This manuscript reports the methods and results of part 2 of a cross-sectional study using the same 98 eligible reviews as in part 1 . To avoid redundancy some sections of this manuscript have been abridged, but all additional information on the research methods and the characteristics of the included reviews can be found in part 1  and in the published protocols of parts 1 and 2 [18,24]. The protocol for this second cross-sectional study  can be consulted via the following link: https://researchintegrityjournal.biomedcentral.com/articles/10.1186/s41073-019-0084-4
The checklist of the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement for cross-sectional studies  was included as additional file 1. The differences between the methods planned in our protocol and those implemented in the final study were reported in Additional file 2. The rationales for these differences were also given. All raw data were reported in the Open Science framework Open Science Framework (https://osf.io/ka7mp/). All eligible studies for this cross-sectional study are reported in Additional file 3. There was no patient or public involvement during the development of the protocol or in the conduct of this study.
The eligibility criteria were published previously in our protocol  and in part 1 of this study  and are given in additional file 2 .
Information sources and search strategy
The Cochrane Database of Systematic reviews  and the websites of 5 leading orthodontic journals were the information sources of this study. These 5 orthodontic journals are: European Journal of Orthodontics [EJO], American Journal of Orthodontics and Dentofacial Orthopedics [AJODO], Angle Orthodontist (AO), The Korean Journal of Orthodontics (KJO), and Orthodontics and Craniofacial Research (OCR). These journals were manually searched from August 1 2009 until July 31 2021 for systematic reviews that fulfilled the eligibility criteria.
Data collection procedures
Two calibrated authors (PS and RMR) independently collected data from the 98 eligible reviews to address the research questions. In the case of disagreement on these procedures the following strategies were implemented and in this sequence: (1) discussions between these operators (2) contacting of authors by email to clarify issues regarding a specific manuscript (3) consultation with a methodologist (NDG) to arbitrate. Data extraction was conducted according to our published protocol  and all data collection forms were reported in (Additional file 2). Data items were collected from the entire eligible review, i.e., the entire manuscript including the abstract, tables, figures, and additional files. We implemented this procedure for all eligible reviews, but did not extract data from the plain language summary of eligible Cochrane reviews.
Assessing adverse effects of orthodontic interventions
Pain and the various categories of adverse effects hypothetically linked to orthodontic interventions as defined by Preoteasa et al.  and modified by Steegmans et al.  were reported in a table in Additional file 2. This table was used as our reference to assess the reporting on adverse effects in the abstract. When additional adverse effects were identified that were not given in this table we included them with rationale. Effects that could be labelled either as ‘beneficial’ or ‘adverse’ were not included unless the review authors labelled these ambiguous effects as ‘adverse’. Explanations for such decisions were given.
Assigning spin of adverse effects of orthodontic interventions in abstracts of systematic reviews
Spin was assigned by comparing whether what was reported in the abstract on adverse effects of orthodontic interventions was congruent with the findings on these effects in the review. Three types of spin were assigned i.e., misleading reporting, misleading interpretation, and misleading (inappropriate) extrapolation on adverse effects of orthodontic interventions in abstracts of systematic reviews . To facilitate this assignment each type of spin was subdivided in categories and spin was assigned when it was identified in one or more of these categories. Spin was assessed in all eligible reviews irrespective of whether these reviews sought adverse effects of interventions or not. Because the pilot tests for our protocol identified only 2 reviews with spin  and because assessing spin is not easy , we conducted additional pilot tests on 10 RCTs to further calibrate the operators (PS and RMR) that assigned spin and to fine-tune the descriptions of spin and the checklists for assigning spin. These fine-tuned descriptions of spin and the pertinent data collection forms to identify spin are reported respectively in Table 2 and Additional file 2.
Types of spin in reviews that did or did not seek adverse effects of interventions
Definitions of the 3 types of spin
Reviews that sought adverse effects of interventions
Reviews that did not seek adverse effects of interventions
Misleading reporting (in the abstract) on adverse effects of interventions:
‘Incomplete or inadequate reporting in the abstract on the results of adverse effects that were not supported by the findings of the review’.
1) Not reporting in the abstract on the results of adverse effects found in the review.
2) Selective reporting in the abstract on the results of adverse effects found in the review.
1) Reporting on results of adverse effects in the abstract when adverse effects were not sought.
2) Reporting in the abstract that adverse effects were sought when they were not sought.
Misleading interpretation (in the abstract) on adverse effects of interventions:
‘Interpretation in the abstract on the results of adverse effects that was not supported by the findings of the review’. This could for example underestimated the adverse effects of the intervention’.
1) Claiming in the abstract that the intervention is safe (has no or minimal adverse effects), despite concerning results on adverse effects found in the review e.g., based on non-statistically significant results on adverse effects with wide confidence intervals .
2) Downgrading in the abstract the importance of the adverse effects, despite concerning results on adverse effects found in the review.
3) Recommendations are made in the abstract for clinical practice that are not supported by the findings in the review on adverse effects’.
1) Claiming in the abstract that the intervention is safe (has no or minimal adverse effects) despite not having sought adverse effects.
2) Downgrading in the abstract the importance of the adverse effects, despite not having sought adverse effects.
3) Recommendations are made in the abstract for clinical practice despite not having sought adverse effects.
Misleading (inappropriate) extrapolation (in the abstract) on adverse effects of interventions:
‘Overgeneralisation in the abstract of the study results to different populations, interventions, outcomes or settings than were assessed in the review despite evidence on adverse effects on a different population, intervention, outcome or setting’.
1) Results are extrapolated in the abstract to another population, intervention, outcome or setting than were assessed in the review despite evidence on adverse effects on a different population, intervention, outcome or setting.
1) Results are extrapolated in the abstract to another population, intervention, outcome or setting than were assessed in the review despite not having sought adverse effects.
In our pilot sample we identified an overall proportion of 14.3% (2/14) of spin of the adverse effects in the abstracts of 14 systematic reviews of orthodontic interventions. For this proportion Epitools software  calculated a required sample size of 48 studies (precision 0.1 and confidence level 0.95), which was fulfilled by our 98 eligible reviews.
Outcomes and statistical analyses
Prevalence proportions were calculated to quantify the answers to our 3 research questions in the 98 selected reviews. According to our published protocol these proportions were also calculated separately for reviews that either did (n = 84) or did not (n = 14) seek any findings related to adverse effects of interventions in the included studies . These statistics were calculated for (1) all journals as a one group (2) the five leading orthodontic journals together and the Cochrane Database of Systematic Reviews separately and (3) each eligible journal separately.
Univariable logistic regression models were built to determine the association between the presence of spin in the abstract and characteristics of the systematic review i.e., journal, year of publication, number of authors, conflict of interest reported, conflict of interest present, funding reported, and type of orthodontic intervention. These analyses were not registered in the protocol and should therefore be interpreted as exploratory. The strengths of associations were quantified using odds ratios (OR), and 95% confidence intervals (95% CI). Multivariable models were built if multiple significant predictors were found in the univariable analysis. Analyses were performed with the use of commercial software (IBM SPSS 22.0, SPSS Inc, Chicago, IL). A two-sided P value of 0.05 was considered to be statistically significant.