Survival analysis of time to quit attempt using data from the international tobacco control survey in Canada...

This study uses longitudinal data from the International Tobacco Control (ITC) Survey. The data were collected from 2002 to 2014. There were 9 waves including 8 wave-to-wave transition periods. The data can be used to explore predictors of time until quit attempt among Canadian adults who smoke. To assess the strength of certain potential predictors of the time of making the first "most recent quit attempt" among people who smoke cigarettes in the ITC Canada Survey participating in at least two waves between Wave 1 and Wave 9, and thereby to illustrate the possibility of applying a survival analysis to these longitudinal survey data using complex survey software in the R language. A longitudinal survey was conducted with a representative sample of adult people who smoke cigarettes from the provinces of Canada, with Wave 1 occurring in 2002 and Wave 9 in 2014. An analysis was carried out to determine the factors that were significantly associated with time of making the first "most recent quit attempt" after joining the sample. We used both the Cox proportional hazard model and non-parametric methods for survival analysis, with and without taking account of the complex survey design. Variables expected to be predictors of the time until making a quit attempt included sex and age group at recruitment, both fixed over time, and previous wave indicators/values of advice on quitting from a doctor or other healthcare provider, intention to quit, and cigarettes used per day. In the survival analysis with fixed covariates and assuming simple random sampling, sex and age group were not significant. When the complex design was accounted for, the results were similar. When the time-varying covariates were added, those aged 55 + were more likely to have made a first quit attempt, although age group was no longer signficant. The square root of previous wave cigarettes per day and previous wave intention to quit were found to affect the time until making the first "most recent quit attempt" significantly, the first associated with a later time and the second with an earlier time. Not receiving advice from a doctor or healthcare provider was associated with a later time but was significant only when the complex design was not accounted for. An international longitudinal study provides an opportunity to study smoking cessation among people who smoke cigarettes and the relationship of time of cessation attempt to predictors such as age, sex, advice from a doctor, intention to quit and cigarettes used per day. Previous wave values of cigarettes used per day and intention to quit were strongly associated with time until making a quit attempt.