Non-Inferiority Trials in Medicine: Practice Changing or a Self-Fulfilling Prophecy?

Non-Inferiority Trials: A Systematic Review on Methodological Quality and Reporting Standards

BACKGROUND

Non-inferiority (NI) trials require unique trial design and methods, which pose challenges in their interpretation and applicability, risking introduction of inferior therapies in clinical practice. With the abundance of novel therapies, NI trials are increasing in publication. Prior studies found inadequate quality of reporting of NI studies, but were limited to certain specialties/journals, lacked NI margin evaluation, and did not examine temporal changes in quality. We conducted a systematic review without restriction to journal type, journal impact factor, disease state or intervention to evaluate the quality of NI trials, including a comprehensive risk of bias assessment and comparison of quality over time.

METHODOLOGY

We searched PubMed and Cochrane Library databases for NI trials published in English in 2014 and 2019. They were assessed for: study design and NI margin characteristics, primary results, and risk of bias for blinding, concealment, analysis method and missing outcome data.

RESULTS

We included 823 studies. Between 2014 and 2019, a shift from publication in specialty to general journals (15% vs 28%, p < 0.001) and from pharmacological to non-pharmacological interventions (25% vs 38%, p = 0.025) was observed. The NI margin was specified in most trials for both years (94% vs 95%). Rationale for the NI margin increased (36% vs 57%, p < 0.001), but remained low, with clinical judgement the most common rationale (30% vs 23%), but more 2019 articles incorporating patient values (0.3% vs 21%, p < 0.001). Over 50% of studies were open-label for both years. Gold standard method of analyses (both per protocol + (modified) intention to treat) declined over time (43% vs 36%, p < 0.001).

DISCUSSION

The methodological quality and reporting of NI trials remains inadequate although improving in some areas. Improved methods for NI margin justification, blinding, and analysis method are warranted to facilitate clinical decision-making.

[Low-dose irradiation of non-malignant diseases: Did we throw the baby out with the bathwater?]

The irradiation of non-malignant diseases, essentially for anti-inflammatory purpose, have been largely proposed and performed worldwide until the 1970-80s. At that time, the better assessment of the radio-induced malignancies, essentially in children and young patients, as well as the efficacy of the new anti-inflammatory drugs (steroids and non-steroids), led to the almost disappearance of those techniques, at least in France. In contrast, our German colleagues are still going on treating about 50,000 patients per year for non-malignant (more or less severe) diseases. After a short historical overview, the present article suggests that we were possibly going too far in the rejection of those low-dose irradiations for benign lesions. The recent emergence of new preclinical data, the better understanding of the risk of radio-induced secondary tumours (almost nil in the elderly), and the severity of some situations, such as the cytokine storm of the COVID-19, should probably lead us to reconsider those low - and sometimes very low (less than 1Gy) - irradiations for well-selected indications in the elderly.

The Overton Window and a Less Dogmatic Approach to Antibiotics

This analysis summarizes a set of recent, innovative trials in infectious diseases that redefine previous, non-evidence based "rules" for antibiotics. Recent trials in infectious diseases have led to reconsidering the traditional treatment of infectious diseases by changing the duration or type of traditional antibiotics or evaluating new antibiotics for approval. These trials have used the noninferiority trial approach. The noninferiority trial design and recent infectious disease trials of relevance are discussed in this viewpoint. My objective was to analyze recent trials in infectious disease and consider needs for future trials.

Value Assessment and Quantitative Benefit-Risk Modelling of Biosimilar Infliximab for Crohn's Disease

BACKGROUND AND OBJECTIVE

Regulatory approval of biosimilars often depends on extrapolating evidence from one clinical indication to all of those of the originator biologic. We aimed to develop a quantitative benefit-risk analysis to assess whether the resulting increase in the uncertainty in the clinical performance of biosimilars (i.e. risk) may be countered by their lower pricing (benefit).

METHODS

A 1-year decision-analytic model was developed for the biosimilar infliximab (Inflectra^®) for Crohn's disease. The perspective was that of the National Health Service in the UK and costs were valued to 2015/16. A hypothetical cohort of biologic-naïve patients with moderate-to-severe Crohn's disease was simulated through the model. Immunogenicity to infliximab was a key modifier, influencing rates of non-response and infusion reactions. Net health benefit was estimated based on quality-adjusted life-years. A range of sensitivity analyses tested the robustness of the results and explored how the biosimilar price must respond to varying immunogenicity to remain the preferred option.

RESULTS

The base-case analysis predicted a positive incremental net health benefit of 0.04 (95% central range 0.00-0.09) favouring the biosimilar, based on 0.803 quality-adjusted life-years, and costs of £18,087 and £19,176 for the biosimilar and originator, respectively. Two-way sensitivity analyses suggested that if 50% of patients developed antibodies, the value-based price of £410 per vial must be lower than that of the originator (£420), but remain higher than the actual market price (£378).

CONCLUSIONS

The model supports the use of Inflecta^® for Crohn's disease in the UK, and provides a framework for the quantitative evaluation of biosimilars in the context of a health technology assessment. Value-based pricing using this methodology could protect health systems from the potential risks of biosimilars where they are untested in the approved populations.