Background:
Rates of smokeless tobacco(ST) use in India are very high and associated with grave health consequences. Research on effective ST cessation approaches is a priority. In a randomized double blind placebo controlled trial of varenicline for ST dependence in AIIMS, India, biochemically confirmed end-of-treatment(EOT) abstinence was greater for varenicline versus placebo(25.2% vs. 19.5%), but this was not statistically different(AOR = 1.6, 95% CI = 0.84-3.1, p = .15). Here, we studied the correlates of successful abstinence in this trial as such an evidence-base is crucial to optimize treatment strategies.

Methods:
237 ST users who were randomized to receive either varenicline(12 weeks, 1 mg, twice per day) or placebo were included in this analysis. Socio-demographics, ST use parameters (age of initiation, duration, chews per day, and FTND-ST scores), abstinence related parameters, baseline urinary cotinine levels, and treatment adherence were analyzed. Multivariate analysis, using biochemically confirmed point prevalence abstinence as the dependent variable, was done to predict the odds of quitting at EOT.

Results:
Participants were predominantly males(97%) and mean age was 34.2(SD± 9)yrs. Using logistic regression on the whole group, baseline urinary cotinine and number of 24 hour quit attempts in the past significantly(p < 0.05) predicted cessation at EOT. The odds of abstinence increased with unit decrease in urinary cotinine levels(Exp(B) = 0.86) and unit increase in number of 24 hour quit attempts (Exp(B) = 0.66). In the varenicline arm average chews per day(Exp(B) = 0.91) and baseline depression scores using HAM-D also significantly predicted successful quitting. Greater adherence increased EOT cessation rates for varenicline (39% vs. 18%, p = .003) but not for placebo (28% vs. 14%, p = .06).

Conclusions:
This has implications for optimizing treatment planning at a program level, especially in resource poor countries. Strategies to enhance treatment adherence and address depression would be important.