Therapy
Selenium in Intensive Care (SIC): results of a prospective randomized, placebo-controlled, multiple-center study in patients with severe systemic inflammatory response syndrome, sepsis, and septic shock.
This review may be edited
Summary
Posted By: Gordon S Doig
E-Mail: gdoig@med.usyd.edu.au
Posted Date: 23 Jan 2007
Title: Selenium in Intensive Care (SIC): results of a prospective randomized, placebo-controlled, multiple-center study in patients with severe systemic inflammatory response syndrome, sepsis, and septic shock.
Authors: Angstwurm MW, Engelmann L, Zimmermann T, Lehmann C, Spes CH, Abel P, Strauss R, Meier-Hellmann A, Insel R, Radke J, Schuttler J, Gartner R.
Reference: Crit Care Med. 2007 Jan;35(1):118-26.
Link: Click here for a direct link to the paper. A password may be required for access to fulltext.
Abstract: OBJECTIVE: Sepsis is associated with an increase in reactive oxygen species and low endogenous antioxidative capacity. We postulated that high-dose supplementation of sodium-selenite would improve the outcome of patients with severe sepsis and septic shock.
DESIGN: Prospective randomized, placebo-controlled, multiple-center trial.
SETTING: Eleven intensive care units in Germany.
PATIENTS: Patients were 249 patients with severe systemic inflammatory response syndrome, sepsis, and septic shock and an Acute Physiology and Chronic Health Evaluation (APACHE) III score >70.
INTERVENTIONS: Patients received 1000 microg of sodium-selenite as a 30-min bolus injection, followed by 14 daily continuous infusions of 1000 microg intravenously, or placebo.
MEASUREMENTS AND MAIN RESULTS: The primary end point was 28-day mortality; secondary end points were survival time and clinical course of APACHE III and logistic organ dysfunction system scores. In addition, selenium levels in serum, whole blood, and urine as well as serum glutathione-peroxidase-3 activity were measured. From 249 patients included, 11 patients had to be excluded. The intention-to-treat analysis of the remaining 238 patients revealed a mortality rate of 50.0% in the placebo group and 39.7% in the selenium-treated group (p = .109; odds ratio, 0.66; confidence interval, 0.39-1.1). A further 49 patients had to be excluded before the final analysis because of severe violations of the study protocol. In the remaining 92 patients of the study group, the 28-day mortality rate was significantly reduced to 42.4% compared with 56.7% in 97 patients of the placebo group (p = .049, odds ratio, 0.56; confidence interval, 0.32-1.00). In predefined subgroup analyses, the mortality rate was significantly reduced in patients with septic shock with disseminated intravascular coagulation (n = 82, p = .018) as well as in the most critically ill patients with an APACHE III score > or =102 (>75% quartile, n = 54, p = .040) or in patients with more than three organ dysfunctions (n = 83, p = .039). Whole blood selenium concentrations and glutathione peroxidase-3 activity were within the upper normal range during selenium treatment, whereas they remained significantly low in the placebo group. There were no side effects observed due to high-dose sodium-selenite treatment.
CONCLUSIONS: The adjuvant treatment of patients with high-dose sodium-selenite reduces mortality rate in patients with severe sepsis or septic shock.
 
Are the Results Valid?
1. Was the assignment of patients to treatments randomized? ( Was allocation concealment maintained?)
Unsure. Although the authors do report how the randomisation sequence was generated and which computer programs were used, they do not specifically state the programs were used in such a fashion that allocation concealment was maintained.
2. Were all patients who entered the trial properly accounted for and attributed at its conclusion?
2a. Was followup complete?
No. Although the authors report using an Intention to Treat (ITT) analysis, outcomes on 11 patients (4.4%) are not reported. Furthermore, we are uncertain as to which group these patients were initially randomised to (incomplete reporting).
Although 5 of these 11 patients were excluded from the ITT due to 'withdrawal' of consent, which is an allowable reason, the authors should report which group they were originally randomised to.
It is NOT acceptable to remove patients from an ITT analysis due to suicide, early termination of treatment, non-compliance or loss to follow-up. Patients who are truly lost to follow-up should be assumed to have died. It would then be acceptable to remove these patients from the Per Protocol Analysis.
Twenty four percent of all randomised patients are excluded from the per-protocol analysis. Due to the failure to present true ITT results and the excessive number of excluded patients, the per protocol analysis should be viewed as hypothesis generating.
2b. Were patients analyzed in the groups to which they were randomized?
Yes. It does not appear that patients were crossed over when treatments were incomplete. They were simply excluded from analysis.
3. Were patients, health workers, and study personnel blind to treatment?
Yes, this was a placebo controlled trial.
4. Were the groups similar at the start of the trial?
True baseline balance would include all patients except for those who withdrew consent.
It is very difficult to determine baseline balance by 'eyeballing' Table 2 but the text does report an imbalance in women, gender-specific age and extremes of BMI.
The authors do nor report the results of a multivariate logistic regression model that controls for this gender, gender-age and BMI imbalance.
5. Aside from the experimental intervention, were the groups treated equally?
Although the authors report ventilator days, need for dialyisis or vasopressors were similar, they do not report p-values or rates of each.
The authors do not report daily nutritional intake by group.
What are the Results?
1. How large was the treatment effect?
Intention to treat
Based on the reported intention to treat analysis, which should have included outcomes from six additional patients, there were no significant differences in the primary outcome between treatments.
Mortality:50% vs 39.7%, p=0.109
Per protocol analysis
The per protocol analysis excludes results from 23% of eligible randomised patients. This analysis should be regarded as hypothesis generating.
Sub-group analysis
Based on my counts, the authors conducted 21 distinct sub-group analysis. Since the authors did not use a true test of treatment x sub-group interactions, the p-values should be discounted using a Bonferroni approach. Ex. If we accept p=0.05 as a reasonable threshold for a Type I error when conducting one single analysis, we should accept 0.05 / 21 = 0.002 when conducting five analyses. None of the sub-groups investigated make this strict cut-off. The link to this paper provides excellent additional information on the interpretation of sub-group analyses.
2. How precise was the estimate of the treatment effect?
The 95% confidence interval about the ITT analysis ranges from a 2.2% increase in mortality due to Se to a 23% reduction.
Will the Results Help Me In Caring For My Patients?
1. Can the results be applied to my patient care?
This is an interesting Level II / Phase II trial with non-significant ITT results. It would be prudent to confirm these findings in a subsequent trial before widespread use is considered.
2. Were all clinically important outcomes considered?
The authors did a reasonably good job of considering major outcomes however explicit values for duration of ventilation and need for dialysis and vasopressors should be reported.
Note - according to the CONSORT statement, an ITT analysis is not appropriate for determining adverse events.
3. Are the likely treatment benefits worth the potential harms and costs?
Uncertain at this point in time.

What other people had to say about:
Selenium in Intensive Care (SIC): results of a prospective randomized, placebo-controlled, multiple-center study in patients with severe systemic inflammatory response syndrome, sepsis, and septic shock.

Add Your Comment
[ Back ]

Any questions or comments please contact Gordon S. Doig
Implemented and designed by John Soer and Gordon Doig
Page last modified on Friday August 10, 2001
www.EvidenceBased.net
  THIS IS A SPACE MESSAGE. THIS IS STILL THE BEST METHOD TO SPACE A TABLE FOR NETSCAPE AND INTERNET EXPLORER. IF YOU SEE THIS MESSAGE CONTACT WEBADMIN@EvidenceBased.net