Tobacco Use and Quit Behaviour Assessment in the Global Adult Tobacco Survey ( GATS ) : Invalid Responses and Implications

Quit attempt among tobacco users is an important tobacco cessation behaviour indicator (Starr et al., 2005; Fong et al., 2006; Global Adult Tobacco Survey Collaborative Group, 2011). It has been defined as an activity by a tobacco user in which the person tries to quit (stop using) with the intention of never using it again (IARC, 2008). The numerator of this indicator includes current tobacco users (smokers/smokeless users) who tried to quit during the past 12 months and former tobacco users (smokers/smokeless users) who have been abstinent for <12 months. The denominator includes the total number


Introduction
Quit attempt among tobacco users is an important tobacco cessation behaviour indicator (Starr et al., 2005;Fong et al., 2006;Global Adult Tobacco Survey Collaborative Group, 2011).It has been defined as an activity by a tobacco user in which the person tries to quit (stop using) with the intention of never using it again (IARC, 2008).The numerator of this indicator includes current tobacco users (smokers/smokeless users) who tried to quit during the past 12 months and former tobacco users (smokers/smokeless users) who have been abstinent for <12 months.The denominator includes the total number RESEARCH ARTICLE Tobacco Use and Quit Behaviour Assessment in the Global Adult Tobacco Survey (GATS): Invalid Responses and Implications Pratap Kumar Jena 1 *, Jugal Kishore 2 , Sanghamitra Pati 3 , Bidyut Kanti Sarkar 4 , Sagarika Das 5 of current tobacco users, and former tobacco users who have been abstinent for <12 month (GTSS, 2009).This indicator has been derived from two separate questions of the GATS survey asked of current and former tobacco users in order to avoid underestimation when only current users are included for assessment.While tobacco use represents falling into the captive of tobacco dependence, the quit attempt represents its opposite dimension.Both the tobacco use prevalence and quit attempt among tobacco users are the measures of effectiveness of tobacco control policies (IARC, 2008).IARC (2008) review on natural history of tobacco use indicates that non smoker would undergo a series of steps 6564 like trial, experimentation, established use, attempting to quit, relapse, and/or maintenance of cessation.Smoking behaviour becomes established when a smoker has smoked at least 100 lifetime cigarettes (Starr et al., 2005).After smoking behaviour gets established, discontinuation of smoking involves an attempt to quit.The outcome of each quit attempt may be relapse or maintenance of cessation.Quit attempts can be planned or sudden and may be done abruptly or gradually with or without any assistance of one or more tobacco cessation methods (IARC, 2008).
Quitting tobacco use has definite and immediate health benefits (USDHHS, 2004;Hughes, 2007;IARC, 2007).However, quitting is not easy; and this process requires first trying to quit (quit attempt) and then being successful in quitting i.e. cessation (Jardin et al., 2012).The nil past history of quit attempt is an indicator of inability of quitting by a smoker and is a component construct of hardcore smoking definition (Kishore et al., 2013).Two theories i.e.Stages of Change (DiClemente et al., 1991;Prochaska et al., 1992) and PRIME theory (West, 2006) have been advocated to understand the process of quitting.Former entails that change of smokers from precontemplation (not interested in quitting) to established non smoking status requires a series of stages of change.The stages between these two opposite stages (smoking vs. non smoking) are contemplation, preparation, action and relapse/maintenance.Quit attempt is the part of "action stage" as advocated by this model.The arbitrary classification of different stages and limitation in explaining abrupt quitting by this theory was addressed in the Robert West's PRIME theory (McEwen et al., 2006;West, 2006).Robert West suggests that smokers make a quit attempt when the desire to stop exceeds the desire to continue smoking.Quit attempt may or may not be preceded by a plan to quit.Quit attempt becomes a rule when smokers tried not to smoke and then stopped smoking (West, 2006).
A recent review by Jardin et al. (2012) suggests that duration and number of previous quit attempts; and intention to quit are strong predictors and necessary precursors of successful quitting.Availability and type of tobacco cessation services also modulates success in smoking quit attempts (Fiore, 2007).However nicotine dependence is the single most consistent predictor of success following a quit attempt (Vangeli et al., 2011).Various key indicators have been defined for population level monitoring and evaluation of tobacco cessation/ control programs, which include classification of smokers by intention and attempts to quit smoking (WHO, 1998;Giovino et al., 2009).Global Adult Tobacco Survey (GATS) is a part of Global Adult Tobacco Surveillance System (GTSS) that monitors tobacco control indicators among adults of 15 years old and above.Quit attempt assessment is a key indicator in GATS to assess tobacco cessation behaviour among current and former users (IIPS, 2010;Global Adult Tobacco Survey Collaborative Group, 2011).Earlier definitional and conceptual problems in GATS have been identified (Jena et al., 2012;2013).The objective of this study is to examine conceptual issues and validity of responses with the GATS assessment of quit attempt.

