This study came up with a very different conclusion on the use of steroid therapy for children acute respiratory distress syndrome (ARDS), compared to studies performed in adult populations. Should we believe it?
Source: Yehya N, Servae S, Thomas NJ, et al. Corticosteroid exposure in pediatric acute respiratory distress syndrome. Intensive Care Med. 2015;41(9):1658-1666; doi:10.1007/s00134-015-3953-4. See AAP Grand Rounds commentary by Dr. Susan Bratton (subscription required).
PICO Question: Among children with acute lung injury, do corticosteroids affect survival or duration of mechanical ventilation?
Question type: Treatment
Study design: Prospective cohort
In medicine, the dictum that children are not just small adults probably has been driven into the brains of every medical student starting with the first day of pediatric clerkship. It is a completely obvious statement, yet very often in pediatrics we lack definitive studies and often need to look into adult clinical trials to see if lessons learned might apply to a younger patient population.
Evidence is fairly solid that corticosteroid therapy provides benefits in management of adult ARDS, or at least no evidence of harm. High-quality pediatric data simply aren't available, and this study is a stab at correcting that. The investigators looked prospectively at children in a pediatric intensive care unit to try to sort out the role of steroids in pediatric ARDS. Their analysis showed that corticosteroid exposure for more than 24 hours was associated with fewer ventilator-free days and longer duration of ventilation. In short, corticosteroid therapy might be detrimental in pediatric ARDS.
Of course, a prospective cohort study is very different from a double blind, randomized placebo controlled trial. We don't really know how clinicians decided to start steroids in their patients, nor how long to continue treatment. Trying to sort out numerous confounding variables in prospective cohort studies is a big task, and these investigators did as good a job as possible, in my opinion.
First, they performed the usual univariate and multivariate analyses, to correct for confounding variables. Second, they performed a propensity analysis (briefly discussed in a November 2014 post), which is a specific technique in nonrandomized trials, including observational and cohort studies, to look for factors associated with the decision to use a treatment. In this study, the propensity analysis suggested that baseline characteristics of the study children did not differ widely according to steroid use decision. Still, we cannot rely on studies of this design to direct our clinical practice. Rather, as the authors state, this study now becomes an important tool to design a definitive randomized controlled trial.
Knowledge of evidence-based medicine (EBM) principles probably is most useful on the front lines of patient care when we lack high quality studies in a particular patient population. In my practice, hardly a day goes by that I don't need to think about how to apply adult data for my pediatric patients with similar disorders. It requires a very thoughtful approach, but the effort is both intellectually stimulating and helpful to my patients. That's how I came to embrace EBM.