Database searches identified 8747 potentially eligible published articles. Following title and abstract screening, 158 published articles underwent full-text screening, resulting in 17 published articles being deemed eligible. Due to several published articles containing multiple studies, a total of 23 separate studies were included in the analysis. Each outcome consisted of the following number of studies and participants––ad libitum meal energy intake: 14 studies, 217 participants; within-lab energy intake: 14 studies, 217 participants; 24-h energy intake: six studies, 83 participants; energy expenditure: eight studies, 69 participants; subjective hunger: 11 studies, 145 participants; acyl-ghrelin: three studies, 36 participants; glucagon-like peptide 1: three studies, 37 participants; peptide YY: two studies, 24 participants. The screening process is summarised in Fig. 2.
Flow diagram depicting process of paper selection.
Ad libitum meal energy intake, within-lab energy intake, 24-h energy intake, energy expenditure, and subjective hunger outcomes contained ≥5 studies and therefore a network meta-analysis was performed. Acyl-ghrelin, glucagon-like peptide 1 and peptide YY contained 3, 3, and 2 studies, respectively; therefore only descriptive data is provided.
Participant and intervention characteristics
Participant characteristics, intervention characteristic and reported outcomes for included studies are summarised in Table 1.
Of the 23 eligible studies, only six included females (five exclusively, one also including males) with the remaining 17 studies including only males. All studies were conducted in young individuals with mean participant age ranging from 11 to 28 years old, of which five were conducted in children (<18 years old; four exclusively, one also including adults). Seven studies included lean participants (BMI < 25 kg/m2), five studies included overweight participants (BMI ≥ 25 kg/m2), and eleven studies did not report BMI values.
Six studies compared FastEx + NoMeal to FedEx + NoMeal, six studies compared FastEx + Meal to FedEx + NoMeal, four studies compared FastEx + Meal to FedEx + Meal, six studies compared FedEx + Meal to FedEx + NoMeal and one study compared FastEx + Meal, with FastEx + NoMeal and FedEx + NoMeal.
The standardised meal provided before exercise typically contained foods associated with breakfast, with porridge (n = 7) and cereal with milk (n = 8) being the most popular options. The energy content of this standardised meal ranged from 733 kJ to 5,675 kJ (175–1,356 kcal), was provided between 20 and 180 min before exercise, and was typically high in carbohydrates (>50% of total energy; n = 16).
The standardised meal provided after exercise was typically a high-carbohydrate beverage (>50% of total energy; n = 13). The energy content of this standardised meal ranged from 529 kJ to 5675 kJ (126 kcal to 1356 kcal) and was given from immediately post-exercise to 60 min post-exercise.
Most studies used a treadmill (n = 18) for the exercise intervention, with the remaining studies using a cycle ergometer (n = 5). Exercise was largely conducted at a fixed intensity (n = 19), lasting between 40 and 60 min (n = 13), and performed at a moderate intensity (40–60% V̇O2 max; n = 14).
Risk of bias analysis
Risk of bias summary tables for each outcome is presented in Supplementary Appendix S2. For studies measuring ad libitum meal energy intake and within-lab energy intake, 50% of studies had a high risk of bias (Figure S1A). Similarly, 67% of studies measuring 24-h energy intake also had a high risk of bias (Figure S1B). Most studies measuring energy expenditure (75%) had an unclear risk of bias (Figure S1C), whereas the majority of studies measuring subjective hunger (64%) had a high risk of bias (Figure S1D). All studies measuring gastrointestinal hormones (acyl-ghrelin, glucagon-like peptide 1, peptide YY) had an unclear risk of bias (Figures S2,A–C).
Ad libitum meal energy intake
There were five studies comparing FastEx + NoMeal to FedEx + NoMeal, one study comparing FastEx + Meal to FedEx + NoMeal, two studies comparing FastEx + Meal to FedEx + Meal, and six studies comparing FedEx + Meal to FedEx + NoMeal. There were no studies comparing FastEx + NoMeal to FedEx + Meal or FastEx + NoMeal to FastEx + Meal (Figure S3,A). Effect estimates and confidence intervals for each study are presented in Figure S4.
