INTRODUCTION

In biomedical research, the randomized controlled trial (RCT) is the scientific gold standard for evaluating effectiveness of healthcare interventions.¹ The precise communication of RCT findings depends on accurate reporting of results.² Although relevant reporting guidelines such as CONSORT (CONsolidated Standards Of Reporting Trials) have been published,^3–5 various problems in published reports still exist.^6–8 One of the problems is that many authors have intentionally or unintentionally misrepresented and misinterpreted their findings, which could “spin” study results and mislead readers.⁸

In the medical literature, “spin” has been used to refer to “specific reporting strategies to distort the presentation and interpretation of results.”⁶ For instance, authors may divert readers’ attention from nonsignificant primary outcomes to significant secondary endpoints. Spin was first systematically surveyed by Boutron and others⁶ in biomedical literature with a representative sample of RCTs indexed in PubMed in December 2006 with nonstatistically significant primary outcomes. In their research, spin strategies were developed for the identification of spin in RCTs with nonsignificant primary outcomes, and spin was highly prevalent in the abstract’s conclusions.⁶ Thereafter, several studies found that spin is common in the biomedical literature, and the frequency of spin in abstracts varies (23% to 85%) by different studies.^7,9–15 In the field of dentistry, several studies have assessed the prevalence and type of spin in RCTs amongst high-impact dental research journals, endodontics, periodontology, and oral implantology and found that nearly 30.7%–85% of abstracts were identified as spin.^7,16,17

Spin in abstracts may be a critical issue, as evidence has shown that abstracts are usually the first and the only part of an article that is accessible to clinicians,¹⁸ and numerous clinicians make medical decisions based solely on the results and conclusions present in abstracts due to time constraints and unavailability of full-text articles.¹⁹ However, to the best of our knowledge, spin has not been assessed in the field of operative dentistry.

Therefore, this study was conducted to (i) assess the existence and characteristics of spin in recently published RCT abstracts in operative dentistry; and (ii) investigate potential factors associated with the presence of spin in abstracts.

METHODS AND MATERIALS

Sample Creation

The PubMed database was searched using a combination of “Dentistry, Operative”, “dental caries”, “dental amalgam”, “composite resins,” and “dental cement” (for full search strategy; see Table 1), to identify RCT abstracts published during 2015–2019 in the field of operative dentistry, which compared two or more groups and have had nonsignificant result for the primary outcome(s).⁶ Predefined inclusion criteria of RCTs were as follows: human participants, interventions associated with health care, experimental studies, presence of a control group, and random assignment of participants to the study or control group.^20,21 As determined a priori, RCTs for which the primary outcome could not be identified were excluded. To identify primary outcomes, the following potential sources were considered in order:¹⁷

1. Explicitly reported primary outcome(s) in the full text

2. The outcome used in sample size calculation

3. Explicitly reported primary outcome(s) in clinical trial registration

4. Outcome(s) reported in the Results section and consistent with the primary/main research objective

Table 1: The Search Strategy Used in This Study (Searched on June 29, 2020)

View Fullscreen

RESULTS

Characteristics of Included Abstracts

Figure 1 demonstrates the literature flow of this study. A total of 77 RCT abstracts with statistically nonsignificant results for the primary outcomes were included in this study (Table 2). Table 3 describes the characteristics of included abstracts. Amongst the 77 RCTs, 33 were from South America, followed by Europe (n=18), Asia (n=17), and others (n=9). Most RCTs were indexed in SCIE (79.2%), conducted in a single center (89.6%), focused on dental materials (adhesives, composite resin, amalgam, and others) (63.6%) and without international collaboration (62.3%). Approximately, a third of the RCTs focused on procedures, such as placement techniques, adhesive application modes, different subgingival restorative margin designs, and others. More than half of the RCTs were two-armed trials (50.6%), with the number of authors being four to seven (72.7%). The source of funding was industry in 14 (18.2%) RCTs, other sources in 38 (49.4%), and unreported/unfunded in 25 (32.5%). Only 18 studies reported the involvement of statisticians. The number of RCTs were mainly published in Journal of Dentistry (23.4%), Clinical Oral Investigations (22.1%), and Operative Dentistry (13.0%). More than half of the trials (68.8%) were at 1–3 year length of follow-up, 16 were at less than 1-year follow-up, and only 8 were followed for more than 3 years.

