1 INTRODUCTION

Tobacco kills nearly six million people annually, and it is estimated to have caused a billion deaths in the last century.1 India is the world's second-largest consumer of tobacco, with 29% of the population consuming some form of tobacco.2 It is an important risk factor for noncommunicable diseases, particularly oral cancer. Tobacco is responsible for 90% of oral cancer cases, with chewing tobacco being the chief culprit.3

Money spent on tobacco use eats into funds that can otherwise go towards meeting basic needs such as food and education. The high cost of treating tobacco-related diseases further exacerbates poverty.4 Characteristics such as socio-economic status (SES), age, sex, income and educational level are strongly associated with nicotine dependence.5-7

Quality-adjusted life-years (QALYs) are summary measures of disease burden that weigh life years lived with preference-based HRQOL scores. One year lived in a reduced health state of the utility value of 0.5 is equal to 0.5 QALYs, the same as living half a year in perfect health. The QALY uses the health utility value to weight years of life lived and so can be used for calculating the economic costs of a condition or a risk factor and for determining the cost-benefit of treatments, interventions and health policies.8 In tobacco cessation/prevention programmes, benefits of quitting smoking may be expressed in QALYs.

Utilities, in health economics, are fundamental values that mirror an individual's preferences for different health states under conditions of uncertainty. The health utility value assesses the value of one health state versus another.9 The value is anchored at 0 for death and 1 for perfect health. Utility ratings or values provide the utility weight needed to calculate quality-adjusted life-years (QALYs).8, 10 There are two methods of measuring utility, namely the indirect and the direct methods. The indirect (or holistic) methods of utility measurement use generic preference-based measures or disease-specific preference measures. The most commonly used indirect measures to classify tobacco-related health states (the Health Utility Index, Quality of Well-Being Scale and EuroQol-5D) have used varied approaches11-13 and produced different findings when assessing different health conditions.12, 14

The direct methods used to measure utilities include the rating scale, the standard gamble and the time trade-off.9, 13 The standard gamble reveals an individual's preference for a particular health state by offering a choice between two options: living with certainty in a health state or taking a gamble on a new intervention whose outcome is unknown.

Direct methods may provide higher utility scores than indirect measures for the same conditions, and they are more suited for utility assessments among patients/individuals suffering from the health condition. By contrast, indirect measures are usually administered to the general population and are simple, easy to administer and less time-consuming. However, they may not fully encapsulate the complex multifactorial conditions and the benefits of mitigating factors like family, friends, work and other influences that only a patient feels.15 Direct measurement of utility ratings and estimates for nicotine dependence-related states have not been reported in the literature.16 Better health utility rating estimates for such states could improve QALY estimates and enhance the effectiveness of cost-utility analyses for tobacco prevention/intervention programmes.17

It has been argued that the terms ‘habit’ and ‘dependence’ denote different things.18 A habit can be considered as an act that is performed repeatedly but is not intractable, whereas dependence denotes impaired control over a habit which has both psychological and physiological underpinnings.19 Research on tobacco-related economic analyses has always focused on the ‘habit’ through metrics like duration, frequency of smoking/chewing and not on the behavioural aspects. Scales such as the Fagerstrom scale assess both the quantitative and the behavioural aspects of nicotine dependence and may be better at evaluating the effectiveness of tobacco control programmes.

As for the Indian situation, there has been only one study where the cost-effectiveness analysis of a school-based smoking prevention programme was reported, and even in that study, preference-based measures had not been used.20 Hence, the objectives of the current study were to measure utility ratings and QALYs for nicotine dependence-related health states using the standard gamble approach among a sample of dental patients. We also investigated the associations of nicotine dependence and studied the influences on tobacco-related health state utilities estimates among patients. We hypothesized that utility ratings would be lower and QALY loss greater in those with more severe nicotine dependence.