Results from the network meta-analysis revealed that FedEx + Meal was ranked as the most effective intervention at reducing ad libitum meal energy intake (P-score = 0.970; Supplementary Appendix S3) and was significantly lower than FedEx + NoMeal (MD: −489 kJ; 95% CI, −898 to −80 kJ; P = 0.019; Fig. 3A). No significant differences in ad libitum meal energy intake were detected between FastEx + Meal and FedEx + NoMeal (MD: 7 kJ; 95% CI, −703 to 717 kJ; P = 0.985) or FastEx + NoMeal and FedEx + NoMeal (MD: 335 kJ; 95% CI, −173 to 843 kJ; P = 0.196). There was also no significant difference in ad libitum meal energy intake between FastEx + Meal and FedEx + Meal (MD: 496 kJ; 95% CI, −173 to 1165 kJ; P = 0.146; Supplementary Appendix S4).
Forest plots of effect estimates for (A) ad libitum meal energy intake, (B) within-lab energy intake, (C) 24-h energy intake, (D) energy expenditure, and (E) subjective hunger. Data are presented as mean differences (MD) ± 95% confidence intervals (CI). FastEx+Meal, fasted exercise with a standardised post-exercise meal; FastEx+NoMeal, fasted exercise without a standardised post-exercise meal; FedEx + Meal, fed exercise with a standardised post-exercise meal; FedEx+NoMeal, fed exercise without a standardised post-exercise meal.
Comparisons based on only indirect evidence showed that ad libitum energy intake was significantly higher during FastEx + NoMeal compared to FedEx + Meal (MD: 824 kJ; 95% CI, 172 to 1477 kJ; P = 0.013). Indirect evidence also showed no significant difference in ad libitum energy intake between FastEx + Meal and FastEx + NoMeal (MD: −328 kJ; 95% CI, −1201 to 545 kJ; P = 0.461; Supplementary Appendix S4).
Results from the SIDE approach showed no significant differences between direct and indirect evidence, suggesting no substantial level of local incoherence (Supplementary Appendix S5). There was however a moderate level of global incoherence (I2 = 56.5%). The comparison-adjusted funnel plot did not suggest evidence of publication bias (Figure S5,A), which was supported by results from Egger’s test (P = 0.383).
Within-lab energy intake
There were five studies comparing FastEx + NoMeal to FedEx + NoMeal, one study comparing FastEx + Meal to FedEx + NoMeal, two studies comparing FastEx + Meal to FedEx + Meal, and six studies comparing FedEx + Meal to FedEx + NoMeal. There were no studies comparing FastEx + NoMeal to FedEx + Meal or FastEx + NoMeal to FastEx + Meal (Figure S3,B). Effect estimates and confidence intervals for each study are presented in Figure S6.
Results from the network meta-analysis found that FastEx + NoMeal was the most effective intervention at reducing within-lab energy intake (P-score = 0.990; Supplementary Appendix S3) and was significantly lower than FedEx + NoMeal (MD: −1326 kJ; 95% CI, −2102 to −550 kJ; P = 0.001; Fig. 3B). There were no significant differences in within-lab energy intake between FastEx + Meal and FedEx + NoMeal (MD: −39 kJ; 95% CI, −1114 to 1036 kJ; P = 0.944) or FedEx + Meal and FedEx + NoMeal (MD: 389 kJ; 95% CI, −250 to 1029 kJ; P = 0.232). Similarly, within-lab energy intake did not significantly differ between FastEx + Meal and FedEx + Meal (MD: −428 kJ; 95% CI, −1441 to 584 kJ; P = 0.407; Supplementary Appendix S4).
