2.1

Studies’ selection and type of studies

Two journals that have ( Journal of Dentistry IF: 3.109 and Clinical Oral Investigations IF: 2.207) and two that have not ( Operative Dentistry IF: 2.819 and Journal of Prosthetic Dentistry IF: 1.515) endorsed the CONSORT were selected.

We contacted by e-mail the editorial team of the journals that have endorsed the CONSORT to ascertain when the CONSORT Statement was endorsed by these journals. We determined endorsement status by cross-checking with the CONSORT group’s endorser database.

A search was performed in MEDLINE via PubMed to identify the RCTs in restorative dentistry published by these two journals two years before the CONSORT Statement was endorsed and two years after (i.e. year of endorsement: 2007; two years before: 2005–2006; and two after: 2009–2010). The RCTs published in the other two journals were selected considering the same date range. We considered as RCTs in restorative dentistry all randomized controlled trials that had at least one restorative intervention irrespective of the design (i.e. parallel, cross-over). We excluded studies that had the primary outcome related to dental implants (i.e. survival of implants). The search strategy used was based on the abbreviation of journal’s name (J Dent, Clin Oral Investig, Oper Dent, J Prosthet Dent) and the selection of years of interest.

Based on the reporting of the editorial team, Journal of Dentistry endorsed the CONSORT Statement during 2003–2004 and Clinical Oral Investigations in 2009. The two journals differ in the endorsement strategy. While Journal of Dentistry recommends the use of CONSORT and submission of the checklist and flow diagram, Clinical Oral Investigations does not recommend the use of reporting guidelines in the instructions to authors. Thus, for Journal of Dentistry published studies were selected from the years of 2001 to 2002 and 2005–2006, for Clinical Oral Investigations from the years of 2007–2008 and 2010–2011 and from Operative Dentistry and Journal of Prosthetic Dentistry from 2001 to 2011.

2.2

Data collection and analysis

2.2.1

Selection of evaluations

All papers published in the four journals in the pre-determined period were uploaded to Distiller Systematic Review Software ^® (Distiller SR), an online program that facilitates screening and data extraction. A team of researchers (RSO and VPN) with content expertise identified articles by first analysing titles and abstracts for relevance and presence of the selection criteria listed above. Retrieved records were classified as i) include, ii) exclude or iii) uncertain. The full text articles of included and uncertain records were obtained for further eligibility screening by the same two reviewers. Discrepancies in eligibility were resolved through discussion between them. In the event of an unsettled disagreement, the opinion of others authors was obtained.

2.2.2

Data extraction and management

First, the data extractors (RSO and VPN) piloted a 10% random sample of included studies using a standardized data extraction form including general characteristics, data describing completeness of reporting and risk of bias. The results of this first step were discussed. Subsequently, data from each remaining study were extracted by one of the researchers. Any doubt was discussed by consensus between the same researchers.

We extracted the general characteristics of the included studies: author, year, and country of publication (classified by continent), interventions tested and design of the study (parallel, cross-over, split-mouth or unclear). We collected the completeness of reporting of RCTs across the 25 items of the CONSORT 2010 checklist. For instance, when evaluating the reporting of ‘title and abstract’, we were evaluating whether reports of RCTs contained the word ‘randomized’ in the title or abstract. The judgment as adequate or inadequate reporting of any of the 25 items was based on the explanation of CONSORT 2010 . If an item was judged as ‘partially’ reported, it was considered ‘inadequate’. We used the CONSORT 2010 considering that this is the last and most complete version.

2.2.3

Validity assessment in included evaluations

The validity of included studies was assessed by one of the data extractors using the Cochrane risk of bias tool considering the judgment of the random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data and selective reporting as ‘Low risk’ of bias, ‘High risk’ of bias or ‘Unclear risk’ of bias.

2.2.4

Data synthesis

For the primary outcome, we collected as dichotomized (‘adequate’ and ‘inadequate’) adherence to each of the 25 items of the CONSORT 2010. We collected the information of the reporting considering if the design of study permit the application of item. For instance, if it is impossible for any reason to apply the blinding (item 11) we considerer that item as ‘adequate’. In situations, where an item was judged as ‘partially’ reported, it was considered ‘inadequate’. For each dichotomous outcome, the proportion of RCTs with adequate reporting of checklist items in each comparison group was calculated. Using these proportions we compared the completeness of reporting between comparison groups (endorsers versus non-endorsers, before versus after endorsement) using a risk ratio (RR) with a 99% confidence interval for each outcome. A RR greater than 1 indicated relatively increased reporting of any CONSORT item. We used a pooled RR with 99% confidence interval to estimate the overall difference between groups within each comparison item following CONSORT endorsement. We did not include in the analyses items with 100% of adequate reporting in both groups (RR = 1) or with 1 adequate item in one group and 0 in another i.e. resulting in a wide confidence interval. Also, there was not situation that the endorsing journal reported 100% of some item and the non-endorsing journals not.

The analyses were performed in Review Manager Software (RevMan Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2014) using a random-effects model for all analyses. The risk of bias was summarized descriptively using the Microsoft Excel Program (Microsoft Corporation).

