This study is a part of the larger Tobacco Control Policy (TCP) India Survey, a prospective cohort study of adult tobacco users (aged 15 +) and non-users from 4 Indian states: Bihar, Madhya Pradesh (MP), Maharashtra, and West Bengal (WB). Within each state, one major city represented an urban area and a surrounding area within 50 km outside the city represented a rural area. At Wave 1, the survey employed a stratified multistage cluster sampling design and was conducted between August 2010 and October 2011. Wave 2 was conducted in October 2011 to September 2013. The survey protocol and questionnaires were first developed in English followed by translation into the dominant languages of each state (Hindi in Bihar and MP, Marathi in Maharashtra, and Bengali in WB). At the end, respondents were debriefed, remunerated, and thanked for their time [20, 21]. Additional details on the construction of survey weights, household enumeration, selection criteria and response rates are available in TCP India Technical Reports [22, 23].
Data for this study were drawn from the TCP India Survey comprising 8940 participants and only baseline smoked tobacco users who participated at both waves of data collection were selected for analysis. Of the 8940 participants sampled at Wave 1, 1255 were smoked tobacco users. Of those 1255 smokers, 1112 were followed up at Wave 2 and reported their tobacco use status. The analytical sample of smokers had an 88.6% retention rate at Wave 2.
Socio-demographic variables measured were age, sex, highest level of educational attainment, monthly household income and urban residence. Education was categorized into low, moderate, and high. Low education included illiterate, primary or middle school education; moderate included secondary school or Industrial Training Institute courses; and high included those who completed college and higher education. Similarly, income level was divided into low, moderate, and high. Low-income category included those earning less than 5000 INR per month; moderate income had those earning between 5000 and 15,000 INR per month and high income included those earning more than 15,000 INR.
Smoking tobacco user
A smoker was defined as anyone who said yes to either of the following questions: “Do you currently smoke cigarettes at least once a month?” or “Do you currently smoke bidis at least once a month?” (Yes/No/Don’t Know).
Tobacco use variables
The tobacco use variables were the use frequency, intention to quit smoking, and quit status.
The cigarette and bidi smoking frequency were measured by two different questions asking: “On average, how often do you smoke cigarettes?” and “On average, how often do you smoke bidis?” The response categories – “Less than once a week/Once a week/Twice a week/3–5 times a week/Every day or almost every day More than once a day” – were combined and reported as daily smoker (Every day or almost every day/More than once a day), less than daily smoker (Once a week/Twice a week/3–5 times a week), and less than weekly smoker (Less than once a week) for cigarette and bidi users separately.
Intention to quit
Intention to quit was measured by asking “Are you planning to quit smoking…” and the response categories were: “Within the next month/Within the next 6 months/Sometime in the future, beyond 6 months/Not planning to quit/Refused/Don’t know.” The responses were recoded as a dichotomous variable with any plans to quit as 1 or Yes and “Not planning to quit/Refused/Don’t know” as 0 or No.
At Wave 2, all smokers from Wave 1 were asked whether they were still smoking. Those who indicated that they had completely quit smoking were categorized as 1 (having quit) and those who continued smoking or transitioned to mixed use were coded as 0 (continuing smoking).
Functional beliefs were assessed using three statements: (F1) You enjoy smoking too much to give it up, (F2) Smoking calms you down when you are stressed or upset, and (F3) Smoking is an important part of your life. Risk-minimizing beliefs were assessed using the following three statements: (R1) The medical evidence that smoking is harmful is exaggerated, (R2) Everybody has got to die of something, so why not enjoy yourself and smoke, and (R3) Smoking is no more risky than lots of other things that people do. These psychosocial beliefs were measured on a five-point Likert-scale ranging from Strongly agree  to Strongly disagree . These beliefs were also dichotomized for frequency analysis where Strongly agree/Agree were coded as 1 (having a belief) and the Neither agree nor disagree/Disagree/Strongly Disagree were coded as 0 (NOT having a belief).
Analyses were conducted using STATA/SE 17. Univariate statistics were used to categorize the sample and bivariate statistics such as paired t-tests were conducted to analyze the difference between justifications and smoking status between two waves. Multivariable logistic regression was used to examine the association between quitting at Wave 2 and functional as well as risk-minimizing beliefs. Two separate models were run to account for these beliefs at Wave 1 and Wave 2 separately. Additionally, models assessing the mean scores of functional and risk-minimizing beliefs were followed by models that analyze each belief item individually. Similar models were also run for the “planning to quit” outcome at Wave 2. Weights were calculated to adjust for disproportionate sampling respondents in subgroups and longitudinal sampling weights were used for regression analysis. The models also included the covariates: age, sex, education, and income.