2 MATERIALS AND METHODS

We conducted a cross-sectional survey among a consecutive sample of dental patients attending university dental clinics. Included in the study were adults aged 18 years and above, who were current or former (quit less than six months ago) consumers of tobacco (chewing or smoking). Those who provided informed consent were interviewed by one of the authors in the Department of Oral Medicine and Radiology. Ethical approval for the study was first obtained from the Institutional Ethics Committee.

The standard gamble method was used in a previous study by the same authors to measure utility ratings for dental health.21 The sample size in the previous study was based on the difference in rating scores (effect size) detected by the standard gamble method. Considering the sensitivity of the standard gamble method (with 80% power and an alpha of 0.05) to detect a difference of 0.6 utility points measured with an estimated standard deviation of 0.2, the necessary sample size of 175 individuals was calculated. This was inflated by 10% to account for incomplete responses, giving a final sample of 193, further rounded up to 200 patients.

Demographic and tobacco-related data (duration of the habit, frequency of consumption, type of tobacco consumed, current or former user) were collected through an interview with open-ended questions. These descriptive questions have been previously used in global tobacco surveys.22 A global rating scale for oral health (Locker 2001) and the Fagerstrom scale for nicotine dependence were also administered. Following this, the standard gamble utility valuation was carried out.

2.1 Global rating scale (RS)

This is a single item scale23 that measures the subjective oral health state of an individual with the question: ‘How would you describe the health of your teeth or mouth?’; it uses the ordinal response options ‘Excellent’, ‘Very good’, ‘Good’, ‘Fair’ or ‘Poor’.

2.2 Fagerstrom scale for nicotine dependence

The six-item scales for both smoking24 and smokeless tobacco25 were translated into the local language (Kannada) and used for smokers and smokeless tobacco users, respectively. Both scales were translated into the local language and back-translated into English to remove any inconsistencies before their use. The classification of dependence was 0-2 (Very low); 3-4 (Low); 5 (Moderate); 6-7 (High) and 8-10 (Very high). Summary scores (ranging from 0 to 10) for both scales were calculated, and the patients were divided into a ‘low to moderate dependence’ group (≤4) and a ‘significant dependence’ group (≥5).

2.3 Standard gamble utilities

The participants were asked to choose between two tobacco-related health scenarios. One of those was to continue with their current health state; the other was to undergo a hypothetical new treatment that would completely reverse all of the damage caused by their tobacco habit back to their habit-free days. However, it could also result in adverse health consequences leading to death. The probability of success for the new treatment was not known. Patients were asked to indicate on a graduated scale, the minimum chance of success (in %) they would require to accept the new treatment. Anything less than this indifference point would be deemed to change their preferences in favour of the first option. The anchor points in the standard gamble scale were 0 and 1. For instance, if a patient was indifferent between continuing in the current state of health and a gamble with a probability of 0.7 (70%) of complete reversal of tobacco-related damages and a complementary probability of 0.3 (30%) of adverse health consequences leading to death, the preference for his or her tobacco-related health state was equal to 0.7. Higher SG scores indicate better health utility and quality of life.

2.4 Data analyses

Summary statistics for all variables were calculated. The main dependent variable was the health utility rating (using the SG technique). QALYs were calculated by multiplying the Standard Gamble scores with the duration of the habit in years. Bivariate analysis to compare utility ratings against socio-demographic and tobacco habit-related variables was done using the Mann-Whitney U test and Kruskal-Wallis ANOVA. Variables with P-values of < .2 were then used in the Poisson regression model to test the association between the Fagerstrom scores and health utility estimates. Poisson regression was used because of the skewness and heteroscedasticity of the SG scores. IBM SPSS version 20 (IBM SPSS Statistics for Windows, Version 20.0. Armonk, NY: IBM Corp) was used for analysis. A P-value of < .05 was considered statistically significant.

