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

Corticosteroid treatment for sepsis: a critical appraisal and meta-analysis of the literature

Cronin L, Cook DJ, Carlet J, Heyland DK, King D, Lansang MA, Fisher CJ Jr.

Crit Care Med 1995; 23: 1430-1439. [abstract]

Reviewed by Varsha Gharpure, MD and Kathleen Meert, MD, Children's Hospital of Michigan

Review posted January 17, 1999

---

I. Are the results of the study valid?

A. Primary questions:

1. Did the overview address a focused clinical question?

Yes. The objective of the meta-analysis is 'To determine the effect of corticosteroid therapy on morbidity and mortality in patients with sepsis'.

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

Yes. Selection criteria used to identify studies were clearly stated as follows:

1. Study design: Randomized clinical trial
2. Population: Adult patients with sepsis or septic shock were included. Authors defined sepsis as evidence of clinical infection and signs of systemic response to the infection (hyperthermia > 38.5°C or hypothermia < 35.5°C), tachycardia (> 90 beats/min), tachypnea (respiratory rate of > 20 breaths/min) and bacteremia. Septic shock was defined as the presence of sepsis accompanied by a sustained decrease in systolic blood pressure to < 90 mm Hg or a decrease of 40 mm of Hg from baseline.
3. Intervention: Treatment with intravenous corticosteroids.
4. Outcome: Mortality and any of the following complications: a) secondary infection (new clinical infection occurring during therapy or within 10 days of discontinuation of antimicrobial agents); b) upper gastrointestinal bleeding; c) progression of organ dysfunction and d) hyperglycemia.

B. Secondary questions:

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

Yes, it is unlikely. The authors searched MEDLINE and EMBASE 1966 to 1993. The Science Citation Index was used to locate additional articles. Personal files and reference lists of all primary studies and review articles retrieved were examined to search for other relevant studies. In addition, manual search of Index Medicus from 1951 forward was also undertaken. Comprehensive list of relevant articles were also sent to the first authors of each study included in the meta-analysis to ask if they knew of any other published or unpublished studies in this field. Primary investigators were contacted to provide more information when data was missing or unclear. There is no mention if they included literature from all languages.

4. Was the validity of the included studies appraised?

Yes. An extensive methodologic quality scoring system was used to critically appraise each of the primary studies. The methodologic assessment included an evaluation of patient selection, baseline characteristics, randomization, blinding, interventions, complications and adherence to protocol. Five studies were deemed to be of high methodologic quality (score 8) and 4 of lower quality (score < 8).

5. Were assessments of studies reproducible?

Yes. Initially, two of the authors reviewed the titles and abstracts of all articles identified in the literature search. Potentially relevant articles were retrieved. These were reviewed for possible inclusion in the meta-analysis by three authors independently. One of the 3 was blinded to the journal, authors, institution and the magnitude and direction of the results. The level of agreement on the selection of relevant studies was excellent (kappa of 0.84). Using specific selection criteria (outlined in question 2 above), 11 papers were identified for inclusion. However, 2 of the 11 were secondary publications reporting on a subset of the original patients. Therefore 9 studies (1232 patients) were included. There was no difference in agreement among the blinded and the two nonblinded reviewers, nor between the 2 nonblinded reviewers (kappa of 1.0).

Two of the authors independently assessed the validity of the 9 studies selected. To avoid bias in the assessments, one reviewer was again blinded. Authors of the primary study were also asked to ensure the accuracy of the methodological scores and to provide important additional information when it was not reported in their publication. The weighted kappa for agreement between reviewers on each quality assessment item was 0.71 to 1.0. Disagreements were resolved by discussion and consensus.

6. Were the results similar from study to study?

Almost. All 9 primary studies except Schumer et al. have showed a trend toward increased mortality with steroid use. However, test for homogeneity demonstrated statistically significant heterogeneity across studies (p = 0.02). Methodologic quality was not responsible for this heterogeneity. The authors do not mention whether exclusion of Schumer's study eliminates heterogeneity. In another meta-analysis published on the same topic by Lefering et al. similar heterogeneity has been reported (1). Lefering mentioned that if Schumer's study was excluded the heterogeneity disappears.

A subset of these 9 studies (n = 6) evaluating steroids in septic shock again showed no beneficial effects. The higher quality studies showed a trend towards increased mortality (RR 1.12, 95% CI 0.95-1.32) whereas lower quality studies showed a significant benefit (RR 0.49, 95% CI 0.25 - 0.98). Thus, for the trials in septic shock, methodologic quality can explain heterogeneity across trial results.

II. What are the results?

1. What are the overall results of the review?

Treatment of sepsis and septic shock with steroids showed a trend towards increased mortality (relative risk 1.13, 95% CI 0.99 to 1.29). In the subset of septic shock studies no effects of steroids on mortality was observed (RR 1.07, 95% CI 0.91 - 1.26).

Regarding morbidity, not all primary studies presented data in a manner amenable to statistical pooling. The following complications were considered:

1. Secondary infections: No increase in secondary infections was reported due to steroids (n = 911, RR 0.92, 95% CI 0.67 - 1.26).
2. GI bleeding: A trend toward increased GI bleeding was observed (n = 527, RR 1.17, 95% CI 0.79 - 1.73).
3. Renal and Hepatic function: Only one of the selected studies described the effect of steroids on renal and hepatic function. BUN was greater in the steroid treated patients upto 7 days post randomization. More steroid treated patients had high bilirubin for the first 24 hours whereas more placebo patients had high bilirubin at 14 days.
4. ARDS: Only one of the selected studies described the effect of steroids on ARDS. Trends toward an increased incidence of ARDS, lower resolution rate and increased mortality were observed.
5. Hyperglycemia: Analysis of data from primary studies was not possible.

2. How precise were the results?

As described above, the confidence interval for the main outcome measure, mortality, is 0.99 -1.29. For the subgroup of patients with septic shock, the confidence interval for the effect of steroid on mortality is 0.91 - 1.26. In both cases the confidence interval includes 1, indicating that there is no significant difference between steroid and placebo treated patients. Confidence intervals for the risk of secondary infection and gastrointestinal bleeding also included one.

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

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

Yes. Out of 9 studies only 1 demonstrates a reduced relative risk of mortality in steroid treated patients. This study however was given a low methodologic quality assessment score. Remaining trials and the pooled results do not justify the use of steroids. A trend towards increased mortality with steroid use is certainly a cause for concern. These studies only include adults. However, there is no reason to believe that pediatric patients will behave differently.

2. Were all clinically important outcomes considered?

Mostly. The authors attempted looking at other variables. They were able to obtain data on rates of secondary infection and gastrointestinal bleeding for analysis. A trend toward increased risk was observed, however confidence intervals were wide and included 1. Risks for hyperglycemia were not assessed since definitions were not explicit across studies and statistical tests could not be applied. The role in steroids in ARDS will need further evaluation.

3. Are the benefits worth the harms and costs? No. This study provides no evidence to support the use of steroids in sepsis or septic shock in adult patients.

References/Related Reviews:

1. Lefering R, Neugebauer EAM: Steroid controversy in sepsis and septic shock: A meta-analysis. Crit Care Med 1995;23:1294-1303. [abstract]

   PedsCCM EB Journal Club Review, by C. Landers

Related Reviews:

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