The backwater effect generated by bridges can significantly contribute to increase the risk of flooding. In this work we compare two different methods to include the effect of bridges in 2D shallow water models. The first method is based on empirical discharge equations that are implemented as internal conditions. The second method is the recently proposed 2D extension of the Two-component Pressure Approach, which accounts for the vertical confinement of the flow. Both approaches are tested and compared using a new set of experimental data obtained in 32 laboratory tests, including 4 different bridge geometries under different flow conditions. The results show that both methods can reproduce the observed bridge afflux for a wide range of flow conditions, but the Two-component Pressure Approach is less dependent on model calibration. On the other hand, both methods fail to correctly reproduce the 2D water depth patterns observed around the bridge.