3 RESULTS

Of the 200 original participants, 7 (3.5%) failed to understand the standard gamble methodology and were unable to complete the interview, leaving 193 to take part. Table 1 presents the mean SG and Fagerstrom scores by participants’ socio-demographic, smoking and health characteristics. A majority were male, were more than 40 years of age, were engaged in unskilled/semi-skilled occupations or had not completed high school. Most were current users of tobacco (especially the smokeless variety) and had previously tried unsuccessfully to quit the habit. A majority of the respondents rated their oral health as ‘fair/poor’, and 13% (n = 25) had a confirmed diagnosis of a precancerous/cancerous condition. Mean Fagerstrom scores (reflecting the degree of nicotine dependence) were higher among the older age groups, the less-educated, users of smokeless tobacco and those with a higher frequency and duration of the habit. Those who reported poor oral health also had higher Fagerstrom scores. Standard gamble scores were higher among those who reported better oral health.

TABLE 1. Mean Standard Gamble(SG) and Fagerstrom scores by demographic and tobacco variables N (%)

SG Score

Mean (SD)

P-value

Fagerstrom score

Mean (SD)

P-value Age 15-24 15 (7.8) 0.8 (0.2) .5 1.7 (1.8) .001 25-39 58 (30.1) 0.8 (0.2) 2.4 (2.3) 40-54 77 (39.9) 0.7 (0.2) 3 (1.9) >54 43 (22.3) 0.7 (0.2) 3.9 (2.4) Sex Male 166 (86.0) 0.7 (0.2) .5 2.9 (2.2) .2 Female 27 (14.0) 0.7 (0.2) 3.4 (2.2) Occupation Unskilled/semi-skilled 115 (59.6) 0.7 (0.2) .9 2.9 (2.3) .3 Skilled/clerical/shop owner 38 (19.7) 0.7 (0.2) 2.6 (2.1) Semi-professional/professional 40 (20.7) 0.8 (0.2) 3.3 (2) Education 1-3 21 (10.9) 0.7 (0.2) .05 3.5 (2.3) .001 4-7 46 (23.8) 0.7 (0.2) 3.3 (2.1) 8-10 53 (27.5) 0.7 (0.2) 3.4 (2.1) 11-12 34 (17.6) 0.7 (0.2) 2.7 (2.3) Graduate/PG 39 (20.2) 0.8 (0.2) 1.8 (2.1) Habit status Quit Recently(within 6 months) 29 (15.0) 0.8 (0.2) .19 2.6 (2.2) .37 Current 164 (85.0) 0.7 (0.2) 3 (2.2) Quit attempts Tried to quit 138 (71.5) 0.7 (0.2) .99 2.8 (2.1) .39 Never tried to quit 55 (28.5) 0.7 (0.2) 3.2 (2.4) Tobacco type Smoking 54 (28.0) 0.7 (0.2) .77 2.4 (2.3) .01 Chewing 139 (72.0) 0.7 (0.2) 3.2 (2.2) Frequency of habit Up to 4 times/day 104 (53.9) 0.8 (0.2) .27 2.3 (2.1) <.001 More than 4 times/day 89 (46.1) 0.7 (0.2) 3.7 (2.1) Duration of habit Up to 8 years 91 (47.2) 0.7 (0.2) .55 2.5 (2.2) .004 More than 8 years 102 (52.8) 0.7 (0.2) 3.4 (2.2) Global oral health rating Poor 36 (18.7) 0.6 (0.2) <.001 4.6 (2.2) <.001 Fair 74 (38.3) 0.7 (0.2) 2.9 (2) Good 28 (14.5) 0.8 (0.2) 2.4 (2.1) Very good 43 (22.3) 0.8 (0.2) 2.4 (2.1) Excellent 12 (6.2) 0.9 (0.7) 1.8 (1.8) Clinical Diagnosis Normal 168 (87.0) 0.7 (0.2) .33 2.9 (2.2) .32 Premalignant/malignant 25 (13.0) 0.7 (0.2) 3.3 (2.1)

At 0.8 (SD, 0.2), the mean SG score of the ‘low to moderate’ (Fagerstrom score of ≤ 4) nicotine dependence group (n = 142, or 73.6%) was higher than the 0.7 (SD, 0.2) of the ‘moderate to high’ (Fagerstrom score of ≥ 5) nicotine dependence group (n = 51, or 26.4%; P < .001).

