Table 4. Summary of GRADE assessment for the certainty in depressive symptoms estimates.
| Comparison Effect | Number of
participants |
Number of direct
comparisons |
Nature of
evidence |
Certainty | Reason for
downgrading |
|---|---|---|---|---|---|
| Aerobic vs. wait-list | 371 vs. 357 | 11 | Mixed | High | None |
| Aerobic vs. usual care | 309 vs. 310 | 9 | Mixed | Moderate | Risk of bias a |
| Aerobic vs. attention-control | 431 vs. 472 | 8 | Mixed | High | None |
| Resistance vs. wait-list | 43 vs. 42 | 2 | Mixed | Moderate | Imprecision b |
| Resistance vs. usual care | 358 vs. 272 | 15 | Mixed | High | None |
| Resistance vs. attention-control | 267 vs. 274 | 5 | Mixed | Moderate | Risk of bias c |
| Mind-body vs. wait-list | 380 vs. 363 | 12 | Mixed | High | None |
| Mind-body vs. usual care | 358 vs. 356 | 10 | Mixed | Moderate | Risk of bias d |
| Mind-body vs. attention-control | 298 vs. 265 | 10 | Mixed | High | None |
| Aerobic vs. resistance | 24 vs. 23 | 1 | Mixed | Low | Imprecision b e |
| Aerobic vs. mind-body | 90 vs. 83 | 4 | Mixed | Moderate | Imprecision e |
| Resistance vs. mind-body | 33 vs. 33 | 1 | Mixed | Low | Imprecision b e |
aPotential attrition bias due to high number of studies with incomplete outcome data. bSmall sample size. cPotential risk of bias due to high number of studies with low adherence. dPotential detection bias due to high number of studies without blinding of outcome assessment. eConfidence intervals include values favouring either treatment.