We know that we should "lift with our legs," but what about pushing heavy objects?
Granata and Bennet found that when cocontraction was not incorporated into the model, the forces on the back were indeed much smaller than those typically encountered in lifting tasks. But when cocontraction was factored in, the force on the spine increased by as much as 400 percent, depending on the height of the handle and the amount of force applied. Cocontraction was greatest when participants bent lower to push on the handle, as they would when pushing the heaviest loads. In these cases, stress on the spine matches stress in lifting tasks and may be what leads to the most injuries.
They also note that as the amount of pushing force increases, the vertical component of the pushing force also increases, because the volunteers need the corresponding downward force on their feet to gain traction. This makes a pushing action more like a lifting action, again potentially increasing the chance of injury.
On a sad note, Kevin Granata [wiki], the co-author of this study, was one of the professors killed in the Virginia Tech shooting.