Is a random urinary albumin concentration a useful screening test in insulin-treated diabetic patients?