PedsCCM: The Pediatric Critical Care Website:
The multidisciplinary and practical resource for Pediatric Critical Care on the Internet Peer reviewed original content integrated with links to selected web resources ... now 19 years old... thanks to everyone for your great support!..


Criteria abstracted from The Users' Guides to the Medical Literature series in JAMA

Interaction of vasopressin infusion, corticosteroid treatment, and mortality of septic shock.

Russell JA, Walley KR, Gordon AC, Cooper DJ, Hébert PC, Singer J, Holmes CL, Mehta S, Granton JT, Storms MM, Cook DJ, Presneill JJ; Dieter Ayers for the Vasopressin and Septic Shock Trial Investigators.

Crit Care Med 2009 Mar;37(3):811-8..[abstract]

Reviewed By: Danny Castro DO, Baylor College of Medicine, Texas Children's Hospital, Houston TX

Review posted March 27, 2010

  1. What is being studied?

    The study objective:

    The objective of this study was to determine whether the interaction of vasopressin and corticosteroid therapy had an effect on the mortality of adult patients with septic shock.

    The study design:

    This was a post hoc analysis of a multi-center, stratified, randomized, double-blinded trial that was performed in a population of adult patients with septic shock that sought to determine whether vasopressin (AVP) decreased 28-day mortality versus Norepinephrine (NE).  (1)

    The patients included:

    Patients greater than 16 years old with septic shock, as defined by the ACCP/SCCM Consensus Conference Committee (2), that was resistant to:

    • Fluids, which was defined as lack of response to 500ml of normal saline or the requirement of vasopressors


    • “Low-dose” NE, defined as 5 mcg/min or the equivalent, for at least 6 consecutive hours in the preceding 24 hours and within the last hour prior to randomization
      • Patients requiring “high” vasopressor doses, defined as NE or the equivalent of 15mcg/min, could be randomized at 3 hours.
      • The NE equivalents, as well as other definitions, were defined in an online supplement of the original article.  (3)

    The patients excluded:

    An extensive list of exclusion criteria were also outlined in the original article’s online supplement. (3)

    The interventions compared:

    Using all patients from the VASST trial that received corticosteroids, the authors compared subjects who received AVP + corticosteroids versus those who received NE + corticosteroids.  Corticosteroid use was defined as the administration of corticosteroids for at least 1 day during the 28 day observation period.  They also compared subjects who received AVP w/o corticosteroids versus those who received NE w/o corticosteroids.

    The outcomes evaluated:

    The primary outcome measurement was 28 day mortality.  Secondary outcomes were 90 day mortality and days alive and free from organ dysfunction over the first 28 days.  They also assessed the interaction of AVP and corticosteroid therapy.

  2. Are the results of the study valid?

      Primary questions:

    1. Was the assignment of patients to treatments randomized?

      No. The subjects were randomized for the purposes of the original study; however, the authors acknowledge that assignment to corticosteroid treatment was neither randomized nor blinded. Administration of corticosteroids was not protocolized and at the discretion of the treating physician, thus introducing a degree of bias.

    2. Were all patients who entered the trial properly accounted for and attributed at its conclusion?

      Yes. All the subjects who were enrolled into the trial were properly accounted for. There was one subject, in the AVP + corticosteroid group, who was lost to follow up and not included in the final analyses.

      Was follow-up complete?

      Yes. Follow-up was sufficiently long and complete for the purposes of this study. As stated previously, only 1 subject was lost to follow up. This loss was not accounted for in the final analyses; however, the loss of this 1 subject is not enough to weaken the strength of inference.

      Were patients analyzed in the groups to which they were randomized?

      There was no randomization for the administration of corticosteroids, the particular intervention of interest. The patients who were originally randomized to either the AVP or NE groups were analyzed in their respective groups.

    3. Secondary questions:

    4. Were patients, health workers, and study personnel "blind" to treatment?

      All staff, except the study pharmacists, were blinded to the treatment assignments of subjects to either the NE or AVP group. The assignment to receive or not to receive corticosteroids was neither randomized nor blinded.

