The second step, based on CDV at 24 h, made it possible to improve the diagnostic accuracy for patients whose risk level at 6 h had not (yet) reached the threshold identified by our study to be associated with adverse outcome. With a cutoff of 2,600 μg/kg at 24 h, the CDV yielded a specificity of 98%, sensitivity of 64%, positive predictive value of 81%, and negative predictive value of 97%. Overall, this two-step strategy identified 18 high-risk patients at 6 h, of whom 15 presented EDASS. The second evaluation at 24 h allowed identification of nine further high-risk patients, of whom six subsequently had EDASS at 72 h. At 72 h, 313 (93%) of the patients classified as low risk were still alive (Fig. 3). Overall, the model yielded sensitivity of 53%, positive predictive value of 78%, specificity of 98%, and a negative predictive value of 96%.
Finally, the two-step strategy defined here remained strongly associated with EDASS (odds ratio = 67.3 [95% confidence interval, 17.9–252.2], P < 0.0001) after adjusting for baseline SOFA score, bacteremia and age. By multivariate analysis, the addition of corticosteroids to vasopressors did not change the overall results, and corticosteroid therapy was not found to be significantly associated with early death.
Our study confirms that the new metric defined in our study, namely early CDV is a strong prognostic factor for mortality attributable to septic shock. The association between vasopressor doses and mortality in sepsis has previously been described in observational studies. In these studies, only the peak dose of vasopressors was considered with a given threshold such as 1 μg/kg/min (10, 11) or 2 μg/kg/min (12), regardless of when this threshold was reached. None of these studies recorded vasopressor dose at different time points over the course of disease. On the contrary, our study is the first, to the best of our knowledge, to propose a metric reflecting the cumulative dose of vasopressors based on hourly measurements, and to show the relation between this metric and the specific causes and timing of death. Our study further showed that vasopressor doses were similar in patients alive at 72 h and in patients who had died from other causes, whereas EDASS patients have much higher CDV, whatever the time point. This confirms the strong relation between early death and vasopressor resistance in sepsis. It was recently demonstrated that vasopressor dependency and refractory shock could be predicted by a mean arterial pressure-phenylephrine dose–response curve (7). Our results suggest that the simple calculation of CDV using a two-step strategy at 6 and 24 h of septic shock could predict the same outcome, probably because the CDV reflects the same underlying phenomenon of catecholamine resistance.
Our study enabled us to design a simple predictive model able to select patients at very high risk of mortality. The 6-h time point was chosen as it allowed an initial rapid and early evaluation of patients at high risk of EDASS. The patients with CDV below the first cutoff underwent a second evaluation at 24 h in order to refine their risk and identify any new patients with an increased risk of EDASS. Indeed, at 24 h, 55% of the patients who would subsequently die by 72 h were still alive, and therefore eligible for further therapeutic intervention. The predictive power of our two-step model remained high, even after adjustment for confounders.
For this model, we chose the thresholds (6 and 24 h) based on the best compromise between specificity and positive predictive value, rather than that of the best agreement rate (maximized at 89% with cutoffs at 300 μg/kg at 6 h and 1,300 μg/kg at 24 h), to select the patients at highest risk, with a low false-positive rate in the perspective of proceeding with a new therapeutic approach in these patients. This two-step strategy was chosen in order to best identify high-risk patients (according to the CDV) in order to guide therapy. Efficient and early selection of patients with a high mortality risk in a very short time is a major challenge if we are to test strategies to reduce early mortality in future trials. We believe that this kind of model might be a robust tool for the selection of patients in future clinical trials targeting early mortality attributable to vasopressor resistance. New vasopressors and vasoactive agents are currently under development, and have the potential to address the issue of catecholamine resistance, although we currently lack the means to identify patients who might yield the most benefit from these treatments. The recently published Angiotensin II for the Treatment of High-Output Shock-3 study reported that angiotensin II effectively increased blood pressure in patients with vasodilatory shock, but the effect on mortality was not statistically significant (8). In this trial, as in others, patient selection was based on an arbitrary norepinephrine flow rate cutoff. We think that in the setting of such trials, the early CDV could help to target, earlier and more accurately, the patients in need of additional vasopressor support.
Further studies with longer follow-up are warranted to consolidate these findings and refine cutoffs.
The absence of consensus regarding the definition of refractory shock was a major impediment to this study. For this reason, we had to use customized criteria to define the “early death” group, although the criteria chosen were very pragmatic. We firmly believe that there is a compelling need to adopt a consensus definition to define this population.
In conclusion, the CDV within the first 6 h following ICU admission is strongly associated with the risk of early death attributable to septic shock. We describe a two-step model, which allows the identification of a very-high risk population in the first 24 h following vasopressor initiation.
Our results propose an original approach to the ill-defined concept of RSS. Although our analysis is limited to the “vasopressor resistance” component, we believe that the population identified by the new strategy described here represents a first step toward a more broadly accepted definition of “RSS.”
