Background: For patients who continue to have seizures despite ongoing treatment, surgical resection of the epileptic focus may be considered, and can result in seizure-freedom. Currently, non-invasive tests provide information to inform the scope and positioning of invasive electroencephalography (EEG) electrodes. However, these technologies could replace intracranial EEG in at least some patients if their ability to accurately locate a seizure focus could be established. In order to inform clinical practice, studies need to investigate the clinical value of a test, and the impact of the results of that test on the decision-making process and subsequently on clinical outcomes. Objectives: The aims of this systematic review were to determine the diagnostic accuracy of non-invasive technologies, how these technologies impact on the decision-making process, associations with surgical outcome, and the gaps in the current evidence base. In addition, a decision-analytical model was designed to consider the potential use of existing data to determine the cost-effectiveness of options for presurgical work-up. Data sources: Eighteen electronic databases were searched without language restrictions [including MEDLINE, EMBASE, BIOSIS Previews, PASCAL, ClinicalTrials.gov, Cochrane Database of Systematic Reviews (CDSR), Cochrane Central Register of Controlled Trials (CENTRAL) and the Cochrane Register of Diagnostic Studies] from 2003 to July 2010. A prior, wider-ranging HTA review in this area conducted by the Centre for Reviews and Dissemination was used as the source for studies prior to 2003. Reference lists of included studies and relevant reviews were also searched, and a citation search of key papers undertaken. Review methods: Systematic reviews of the diagnostic accuracy, clinical utility and costeffectiveness of non-invasive technologies used to define the seizure focus in patients with refractory epilepsy being considered for surgery were undertaken according to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines. Thirteen diagnostic accuracy studies, seven outcome prediction studies and one study reporting the impact of test results on the decision-making process ('decision study') were included. The decision study was used to aid the development of a decision-analytical model to illustrate how data from appropriately designed clinical studies can be utilised. Results: Data from the diagnostic accuracy studies could not determine the contribution of the tests to the decision-making process. The number of index tests that could not be classified as correctly, non- or wrongly localising as indicated by a surgical outcome was high, up to 53%. The decision study reported fluorodeoxyglucose positron emission tomography influencing the decision for or against surgery in 78 of the 110 patients. The constructed decision-analytical model provided provisional cost-effectiveness results from the included diagnostic strategies. It demonstrated the feasibility of extending such analysis to all diagnostic strategies if suitable data were to become available. Limitations: There were a number of limitations of the available evidence, and overall, the quality of the available evidence was poor; only one study met the inclusion criteria that evaluated the use an index test on the decision-making process. Most of the available data was from the diagnostic accuracy studies; those currently available did not provide information on either the diagnostic accuracy or clinical utility of the tests being evaluated. Further limitations were the generally small study sizes, patient selection bias and the substantial clinical heterogeneity across the studies. Conclusions: The current evidence base is abundant but not adequately informative; there is no acceptable reference standard, reporting of clinical outcomes tends to be only following surgery, and decision level and clinical effectiveness studies are lacking. The additional value of diagnostic technologies for the localisation of epileptic foci is related to the impact on treatment decisions and the value of the treatments themselves; this needs to be considered fully in informing cost-effectiveness. Appropriately designed studies are needed to determine the added value of diagnostic regimens. Ultimately, how research informs the actual decision problem(s) faced by clinicians and the NHS needs to be considered; decision modelling is central to this issue. Funding: The National Institute for Health Research Health Technology Assessment programme.