QALYs were calculated considering the duration of the habit as the time spent in the tobacco-related health state multiplied by the utility weight (SG score) for both the ‘low to moderate’ dependence group and the ‘significant dependence’ group (Table 2). The mean QALYs for the two groups were 8.8 (SD, 8.9) and 9.6 (SD, 7.5) with the mean duration of the habit for the two groups being 11.4 (SD, 10.6) and 16.3 (SD, 13.3), respectively. QALYs lost due to the habit among the ‘low to moderate’, and ‘significant’ dependence groups were 2.7 (SD, 3.7) and 6.7 (SD, 8.0), respectively.

TABLE 2. QALY by nicotine dependency status and QALY loss due to the habit Fagerstrom scores P-value Low to moderate dependency Significant dependence Mean(SD) Mean(SD) Duration of habit in years 11.4 (10.6) 16.3 (13.3) QALY for the duration of habit 8.8 (8.9) 9.6 (7.5) .2 QALY lost 2.7 (3.7) 6.7 (8.0) .001 Abbreviation: QALY, Quality adjusted life year.

Table 3 presents the Poisson regression model for the SG score. Poisson regression was done as the SG variable was count with a large number of data points for just a few values.

TABLE 3. Poisson regression model to test the association between the Fagerstrom scores and health utility estimates (Standard Gamble scores) through probability ratios Parameter PR 95% CI Age 1 0.99-1 Gender Male 0.99 0.94-1.05 Female 1 Education 1-3 1.02 0.95-1.1 4-7 0.9 0.85-0.95 8-10 0.96 0.91-1.01 11-12 0.98 0.93-1.03 Graduate /PG 1 Fagerstorm score Low to moderate 1.16 1.11-1.21 Significant dependence 1 Global rating Poor 0.79 0.73-0.85 Fair 0.86 0.8-0.92 Good 0.90 0.84-0.97 Very good 0.94 0.87-1.01 Excellent 1 Habit status Quit 1.07 1.02-1.12 Current 1 Note Dependent Variable: Standard Gamble Score. Predictors which were significant at P < .2 included. Abbreviation: PR, Probability ratio.

This model was selected based on the chi-square goodness-of-fit test. Predictors that were significant at P = .2 in the bivariate analysis were included in the Poisson regression model for the SG score (health utility estimate) after adjusting for age, sex and occupation. Those with lower nicotine dependence, better oral health and those who had quit the habit had better health utility estimates (reflected in higher SG scores) than others.

4 DISCUSSION

We investigated the influences on health state utilities estimates among tobacco consumers and found higher health utility ratings among those with ‘low to moderate’ nicotine dependence, better oral health status or among those who had quit the habit. QALY loss for those with ‘significant’ nicotine dependence was more than double that of those with less dependence.

Ours was the first study to directly measure health utility ratings for different tobacco-related health states from a clinical sample of patients. This was in-line with the recommendations made by Arnold et al (2009)15 that direct methods of utility measurement were more suitable for use with actual patients. Dental patients were studied considering the high incidence of oral cancer and precancer in the source population, for which consumption of chewing tobacco was the overwhelming cause.26 Another reason was the assumption that dental patients in a dental clinic environment would more likely give honest and accurate information to the dentists about their tobacco consumption.

In contrast to studies where participants provided utility ratings for hypothetical health states, the current study measured the utility ratings of individuals who were actually experiencing the health states in question.20 Obtaining utility ratings from patients has its advantages and disadvantages. A review found that patients and the general population provide similar values for hypothetical health states. However, the findings differed when patients valued their own health state but the general population valued a hypothetical state, in that patients gave higher values.27 It is also possible that individuals actually experiencing a less-than-ideal health state may have gradually habituated to it and may lose their perspective of a perfect health state, thus lowering their expectations. However, utility values derived from the general population are ideal for assessing the societal costs of a disease and our findings may not be generalizable to the source population.

