Criteria abstracted from The Users' Guide to Medical Literature, from the Health Information Research Unit and Clinical Epidemiology and Biostatistics, McMaster University

Highlighted lines and questions below provide links to the pertinent description of criteria in The EBM User's Guide, now available at the Canadian Centres for Health Evidence

Article Reviewed:

Please visit the new Evidence Based Journal Club Reviews

Steroid controversy in sepsis and septic shock: A meta-analysis.

Lefering R, Neugebauer E.

Crit Care Med 1995; 23: 1294-1303. [abstract]

Reviewed by Cheri Landers, MD, Texas Children's Hospital

Review posted June 11, 1999

---

I. Are the results of the study valid?

A. Primary questions:

1. Did the overview address a focused clinical question?

Yes. The authors wanted to assess the therapeutic use of glucocorticoids in patients with sepsis, septic shock, or severe infections. They looked at overall steroid effect, differences in drug used, steroid dose, time of administration, type of infection, and adverse events.

2. Were the criteria used to select articles for inclusion appropriate?

Yes. The authors sought to include articles that dealt with clinical trials in human subjects that were critically ill with sepsis, septic shock or severe infection that were treated with glucocorticords. This yielded 49 reports. No attempt was made to select articles with the same definition of sepsis as there was too much variability between studies in their inclusion criteria. Definitions of septic shock or sepsis varied in individual studies from positive blood culture to a clinical definition such as septic history or lowered blood pressure. Only two trials used a detailed definition such as "sepsis syndrome" or "systemic sepsis".

These 49 reports were then evaluated to select comparable trials. To be included, the report had to have the primary goal to evaluate the effectiveness of corticosteroids and be a randomized prospective trial. The patients had to have a general syndrome of sepsis or septic shock rather than specific types of infection. The type, amount and route of drug must be mentioned and mortality had to be an outcome addressed. Ten of the 49 reports met all of these criteria. Two of the ten were abstracts.

B. Secondary questions:

3. Is it unlikely that important, relevant studies were missed?

Uncertain. The authors specified only that they conducted a search of MEDLINE from 1966 to 1992, reviewed their own computerized files and citations of articles retrieved, and contacted experts in the field. However there was no mention of attempts to search to European or other non-MEDLINE sources, and the exact search strategy was not defined.

4. Was the validity of the included studies appraised?

Yes. A lengthy questionnaire was used to quantify the design, conduct and presentation quality of the trials. Both authors evaluated the trials (minus 2 abstracts) independently with reqard to this questionnaire and final rating was determined by consensus discussion.

5. Were assessments of studies reproducible?

This was not addressed by the article, only that there was a "consensus discussion" regarding the quality rating obtained.

6. Were the results similar from study to study?

No, a source of heterogeneity was found to be the Schumer trial (1). A significant test for heterogeneity indicates that the studies are significantly different and analyzing them together may not be appropriate. Once the Schumer trial was eliminated the differences between the studies evaluated in this article were not significantly different.

The authors were unable to identify the source of heterogeneity in the Schumer study other than the subjects were almost all male as it was done at a Veterans Administration hospital. There was no obvious extreme in steroid dose or drug used and not an identifiable difference in the patient characteristics. However, the Schumer article used retrospective analysis to look at a similar group of patients and steroid use. In this group there was no randomization for steroid therapy; the patients were given steroids whenever a benefit was expected. This may lead to the conclusion that steroid administration is beneficial.

Therefore, it does not appear that the prospective portion of the Schumer trial was designed markedly different from the other individual trials reviewed. Because the results of Schumer's article were so different, it might be argued that there must have been a difference in patient characteristics or design not identified. Because of this, it is appropriate to exclude this trial.

II. What are the results?

1. What are the overall results of the review?

Treatment effect was expressed as differences in mortality rates (at 14 days or at hospital discharge) between steroid treated and control groups. After omitting the Schumer trial, the rate difference was in favor of controls (4.8%, i.e., the mortality rate amongst the controls was 4.8% less than the steroid treated patients). Only one study showed significant benefit with steroids (Schumer). Including the Schumer trial, the mortality rate of the treated patients was 0.2% less than controls.

To assess type and dose of steroid, the equivalent hydrocortisone dose for a theoretical patient weighing 70 kg in each study was calculated and the studies were then classified into low dose or high dose groups. Low dose was defined as an equivalent dose of < 20 g equivalents of hydrocortisone in the first 24 hours and high dose was doses at or above 20 g. The low dose group had a trend toward steroid benefit (1.9% mortality rate difference), while the high dose groups tended toward a control benefit (3.6% risk difference).

Only four of ten studies assessed subgroups of patients with gram-negative and gram-positive infections. In patients with gram negative sepsis there was a trend toward steroid benefit (5.6% rate difference) but in gram positive sepsis, there was a trend toward controls (1.8% difference).

The authors were unable to assess the effect of time of drug administration due to lack of information from studies.

GI bleeding had a trend toward benefit of control (2.3% risk difference) in the five studies that recorded this adverse event. Secondary infections in seven studies were equal between groups.

2. How precise were the results?

For every result evaluated and listed in the above question, 95% confidence intervals were given. Each confidence interval for difference in mortality rates, dose and type of drug, type of infection and adverse events included zero. When the confidence interval includes zero, there is no significant difference between results that were compared.

Only 4 individual studies had enough power to reject a 25% steroid-induced reduction of mortality rates.

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

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

Application to the pediatric population is always difficult in studies done with adult subjects. However, if this is overlooked, this data can most likely be applied to patients with sepsis, septic shock, and/or severe infections, although the definitions of these diagnosis must be better defined.

2. Were all clinically important outcomes considered?

Mortality is, of course, the ultimate in clinical outcomes but other important endpoints such as resolution of hypotension or hemodynamic instability reversal, length of hospital stay, length of ICU stay, cost of hospital stay or other cost analysis were not considered and were likely not included in any of the studies evaluated.

3. Are the benefits worth the harms and costs?

No. No benefits were shown with steroid treatment, although no difference in adverse events were noted either.

References:

1. Schumer W. Steroids in the treatment of clinical septic shock. Ann Surg 1976; 184:333-339. [abstract]

Related Reviews

Cronin L, Cook DJ, Carlet J, et al. Corticosteroid treatment for sepsis: a critical appraisal and meta-analysis of the literature. Crit Care Med 1995; 23: 1430-1439. [abstract]

Review, by V. Gharpure, K. Meert

Bollaert PE, Charpentier C, Levy B, et al. Reversal of late septic shock with supraphysiologic doses of hydrocortisone. Crit Care Med 1998; 26:645-650. [abstract]

Review, by A. Torres

Comments

Back to the EB Journal Club Index