Analysis of HIV-1 clinical trials: statistical magic?