Evidence Based Review Article

The Journal of Informed Pharmacotherapy 2004;15:201.

Blood Pressure Lowering or Just Candesartan in Acute Ischemic Stroke? The ACCESS trial

Reviewer: Peter Loewen, Pharm.D.
Reviewer’s e-mail address:
Reviewer's profession/specialty: Pharmacotherapeutic Specialist - Adult Medicine

Original Citation

Schrader J, Luders S, Kulschewski A, et al.  The ACCESS Study: evaluation of Acute Candesartan Cilexetil Therapy in Stroke Survivors. Stroke 2003;34:1699-1703. PubMed Citation

Overall Study Question

The objective of this study was to assess the safety of modest blood pressure reduction using candesartan in acute ischemic stroke.  Patients with acute ischemic stroke (hemorrhage ruled out by CT) and hypertension such that blood pressure (BP) reduction was indicated, based on current (German) guidelines (SBP >200 mmHg and/or DBP >110 mmHg 6-12 hours after onset or SBP >180 mmHg and/or DBP >105 mmHg 24-36 hours after admission).  

This study was conducted in two phases.  PHASE 1: In the experimental arm, the intent was to produce a 10-15% BP reduction within 24 hours using candesartan started on within 24h of admission. In reality, the treatment was started about 30 hours after symptom onset. The dose was titrated from 4 mg daily to 16 mg over the first 48h, according to need. If BP remained very high (SBP >230 mmHg or DBP >115 mmHg on day 1 or SBP >200 mmHg or DBP >110 mmHg on following days) despite candesartan, urapidil (an alpha/beta receptor blocker used commonly in Europe) could be electively added. This portion of the study lasted 7 days, and the control group received placebo during this time. PHASE 2: At 7 days post-stroke, candesartan patients whose BP was still >135/85 mmHg had their dose increased or additional antihypertensives added (hydrochlorothiazide, metoprolol or felodipine). In placebo-treated patients with BP >140/90 mmHg, candesartan was started. This was required in 164/166 patients in the placebo arm. Phase 2 continued for one year. Note that the BP levels justifying intervention used here, based apparently on German guidelines, are slightly lower than the thresholds in the ACLS guidelines used in North America (SBP >220 mmHg, DBP >120 mmHg). 

The primary outcomes were mortality or disability (using Barthel Index) at 30 days post-stroke. Secondary outcomes included mortality plus cerebrovascular or cardiovascular events at one year.

What was the declared relationship between authors and sponsors for this study?

Although the study was “supported” by AstraZeneca Germany (makers of candesartan), the authors report that this was an investigator-initiated study. None of the authors were employees of the sponsor. No conflict of interest declarations are included in the manuscript.

Are the Results of the Study Valid?

1. Was assignment of patients randomized?

Yes, during Phase 1, which applied to the primary endpoint. Phase 2 treatment was dictated based on presence or absence of hypertension at 7 days post-stroke and thus was not randomized.

2. Were all patients who entered the trial properly accounted for and attributed at its conclusion?

The plan was to enroll 500 patients. The data safety monitoring board recommended stopping the study when 342 subjects had been enrolled. Only 339 subjects were included in the analysis, although reasons for exclusion of the 3 patients are given (carotid stenosis, brain metastasis). Thus, the analysis was not performed using intention-to-treat methods. The exclusion of 3 subjects is not likely to influence the results meaningfully.

3. Were patients, their clinicians, and study personnel 'blind' to treatment?

Yes, in Phase 1. Though the authors do not differentiate between Phases 1 and 2 in this regard, it is difficult to imagine how blinding could have been maintained in phase 2. It is somewhat irrelevant, however, since all patients except two were on candesartan during this phase. 

4. Were the groups similar at the start of the trial?

For the baseline characteristics described, they appeared to be similar.

5. Aside from the experimental intervention, were the groups treated equally?

Yes, during Phase 1. In Phase 2, however, not much detail is provided. It is reported that use of concomitant antihypertensives and aspirin was similar between the two groups. 

What were the Results?

1. How large was the treatment effect? 

Primary endpoint (30 days): No mortality results are provided, aside from what can be observed from the 12-month survival curves, which does show divergence in favour of the experimental group. The Barthel Index scores showed no difference at this point, although the authors report that during the analysis phase they decided that it was invalid to use the Barthel Index based on some new data which had become available. The amount of urapidil required was not reported.

Secondary Endpoint (12 months): Death plus cerebrovascular or cardiovascular events occurred in 9.8% of experimental arm and 18.7% in control arm (p=0.026). Mortality alone occurred in 2.9% of patients in experimental compared to 7.2% in control arm (p=0.07). Cerebrovascular events occurred in 7.5% in experimental compared to 11.5% in control arm while cardiovascular events occurred in 1.2% of the experimental group compared to 6.0% in control arm. The survival curves for the composite secondary endpoint were significantly different (p=0.026), favouring the experimental arm. The odds ratio for the secondary endpoint was 0.475 (95% CI 0.252-0.895).

