Impact of different training modalities on anthropometric outcomes in patients with obesity: A systematic review and network meta‐analysis

Summary Obesity management guidelines consistently advise aerobic training for weight loss, whereas recommendations for other training modalities are sparse. This systematic review and network meta‐analysis (NMA) aimed to compare the long‐term effects of different training modalities on anthropometric outcomes in patients with obesity. MEDLINE, Cochrane CENTRAL, and Web of Science were searched to identify the following: (1) randomized controlled trials (RCTs); (2) conducted in adults with a mean body mass index (BMI) ≥30 kg/m2; (3) comparing aerobic, resistance, combined, or high‐intensity interval training head‐to‐head or to control for ≥6 months; and (4) reporting changes in body weight (BW), BMI, waist circumference (WC), fat mass (FM), or fat‐free mass (FFM). Random‐effects NMA models were fitted in a frequentist approach. GRADE framework was used to assess certainty of evidence. Thirty‐two RCTs with 4774 participants with obesity were included in this review. Aerobic training was ranked as best for improving BW, BMI, and WC and combined training for improving FM, as well as equally with resistance training most effective for improving FFM. Low to moderate certainty of evidence supports use of aerobic training to improve anthropometric outcomes in obesity, and its combination with resistance training provides additional benefit for reducing FM and increasing FFM.


Content
Pages Table S1. Search strategy.
3 Table S2. Additional data received from authors. 4 Table S3. Reasons for risk of bias assessment judgement. 5 Table S4. Reasons for study full-text and study arms exclusion and supplementary references.

Reference
Data received Donnelly et al. 2013 Waist  Table S3. Reasons for risk of bias assessment judgement in the present systematic review.
Abbreviations: DXA -dual energy x-ray absorptiometry, ITT -intention-to-treat, mITT -modified intention-to-treat, MRImagnetic resonance imaging; NA -not applicable; Presented judgements are based on assessment algorithms available in the guidance document for the revised Cochrane risk-of-bias tool for randomized trials (RoB2).

Reasons for judgement
Bias arising from the randomization process Low risk Available description of random sequence generation and allocation concealment (in the manuscript or trial protocol) as well as no evidence for baseline imbalances.
Some concerns Missing description of allocation concealment, but no evidence for baseline imbalances.

High risk
Missing description of allocation concealment but there is an evidence that baseline imbalances were due to failure to proper random allocation of participants. Bias due to deviations from the intended interventions Low risk Study used ITT (or mITT) approach and reported high adherence of participants to exercise programs (i.e 80% of session completed).

Some concerns
Study used ITT (mITT) or per-protocol approach and reported moderate non-adherence of participants to exercise program (i.e. <80% of sessions completed).

High risk
Study used per-protocol approach in case of serious nonadherence or unwillingness to continue exercise program. Bias due to missing outcome data Low risk Study's drop-out rate ≤20%, with stated reasons, and dropouts were likely not dependent on their true value.

Some concerns
Study's drop-out rate >20%, but reasons for dropouts missing, and dropouts were likely not dependent on their true value.