View Figure

• Download as Powerpoint
• Download Figure

Table 2: A List of Included RCTs in the Study

View Fullscreen

Table 2: A List of Included RCTs in the Study

View Fullscreen

Table 2: A List of Included RCTs in the Study

View Fullscreen

Table 2: A List of Included RCTs in the Study

View Fullscreen

Table 2: A List of Included RCTs in the Study

View Fullscreen

Table 2: A List of Included RCTs in the Study

View Fullscreen

Table 3: Characteristics of Included RCT Abstracts (N=77)

View Fullscreen

Spin Assessment

Spin was identified in 58 (75.3%) of the 77 included abstracts. Nineteen RCTs (24.7%) presented spin in both the Results and Conclusions section of abstracts, and more than one type of spin strategy was found in 25 (32.5%) abstracts. Frequencies of each spin strategy are presented in Table 4.

Table 4: Frequencies of Spin Strategies in the Results and Conclusions Section (N=77)

View Fullscreen

Spin was observed in the Results section of 32 (41.6%) abstracts. The most frequent spin strategies in the Results section were focusing on significant within-group comparisons for primary outcomes (17, 22.1%) and focusing on a statistically significant primary outcome when there are several co-primary outcomes (10, 13.0%).

The Conclusions section of 45 (58.4%) abstracts presented with spin. Claiming equivalence/noninferior/comparable/similar for statistically nonsignificant results was the most common spin strategy in the Conclusions section (16, 20.8%), followed by focusing on a statistically significant primary outcome when there are several co-primary outcomes (13, 16.9%) and focusing on only statistically significant results (ie, secondary endpoints, subgroup analysis, and within-group analysis) (11, 14.3%). Other spin strategies identified included claiming benefit with no recognition of the statistically nonsignificant results for the primary outcome (5, 6.5%), acknowledge statistically nonsignificant results for the primary outcome but emphasize the beneficial effect of treatment (4, 5.2%), and recommendation to use the treatment (4, 5.2%).

Predictors of Spin

In the multivariable logistic analyses, RCTs with unreported/unfunded source of funding were significantly more likely to present spin in the Results section of abstracts (OR=8.10, 95% CI: 1.31–50.16; p=0.025), and RCTs with more than two treatment arms were significantly more likely to present spin in the Conclusions section of abstracts (OR=5.66, 95% CI: 1.71–18.80; p=0.005). Other factors, namely SCIE indexing, international collaboration, number of centers, the topic of each study (materials and procedures), trial registration, number of authors, statistician involvement, continent of origin, and length of follow-up were not significantly associated with the presence of spin (p>0.05) (Table 5).

Table 5: Binary Logistic Regression-derived OR and 95% CI, with Presence of Spin in the Results and the Conclusions Sections as the Dependent Variables for the Included 77 Abstracts

View Fullscreen

Table 5: Binary Logistic Regression-derived OR and 95% CI, with Presence of Spin in the Results and the Conclusions Sections as the Dependent Variables for the Included 77 Abstracts

View Fullscreen

DISCUSSION

This is the first study to evaluate spin and analyze factors associated with the presence of spin in the field of operative dentistry. We concentrated our analysis on the abstracts of RCTs (with statistically nonsignificant primary outcomes) published from 2015 to 2019. Spin was identified in 75.3% abstracts in one form or another, which exceeded the occurrence of spin in most of the studies (30.7%–70%)^11,12,16 but was less than that identified in endodontic RCTs (85.0%).⁷ Some explanations for the difference among studies may be RCTs from different specialties, diverse instruments for evaluation of spin (the Boutron instrument or its variations), and some bias in assessment among investigators. The spin strategies manifest in various ways. The most common strategy in the Results section was focusing on significant within-group comparisons for primary outcomes rather than between-group comparisons. While claiming equivalence for statistically nonsignificant results was the most frequent spin strategy in the Conclusions section. Spin was more prevalent in the Conclusions section (58.4%) than the Results section (41.6%). Previous studies also reported that the Conclusion sections of abstracts were more susceptible to spin than the other sections.^7,10

