University of Ioannina
Αποτίμηση σχέσεων αιτιότητας σε μετα-αναλύσεις τυχαιοποιημένων κλινικών δοκιμών
Abstract
dc:descriptionThe main aim of this doctoral thesis is the identification, systematic evaluation and the prospect of generalization of methodologies and of specific effect metrics which allow for a causal interpretation of the results from meta-analyses of randomized clinical trials (RCTs). Three separate studies were carried out on both methodological and empirical evaluation, with specific objectives (a) the identification and implementation of a causally explicit framework for meta–analyses, (b) the evaluation of the credibility of potentially causal associations from meta-analyses of RCTs and (c) the description and quantification of known causal associations using a clinically relevant metric. In the first research paper, a systematic review of methodologies pertaining to the implementation of a causally explicit framework for meta-analyses of RCTs was performed. A total of three distinct methodological approaches were identified. Such methodologies have begun to emerge in the literature only during the last couple of years. Application of such methodologies may allow for the interpretation of meta-analysis results as causal estimates and could potentially lead to higher levels of evidence. However, methodological limitations stemming from the assumptions regarding the data generation process and sampling of the RCTs may hamper any attempt for a valid causal interpretation or generalization of the estimands. The second research paper investigated the possibility of an empirical evaluation of potential causal relationships by implementing a series of complementary analyses in order assess the epidemiological credibility of associations from meta-analyses of RCTs. Omega-3 fatty acid supplementation in relation to the reduction of the risk of “hard” cardiovascular outcomes was used as a clinical example. At this stage, the disadvantage of evaluating all interventions as an overall treatment group, regardless of the dosing regimen, emerged. Omega-3 supplements at low doses (<2 g/day) did not appear to be associated with cardiovascular outcomes. This conclusion seems unlikely to change from future research, as it is supported by sufficient scientific evidence. In contrast, the evidence for higher doses was limited and any emerging benefit of high-dose omega-3 supplementation should be further evaluated. The final research paper evaluated the empirical evidence base of the top 15 bestselling drugs using both meta-analyses of RCTs and individual RCTs. The Number Needed to Treat (NNT) metric was used as a clinically relevant measure of effect. Overall, two separate drug profiles emerged, focusing either a) on efficacy, where the observed beneficial NNTs on efficacy were remarkably small, or b) on safety and ease of use, with much higher beneficial NNTs on efficacy but with good safety profiles and without any serious adverse effects. However, despite the abundance of available clinical comparisons, the high heterogeneity between the studied populations along with the moderate quality of the evidence make hamper any attempts to generalize NNTs while maintaining a causal interpretation. Overall, the present thesis highlights a) the challenges of describing causal associations using meta-analyses and b) the fact that although meta-analysis techniques fail to explicitly demonstrate causality, the possibility such associations holding can be supported by strong epidemiological evidence. However, accounting for potential sources of heterogeneity is crucial in terms of generalizability and clinical significance of the findings.
Degree
thesis:*- Grantor dc:publisher
- University of Ioannina
- Year dc:date
- 2020
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Markozannes, Georgios
- Μαρκοζάννες, Γεώργιος
Subjects
dc:subject × 18- Αιτιότητα
- Αιτιώδης συμπερασματολογία
- Μετα-ανάλυση
- Τυχαιοποιημένες κλινικές δοκιμές
- Ετερογένεια
- Causality
- Causal inference
- Meta-analysis
- Randomized clinical trials
- Heterogeneity
- Επιστήμες Υγείας
- Ιατρική και Επιστήμες Υγείας
- Επιδημιολογία
- Επιστήμες υγείας, άλλοι τομείς
- Health Sciences
- Medical and Health Sciences
- Epidemiology
- Health sciences, miscellaneous
Rights
dc:rights- Statement dc:rights
-
- BY_NC
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/48822