Can we use routinely collected primary care medical records to quantify and explain variations in amputation rates across the UK?