5.4	−35.6, 38.9	Trivial	−8.1	−20.3, 6.0	Small↓^*
Diet in control group	−43.1	−58.9, −21.3	Small↓^***	−21.9	−32.8, −9.2	Small↓^**
Diet in exercise group	−19.5	−44.5, 16.6	Trivial	−11.1	−20.6, −0.4	Small↓^*

Evaluated at mean baseline (HOMA-IR=moderate, FAI = 8.4%), training time = 30 h, and no dietary co-intervention.

Modifying effect of baseline is evaluated for high versus low baseline.

Modifying effect of baseline is evaluated for a 3.0-fold difference in baseline.

90% CL 90% compatibility limits, ↑ increase, ↓ decrease.

Effects are shown with their observed magnitudes, determined by standardisation. Clear effects are show with the probability of either a true substantial change (^*possibly, ^**likely, ^***very likely) and/or a true trivial change (⁰⁰likely, ⁰⁰⁰very likely). Magnitudes in bold are clear with 99% compatibility intervals.