Article type
Year
Abstract
Background: studies published as systematic reviews/meta-analyses (SR/MAs) on nutrition in cancer prevention may face problems similar to those documented in other fields, that is, poor overall quality, but no study has systematically examined this.
Aims: the overall aim of this research is to examine the risk of bias (ROB) of articles published as SR/MAs on nutritional interventions in cancer prevention. We documented the ROB using the ROBIS instrument, and examined the associations between characteristics of SR/MAs and their ROB.
Methods: we searched the Cochrane Library, MEDLINE and Embase from 2010 to 2018 for SR/MAs. Eligible for inclusion were articles identified as SR/MAs in the title or abstract that included primary studies with a control group (randomized and non-randomized studies) assessing the effects of any nutritional intervention in the prevention of cancer incidence or mortality in the general population or people at risk for cancer.
Two review authors independently screened titles, abstracts and full texts. We resolved conflicts by discussion and if required by an arbiter. We designed and pilot tested extraction forms, which enabled us to independently retrieve all population, intervention, comparator and outcome data among eligible reviews. Our abstraction tool used the ROBIS (domains including study eligibility criteria, identification and selection of studies, data collection and study appraisal, synthesis and findings) to evaluate ROB. We registered the protocol in PROSPERO (CRD42019121116).
Results: from 750 SR/MA included we randomly selected a sample of 101 articles using the RAND procedure in Excel. We are in the process of data extraction and evaluation of the ROB of the studies. Preliminary data on 17 studies extracted thus far show overall high ROB due to concerns identified in each ROB domain using the ROBIS instrument. The most common concerns identified included no reference to protocol (100% of reviewed studies), no search for unpublished studies (100%), no information or lack of duplicate selection of studies and data extraction (71% and 59%), lack of formal assessment of quality of primary studies (71%). Studies commonly measured statistical heterogeneity, but often combined different study designs in one meta-analysis (88%). The identified ROB issues were generally not addressed in the interpretation of findings. We will present full results at the Colloquium.
Conclusions: the project will bring important information about overall ROB of SR/MAs addressing nutritional interventions in cancer prevention.
Improvement of a patient-focused health outcome: our systematic survey focuses on patient-important cancer outcomes. Summary findings of SR/MAs that are at high ROB may be misleading for users’ wishing to optimally understand the best summary evidence for cancer prevention.
Funding: Project funded by National Science Centre, No. UMO-217/25/B/NZ7/01276
Aims: the overall aim of this research is to examine the risk of bias (ROB) of articles published as SR/MAs on nutritional interventions in cancer prevention. We documented the ROB using the ROBIS instrument, and examined the associations between characteristics of SR/MAs and their ROB.
Methods: we searched the Cochrane Library, MEDLINE and Embase from 2010 to 2018 for SR/MAs. Eligible for inclusion were articles identified as SR/MAs in the title or abstract that included primary studies with a control group (randomized and non-randomized studies) assessing the effects of any nutritional intervention in the prevention of cancer incidence or mortality in the general population or people at risk for cancer.
Two review authors independently screened titles, abstracts and full texts. We resolved conflicts by discussion and if required by an arbiter. We designed and pilot tested extraction forms, which enabled us to independently retrieve all population, intervention, comparator and outcome data among eligible reviews. Our abstraction tool used the ROBIS (domains including study eligibility criteria, identification and selection of studies, data collection and study appraisal, synthesis and findings) to evaluate ROB. We registered the protocol in PROSPERO (CRD42019121116).
Results: from 750 SR/MA included we randomly selected a sample of 101 articles using the RAND procedure in Excel. We are in the process of data extraction and evaluation of the ROB of the studies. Preliminary data on 17 studies extracted thus far show overall high ROB due to concerns identified in each ROB domain using the ROBIS instrument. The most common concerns identified included no reference to protocol (100% of reviewed studies), no search for unpublished studies (100%), no information or lack of duplicate selection of studies and data extraction (71% and 59%), lack of formal assessment of quality of primary studies (71%). Studies commonly measured statistical heterogeneity, but often combined different study designs in one meta-analysis (88%). The identified ROB issues were generally not addressed in the interpretation of findings. We will present full results at the Colloquium.
Conclusions: the project will bring important information about overall ROB of SR/MAs addressing nutritional interventions in cancer prevention.
Improvement of a patient-focused health outcome: our systematic survey focuses on patient-important cancer outcomes. Summary findings of SR/MAs that are at high ROB may be misleading for users’ wishing to optimally understand the best summary evidence for cancer prevention.
Funding: Project funded by National Science Centre, No. UMO-217/25/B/NZ7/01276