Essentially, there was a panic overreaction by the HK medical establishment that meant patients were given huge doses of steroids. The HK "experts" are, of course, not proud of this, so it doesn't get talked about now. But it did slip out at the time and was reported in the South China Morning Post on 9 May 2003.
Sure, HK tried marinating their critical patients with steroids, and that didn't work. No argument.
But Singapore and Toronto didn't marinate, and their mortality was much closer to HK than mainland China.
China was very much the statistical outlier in the SARS dataset, not HK.
Its hard to work out what is happening with all the changing data.
With a dramatic increase in the number of cases in Hubei, the Hubei mortality has decreased from to 3% (in Hubei) and 4.5% (Wuhan city) to about 2.5%
Meanwhile, the mortality outside Hubei is slowly drifting up (from 0.16% to o.38% to 0.47% to 0.52%).
Overall, it may be that these figures are going to converge at about 2% mortality (compared to 0.1% for a bad 'flu season, and perhaps 2.5% for the 1918 "Spanish 'flu").