By Michael Keane.
A recent, much publicized, randomized controlled trial (RCT) demonstrated that the corticosteroid, dexamethasone, reduced mortality from CoViD-19. This was hailed as a practice-changing breakthrough.
The authors announced that “RECOVERY is a large, pragmatic, randomized, controlled adaptive platform trial designed to provide rapid and robust assessment of the impact of readily available potential treatments for COVID-19” for which, hitherto, “no therapeutic agents have been shown to reduce mortality.”
However, this study highlights another phenomenon: during the first 6 months of this pandemic the paradigm of discovering and implementing treatments based on the RCT has lagged significantly behind biologically-rational clinical innovation; where such “unproven” treatments (refined by continuous adaption based on world-wide feedback) have demonstrated competitive survival rates.
In this context we need to reconsider the primacy of the RCT during a fast-paced pandemic and consider more efficient and ethical approaches to clinical knowledge acquisition.
The co-ordination involved with RECOVERY was heroic. But, is it biologically optimal to treat patients with a non-titrated (without regards to, or adjustment in response to, the inflammatory markers of the individual patient), one-size-fits-all (do 50kg and 130kg patients get the same dose?), curiously low dose of a steroid? Furthermore, in this open-label study, the mortality in the control group ranged from 13% in even those patients who initially needed no supplemental oxygen, up to 41% in those who were mechanically ventilated. Similarly, Remdesivir wasn’t available until towards the end of the study.
Consider the reproducibility crisis in even benchtop biology, where the slightest change in the experimental milieu can completely change the outcome of the experiment. Now consider what happens if we drop 6mg of dexamethasone in different systems where the mortality is substantially less, Remdesivir is available, CoViD-specific anti-coagulation is refined, countless other modalities and improvements in care may be present and the individual patient is considered regarding dose, clinical course and markers of effect? And the effect of dexamethasone in RECOVERY was an average effect across age ranges. How reproducible is the effect for my given patient’s age, in this highly age-dependent disease?
In this situation, we are completely back to conjecture; we have no certainty of reproducibility of “the evidence” in this new clinical “milieu”. Therefore, using dexamethasone would be akin to trying something reasonable in the mix and seeing what happens; exactly what clinical innovators had already been doing with steroids. Measurement, adaption and feedback compared to differing regimens would then determine if an all-comers 6mg dose of dexamethasone would survive in the clinical mix.
CoViD-19 is a disease of different stages. Based on an increasing understanding of the biology, cytokine biochemistry, resulting pathology and survival advantage of treatment, it has been proposed that CoViD-19 is a “steroid responsive disease: however, timing is critical.” Thus, many clinical units were already using steroids as part of their treatment protocols and were achieving highly competitive survival rates; saving lives through rational innovation while continuously updating knowledge of drug and disease-specific effects.
Once the pulmonary stage starts, “Providing supportive care (with ventilators that themselves stoke the fire) and waiting for the cytokine fire to burn itself out simply does not work… this approach has failed and has led to the death of tens of thousands of patients.”
The systematic use of an early-intervention protocol including methylprednisolone, a steroid, at higher doses than the once-size-fits-all dexamethasone dose, and titrated according to response, “has reduced the hospital mortality from COVID-19 to approximately 3.5%”
The clinical results of methylprednisolone have already been observed in different parts of the world from the US to India.
Would patients want to know the entirety of available information about the biochemistry, histopathology and survival rates of doing essentially nothing but supportive care, before enrolling in a trial where they might only get supportive care? While there may be variation in each centre’s mortality rate, 13% up to 41% versus a total of 3.5% is a significant difference.
What is the fidelity of the knowledge gained from this study? What to offer patients? The “evidence-based” non-titrated, low dose dexamethasone or treatment with titrated doses of methylprednisolone?
The prevailing orthodoxy within the academic establishment is that we should not encourage the use of new treatments for CoViD-19 until those treatments have demonstrated statistically significant clinical benefits in an RCT. Until that point, only patients enrolled in clinical trials should have access to new treatments, no matter how promising the treatment and no matter whether the patient might otherwise be at high risk of death.
However, the fundamentalist belief in the RCT fails to acknowledge the limitations of the RCT in a rapidly changing, highly complex, enormously combinatorial and evolving system in which knowledge is continually being generated by clinicians around the world; the situation during the CoViD-19 pandemic.
From an ethical perspective, discouraging or prohibiting access to off-label/compassionate-use therapies outside of the RCT creates the potential for large disequilibria of the needs of current and future patients, in addition to forsaken clinical information from wide-scale observation.
Depending on the pre-test totality of all knowledge of a treatment, it is difficult to even know if the result of an RCT is a true or false result. And importantly, depending on the pre-test probability, the result of the RCT might only give marginal (or no) extra surety regarding the effect of the studied treatment.
A crucial, existential concept related to knowledge acquisition within a complex system is whether the cumulative observation of clinical effects without a control provides any useable information.
The RCT attempts to isolate a variable using the scientific method. Yet there are systems constraints within acute clinical medicine regarding reproducibility, and this can significantly affect the fidelity of knowledge that the RCT is able to provide. For CoViD-19, multiple therapies and disease and patient variation can easily run into the millions of combinations of experimental conditions. The worldwide concurrent collective experience can often provide higher fidelity to tease out these factors than individual experiments.
In this regard, the RCT has proved to be disappointing in providing efficient clinical knowledge in acute care medicine over the decades. In perioperative medicine, for example, even proponents of the RCT have conceded that there has been a paucity of large RCTs that have shown a clear beneficial treatment effect. Prior to CoViD-19, a major funding body had de-prioritized the RCT for sepsis research.
Should patients be empowered with the complete accumulated knowledge of an intervention as part of informed consent, before enrolling in an RCT?
To be sure, the RCT has a role. However, with its history of being an inefficient clinical-knowledge-gaining paradigm, we should not necessarily wait for clinical trials before introducing reasoned treatment approaches for our most vulnerable CoViD-19 patients.
Author: Michael Keane
1) Swinburne University, Melbourne, Australia.
2) Monash University, Melbourne, Australia.
3) Casey Hospital, Monash Health, Melbourne, Australia
Competing interests: I declare that I am on the Medical Advisory Board of Cannvalate, a medicinal cannabis company.