In support of this hypothesis, perceived drug together efficacy did not differ by treatment assignment and neither did dropout rates. In the literature, dropout rates in placebo arms of randomized controlled clinical trials are consistently reported to be higher than those observed in active treatment arms when the treatment is efficacious, suggesting that lack of efficacy plays a role in influencing dropout (Gonzales et al., 2006; Jorenby et al., 2006; Jorenby et al., 1999; Oncken et al., 2006). In a pilot study of gabapentin for smoking cessation, 35% (N = 6) of subjects receiving gabapentin failed to complete the 6-week study and 12% (N = 2) dropped out due to adverse events (White et al., 2005). In another pilot study of cigarette smokers (N = 50) receiving gabapentin for 8 weeks, 22% (N = 11) of subjects dropped out prior to the end of the medication phase (Sood et al.

, 2007). Dropout rates in both of these studies were lower than in the current study; however, both studies were open-label and had a shorter duration of therapy. The finding that a difference in dropout was not observed between the active and control arms of this study is consistent with the hypothesis that gabapentin is no more efficacious than placebo for smoking cessation. However, we cannot rule out the possibility that gabapentin may be deleterious to smoking cessation. Several drugs have been explored for the treatment of tobacco dependence based upon theoretical constructs. Methylphenidate has been proposed as a possible treatment option because of similar neurobiological effects as nicotine and early suggestive data (Robinson et al.

, 1995). However, methylphenidate was subsequently observed to enhance the abuse-related behavioral effects of nicotine in animal models (Wooters, Neugebauer, Rush, & Bardo, 2008), and human laboratory studies observed that methylphenidate increased tobacco smoking (Rush et al., 2005). One interpretation of this data is that methylphenidate attenuates the reinforcing effect of nicotine leading to increased nicotine intake to overcome decreased reinforcement. Given the design of the current trial, we could not evaluate this hypothesis. The small sample size coupled with the high dropout rate in this study limits our ability to draw any definitive conclusions about gabapentin for the treatment of tobacco dependence.

Our original intent was to enroll a total of 120 subjects, which was based on a hypothesized end-of-treatment abstinence rate of 15% for placebo and determined to provide statistical power (one-tailed, �� = .05) of 82% to detect an abstinence rate of 40% or greater for gabapentin compared with placebo. Due to the high dropout, a review of the primary endpoint was performed after 80 subjects AV-951 were enrolled. For the hypothesized abstinence rates, this effective sample size provided statistical power of 90% for a one-tailed �� = .