If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Completeness of Reporting Is Suboptimal in Randomized Controlled Trials Published in Rehabilitation Journals, With Trials With Low Risk of Bias Displaying Better Reporting: A Meta-research Study
Department of Health Sciences, Faculty of Science, Vrije Universiteit Amsterdam, Amsterdam Movement Sciences, the NetherlandsGIMBE Foundation, Bologna, Italy
Division of Occupational Medicine, IRCCS Azienda Ospedaliero-Universitaria di Bologna, Bologna, ItalyDepartment of Biomedical and Neuromotor Sciences (DIBINEM), Alma Mater Studiorum, University of Bologna, Bologna, Italy
GIMBE Foundation, Bologna, ItalyDepartment of Neuroscience, Rehabilitation, Ophthalmology, Genetics, Maternal and Child Health, University of Genoa, Genoa, Italy
Department of Clinical Science and Translational Medicine, University of Rome Tor Vergata, Rome, ItalyDepartment of Medicine and Health Science “Vincenzo Tiberio,” University of Molise, Campobasso, Italy
Department of Health Sciences, Faculty of Science, Vrije Universiteit Amsterdam, Amsterdam Movement Sciences, the NetherlandsDepartment of Epidemiology and Data Science, Amsterdam UMC, Location VUmc, Amsterdam Movement Sciences Research Institute, the Netherlands
Department of Health Sciences, Faculty of Science, Vrije Universiteit Amsterdam, Amsterdam Movement Sciences, the NetherlandsDepartment of General Practice, Erasmus MC, University Medical Center, Rotterdam, the Netherlands
Primary: To evaluate the completeness of reporting of randomized controlled trials (RCTs) published in rehabilitation journals through the evaluation of the adherence to the Consolidated Standards of Reporting Trials (CONSORT) checklist and investigate the relationship between reporting and risk of bias (ROB). Secondary: To study the association between completeness of reporting and the characteristics of studies and journals.
Data Sources
A random sample of 200 RCTs published between 2011 and 2020 in 68 rehabilitation journals indexed under the “rehabilitation” category in the InCites Journal Citation Report.
Study Selection
One reviewer evaluated the completeness of reporting operationalized as the adherence to the CONSORT checklist. Two independent reviewers evaluated the ROB using the Cochrane risk-of-bias 2.0 tool.
Data Extraction
Overall adherence and adherence to each CONSORT section were calculated. Regression analyses investigated the association between completeness of reporting, ROB, and other characteristics (quartile range, publication modalities, study protocol registration).
Data Synthesis
The mean overall CONSORT adherence across studies was 65%. Studies with high ROB have less adherence than those with low ROB (−5.5%; CI, −10.9 to 0.0). There was a 10.2% (% CI, 6.2-14.3) increase in adherence if the RCT protocol was registered. Studies published in first quartile journals displayed an overall adherence of 11.7% (% CI 17.1-6.4) higher than those published in the fourth quartile.
Conclusions
Reporting completeness is still suboptimal and is associated with ROB, journal impact ranking, and registration of the study protocol. Trial authors should improve adherence to the CONSORT guideline, and journal editors should adopt new strategies to improve the reporting.
Randomized controlled trials (RCTs) are considered the preferable source of evidence to assess the effectiveness of an intervention for a health condition.
Evaluating the validity of a trial requires essential information to be included in a study report so that readers can adequately assess the generalizability of the study and decide on whether it is appropriate to apply the findings to their patient population.
Beyond limiting the applicability of study results, one of the consequences of poor reporting is its possible influence on the risk of bias (ROB) evaluation. Study methods are frequently not described adequately, and results are presented ambiguously, incompletely, or selectively.
The consequence is that the internal validity of a study, in terms of ROB, is hard to judge. Therefore, these studies are difficult to use for decision-making purposes, such as developing clinical guidelines. To overcome these problems, reporting guidelines (RGs) have been developed to support authors in reporting research methods, results, and other sections of a scientific manuscript.
and currently consists of 25 key items and 11 subitems (for a total of 37 items) that guide the reporting of an RCT. Assessing published RCTs for their completeness of reporting (ie, adherence to the CONSORT checklists) is essential for drafting further publication policies and minimizing the risk of publication bias.