APPENDIX: LIST OF INVESTIGATORS IN THE EPISS STUDY
J.P. Quenot, P.E. Charles, S. Prin, A. Pavon, S. Barbar, University Hospital Bocage, Dijon, France; K. Kuteifan, J. Mootien, P. Guiot, Centre Hospitalier, Mulhouse, France; F. Kara, Centre Hospitalier, Haguenau, France; M. Hasselmann, P. Sauder, F. Ganster, O. Martinet, NouvelHopital Civil, Strasbourg, France; V. Castellain, F. Schneider, HopitalHautepierre, Strasbourg, France; J.C. Navellou, G. Capellier, Centre HospitalierUniversitaire, Besancon, France; A. Noirot, P. Daoudal, Centre Hospitalier, Vesoul, France; O. Ruyer, M. Feissel, J.P. Faller, Centre Hospitalier, Belfort, France; B. Levy, A. Gerard, J. Perny, P. Perez, HopitalBrabois, Nancy, France; S. Gibot, P.E. Bollaert, D. Barraud, A. Cravoisy, Hopital Central, Nancy, France; A.M. Gutbub, P. Rerat, G. Laplatte, H. Lessire, Centre Hospitalier, Colmar, France; C. Mezher, Centre Hospitalier, Montbeliard, France; J. Cousson, T. Floch, Hopital Robert Debré, Reims, France; G. Louis, J.F. Poussel, Centre Hospitalier, Metz, France.
1. Quenot JP, Binquet C, Kara F, Martinet O, Ganster F, Navellou JC, Castelain V, Barraud D, Cousson J, Louis G, et al. The epidemiology of septic shock
in French intensive care units: the prospective multicenter cohort EPISS study. Crit Care
2013; 17 2:R65.
2. Daviaud F, Grimaldi D, Dechartres A, Charpentier J, Geri G, Marin N, Chiche JD, Cariou A, Mira JP, Pene F. Timing and causes of death in septic shock
. Ann Intensive Care
2015; 5 1:16.
3. Hotchkiss RS, Monneret G, Payen D. Sepsis-induced immunosuppression: from cellular dysfunctions to immunotherapy. Nat Rev Immunol
2013; 13 12:862–874.
4. Blair E, Cowley RA, Tait MK. Refractory septic shock
in man. Role of lactate and pyruvate metabolism and acid-base balance in prognosis
. Am Surg
5. Bellissant E, Annane D. Effect of hydrocortisone on phenylephrine—mean arterial pressure dose-response relationship in septic shock
. Clin Pharmacol Ther
2000; 68 3:293–303.
6. Benedict CR, Rose JA. Arterial norepinephrine changes in patients with septic shock
. Circ Shock
1992; 38 3:165–172.
7. Conrad M, Perez P, Thivilier C, Levy B. Early prediction of norepinephrine dependency and refractory septic shock
with a multimodal approach of vascular failure. J Crit Care
2015; 30 4:739–743.
8. Khanna A, English SW, Wang XS, Ham K, Tumlin J, Szerlip H, Busse LW, Altaweel L, Albertson TE, Mackey C, et al. ATHOS-3 Investigators. Angiotensin II for the treatment of vasodilatory shock. N Engl J Med
2017; 377 5:419–430.
9. Ranieri VM, Thompson BT, Barie PS, Dhainaut JF, Douglas IS, Finfer S, Gardlund B, Marshall JC, Rhodes A, Artigas A, et al. Drotrecogin alfa (activated) in adults with septic shock
. N Engl J Med
2012; 366 22:2055–2064.
10. Brown SM, Lanspa MJ, Jones JP, Kuttler KG, Li Y, Carlson R, Miller RR 3rd, Hirshberg EL, Grissom CK, Morris AH. Survival after shock requiring high-dose vasopressor therapy. Chest
2013; 143 3:664–671.
11. Martin C, Medam S, Antonini F, Alingrin J, Haddam M, Hammad E, Meyssignac B, Vigne C, Zieleskiewicz L, Leone M. Norepinephrine: not too much, too long. Shock
2015; 44 4:305–309.
12. Jenkins CR, Gomersall CD, Leung P, Joynt GM. Outcome of patients receiving high dose vasopressor therapy: a retrospective cohort study. Anaesth Intensive Care
2009; 37 2:286–289.
13. Gaies MG, Gurney JG, Yen AH, Napoli ML, Gajarski RJ, Ohye RG, Charpie JR, Hirsch JC. Vasoactive-inotropic score as a predictor of morbidity and mortality in infants after cardiopulmonary bypass. Pediatr Crit Care Med
2010; 11 2:234–238.
14. Wernovsky G, Wypij D, Jonas RA, Mayer JE Jr, Hanley FL, Hickey PR, Walsh AZ, Chang AC, Castaneda AR, Newburger JW, et al. Postoperative course and hemodynamic profile after the arterial switch operation in neonates and infants. A comparison of low-flow cardiopulmonary bypass and circulatory arrest. Circulation
1995; 92 8:2226–2235.