28% of all autopsies in the study revealed evidence of at least one missed diagnosis and 8% evidence of a Class I (Goldman criteria) error which means that the misdiagnosis could have contributed to or directly caused death (the most common of which were heart attack, pulmonary embolism, artery blockage in the lungs, pneumonia and aspergillosis)
This is clearly a serious issue and one estimate given for the United States is that 40 500 deaths occur in intensive care units every year as a result of misdiagnosis. Given that aspergillosis is one of the most common of these (up to 30% of Class I misdiagnoses - though the paper isn't clear on this), this could account for up to 13 000 deaths a year.
For a rare infection such as aspergillosis that number of deaths would represent a substantial increase in the number of recorded fatal infections per year.
The authors suggest several causes of underdiagnoses but the priniciple factor is suggested to be the lack of autopsies carried out to check clinicians original diagnosis - the rate of autopsies carried out for this purpose has been steadily falling over many years in many countries including the US and EU.