Consolidated Standards of Reporting Trials (CONSORT) and the completeness of reporting of randomised controlled trials (RCTs) published in medical journals.
confirmed that most of the authors (∼88%) publishing RCTs in high impact rehabilitation journals did not mention RG use and that 58% of those who declared using a RG did not do it appropriately. This occurred despite many rehabilitation journals endorsing RGs during the submission process.
In fact, in 2014, a total of 28 rehabilitation journals simultaneously published an editorial highlighting the need of using RGs to ensure the quality and completeness of studies in the field. To our knowledge, adherence to the CONSORT checklist and the relationship between the completeness of reporting and ROB of RCTs published in rehabilitation journals have not been systematically evaluated.
The primary objectives of this meta-research study were (1) to evaluate the completeness of reporting of RCTs published in rehabilitation journals through the evaluation of the adherence to the CONSORT checklist and (2) to investigate the relationship between the completeness of reporting and the ROB assessed with the Cochrane risk-of-bias tool 2.0 (RoB 2) tool, which is the most updated Cochrane tool to assess the ROB in RCTs.
The secondary objective was to study the association between the completeness of reporting and the characteristics of studies and journals.
Methods
A cross-sectional analysis was conducted in a random sample of 200 RCTs published between 2011 and 2020 in the 68 journals indexed under the “rehabilitation” category in InCites Journal Citation Report.
This sampling method was chosen to include all journals in the rehabilitation field. The 68 rehabilitation journals indexed in InCites are reported as an annex (supplemental table S1, available online only at http://www.archives-pmr.org/). The project protocol was publicly posted before data extraction on the medRxiv preprint server.
Innocenti T, Giagio S, Salvioli S, et al. Adherence to the CONSORT Statement and risk of bias assessment in randomized controlled trials in rehabilitation journals: a protocol for a meta-research study. medRxiv 2021:2021.01.21.21250256.
RCTs with a parallel-group design published from 2011-2020 as full-text scientific articles in the rehabilitation journals as mentioned earlier were included. The following were excluded: observational or cohort studies, interim analyses, economic analyses of RCTs, RCT protocols, quasi-experimental design, posttrial follow-up studies, subgroup and secondary analyses of previously reported RCTs, RCTs with crossover design, pilot feasibility RCTs, n-of-1 trials, cluster trials, editorials, letters, and news reports.
Study selection process
Journal tags for the journals were identified in MEDLINE, and a detailed search strategy was created to find all RCTs published from 2011-2020 in this database (supplemental table S2, available online only at http://www.archives-pmr.org/). The search was performed only in MEDLINE because all rehabilitation journals are indexed in this database. Titles and abstracts of the selected articles were screened for eligibility in a blinded process by 2 independent reviewers who selected potential articles based on inclusion criteria. A third reviewer resolved any disagreement. After this process, 200 reports (our study sample) were randomly selected by an independent author using a computerized random sequence generator.
) to include an equal number of studies from 2011-2015 (n=25 for each quartile range) and from 2016-2020 (n=25 for each quartile range). The study selection process is summarized in fig 1.
Full texts were stored in EndNote X7.a A data extraction form was used and the following data were extracted: first author, publication year, journal characteristics (open access vs hybrid), country, rehabilitation field, protocol registration (yes/no), completeness of the reporting (see below), and ROB with the Cochrane RoB 2 tool (see below).
Methods for extracting data and assessing the completeness of reporting
The completeness of reporting was calculated as adherence to each item on the CONSORT checklist,
and it was assessed using a 25-item checklist. Preliminarily, interrater agreement was evaluated in a sample of 30 RCTs (15% of the entire sample), which were scored independently by 2 authors with postgraduate training in clinical epidemiology and critical appraisal. This method was chosen to ensure reproducibility and to align the appraisers. However, the CONSORT checklist is not an assessment checklist; therefore, it does not need to be performed independently by 2 authors. We calculated the Cohen kappa for the couple of raters to measure the interrater agreement. The agreement was classified as suggested by McHugh
: values <0 indicate no agreement, 0-0.20 none to slight, 0.21-0.40 fair, 0.41-0.60 moderate, 0.61-0.80 substantial, and 0.81-1 almost perfect agreement. In this sample, the agreement was almost perfect (κ=0.825; P<.001).
