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

A prospective, randomized, and controlled study of fluid management in children with severe head injury: lactated Ringer's solution versus hypertonic saline

Simma B, Burger R, Falk M, Sacher P, Fanconi S.

Crit Care Med 1998; 26: 1265-1270. [abstract]

Reviewed by Stephanie Sudikoff, MD, Yale University

Review posted March 9, 1999

---

I. What is being studied?:

The study objective:

To determine whether children with severe head injury treated with hypertonic saline have lower ICP and higher CPP as compared with children treated with lactated Ringer's solution.

The study design:

An open, randomized, and prospective study.

The patients included:

35 consecutive head injured children, aged < 16 years, and a Glasgow Coma Scale score of < 8 at the scene of the accident.

The patients excluded:

No specific exclusion criteria were listed.

The interventions compared:

Administration of either lactated Ringer's (Group 1) or hypertonic saline (Group 2) as maintenance fluid and as the fluids used for volume challenges when clinically indicated. The intervention fluid was used over a 72 hr period from the time of ICP monitoring device insertion. The amount of fluid was listed as being restricted to 1200 ml/m2, but the hypertonic saline groups fluid management was aimed to achieve serum Na+ goal of 145 to 150 mmol/L, although there is no mention of how the hypertonic fluid was administered differently to achieve this goal.

The outcomes evaluated:

Primary: Correlations between ICP and serum sodium concentration and between CPP and serum sodium concentration in both groups.

Secondary: The number of interventions necessary to keep ICP < 15 mm Hg, additional changes in fluid volumes, and clinical outcome variables, including days of mechanical ventilation, ICU days, survival rate, and number of predefined complications.

Therapeutic interventions to reduce high ICP were given in a predefined sequence of grades 1 - 5 as 1) manual hyperventilation, 2) additional sedation and/or analgesia, 3) mannitol (0.5 g/kg), 4) thiopental bolus (1-5 mg/kg), and 5) thiopental continuous infusion (1-5 mg/kg/hr).

II. Are the results of the study valid?

Primary questions:

1. Was the assignment of patients to treatments randomized?

Yes, by an independent investigator.

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

Was followup complete?

Almost. Three patients were excluded from the final analysis due to GCS scores that did not meet inclusion criteria.

There appears to be a typographical error in the article, as they state that three patients were excluded from analysis for a GCS < 8, where presumably they meant > 8, since a CGS < 8 was the criterion for inclusion in the study. The remaining 32 patients were all apparently followed until death or hospital discharge; there is no mention of patients lost to follow-up.

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

Yes. No patients crossed over to the other group.

Secondary questions:

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

No. The authors state that since the goal in Group 2 was to achieve a serum sodium of 145 to 150 mmol/L, the caregivers could not be blinded. However, since they did not state how the hypertonic fluid was administered differently than lactated Ringer's to achieve this goal, one wonders why the caregivers could not have been blinded. Since the unblinded caregivers were making the decisions on interventions to treat ICP and CPP changes, this lack of blinding detracts significantly from the study's validity.

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

No statistically significant differences existed between the two groups. However, the differences in initial CT result of diffuse head injury (6 in Group 1 vs. 3 in Group 2) and initial ICP (13 in Group 1 vs. 9 in Group 2) stood out.

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

Not completely. Routine care, encompassing positioning, temperature control, sedation, ventilation parameters, and level of oxygenation, was standardized. In addition, all episodes of low CPP due to high ICP were treated with a predefined sequence of therapeutic interventions (grades 1-5) in both groups. A 20 cc/kg fluid bolus, presumably using the fluid the patient was assigned to receive for maintenance, was given for low CPP (< 45 mm Hg in infants and < 50 mm Hg in children) with normal ICP (< 15 mm Hg). Group 1 patients received significantly more fluid in the first 24 hours than Group 2. Whether this was because more fluid was needed to maintain MAP in these patients is not stated.

III. What were the results?

1. How large was the treatment effect?

Difficult to assess accurately. The authors report significant inverse correlation between [Na]/ICP at all times in both groups, and [Na]/CPP only after excluding the first 8 hours of study (why this time period was chosen for analysis was not stated, and appears to clearly be a post hoc assessment) and only in Group 2. However, their statistical analysis is inappropriate. When assessing measurements of two variables in the same patient, but at many different times, they pooled all the data together as if it were one sample, therefore analyzing 260-290 data points, rather than 15-17 data points. When calculating correlation coefficients, the greater the number of data points, the smaller the r value can be and still achieve statistical significance. Hence, the magnitude of the treatment effect is likely smaller than the numbers presented.

The authors state that Group 1 patients required more interventions to keep ICP < 15 mm Hg than Group 2 patients. In a very creative pooling of numbers of interventions for Grades 1-4 and total mg/kg of thiopental administered for Grade 5, they arrive at a unitless value of 1047 for Group 1 and 662 for Group 2, statistically different by the Mann-Whitney U test. This is an outstanding example of how numbers and statistics can be manipulated inappropriately to whatever outcome an author desires.

There is no difference between groups for grades 1-3 interventions. Group 1 received, in total, 18 boluses of thiopental, compared to 2 in Group 2. And Group 1 received 420 mg/kg of thiopental by infusion compared to 15 mg/kg in Group 2. These appear to be significant differences, although a single patient in Group 1 who "required" many boluses of thiopental and then a prolonged thiopental infusion could easily have accounted for these differences. The authors should have stated the numbers of patients receiving these interventions, not simply a pooled number from each group.

There was no statistically significant difference in overall survival rate or hospital length of stay between the two groups. The only statistical difference quoted with a p value of < .05 was length of ICU stay, which was significantly lower in the hypertonic saline group (11.6 +/- 6.1 vs. 8.0 +/- 2.4 days). There does not appear to be any significant difference in complication rate between the two groups, although another typographical error confuses the issue. The p value for the incidence of ARDS is quoted as p=0.1 in the summary table and as p=0.01 in the text (4 cases in Group 1 vs. no cases in Group 2).

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

It is inappropriate to comment on the precision of their primary outcome as their statistical analysis was flawed. Confidence intervals were not given for the numbers of interventions required between groups and cannot be calculated from the data as presented.

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

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

The patients in this study appear to be similar to most severely head injured pediatric patients. If one accepts their conclusions, then the results could be applicable for our PICU patients. This does not guide us in the care of less severely injured patients.

2. Were all clinically important outcomes considered?

No. There was no assessment of neurologic/functional outcome included in this study. Even if one accepts their results at face value, one can still not draw any conclusions about whether hypertonic saline improves the outcome of such patients.

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

Not likely. Although the risk of potential harm seems small, the clinically important benefits are even smaller. No improvement in overall survival was shown and no assessment of functional outcome was presented.

References

1. Bland M. An Introduction to Medical Statistics. Oxford University Press 1995. Pp. 192-99.

Comments

Back to the EB Journal Club Index