A remark on Getzler's semi-classical approximation