Reporting quality of randomized controlled trial abstracts presented at European Orthodontic Society congresses

Transcription

1 European Journal of Orthodontics, 2016, doi: /ejo/cjv094 Advance Access publication 27 December 2015 Original article Reporting quality of randomized controlled trial abstracts presented at European Orthodontic Society congresses Fang Hua, Tanya Walsh, Anne-Marie Glenny and Helen Worthington Cochrane Oral Health Group, School of Dentistry, The University of Manchester, UK Correspondence to: Fang Hua, Cochrane Oral Health Group, School of Dentistry, The University of Manchester, JR Moore Building, Oxford Road, Manchester M13 9PL, UK. Summary Objectives: To assess the reporting quality of randomized controlled trial (RCT) abstracts presented at the European Orthodontic Society (EOS) congresses, investigate any improvement after the release of CONSORT (CONsolidated Standards Of Reporting Trials) for Abstracts guidelines, and identify factors associated with reporting quality. Methods: Abstract books of the (Pre-CONSORT period) and (Post-CONSORT period) EOS congresses were obtained from the official website of European Journal of Orthodontics. A hand-search was conducted to identify RCTs. Reporting quality was assessed and scored using the original 17-item CONSORT for Abstracts checklist. Risk ratio and the t-test were used to compare the adequate reporting rate of each item and the overall quality in two periods, respectively. Univariate and multivariate regressions were used to identify predictors of reporting quality. Results: A total of 138 RCT abstracts were included and assessed. The mean overall CONSORT score was 4.10 (SD: 1.32) and 4.48 (1.31) in the Pre- and Post-CONSORT samples, respectively [P = 0.099; mean difference, 0.39 (95% CI: 0.84, 0.07)]. Only three CONSORT items ( objective, interventions, and conclusions ) were adequately reported in most abstracts (>75 per cent). No abstract provided information regarding the corresponding author, trial registration, and source of funding. Less than 1.5 per cent of the included abstracts sufficiently reported randomization, recruitment, and outcome in the results section. In the multivariate analysis, greater word count (P = 0.036) and provision of the exact P value (P = 0.006) were significantly associated with higher reporting quality. Limitations: Our final regression model can explain only about 8 per cent of the variance of reporting quality. Other predictors not included in this study may be investigated in analogous studies. Conclusions: The reporting quality of RCT abstracts presented at EOS congresses was suboptimal. Joint efforts by authors and conference committees to improve reporting are needed. Introduction Poor reporting is a breach of the ethical framework of medical research conduct (1, 2), and an avoidable waste of invested time and resources (3). A previous article indicated that at least 50 per cent of research reports in biomedicine were sufficiently poor and therefore unusable, representing a waste of tens of billions of pounds (4). Such a high amount of waste warrants research into the current reporting quality of medical research, and active implementation of the findings to improve reporting and reduce relevant research waste (3). The Author Published by Oxford University Press on behalf of the European Orthodontic Society. All rights reserved. For permissions, please journals.permissions@oup.com 584

2 F. Hua et al. 585 High-quality, well-designed, and well-conducted randomized controlled trials (RCTs) are considered evidence of the highest grade in the hierarchy of research designs (5) and the gold standard to investigate benefits and harms of health care interventions (6). Frequently, the results of RCTs are first presented at medical conferences in the form of abstracts (7). Often included in meta-analyses (8) and clinical textbooks (9), these abstracts can exert profound influence on the evidence base and clinical practice. In addition, abstracts presented at conferences can be the only available reports of the corresponding studies. Livas et al. (10) found that only 52.2 per cent of abstracts presented at two European Orthodontic Society (EOS) congresses were published in full, and a Cochrane review (11) suggested that only 57.5 per cent of RCT abstracts presented at biomedical conferences reached full publication. Therefore a well-written conference abstract should contain sufficient details about the study and help readers (clinicians, systematic reviewers, policy makers etc.) assess the reliability and relevance of the findings (12). In 2008, an extension of the CONSORT (CONsolidated Standards Of Reporting Trials) statement was released to guide the reporting of RCTs in journal and conference abstracts (12). However, after the publication of these guidelines, the reporting quality of RCT abstracts presented at medical specialty conferences remained suboptimal (13, 14). To our knowledge, the reporting quality of RCT abstracts presented at dental conferences has not been assessed. Therefore, the objectives of this study were to 1. assess the reporting quality of RCT abstracts presented at EOS congresses; 2. investigate any improvement after publication of the CONSORT for Abstracts guidelines; and 3. identify factors associated with abstract reporting quality. Data extraction The following data and descriptive information were extracted from each abstract as potential predictors of reporting quality: presentation type (oral versus poster), number of authors, continent of origin (first author), publication date, word count, significance of the main finding (positive