Sorry, I didn't get notified for this so I've only just seen it. This threw me as well as I've never seen this mentioned in any other study that I've read. It would have been helpful if the article had elaborated a bit more.
Most studies state that anything above 20 dB SL is really rare, and anything above 30 dB SL is the very upper level. Put it this way, I've never seen anyone mention anything above 50 dB SL and it's generally accepted that the vast majority fall somewhere between 4 dB SL and 10 dB SL. This does not always directly correlate to loudness, though. Some can have a lower sensation level but have a higher psychoacoustic measurement in sones.
As I said earlier, the variables make this a complex web to untangle.
There was a study done, which I found interesting, which compared real tinnitus candidates with people who were simulating that they had it (based on previous experience, having had temporary tinnitus after attending a gig or other loud event). The simulators had to have had tinnitus within the last few weeks so that it was fresh in their memory.
What they found was that the simulators all had a sensation level around 10 dB SL higher than those who really had it. The fakers also tended to rate lower frequencies as being more similar to their tinnitus than those who really had it. They found it consistent enough that they could predict who really had tinnitus and who didn't. I read this a while back so it took me some time to track it down again but I'll link to it below.
https://www.frontiersin.org/articles/10.3389/fneur.2017.00038/full
Here's one of the result tables comparing the fakers to the real tinnitus candidates. Obviously, a much bigger sample size would be needed to draw meaningful conclusions from this, but I found it interesting nonetheless.
View attachment 41971