1. Selection of journals and sample of abstracts of RCTs

We selected five leading Chinese medical journals indexed in MEDLINE from 1998 to 2007 with top ranking impact factor (IF) in each field based on the “2007 data for Sci-tech Journal Citation Reports of China”(Table 1) and with both English and Chinese abstracts: the National Medical Journal of China, Chinese Journal of Internal Medicine, Chinese Journal of Obstetrics and Gynecology, Chinese Journal of Pediatrics, and the Chinese Journal of Surgery.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Number of abstracts identified and included from five Chinese medical journals.

https://doi.org/10.1371/journal.pone.0011926.t001

The CBM (Chinese Biomedical Database) disk database was searched using an extended version of the Cochrane Highly Sensitive Search Strategy [5] (Table 2) to identify RCTs (searched April, 2008) for Chinese abstracts in each of the five leading journals. Based on the citation of each RCT, we hand searched these five printed journals for the English abstracts. We included all publications reporting RCTs,which was defined as a trial where the allocation of participants to interventions was described by the words random, randomly, randomized, or randomization in all disease areas and all types of interventions, dealing with patients or volunteers.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Search strategy to identify RCTs in five leading Chinese medical journals.

https://doi.org/10.1371/journal.pone.0011926.t002

2. Evaluation Method

We assessed the quality of reporting in abstracts of RCTs based on the CONSORT for Abstracts checklist [6]. Refined items from the CONSORT for Abstracts are presented in the Table 3. We assessed each item as “reported” or “not reported” according to whether the author reported all the contents listed in the refined items or not. We checked for any differences in reporting between the Chinese and English language version of the same abstract.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Adapted CONSORT for Abstracts checklist items reported in the 332 journal abstracts of RCTs.

https://doi.org/10.1371/journal.pone.0011926.t003

We determined the number and proportion (%) of RCT abstracts that reported each of the CONSORT for Abstracts checklist items. Four assessors (YL Chen, CL Ai, L Wang and YR Duan) independently rated each RCT abstract and inter-rater agreement for each checklist item was determined using the Kappa statistic. Overall, this produced good agreement between the reviewers (0.7). Discrepancies were resolved by consensus and or discussed. Data analysis was carried out using Excel 2003, Epidata 3.02 and Stata version 9.0.

1. Quality of reporting in Chinese abstracts

We identified 692 abstracts of RCTs during our initial search and subsequently determined that 332 pairs were eligible for analysis based on our criteria of findings being reported both Chinese and English (Figure 1). We have presented the results of our findings in Table 3. Overall only 10% (34/332) of the abstracts could be identified as an RCT based on the title. No abstracts mentioned the method of allocation concealment and, while 49 (15%) RCT abstracts reported on blinding, only one indentified who was blinded. Although 65 (20%) of the abstracts mentioned adverse events or side effects, less than 10% reported specific symptoms (n = 20) or rates (n = 26). Lastly, 246 (75%) abstracts did not report their sources of funding.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Identification of RCT abstracts from the Chinese Biomedical Database.

https://doi.org/10.1371/journal.pone.0011926.g001

2. Comparison the quality of reporting between the Chinese with English abstracts

The results of the English language version of the same RCT abstracts are also presented in Table 3. In terms of the proportion of RCT abstracts fulfilling a criterion, the absolute difference (percentage points) between the Chinese and English abstracts was 10% (ranging from 0 to 25%) on average, per item. Some items such as details of the primary outcome, were only reported in the English version of the abstract. Some items were complementary between the Chinese and English abstracts, for example, some Chinese abstracts only reported the “numbers randomized” while their corresponding English abstracts only reported the “numbers analyzed”.

1. General quality of reporting in abstracts and its possible reasons

Our study indicates that the quality of reporting in abstracts of RCTs published in these five leading Chinese medical journals is insufficient. Overall, the average number of reported items in each RCT abstract, based on the CONSORT for Abstracts checklist, was three. None of the abstracts adequately reported the key features of the trial design, including the method of generating the allocation sequence, concealment of the allocation, or the estimated effect size and its precision for the primary outcome. Less than half of the RCT abstracts reported the study as randomized in the title, the number of participants analysed, details of harms, trial registration and source of funding.