versus negative), and reporting of the exact P value (e.g. P = rather than P < 0.05). Information regarding abstract authors (number of authors, continent of origin) was obtained after the quality assessment process. Significance of each abstract s main finding was determined by FH and one of the other authors independently and in duplicate, with all discrepancies resolved by discussion. A positive main outcome was defined as a statistically significant difference between the experimental treatment and the control treatment, or a stated preference for either treatment arm (11). Statistical analysis Risk ratio (RR) was used to compare the adequate reporting rate of each CONSORT item and sub-item in two periods (Pre- versus Post-CONSORT). Improvements in mean OCS by characteristics were assessed with t-tests. Additionally, univariate and multivariate linear regression analyses were performed to determine the association between reporting quality (OCS) and potential predictors. We carried out the univariate analyses first, and then individually entered all significant predictors from the univariate analyses into the multivariable modelling. Significant violation of normality was not indicated in assessment of residuals. For the multivariate analysis, tolerance and variance inflation factor (VIF) were used to detect multicollinearity. Any predictor with a tolerance less than 0.1 and/ or VIF above 10 was excluded from the final model (18). Statistical significance was defined as P < Materials and methods Collection and selection of abstracts As the CONSORT for Abstracts guidelines were released in January 2008 (12), we aimed to assess RCT abstracts presented before or at least 2 years after this date, allowing for the dissemination and uptake of these guidelines (15, 16). Therefore, we retrieved the abstract books of the (Pre-CONSORT period) and (Post-CONSORT period) EOS congresses from the official website of European Journal of Orthodontics (ejo.oxfordjournals.org). A hand-search was conducted to identify RCT abstracts. Predefined inclusion criteria were: human participants, interventions associated with health care, experimental studies, presence of a control group, as well as randomization of participants to the study/control group (15). All included abstracts were compiled into a standardized Word document with all identifiers (author name and affiliation) removed to ensure blinded quality assessment (17). Assessment of reporting quality Quality assessment was carried out by FH and one of the other authors independently and in duplicate, using the original 17-item CONSORT for Abstracts checklist and associated explanations (12). Discrepancies were resolved by discussion. An individual item was scored 1 if the item was reported adequately and 0 if the reporting was inadequate. Then for each abstract, scores for all 17 items were totalled to calculate an overall CONSORT score (OCS; score range: 0 to 17). In addition, reporting of 11 sub-items of applicable CONSORT items, as suggested in the associated explanations of CONSORT for Abstracts (12), was also documented to provide supplementary information. Results Characteristics of included abstracts A total of 138 RCT abstracts were identified and included. The number of RCT abstracts presented at EOS congresses increased by 70.6 per cent from the Pre-CONSORT (n = 51) to Post-CONSORT period (n = 87). Most of the included abstracts were presented as posters (69.6 per cent), by authors from Europe (80.4 per cent), in 200 to 300 words (69.6 per cent), and with a positive main outcome (60.1 per cent) (Table 1). Reporting of general items Assessment results of the reporting of each CONSORT item and sub-item are shown in Table 2 and Figure 1. For both periods, only about one-fourth of the abstracts (Pre-CONSORT, 25.5 per cent; Post-CONSORT, 27.6 per cent) can be identified as randomized through their titles, and less than one-sixth (Pre-CONSORT, 15.7 per cent; Post-CONSORT, 16.1 per cent) described the trial design. None of the included abstracts provided contact details of the authors, information regarding trial registration, or the source of financial support. Reporting of trial methodology Most abstracts adequately reported the CONSORT item objective (Pre-CONSORT, 96.1 per cent; Post-CONSORT, 93.1 per cent) and interventions (Pre-CONSORT, 82.4 per cent; Post-CONSORT, 90.8 per cent). However, only 21.6 per cent of the Pre-CONSORT abstracts and 32.2 per cent of the Post-CONSORT abstracts clearly defined the primary outcome. For details about participants,

3 586 European Journal of Orthodontics, 2016, Vol. 38, No. 6 Table 1. Characteristics of included abstracts. Characteristic Category Pre-CONSORT Post-CONSORT Overall (N = 51), n (%) (N = 87), n (%) (N = 138), n (%) Presentation type Oral 19 (37.3) 23 (26.4) 42 (30.4) Poster 32 (62.7) 64 (73.6) 96 (69.6) Continent Europe 40 (78.4) 71 (81.6) 111 (80.4) Asia 6 (11.8) 7 (8.0) 13 (9.4) Oceania 3 (5.9) 6 (6.9) 9 (6.5) Others 2 (3.9) 3 (3.4) 5 (3.6) Word count <200 4 (7.8) 6 (6.9) 10 (7.2) (35.3) 27 (31.0) 45 (32.6) (39.2) 31 (35.6) 51 (37.0) >300 9 (17.6) 23 (26.4) 32 (23.2) No. of authors 3 33 (64.7) 30 (34.5) 63 (45.7) (35.3) 57 (65.5) 75 (54.3) Exact P value Yes 12 (23.5) 33 (37.9) 45 (32.6) No 39 (76.5) 54 (62.1) 93 (67.4) Trial outcome Positive 30 (58.8) 53 (60.9) 83 (60.1) Negative 21 (41.2) 34 (39.1) 55 (39.9) although eligibility criteria were reported in most abstracts (Pre- CONSORT, 84.3 per cent; Post-CONSORT, 80.5 per cent), mentioning of the data collection setting was rare (Pre-CONSORT, 9.8 per cent; Post-CONSORT, 6.9 per cent). For information regarding randomization, although