Previous studies have assessed whether particular factors were associated with the presence of spin, including financial support, journal impact factor, intervention type, trial phase, trial type, statistician involvement, number of authors, sample size, international collaboration, number of centers, the number of treatment arms, reporting of trial registration, article citations, and the conflict-of-interest disclosures.^{10,12,14–17} Amongst these, most of the studies consistently found no factors to be significantly associated with spin,^{10,12,14–16} but only one study¹⁷ concluded that multicenter RCTs were less likely to present spin in abstracts. In this study, the presence of spin in the Results section of abstracts was significantly associated with source of funding, which was inconsistent with the previous studies.^10,16 Nonindustry-sponsored RCTs may lack specialized teams with the involvement of methodological experts and statistician, or they may be exploratory research with small sample sizes due to limited financial support. Therefore, more statistically nonsignificant results may be found, and spin was more likely to occur. It should also be noted that the difference among studies may be explained by different subjects or sampling, and further study should be conducted to draw conclusions. Furthermore, number of treatment arms was associated with the presence of spin in the Conclusions section of abstracts. RCTs with more than two arms were more likely to focus only on significant between-group results, which was consistent with the findings that focusing on a statistically significant primary outcome when there are several co-primary outcomes (13, 16.9%) was a common spin strategy in the Conclusions section.

Spin in abstracts is particularly crucial, because readers often focus on this concise summary to determine whether the literature is worthy of full-text review. Distorted results can affect researchers’/clinicians’ interpretation of the experimental treatment. Boutron and others²² conducted an RCT to evaluate the impact of spin in abstracts of cancer RCTs on clinicians’ interpretation of treatment benefit. When abstracts present with spin, readers considered the treatment as more beneficial to patients (mean difference, 0.71; 95% CI, 0.07–1.35; p=0.030), and clinicians were more interested in reading the full text (mean difference, 0.77; 95% CI, 0.08–1.47; p=0.029). Furthermore, RCTs with spin in abstracts were more likely to be cited compared with those without spin.¹⁷ Spin may not only mislead readers by distorting results but also exert adverse influence on further research.²³ This study might raise awareness among the readers of operative dentistry about spin in published reports.

Manuscripts with statistically significant results are more likely to be published.²⁴ This common phenomenon may prompt some researchers intentionally or subconsciously to spin results and conclusions in order to attract peer reviewer attention.⁸ Reporting guidelines like CONSORT 2010 were developed to help authors improve the reporting quality of manuscripts.^4,20,21 However, guidelines on avoiding spin are not available. Thus, present guidelines need to be expanded to minimize the occurrence of spin. Peer reviewers and editors should be aware of spin in abstract reporting and provided with specific instruments to help identify it in manuscripts. Manuscript authors are supposed to report and interpret results objectively, and so guidelines for authors should also reflect this.

This study has several strengths and limitations. First, 5 years of published RCTs in the field of operative dentistry were evaluated to provide a comprehensive view of the issues related to the spin in this field. Second, as far as we know, this is the first study to investigate potential factors associated with presence of spin in the Results and Conclusions sections, respectively. One limitation of this study is that spin was not evaluated in the main text of included RCTs. A recent study assessed spin in the abstract and the full text of dental RCTs, and found 78.7% of the included RCTs presented some type of spin in the main text.¹⁴ However, the small sample size may not be able to provide highly accurate results. Further study is needed to identify spin in the full texts of dental literature.

CONCLUSIONS

The occurrence rate of spin (75.3%) in the sample of operative dentistry RCTs abstracts is high. Source of funding (OR=8.10; p=0.025) and number of treatment arms (OR=5.66; p=0.005) were associated with the presence of spin in the Results and the Conclusions sections of abstracts, respectively. Readers and other stakeholders need to be aware of the existence of spin in RCT abstracts, and appraise the results and conclusions of RCT abstracts critically.