The biggest problem with that statistic is the number of infected people, which determines the other variables you mention. Accuracy here requires for everyone to be tested, because many of those infected only show minor symptoms.
100% accuracy is neither feasible or required. Only a statistically valid random testing is necessary. At the present time in most of this country there are only enough tests available for those who are showing a specific symptomology and/or are in high risk groups. So any statistics are heavily biased in favor of being infected.
If we knew what it was we were doing, it wouldn't be called research, would it?
