Article type
Year
Abstract
Background: A number of 'claimed' randomized controlled trials (RCTs) were published with a high speed increasing over time in China and always with positive results. However, their quality was frequently oppugned. Objectives: We conducted an investigation with the aim to identify how many 'real' RCTs there were actually in 'claimed' RCTs and to assess the quality of the 'real' RCT reports.
Methods: Twenty-one investigators were trained in relevant knowledge, communication skills and quality control. Twenty-four commonly encountered diseases were selected and we searched the China National Knowledge Infrastructure (CNKI) database (from 1994 to June, 2005). We interviewed by telephone the authors of these 'claimed' RCTs about randomization methods and related quality features of trials, and used this information to judge the credibility of 'claimed' RCTs. We further assessed the report quality of 'real' RCTs using the CONSORT (Consolidated Standards of Reporting Trials) items.
Results: Three thousand one hundred and thirty-seven 'claimed' RCTs were included for telephone interview. Finally, only 207 studies were identified as 'real' RCTs (6.8%, 95%CI 5.9 to 7.7). There was no statistically significant difference in quality between trials of Traditional Chinese Medicine (TCM) and Western Medicine (WM). 'Real' RCTs were conducted in medical university affiliated hospitals more frequently than in level three and level two hospitals (RR 1.58, 95%CI 1.18 to 2.13 and RR 14.42, 95%CI 9.40 to 22.10, respectively), and level three hospitals were higher than level two hospitals (RR 9.32, 95%CI 5.83 to 14.89); all trials of new drugs were identified as 'real' RCTs; trials funded by government and other official sources of support had rates of 51.6%, (95%CI 39.2 to 64.1) and 56.3% (95%CI 32 to 81.0), respectively; more than half of studies (56.0% (116/207), 95%CI 49.2 to 62.8) reported CONSORT items unsatisfactorily with the rate of 3% to 35%, only 10 RCTs (4.8%, 95%CI 1.9 to 7.7) had high rates of reporting of 70% to 82% of CONSORT items.
Conclusions: Most Chinese authors of 'claimed' RCTs did not know exactly about the randomization approaches but incorrectly used the term 'randomization' to claim non-RCTs; few authors have the knowledge of RCTs but incorrectly used the term 'randomization' to claim non-RCTs due to an absence of serious scientific attitude; critical registration and an auditing system can result in well-designed RCTs. Most of the 'real' RCTs were conducted by medical university affiliated hospitals. Selection bias is an important interpretable reason of false positive results always in Chinese claimed RCTs.
Methods: Twenty-one investigators were trained in relevant knowledge, communication skills and quality control. Twenty-four commonly encountered diseases were selected and we searched the China National Knowledge Infrastructure (CNKI) database (from 1994 to June, 2005). We interviewed by telephone the authors of these 'claimed' RCTs about randomization methods and related quality features of trials, and used this information to judge the credibility of 'claimed' RCTs. We further assessed the report quality of 'real' RCTs using the CONSORT (Consolidated Standards of Reporting Trials) items.
Results: Three thousand one hundred and thirty-seven 'claimed' RCTs were included for telephone interview. Finally, only 207 studies were identified as 'real' RCTs (6.8%, 95%CI 5.9 to 7.7). There was no statistically significant difference in quality between trials of Traditional Chinese Medicine (TCM) and Western Medicine (WM). 'Real' RCTs were conducted in medical university affiliated hospitals more frequently than in level three and level two hospitals (RR 1.58, 95%CI 1.18 to 2.13 and RR 14.42, 95%CI 9.40 to 22.10, respectively), and level three hospitals were higher than level two hospitals (RR 9.32, 95%CI 5.83 to 14.89); all trials of new drugs were identified as 'real' RCTs; trials funded by government and other official sources of support had rates of 51.6%, (95%CI 39.2 to 64.1) and 56.3% (95%CI 32 to 81.0), respectively; more than half of studies (56.0% (116/207), 95%CI 49.2 to 62.8) reported CONSORT items unsatisfactorily with the rate of 3% to 35%, only 10 RCTs (4.8%, 95%CI 1.9 to 7.7) had high rates of reporting of 70% to 82% of CONSORT items.
Conclusions: Most Chinese authors of 'claimed' RCTs did not know exactly about the randomization approaches but incorrectly used the term 'randomization' to claim non-RCTs; few authors have the knowledge of RCTs but incorrectly used the term 'randomization' to claim non-RCTs due to an absence of serious scientific attitude; critical registration and an auditing system can result in well-designed RCTs. Most of the 'real' RCTs were conducted by medical university affiliated hospitals. Selection bias is an important interpretable reason of false positive results always in Chinese claimed RCTs.
PDF