| ObjectiveTo set up a scale for assessing the statistical quality of randomized controlled trial article and standard of score by expert consultation and group discussion. Stratified random sampling technique was used to select articles to assess its validity, reliability and acceptability. The scale was used to evaluate the statistical quality of randomized controlled trials from journal of Zhengzhou University (Medical Sciences) and to provide the reference for medical editor.MethodsThe item pool was obtained through reviewing literature material and discussing in group. The scale for assessing the statistical quality of randomized controlled trial article was obtained by expert consultation. Stratified random sampling method was used to select samples. According to the survey findings and expert consultation, delphi method, variation coefficient method and related coefficient was used to select scale items. Finally, the scale for assessing the statistical quality of randomized controlled trial article contained 3 dimensions and 20 items.Stratified random sampling technique was used to select 50 non-core periodical articles,50 core periodical articles and 34 SCI articles. Internal consistency reliability, test-retest reliability, rater reliability and discriminant validity was calculated.Cluster sampling was used to select samples from journal of Zhengzhou University (Medical Sciences). This scale was used to evaluate the statistical quality and understand the state of them.All data were input into Epidata 3.0, the analysis tool was SPSS16.0.α=0.05. The main statistical methods used including variation degree analysis, correlation analysis, Kappa coherency tests, Cronbach coefficientα, consistence examination, analysis of variance and ROC curves.Results1. The scale for assessing the statistical quality of randomized controlled trial article contained 3 dimensions and 20 items. Statistical design contained 8 items, statistical process contained 5 items and statistical results contained 7 items. 2. The results of the investigation showed that the internal consistency alpha coefficient was 0.801, Kappa coherency tests indicated good test-retest reliability and rater reliability. The Kappa uniformity coefficients of each items were larger than 0.75. The result of paired t test showed there was no statistically difference.3. Three types of periodicals could be distinguished(F=173.81, P<0.001). The score of non-core journal articles, core journal articles and SCI articles were 60.20±4.68,65.68±8.44 and 86.91±6.06 separately. The results of correlation analysis revealed that the correlation coefficient was 0.948.4. The recovery of scale was 100%. The time that the researchers completed the scale was between 9 minutes and 15 minutes. The mean time was 12.03±1.26 minutes. The acceptability of the scale was good.5. The results of the investigation of the journal of Zhengzhou University (Medical Sciences) entries randomized controlled clinical trial papers showed that the statistical quality were between 40 points and 78 points. The average was 57.28 points, and the standard deviation was 7.689. The scores for hired was 60 points or more than 60 points. The sensitivity was 0.714, the specificity was 0.872 and the Youden's index was 0.586.Conclusion1. This study established a scale for assessing the statistical quality of randomized controlled trial articles, and it contained 3 dimensions (statistical design, statistical process and statistical results) and 20 items. The results of the investigation showed that the reliability, validity and acceptability of the scale were good.2. The results of assessing the randomized controlled clinical trials from journal of Zhengzhou University (Medical Sciences) show statistical quality scores follow a normal distribution statistics. ROC curve analysis show that the article of 60 points or more than 60 points can be considered as accepted papers. |
PDF Full Text Request