That was in 2006. They changed their policy and stopped multiplying small dose rates by large populations to get large numbers after that because the linear-no-threshold assumption has not been shown to be applicable at very low doses. Their later reports do not give any large numbers like this.
But that's just even more arbitrary, right? It just means nobody knows how many deaths it caused, and there is an uncertainty of an order of magnitude based on how the effects of very low doses.