There is an established body of evidence about the positive impact of physical activity on the prevention of non-communicable diseases, and the corresponding reduction in physical capacity caused by disease and poor health. The growing use of commercial and research-grade physical activity monitors may allow for an understanding of physical behaviour that could provide health related insights in clinical populations that are not easily accessible from existing clinical measures. In this paper we investigated the relationship between a free-living measure of physical performance, peak two-minute step count, using data obtained from a thigh worn activity monitor, and selected biomarkers, blood pressure and glycated haemoglobin, used in clinical diagnosis, treatment selection and monitoring. We were able to quantify peak two-minute step count in a population representative sample of 2,850 individuals from the longitudinal BCS70 study. We also observed that clinically relevant reductions in our selected biomarkers were associated with significant decreases in peak two-minute step count, which persisted after controlling for potentially confounding anthropometric and behavioural factors. We conclude that meaningful digital outcomes can be derived from physical behaviour data obtained from body-worn sensors and could play an important role in the development and delivery of disease prevention and treatment.