Results based on only indirect evidence revealed that within-lab energy intake was significantly lower during FastEx + NoMeal compared to FedEx + Meal (MD: −1715 kJ; 95% CI, −2721 to −710 kJ; P = 0.001), but not significant difference between FastEx + Meal and FastEx + NoMeal (MD: 1287 kJ; 95% CI, −39 to 2613 kJ; P = 0.057; Supplementary Appendix S4).
Results from the SIDE approach showed no significant differences between direct and indirect evidence, suggesting no substantial level of local incoherence (Supplementary Appendix S5). Nevertheless, there was a high level of global incoherence (I2 = 81.9%). The comparison-adjusted funnel plot (Fig. S5,B) and the results from Egger’s test (P = 0.517) did not suggest evidence of publication bias.
24-h energy intake
Two studies compared FastEx + NoMeal to FedEx + NoMeal, one study compared FastEx + Meal to FedEx + NoMeal, two studies compared FastEx + Meal to FedEx + Meal, and one study compared FedEx + Meal and FedEx + NoMeal. There were no studies comparing FastEx + NoMeal to FedEx + Meal or FastEx + NoMeal to FastEx + Meal (Fig. S3,C). Effect estimates and confidence intervals for each study are presented in Figure S7.
Results from the network meta-analysis revealed that FastEx+NoMeal was the most effective intervention at reducing 24-h energy intake (P-score = 0.964; Supplementary Appendix S3) and was significantly lower than FedEx + NoMeal (MD: −2095 kJ; 95% CI, −3910 to −280; P = 0.024; Fig. 3 C). However, no significant differences in 24-h energy intake between FastEx + Meal and FedEx + NoMeal (MD: −161 kJ; 95% CI, −2120 to 1798 kJ; P = 0.872), FedEx + Meal and FedEx + NoMeal (MD: 680 kJ; 95% CI, −1194 to 2553 kJ; P = 0.477), or FastEx + Meal and FedEx + Meal (MD: −841 kJ; 95% CI, −2361 to 679 kJ; P = 0.278; Supplementary Appendix S4) were detected.
Comparisons based on only indirect evidence found that 24-h energy intake was significantly lower during FastEx + NoMeal compared to FedEx + Meal (MD: −2775 kJ; 95% CI, −5383 to −166 kJ; P = 0.037). However, indirect evidence revealed no significant difference in 24-h energy intake between FastEx + Meal and FastEx + NoMeal (MD: 1934 kJ; −736 kJ to 4605 kJ; P = 0.156; Supplementary Appendix S4).
Results from the SIDE approach showed no significant differences between direct and indirect evidence, suggesting no substantial level of local incoherence (Supplementary Appendix S5). There was however a moderate level of global incoherence (I2 = 63.5%). The comparison-adjusted funnel plot did not suggest evidence of publication bias (Figure S5,C). Egger’s test was not performed because of the small number of studies reporting 24-h energy intake (n = 6).
Energy expenditure
One study compared FastEx + NoMeal to FedEx + NoMeal, six studies compared FastEx + Meal to FedEx + NoMeal, one study compared FastEx + Meal to FedEx + Meal, and one study compared FastEx + NoMeal to FastEx + Meal. There were no studies comparing FastEx + NoMeal to FedEx + Meal or FedEx + Meal to FedEx + NoMeal (Fig. S3,D). Effect estimates and confidence intervals for each study are presented in Figure S8.
The results of the network meta-analysis revealed that FastEx + Meal is the most effective intervention at increasing energy expenditure (P-Score = 0.744; Supplementary Appendix S3) but was not significantly different from FedEx + NoMeal (MD: 0.07 kJ/min; 95% CI, −0.15 to 0.30 kJ/min; P = 0.529; Fig. 3D) or FedEx + Meal (MD: 0.00 kJ/min; 95% CI, −0.70 to 0.69 kJ/min; P = 0.992). Energy expenditure during FastEx + NoMeal was however significantly lower than during both FedEx + NoMeal (MD: −0.67 kJ/min; 95% CI, −1.10 to −0.23 kJ/min; P = 0.003) and FastEx + Meal (MD: −0.74 kJ/min; 95% CI, −1.18 to −0.31 kJ/min; P < 0.001; Supplementary Appendix S4).
