If that 1 out of 50 was picked by random, then that might be correct, but the article makes it seem as though that isn't the case.

Since most of these autopsies are conducted when foul play is suspected or when the cause of death is unknown, that doesn't help doctors improve in the cases where they thought they knew the cause of death but were wrong.

1. Many of them are natural causes, third party (murder, suicide, accidents), and so on that are not necessarily caused by diseases or unexpected pathogens of which the doctors want to learn about. Remember the early autopsy reasons in the article: "In the United States, we usually don’t autopsy people unless the cause of death is mysterious or foul play is suspected." 2. What about races, gender, blood types, ages, and so on? The progression of the disease is different in each combination, so when factoring these factors in, 1 out of 50 really isn't that big of a number.