Systematic Searches and Literature Review
The systematic identification of relevant literature was conducted according to National Health and Medical Research Council (NHMRC) standards between July and November 2006. Previous international and national stroke guidelines were identified and evaluated using the AGREE tool. Guidelines developed by the Royal College of Physicians in the United Kingdom (UK) in 2004 were deemed the most recent and robust guidelines and hence were used as a basis for updating the literature searches. An external consultant was used to undertake all the electronic database searches.
Question Formulation
89 clinical questions were developed by the Expert Working Group (EWG) to address interventions relevant to acute stroke care. The questions generally queried the effects of a specific intervention and were developed in three parts: the intervention, the population and the outcomes. An example is "What is the effect of anticonvulsant therapy on reducing seizures in people with post-stroke seizures?" In this example, anticonvulsant therapy is the intervention, reduction of post-stroke seizures is the outcome, and the population is people with post-stroke seizures.
Finding Relevant Studies
For this guideline searching, there could be no single search coverage for all 89 questions: some sections of the guidelines need updating only from 2003, some are topics not previously addressed in the guidelines, some have already been well researched by other reputable guidelines authorities while some have no comprehensive meta-analysis relating to them.
In order to have some structure to the searching and to make filtering of the references more manageable, the questions were searched and stored in separate Endnote libraries by broad topics:
- Organisation of care
- Discharge planning, transfer of care and integrated community care
- Pre hospital care
- Early diagnostic assessment
- Management in the emergency phase
- Assessment and management of consequences of stroke
- Prevention and management of complications
- Early secondary prevention
- Palliation and death
- Transient ischemic attack (TIA)
Each reference within the library was then marked with the questions for which it was relevant. For Australasian Medical Index, EMBASE, Medline and Medline in-process & other non-indexed citations searching was conducted in four broad steps:
- Terms for the patient group (P) were abridged from the Cochrane Collaboration's Stroke Group.
- Where appropriate, intervention or other factor terms were added.
- Relevant evidence filters (Cochrane sensitive filter or Medline diagnostic filter) were applied to the basic search strategies.
- If the search was for an update only to National Stroke Foundation (NSF) or other authoritative meta-analysis, the references were limited to years 2003 onwards.
For brevity, search strategies are not included in the original guideline document but are available from the NSF. Table 3 in Appendix A of the original guideline document outlines the number of articles found for each 10 topic areas listed above.
A systematic process for choosing relevant articles occurred. At first, relevant systematic reviews were initially identified. Where no systematic review was found, primary studies were then searched. This initial process was conducted by one member of the working group. Final decision to include and review articles was made by two members of the working group after abstracts were scrutinised. Reference lists of identified articles and other guidelines were then used to identify further trials. The table of contents of a number of key journals for the last 6 months was also conducted. The following journals were chosen: Stroke, Cerebrovascular Disease, Lancet (and Lancet Neurology), and Archives of Physical Medicine and Rehabilitation. For a number of topics a general internet search was then undertaken (using the "Google" search engine). Finally, where possible, experts in the field were contacted to review the identified studies and suggest other new studies not identified. Hand searching continued until May 2007 and significant studies were included.
Cost Analysis
The Guidelines project officer conducted a separate systematic review for this section. The economic literature was searched with a total of 1484 references retrieved after deduplication (see Table 4 in Appendix A of the original guideline document). One person sorted these and selected 70 potentially relevant articles. These abstracts were scrutinised for omissions by two people and appropriate papers were retrieved and reviewed (n=30).