Limitations of indicator-dilution methods in experimental shock.