This updated systematic review concluded that patient history and clinical symptoms (particularly acute facial paresis, arm drift and abnormal speech) can provide the basis for diagnosing patients suffering from a possible stroke. The authors' conclusions follow from the evidence presented. However, owing to the limited search and the lack of detail about the methods used, it is not possible to verify these conclusions.

To update a systematic review that assessed the accuracy and reliability of symptoms and neurological examination findings in the evaluation of patients with suspected stroke or transient ischaemic attack.

This was an update of a 1994 report and the search was restricted to the time since the last review (see Other Publications of Related Interest). MEDLINE was searched from 1994 to 2005 using the search terms reported. This search was restricted to English-language articles. In addition, bibliographies of textbook chapters, review articles, and personal files were manually checked for any additional relevant literature.

Study designs of evaluations included in the review

Case reports and reviews of other studies were not eligible for inclusion; no further inclusion criteria were specified. The studies included in the review were independent, blind comparisons of the index test with a valid 'gold' standard, with participants enrolled consecutively or non-consecutively.

Specific interventions included in the review

Studies providing data on the accuracy and/or reliability of tests for the diagnosis of stroke were eligible for inclusion. The specific tests used in the included studies were patient history or physical examination for the clinical assessment of stroke: either diagnosis or estimation of short-term prognosis (mortality or functional disability).

Reference standard test against which the new test was compared

The studies had to include a description of a final diagnosis of stroke, as established by a clinical expert who had reviewed all clinical data, neuro-imaging and other relevant laboratory tests.

Participants included in the review

Studies that described the study site and/or the participants included in the study were eligible for inclusion. The participants were defined as either clinicians or patients; no further details were specified.

Outcomes assessed in the review

To be eligible for inclusion, the studies had to present primary data or appropriate summary statistics of the accuracy or reliability of the tests, or an estimation of the short-term prognosis, such as likelihood ratios (LRs).

How were decisions on the relevance of primary studies made?

The authors did not state how the papers were selected for the review, or how many reviewers performed the selection.

The quality of the studies was assessed by evaluating whether the clinical examination was performed masked to the neuro-imaging results. The authors did not state how many reviewers performed the quality assessment.

The authors did not state how many reviewers performed the data extraction. The outcome measures included in the review were the sensitivity, specificity and LRs. These were extracted directly from each article, where available, or calculated from primary data. For studies with outcomes on the precision of diagnosis, the kappa statistic (for paired observers) or intraclass correlation coefficient (more than 2 examiners) was extracted.

How were the studies combined?

Summary estimates for the LRs of the accuracy studies, with corresponding 95% confidence intervals (CIs), were calculated using a random-effects model.

How were differences between studies investigated?

The authors did not state how differences between the studies were investigated.

Fourteen diagnostic accuracy studies (n=5,484) reporting sensitivity and specificity data were included in the review. The number of included reliability studies was unclear and their results are omitted from this abstract.

The authors used a prior probability of a stroke of 10% for patients with present neurologically relevant symptoms. They concluded that the presence of acute facial paresis, arm drift or abnormal speech (as evaluated by physician assessment) increased the likelihood of stroke (LR of at least one finding 5.5, 95% CI: 3.3, 9.1). The absence of all three symptoms decreased the odds of a stroke (LR of zero findings 0.39, 95% CI: 0.25, 0.61). The early mortality from stroke was increased with the presence of impaired consciousness, hemiplegia or conjugate gaze palsy (LR of at least one finding 1.8, 95% CI: 1.2, 2.8). The absence of all three symptoms decreased the odds of mortality (LR of zero findings 0.36, 95% CI: 0.13, 1.0). It was unclear how many studies these results were based upon.

The included studies ranged from highest quality with independent blind comparisons of the test with a valid 'gold' standard and a large number of patients consecutively enrolled (1 study), to lower quality with independent blind comparisons of the test with a valid 'gold' standard and with patients enrolled non-consecutively (8 studies).

Patient history and clinical findings provide the basis for diagnosing patients suffering from a possible stroke. Focusing on three findings (acute facial paresis, arm drift and abnormal speech) may improve the accuracy and reliability of the diagnosis of stroke.

This review addressed a clear question in terms of the index and reference tests used in studies eligible for inclusion and the outcomes included in the review. Limited details on the eligibility of various study designs was given. The search was limited to one electronic database, supplemented by handsearches, and only studies published in English were eligible for inclusion. It is therefore possible that some relevant studies, either unpublished or published in languages other than English, might not have been identified by the search. There were no details on how the studies were selected, how quality was assessed, or how the data were extracted. This means that the review cannot be assessed for any potential errors or reviewer bias that might have occurred.

Since there were limited details on the studies themselves and no results from individual studies were reported, it was difficult to assess whether the statistical analyses were appropriate. Differences between the studies were not formally assessed or discussed. The limited search strategy and the lack of detail on the review methodology and on the individual studies mean that it is not possible to assess the reliability of the authors' conclusions.

Practice: The authors stated that history and neurological examinations are critical tools for the identification and treatment of patients with suspected stroke.

Research: The authors did not state any implications for future research.

Goldstein LB, Matchar DB. Clinical assessment of stroke. JAMA 1994;271:1114-20.

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.