Variations in institutional review board decisions for HIV quality of care studies: a potential source of study bias.