Stata se 11

Data were analysed using Stata SE 11.1 (StataCorp, Texas) during 2013 and weighted to the population on sex, age, place of residence [66 ] and educational attainment [67 ]. Chi-square analysis assessed whether demographic characteristics differed between states at each survey. Logistic regression analysis tested interactions by state (WA versus Victoria) and subsequent surveys (baseline versus W1; baseline versus W2) controlling for socio-economic status (SES) and BMI. Interactions were tested for the full sample and the sub-sample of overweight/obese respondents. Overweight/obese and healthy weight/underweight categories were combined and are referred to as ‘overweight’ and ‘not overweight’, respectively. Logistic regression was undertaken to determine whether campaign recall differed according to sex, age, BMI, parental status, place of residence and SES when controlling for all other factors and time spent viewing commercial television. Differences in ratings of perceived effectiveness by ad type (principal versus supplementary) were tested using logistic regression controlling for individual-level clustering and the above demographics.

The data analysis and statistical software, Stata/SE 11.1 (College Station, Texas, USA), was used to perform all analysis. Statistical significance was defined as a two-tailed p value of <0.05. The Student's t-test was used to analyse continuous variables. These variables were reported as a mean±SD if normally distributed or as a median and range if skewed. The χ² test was used to analyse categorical variables. These variables were reported as a proportion (%) of the overall cohort.
The Kaplan–Meier method was used to approximate the overall survival, and the significance of survival differences between the sexes was determined using the log-rank test. Differences in survival among men and women within different diagnosis scenarios and aetiologies, including hepatitis B, hepatitis C, and non-viral diseases were also evaluated using the Kaplan-Meier method and log-rank test. A stepwise multivariate Cox proportional regression was used to estimate HRs with a 95% CI, which related patient survival to baseline predictors that were found to have a significant effect on survival during univariate analysis.

Statistical analyses were performed using the SPSS (SPSS, Inc., Chicago, IL, USA) and STATA (STATA Corp., College Station, TX, USA) software packages (SPSS for Windows, version 18.0.1 and STATA/SE 11.1). Differences in the means of continuous variables were analysed using non-parametric tests (Mann–Whitney or Kruskal–Wallis) for two and multiple independent samples, respectively. Univariate survival analysis for the outcome measure (disease-specific survival [DSS] or disease-free survival [DFS]) was based on the Kaplan–Meier method, with log-rank (Mantel–Cox) comparison test. DSS and DFS were calculated, based on the time from diagnosis to death (due to disease), and on the time from diagnosis to the appearance of metastatic disease, respectively. In all tests, the values p < 0.05 were regarded statistically significant.