Democritus University of Thrace (DUTH)
Σύγχρονες στατιστικές μέθοδοι μετά-ανάλυσης (network meta-analysis) για την αξιολόγηση της ασφάλειας και της αποτελεσματικότητας θεραπευτικών παρεμβάσεων σε χρόνια νοσήματα και τον έλεγχο πιθανών προγνωστικών παραγόντων
Abstract
dc:descriptionThe scope of this Ph.D. thesis was the evaluation of the effectiveness of treatments reported by published randomized clinical trials, primarily in the field of vascular surgery, that often lack direct comparisons between treatments, using the methods of "network meta-analysis" (NMA) and "umbrella review".The past few years, the synthesis of evidence through NMA methods has be-come increasingly popular, showing an exponential increase in the number of publications employing NMA. Network meta-analysis is an extension of the meta-analysis method and allows the evaluation of multiple treatments in a single analysis, combining direct and indirect evidence. In the general part of this thesis, the methodology of the conventional meta-analysis is described, followed by an extensive presentation of the methodology of NMA, as well as the "umbrella review". Next, practical applications of the aforementioned methods was performed and yielded three studies on the field of vascular surgery. The first study concerns carotid endarterectomy closure techniques, aiming to re-evaluate the potential of different closure techniques. The meta-analysis concludes that reverse endarterectomy (EvE) is associated with a reduced risk of early postoperative combined 30-day stroke and death rate when compared with other arterial occlusion methods, apart from PTFE and bovine pericardium patching. The SUCRA analysis indicated that the EvE method was associated with the highest probability of being the best closure method. The second study compared the efficacy of first-line endovascular therapies for the treatment of arteriovenous fistula stenosis and concluded that DCBA ranks as the best treatment option in terms of 6-month failure rate, followed by CtBA. Finally, an umbrella review was performed to compare simple angioplasty versus paclitaxel-coated balloon in arteriovenous fistula and graft stenosis. Its purpose is to summarize and to enhance evidence reliability, given the growing number of existing meta-analyses and systematic reviews on this topic. The implementation of network meta-analysis studies engaged the random effects model that included the following steps: direct comparison of treatments, indirect comparison and a combination of direct and indirect comparison, as well as the Stata routines. In addition, the predictive intervals (PIs), which are important tool that predicts the actual treatment effect to be expected in future studies, were also calculated. In the study that compares carotid endarterectomy closure techniques, PIs in-creased confidence in comparisons of reverse endarterectomy versus polyurethane and Dacron grafts, an important finding for future analyzes. In five statistically significant results of the "umbrella review", the estimation of PIs included null, except from the assessment of TLPP at 12 months. The thesis demonstrated that the use of modern methods, such as network me-ta-analysis and "umbrella review", for evidence synthesis are attractive methodologies for the evaluation of multiple treatments, the combination of both direct and indirect comparisons of these treatments in a single analysis, aspects that are useful for re-searchers, clinicians and decision makers across the clinical domain.
Degree
thesis:*- Grantor dc:publisher
- Democritus University of Thrace (DUTH)
- Year dc:date
- 2024
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Christaina, Eleni
- Χρήσταινα, Ελένη
Subjects
dc:subject × 14- Μετα-ανάλυση δικτύου
- Network meta-analysis
- Μαθηματικά
- Φυσικές Επιστήμες
- Μαθηματικά, άλλοι τομείς
- Επιστήμες Υγείας
- Ιατρική και Επιστήμες Υγείας
- Επιστήμες υγείας, άλλοι τομείς
- Mathematics
- Natural Sciences
- Mathematics (miscellaneous)
- Health Sciences
- Medical and Health Sciences
- Health sciences, miscellaneous
Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/58090