Stop Predicting, Start Testing: A Dynamic Approach for Using Dobutamine in Septic Shock

Dobutamine is recommended by current Surviving Sepsis Campaign guidelines for patients with septic shock who have myocardial dysfunction and persistent hypoperfusion despite adequate volume status and mean arterial pressure (MAP) optimization (1). Yet the supporting evidence remains limited, and no validated bedside parameter identifies, in real time, which patients will respond. In this issue of Critical Care Explorations, Tarbaghia et al (2) address this gap in a single-center retrospective cohort study conducted at a quaternary academic center. Among 3407 admissions for septic shock between 2011 and 2020, 251 patients received dobutamine together with vasopressors; 104 who underwent transthoracic echocardiography within 12 hours before dobutamine initiation were included in the analysis. Hemodynamic response was defined by both an increase in central venous oxygen saturation (ScvO₂) of at least five percentage points and a reduction in norepinephrine-equivalent (NEE) dose 2 hours after dobutamine initiation. Thirty-seven patients (36%) met this composite criterion.

Baseline demographics, illness severity, lactate, and vasopressor dose were broadly similar between responders and nonresponders, although baseline ScvO₂ was modestly lower among responders. Importantly, the separation between groups was not confined to the 2-hour window used to define response. At 6 hours, responders continued to show a greater reduction in NEE dose (ΔNEE, −0.13 vs. 0.04 µg/kg/min; p < 0.001), a higher MAP/NEE ratio (380 vs. 166; p = 0.003), and a greater increase in ScvO₂ (+12 vs. +6 percentage points; p = 0.014). Unadjusted 28-day ICU mortality did not differ significantly between responders and nonresponders (41% vs. 58%; p = 0.13). After multivariable adjustment, however, response was associated with lower mortality (OR, 0.18; 95% CI, 0.04–0.68). Responders also had more ICU-free days (median, 17 d [IQR, 0–22 d] vs. 0 d [IQR, 0–15 d]; p = 0.017), whereas the requirement for renal replacement therapy did not differ (43% vs. 37%; p = 0.7). No baseline echocardiographic parameter, including left ventricular ejection fraction (LVEF), left ventricular outflow tract velocity-time integral (LVOT VTI), or tricuspid annular plane systolic excursion, distinguished responders from nonresponders.

Several features of the revised statistical approach are reassuring, including clinically selected covariates, formal collinearity assessment, and a backward-elimination sensitivity analysis that yielded a similar estimate. Nevertheless, the adjusted association warrants caution. First, it is based on only 37 responders among 104 patients, and its wide 95% CI encompasses markedly different plausible effect sizes; the estimate should therefore be regarded as hypothesis-generating rather than confirmatory. Second, the association was not statistically significant in the unadjusted analysis and became substantially stronger after adjustment, indicating considerable model dependence. Moreover, mechanical ventilation, renal replacement therapy, and maximum NEE dose within 24 hours may include measurements obtained after the 2-hour response assessment, and adjustment for variables occurring after treatment initiation or downstream of the evolving shock trajectory may introduce post-treatment bias. Third, baseline ScvO₂ contributes mathematically to the change used to define response. Its identification as a predictor therefore partly reflects mathematical coupling rather than an independent physiologic signal.

The absence of an identified echocardiographic predictor also has a physiologic interpretation. Although LVEF and LVOT VTI are central to the evaluation of left ventricular performance, neither is a direct measure of intrinsic myocardial contractility, and both are strongly influenced by loading conditions and ventriculo-arterial coupling. Consequently, the left ventricle may appear deceptively normal, or even hyperdynamic, despite impaired intrinsic contractile function, as commonly occurs during early septic vasoplegia. Using transesophageal echocardiography, Vieillard-Baron et al (5) described unexpectedly preserved or hyperdynamic left ventricular function during early resuscitation, when afterload was lowest (3). As vasoplegia recedes with resuscitation and treatment of the underlying infection, previously masked myocardial dysfunction may become apparent; the reported incidence of septic cardiomyopathy increased from approximately 18% at 6 hours to 60% at 72 hours (4, 5). In the cohort of Tarbaghia et al (2), echocardiography was performed at variable points in the resuscitation course across a 10-year study period. A single resting examination was therefore unlikely to capture this dynamic, afterload-dependent phenomenon consistently—a limitation related at least as much to timing as to technique.