Our study did have some notable limitations. We used separate nicotine dependence scales for smokers and smokeless tobacco users. Both these scales, however, had identical questions, nearly identical responses, had the same maximum summary scores and the same scoring ranges to denote levels of nicotine dependence. The only difference was the substitution of smoking with chewing tobacco. The nicotine dependence groups were collapsed into two categories instead of the original five levels, in order to obtain sufficient numbers for comparison. Hence, we could describe only two tobacco-related health states, and this was less than ideal. Tobacco status was based on self-report and was not confirmed by objective testing. This method is prone to bias leading to under-reporting of the habit.28, 29 This may be a reason for the low number of self-reported female smokers/chewers in our study as well. Among those who self-report, it is possible that the habit may not be continuous, with periods of abstinence/reduction in between, and the actual duration of the habit may be less than our estimated value. Chewing tobacco users vastly outnumbered the smokers in our study. This was in broad agreement with a recent GATS (2016-17) survey, which reported the prevalence of chewing tobacco in the state of Karnataka to be more than double that of smoking tobacco.30 Clinical oral examination was substituted with the global oral health item as an examination of its validity found that it did indeed provide an adequate and valid summary of people's oral health.31

Our finding that socio-demographic characteristics such as age and education are associated with nicotine dependence was consistent with observations from other studies, where tobacco use tends to be higher among older and less-educated individuals.32-35 Nicotine dependence was also higher among smokeless tobacco users, among those who consumed tobacco more frequently and those who had a longer duration of the habit, consistent with earlier findings.36-39 Health utility ratings, as measured by the standard gamble methods, were not associated with socio-demographic characteristics, and this was consistent with previous investigations.23, 40-43 Tobacco habit-related factors like frequency, duration and the type of tobacco consumed that were found to be associated with nicotine dependence were not associated with the patients’ utility ratings. This could be explained by the reasoning that the relationship between tobacco habit severity or duration, and adverse health outcomes may not necessarily be linear. A person who has a comparatively shorter history or lower frequency of tobacco consumption but has experienced adverse symptoms due to the habit and then quit may have a different utility rating than someone who smokes/chews a lot but has no symptoms. However, the utility ratings were slightly higher in the ‘low/moderate’ group than the ‘significant’ group. This could be due to the fact that the Fagerstrom scale measures ‘dependence’ through a mainly qualitative perspective with questions like, ‘how soon after you wake up do you take your first cigarette?’ or ‘which cigarette would you hate most to give up?’ which may not match with numerical data of frequency or duration every time. The presence or absence of premalignant/malignant lesions did not seem to have any influence on the patients’ tobacco-related health utility ratings. Previous studies have shown that factors like age at diagnosis, comorbidities and stage of cancer at diagnosis to be associated with health utility ratings.44 The cases of premalignant/malignant lesions in our study had a wide variation with respect to all the variables mentioned above, which may have led to this finding.

QALYs were calculated by multiplying the utility score (SG) with the duration of tobacco habit in years for each patient in both the ‘low-moderate’ group as well as the ‘significant’ dependence groups. The mean QALY for the two groups were broadly similar, at 8.7 and 9.6, but the QALYs lost due to the habit were considerably higher (6.7) in the ‘significant’ group than in the ‘low to moderate’ group (2.7). A study among US adults also showed the QALY loss among current smokers to be almost twice that of former smokers.17 Higher duration of habit in the ‘significant’ dependence group could explain this difference. Moving from the low to moderate dependence profile to a ‘significant’ dependence profile was associated with a utility score that was lower by 0.1 points.

