Sarah Khan edited untitled.tex  almost 9 years ago

Commit id: 75c745f0df9b684477a1eae4332a3947bd099314

deletions | additions      

       

\section{Abstract and Key Words}  \subsection{{\textbf{Aim:}}}To evaluate the reporting and utilization  of methodological quality measures in systematic reviews addressing low quality  and meta-analyses risk of bias  in major  oncology journals. \subsection{\textbf{Methods:}}  We formulated an 11-item quality measure abstraction manual for systematic reviews and meta-analyses. We performed a systematic search for the articles through PubMed on the 18 and 26 of May 2015. Covidence was used to screen articles based on the title and abstract. The methodological quality and reporting of risk of bias were evaluated by three rounds of coding from two independent reviewers using the same checklist.Studies that were excluded include meta-analyses that were not the focus of the study,scoping reviews,case reports, individual patient data meta-analyses, and narrative reviews. Differences in assessment were resolved through group consensus between the two independent reviewers. Data analyses were conducted through Microsoft Excel 2007, and STATA 13.1 software.   \subsection{\textbf{Results:}}  258 articles were identified to meet initial search criteria. Quality assessment was conducted on 182 articles after exclusion of ineligible articles. Quality and risk of bias assessment was (50 percent) of articles assessed. Most common tools used were those that were of some unique method created by the author (27.47 adapted from other sources (24.47  percent, n=25) and those n=25/91), author  independently assessedby authors  (20.88 percent, n=19). n=19/91) and unspecified (13.19 percent, n=12/91).  In assessing risk of bias, high/medium/low scale was used most (18.97 percent, n=11), n=11/58) followed by high/medium/unclear (13.79 percent, n=8/58),  and quality was assessed through author created scales (29.31 percent, n=17). n=17/58) and the Jadad scale (15.52 percent, n=9/58). Low Quality or High risk of bias studies were found in 46 studies with 76 percent (n=35/46) including those studies.From included studies, subgroup analysis was conducted in 17.58 percent, meta regression in 8.79 percent, and sensitivity analysis in 17.58 percent. In 37 studies reporting of low quality or high risk of bias was unknown. Quality measures were articulated largely in narrative format (47.25 percent, n=43/91) or not at all (39.56 percent, n=36/91).  \subsection{\textbf{Conclusions:}}  Quality and risk of bias were assessed in half of the systematic reviews and meta-analyses coded, however methods of assessment are determined by authors independently rather than following well known scales such as Jadad and Newcastle-Ottawa Scale.High risk of bias and low quality studies were included in most of these studies.This analysis provides further evidence of the lack of consistency in reporting quality measures for clinical findings in the field of oncology. Differences between assessment of bias and quality reporting could negatively impact the clinical application of findings in oncology journals.  \subsection{\textbf{Keywords:}}