Actually if you look at their full sample data you'll see that women physicians had higher survival rates for both make and female patients. It's just that only

Actually if you look at their full sample data you'll see that women physicians had higher survival rates for both make and female patients. It's just that only female patients were statistically different from zero. And note that just because female-female was statistically significant and female-male was not, that doesn't by itself imply that female-female was statically distinguishable from female-male. (Though it may have been, I don't understand all the controls.)