Louis de Bernard (he/him)

#ICU Evidence #002: Earlier vasopressin in septic shock (VASSPR trial): are we treating the patient or the norepinephrine dose? Add vasopressin at a norepinephrine-equivalent dose of 0.1 or 0.4 µg/kg/min? VASSPR tested these strategies across 13 US hospitals and nearly 4,000 patients. It changed practice, but not outcomes. I think it also raises a more interesting question: Should vasopressor dose alone tell us when to add another hemodynamic intervention? This was a pragmatic cluster-crossover trial. Hospitals alternated monthly between recommending fixed-dose vasopressin at a norepinephrine-equivalent dose ≥0.1 µg/kg/min (0.1 arm) or ≥0.4 (0.4 arm). Clinicians retained discretion, so this was not vasopressin vs placebo or a protocol that overrode clinical judgment. Among patients reaching the relevant vasopressor range, vasopressin was used in 68.0% in the 0.1 arm vs 42.7% in the 0.4 arm. But 28-day mortality was 31.8% vs 33.0%, 90-day mortality 37.9% vs 38.4%, and clinically significant arrhythmias 18.4% vs 18.1%. None of the other secondary clinical outcomes materially differed either. Here is what bothers me. Two patients on exactly the same norepinephrine dose may have profoundly different circulations: vasoplegia with preserved cardiac output, septic cardiomyopathy with impaired forward flow, RV dysfunction, persistent tissue hypoperfusion despite acceptable macrohemodynamics, or some combination. A norepinephrine dose tells us how much drug we are giving, not WHY the patient remains in shock. So I struggle with using a drug-dose threshold to trigger another major hemodynamic intervention. Starting vasopressin changes vascular tone and therefore the loading conditions under which the heart and circulation operate. Before treating the blood pressure, I want to understand the circulation. POCUS is an excellent bedside tool here: LV and RV function, chamber size, important valvular or pericardial pathology, and clues about filling and forward flow, integrated with examination, perfusion, lactate trajectory, and other hemodynamic data. This also brings to mind the four-hemodynamic-interface model proposed by @ThinkingCC and colleagues. Increasing vascular tone may be exactly what a profoundly vasoplegic patient needs. In another phenotype, excessive vasoconstriction could adversely affect ventricular-arterial coupling (Interface I, the heart pumping into the arterial system) and potentially arterial-to-microcirculatory coupling (Interface II, translating arterial pressure and flow into tissue perfusion). A higher MAP is not necessarily better hemodynamics. And we should look again after starting vasopressin. Did MAP rise? Fine. But what happened to cardiac function, forward flow and tissue perfusion? Is the RV tolerating the new loading conditions? Can catecholamines actually be reduced? Otherwise, hemodynamic management risks becoming “set it and forget it”: reach a norepinephrine threshold → add vasopressin → celebrate the MAP → move on. One safety finding caught my attention. Clinically diagnosed cardiogenic shock occurred in 7.4% in the 0.1 arm vs 6.3% in the 0.4 arm (adjusted OR 1.19, 95% CI 0.93–1.54). This was an exploratory endpoint and the difference was not statistically significant, so I would not interpret it as evidence that earlier vasopressin causes cardiogenic shock. I do question the endpoint itself. Cardiogenic shock was captured as a clinical diagnosis rather than defined by a pre-specified objective hemodynamic phenotype. I appreciate that systematically phenotyping cardiac function in a pragmatic 4,000-patient trial is not trivial. I’d still want something more objective than a clinical diagnosis, especially with echo this accessible. There’s some irony here: we study when to alter vascular tone, see a hint of a cardiac signal, and still know little about what the heart and circulation were actually doing. My criticism is of the paradigm, not the investigators. Running a cluster-crossover trial across 13 hospitals and actually changing bedside vasopressin use is an impressive accomplishment. VASSPR answers the pragmatic policy question it was designed to answer. What it cannot tell us is whether earlier vasopressin might benefit one hemodynamic phenotype while being unhelpful, or potentially disadvantageous, in another. Perhaps our next trials should ask less “What norepinephrine dose have we reached?” and more: “Why does this patient still need norepinephrine, and what happened after we intervened?” “Pragmatic” explains the design. It does not replace physiology. #JAMA #FOAMcc #FOAMed