most abstracts mentioned random allocation, only one abstract (0.7 per cent) provided information about allocation concealment. Besides, among all included abstracts only 14 (10.1 per cent) provided information about blinding. Only one abstract from the Pre-CONSORT sample (2.0 per cent) and nine from the Post-CONSORT sample (10.3 per cent) specified who were blinded. Reporting of trial results The number of participants randomized was reported in over half of the included abstracts (Pre-CONSORT, 58.8 per cent; Post- CONSORT, 58.6 per cent). But only one Pre-CONSORT abstract (2.0 per cent) and nine Post-CONSORT abstracts (10.3 per cent) provided the number of participants analysed in each group. Only one abstract indicated adoption of the intention-to-treat or per-protocol analysis. Result for the primary outcome in each group was provided significantly more often during the Post-CONSORT period (RR = 8.79, 95% CI = 1.20, 64.61), but still this information was given by only 15 Post-CONSORT abstracts (17.2 per cent). Only 2 abstracts from the Post-CONSORT sample (2.3 per cent) provided both the effect size and its precision for the primary outcome. Additionally, for both periods, adverse events or side-effects were reported only by about 17 per cent of the abstracts. Reporting of trial conclusions In 76.5 per cent of the Pre-CONSORT abstracts and 81.6 per cent of the Post-CONSORT abstracts the conclusions were reported adequately and consistent with the trial results. But only 17.6 per cent of the Pre-CONSORT sample and 12.6 per cent of the Post- CONSORT sample provided in the conclusions section both benefits and harms of the trial. Overall CONSORT score Table 3 shows the mean OCS by period and by characteristics. The mean OCS of abstracts presented during the Post-CONSORT period (mean ± SD, 4.48 ± 1.31) was slightly higher than that of abstracts presented during the Pre-CONSORT period (4.10 ± 1.32), with no statistical significance (mean difference, 0.39; 95% CI: 0.84, 0.07; P = 0.099). Factors associated with reporting quality In the univariate analyses, only word count (P = 0.026) and reporting of the exact P value (P = 0.004) were found to be significantly associated with reporting quality. We entered these two predictors into the multivariate analysis. Both greater word count (B = 0.01; 95% CI: 0.00, 0.01; P = 0.036) and provision of the exact P value (B = 0.65; 95% CI: 0.19, 1.11; P = 0.006) remained significant predictors of higher reporting quality. However, the relatively low R 2 (8.9 per cent) and adjusted R 2 (7.6 per cent) values of the final model indicated that the OCS is likely to be influenced by other factors not included in this model (Table 4). Discussion In the present study, 138 RCT abstracts presented at 10 EOS congresses were identified and assessed. The overall reporting quality of these abstracts was suboptimal, and did not improve significantly after publication of the CONSORT for Abstracts guidelines. Only three CONSORT quality items ( objective, interventions, and conclusions ) were adequately reported in most abstracts (>75 per cent). No abstract provided information regarding the corresponding author, trial registration, and funding sources. Less than 1.5 per cent of the included abstracts sufficiently reported randomization, recruitment, and outcome in the results section, and less than 10 per cent adequately reported participant, blinding, and numbers analysed. Reporting of orthodontic trial abstracts Previously, Fleming et al. (19) assessed the reporting quality of RCT abstracts published in four leading orthodontic journals during In their study, four CONSORT items including interventions, objective, numbers randomized, and conclusions were reported adequately by the majority of included abstracts (>93 per cent); trial registration and funding were never reported; and other information including adverse events, allocation concealment,

4 F. Hua et al. 587 Table 2. Reporting of each CONSORT checklist item and sub-item by publication period (12). Items Criteria and sub-items Pre-CONSORT (N = 51), n (%) Post-CONSORT (N = 87), n (%) Risk ratio (95% CI) 1. Title Identification of the study as randomized 13 (25.5) 24 (27.6) 1.08 (0.61, 1.93) 2. Authors Contact details for the corresponding author 0 (0) 0 (0) NE*** 3. Trial design Description of the trial design (e.g. parallel, 8 (15.7) 14 (16.1) 1.03 (0.46, 2.28) cluster, non-inferiority) 4. Participant Eligibility criteria for participants and the settings 5 (9.8) 5 (5.7) 0.59 (0.18, 1.93) where the data were collected 4a. Eligibility criteria for participants 43 (84.3) 70 (80.5) 0.95 (0.82, 1.12) 4b. Settings of data collection 5 (9.8) 6 (6.9) 0.70 (0.23, 2.19) 5. Interventions Interventions intended for each group 42 (82.4) 79 (90.8) 1.10 (0.96, 1.27) 6. Objective Specific objective or hypothesis 49 (96.1) 81 (93.1) 0.97 (0.89, 1.05) 7. Outcome 1* Clearly defined primary outcome for this report 11 (21.6) 28 (32.2) 1.49 (0.81, 2.73) 8. Randomization How participants were allocated to interventions 0 (0) 1 (1.1) NE 8a. Random assignment 48 (94.1) 82 (94.3) 1.00 (0.92, 1.09) 8b. Sequence generation 1 (2.0) 3 (3.4) 1.76 (0.19, 16.46) 8c. Allocation concealment 0 (0) 1 (1.1) NE 9. Blinding (masking) Whether or not participants, care givers, and 1 (2.0) 9 (10.3) 5.28 (0.69, 40.44) those assessing the outcomes were blinded 9a. Generic description only (e.g. single-blind, 0 (0) 4 (4.6) NE double-blind) 10. Numbers randomized Number of participants randomized to each 30 (58.8) 51 (58.6) 1.00 (0.75, 1.33) group 11. Recruitment Trial status 1 (2.0) 1 (1.1) 0.59 (0.04, 9.17) 12. Numbers analysed