Table 2.
GRADE summary of quality of evidence for primary outcomes: tranexamic acid vs. control treatment in aneurysmal subarachnoid hemorrhage.
| Quality assessment | No of patients | Effect | Quality | Importance | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| No of studies | Design | Risk of bias |
In consis tency |
In direct ness |
Impre cision |
Other conside rations |
TXA | Control | Relative (95% CI) | Absolute | ||
| All-cause mortality | ||||||||||||
| 12 | RCTs | Seriousb | not serious | not serious | Seriousc | none | 417/1702 (24.5%) | 426/1679 (25.4%) | RR 0.97 (0.86 to 1.08) | 8 fewer per 1,000 (from 36 fewer to 20 more) | ⊕⊕○○ LOW | CRITICAL |
| Poor outcome | ||||||||||||
| 5 | RCTs | seriousb | not serious | not serious | Seriousc | none | 503/1249 (40.3%) | 481/1242 (38.7%) | RR 1.04 (0.95 to 1.15) | 15 more per 1,000 (from 19 fewer to 58 more) | ⊕⊕○○ LOW | CRITICAL |
| Rebleeding | ||||||||||||
| 11 | RCTs | serious | Seriousa | not serious | not serious | none | 162/1423 (11.4%) | 284/1416 (20.1%) | RR 0.59 (0.43 to 0.82) | 82 fewer per 1,000 (from 36 fewer to 114 fewer) | ⊕⊕○○ LOW | CRITICAL |
| Cerebral ischemia | ||||||||||||
| 8 | RCTs | Seriousb | Seriousa | not serious | Seriousc | none | 384/1328 (28.9%) | 340/1318 (25.8%) | RR 1.17 (0.95 to 1.46) | 44 more per 1,000 (from 13 fewer to 119 more) | ⊕○○○ VERY LOW | IMPORTANT |
| Hydrocephalus | ||||||||||||
| 7 | RCTs | Seriousb | not serious | not serious | Seriousc | none | 436/1091 (40%) | 399/1089 (36.6%) | RR 1.09 (0.99 to 1.2) | 33 more per 1,000 (from 4 fewer to 73 more) | ⊕⊕○○ LOW | IMPORTANT |
| VTE | ||||||||||||
| 7 | RCTs | Seriousb | not serious | not serious | Seriousc | none | 40/1078 (3.7%) | 34/1073 (3.2%) | RR 1.16 (0.75 to 1.8) | 5 more per 1000 (from 8 fewer to 25 more) | ⊕⊕○○ LOW | IMPORTANT |
a
Serious imprecision due to high I2 (>50%) and non-overlapping confidence intervals.
b
Serious risk of bias due to lack of blinding of participants and personnel in some studies.
c
Serious imprecision due to confidence interval including benefit and harm or low number of events below optimal information size.