Following the explanation and elaboration statements of the CONSORT checklist, each item was marked with “1” if it was well described, with “0” if incomplete or missing, and with “NA” if not applicable. An item was defined as “NA” when its authors did not describe it, but it was clear (by reading the protocol and/or the full text) that such information was justifiably missing (eg, if authors did not perform any additional analyses, such as a priori subgroup analysis because they did not plan them in the protocol, item 12b was not applicable). Total adherence to each item and overall adherence to the CONSORT checklist for each study were calculated (from 0%-100% with 0% representing 0 adherence and 100% full adherence), weighting for the number of applicable items. Given the aim of this study to investigate the completeness of reporting, the authors of the studies included were not contacted for information omitted from articles.
Methods for extracting data and assessing the risk of bias
was used to assess the ROB in the studies included. Because this assessment is required for each outcome measure considered critical or important, the primary outcome of each RCT was considered for evaluation in this study. For each RCT, a ROB level was assigned for each of the 5 domains (bias arising from the randomization process, bias because of deviations from intended interventions, bias because of missing outcome data, bias in the measurement of the outcome, bias in the selection of the reported result), as suggested by the Cochrane Collaboration tool.
Possible ROB level was “high,” “low,” or “some concerns.” An overall ROB score for each study was assigned as suggested by the tool's instructions. The results were graphically summarized through the ROB graph obtained with the ROBVIS Tool.
Two reviewers performed the ROB assessment for each RCT independently; a third reviewer resolved the disagreement when necessary.
Data analysis
The primary analysis addressed (1) the completeness of reporting in each study (see above). Overall adherence to the CONSORT checklist was calculated (in percentage) as the total number of items described and reported out of the total number of applicable items. (2) The adherence to each item and each section of the CONSORT checklist in all studies, calculated as (for each item; in percentage) the number of times that 1 item is described and reported out of the total number of studies in which the item could potentially appear in. (3) The relationship between the completeness of reporting for each study (see above) and studies with low or high ROB. We hypothesized that studies with domains with a low ROB display better reporting compared with those with a high ROB. The relationship was investigated both through descriptive statistics and with multivariable linear regression analysis between the adherence to the CONSORT checklist (both overall adherence and adherence to each checklist section) as dependent variable(s) and the ROB (for each RoB 2 domain and overall ROB) as independent variables.
A multivariable linear regression analysis was performed as a secondary analysis to assess the association between the adherence to the CONSORT checklist (both overall adherence and adherence to each section) as dependent variable(s) and the following characteristics as independent variables: publication year; journal ranking (quartile range: Q1, Q2, Q3, Q4); publication options (Open Access vs Hybrid); and study protocol registration (yes/no).
The assumptions of linearity, homoscedasticity, independence of observations, and normality were checked for each linear regression model.
Results
The journals and studies' characteristics are shown in supplemental table S3 (available online only at http://www.archives-pmr.org/), and their references are listed in supplementary table S4 (available online only at http://www.archives-pmr.org/). For all regression analyses the assumptions were met.
Adherence to the CONSORT checklist
Overall, the mean adherence varied across items (table 1). The most reported item was the background (#2a) in 100% of cases; the least reported one was the description of the trial design (#3a) in 11% of cases. Some items were not applicable in most RCTs (see table 1), and this may have influenced the calculation of adherence of some items (eg, for #14b where adherence was 0% and NA=184). More in general, the highest adherence rates were in the introduction section (99%), and the lowest were in the “other information” section (37%) (fig 2). The mean overall adherence across studies was 65%, with great variability among them. The highest adherence was 97%, and the lowest was 25% (fig 3).
