Some consequences of using the Horsfall-Barratt scale for estimating disease severity compared to nearest percent estimation