Use of the proportion of patients dying on an End of Life Pathway as a quality marker: considerations for interpretation