Number of participants analysed in each group 1 (2.0) 9 (10.3) 5.28 (0.69, 40.44) 12a. Intention-to-treat analysis or per-protocol 0 (0) 1 (1.1) NE analysis 13. Outcome 2** For the primary outcome, a result for each group 0 (0) 2 (2.3) NE and the estimated effect size and its precision 13a. Primary outcome result for each group 1 (2.0) 15 (17.2) 8.79 (1.20, 64.61) 13b. Estimated effect size 0 (0) 4 (4.6) NE 13c. Precision of the estimate (e.g. CI 95%) 0 (0) 3 (3.4) NE 14. Harms Important adverse events or side-effects 9 (17.6) 15 (17.2) 0.98 (0.46, 2.07) 15. Conclusions General interpretation of the results 39 (76.5) 71 (81.6) 1.07 (0.89, 1.28) 15a. Benefits and harms balanced 9 (17.6) 11 (12.6) 0.72 (0.32, 1.61) 16. Trial registration Registration number and name of trial register 0 (0) 0 (0) NE 17. Funding Source of funding 0 (0) 0 (0) NE *Outcome reported in Materials and Methods section. **Outcome reported in Results section;. ***NE: Not estimable due to zero cell counts. blinding, and confidence intervals were rarely described (<10 per cent). This pattern of inadequate reporting is generally in keeping with our findings, suggesting a lack of awareness of the CONSORT for Abstracts guidelines among orthodontic researchers (19). However, there are also some differences between Fleming et al. s (19) assessment results and ours. For instance, in the present study the adequate reporting rate was much lower for numbers randomized (59 per cent) and higher for harms (17 per cent). These could be attributed to the differences in requirements and peerreview processes between orthodontic conferences and orthodontic journals. On one hand, with the use of more detailed instructions for authors and a stricter, usually iterative peer-review process, editors, and reviewers of orthodontic journals have better chance to help improve the reporting of submitted RCT abstracts. A former study showed that active implementation of the CONSORT for Abstracts guidelines had led to better abstract reporting in leading medical journals (20). In fact, some orthodontic journals such as the European Journal of Orthodontics, AJO-DO and Progress in Orthodontics have already endorsed the CONSORT for Abstracts guidelines. Moreover, recently the editors of AJO-DO have published an adaptation of the CONSORT guidelines for orthodontic trials, including guidance on trial abstract reporting (21). On the other hand, the word limit for abstracts is usually smaller for journals than for conferences. At the time of Fleming et al. s (19) study, leading orthodontic journals had a word limit of either 200 or 250 for abstracts; while the word limit for recent EOS congresses has been In fact, in our sample, 60.2 per cent of the conference abstracts had more than 250 words, and 23.2 per cent had more than 300 words. This means that, before submission for full publication, authors may have to shorten their conference abstracts to meet journals requirements. In this process, without knowledge about the CONSORT for Abstracts checklist, authors may delete some key details (e.g. adverse events) while keeping trivial information (e.g. statistical software used). Nevertheless, before making any comparison between Fleming et al. s (19) results and ours, readers should note the methodological differences between these two studies. In the present study, the original 17-item CONSORT for Abstracts checklist and its official explanations (12) were used to assess quality of reporting. Like in most similar research (7, 16, 17, 22 26), a simple dichotomous scoring

5 588 European Journal of Orthodontics, 2016, Vol. 38, No. 6 currently there is no guideline for conducting assessments of reporting quality. The interpretation of some items of the CONSORT for Abstracts checklist (e.g. outcome, recruitment, participant, harms, and conclusions ) could be contentious and further clarification may be necessary (29). In addition, direct comparison among reporting quality assessment studies is also complicated by the difficulty in determining the stringency with which relevant reporting criteria were applied in each study (19). Since a conference abstract could be the only available report of a trial (11), readers (especially systematic reviewers) need the contact details of authors and registration information to retrieve additional data and assess the selective reporting of outcomes (12). Besides, as previous research has shown that studies funded by industry tend to have outcomes favouring sponsors (30), and that conference abstracts are more likely to report interim analyses than full papers (31), sources of funding and the trial status are also essential for assessments of trials. However, in this study, reporting of these items was either absent or extremely rare. Nevertheless, it is worth noting that absence of authors contact details could be due to the current EOS policy only, and therefore should be relatively straightforward to improve. Figure 1. Reporting quality of each CONSORT item and sub-item. *Sub-items of applicable CONSORT items. system (0 = no or inadequate; 1 = adequate) was used to quantify reporting quality. In contrast, Fleming et al. (19) assessed the quality by referring to a 21-item checklist modified from the CONSORT for Abstracts, and used an ordinal system (0 = no description; 1 = inadequate; 2 = adequate) for quality categorization. Reporting of trial abstracts in conferences Since 2010, there have been several studies assessing the reporting quality of RCT abstracts presented at conferences in other fields of medicine, using the CONSORT for Abstracts checklist (7, 14, 24 28). Among these, about five studies