An interesting paper by Charles Haas suggests that calculating the quarantine period for Ebola at 21 days may result in a 0.2% up to 12% chance of release of the virus.
The precise origin of this assessment (21 days) is unclear, however it is possibly based on the study of the either the 1976 Zaire outbreak11 or 2000 Uganda outbreak12 both of which reported (without detailed analysis) a maximum observed incubation time of 21 days.
The WHO Response Team17 has just published an incubation time distribution based on the first 9 months of the West Africa outbreak (total of 4010 confirmed and probable cases with usable data). They reported a mean incubation period of 11.4 days with an upper 95th percentile of 21 days — and they were able to fit the data to a gamma distribution.
For greater security, Haas argues, a quarantine period of 31 days is preferable. However, all quarantines run up against a trade-off:
So, we need to look at the cost of enforcing quarantine (Police wages plus cost of food and medical supplies plus loss of earnings for quarantined person) as opposed to the costs of treating contacts who may get infected, and the costs of lives that are lost. Time to call in the accountants.