Materials and Methods
GATS-India (IIPS, 2010) survey tool Section B and Section D1assessed smoking tobacco use and its quit attempt behaviour respectively.Similar assessment for smokeless tobacco use was done under section C & D2 of the GATS tool.The concept of tobacco use behaviour and quit attempt assessment as in GATS are graphically represented in Figure 1.
GATS has assessed three category of current (daily, less than daily and not at all) users of tobacco.Current not at all but former users and 'current less than daily' users were assessed for their past tobacco use behaviour.Current daily users were not assessed for their past use behaviour.Unlike other surveys, the definition of current use in GATS does not have definite timeframe of assessment.The term 'current daily smoker' has been referred as the person currently smokes at least one tobacco product every day, over a period of one month or more (IIPS, 2010).Taking cue from this definition, current user would mean that users in the past 30 days or beyond preceding the survey.Thus the time of reference was limited to 30 days preceding the survey to characterize current users in this study, which is similar to other surveys (IARC, 2008).
Items numbered D01 to D03 under cessation section of the GATS questionnaire measured the duration of the last quit attempt among the current smokers considering the quit attempts in the past 12 months preceding the survey.Similar measurement for smokeless users was assessed by the items D09, D10 and D11.GATS had also assessed duration of abstinence (successful quit attempt) among former tobacco users (Items B13 and C13).These items assessed self reported tobacco use behaviour and duration of last quit attempt or abstinence.Such assessment was subjected to stratified analysis to identify invalid responses without using GATS sampling weights.
The measurement of duration of the last quit attempt among current users were collected by recording time units like 'Month', 'Week', 'Day' and 'less than 24 hours' etc and its frequency.Similar time units along with 'Year' were used to assess the duration since abstinence among former users.These time units represent ranked data and sum of fixed number of lower units result in formation of higher unit i.e. seven days would result in one week, four weeks would result in one month and so on.As far as the duration of the last quit attempt during past 12 months among the current users is concerned, it can't be >11 months i.e. one year.More so, the duration of last quit attempt among the less than daily users when expressed in terms of 'less than 24 hours' or 'days', it would also indicate invalid responses, as these users may be weekly user or monthly user etc.The duration of quit attempt when expressed in 'more than 7 days', 'more than 4 weeks', 'more than 12 months' etc., would result in erroneous interpretation owing to variation in unit of measurements of time and inter unit relationship.Therefore for the purpose of this study we assumed logics as in Figure 2 to identify invalid responses.Category A included absolute invalid responses when time-frame of assessment of current tobacco use and less than daily tobacco use were violated.Category B included responses that violated unit of measurement of time.

Results
GATS had assessed both current and past use behaviour of tobacco users.Table 1 represents distribution of respondents according to current and past use of tobacco.These figures in the table clearly indicate that 'daily use' was four times more prevalent than 'less than daily' use among current smokers and smokeless tobacco users.Among the current less than daily users, around one out of four, were daily users in the past.Daily smoking was predominant among former smokers but reverse trend was observed among former smokeless tobacco user.
Table 2 represents invalid responses identified by stratified analysis of duration of the last quit attempt among the current users of tobacco.It is to be noted that, the denominator of current user (use of tobacco within 30 days preceding the survey) and duration of last quit attempt in the past 12 months are different.Responses of 3102 current smokers and 4036 current smokeless users were analyzed to quantify invalid responses.Analysis indicated overall 21.24% (category A: 7.32%; category B: 17.67%) and 22.7% (category A: 8.05%; category B: 18.09%) invalid responses among current smokers and smokeless users respectively regarding their duration of last quit attempt among current users.
Table 3 represents invalid responses identified by stratified analysis of duration of abstinence among the former users of tobacco.Responses of 1904 former smokers and 1343 former smokeless users were analyzed to quantify invalid responses with respect to their duration of abstinence.Analysis indicated overall 6.62% (category A: 4.7%; category B: 2.3%) and 10.57% (category A: 8.6%; category B: 3.5%) invalid responses were identified among former smokers and smokeless users respectively regarding their duration of abstinence.

