The Land of Irrelevance #2 - The MENDS2 Study
The Thoughtful Intensivist is back to strike nonsensical research! Today I recall the MENDS2 Study, one study I just couldn't forget since NEJM published it in 2021. It is so remarkable because it illustrates what happens when researchers blinded to marginal utility try to build a house on the Sepsis swamp. Okay, I know I tend to over-explore metaphors. Let's get more medical: this study was designed during a confusional state. Only God knows why it went on and got the pages of NEJM.
The study follows the path of the MENDS study, which confirmed the strong plausibility assumption that lorazepam would cause more delirium and coma than dexmedetomidine. That was a scenario of high biological plausibility, the study question held to the pharmacodynamics properties of the drugs, and the outcome metrics used had strong internal validation to answer the question. MENDS2 also compares dexmedetomidine with a GABAergic sedative, however, the research hypothesis became a mess when they decided to study sepsis.
Title: Dexmedetomidine or Propofol for Sedation in Mechanically Ventilated Adults with Sepsis
Scenario: Mechanically Ventilated adults with sepsis
Dominant Cause: “dysregulated” immune response to infection and ventilatory insufficiency
Dominant intervention: Infection treatment and any form of mechanical ventilation
Hypothesis: Dexmedetomidine's superior immunomodulatory and neurologic effects will reduce mortality and delirium incidence
Causal chain: That's a hard one. The article assumes that (1) Dexmedetomidine causes less inflammation and increased bacterial clearance and (2) Dexmedetomidine causes less neuronal apoptosis and promotes biomimetic sleep.1 The rest is left to the reader to figure out. I think they implicitly assumed that (3) Less inflammation and increased bacterial clearance cause improvement in Sepsis; (4) Sepsis causes delirium and death; (5) Neuronal apoptosis causes delirium (bold proposition); (6) Lack of biomimetic sleep causes delirium; (6A) Biomimetic sleep is as good as natural sleep to prevent delirium; (7) The combined effects of (4), (5), and (6) cause less delirium and mortality in 14 days.
The tricky part is that (1), (2), and (4) are true, inciting the reader to accept the whole set of propositions and the conclusion (7). However, not even (1) and (2) are solid enough to build on. You can state (1), but it is absurd to state (1) is true in all and only in patients receiving dexmedetomidine. The propofol patients display compensatory anti-inflammatory response and bacterial clearance. The same applies to (2). Moreover, to what extent do the effects mentioned in (1) and (2) manifest? No one knows. Before I get lost in this maze, I will wrap it up as a long and confusing chain of weak assumptions.
Case: Marginal Treatment
The trial studies the marginal effect of the two sedatives in patients at the same degree of sedation. It targets some inflammatory pathways in the overwhelmingly complex immune response present in septic patients. It also targets some neurological… things, completely blind to the fact that these are not dominant causes of mortality and delirium. In reality, no one knows what are the dominant causes of delirium.
Narrative:
The dominant causes of mortality are clear: infection and ventilatory failure. Both causes are addressed in both study arms. All other interventions, including the choice of sedatives, are marginal causes of mortality. Delirium, the other part of the primary outcome, has no biological hallmarks to provide a treatable dominant cause. Delirium is more a symptom than a disease. People should discuss in the first place if delirium is a disease or a common presentation of many conditions.
MINDS2 assumes dexmedetomidine has some neurologic properties that reduce delirium. The researchers were blind to the fact that delirium and death are multifactorial, and in such a scenario one must account for the marginal contribution of each cause. The other naïve assumption was that the “immunomodulation” caused by dexmedetomidine would hit a dominant cause of “dysregulated immune response”, the undefinable condition that defines sepsis, and ultimately reduce sepsis mortality.
So far, the MENDS2 rationale is a bizarre display of something we are used to seeing, the confusion between biological plausibility and clinical relevance. The study design finally entered the realm of a mental condition when it stated it was powered to detect a 12 percentage-point absolute difference in mortality at 90 days.2 It would be the most effective intervention in decades.
The study is remarkable because its rationale combines two conditions, sepsis, and delirium, whose pathology are not described, hence providing no target for intervention. Why did they assume that dexmedetomidine would be a dominant treatment for both? And where did the 12% reduction in mortality come from? Wait. I will rephrase it. Will a stone float on water?
For mistaking biological plausibility for clinical relevance, MENDS2, frankly confusional, crossed the borders of the Land of Irrelevance and disappeared in the Mists of Bad Hypotheses
The induction and maintenance of rapid eye movement (REM) and non-REM stages 1–3 (N1–N3) brain states that neurophysiologically approximate natural sleep. I had to google this one
Assuming an expected mortality of 30% in the propofol group.