Worse outcomes with general anaesthesia in pooled analysis of stroke patients


Data from the pooled analysis of the HERMES collaboration have shown that stroke patients given endovascular treatment under general anaesthesia experienced worse outcomes than those treated without general anaesthesia. Bruce Campbell (Royal Melbourne Hospital, University of Melbourne, Australia) presented the results of the analysis at the International Stroke Conference (ISC; 22–24 February, Houston, USA), concluding that this suggests general anaesthesia should be avoided whenever possible pending further randomised trial results.

Investigators examined the association of general anaesthesia (GA) and outcome in the five HERMES trials (MR CLEAN, ESCAPE, REVASCAT, SWIFT PRIME, and EXTEND IA) published in early 2015. According to Campbell, two statisticians collated and pooled the clinical data independent of the investigators; and a statistical analysis plan was formulated and finalised prior to data analysis. Logistic regression models were adjusted for pre-specified prognostic variables (age, sex, baseline stroke severity on NIHSS, time from symptom onset to randomisation, baseline ASPECTS, baseline site of occlusion, and whether the patient received alteplase) and the source trial.

General anaesthesia was used in 25% (n=153) of the patients and they were well matched with no statistically important differences between these key variables. “If anything, age favoured the general anaesthesia group, NIHSS was the same, ASPECTS was the same and the onset to randomisation also slightly favoured the general anaesthesia group,” Campbell reported.

Baseline characteristics were as follows: in the general anaesthesia group (n=153), the mean age was 65.4 years, 45.1% female with an NIHSS score of 17 (median) and a median ASPECTS score of 9, and a mean onset to randomisation time of 208.7 minutes. In the non-GA group (n=456), the mean age was 67.2 years, and 48% were female with an NIHSS score of 17 (median), a median ASPECTS score of 9, and a mean onset to randomisation time of 216.8 minutes.

“Patients admitted directly to the treating centre were more likely to be treated awake. The distribution of vessel occlusions was the same, and alteplase was more frequently administered in the GA group,” he noted.

Bruce Campbell

In terms of functional outcome, Campbell reported that the non-GA endovascular group had a massive shift towards reduced disability versus standard care group (which included IV t-PA in 87% of patients), with a common odds ratio of 2.62, p<0.001 in ordinal analysis of the day 90 modified Rankin Scale (mRS). “If you look at the general anaesthesia versus control group, the common odds ratio is 1.73, reduced compared to the awake patients, but still a statistically significant benefit. That is in contrast to the previously reported finding that GA abolished endovascular treatment benefit in the MR CLEAN trial. If you look at endovascular treatment under general anaesthesia versus non-GA, there was a statistically significant benefit of being awake (common odds ratio 1.59, p=0.01 favoring non-GA ),” he added.

The data also showed that there were statistically significant differences in independent functional outcome (mRS 0–2) between standard care (26.5%), general anaesthesia thrombectomy (35.9%) and awake thrombectomy (50%). Further, excellent functional outcome (mRS 0–1) among the three groups saw statistically significant differences with 12.9% for standard care, 19% for the general anaesthesia group and 30% for the non-GA group. A similar observation was made in terms of early neurological recovery (8 point reduction in NIHSS or 0–1 at 24 hours) with 21.2% in the standard care group, 37.9% in the general anaesthesia group and 54.6% in the non-GA group. Investigators also did a propensity based analysis which gave exactly the same result.

Mortality, Campbell pointed out, showed some interesting findings. “There was a significant reduction in mortality in the awake group (13.6% versus standard care 18.9%, p=0.03) that was not seen in the general anaesthesia group (18.3%). For people who have concerns about the safety of doing things awake, there is certainly no evidence of that increasing the rate of death, and likewise on the rate of symptomatic intracerebral haemorrhage (ICH) (4.2% in the non-GA group versus 5.2% in the GA group). There was no also no significantly increased pneumonia in the GA patients and similar rates of vessel perforation.

The time interval between randomisation and reperfusion was greater in the GA group compared to the non-GA group (median 98 vs. 75 minutes, p˂0.001). Despite this, the similar magnitude difference in onset to reperfusion time did reach statistical significance (median 303 vs. 280 minutes in the GA and non-GA groups, respectively, p=0.10) but there was a nonsignificant imbalance in the time from stroke onset to randomisation (208min GA vs. 217min non-GA).

