GRADE evidence profile
| Patients: aged 18 years and above suffering massive MCA Intervention: decompressive hemicraniectomy surgery Comparator: best (standard) medical management Outcome: death and/or disability at 12 months follow-up based on mRS scores | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Quality assessment | Summary of findings | |||||||||||
| Number of patients | Effect | Quality/certainty of evidence | ||||||||||
| mRS cut-off point; n of studies | Design | Risk of bias | Consis-tency | Direct-ness | Preci-sion | Publication bias | Hemi-craniectomy surgery | Medical care | Relative (95% CI) | Absolute effect | ||
| Hemi | Med | |||||||||||
| mRS 0-5 vs 6 (death); n=7 | Randomised controlled trials* | No Serious risk of bias | No serious inconsis-tency† | No serious indirect-ness‡ | No serious imprecision | None detected§ | n=165 | n=173 | RR (95% CI) 2.05 (1.54 to 2.72) | 697 per 1000 | 307 per 1000 | HIGH CONFIDENCE/CERTAINTY |
| 390 more per 1000 patients; 95% CI from 165 to 527 | ||||||||||||
| mRS 0-3 vs 4-6; n=7 | Randomised controlled trials* | Serious¶ | No serious inconsis-tency† | No serious indirect-ness‡ | Serious imprecision** | None detected§ | n=165 | n=173 | RR (95% CI) 1.58 (1.02 to 2.46) | 267 per 1000 | 139 per 1000 | LOW CONFIDENCE/CERTAINTY |
| 128 more per 1000 patients; 95% CI from 3 more to 203 more | ||||||||||||
| mRS 0-4 vs 5 and 6; n=7 | Randomised controlled trials* | Serious¶ | No serious inconsis-tency† | No serious indirect-ness‡ | No serious imprecision** | None detected§ | n=165 | n=173 | RR (95% CI) 2.25 (1.51 to 3.35) | 588 per 1000 | 237 per 1000 | MODERATE CONFIDENCE/CERTAINTY |
| 351 more per 1000 patients; 95% CI from 121 more to 557 more | ||||||||||||
*Six trials that reported complete 12-month follow-up mRS data and one trial based on 6-month follow-up data from the pooled analysis; note while we judged low risk of bias, the reporting of sequence generation could be substantially improved.
†Statistical consistency (heterogeneity): χ2 tests were not significant and I2s were generally low (<50%).
‡Directness: we judged that there was directness as there was clear applicability of study patients to the research question (similar patients); there were no indirect comparisons reported as part of the included trials.
§Based on our exhaustive literature search and the absence of problems of industry funding, we judged that the risk of important publication bias was low.
¶We rated down for risk of bias because in four studies allocation was not concealed, in three studies outcome assessors were not blind to allocation and all but two studies stopped early for benefit. We did not rate down for the outcome of mortality because it is not subject to bias in outcome assessment.
**Precision: we rated down particularly due to imprecision of estimates as a result of total small sample size and small number of events (particular imprecision was for mRS 0-3).
MCA, middle cerebral artery infarction; mRS, modified Rankin Scale; RR, risk ratio.