    5. Were the groups similar at the start of the trial?

      Yes, the groups were fairly similar. Of note, within the corticosteroid group, there was a trend towards subjects being significantly younger and having more hematologic dysfunction. In the no corticosteroid group, there was also a trend towards subjects having significantly more hematologic dysfunction. Although there was no statistical significance, there again was an age discrepancy in this group with the AVP subgroup being younger by about 3 years. The findings of the AVP subgroups of both the + steroids and – steroid groups being younger than their respective NE subgroups might have clinical significance as more elderly individuals may have more co-morbidities. The lack of statistical significance could be due to the smaller sample size in the no corticosteroid group.

      Additionally, the original article defined and stratified (a priori) patients with septic shock into a “more severe shock” stratum and a “less severe shock” stratum. The patients who received corticosteroids had a higher proportion of patients who were in the “more severe shock” stratum than the group who received no corticosteroids. Due to this fact, the authors decided to not compare the corticosteroid treatment group with the no corticosteroid treatment and instead compare AVP versus NE within each group (+ steroid and – steroid) and test for interaction.

    6. Aside from the experimental intervention, were the groups treated equally?

      There is no mention of any differential treatment within the no steroid group. As for the steroid group, several types of steroids were used but the authors state that there were no differences between the AVP and NE subgroups in the hydrocortisone dose received and the numbers of patients treated each day. However, although the number of patients receiving corticosteroids each day was roughly similar between the NE and AVP subgroups, for most of the days in the 28 day observation period the hydrocortisone dose in the AVP subgroup was less.

      As stated previously, corticosteroid use was defined as the administration of corticosteroids for at least 1 day during the 28 day observation period. No data was given on average or median days of steroid therapy thus duration of steroid therapy could potentially have been different for either subgroups within the + steroid group.

  3. What were the results?

    1. How large was the treatment effect?

      Neither absolute risk reduction, relative risks nor relative risk reductions were reported in this article. The authors reported a statistically significant difference between the AVP + steroid group and the NE + steroid group in 28 day mortality (35.9% vs. 44.7%; p = 0.03) and 90 day mortality (45.2% vs. 55.5%; p = 0.01). When calculated, the ARR, RR and RRR are 0.088 (95% CI 0.009 to 0.167), 0.8 and 0.2 (0.02 to 0.34), respectively, for 28 day mortality and 0.10 (95% CI 0.02 to 0.18), 0.81 and 0.19 (95% CI 0.04 to 0.31, respectively, for 90 day mortality.

      Additionally, they reported a statistically significant difference in days alive and free of any organ dysfunction (median [IQR]) between the AVP + steroid group (0 [0,3]) and the NE + steroid group (0 [0,9]), p = 0.02. More specifically, there was a significant difference between the AVP + steroid group and the NE + steroid group in days alive and free of ventilation (8 vs. 3, p = 0.03) and a trend towards significance in days alive and free of cardiovascular (18.5 vs. 12, p = 0.09) and renal dysfunction (21 vs. 15, p = 0.07).

      Lastly, the author's reported a statistically significant interaction between vasopressin treatment and corticosteroid treatment (p = 0.008) even after adjustment for age (p = 0.007) and age, gender, APACHE II score and severity of shock (p ≤ 0.02).

    2. How precise was the estimate of the treatment effect?

      For 28 day mortality, the true estimate of the ARR between the AVP + steroid group and AVP + NE group may be as low 0.009 or as high as 0.167 and the true estimate of the RRR may be as low as 0.02 or as high as 0.34. For 90 day mortality, the true estimate of the ARR between the AVP + steroid group and NE + steroid group may be as low as 0.02 or as high as 0.18 and the true estimate of the RRR may be as low as 0.04 or as high as 0.31. Thus, for both 28 day and 90 day mortality, the 95% CI for the ARR and the RRR have wide ranges around the point estimate and are therefore not very precise. However, since the 95% CI of both the ARR and the RRR do not cross zero, they do achieve statistical significance.

