Specifically, they hypothesized that surfactant protein genetic variation may influence outcomes in influenza. They evaluated a chosen set of SNPs, searching plausible candidates related to surfactant proteins for an association with severe disease within rigorously controlled conditions. Finally, they assessed the significance of univariate relationships in a more diverse and, therefore, more generalizable group of subjects by means of multivariate analysis to find the relative importance of their selected SNP compared with other independent, potentially causal factors for severe disease. Hypothesis-driven exploration, carefully corrected for multiple comparisons, produced a borderline P value, above the arbitrary threshold of .05. But the hypothesis-driven nature of the study allowed a thoughtful interpretation that identified the SNP as worthy of additional investigation. Multivariate analysis within a larger population of patients replicated the real-world mix of young and old patients with chronic pulmonary, neurologic, cardiac, metabolic, and endocrine disorders and demonstrated true significance.