This interpretation also reframes what an informative echocardiographic assessment might entail. Earlier dobutamine-challenge studies suggest that the response to the drug itself, rather than resting cardiac function, may be more prognostically informative. Kumar et al (6) found that changes in cardiovascular performance during a dobutamine stress test, rather than baseline measurements, distinguished survivors from nonsurvivors with severe sepsis or septic shock. Creteur et al (7) similarly used a dobutamine test to reveal hepato-splanchnic hypoperfusion not apparent from resting measurements. Direct reassessment after dobutamine—using stroke volume index or global longitudinal strain, a deformation-based measure that may predict mortality more reliably than LVEF in sepsis and septic shock (8)—could characterize physiologic responsiveness more precisely than the resting parameters examined here. The ongoing Adjunctive DobutAmine in sePtic Cardiomyopathy With Tissue Hypoperfusion (ADAPT) trial applies this principle prospectively by restricting enrollment to patients with objectively impaired cardiac performance and low flow (LVEF ≤ 40% and LVOT VTI < 14 cm) rather than an unselected population (9).

A more fundamental point, however, is that macrohemodynamic optimization is a means, not an end. Septic shock reflects a complex interaction among pathologic vasodilation, relative or absolute hypovolemia, microvascular flow abnormalities, and cardiac dysfunction; hemodynamic resuscitation must therefore be individualized. Restoration of systemic hemodynamics is frequently dissociated from improvement in microcirculatory flow, and persistent microcirculatory abnormalities are associated with organ dysfunction and mortality in sepsis (10). Conversely, low-dose dobutamine may improve tissue perfusion without substantial changes in systemic hemodynamics. De Backer et al (11) found that dobutamine improved sublingual capillary perfusion independently of changes in cardiac index or arterial pressure. Using intravital microscopy of the intestinal microcirculation, Ospina-Tascón et al (12) subsequently showed that low-dose dobutamine reduced microvascular flow heterogeneity and the splanchnic oxygen extraction ratio despite unchanged systemic oxygen delivery. These findings are consistent with a microcirculatory effect not fully explained by changes in systemic hemodynamics and support targeting tissue perfusion rather than macrohemodynamic surrogates alone.

Translating this principle to the bedside, capillary refill time (CRT), which guided resuscitation in ANDROMEDA-SHOCK (13), was later adopted in ANDROMEDA-SHOCK-2 (14) as the target of a personalized resuscitation protocol, in which dobutamine served two purposes. First, it was an optimization option when basic echocardiography identified ventricular dysfunction at the start of Tier 1; CRT normalized in 23 of 65 patients (35.4%) after dobutamine. Second, and perhaps more important, it was used as a fixed-dose, 1-hour hemodynamic test at the end of Tier 2, in patients whose CRT remained prolonged despite no detected ventricular dysfunction; CRT normalized in 17 of 43 patients (39.5%). The latter response could reflect two situations: 1) basic echocardiographic screening, which relies on elementary parameters rather than deformation indices such as global longitudinal strain, may have missed the same afterload-dependent contractile impairment discussed above, so a normal screening echocardiogram did not rule out a genuine dobutamine responder; or 2) the fixed low dose of dobutamine may have acted directly on the microcirculation, independently of preintervention macrohemodynamics. In either case, a perfusion-anchored dobutamine trial may characterize response more meaningfully than a single resting echocardiographic assessment or changes in vasopressor dose and ScvO₂ alone.

None of these concerns detracts from the contribution of Tarbaghia et al (2). Retrospective analyses of dobutamine physiology in septic shock are scarce, and their trajectory-based responder framework, despite its methodological limitations, appropriately shifts attention from a static measurement toward the patient’s dynamic response. That conceptual shift, rather than the specific adjusted odds ratio, may be the study’s most durable contribution. It is also consistent with a broader movement in the field: a recent position article explicitly advocated personalizing rather than protocolizing hemodynamic management in septic shock (15).

What ultimately withstands scrutiny, therefore, is not the identification of a baseline echocardiographic predictor, but the observation that the early hemodynamic trajectory may convey more prognostic information than a single snapshot. This finding does not justify changing practice based on the adjusted mortality estimate, which remains fragile, model-dependent, and hypothesis-generating. It does, however, support prospective evaluation of a time-limited, perfusion-anchored dobutamine trial followed by reassessment with repeated hemodynamic or perfusion measures, such as stroke volume or CRT. Framed in this way, the principal contribution of Tarbaghia et al (2) is not a statistical rule predicting who will respond, but a demonstration that observed response—rather than baseline cardiac anatomy alone—may ultimately prove more useful in determining whether to continue dobutamine.