Evaluation Of Biopsy Randomized Controlled Trials

BackgroundWith the development of evidence-based medicine, reporting bias has received widespread recognition. It affects the results of trials and leads to a reduction in the strength of evidence for conducting systematic reviews and developing clinical practice guidelines.As the impotortant part of complementary and alternative therapies, acupuncture has been widely used in the world. At the present, many clinical trials have been conducted to assess the effect and safety of acupuncture, and a large number of randomized clinical trials (RCTs) have been published. Therefore, it’s of significance to conduct research focusing on these literatures.How about the status quo and the existing problems about methodological qualtity and reporting quality of acupuncture RCTs published in English? How about the prevalence and factors related to positive results of acupuncture RCTs? Does outcome reporting bias exist in the registered RCTs on acupunture? How about views on reporting bias of editors and peer reviewers from complementary and alternative medicine (CAM) journals in China? These questions are all worthy of further study.ObjectivesTo explore the statue quo of acupuncture RCTs published in English by bibliometrics, and to assess the prevalence of positive results and examine factors potentially influencing the results to provide reference for acupuncture researcher to understand the development and the existing problems of acupuncture; to identify outcome reporting bias in the regeristed acupuncture RCTs, and to investigate views on reporting bias of editors and peer reviewers from CAM journals in China, in order to provide reference for taking measures to reduce reporting bias.Methods1 Literature searchingThree English databases including PubMed, EMBASE and Cochrane Library were systematically searched from their inception to 28th Feburary 2014 to identify acupuncture RCTs published in English. Furthermore, a systematic search of 15 registries was conducted to identify registered records of the included acupuncture RCTs.2 Bibliometric analysesWe conducted bibliometric analysis of the included trials, and explored the status quo and the potential problems.3 Critical appraisalsWe assessed the reporting quality of the included trials by using the checklist of Standards for Reporting Interventions in Controlled Trials of Acupuncture (STRICTA). The Cochrane Risk of Bias was used to assess the methodological quality of the included trials, including random sequence generation, allocation concealment, blinding, incomplete outcome reporting, selective reporting and other sources of bias.4 Questionnaire surveyA questionnaire about reporting bias was developed and sent to editiors and peer reviewers from CAM journals in China.Results1 RCTs on acupuncture published in English:a bibliometric analysis and assessment of methodological and reporing qualityA totally of 855 acupuncture RCTs (involving 109,698 participants) were included and published from 1975 to 2014. The sample size ranged from 6 to 14,161 with an average of 128.30 per trial. The number of publications showed a tendency of sustained growth, especially after 2006. The included trials originated from 35 different countries, with most coming from China (n=223,26.08%) and USA (n=113,13.22%). All of the trials were published in 258 different journals, of which 338(39.53%) were published in CAM journals and only 53(6.20%) appeared in high impact-factor Journals (IF>10). The majority were single-center (n=563, 65.85%), parallel-group (n=784,91.70%), two-arms (n=591,69.12%), superiority design (n=694,81.17%) and placebo/sham-controlled (n=503,58.83%). The most frequent condition was musculoskeletal system and connective tissue (n=186, 21.75%). Only 111(12.98%) provided the information on trial registration. Only 368(43.04%) specified the primary outcome and secondary outcome.695(81.29%) reported clinical relevant outcomes while only 20.70%(177/855) and 6.08% (52/855) reported safety outcomes and health-economic outcomes, respectively. Generally, the methodological information of included trials was insufficient. Only 49.71%(425/855),38.95%(333/855),66.55%(569/855) reported generation of allocation sequence, allocation concealment and blinding, respectively. Incomplete outcome reporting and selective outcome reporting existed in 10.76% (92/855) and 14.74% (126/855), respectively. Furthermore, the reporting quality of acupuncture intervention was poor. Only three articles provided detailed information on all the items of STRICTA. The least reported item was "Extent to which treatment was varied" (n=123,14.39%).2 Prevalence and factors related to positive results of acupuncture RCTs published in EnglishAmong the included 855 articles, positive results were identified in 68.65% (587/855). The results of trials from Thailand (n=4), Greece (n=3), Belgium (n=2), Croatia (n=2), Egypt (n=1), Hungary (n=1), Netherlands (n=1) and Singapore (n=1) were all positive, followed by Spain (7/8,87.50%) and China (191/223,85.68%). In univariate analysis, positive results were associated with publication year, country origin, journal type, number of centers, adequate allocate concealment and comparison type (P<0.05). However, journal impact factor, sample size, trial registration, random sequence generation, blinding and funding did not affect