Mortality rates are low because hospitals are working. Intensive care beds in Italy are projected to be full in less than 10 days. You’ll see the rate grow by then.

Areas that were locked down exactly 15 days ago, now seem to have seen a significant decrease in infection rates in the last few days. So there is hope that the system will fire on all cylinders and hold the line just long enough to weather the storm. Also, Southern Italy is significantly drier and warmer than Northern Italy, which might (might) help.

One of the issues working against this scenario is the fact that the worst-affected regions are also the ones with the best health services. The other regions might struggle to respond as efficiently as the "pioneers", although they will benefit from the accumulated experience.

The range varies a lot depending on the quality of the treatment. The early initial estimates were based on data from Wuhan, where hospitals were overwhelmed early. The later, lower numbers were from areas where all new cases had hospital beds and ICUs available if they needed them.

Problem is, given the current numbers, the projected total number of people who will require hospitalization, and of those that will require ICU, significantly exceed availability. So if it spreads too fast, and all these people get sick at the same time, we won't have enough beds to go around - and that will cause the mortality rate to spike again.