provided the adequate reporting rate of each CONSORT quality item (Table 5). The findings of these studies are generally comparable with ours. Adequate reporting of objective, interventions, conclusions, and absence of information for authors, trial registration, and funding seem to be universal across medical specialties. It also seems that, compared with our sample, conference abstracts from other medical specialties did better in the reporting of participant, trial design, blinding, and outcome in the results section. However, it is hard to determine whether such differences were due to true differences in abstract reporting or differences in the interpretation of the CONSORT for Abstracts guidelines. Despite the existence of reporting guidelines (e.g. CONSORT, PRISMA), Factors associated with reporting quality To our knowledge, so far there has been no study assessing the association between potential factors and the reporting quality of RCT abstracts presented at medical conferences. Through univariate and multivariate linear regression analyses, we found that greater word count and provision of the exact P value were significantly associated with better abstract reporting. These findings are consistent with previous studies regarding RCT abstracts in cancer nursing journals (32) and general dental journals (15), respectively. Besides, our results suggested that continent of origin, number of authors, and significance of main outcome were not significant predictors of reporting quality. These are also in keeping with previous research on the reporting of RCT abstracts in orthodontic (19), dental (15, 29, 33), and medical (16, 32) journals. Restriction on word count has been commonly considered a key constraint for detailed reporting (15). On one hand, a higher word limit means an opportunity to provide more information; but on the other, greater word count does not guarantee good reporting quality (32). The CONSORT for Abstracts guidelines stated that words would be adequate to address all items in the checklist (12). The word limit for recent EOS congress abstracts has been , which should be enough to adequately report an abstract of orthodontic trial (19). However, in this study, 76.8 per cent of the abstracts were less than 300 words and 39.8 per cent were less than 250 words. This highlights the importance that conference committees adopt a sufficient word limit in their online instructions, and authors make full use of the space allowed when preparing their abstracts. Reason for the significant association between provision of the exact P value and better reporting quality is unknown. One explanation could be that it is a proxy for authors knowledge about reporting statistics, abstracts, and/or trials, since reporting of the exact P value has been recommended by classic statistical guidelines (34), classic guidelines for reporting abstracts (35), and the CONSORT guidelines for reporting trials (36). Besides, in this study, both linear regression and the t-test indicated that the reporting quality did not improve significantly after the release of the CONSORT for Abstracts guidelines. Judging by the RRs for each CONSORT quality item, only one sub-item (primary outcome

6 F. Hua et al. 589 Table 3. Mean overall CONSORT score (OCS) by characteristics. Characteristic Category Mean OCS ± SD Difference P value Pre-CONSORT Post-CONSORT Mean 95% CI (t-test) Presentation type Oral 4.21 ± ± ( 1.49, 0.08) Poster 4.03 ± ± ( 0.87, 0.27) Continent Europe 4.22 ± ± ( 0.85, 0.20) Asia 3.50 ± ± ( 2.05, 0.19) Oceania 3.33 ± ± ( 3.25, 1.91) Others 4.50 ± ± ( 4.39, 5.39) Word count < ± ± ( 0.87, 2.03) ± ± ( 0.87, 0.54) ± ± ( 1.55, 0.04) > ± ± ( 1.45, 1.03) No. of authors ± ± ( 0.93, 0.47) ± ± ( 1.37, 0.03) Exact P value Yes 4.50 ± ± ( 1.20, 0.38) No 3.97 ± ± ( 0.81, 0.31) Trial outcome Positive 4.07 ± ± ( 1.16, 0.12) Negative 4.14 ± ± ( 0.83, 0.47) Overall 4.10 ± ± ( 0.84, 0.07) Table 4. Univariate and multivariate linear regression derived coefficients (B) and 95% confidence intervals, with overall CONSORT score (OCS) as the dependent variable for the included 138 abstracts. VIF, variance inflation factor. Univariate Multivariate* Predictor Category/unit B 95% CI P value B 95% CI Tolerance VIF P value Presentation type Oral Reference Poster 0.37 ( 0.85, 0.12) Continent Europe Reference Asia 0.43 ( 1.20, 0.33) Oceania 0.66 ( 1.56, 0.25) Others 0.23 ( 1.43, 0.96) Publication date Pre-CONSORT Reference Post-CONSORT 0.39 ( 0.07, 0.84) No. of authors 1 person 0.01 ( 0.21, 0.19) Word count 1 word 0.01 (0.00, 0.01) (0.00, 0.01) Exact P value No Reference Reference Yes 0.68 (0.22, 1.14) (0.19, 1.11) Trial outcome Negative Reference Positive 0.14 ( 0.31, 0.60) *For multivariate analysis, constant = 2.886, R 2 = 0.089, adjusted R 2 = 0.076, P = result for each group) was reported significantly better during the Post- CONSORT period. This finding is generally in keeping with a former study regarding trial abstracts at a urology meeting (13). Previously, Hopewell et al. s (20) study found that reference to the CONSORT for Abstracts guidelines in instructions to authors, combined with an active policy to implement these guidelines, has led to significant increase in the reporting quality of RCT abstracts in leading medical journals. However, to our knowledge, the association between endorsement of CONSORT for Abstracts in conference instructions and the reporting quality of conference RCT abstracts has not been studied. The question of how to improve the reporting of conference abstracts remains to be explored, preferably by conference committees and researchers collaboratively using prospective