High-Frequency Oscillation and Chronic Lung Disease in Very Low Birth Weight Infants

Abstract

To the Editor. The study by Rimensberger et al1 suggests that early institution of high-frequency oscillation (HFO) for the treatment of respiratory distress syndrome improves the pulmonary outcome in ventilated very low birth weight infants. Unfortunately, there are flaws in study design that limit the validity of this conclusion. First, the 2 study cohorts were not concurrent. Most of the infants in the experimental cohort treated with HFO were born in 1997 and 1998, whereas those in the historical control cohort treated with conventional mechanical ventilation were born in 1994 and 1995. Subtle differences in clinical management might have been unavoidable in the 2-year period that separates these cohorts and contributed to the difference in the pulmonary outcome. Second, the statistical basis for selecting the specific number of infants in each cohort is not provided. Without appropriate power analysis, the statistical validity of comparisons between cohorts is difficult to confirm. Third, the retrospective nature of the study raises questions about uncontrolled data collection and problems with data retrieval, both inherent problems of a retrospective study design. Fourth, what is most striking is that there is no discussion of the common complications, such as symptomatic patent ductus arteriosus and airway infection, both of which are strong contributors to the development of chronic lung disease. And fifth, there is no discussion of the nutritional management of the infants. Nutrition plays a big role in the process of healing from lung injury. Thus, although the zero incidence of chronic lung disease in very low birth weight infants ventilated with HFO appears to be a remarkable achievement indeed, the enthusiasm for accepting this conclusion is dampened by the flawed study design. The subtle message that a single ventilation strategy can eliminate chronic lung disease, a disease of diverse causes, is misleading. What is …