At several schools such as UCSF (where I'm attending), MSTP students can do a longitudinal clinical clerkship, for which they'll receive credit, during their Ph.D. years. This is usually arranged by the student in conjunction with a preceptor of interest. One setup would be to attend a weekly clinic. This enables the student to stay in contact with clinical medicine, while requiring only a minimal time committment.
Most MD/PhD students do laboratory benchwork during their thesis. Some students enter "dry labs." Few actually do clinical research. The idea of MD/PhD programs is to generate physician-scientists who will conduct basic biomedical research and help translate their discoveries into potential targets for therapy.
Of note, clinical research (i.e. measuring the efficacy of a drug versus another drug or placebo), often takes place without a precise knowledge of the underlying mechanisms. There are many drugs available for which the mechanism of action is not really known. In clinical research, elucidating the exact mechanism is not the primary goal--it is to determine efficacy, dosage, side effects, and other information.
In order to elucidate the mechanisms of drug action and hence pathways affected by disease, basic research is needed. A clinical study may raise interesting questions that can then be investigated at a more basic level, and thus we can learn more about pathogenic mechanisms and identify additional targets for treatment.
Just my 2 cents... well, maybe a nickel...