Discussion
The responses for the duration of last quit attempt by the current users are invalid in 1 out of 5 cases.Similarly the measurement of duration since abstinence among former user has yielded 6.6-10.6%invalid responses.The invalid responses may be due to fact that, the time frame of 'quit attempt', 'current use' and 'former use' are different.This is attributable to the ill defined current use and former use definition which formed the denominator of quit attempt indicator.Overall, the quit attempt indicator needs to be improved to measure what it intended to measure.
If we assume that Category A responses are valid one, these responses would indicate that former user are those who used tobacco just one day or one week or one month before the survey.However, in such instances current daily use definition (daily use for ≥1 month) as well as former use definition in GATS would be contradicted.Similarly if we assume that responses for duration of last quit attempt among current users is ≥11 months, it would defeat the very purpose of assessment of quit attempt in the past 12 months among the current daily users.
The self-reported data of tobacco cessation behaviour like quit attempt is not same as simple answers to the question.Measurement of self-reported duration of quit attempt in the past may stretch the capacity of the respondent to recall the past events and hence this measure is subject to a range of possible biases as indicators of their target constructs (IARC, 2008).The same biases are also applicable to measurement of current or past use of tobacco.However owing to time frame as per definition of various indicators, the biases may also vary.The biases may be less for current use assessment where time frame of assessment is limited to the use of tobacco within 30 days preceding the survey than the measurement of quit attempt in the past 12 months.The ill defined time frame of current use in GATS survey wouldn't help to identify the magnitude of bias in this measurement.It has been observed that respondents appear to forget many quit attempts of short duration, specifically, if the attempt took place more than three months before the interview (Gilpin et al., 1994;West et al., 2007).GATS measured the quit attempt considering past use of tobacco beyond 3 months and also imprecisely defined the current use time frame.This would lead to facilitation of occurrence of bias and failure of measuring the magnitude of the bias.
Having quit for more than 7 days during the previous 12 months has been considered a strong quitting history leading to cessation in future (Pierce et al., 1998).Therefore authors support 12 month timeframe of assessment for quit attempts but not the ill defined time frame of assessment for current and former users in the GATS tool.As per natural history of tobacco use, quit attempt is possible among established user, i.e. lifetime use of 100 cigarettes.These established users may be daily or less than daily.While it is easy to identify short duration of quit attempt among daily users, it won't be the same for less than daily users.Therefore assessment of quit attempt among less than daily users should at least be expressed in week time unit not in day or <24 hr time unit.Moreover GATS being a policy influencing tool, we should avoid recording of micro details of short duration (<7 days) of quit attempt in the global adult survey.
The number of prior quit attempts also predicts the future quit attempts (Zhou et al., 2009;Jardin et al., 2012) and increases the probability of cessation (West et al., 2001;Tobias et al., 2010).The fact that individuals with higher number of quit attempts are repeatedly trying to quit tobacco use indicates that they are motivated (Joseph et al., 2004), but are just unable to maintain long-term abstinence.Successful smoking cessation programs must identify this group of smokers as having many failed attempts may result in frustration, fear, defiance, and loss of interest in quitting (Wolburg et al., 2006).Hence GATS should additionally include number of prior quit attempts in the past 12 months preceding the survey.Spontaneous quit attempt has more success rate than planned one (West et al., 2006).Similarly abrupt quit attempt has more success rate than gradual reduction (Gritz et al., 1999).These assessments have been given importance in ITC (2007) and Smoking Tool Kit Study (West, 2006) but not in GATS tool.
GATS is a nationally representative cross sectional survey and hand held devices were used to record the responses digitally.This digital interface in the hand held device could be used to identify invalid responses during response recording and the 'prompt or pop up' that generated could be used for rectify the responses instantly.However current study results indicate that there was inadequate utilization of the opportunity provided by digital data recording interface to collect the valid responses.
The credibility of a survey tool depends on its ability to collect the valid and reliable data.Therefore it is pertinent to identify the items that failed to collect valid and reliable information.Quit attempt is a measure of tobacco control policy and hence it is important to have credible information on this indicator at all times.GATS has used multiple tobacco control policy measures that could result in increased internal validity of the survey for evaluating tobacco control policies.However unless each indicator captures the true information, it would be difficult to internally validate using various indicators.
In situations where tobacco control evaluation studies have to depend on cross-sectional data collection, measures of quit attempts could be considered as proxies of future behavior.Further, being more responsive to policy changes in the short run compared to populationlevel prevalence rates of smoking (Fong et al., 2006a;2006b) quit attempt assessment is taken as intermediate evaluation outcomes.GATS has limitations for assessing quit attempt, similar to assessment of cigarette per day (Jena et al., 2013) due to definitional issues and failure to utilize opportunity of digital interface to record responses in the survey directly.Based on results of this study authors recommend following revisions in the GATS tool for assessment of quit attempt.
In recommendations, i) Quit attempt should be redefined considering the appropriate time-frame of assessment of current use, former use and quit attempt; ii) The GATS should assess the number of quit attempts that could be used as a proxy for their future success in quitting; iii) Validity and reliability of the tool should be tested before the actual survey; iv) The digital interface should be validated using predefined logic for each cell used for data recording.The digital data collection interface should be designed to convert all time units (year, month, week, day) into single time unit like day to avoid confusion.

Figure 1 .
Figure 1.Assessment of Tobacco Use and Quit Attempt in Global Ault Tobacco Survey

Figure 2 .
Figure 2. Logic Used to Identify Invalid Responses in Measurement of Duration

Table 3 . Stratified Analysis of Duration of Abstinence among Former Tobacco Users to Identify Invalid Responses
Both Category A & B Invalid Responses B03, B13, C03, C13 represents respective items in GATS-India survey (2009-10) tool

Table 2 . Stratified Analysis of Duration of the Last Quit Attempt among Current Tobacco Users to Identify Invalid Responses
Thinking about the last time you tried to quit, how long did you stop smoking?[Enter Unit] :http://dx.doi.org/10.7314/APJCP.2013.14.11.6563Tobacco Use and Quit Behaviour Assessment in the Global Adult Tobacco Survey: Invalid Responses and Implications D10a.Thinking about the last time you tried to quit, how long did you stop using smokeless tobacco?[enter unit] Asian Pacific Journal of Cancer Prevention, Vol 14, 2013 6567 DOI