One of the procedures that is heavily being researched these days is fusion of lumbar vertebrae. The theory revolves around the idea that a painful segment is dysfunctional due to "decreased disc space" or hypermobility. The procedure involves decompression and the fusion of the segments, so that pain and compression can, theoretically, no longer come from the segment.
One of the obvious issues with this procedure is that it almost guarantees further degeneration and pain at the above and below segments (if you are looking at it pathoanatomically). Why? Because, in a given motion, like lumbar flexion, the force and mobility is supposed to be relatively evenly dispersed throughout the entire region. When one of the segments becomes fused, greater motion must occur at each joint, but especially at the ones directly above and below. This hypermobility leads to increased degeneration and then hypomobility and possibly pain (again assuming pathoanatomical perspective).
The other major concern is that the surgery typically doesn't have that favorable results. Previous research had said that lumbar fusion had only a 25% success rate. With the risks of any invasive procedure, I know I wouldn't want to pursue it. More recently, a study revealed that fusion is no better than non-operative care for chronic low back pain. Again, with the risks, why go through with surgery?
Personally, I recommend avoiding lumbar surgery in almost all cases. The risks and low success of surgery typically aren't with it, in my opinion. However, there are cases where I do recommend fusion of the spine. For example, worsening neurological signs that do not improve with conservative management, fracture, and other similar findings. There are cases where severe future injury can be avoided with surgery, like cauda equina syndrome. Lack of progress may indicate the need to switch treatment approaches, but progressively worsening symptoms may indicate the need for a surgical consultation.