Table 1Mean adherence across each item of the CONSORT checklist in RCTs published in rehabilitation journals (N=200)
CONSORT Items
Studies Where Item Was Not Applicable (n)
Mean Adherence (%) Calculated in Studies Where Item Was Applicable
Title and abstract
1a
Identification as a randomized trial in the title
0
95
1b
Structured summary of trial design, methods, results, and conclusions (for specific guidance see CONSORT for abstracts)
0
99
Introduction
Background and objectives
2a
Scientific background and explanation of rationale
0
100
2b
Specific objectives or hypotheses
0
98
Methods
Trial design
3a
Description of trial design (such as parallel, factorial) including allocation ratio
0
11
3b
Important changes to methods after trial commencement (such as eligibility criteria), with reasons
195
60
Participants
4a
Eligibility criteria for participants
0
97
4b
Settings and locations where the data were collected
0
67
Interventions
5
The interventions for each group with sufficient details to allow replication, including how and when they were actually administered
0
83
Outcomes
6a
Completely defined prespecified primary and secondary outcome measures, including how and when they were assessed
0
66
6b
Any changes to trial outcomes after the trial commenced, with reasons
193
43
Sample size
7a
How sample size was determined
0
56
7b
When applicable, explanation of any interim analyses and stopping guidelines
196
25
Sequence generation
8a
Method used to generate the random allocation sequence
0
57
8b
Type of randomization; details of any restriction (such as blocking and block size)
0
36
Allocation concealment mechanism
9
Mechanism used to implement the random allocation sequence (such as sequentially numbered containers), describing any steps taken to conceal the sequence until interventions were assigned
8
36
Implementation
10
Who generated the random allocation sequence, who enrolled participants, and who assigned participants to interventions
0
25
Blinding
11a
If done, who was blinded after assignment to interventions (eg, participants, care providers, those assessing outcomes) and how
23
66
11b
If relevant, description of the similarity of interventions
142
83
Statistical methods
12a
Statistical methods used to compare groups for primary and secondary outcomes
0
83
12b
Methods for additional analyses, such as subgroup analyses and adjusted analyses
175
44
Results
Participant flow
13a
For each group, the numbers of participants who were randomly assigned, received intended treatment, and were analyzed for the primary outcome
0
82
13b
For each group, losses and exclusions after randomization, together with reasons
24
70
Recruitment
14a
Dates defining the periods of recruitment and follow-up
0
50
14b
Why the trial ended or was stopped
184
0
Baseline data
15
A table showing baseline demographic and clinical characteristics for each group
0
90
Numbers analyzed
16
For each group, number of participants (denominator) included in each analysis and whether the analysis was by original assigned groups
0
62
Outcomes and estimation
17a
For each primary and secondary outcome, results for each group, and the estimated effect size and its precision (such as 95% CI)
0
66
17b
For binary outcomes, presentation of both absolute and relative effect sizes is recommended
188
75
Ancillary analyses
18
Results of any other analyses performed, including subgroup analyses and adjusted analyses, distinguishing prespecified from exploratory
177
57
Harms
19
All important harms or unintended effects in each group (for specific guidance see CONSORT for harms)
57
34
Discussion
Limitations
20
Trial limitations, addressing sources of potential bias, imprecision, and, if relevant, multiplicity of analyses
0
82
Generalizability
21
Generalizability (external validity, applicability) of the trial findings
0
67
Interpretation
22
Interpretation consistent with results, balancing benefits and harms, and considering other relevant evidence
0
93
Other Information
Registration
23
Registration number and name of trial registry
0
38
Protocol
24
Where the full trial protocol can be accessed, if available
20
21
Funding
25
Sources of funding and other support (such as supply of drugs), role of funders
Relationship between completeness of reporting and risk of bias
Overall, studies with low ROB showed better reporting than those with high ROB (table 2 and supplemental table S5, available online only at http://www.archives-pmr.org/). The results of RoB 2 analysis are reported in supplemental table S6 (available online only at http://www.archives-pmr.org/). The descriptive analysis showed that the reporting of the results section presented the largest differences between studies at high and low ROB, especially in domain 1 (ROB arising from the randomization process), domain 2 (ROB because of deviations from the intended interventions), and domain 5 (ROB in the selection of the reported result). Domain 5 showed that the greatest differences between high and low risk studies occurred in terms of the reporting of results, discussion, other information (see table 2). Regression analyses also confirm these trends (table 3, primary analysis). Studies with high ROB in domain 1 have an overall adherence to the CONSORT checklist 6.86% (CI, −12.34 to −1.39) lower than those with low ROB. In domain 2, studies with high ROB have 6.99% (CI, −12.67 to −1.28) lower adherence than those with low ROB. If we look at the overall ROB, studies with high ROB have 5.46% (CI, −10.91 to −0.02) lower adherence than those with low ROB.