To conclude, tobacco consumption was negatively associated with health-related quality of life among the study participants. This study demonstrated the effectiveness of a direct method of utility assessment in measuring health utility ratings of a clinical sample of patients. This technique could represent a faster and simpler way of generating quality of life weights from clinical and general population samples that could be used for cost-utility analyses of tobacco cessation /prevention interventions in different settings and cultures and countries.

CONFLICTS OF INTEREST

The authors declare that they have no conflict of interest related to this study.

AUTHOR CONTRIBUTIONS

All authors contributed to the study conception and design. SA conceived the study topic, the study design, described the methodology and prepared the manuscript. ShA conducted all the interviews, collected and collated the study data and got the permissions and approvals. KP analysed and interpreted the data and involved in manuscript preparation and editing. MT provided expert guidance in all aspects of the study from the conception of the topic, study design data analyses and manuscript editing.

ETHICS APPROVAL

The research was conducted in full accordance with the World Medical Association Declaration of Helsinki. Ethical approval was obtained from the Institutional Ethics Committee before the start of the study.

COMPLIANCE WITH ETHICAL STANDARDS

All procedures performed in this study were in accordance with the ethical standards of the institutional research committee (Kasturba Medical College and Kasturba Hospital Ethics Committee, Reference number: 92/2019) and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. The authors declare no conflict of interest, and no external funding from any source was obtained for the conduct of this study. Informed consent was obtained from all participants prior to participating in the study.

Data will be made available on request.

REFERENCES

1 WHO report on the global tobacco epidemic 2013 [Internet]. WHO. World Health Organization; 2016. https://www.who.int/tobacco/global_report/2013/en/Accessed December 24, 2019. 2 GATS. GATS 2 Global Adult Tobacco Survey [Internet]. 2016. Available from: https://www.who.int/tobacco/surveillance/survey/gats/GATS_India_2016-17_FactSheet.pdfAccessed December 24, 2019. 3Asthana S, Vohra P, Labani S. Association of smokeless tobacco with oral cancer: A review of systematic reviews. Tob Prev Cessat. 2019; 5. 4John R, Rout S, Kumar B, Arora M. Report on economic burden of tobacco related diseases in India. Ministry of health and family welfare, Government of India; 2014. 5Shahwan S, Abdin E, Shafie S, et al. Prevalence and correlates of smoking and nicotine dependence: results of a nationwide cross-sectional survey among Singapore residents. BMJ Open. 2019; 9:e032198. 6Singh PK, Yadav A, Singh L, Singh S, Mehrotra R. Social determinants of dual tobacco use in India: an analysis based on the two rounds of global adult tobacco survey. Prev Med Reports. 2020; 18: 101073. 7Mbulo L, Palipudi K, Smith T, et al. Patterns and related factors of bidi smoking in India. Tob Prev Cessat. 2020; 6. 8Johannesson M, Jönsson B, Karlsson G. Outcome measurement in economic evaluation. Health Econ. 1996; 5: 279- 296. 9Torrance GW. Measurement of health state utilities for economic appraisal. A review. J Health Econ. 1986; 5: 1- 30. 10Gold MR, Stevenson D, Fryback DG. HALYs and QALYs and DALYs, Oh My: similarities and differences in summary measures of population health. Annu Rev Public Health. 2002; 23: 115- 134. 11Raisch DW. Understanding quality-adjusted life years and their application to pharmacoeconomic research. Ann Pharmacother. 2000; 34: 906- 914. 12Kaplan RM, Feeny D, Revicki DA. Methods for assessing relative importance in preference based outcome measures. Qual Life Res. 1993; 2: 467- 475. 13Drummond MF, Sculpher MJ, Claxton K, Stoddart GL, Torrance GW Methods for the Economic Evaluation of Health Care Programmes. Oxford, UK: Oxford University Press; 2005. 14Luo N, Johnson JA, Shaw JW, Feeny D, Coons SJ. Self-reported health status of the general adult U.S. population as assessed by the EQ-5D and health utilities index. Med Care. 2005; 43: 1078- 1086. 15Arnold D, Girling A, St