TABLE 7.
Key Outcomes for Comparison 2: ICM Versus ECC
| Outcomes | No. Participants (Studies) Follow-up | Certainty of the Evidence | Relative Effect (95% CI) | RD/MD (95% CI) | I2, % |
|---|---|---|---|---|---|
| Neonatal outcomes | |||||
| Survival to discharge from hospital | 945 (10 RCTs) | ⊕⊕⊕⊝ Moderatea,b,c | RR: 1.02 (0.98 to 1.06) | RD: 0.02 (−0.01 to 0.05) | 24 |
| Severe IVH: ultrasound diagnosis grades III, IV | 889 (10 RCTs) | ⊕⊕⊝⊝ Lowd,e | RR: 0.72 (0.44 to 1.19) | RD: −0.02 (−0.05 to 0.01) | 0 |
| CLD: oxygen at 36 wk PMA | 685 (7 RCTs) | ⊕⊕⊝⊝ Lowf,g | RR: 1.02 (0.63 to 1.65) | RD: −0.02 (−0.12 to 0.08) | 60 |
| NEC (Bell’s stage ≥II or any grade47; requiring surgery) | 843 (9 RCTs) | ⊕⊕⊕⊝ Moderatea,h | RR: 0.80 (0.55 to 1.18) | RD: (−0.06 to 0.02) | 0 |
| Peak Hb concentrations within the first 24 h after birth | 914 (10 RCTs) | ⊕⊕⊕⊝ Moderatei,j | Continuous outcome | MD: 1.18 (0.65 to 1.71) | 71 |
| Peak Hct within the first 24 h after birth | 774 (7 RCTs) | ⊕⊕⊕⊝ Moderatei,j | Continuous outcome | MD: 3.04 (1.28 to 4.80) | 69 |
| Peak Hb concentrations within 7 d after birth | 54 (1 RCT) | ⊕⊕⊝⊝ Lowa,k,l,m | Continuous outcome | MD: 0.60 (−0.57 to 1.77) | Not estimable |
| Peak Hct within 7 d after birth | 54 (1 RCT) | ⊕⊕⊝⊝ Lowa,k,l,m | Continuous outcome | MD: 1.00 (−2.32 to 4.32) | Not estimable |
| Hyperbilirubinemia (treated by phototherapy) | 480 (5 RCTs) | ⊕⊕⊕⊕ Higha | RR: 1.04 (0.94 to 1.16) | RD: 0.03 (−0.04 to 0.10) | 10 |
| Infant outcomes | |||||
| Moderate to severe neurodevelopmental impairment in early childhood | 26 (1 RCT) | ⊕⊝⊝⊝ Very lown,o,p,q | RR: 0.75 (0.21 to 2.71) | RD: −0.08 (−0.42 to 0.26) | Not estimable |
| Cerebral palsy in early childhood | 161 (1 RCT) | ⊕⊝⊝⊝Very lown,o,p,q | RR: 2.65 (0.88 to 7.97) | RD: 0.08 (−0.00 to 0.17) | Not estimable |
| Significant mental developmental delay in early childhood | 0 (0 studies) | — | Not estimable | Not estimable | Not estimable |
| Legal blindness in early childhood (<20/200 visual acuity) | 0 (0 studies) | — | Not estimable | Not estimable | Not estimable |
| Hearing deficit in early childhood (aided or <60 dB on audiometric testing) | 0 (0 studies) | — | Not estimable | Not estimable | Not estimable |
| Maternal outcomes | |||||
| PPH (clinically estimated blood loss of ≥ 500 mL) | 0 (0 studies) | — | Not estimable | Not estimable | Not estimable |
| Maternal death or severe morbidity | 0 (0 studies) | — | Not estimable | Not estimable | Not estimable |
| Severe PPH (blood loss ≥1000 mL) | 266 (2 RCTs) | ⊕⊝⊝⊝ Very lowr,s | RR: 2.83 (0.12 to 67.01) | RD: 0.01 (−0.02 to 0.03) | Not estimable |
| Use of therapeutic uterotonic agents | 0 (0 studies) | — | Not estimable | Not estimable | Not estimable |
| Blood transfusion (maternal) | 66 (1 RCT) | ⊕⊝⊝⊝ Very lowr,t,u | RR: 2.83 (0.12 to 67.01) | RD: 0.03 (−0.05 to 0.11) | Not estimable |
| Manual removal of the placenta | 0 (0 studies) | — | Not estimable | Not estimable | Not estimable |
| Additional treatment of PPH (uterine tamponade, embolization) | 0 (0 studies) | — | Not estimable | Not estimable | Not estimable |
| Postpartum infection | 0 (0 studies) | — | Not estimable | Not estimable | Not estimable |
CLD, chronic lung disease; Hb, hemoglobin; Hct, hematocrit; PMA, postmenstrual age; PPH, postpartum hemorrhage; RD, risk difference; —, not applicable; ⊕, positive; ⊝, negative.
No downgrade despite concerns due to blinding of intervention and selective outcome reporting bias. Blinding less likely to affect this outcome. Selective outcome reporting bias less likely to affect this estimate because this is a null result. This is a borderline decision.
Some inconsistency, but not sufficient to downgrade. I2 = 24%.
Effect ranges from clinically important reduction to clinically important increase of survival. Downgrade by 1 for imprecision.
Effect ranges from clinically important reduction to clinically important increase. Low events and low participants. Downgrade by 2 for imprecision.
No downgrade despite concerns because of blinding of intervention and selective outcome reporting bias. Blinding less likely to affect this outcome. Biggest and majority of studies were blinded for outcome assessment. Selective outcome reporting bias less likely to affect this estimate because this is a null result. This is a borderline decision.
Effect ranges from clinically important reduction to clinically important increase. Downgrade by 1 for imprecision.
Moderate heterogeneity downgrade by 1 for inconsistency.
Wide CI and relatively low event rates. Downgrade by 1 for imprecision.
No downgrade despite concerns due to blinding of intervention and selective outcome reporting bias. Blinding less likely to affect this outcome. This is a borderline decision.
Substantial heterogeneity, all but one effect estimates point in the same direction. Downgrade by 1 for inconsistency.
Unable to assess inconsistency (only 1 study). No downgrade.
Only 1 small study, wide CI. Downgrade by 1 for imprecision.
Only 1 single-center study impairs generalizability. Downgrade by 1 for indirectness.
No downgrade despite concerns due to blinding of intervention and selective outcome reporting bias. Blinding less likely to affect this outcome. Selective outcome reporting bias less likely to affect this estimate because this is a null result. This is a borderline decision.
Unable to assess inconsistency (only 1 study). No downgrade.
Only 1 small study, low event numbers, very wide CI. Downgrade by 2 for imprecision.
Only 1 single-center study impairs generalizability. Downgrade by 1 for indirectness.
Very wide CI, only 1 event. Downgrade by 2 for imprecision.
All studies unblinded for intervention and outcome assessment. Subjective outcome, may have been influenced by lack of blinding. Downgrade by 1 for risk of bias.
No downgrade despite concerns due to blinding of intervention and selective outcome reporting bias. Blinding less likely to affect this outcome. Selective outcome reporting bias less likely to affect this estimate because this is a null result. This is a borderline decision.
Only 1 single-center study, this impairs generalizability. Downgrade by 1 for indirectness.