Table 2Mean section adherence for each domains of RoB 2 tool and difference between studies at low risk and high risk
ROB
D1
D2
D3
D4
D5
Low
High
Differences
Low
High
Differences
Low
High
Differences
Low
High
Differences
Low
High
Differences
CONSORT
Adherence (%)
Title and abstract
99.0
92.9
−6.1
97.5
92.3
−5.2
96.2
96.9
0.7
96.5
100.0
3.5
97.4
95.0
−2.4
Introduction
99.0
97.6
−1.4
100.0
100.0
0.0
98.6
98.4
−0.2
99.2
100.0
0.8
98.7
100.0
1.3
Methods
65.7
46.5
−19.2
65.2
47.7
−17.5
59.1
51.2
−7.9
58.5
55.2
−3.3
61.6
47.1
−14.5
Results
73.1
48.8
−24.3
76.2
49.2
−27.0
67.9
56.3
−11.6
67.6
55.8
−11.8
74.1
45.5
−28.6
Discussion
84.1
82.5
−1.6
89.3
76.9
−12.4
80.0
82.3
2.3
82.8
70.0
−12.8
82.0
63.3
−18.7
Other information
43.9
26.6
−17.3
43.5
29.5
−14.0
37.6
31.3
−6.3
41.3
27.0
−14.3
49.9
20.0
−29.9
Overall Adherence
71.6
55.3
−16.3
72.6
55.7
−16.9
66.3
60.0
−6.3
66.8
60.2
−6.6
70.4
52.9
−17.5
Abbreviations: D1, ROB arising from the randomization process; D2, ROB because of deviations from the intended interventions; D3, ROB because of missing outcome data; D4, ROB in measurement of the outcome; D5, ROB in selection of the reported result.
Table 3Multivariable linear regression models for primary and secondary study analysis with association between independent variables and overall adherence to the CONSORT checklist
NOTE. Dependent variable: overall adherence to the CONSORT checklist.
Abbreviations: D1, ROB arising from the randomization process; D2, ROB because of deviations from the intended interventions; D3, ROB because of missing outcome data; D4, ROB in measurement of the outcome; D5, ROB in selection of the reported result; high, high risk of bias; low, low risk of bias; Overall ROB, overall risk of bias; Q, quartile.
The descriptive statistics on adherence to each item in a study at low and high risk for each ROB domain are detailed in supplementary table S5. Regression coefficients for adherence to the individual CONSORT sections are reported in supplemental table S7 (available online only at http://www.archives-pmr.org/).
Secondary analysis
There was a significant relationship between adherence to the CONSORT checklist and protocol registration, publication year, and journal quartile (see table 3, secondary analysis). For protocol registration, there was a 10.21% (% CI, 6.21-14.29) increase in overall adherence if the protocol was registered. Overall adherence increased by 1.17% (% CI, 0.50-1.84) for each subsequent year, and studies published in Q1 have an overall adherence of 11.76% (% CI, 17.13-6.43) higher than Q4.
As for the relationship between adherence to each CONSORT section and the same independent variables, there was a 9.76% (% CI, 3.98-15.55) increase in adherence with the methods section if the protocol was published, and studies published in Q1 complied 10.18% (% CI, 17.87-2.48) more than those published in Q4 (detailed regression coefficients for adherence to the CONSORT sections are reported in supplemental table S8, available online only at http://www.archives-pmr.org/). There was also a significant correlation between adherence in the results, discussion, other information sections, protocol registration, journal quartile, and publication year.
Discussion
This meta-research study evaluated the completeness of reporting in a representative sample of RCTs published in rehabilitation journals, and the existing correlations with ROB and other important characteristics were investigated. Overall, the completeness of reporting was suboptimal, with a mean overall adherence to the CONSORT checklist equal to 65%, with a range from 25%-97% (see fig 3). The results of this meta-research study are consistent with others in the more general biomedical field.
Does the medical literature remain inadequately described despite having reporting guidelines for 21 years? - A systematic review of reviews: an update.
concludes that, despite having RGs for 21 years, the reporting of RCTs is still suboptimal.
To our knowledge, this is the first study to compare the completeness of reporting of RCTs to the ROB of these studies in the rehabilitation field. Our results show a relationship between them, although the potential causality in this association remains to be proven. This implies that authors of RCTs should pay more attention when writing their manuscripts because their work will be evaluated based on what they have clearly stated. As Douglas Altman reported, “Readers should not have to infer what was probably done, they should be told explicitly.”
Studies with low ROB showed better overall reporting than those with high ROB, especially for the RoB 2 domains regarding selection methods (domains 1 and 2). We investigated this relationship through a multivariable regression analysis (see table 3) with all ROB domains and overall ROB score in the same model. We adopted this solution to clearly show how reporting can be influenced by both individual domains and the overall score without multicollinearity.
