Reporting Quality And Statistical Reporting Quality Of Observational Studies In The Journal Of High SCI Factors In Urology

Objective(1)To evaluate the reporting quality of observational studies published by the field of urology with high impact factors,explore the existing report defects and the factors that affect the report quality.(2)To evaluate the quality of statistical reporting of observational studies published by the field of urology with high impact factors,and explore the deficiencies of statistical reporting and the factors that affect the quality of statistical reporting.(3)To analyses the types,number and trends of statistical methods used in observational studies published in the field of urology with high impact factors.(4)To provide rational suggestions and reporting guidelines for medical researchers to correctly select statistical methods and clearly,completely report content and statistical content.Methods(1)The method of cross-sectional study is adopted in this paper,the five urology journals were selected according to the impact factors of Web of Science in 2018:Journal Of Urology(IF=5.157),BJU International(IF=4.688),Prostate Cancer And Prostatic Diseases(IF=4.099),Prostate(IF=3.820).Based on inclusion and exclusion criteria,all observational studies published in five journals in the PubMed were retrieved from January 1,2008 to September 31,2019,retrieval language in English.(2)Two investigators independently sifted through the literatures based on inclusion and exclusion criteria and evaluate the reporting quality and statistical reporting quality of included articles according to the STROBE statement(The Strengthening the Reporting of Observational Studies in Epidemiology Statement)and the statistical reporting quality evaluation checklist.(3)General information extracted included:name of journal,publication time,origin of corresponding author,type of study,number of authors,impact factor,funding support,the affiliation of corresponding author,international collaborative authorship,and participation of statistician or epidemiologist et al.(4)Theχ~2 test or Fisher exact probability method was used to compare the composition ratios and rates between different groups,the t test and analysis of variance were used to compare the mean between different groups,and the Mann-Whitney U test and Kruskal-Wallis H test were used to analyze the variables with skewed distribution.Radar charts are used to describe the report status of STROBE statements and statistical report quality checklists for different types of studies.After the STROBE score and statistical report quality score were divided into high quality group and low quality group,Logistic regression analysis was used to analyze the factors affecting the quality of STROBE and statistical report.Data entry and analysis were performed using SPSS 19.0.Results1.Basic characteristics of the included studiesA total of 8605 related documents were found in the initial search.After screening,a total of 193 related documents were included in the study,of which 31(16.06%)were from European Urology,54(27.98%)were from Journal of Urology,and 47(24.35%)From BJU International,26(13.47%)from Prostate Cancer and Prostatic Diseases and 35(18.13%)from European Urology.79(40.93%)were cohort studies,58(30.05%)were cross-sectional studies,and 56(29.02%)were case-control studies.2.Reporting quality of included studies:(1)The median reporting rate for all items declared by STROBE is 74.62%.The reporting rate of 3 items reached 100%:item 2(make a scientific explanation of the research background),item 4(describe the factors of the study design earlier in the paper),item 22(give the funding source of the current study and funders).The reporting rate of11 items/sub-items(32.35%)exceeded 95%,and the reporting rate of 10 items(29.41%)was less than 50%.The five items with the lowest reporting rate are:absolute risk(1.04%),treatment of lost follow-up(2.59%),treatment of missing values(8.81%),number of missing values(9.84%),and the reasons for the selecting of independent variables(11.92%).(2)The range of STROBE score of included studies was 9.53-20.73,with an average score of 15.69±2.25.The STROBE scores of the cohort studies and the case-control studies were 1.01 and 0.92 points higher than those of the cross-sectional studies,(P<0.05).Articles with statisticians or epidemiologists had 1.07 points higher than those without(P<0.05).(3)The results of multiple linear regression showed that the STROBE statement score was statistically significant in the following influencing factors(P<0.05):case-control studies(MD=0.839,95%CI:0.039-0.042,P=0.039),cohort studies(MD=0.815,95%CI:0.077-1.553,P=0.031),publication time(MD=0.141,95%CI:0.026-0.056,P=0.017),participation of statistician or epidemiologist(MD=0.969,95%CI:0.335-1.602,P=0.003),number of statistical methods(MD=0.760,95%CI:0.176-1.374,P=0.015).(4)Multivariate Logistic regression results showed that the quality of studies reporting with participation of professional epidemiological or statistical was higher(OR=1.82,95%CI:1.05-3.67,P=0.048);cohort studies(OR=2.48,95%CI:1.06-6.41,P=0.042)and case-control studies(OR=3.63,95%CI:1.35-9.76,P=0.011)are better than cross-sectional studies.3.Types,quantities and trends of statistical methods used in the included studies:(1)Of the 193 articles,6(3.1%)did not use any statistical methods or used only descriptive statistics.36(18.7%)articles only used simple statistical methods such as t-test,chi-square test,completely random design analysis of variance,et al.(2)Logistic regression(93,48.2%)is the most commonly used statistical method in observational research,followed by Chi-square test(90,46.6%).(3)The median of the statistical methods included in the article was 2.0(2.0),with a maximum of 9 and a minimum of 0.(4)The trend of statistical methods in the past decade is that the rating of complex statistical methods is always greater than that of simple statistical methods,but there is no obvious trend change in past years.4.Reporting quality of included studies(1)Outliers(2,1.0%)and sample size(13,6.7%)are often ignored in statistical reporting.Among the literatures using univariate analysis,17.1%(7)reported whether the t test condition was met,and 4.0%(1)reported whether the condition of variance analysis was met.Among the articles using regression analysis,29.1%(43)reported whether there was an interaction,3.4%(5)reported collinearity of independent variables,and 10.1%(15)reported inclusion of independent variables.(2)The statistical reporting quality score of the literatures included in the study ranged from 0.11-0.82,with an average score of 0.42.Only 6(3.1%)articles scored higher than 0.75,and 132(68.4%)articles have statistical reportig quality scores lower than 0.5.(3)The results of multiple linear regression showed that the reporting quality scores were statistically significant in the number of statistical methods(MD=0.070,95%CI:0.028-0.113,P=0.001).(4)Multivariate Logistic regression results showed that the quality of literature reports with epidemiological or statistical professional participation was higher(OR=1.73,95%CI:1.18-3.39,P=0.019),and observational studies with a number of statistical methods greater than 2 has a higher statistical reporting quality(OR=2.71,95%CI:1.50-4.91,P<0.001),and the quality of statistical reporting of articles with authors greater than or equal to 7 also improved significantly(OR=2.06,95%CI:1.04-4.08,P=0.038).ConclusionThe reporting quality of observational studies in urology is acceptable.The use of statistical methods is mainly Logistic regression andc~2 test.In recent years,the use of statistical methods has not been diversified and complicated,and the quality of statistical reports is relatively inadequate.The deficiencies in literature reports of urinary observational research are mainly concentrated in the methods and results;among them,the deficiencies are mainly reflected in several items such as sample size,missing data,flowcharts,and absolute risk.There is a lot of room for improvement in the quality of statistical reports.The shortcomings of t-tests and analysis of variance are mainly reflected in the failure to report their distribution assumptions.The shortcomings of non-parametric tests are mainly the reasons for the non-parametric tests not reported.The report also has significant shortcomings.There is a positive correlation between the report quality of the thesis and the quality of the statistical report.The two complement each other,and improving one will promote the other.The participation of epidemiology or statistics professionals,the type of study design as cohort studies or case-control studies are related to the high report quality of observational studies;the quality of statistical reports of literatures with epidemic or statistics professionals participation and the number of authors equal to or greater than 7 improved significantly.