The results of this analysis illustrate three points. First, that for predictions of AIDS cases four to five years into the future, the back projection method is largely insensitive to the assumption one makes for the incubation period distribution. The two extreme distributions considered represent the fast and slow extremes of incubation period distribution usually proposed; distributions that lie between these two give predictions within the range of predictions that the two generate. The estimated number of new HIV infections, however, is highly sensitive to the assumed incubation period distribution; prediction of AIDS cases in the long term will be similarly sensitive.
Fractional extreme distributions
