Re: Flatback or something else?
Harringtons are generally much easier to remove than more modern hardware is, because as you say they are only attached at the ends, but removal very much depends on the individual patient. In some cases removal is very easy, but in others it can be very difficult due to the rod becoming encased in bone (again, as you say). Your surgeon will weigh up the risks of removing the rod vs the benefits of having it out.
I had my original Harrington (placed in 1986, when I was 10) removed and replaced with a longer one in 1994, to extend my fusion to include my lumbar curve. After just eight years that original Harrington was very deeply embedded and they had to break the bone away and actually cut the rod into three pieces in order to get it out. That was a very difficult removal, but my surgeon felt it was necessary.
A few years back I needed to have my fusion extended down one level (due to disc degeneration at the bottom of the fusion) and I requested that my surgeon remove my Harrington whilst he was in there, if he could - simply because I'd prefer not to have it. However, when he took a look at it during the surgery he saw that this rod was also very deeply encased in bone; he didn't attempt to remove it because there was no real clinical need to do so, and the risks outweighed the benefits. Fair enough!
37 years old, diagnosed with infantile idiopathic scoliosis at 6 months old with curves of 62(T) and 40(L) degrees. Casting and Milwaukee braces until surgery at 10 - ant release/pos fusion T1-T12, halo traction. Post op cast and then TLSO. Further surgery at 18 (ant release/pos fusion extended to L3 to include lumbar curve, costoplasty) and 25 (another costoplasty). Fusion extended to L4 at 33 (XLIF with 4 pedicle screws and two short rods). Pre-op curves: 76(T) and 70(L). Post-op curves: 45(T) and 35(L). Diagnosed with Ehlers-Danlos Syndrome aged 34; scoliosis almost certainly due to this rather than being idiopathic.