Patient outcomes in historical comparators compared with randomised-controlled trials

Int J Stroke. 2010 Feb;5(1):10-5. doi: 10.1111/j.1747-4949.2009.00393.x.

Abstract

Background: Decompressive hemicraniectomy for malignant middle cerebral artery infarction has long been controversial. Recently, data from randomised-controlled trials have shown that the procedure is life-saving and improves outcome. However, these randomised-controlled trials were difficult to conduct, because of ethical considerations due to high mortality in control groups. While the use of historical comparators may not be ideal for phase III efficacy trials, these data may be useful to inform the selection of trial populations. We sought to replicate the findings of the DESTINY trial of decompressive surgery in malignant middle cerebral artery infarction using the Virtual International Stroke Trials Archive, to determine whether historical comparators could be used as an alternative to control groups in situations where randomised-controlled trials are infeasible or regarded as unethical due to the high mortality under conservative treatment.

Methods: We extracted data on patients from Virtual International Stroke Trials Archive who displayed signs of malignant middle cerebral artery infarction (baseline National Institutes of Health Stroke Scale> or =20, LOC1A score of > or =1 on the National Institutes of Health Stroke Scale at baseline, lesion volume > or =145 cm(3)). We used a chi(2)-test and logistic regression (adjusting for baseline National Institutes of Health Stroke Scale) to compare the functional outcomes (modified Rankin scores and Barthel index) at the last available follow-up assessment between the DESTINY surgical and the Virtual International Stroke Trials Archive comparator groups. We assessed 90-day survival rates using a Kaplan-Meier analysis and Cox proportional hazards modelling (adjusting for the baseline National Institutes of Health Stroke Scale score).

Results: Fewer patients in the Virtual International Stroke Trials Archive comparator group (n=6/32, 19% with a 90-day follow-up) achieved a good functional outcome by mRS at the final follow-up, when compared with the DESTINY surgical group (n=8/17, 47% with a 6-month follow-up; chi(2)-test, P=0.04). This difference persisted after adjusting for baseline National Institutes of Health Stroke Scale (logistic regression, P=0.04), but not when accounting for patient age (P=0.66). Analysis of Barthel index at the final follow-up revealed no significant difference between the two groups (chi(2)-test, P=0.07), although a trend towards a better outcome in the DESTINY group was observed. In contrast with the findings of the DESTINY trial, we found no significant difference in 90-day survival rates between the surgical (88%) and the Virtual International Stroke Trials Archive (72%) comparator groups (Cox proportional hazards model, P=0.24).

Conclusion: The beneficial effects of decompressive hemicraniectomy on survival were not confirmed using a historical comparator dataset. Our observations might be due to the fact that patients with malignant middle cerebral artery infarction are usually excluded from clinical trials of drug efficacy, and patients identified from Virtual International Stroke Trials Archive may not have been truly representative of patients with malignant middle cerebral artery infarction. This mismatch could be rectified through recruitment of population-based studies and stroke registries to Virtual International Stroke Trials Archive to increase the number of patients eligible for entry into the comparator patient data pool.

MeSH terms

  • Adult
  • Computer Simulation
  • Craniotomy
  • Data Interpretation, Statistical
  • Databases, Factual
  • Decompression, Surgical
  • Female
  • Follow-Up Studies
  • Humans
  • Infarction, Middle Cerebral Artery / surgery
  • Infarction, Middle Cerebral Artery / therapy
  • Kaplan-Meier Estimate
  • Logistic Models
  • Male
  • Middle Aged
  • Neurosurgical Procedures
  • Proportional Hazards Models
  • Randomized Controlled Trials as Topic*
  • Research Design*
  • Stroke / surgery
  • Stroke / therapy*
  • Survival Analysis
  • Treatment Outcome*