2.1

Studies’ selection and type of studies

Two journals that have ( Journal of Dentistry IF: 3.109 and Clinical Oral Investigations IF: 2.207) and two that have not ( Operative Dentistry IF: 2.819 and Journal of Prosthetic Dentistry IF: 1.515) endorsed the CONSORT were selected.

We contacted by e-mail the editorial team of the journals that have endorsed the CONSORT to ascertain when the CONSORT Statement was endorsed by these journals. We determined endorsement status by cross-checking with the CONSORT group’s endorser database.

A search was performed in MEDLINE via PubMed to identify the RCTs in restorative dentistry published by these two journals two years before the CONSORT Statement was endorsed and two years after (i.e. year of endorsement: 2007; two years before: 2005–2006; and two after: 2009–2010). The RCTs published in the other two journals were selected considering the same date range. We considered as RCTs in restorative dentistry all randomized controlled trials that had at least one restorative intervention irrespective of the design (i.e. parallel, cross-over). We excluded studies that had the primary outcome related to dental implants (i.e. survival of implants). The search strategy used was based on the abbreviation of journal’s name (J Dent, Clin Oral Investig, Oper Dent, J Prosthet Dent) and the selection of years of interest.

Based on the reporting of the editorial team, Journal of Dentistry endorsed the CONSORT Statement during 2003–2004 and Clinical Oral Investigations in 2009. The two journals differ in the endorsement strategy. While Journal of Dentistry recommends the use of CONSORT and submission of the checklist and flow diagram, Clinical Oral Investigations does not recommend the use of reporting guidelines in the instructions to authors. Thus, for Journal of Dentistry published studies were selected from the years of 2001 to 2002 and 2005–2006, for Clinical Oral Investigations from the years of 2007–2008 and 2010–2011 and from Operative Dentistry and Journal of Prosthetic Dentistry from 2001 to 2011.

2.2

Data collection and analysis

2.2.1

Selection of evaluations

All papers published in the four journals in the pre-determined period were uploaded to Distiller Systematic Review Software ^® (Distiller SR), an online program that facilitates screening and data extraction. A team of researchers (RSO and VPN) with content expertise identified articles by first analysing titles and abstracts for relevance and presence of the selection criteria listed above. Retrieved records were classified as i) include, ii) exclude or iii) uncertain. The full text articles of included and uncertain records were obtained for further eligibility screening by the same two reviewers. Discrepancies in eligibility were resolved through discussion between them. In the event of an unsettled disagreement, the opinion of others authors was obtained.

2.2.2

Data extraction and management

First, the data extractors (RSO and VPN) piloted a 10% random sample of included studies using a standardized data extraction form including general characteristics, data describing completeness of reporting and risk of bias. The results of this first step were discussed. Subsequently, data from each remaining study were extracted by one of the researchers. Any doubt was discussed by consensus between the same researchers.

We extracted the general characteristics of the included studies: author, year, and country of publication (classified by continent), interventions tested and design of the study (parallel, cross-over, split-mouth or unclear). We collected the completeness of reporting of RCTs across the 25 items of the CONSORT 2010 checklist. For instance, when evaluating the reporting of ‘title and abstract’, we were evaluating whether reports of RCTs contained the word ‘randomized’ in the title or abstract. The judgment as adequate or inadequate reporting of any of the 25 items was based on the explanation of CONSORT 2010 . If an item was judged as ‘partially’ reported, it was considered ‘inadequate’. We used the CONSORT 2010 considering that this is the last and most complete version.

2.2.3

Validity assessment in included evaluations

The validity of included studies was assessed by one of the data extractors using the Cochrane risk of bias tool considering the judgment of the random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data and selective reporting as ‘Low risk’ of bias, ‘High risk’ of bias or ‘Unclear risk’ of bias.

2.2.4

Data synthesis

For the primary outcome, we collected as dichotomized (‘adequate’ and ‘inadequate’) adherence to each of the 25 items of the CONSORT 2010. We collected the information of the reporting considering if the design of study permit the application of item. For instance, if it is impossible for any reason to apply the blinding (item 11) we considerer that item as ‘adequate’. In situations, where an item was judged as ‘partially’ reported, it was considered ‘inadequate’. For each dichotomous outcome, the proportion of RCTs with adequate reporting of checklist items in each comparison group was calculated. Using these proportions we compared the completeness of reporting between comparison groups (endorsers versus non-endorsers, before versus after endorsement) using a risk ratio (RR) with a 99% confidence interval for each outcome. A RR greater than 1 indicated relatively increased reporting of any CONSORT item. We used a pooled RR with 99% confidence interval to estimate the overall difference between groups within each comparison item following CONSORT endorsement. We did not include in the analyses items with 100% of adequate reporting in both groups (RR = 1) or with 1 adequate item in one group and 0 in another i.e. resulting in a wide confidence interval. Also, there was not situation that the endorsing journal reported 100% of some item and the non-endorsing journals not.

The analyses were performed in Review Manager Software (RevMan Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2014) using a random-effects model for all analyses. The risk of bias was summarized descriptively using the Microsoft Excel Program (Microsoft Corporation).