Comparisons based on only indirect evidence showed that energy expenditure was not significantly different between FastEx + NoMeal and FedEx + Meal (MD: −0.75 kJ/min; 95% CI, −1.57 to 0.08 kJ/min; P = 0.075) or FedEx + Meal and FedEx + NoMeal (MD: 0.08 kJ/min; 95% CI, −0.66 to 0.81 kJ/min; P = 0.837; Supplementary Appendix S4).
Results from the SIDE approach showed no significant differences between direct and indirect evidence, suggesting no substantial level of local incoherence (Supplementary Appendix S5). There was however a high degree of global incoherence (I2 = 81.3%). There was no evidence of publication bias following a visual inspection of the comparison-adjusted funnel plot (Fig. S5,D). Egger’s test was not performed due to the small number of studies assessing energy expenditure (n = 8).
Subjective hunger
Four studies compared FastEx + NoMeal to FedEx + NoMeal, two studies compared FastEx + Meal to FedEx + NoMeal, three studies compared FastEx + Meal to FedEx + Meal, and two studies compared FedEx + Meal to FedEx + NoMeal. There were no studies comparing FastEx + NoMeal to FedEx + Meal or FastEx + NoMeal to FastEx + Meal (Figure S3,E). Effect estimates and confidence intervals for each study are presented in Figure S9.
Results from the network meta-analyses found that FedEx + Meal was the most effective intervention at reducing subjective hunger (P-score = 0.980; Supplementary Appendix S3) but was not significantly different from FedEx + NoMeal (MD: −6 mm; 95% CI, −14 to 2 mm; P = 0.119; Fig. 3E). Compared to FedEx + NoMeal, subjective hunger was significantly higher during both FastEx + Meal (MD: 13 mm; 95% CI, 5 to 21 mm; P = 0.001) and FastEx + NoMeal (MD: 23 mm; 95% CI, 16 to 30 mm; P < 0.001). Likewise, subjective hunger was significantly higher during FastEx + Meal compared to FedEx + Meal (MD: 19 mm; 95% CI, 12 to 26 mm; P < 0.001; Supplementary Appendix S4).
Results based on only indirect evidence revealed that subjective hunger was significantly higher during FastEx + NoMeal compared to FedEx + Meal (MD: 29 mm; 95% CI, 18 to 39 mm; P < 0.001). However, indirect evidence found no significant difference between FastEx + Meal and FastEx + NoMeal (MD: −10 mm; −20 to 0 mm; P = 0.062; Supplementary Appendix S4)
Results from the SIDE approach showed no significant differences between direct and indirect evidence, suggesting no substantial level of local incoherence (Supplementary Appendix S5). Global incoherence was high (I2 = 76.5%) and visual inspection of the comparison-adjusted funnel plot did not indicate publication bias (Fig. S5,D). Egger’s test also suggested that publication bias was not present (P = 0.644).
Gastrointestinal hormones
All three studies measuring acyl-ghrelin concentrations compared FastEx + NoMeal to FedEx + NoMeal. For glucagon-like peptide 1, two studies compared FastEx + Meal to FedEx + Meal and one study compared FedEx + Meal to FedEx + NoMeal. For peptide YY, one study compared FastEx + Meal to FedEx + NoMeal, and one other study compared FastEx + Meal to FedEx + Meal. Network meta-analysis were therefore not performed for these outcomes and only individual study effects are presented (Fig. S10–S12).
Confidence in evidence
Confidence in evidence for all effect estimates is displayed in Table 2. Full details of the evaluation are provided in Supplementary Appendix S6.