After new confirmed cases, deaths are probably the second most important metric. After all, if no one died from COVID-19, this whole pandemic thing would be a lot less scary.
A seven-day moving average is used to smooth the trend, as the data for new deaths has some seasonality depending on the day of the week (like new confirmed cases). I also include new confirmed cases on this chart (with a one week lag as deaths obviously come a bit after initial diagnosis) to show the relationship and subsequent divergence:
Note: I do exclude one data point from New Jersey that represents about 1,800 deaths as they were backdated deaths applied to one single day. So, it causes some noise to the series and the focus here is the general trend.