Campbell acknowledged that there are limitations to this study, noting that patients were not randomised to anaesthetic strategy and therefore there may be unmeasured confounding factors that led to inclusion of sicker patients in the GA group and potentially biased the analysis.

“However, all of that taken into account,” Campbell stated, “general anaesthesia was associated with worse outcomes after endovascular thrombectomy, after adjustment for baseline prognostic variables. We also did a propensity-based analysis which gave exactly the same result. There was no safety advantage of general anaesthesia—there was increased pneumonia, similar perforation and ICH rates. Together, if you synthesise those results, I would say there is evidence to avoid general anaesthesia wherever possible pending further randomised trial results. If you need general anaesthetic, however, you should not avoid doing the procedure as it still remains effective versus standard care.”

Single-centre RCT contradicts HERMES anaesthesia findings

On the other hand, the SIESTA study (Sedation vs. intubation for endovascular stroke treatment), published in the Journal of the American Medical Association (JAMA) found that conscious sedation did not result in better early neurological outcome compared with general anaesthesia. Study authors, Silvia Schönenberger et al, maintain that the study findings do not support an advantage for the use of conscious sedation.

The single-centre, randomised, parallel-group, open-label treatment trial with blinded outcome evaluation conducted at Heidelberg University Hospital in Germany (April 2014–February 2016) included 150 patients with acute ischaemic stroke in the anterior circulation, higher National Institutes of Health Stroke Scale (NIHSS) score (>10), and isolated/combined occlusion at any level of the internal carotid or middle cerebral artery.

Patients were randomly assigned to an intubated general anaesthesia group (n=73) or a non-intubated conscious sedation group (n=77) during stroke thrombectomy. Primary outcome was early neurological improvement on the NIHSS after 24 hours (0–42). Secondary outcomes were functional outcome by modified Rankin Scale (mRS) after three months (0–6), mortality, and peri-interventional parameters of feasibility and safety.

“Among 150 patients (60 women [40%]; mean age, 71.5 years; median NIHSS score, 17), primary outcome was not significantly different between the general anaesthesia group (mean NIHSS score, 16.8 at admission vs. 13.6 after 24 hours; difference, -3.2 points [95% CI, -5.6 to -0.8]) vs. the conscious sedation group (mean NIHSS score, 17.2 at admission vs. 13.6 after 24 hours; difference, -3.6 points [95% CI, -5.5 to -1.7]); mean difference between groups, -0.4 (95% CI, -3.4 to 2.7; p=0.82),” Schönenberger et al reported.

Further, they write that in the general anaesthesia group substantial patient movement was less frequent than in the conscious sedation group, (0% vs 9.1%; difference, 9.1%; p=0.008), but post-interventional complications were more frequent for hypothermia (32.9% vs. 9.1%; p<0.001), delayed extubation (49.3% vs. 6.5%; p<0.001), and pneumonia (13.7% vs. 3.9%; p=0.03) in the general anaesthesia group. More patients were functionally independent (unadjusted mRS score, 0 to 2 after three months [37% in the general anaesthesia group vs. 18.2% in the conscious sedation group; p=0.01]). There were no differences in mortality at three months (24.7% in both groups).

“Among patients with acute ischaemic stroke in the anterior circulation undergoing thrombectomy, conscious sedation versus general anaesthesia did not result in greater improvement in neurological status at 24 hours. The study findings do not support an advantage for the use of conscious sedation,” the authors concluded.

Commenting on the SIESTA results compared to the HERMES results, Campbell stated “It is important to note the excellent control of physiologic parameters in SIESTA, particularly blood pressure which was maintained at >140mmHg systolic. The approach of giving the same medications for the conscious sedation group as the general anaesthesia group in SIESTA also differs substantially from common practice in the HERMES trials where many non-GA patients had no sedation at all and this may have implications for functional outcome. Together, if you synthesise these results, I would still say there is good reason to avoid general anaesthesia unless is it medically necessary to undertake the procedure, pending results of further randomised trials. There is no suggestion of a safety advantage of general anaesthesia and both studies found significantly increased pneumonia in the general anaesthesia patients. However, if general anaesthetic is required due to airway compromise or marked agitation, physicians should feel reassured that the thrombectomy procedure is likely to remain effective versus standard care but pay very close attention to maintaining blood pressure.”