Berwanger and colleagues [7] evaluated the quality of reporting in 227 abstracts describing RCTs published in the New England Journal of Medicine, JAMA, the BMJ and the Lancet in 2006. In their paper, the authors identified that details regarding methodological quality were also poorly reported as only one RCT abstract described the method of allocation concealment; 21 (9%) clearly specified who was blinded; 51 (23%) described intention-to-treat analysis; and 32 (14%) outlined losses to follow-up.

In the mid-1990s, Anrong [8] and Bili [9] reported that the quality of abstracts in Chinese medical journals was poor. They suggested that poor abstract reporting may indicate suboptimal study design, inadequate reporting, or both. Previous work has shown that abstracts frequently under report key features of study design [10]–[11] and omit important results. It is suggested that the poor quality of reporting of Chinese abstracts may also be an indication of poor reporting in full-text RCTs in China [12]–[18]. When an item isn't reported in the full publication, it is likely to also be missing from the abstract. A lack of detailed requirement for abstracts in a journal's ‘Instruction for Authors’ is also a potential problem. Although most of the journals included in our study have adopted structured abstracts, the structured format alone is not sufficient to guide authors regarding their content. Thus, some of the most salient information might be missed if no detailed instructions for each heading are given. Finally, a lack of sufficient knowledge of research methodology might also attribute to the poor quality of reporting.

Abstracts should contain the most important information and accurate study results relevant to clinical practice. While we did find that most abstracts reported the purported “study benefits,” we also observed that the primary outcome, the estimated effect size, and the precision and harms of the study were rarely reported. Cumulatively, this lack of information hampers readers in making clinical decisions [19]–[20] and presents clinicians many difficulties when trying to evaluate the benefits and harms of any intervention. Unfortunately, only two RCT abstracts reported details of trial registration.

While, not included in our study, the Chinese Journal of Evidence-Based Medicine recently introduced the CONSORT for Abstracts guidelines into its ‘Instructions for Authors’ [21]. The application of the CONSORT Statement has demonstrated benefits in improving the quality of reporting in abstracts [22]–[24], and we hope that it will do the same with the introduction of CONSORT for Abstracts. We very much hope that the five journals included in our study will also consider including such guideline to their ‘Instructions for Authors’.

2. Limitations

This study has several limitations, firstly, we did not take a random sample of all RCT abstracts published in Chinese medical journals and therefore our findings may not reflect all abstracts of RCTs published in China or in other countries. Secondly, we did not compare abstracts with their corresponding full articles, However, previous studies have shown that the quality of reporting of the full-text is also poor in the same medical journals [12]–[18]. We further wish to point out that the WHO International Clinical Trial Register Platform only approved The Chinese Clinical Trial Register (ChiCTR) in 2007. Thus, it may well be that many researchers do not know the importance of prospective trial registration [25]. While not included in our current study, the Chinese Journal of Evidence-Based Medicine recently introduced the CONSORT for Abstracts guidelines into its ‘Instructions for Authors’ [21]. This adoption has improved the quality of reporting in other journals [22]–[24] and it is our sincere hope that the five journals included in our study, as well as other Chinese journals, adopt the CONSORT statement for Abstracts into their ‘Instructions for Authors.’ We believe that this adoption will markedly improve both the interpretation and practice of medicine in China in the future.

Conclusions

In summary, the quality of reporting in abstracts of RCTs published in these five leading Chinese medical journals requires substantial improvement to meet the recommendations set out in CONSORT for Abstracts guidelines. With the publication of CONSORT for Abstracts checklist, we suggest that Chinese medical journals should adopt these recommendations and do more to ensure that authors apply to meet internationally agreed standards, thereby allowing the conduct of their studies to be monitored and improved.