the positive results (P>0.05). In multivariate analysis, positive results of acupuncture RCTs were associated with country origin, journal type, number of centers and comparison type (P<0.05).3 To identify outcome reporting bias in regeristed RCTs on acupunctureWe searched 15 major international trial registries to identify the registered records of 855 included trials based on the registration number, the first author, the corresponding author or title. A total of 120 RCTs on acupuncture with 137 published reports were identified. The majority of trial protocols were registered in either Clinicaltrials.gov (57/120,47.50%) or ISRCTN (37/120,30.83%). Only 19.17%(23/120) were prospective registration while 80.83%(97/120) were retrospective registration, and 36 trials were registered even after the completion of the study. The trial registration number was unavailable in 27 published reports (19.71%) and the trial registration number was incorrect in 5 published reports (3.65%). A comparison of registered and published primary outcomes was conducted in 88 publications where outcomes were specified as primary outcome or secondary outcome. The primary outcomes of 35 articles (39.77%) were inconsistent with those specified in registered records. The discrepancies involved in a registered primary outcome omitted in the published article (20/88,22.73%), a registered primary outcome defined as a secondary outcome in the published article (12/88,13.64%), a published primary outcome registered as a secondary outcome (4/88,4.55%), an absent primary outcome in the registry defined as a primary outcome in published article (17/88, 19.32%), and different timing of assessment of primary outcome in the published article and the registry (14/88,15.91%). Furthermore,57.14% (20/35) had discrepancies that favored statistically significant primary outcomes. The other inconsistencies were identified in inclusion criteria (62.22%), allocate concealment(57.14%)and exclusion criteria (55.00%).4 Views on reporting bias of editors and peer reviewers from CAM journals in China: a questionnaire surveyA total of 89 editors and peer reviewers were surveyed from 10 CAM journals in China.83.15% (74/89) survey respondents have a higher recognition of trial registration, but only 65.17% (58/89),47.19% (42/89),48.31% (43/89) have a higher recognition of the conception, types and sources of reporting bias, respectively.57.30% (51/89) respondents would like to give preference to publish papers with positive results, and 87.64%(78/89) respondents would not like to giver preference to publish papers with negative results while 7.87% (7/89) considered the publication or non-publication of research findings depended on the nature and direction of the results. The respondents thought authors’main reasons for omitting outcomes were fear of the negative results rejected by journals (79/89,88.76%), poor study quality (64/89,71.91%), and the controversial results in view of the current knowledge in the field (38/89, 42.70%).15.73%(14/89) respondents thought space constraints might be the one of authors’reasons for inadequate reporting of trial results. Reporting bias has been encountered by 48 (53.93%) of the respondents, of which 8.99% (8/89) rejected those articles, and 43.82% (39/89) asked the authors to give the reason. All of the respondents thought it was significant for trial registration to decrease reporting bias. However,15.73%(14/89) expressed doubts that the current practice of trial registration and posting of trial results in public databases would decrease reporting bias. Furthermore, only 55.06% (49/89) respondents cross-checked the consistency between the submitted manuscripts and the registered records to identify any discrepancies. The resondents explained the main reasons for not check the consistency between the registered protocols and the subsequent publications were lack of time (17/25,68.00%), trust in authors (8/25,32.00%) and ignorance (6/25,24.00%). Only 43.82% (39/89) thought the journals encouraged trial registration, and the main reasons for not requiring trial registration of journals were:few trials registered in China (68/89,76.40%), fear of discouraging research submission (9/89,10.11%).65.17% (58/89) respondents suggested the journals should adopt trial registration policy.ConclusionsA lot of measures have been conducted to reduce reporting bias. However, the methodological quality and reporting quality of acupuncture RCTs are generally suboptimal, though the quantity of acupuncture RCTs is substantial and increasing. Positive results and selective outcome reporting of acupuncture RCTs are prevalent, and the change of primary outcomes intends to favor statistical significance. This study also showes inadequacies in the knowledge of editors and peer reviewers from CAM journals in China regarding reporting bias. Editors and peer reviewers doubt that the current practice of trial registration and posting of trial results in public databases would decrease reporting bias, and they are lack of awareness of cross-check the consistency between the submitted manuscripts and the registered records. To ensure publication of reliable and unbiased results, and then to reduce reporting bias, further promotion and implementation of trial registration, more stringent training and education for editors and peer reviewers are sitll needed.