Irrespective of this limitation, the critical importance of evaluating the intrinsic progressiveness of disease in individual patients cannot be overemphasized. Thresholds for the severity of disease do not, in isolation, meet this need. Biomarker data should, in principle, identify patients at higher risk of progression, irrespective of disease severity, but to date, despite promising findings in IPF and systemic sclerosis, no current biomarker is routinely fit for clinical purpose. It appears intuitively likely, if it is accepted that changes in pathogenesis occur as disease progresses, that accurate prognostication will require the integration of severity and biomarker data. For example, in the most compelling IPF biomarker study to date, four of five serum proteins that were found to predict the course of disease were discarded when severity information was incorporated as an independent prognostic determinant in a final model.11 Short-term changes in disease severity also add substantially to prognostic evaluation. In a pharmaceutical cohort, outcome in IPF was linked equally to the baseline severity of disease and events in the next year, including decline in FVC and hospital admissions.12 As clinicians who look after patients with IPF and other chronic ILDs know, the likely prognosis, in IPF and other chronic ILOs, can change radically during early follow-up, especially when progression, despite treatment, is apparent.