It is critical to note that despite the use of candesartan, there was no difference blood pressure lowering between the experimental and control groups over the first 7 days. During the 1 year follow-up, the two groups were on roughly the same treatment, so as expected, there were no overall differences in blood pressure here either.

2. How precise was the estimate of the treatment effect?

Unfortunately, no evaluable results for the primary endpoint were provided.  The 95% confidence interval around the absolute risk reduction (ARR) for the secondary endpoint (8.9%) was 1.44-16.25%. Thus, this result is consistent with a fairly large and clinically meaningful treatment effect as well as a small and possibly marginally important one. 

Will the Results Help Me in Caring for My Patients?

1. Can the results be applied to my patient care?

Because of the preliminary nature of this data (a Phase II trial with incomplete reporting of primary outcome results) and its failure to test the blood pressure-lowering hypothesis, it would be premature to conclude that candesartan in the preferred agent for pharmacologically lowering BP in the acute stroke phase. It was not effective for lowering blood pressure. However, its use could be considered as an adjunct to standard acute BP-lowering therapy in hopes of replicating the results show in this trial.

2. Were all clinically important outcomes considered?

The outcomes considered were appropriate, however more rigorous endpoint evaluations would have included other measures of function (e.g. Modified Rankin scale) and neurologic outome (e.g. NIH Stroke Scale) as is typically done in acute stroke trials. Curiously, an earlier publication by the authors states that they planned to use exactly these measures.(2)  Using these could have helped the primary endpoint analysis since the Barthel Index had to be abandoned.  

3. Are the likely treatment benefits worth the potential harms and costs?

The limitations in reporting of the primary endpoint make it difficult to determine whether the benefits outweigh the harms overall. From a vascular events viewpoint, they certainly appeared to at the 1 year endpoint. Unfortunately, insufficient data is provided to judge whether patients were any better off from a functional or neurological viewpoint at 3 months or 1 year.


This trial is the first systematic evaluation of blood pressure lowering in patients with very high BP in the immediate post-stroke period. Current guidelines state that no attempts to lower blood pressure should be made in this phase unless the pressure is very high or there is concomitant myocardial ischemia, heart failure, or arterial dissection. (1) Even then, however, the effects of doing so have not been formally evaluated. The main concern about lowering blood pressure is that it may interfere with a compensatory response to cerebral hypoperfusion and thereby cause harm. 

What is remarkable about this trial is that a relatively large effect on vascular events was achieved by starting candesartan immediately, even though blood pressure was not lowered any more than it was in the placebo group. This strongly suggests another mechanism for candesartan’s efficacy in acute stroke. Furthermore, even though both groups were on similar BP-lowering regimens over the next year with similar BP, vascular events continued to accrue at a much higher rate in the group which received placebo during the first 7 days than in the initial candesartan recipients. Looking at the survival curves, it appears that the divergence occurred after the first 7 days. 

The authors hypothesize that early vascular remodeling is favourably influenced by angiotensin blockade and the effects of this are manifest over the subsequent months.

While it is clear is that a larger more rigorously reported trial is justified and probably required to confirm the promising hypotheses raised in this trial, in the meantime clinicians are left in a difficult position. For patients who require blood pressure lowering according to ACLS guidelines, should they use candesartan in hopes of replicating the outcome benefits shown in this trial even though it may be inefficacious in meeting blood pressure goals? Or, should they use nitroprusside or labetalol and titrate upward until the BP drops by 10-20%, as the ACLS guidelines currently recommend, despite no clinical trials to suggest benefit? A third option would be to start candesartan in hopes of gleaning its remodeling benefits and use standard BP-lowering drugs as per ACLS to do the “heavy lifting” of BP reduction. I suspect that the third option is justified, particularly in patients with a history of hypertension requiring chronic drug therapy anyway.

Unfortunately this trial did not test its primary hypothesis of whether it is safe to lower blood pressure in the acute stages of ischemic stroke. Nonetheless candesartan appeared to confer significant benefits when initiated early. 

Reviewer Competing Interests

None declared.


  1. ACLS 2000 Guidelines. Circulation 2000;102(supp1):I204-I216.
  2. Schrader J, Rothemeyer M, Luders S, Kollmann K. Hypertension and stroke--rationale behind the ACCESS trial. Acute Candesartan Cilexetil Evaluation in Stroke Survivors. Basic Res Cardiol 1998;93Suppl 2:69-78.

Copyright © 2004 by the Journal of Informed Pharmacotherapy. All rights reserved.