It makes sense when you think that this removes the individual variation per hospital/doctor writing a death certificate, you don't rely on professional opinion, you count all of them as it is easier to later clean this statistic with models based on historical data than the opposite (to figure out the missing data for deaths from CoViD-19 after the fact).

You need to be aware that Sweden has a long history of keeping demographical statistics [0] so there is a method to the apparent madness of counting.

[0] https://www.scb.se/en/About-us/main-activity/history-of-stat...