Abstract
Among all patients who present to the emergency department with ‘the worst headache’ of their lives, approximately 12%–25% have a subarachnoid haemorrhage (SAH) (1, 2). The first haemorrhage can be fatal or result in permanent neurological disability. Because early surgery to repair aneurysms reduces short-term complications and improves outcomes, an accurate early diagnosis is critical (3). Unfortunately, misdiagnosis of SAH remains common.
Patients with a SAH frequently describe a distinct severe headache in the days or weeks before the index episode of bleeding, which is often ignored. The results of research on this topic are highly varied, yielding estimates of incidence of sentinel headache as high as 60% (4, 5) and as low as 10% (6). Before additional widespread public and healthcare education campaigns to reduce the misdiagnosis of SAH are designed and implemented, the true incidence must be better estimated.
A systematic review is a summary that attempts to answer a focused clinical question using methods designed to reduce the likelihood of bias. Meta-analysis describes systematic reviews that use quantitative methods to summarize the results. The process of conducting a systematic review involves 1. defining the question, 2. conducting a literature search, 3. identifying relevant studies, 4. applying inclusion and exclusion criteria, 5. appraising the studies, 6. abstracting data and 7. conducting analysis by combining results if appropriate (7). Systematic reviews and meta-analysis are being used most frequently to combine results of randomized controlled trials (RCT). However, in many situations, a RCT design is either not feasible or not appropriate to answer certain clinical questions. Answering a question about incidence falls in the realm of epidemiologic or observational research. Systematic reviews have a role here also. The number of meta-analyses of observational studies in health has increased substantially in the last decade (8). Recently, systematic reviews of the incidences of upper gastrointestinal bleeding in the general population (9), schizophrenia (10), intracerebral haemorrhage (11) and epilepsy (12) have been published during 2002 alone. Systematic reviews of observational data pose particular and unique challenges due to inherent biases and differences in study designs (13). To aid authors, reviewers, editors, and clinicians, a special consensus statement proposal for the uniform reporting of systematic reviews and meta-analysis of observational studies was published (14).
Polmear has conducted a systematic review of the literature in order to determine the true incidence of sentinel headaches in aneurysmal subarachnoid haemorrhage. In the introduction, the author clearly defines the problem under study. The methods outline the type of studies under consideration and the study characteristics which explicitly define the inclusion and exclusion criteria. The search for relevant studies was exhaustive and included multiple large bibliographic databases, English and non-English publications, reference lists, and experts in the field. Two reviewers independently applied inclusion criteria, appraised the quality of the studies, and extracted the data with very good agreement (kappa 0.88). In some cases, in the presence of marked heterogeneity, a single summary estimate (in this case, an incidence) may well be inappropriate. Investigation of heterogeneity, which was conducted by the authors, is an important feature of systematic reviews. The results are succinctly reported and the discussion includes relevant information related to the dangers of bias in observational studies. Overall, the quality of Polmear's systematic review is high and the results are valid: The best estimate of the true incidence of sentinel headaches in aneurysmal SAH is in the range of 10% to 43%. The variation in incidences is most likely due to differences in rates of misdiagnosis from region to region.
Given the morbidity and mortality of aneurysmal subarachnoid haemorrhage, the finding that sentinel headaches occur in at least 10% of these patients underscores both the opportunity and clinical mandate to identify this group of patients who may present early to medical attention.
Systematic reviews of observational studies will increasingly be utilized to derive the best estimates of a true incidence in a given population. Although there are unique challenges to conducting a systematic review and meta-analysis of observational research, consensus guidelines exist to improve the quality of such publications, and a well-conducted systematic review can have far-reaching clinical implications.