Some superiority trials with nonsignificant results published in high impact factor journals correspond to noninferiority situations: a research-on-research study.

OBJECTIVES

Many negative randomized controlled trials (RCTs) report spin in their conclusions to highlight the benefits of the experimental arm, which could correspond to a noninferiority (NI) objective. We aimed to evaluate whether some negative superiority RCTs comparing 2 active interventions could correspond to an NI situation and to explore associated trial characteristics.

STUDY DESIGN AND SETTING

We searched PubMed for superiority RCTs comparing 2 active interventions with non-statistically significant results for the primary outcome that were published in 2021 in the 5 journals with the highest impact factor in each medical specialty. Three reviewers independently evaluated whether trials could correspond to an NI situation (ie, an evaluation of efficacy as the primary outcome, with the experimental intervention presenting advantages including better safety profile, ease of administration, or decreased cost as compared with the control intervention).

RESULTS

Of the 147 trials included, 19 (12.9%, 95% CI [7.9%, 19.4%]) corresponded to a potential NI situation. As compared with trials not in a potential NI situation, they were published in a journal with a lower impact factor (median impact factor 8.7 vs 15.6), were more frequently rated at high or some concerns regarding risk of bias (n = 14, 73.7% vs n = 69, 53.9%) and reported spin in the article conclusions (n = 11, 57.9% vs n = 24, 18.8%).

CONCLUSION

A non-negligible proportion of superiority negative trials comparing 2 active interventions could correspond to an NI situation. These trials seemed at increased risk of bias and frequently reported spin in the conclusions, which may distort the interpretation of results.

PLAIN LANGUAGE SUMMARY

Noninferiority trials are designed to show that a new intervention is not worse in terms of efficacy than the reference intervention. It is adapted when the new intervention has an advantage in terms of safety, ease of use or cost over the reference one. However, the literature displayed some superiority negative trials comparing 2 active interventions that could correspond to a potential noninferiority situation. Our study aimed to assess whether some superiority trials with nonsignificant results for the primary outcome could correspond to an NI situation and to explore associated trial characteristics. Our findings indicate that a non-negligible proportion of superiority negative trials could correspond to a noninferiority situation. Moreover, those trials seemed at increased risk of bias and frequently reported spin in the conclusions.