Relationship between Respiratory Syncytial Virus Load and Illness Severity in Children

To the Editor—We read with interest the paper by Wright et al. [1], which provides several useful insights into the nature of respiratory syncytial virus (RSV) infection and the nature of the humoral immune response to it. However, the data presented fail to support one of the conclusions—that RSV load is not related to disease severity. Because the original study was not intended to address this particular question, several aspects of the study design limit its ability to exclude the existence of such a relationship. The first limiting factor is that the study population was a heterogeneous group of subjects, the majority of whom had underlying conditions that would, theoretically, disrupt the putative relationship between RSV load and disease severity. These conditions included congenital heart disease, chronic lung disease, and premature birth. Because the extent and severity of these underlying conditions is variable, disease severity in these patients is more likely to be a function of their underlying disease than of their RSV load. A second limiting factor is the exclusion of subjects who were ventilator dependent at enrollment. This exclusion effectively narrowed the range of RSV disease severity that could be evaluated in the study. Narrowing this disease severity range limits the ability to determine a relationship between disease severity and RSV load. A third limiting factor involves the calculation of disease severity, which was based mainly on respiratory illness scores determined for up to 13 days after the initial enrollment assessment. Unfortunately, approximately half the subjects received RSV immune globulin infusions after this initial assessment but were still included in these disease severity calculations. RSV immunoglobulin therapy could have affected disease severity, thus confounding any correlation between severity of disease and RSV load. Finally, the authors do not comment on whether their sample size is adequate to exclude the possibility of a type 2 error. Presumably, this is because the study was originally designed (and powered) to address a different clinical question. Without a power analysis, the results of a study reporting “a lack of correlation” are difficult to interpret. We recently reported the results of a study that was designed specifically to examine the relationship between RSV load and disease severity [2]. We excluded infants with underlying conditions or exposures that might disrupt the relationship between disease severity and RSV load (e.g., chronic lung disease, congenital heart disease, immunodeficiency, and prior treatment with systemic glucocorticoids, ribavirin, or RSV‐specific antibody preparations). We also designed the study to compare RSV loads between groups of hospitalized patients who had the greatest degree of difference in their RSV disease severity. We, therefore, compared infants who had RSV‐induced respiratory failure with hospitalized infants who did not have respiratory failure. Furthermore, we used linear modeling to control for potential effect modifiers. After controlling for gestational age and ethnicity, we found that infants with severe disease had ∼10‐fold (1.15 log pfu/mL) greater RSV loads than did their counterparts with nonsevere disease (⁠P=.02⁠). There are also other studies reporting an association between RSV load and disease severity. An article by Hall et al. [3] found a significant association between RSV load and radiographic consolidation and an association between RSV load and hypoxia that fell just short of statistical significance. Of note, this and another study from the 1970s [4] were limited by some of the same problems affecting the study by Wright et al. [1], including lack of statistical power and failure to control for potential effect modifiers. In addition, we subsequently reported the results of a study of 80 RSV‐infected infants [5] in which linear regression showed that both host factors (e.g., premature birth, birth weight, and white ethnicity) and initial nasal RSV load were significantly associated with longer duration of hospitalization. In summary, the article by Wright et al. [1] provides information that is useful in unraveling the components of the RSV disease process, but several aspects of the study design limit its ability to examine the relationship between RSV load and disease severity. The nature of this relationship is critical in determining the potential utility of antiviral treatment strategies for this disease. Rather than conclude that no association exists, researchers who study this disease should be cautious and evaluate the results from studies specifically designed to determine this association. We believe that RSV disease severity is likely to be determined by an interplay between host factors (e.g., underlying conditions and nature and magnitude of the inflammatory response) and virus factors, including RSV load. More study clearly is needed in this important area.