Generalizing results of randomized trials to clinical practice: reliability and cautions