**7. Artificial urinary sphincter**

The AUS in the female population differs from the suburethral sling in that it does not provide a backboard or urethral support, but rather it attempts to mimic the sphincter mechanism of the urethra with circumferential compression. (Light & Scott 1985) The initial use of the AUS in women with incontinence was described in 1985 (Light & Scott 1985) and has since been well documented in the literature. (Vayleux et al. 2011)

Costa and colleagues evaluated the efficacy of the AMS 800 AUS in women with Type III incontinence and a negative Marshall test. (Costa et al. 2001) They described a modified surgical procedure through an abdominal approach. Of the 190 patients with working devices, continence was achieved in 88.7% and 81.8% of those with non-neurogenic and neurogenic bladders, respectively at a mean follow-up of 3.9 years. 51 patients had perioperative complications and a high percentage of patients had not undergone prior surgical therapy. Thomas and associates reported 12 year follow-up in 68 patients who underwent an AUS. (Thomas et al. 2002) Despite an 81% continence rate, 46% required removal or replacement for erosion or infection. They concluded by recommending an AUS in patients with SUI after failure of one anti-stress incontinence operation and rather than as a last resort. More recent series reports demonstrate the safety, efficacy and complication rates associated with female AUS implantation. (Chung et al. 2010; Vayleux et al. 2011) Importantly, while there is a high proportion of patients that continue to use the AUS after implantation, about 50% of them have required revision or replacement, usually within the first several years of implantation. Continence rate in these two studies indicates 65-70% for no pads and 73.5-83% for 0 to 1 pad per day. Satisfaction rates were high with a majority of patients reporting they would undergo AUS placement again if necessary. Additional modifications such as insertion of an AUS laparoscopically (Roupret et al. 2010) or with a large cuff (Revaux et al. 2010) have been proposed, however long-term data is lacking at this time.