High risk
Study's drop-out rate >20%, and dropouts were likely dependent on their true value (i.e., rates not equal between groups). Bias in the measurement of the outcome Low risk Anthropometrics and body composition assessed using recommended reliable methodology (i.e. DXA, MRI) Some concerns Anthropometrics and body composition assessed using less reliable methodology (i.e. bioimpedance analysis) High risk NA Bias in the selection of the reported result Low risk No evidence for results selection and pre-specified trial protocol available.
1 downgraded due to inconsistency (I 2 >50%); 2 downgraded due to risk of bias (approximately 1/3 of included RCTs rated with high risk of bias); 3 not downgraded due to incoherence (dominant estimate similar to network estimate); 4 direct evidence contributing more to the NMA estimate (>50%); 5 downgraded due to imprecision (95% CI overlaps important minimal important difference: -0.5 kg/m 2 ; and/or important harm: +0.5 kg/m 2 ); *Direct estimates were evaluated with the following GRADE criteria: risk of bias, indirectness, inconsistency and publication bias. As suggested recently by the GRADE working group, consideration of imprecision is not necessary when rating the direct and indirect estimates to inform the rating of NMA estimates.   AET -aerobic exercise training, CT -combined training, MI -minimal intervention, RT -resistance training, MD -mean difference, CI -confidence interval; ⨁⨁⨁⨁ High; ⨁⨁⨁◯ Moderate; ⨁⨁◯◯ Low; ⨁◯◯◯ Very low. Table S11. Relative ranking* of training effects on anthropometric outcomes.
*P-scores were calculated and presented to obtain relative ranking of training modalities. Higher P-score value indicates greater benefit (larger decrease or increase in outcome of interest) of certain intervention.
Ranking for all outcomes combined (assuming their equal importance) was obtained by taking an average of all P-scores for certain intervention.
Bolded are training modalities identified as the best for the given outcome.
AET -aerobic exercise training, CT -combined training, MI -minimal intervention, RT -resistance training; ↑ -increase is the effect of interest; ↑ -decrease is the effect of interest; Table S12. Results of node-splitting approach to assess inconsistency for anthropometric outcomes.
D -difference between direct and indirect estimates presented in Supplementary Tables 4-8; AET -aerobic exercise training, CT -combined training, MI -minimal intervention, RT -resistance training;  Figure S3. Net-heat plot* to assess inconsistency for body weight.
AET -aerobic exercise training, CT -combined training, MI -minimal intervention, RT -resistance training; *This plot is a heat map where the colours on the diagonal represent the inconsistency contribution of the corresponding design and the colours on the off-diagonal are associated with the change in inconsistency between direct and indirect evidence in a network estimate in the row after relaxing the consistency assumption for the effect of a design in the column. A blue coloured element indicates that the evidence of the design in the column supports the evidence in the row; a red coloured element indicates that the evidence of the design in the column contrasts to the evidence in the row. Figure S4. Net-heat plot to assess inconsistency for body mass index.
*This plot is a heat map where the colours on the diagonal represent the inconsistency contribution of the corresponding design and the colours on the off-diagonal are associated with the change in inconsistency between direct and indirect evidence in a network estimate in the row after relaxing the consistency assumption for the effect of a design in the column. A blue coloured element indicates that the evidence of the design in the column supports the evidence in the row; a red coloured element indicates that the evidence of the design in the column contrasts to the evidence in the row. Figure S5. Net-heat plot to assess inconsistency for waist circumference.
*This plot is a heat map where the colours on the diagonal represent the inconsistency contribution of the corresponding design and the colours on the off-diagonal are associated with the change in inconsistency between direct and indirect evidence in a network estimate in the row after relaxing the consistency assumption for the effect of a design in the column. A blue coloured element indicates that the evidence of the design in the column supports the evidence in the row; a red coloured element indicates that the evidence of the design in the column contrasts to the evidence in the row. Figure S6. Net-heat plot to assess inconsistency for fat mass.
*This plot is a heat map where the colours on the diagonal represent the inconsistency contribution of the corresponding design and the colours on the off-diagonal are associated with the change in inconsistency between direct and indirect evidence in a network estimate in the row after relaxing the consistency assumption for the effect of a design in the column. A blue coloured element indicates that the evidence of the design in the column supports the evidence in the row; a red coloured element indicates that the evidence of the design in the column contrasts to the evidence in the row. Figure S7. Net-heat plot to assess inconsistency for fat-free mass.
*This plot is a heat map where the colours on the diagonal represent the inconsistency contribution of the corresponding design and the colours on the off-diagonal are associated with the change in inconsistency between direct and indirect evidence in a network estimate in the row after relaxing the consistency assumption for the effect of a design in the column. A blue coloured element indicates that the evidence of the design in the column supports the evidence in the row; a red coloured element indicates that the evidence of the design in the column contrasts to the evidence in the row.      AET -aerobic exercise training, CT -combined training, MI -minimal intervention, RT -resistance training; AET -aerobic exercise training, CT -combined training, MI -minimal intervention, RT -resistance training; AET -aerobic exercise training, CT -combined training, MI -minimal intervention, RT -resistance training; AET -aerobic exercise training, CT -combined training, MI -minimal intervention, RT -resistance training; AET -aerobic exercise training, CT -combined training, MI -minimal intervention, RT -resistance training; AET -aerobic exercise training, CT -combined training, MI -minimal intervention, RT -resistance training;