Thromboprophylaxis in medical inpatients reduces pulmonary embolism, increases the risk of major haemorrhage and does not reduce total mortality, producing questionable net clinical benefit