Table 2

Quality assessment and summary of findings for the main comparison

Quality assessmentNo of patientsEffectQualityImportance
No of studiesStudy designRisk of biasInconsistencyIndirectnessImprecisionOther considerationsEarly rehabilitationControlRelative
(95% CI)
Absolute
(95% CI)
Incidence of ICU-AW
2Randomised trialsSerious*Not seriousNot seriousSerious†None22/78 (28.2%)37/76 (48.7%)OR 0.42
(0.22 to 0.82)
202 fewer per 1000
(from 49 fewer to 314 fewer)
⨁⨁◯◯
Low
Critical
MRC
3Randomised trialsSerious*Not seriousNot seriousSerious†None9799SMD 0.38 higher
(0.1 higher to 0.66 higher)
⨁⨁◯◯
Low
Critical
Delirium-free days
3Randomised trialsSerious*Not seriousNot seriousSerious†None164162SMD 0.02 lower
(0.23 lower to 0.2 higher)
⨁⨁◯◯
Low
Critical
Incidence of HADS/HAS
2Randomised trialsSerious*SeriousNot seriousSerious†None10/48 (20.8%)11/44 (25.0%)OR 0.79
(0.29 to 2.12)
42 fewer per 1000
(from 162 fewer to 164 more)
⨁◯◯◯
Very low
Critical
Health related QOL score (EQ5D)
2Randomised trialsSerious*Serious‡Not seriousSerious†None3528SMD 0.11 SD higher
(0.86 lower to 1.09 higher)
⨁◯◯◯
Very low
Critical
SF-36 PF
2Randomised trialsSerious*Not seriousNot seriousSerious†None9398SMD 2.41 higher
(0.75 lower to 5.58 higher)
⨁⨁◯◯
Low
Critical
  • *Most of the studies are classified as unclear or high risk of bias (performance bias and attrition bias).

  • †CIs for absolute effects are wide and include no effect and important harm and/or small sample size (the number of events is <300).

  • ‡Substantial heterogeneity present.

  • AW, acquired weakness; EQ5D, EuroQol 5 Dimensions; HADS/HAS, Hospital Anxiety and Depression Scale; ICU, intensive care unit; MRC, Medical Research Council; QOL, quality of life; SF-36 PF, 36-Item Short Form Health Survey Physical Function scale; SMD, standardised mean difference.