Generalizability of the results of randomized trials.