As the data moves to a flatter curve, I have changed the linear regression to loess regression. Each day (apart from one outlier) the reading is less than the day before – black data points. The regression line has an uncertainty which just touches the growth line (grey area).
We have now had a calendar month of a slow decline in daily hospital mortality. The problem with Covid19 mortality rates is the lag time. It takes many weeks for susceptible individuals to become seriously unwell after contracting the infection.