Abstract
Clinical nerve injury has been reported in 0.6–4.8% of shoulder arthroplasties. Classical teaching is that 70–85% of injuries recover. Despite recovery of motor function, overall shoulder function may be negatively affected and residual pain is common. Complex regional pain syndromes may develop and become permanent. Consequently, methods to limit nerve injury have been investigated.
In the early 2000's I became concerned about the incidence of nerve injuries in my arthroplasty practice. I became intrigued with the idea of peripheral nerve monitoring as a method to alert the surgeon intra-operatively about impending nerve insults so that evasive measures could be taken to prevent any clinically significant nerve injuries. The results of our first 30 consecutive patients were published in JSES in 2007. Seventeen patients (56.7%) had 30 episodes of nerve dysfunction (i.e. nerve alerts) during surgery. Twenty-three of thirty alerts (76.7%) returned to normal after repositioning the arm to a neutral position. Post-operative EMG was positive in 4 of 7 (57.1%) patients who did not have a return to normal motor latency intra-operatively and in 1 of 10 (10%) patients whose intra-operative nerve function did return to normal. None had clinical nerve injuries. This early experience indicated that nerve injury was potentially more common than previously thought but intra-operative nerve monitoring seemed to have a relatively high false positive rate.
Our group subsequently studied 440 shoulder arthroplasty cases. The protocol used to identify a nerve alert was made more restrictive than the first study as an attempt to decrease the false positive rate. In this larger group, nerve alerts occurred in 185 cases (42.0%), and 37 (8.4%) cases did not have signals return to above the alert threshold at closure. There were no permanent post-operative nerve injuries and 5 transient nerve injuries (1.1%). Cases in which MEP amplitudes remained below alert threshold were significantly more likely to have a post-operative nerve injury (p = 0.03). There were no false negatives, (i.e. a post-operative nerve injury occurred while MEPs were normal at closure), making sensitivity 100%. There were 32 false positives, leading to a specificity of 92.6%, a positive predictive value (PPV) of 13.5%, a negative predictive value (NPV) of 100%, and an accuracy (ACC) of 92.3%. In my opinion, the high false positive rate and the low PPV make the technique difficult to justify for routine clinical use.
