Quality of systematic reviews of diagnostic test accuracy on Alzheimer’s disease dementia and other dementias

Article type
Authors
Arevalo-Rodriguez I1, Segura O1, Solá I2, Sánchez E1, Bonfill X2, Alonso-Coello P3
1Fundación Universitaria de Ciencias de la Salud, Hospital de San José /Hospital Infantil de San José , Bogotá D.C., Colombia
2Iberoamerican Cochrane Centre - Institute of Biomedical Research (IIB Sant Pau), CIBER Epidemiología y Salud Pública (CIBERESP), Spain - Universitat Autónoma de Barcelona, Barcelona, Spain
3Iberomerican Cochrane Centre, Clinical Epidemiology and Public Health Department, Institute of Biomedical Research (IIB Sant Pau), Spain. CIBER de Epidemiología y Salud Pública (CIBERESP), Spain
Abstract
Background: The Cochrane Handbook of Diagnostic Test Accuracy (DTA) has introduced important guidelines in order to develop high-quality DTA reviews. In dementia, clinical diagnosis has become a paramount issue related to early and adequate management. Systematic reviews (SRs) of accuracy of dementia tests can establish which diagnostic tools should be used in daily-practice. It is important to know the rigor of these reviews in order to know their role to guide decision-making process.

Objectives: To describe the characteristics and quality of DTA SRs in the field of Alzheimer disease (AD) and other dementias.

Methods: We searched MEDLINE, EMBASE, The Cochrane Library and DARE (from their inception to March 2013), as well as list of references of included studies. Two researchers independently assessed eligibility, extracted data and assessed risk of bias (RoB) using the AMSTAR-II tool, according to the Cochrane methodology for DTA reviews.

Results: We included 23 SRs with sample sizes ranging from 160 to 26 019 participants. Only ten reviews (43%) assessed RoB of included studies, being patient’s spectrum and incorporation bias the most frequent biases. One review (4.3%) reported to have a protocol, 20 (86.9%) not provided a list of included/excluded studies and 15 (65.2%) did not reported a duplicate study selection/data extraction. Eight reviews pooled data with methods addressed to SRs of interventions (34.7%). Five reviews (21.7%) analyzed publication bias by means of funnel plots and/or Egger’s Test. All SRs did not report consequences of inconclusive results, adverse events or use of resources.

Conclusions: The set of assessed SRs shows important flaws with an impact in the confidence of their results. In order to improve the quality of DTA SRs is necessary to value the nature of the diagnostic evidence, apply appropriate statistical methods for summarize the evidence and provide an assessment of quality of included studies.