      For days alive and free of ventilation, although the difference in median days was statistically significant, the interquartile range (IQR) were similar between the AVP + steroid group and the NE + steroid group [(0, 20) vs. (0, 18.75), respectively]. The same could be said of the trends toward significance appreciated for days alive and free of cardiovascular and renal dysfunction. This raises the question of whether there is a true clinical difference. The IQR for days alive and free of any organ dysfunction are (0,9) and (0,3), for the AVP + steroid group and the NE + steroid group, respectively. Although the median for both are zero, the IQR seems to imply more of a true difference as the upper limit of the IQR is 9 days for the AVP + steroid group as opposed to 3 days for the NE + steroid group.

      With regards to their logistic regression analysis of the interaction effect between corticosteroids and AVP, the p value was highly significant even after adjustment for multiple variables, including severity of shock and age, which were clearly different between the “steroid” group and the “no steroid” group. However, odds ratios and accompanying confidence intervals were not provided. Presumably the odds ratios and CI's could have been provided, however, there would be debate as to how they were derived and whether they were truly meaningful. (4)

  4. Will the results help me in caring for my patients?

    1. Can the results be applied to my patient care?

      No. The findings of a highly statistically significant interaction test between AVP and corticosteroids, as well as, AVP levels being significantly increased by corticosteroid therapy are very intriguing. However, several study limitations pose questions as to whether the results are of clinical utility.

      First, this was a post hoc analysis of a study whose primary objective was completely different from this study’s. Thus, there was a lack of randomization to the intervention of interest. Additionally, but not surprisingly, this resulted in a selection bias of the more severely ill into the steroid intervention group.

      Lastly, their definition of “use of corticosteroid”, “administration of corticosteroids for at least 1 day…”, allowed for significant variance in steroid administration. There was no average or median days of steroid administration given.

    2. Were all clinically important outcomes considered?

      Yes, for the most part. Mortality and days alive and free of organ dysfunction are clinically important outcomes. More importantly, the author’s also looked at days alive and free of organ dysfunction in survivors. There were no statistically significant differences between the NE and AVP subgroups in the steroid group and the NE and AVP subgroups in the no steroid group. The median days and IQR between the four groups were very similar. Serious adverse events were also considered, however, superinfection would have been a clinically important adverse event to report as the CORTICUS trial demonstrated more episodes of superinfection in their steroid group. (5)

    3. Are the likely treatment benefits worth the potential harms and costs?

      Unclear. This study suggests that an interaction between vasopressin and corticosteroids is present. However, given the study design, this should only be used for hypothesis generation, not convincing evidence of benefit.


  1. Russell, James A., Walley, Keith R., Singer, Joel, Gordon, Anthony C., Hebert, Paul C., Cooper, D. James, Holmes, Cheryl L., Mehta, Sangeeta, Granton, John T., Storms, Michelle M., Cook, Deborah J., Presneill, Jeffrey J., Ayers, Dieter, the VASST Investigators.  Vasopressin versus Norepinephrine Infusion in Patients with Septic Shock.  N Engl J Med 2008 358: 877-887.
  2. American College of Chest Physicians/Society of Critical Care Medicine Consensus Conference: definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. Crit Care Med 1992; 20: 864-874.
  3. Online Supplementary Appendix. Russell JA et al. Vasopressin versus Norepinephrine Infusion in Patients with Septic Shock. N Engl J Med 2008; 358: 877-87.
  4. Chen JJ.  Communicating Complex Information: The Interpretation of Statistical Interaction in Multiple Logistic Regression Analysis.  Am J Public Health 2003; 93(9): 1376-77.
  5. Sprung, Charles L., Annane, Djillali, Keh, Didier, Moreno, Rui, Singer, Mervyn, Freivogel, Klaus, Weiss, Yoram G., Benbenishty, Julie, Kalenka, Armin, Forst, Helmuth, Laterre, Pierre-Francois, Reinhart, Konrad, Cuthbertson, Brian H., Payen, Didier, Briegel, Josef, the CORTICUS Study Group,
    Hydrocortisone Therapy for Patients with Septic Shock N Engl J Med 2008 358: 111-124. founded 1995

Questions or comments?

Provide Feedback | Editorial Notes | About Us