|
Is MRI better than CT for detecting a vascular component to dementia? A systematic review and meta-analysis |
Beynon R, Sterne JA, Wilcock G, Likeman M, Harbord RM, Astin M, Burke M, Bessell A, Ben-Shlomo Y, Hawkins J, Hollingworth W, Whiting P |
|
|
CRD summary This well-conducted review concluded that there was insufficient evidence to suggest that magnetic resonance imaging was superior to computed tomography in identifying cerebrovascular changes, in autopsy-confirmed and clinical cohorts of patients with dementia or Alzheimer's disease. The authors' conclusions were suitably cautious, given the limitations of the available evidence. Authors' objectives To investigate the accuracy of magnetic resonance imaging (MRI) and computed tomography (CT), and whether MRI was superior to CT, in detecting a vascular component of dementia. Searching MEDLINE, EMBASE, BIOSIS, Science Citation Index, Zetoc, NTIS, Dissertation Abstracts, and the GrayLIT Network were searched, without language restrictions, for published articles, from database inception to February 2011; search terms were reported. Study selection Studies that assessed the accuracy of MRI, CT, or both, compared with one of the specified reference standards, for the detection of cerebrovascular changes, in patients with vascular dementia or Alzheimer's disease, were eligible for inclusion. Studies of patients with mixed pathology were included, regardless of the reference standard. Studies had to provide sufficient data to produce 2x2 tables of test performance. In the included studies, where reported, the mean age of participants ranged from 66 to 85 years. Most studies used the criteria of the National Institute of Neurological Disorders and Stroke, and the Association Internationale pour la Recherche et l’Ensignement en Neurosciences, or the Diagnostic and Statistical Manual of Mental Disorders (DSM)-III or -III-R as the reference standard. Two reviewers independently screened titles and abstracts; one reviewer screened full papers with decisions checked by a second. Disagreements were resolved by discussion. Assessment of study quality One reviewer assessed study quality, using a 10-point adapted version of the QUADAS; a second reviewer checked the assessment and disagreements were resolved by discussion. Data extraction The data were extracted by one reviewer to construct 2x2 tables of test performance for general infarcts, lacunar infarcts, non-lacunar infarcts, white-matter hyperintensities, periventricular hyperintensities, basal ganglia hyperintensities, and an overall assessment, such as the presence of two or more findings. Sensitivity, specificity, and the diagnostic odds ratio were calculated. A second reviewer checked the accuracy of the extracted data; disagreements were resolved by discussion. Methods of synthesis Summary estimates of sensitivity and specificity, with 95% confidence intervals, were calculated using the bivariate or hierarchical summary receiver operating characteristic model, where at least four studies were available. Where fewer than four studies were available, univariate random-effects meta-analyses were conducted. Positive and negative likelihood ratios were derived from these summary estimates. Summary diagnostic odds ratios, with 95% confidence intervals, were calculated using a standard random-effects meta-analysis; the ratio of diagnostic odds ratios for MRI, compared with CT, was calculated using meta-regression, and from studies where direct comparisons were made. Τ² was used to assess heterogeneity. Sensitivity analyses were conducted to investigate the impact of spectrum and incorporation bias. Results of the review Thirty-eight studies (4,377 patients; range 23 to 683) were included in the review: 26 studies (37 data sets) assessed CT, and 16 studies (33 data sets) assessed MRI; four studies assessed both. Twenty studies were prospective cohorts, six were retrospective cohorts, and 12 were case–control studies. Most studies (61%) did not enrol an appropriate patient spectrum, and there was a risk of incorporation bias in 23 (61%) of the non-autopsy studies. Other QUADAS items were classified as adequate or unclear in most studies. Progression bias could have been a problem in the autopsy studies. Autopsy as the reference standard: Seven studies were available. Overall, for CT (six studies) sensitivity ranged from 11% to 100% and specificity from 24% to 100%. For MRI (one study) sensitivity was 79% and specificity 28%. When comparing CT (two studies) with MRI (one study) the ratio of diagnostic odds ratios, for white-matter hyperintensity, was 0.28 (95% CI zero to 55,849). Other reference standards: For white-matter hyperintensity, the summary estimate of sensitivity was 71% (95% CI 53 to 85) and specificity was 55% (95% CI 44 to 66) for CT (11 studies), and sensitivity was 95% (95% CI 87 to 98) and specificity was 26% (95% CI 12 to 50) for MRI (six studies). For general infarcts, sensitivity was 53% (95% CI 36 to 70) and specificity was 96% (95% CI 94 to 97) for MRI, and sensitivity was 52% (95% CI 22 to 80) and specificity was 96% (95% CI 93 to 98) for CT. MRI had greater accuracy than CT for six of the seven imaging findings assessed; further results were presented. Authors' conclusions There was insufficient evidence to suggest that MRI was superior to CT in identifying cerebrovascular changes, in autopsy-confirmed and clinical cohorts of patients with Alzheimer's disease or dementia. CRD commentary The review addressed a clear question, supported by reproducible inclusion criteria. Several sources were searched for published or unpublished studies, with no date or language restrictions, and no methodological filters, minimising the risk of missing relevant studies. Each stage of the review was conducted by two people separately, reducing the risks of error and bias. Study quality was assessed, using appropriate criteria; summary results were presented and these results were considered in the analysis. Appropriate methods of synthesis were used. The authors' conclusions were suitably cautious, given the evidence available. Implications of the review for practice and research Practice: The authors did not state any implications for practice. Research: The authors stated that studies should: assess MRI and CT in the same patients, with symptoms of early dementia; be large enough to allow precise estimates of relative diagnostic accuracy; use a reference standard of accepted diagnostic criteria, without incorporating imaging findings, ideally with autopsy confirmation; use global assessment criteria for MRI and CT, based on the most useful individual imaging findings, indicative of a vascular component to dementia; and quantify diagnostic accuracy. Funding Funded by the UK Medical Research Council. Bibliographic details Beynon R, Sterne JA, Wilcock G, Likeman M, Harbord RM, Astin M, Burke M, Bessell A, Ben-Shlomo Y, Hawkins J, Hollingworth W, Whiting P. Is MRI better than CT for detecting a vascular component to dementia? A systematic review and meta-analysis. BMC Neurology 2012; 12: 33 Indexing Status Subject indexing assigned by NLM MeSH Cerebrovascular Disorders /diagnosis /epidemiology; Comorbidity; Dementia /diagnosis /epidemiology; Humans; Magnetic Resonance Imaging /statistics & Prevalence; Reproducibility of Results; Risk Assessment; Sensitivity and Specificity; Tomography, X-Ray Computed /statistics & numerical data; numerical data AccessionNumber 12012041460 Date bibliographic record published 04/01/2013 Date abstract record published 12/08/2013 Record Status 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. |
|
|
|