and controlled studies (37). Limitations and strengths This study has several limitations. First, as there is usually a time lag of months between the submission of abstracts and their presentation at conferences, the submitted abstracts may be different from their subsequent version being disseminated in oral/poster sessions (38). However, we argue that abstract books (or conference proceedings) are more widely disseminated since they are the only permanent records of the presented studies, which will be available after conferences for both conference attendees and other stakeholders. This highlights authors responsibility to ensure the quality of their abstracts before submission to scientific conferences. Second, through univariate and multivariate linear regression analyses we found that word count and provision of the exact P value were significant predictors of the reporting quality of conference abstracts. But according to the R 2 values, our final model can only explain about 8 per cent of the variation of OCS. Other factors, such as the number of centres (single centre versus multi-centre), total sample size and involvement of statisticians may also influence reporting quality (29, 39, 40). However, these are beyond the scope of this study and could be investigated in future research. In addition, our

7 590 European Journal of Orthodontics, 2016, Vol. 38, No. 6 Table 5. Comparison of assessment results between this study and previous similar research regarding conferences in other fields of medicine. Study ID Current study Yoon (2012) (25) Yoon (2012) (24) Knobloch (2011) (26) Mann (2011) (14) Autorino (2010) (7) Area Orthodontics Sports injury Sports injury Burn Gerontology Endourology Sample size Items Adequate reporting rate (%) 1. Title Authors Trial design Participant *** Interventions Objective Outcome 1* Randomization Blinding Numbers randomized Recruitment Numbers analysed Outcome 2** *** Harms Conclusions Trial registration Funding *Outcome reported in Materials and Methods section. **Outcome reported in Results section. ***The lowest adequate reporting rate of the corresponding sub-items (as only the reporting rates of sub-items were provided). OCS assumed that each of the 17 CONSORT items were equally important. This may not always be the case, but the relative importance of each item depends on the reader s perspective (20). Besides, the method of equally weighting CONSORT items has been widely used in previous similar research (15, 16, 20, 24 26, 32, 39). Despite these limitations, our study has several strengths. To our knowledge this is the first study to assess the reporting quality of abstracts presented at dental conferences, and the first to investigate predictors of the reporting quality of conference trial abstracts using linear regression analyses. One hundred and thirty-eight RCT abstracts presented at 10 EOS congresses, across a 12-year time span, were included. Quality assessment was carried out using the original CONSORT for Abstracts checklist. Implications for conference committees The EOS congresses and other dental conferences have been and will continue to be major platforms for communication among clinicians, researchers, students, educators, policymakers, as well as industry. We recommend that the organizing committees of these conferences endorse the CONSORT for Abstracts guidelines in their online instructions for authors. For example, this could be done by adding a paragraph specifically for authors submitting abstracts describing RCTs, which: 1. Introduces the CONSORT for Abstracts checklist with reference to relevant publications (12) and webpages ( 2. Requires the contact detail (e.g. address) of at least one corresponding author. 3. Requires information regarding trial recruitment, trial registration (register and registration number), and source of funding. 4. Sets a reasonable word limit for trial abstracts (at least 300 is recommended). Conference committees should also ensure the publication of the above-mentioned details together with their corresponding trial abstracts, in the conference proceedings/abstract books. Implications for authors of abstracts Complete, clear, and transparent reporting is a moral obligation of all researchers (1, 2). Besides, a conference abstract could be the only report of a trial that is available to health care professionals and other stakeholders (11, 12). Therefore, when preparing a trial abstract for conference presentation, authors should familiar themselves with the CONSORT for Abstracts guidelines, make full use of the space allowed by the conference committee, and try to incorporate key information as much as possible within the word limit. Implications for conference attendees At the level of primary research, RCTs are the best studies to inform clinical decisions. However, the quality of RCTs could be compromised by poor design and/or reporting. Thus, before applying the findings of an RCT into practice, clinicians need to appraise the RCT s reliability and relevance (41, 42). The results of this study and previous research have shown that many key details essential to critical appraisal were under-reported or unreported in RCT abstracts presented at some dental and medical conferences. Therefore, when faced with a conference abstract describing an RCT, conference attendees, and other readers should use the CONSORT for Abstracts checklist (12) to initially assess the abstract, and then decide whether to carry out a critical appraisal directly or retrieve additional information first. Conclusions The reporting quality of RCT abstracts presented at EOS congresses was suboptimal.