For almost a decade, most international scientific organizations have highlighted the importance of publishing research findings and study protocols
to ensure research reproducibility and integrity and to avoid publication bias. It is well established that the prospective registration of clinical trials and the endorsement of reporting guidelines may reduce research dissemination bias in clinical research.
Our results showed that protocol registration was positively correlated with better reporting, and this may add relevant knowledge to the rehabilitation field because these results emphasize the role of study protocols and the necessity of their registration to improve each aspect of research quality, from the design to the reporting and the ROB.
The completeness of reporting seems to be also correlated with publication year, and this leads to more room for improvement. This trend could be explained by rehabilitation journals' progressive endorsement of RGs, even if the evidence on this topic is inconclusive.
Moreover, the impact factor (IF) of the journals seems to correlate with the completeness of reporting: studies published in journals in the first quartile have an overall adherence of 11.7% higher than those published in the fourth (see table 3). Although IF metrics are not an indication of the quality of the studies published in a journal,
our results suggest that reporting is more complete in journals with high IF.
Journals could take several steps to ensure better adherence to the CONSORT checklist. First, journals requiring RGs could use a standardized process to assess whether an article follows RGs. Unfortunately, recent studies show that most of the strategies adopted by journals to improve CONSORT use (such as requiring authors to follow the RG in the Instructions to Authors) do not appear to be always effective.
This phenomenon probably happens because editors and reviewers do not check the accuracy of the information and because there are no electronic systems that could check the accuracy of reporting.
Additionally, journals may not thoroughly check reviewers' work, and reviewers are not provided with incentives for using (or not using) RGs during their peer-review, although asking authors to provide information about incomplete or missing items from reporting guidelines could improve the manuscript quality.
There is a need for more attention to each step of evidence production in rehabilitation research. Researchers must improve compliance with the reporting standard. They must increase the readability and transparency of their research reports by submitting the CONSORT checklist alongside their manuscripts, as already required by some rehabilitation journals. Given the lack of incentives for peer-reviewers, journal editors should pay more attention to the peer-review process and, if possible, adopt electronic systems (eg, through artificial intelligence) to check the accuracy of reporting in a scientific paper. In the last years, many studies focused on the use of artificial intelligence in the peer-review process, with promising results in reducing the editor's workload and speeding up the review process.
These systems may run checks during the submission process, detecting and prompting for corrections before submission and explicitly requiring that authors follow RGs indications.
Study limitations
To our knowledge, this is the first study that has thoroughly investigated the use of CONSORT in RCTs published in rehabilitation journals. The major strength is the inclusion of all journals indexed under the rehabilitation category in InCites (without IF restrictions). Therefore, we have included a sample that could potentially represent the entire population of rehabilitation journals. In addition, to our knowledge, this is the first study to compare the completeness of reporting with the ROB in the included studies.
However, our study also has some limitations. We used the CONSORT checklist to assess the completeness of reporting, despite it being a guide for writing and not for evaluating quality. We are aware that despite being in line with previous meta-research studies,
To overcome this issue, we have tried to clearly define how we have reached the scores, and we run a preliminary pilot test to increase interassessor agreement, especially on items requiring interpretation (sometimes indicated by phrases like “if relevant” or “if applicable”).
Conclusions
Despite having RGs for more than 20 years, the completeness of reporting in RCTs in rehabilitation research is still suboptimal. RCTs with low ROB seem to be positively correlated with better reporting, and the registration of study protocols and higher journal ranking are also related to complete reporting. Researchers should improve compliance with the reporting standards, and journals editors should implement new strategies in the peer-review process to ensure more attention is devoted to reporting issues.
Consolidated Standards of Reporting Trials (CONSORT) and the completeness of reporting of randomised controlled trials (RCTs) published in medical journals.
Innocenti T, Giagio S, Salvioli S, et al. Adherence to the CONSORT Statement and risk of bias assessment in randomized controlled trials in rehabilitation journals: a protocol for a meta-research study. medRxiv 2021:2021.01.21.21250256.
Does the medical literature remain inadequately described despite having reporting guidelines for 21 years? - A systematic review of reviews: an update.
Statistically significant.
00227-1&id=gr1.jpg){kind=link}
00227-1&id=gr2.jpg){kind=link}
00227-1&id=gr3.jpg){kind=link}