8 F. Hua et al. 591 No evidence suggested significant improvement in reporting since the publication of CONSORT for Abstracts guidelines. Greater word count and provision of the exact P value were found to be significantly associated with better reporting quality. Joint efforts by authors and conference committees to improve the reporting of RCT abstracts presented at conferences are needed. Funding President s Doctoral Scholar Award from the University of Manchester to FH. References 1. Needleman, I., Moher, D., Altman, D.G., Schulz. K.F., Moles. D.R. and Worthington H. (2008) Improving the clarity and transparency of reporting health research: a shared obligation and responsibility. Journal of Dental Research, 87, Moher D. (2007) Reporting research results: a moral obligation for all researchers. Canadian Journal of Anaesthesia, 54, Glasziou, P., Altman, D.G., Bossuyt, P., Boutron, I., Clarke, M., Julious, S., Michie, S., Moher, D., Wager, E. (2014) Reducing waste from incomplete or unusable reports of biomedical research. Lancet, 383, Chalmers, I. and Glasziou, P. (2009) Avoidable waste in the production and reporting of research evidence. Lancet, 374, Concato, J., Shah, N. and Horwitz, R.I. (2000) Randomized, controlled trials, observational studies, and the hierarchy of research designs. The New England Journal of Medicine, 342, Clancy, M.J. (2002) Overview of research designs. Emergency Medicine Journal, 19, Autorino, R., Borges, C., White, M.A., Altunrende, F., Perdona, S., Haber, G.P., De Sio, M., Khanna, R., Stein, R.J., Kaouk, J.H. (2010) Randomized clinical trials presented at the World Congress of Endourology: how is the quality of reporting? Journal of Endourology, 24, Cook, D.J., Guyatt, G.H., Ryan, G., Clifton, J., Buckingham, L., Willan, A., Mcllroy, W., Oxman, A.D. (1993) Should unpublished data be included in meta-analyses? Current convictions and controversies. JAMA, 269, Bhandari, M., Devereaux, P.J., Guyatt, G.H., Cook, D.J., Swiontkowski, M.F., Sprague, S., Schemitsch, E.H (2002) An observational study of orthopaedic abstracts and subsequent full-text publications. The journal of Bone Joint Surgery American Volume, 84-A, Livas, C., Pandis, N. and Ren, Y. (2014) Full-text publication of abstracts presented at European Orthodontic Society congresses. European Journal of Orthodontics, 36, Scherer, R.W., Langenberg, P. and von Elm, E. (2007) Full publication of results initially presented in abstracts. Cochrane Database Syst Rev (2):MR Hopewell, S., Clarke, M., Moher, D., Wager, E., Middleton, P., Altman, D.G., Schulz, K.F., CONSORT Group (2008) CONSORT for reporting randomized controlled trials in journal and conference abstracts: explanation and elaboration. PLoS Medicine, 5, e De Sio, M., et al. (2012) Reporting quality of abstracts presented at the European Association of Urology meeting: a critical assessment. Journal of Urology, 188, Mann, E. and Meyer, G. (2011) Reporting quality of conference abstracts on randomised controlled trials in gerontology and geriatrics: a cross-sectional investigation. Zeitschrift für Evidenz, Fortbildung und Qualität im Gesundheitswesen, 105, Hua, F., Deng, L., Kau, C.H., Jiang, H., He, H. and Walsh, T. (2015) Reporting quality of randomized controlled trial abstracts: survey of leading general dental journals. Journal of the American Dental Association (1939), 146, e Ghimire, S., Kyung, E., Lee, H. and Kim, E. (2014) Oncology trial abstracts showed suboptimal improvement in reporting: a comparative before-andafter evaluation using CONSORT for Abstract guidelines. Journal of Clinical Epidemiology, 67, Hopewell, S., Boutron, I., Altman, D.G. and Ravaud, P. (2015) Deficiencies in the publication and reporting of the results of systematic reviews presented at scientific medical conferences. Journal of Clinical Epidemiology, 68, Norman, G. and Streiner, D. (2008) Biostatistics: The Bare Essentials. BC Decker Inc., Hamilton, Ontario, 3rd edn. 19. Fleming, P.S., Buckley, N., Seehra, J., Polychronopoulou, A. and Pandis, N. (2012) Reporting quality of abstracts of randomized controlled trials published in leading orthodontic journals from 2006 to American Journal of Orthodontics and Dentofacial Orthopedics, 142, Hopewell, S., Ravaud, P., Baron, G. and Boutron, I. (2012) Effect of editors implementation of CONSORT guidelines on the reporting of abstracts in high impact medical journals: interrupted time series analysis. BMJ, 344, e Pandis, N., Fleming, P.S., Hopewell, S. and Altman, D.G. (2015) The CONSORT Statement: Application within and adaptations for orthodontic trials. American Journal of Orthodontics and Dentofacial Orthopedics, 147, Turpen, R.M., Fesperman, S.F., Smith, W.A., Vieweg, J. and Dahm, P. (2010) Reporting quality and information consistency of randomized, controlled trials presented as abstracts at the American Urological Association annual meetings. Journal of Urology, 184, Faggion, C.M., Jr and Giannakopoulos, N.N. (2012) Quality of reporting in abstracts of randomized controlled trials published in leading journals of periodontology and implant dentistry: a survey. Journal of Periodontology, 83, Yoon, U. and Knobloch, K. (2012) Assessment of reporting quality of conference abstracts in sports injury prevention according to CONSORT and STROBE criteria and their subsequent publication rate as full papers. BMC Medical Research Methodoloy, 12, Yoon, U. and Knobloch, K. (2012) Quality of reporting in sports injury prevention abstracts according to the CONSORT and STROBE criteria: an analysis of the World Congress of Sports Injury Prevention in 2005 and British Journal of Sports Medicine, 46, Knobloch, K., Yoon, U., Rennekampff, H.O. and Vogt, P.M. (2011) Quality of reporting according to the CONSORT, STROBE and Timmer instrument at the American Burn Association (ABA) annual meetings 2000 and BMC Medical Research Methodology, 11, Knobloch, K., Yoon, U. and Vogt, P.M. (2010) Clinical trial quality and reporting quality in American Society of Plastic Surgeons and German conference abstracts: a transatlantic perspective. Plastic and Reconstructive Surgery, 126, 281e 282e. 28. Knobloch, K., Yoon, U. and Vogt, P.M. (2011) Randomized controlled trials and consolidated standards of reporting trials reporting quality: an evidence-based analysis of the American Society of Plastic Surgeons, American Burn Association, and American Society for Surgery of the Hand conference meetings. Plastic and Reconstructive Surgery, 127, 175e 176e. 29. Seehra, J., Wright, N.S., Polychronopoulou, A., Cobourne, M.T. and Pandis, N. (2013) Reporting quality of abstracts of randomized controlled trials published in dental specialty journals. Journal of Evidence Based Dental Practice, 13, Lexchin, J., Bero, L.A., Djulbegovic, B. and Clark, O. (2003) Pharmaceutical industry sponsorship and research outcome and quality: systematic review. BMJ, 326, Hopewell, S., Clarke, M. and Askie, L. (2006) Reporting of trials presented in conference abstracts needs to be improved. Journal of Clinical Epidemiology, 59, Guo, J.W. and Iribarren, S.J. (2014) Reporting quality for abstracts of randomized controlled trials in cancer nursing research. Cancer Nursing, 37, Kiriakou, J., Pandis, N., Madianos, P. and Polychronopoulou, A. (2014) Assessing the reporting quality in abstracts of randomized controlled trials

9 592 European Journal of Orthodontics, 2016, Vol. 38, No. 6 in leading journals of oral implantology. Journal of Evidence Based Dental Practice, 14, Altman, D.G., Gore, S.M., Gardner, M.J. and Pocock, S.J. (1983) Statistical guidelines for contributors to medical journals. British medical journal (Clinical Research Ed.), 286, Haynes, R.B., Mulrow, C.D., Huth, E.J., Altman, D.G. and Gardner, M.J. (1990) More informative abstracts revisited. Annals of Internal Medicine, 113, Moher, D., Hopewell, S., Schulz, K.F., Montori, V., Gotzsche, P.C., Devereaux, P.J., Elbourne, D., Egger, M., Altman, D.G. (2010) CONSORT 2010 explanation and elaboration: updated guidelines for reporting parallel group randomised trials. BMJ, 340, c Stevens, A., et al. (2014) Relation of completeness of reporting of health research to journals endorsement of reporting guidelines: systematic review. BMJ, 348, g Buchan, J.C. and Spokes, D.M. (2010) Do recorded abstracts from scientific meetings concur with the research presented? Eye (London, England), 24, Sandhu, S.S., Sandhu, J. and Kaur, H. (2015) Reporting quality of randomized controlled trials in orthodontics-what affects it and did it improve over the last 10 years? European Journal of Orthodontics, 37, Pandis, N., Polychronopoulou, A. and Eliades, T. (2010) An assessment of quality characteristics of randomised control trials published in dental journals. Journal of Dentistry, 38, Brignardello-Petersen, R., Carrasco-Labra, A., Glick, M., Guyatt, G.H. and Azarpazhooh, A. (2015) A practical approach to evidence-based dentistry: III: How to appraise and use an article about therapy. Journal of the American Dental Association, 146, e Richards, D. (2009) Critically appraising randomised trials. Evidence- Based Dentistry, 10,