The health, poverty, and financial consequences of a cigarette price increase among 500 million male smokers in 13 middle income countries: compartmental model study

Abstract Objective To examine the impact of a 50% increase in market prices of cigarettes on health, poverty, and financial protection. Design Compartmental model study. Setting 13 middle income countries, totalling two billion men. Participants 500 million male smokers. Main outcome measures Life years gained, averted treatment costs, number of men avoiding catastrophic healthcare expenditures and poverty, and additional tax revenue by income group. Results A 50% increase in cigarette prices would lead to about 450 million years of life gained across the 13 countries from smoking cessation, with half of these in China. Across all countries, men in the bottom income group (poorest 20% of the population) would gain 6.7 times more life years than men in the top income group (richest 20% of the population; 155 v 23 million). The average life years gained from cessation for each smoker in the bottom income group was 5.1 times that of the top group (1.46 v 0.23 years). Of the $157bn (£113bn; €127bn) in averted treatment costs, the bottom income group would avert 4.6 times more costs than the top income group ($46bn v $10bn). About 15.5 million men would avoid catastrophic health expenditures in a subset of seven countries without universal health coverage. As result, 8.8 million men, half of them in the bottom income group, would avoid falling below the World Bank definition of extreme poverty. These 8.8 million men constitute 2.4% of people living in extreme poverty in these countries. In contrast, the top income group would pay twice as much as the bottom income group of the $122bn additional tax collected. Overall, the bottom income group would get 31% of the life years saved and 29% each of the averted disease costs and averted catastrophic health expenditures, while paying only 10% of the additional taxes. Conclusions Higher prices of cigarettes provide more health and financial gains to the poorest 20% than to the richest 20% of the population. Higher excise taxes support the targets of the sustainable development goals on non-communicable diseases and poverty, and provides financial protection against illness.


Derivation of outcomes
We estimated the impact of a 50% price increase in cigarette prices on the following health and financial outcomes for each of the 13 countries: a. Baseline number of male smokers by age and five income groups (fifths) b.
Years of life gained after price intervention c.
Treatment cost averted d.
Individuals averting catastrophic health expenditures and poverty e.
Additional tax revenue

Baseline number of male smokers by age and five income groups (fifths)
Data Sources: (1) 2015 population from UN Population Division; (2) smoking prevalence, by five income groups (fifths) and age-group (5-year) from GATS and similar local surveys.
We defined a current smoker as one who smokes cigarettes either daily or at least once every week. We focused only on manufactured cigarettes and not on bidis, small and locally-grown cigarettes sold commonly in India and Bangladesh. We used asset index as measure of income. For countries without readily (Available asset index in their respective surveys, we used educational attainment as proxy, and applied the relative prevalence of smoking among illiterate or completion of primary, secondary or high school or college. Procedure: A price increase results in reduction of number of smokers and is subject to the responsiveness of smoker to price change. The price elasticity, of a smoker in turn is influenced by and . As per the literature, the for cigarettes is about -0.4 meaning a 50% price increase will reduce smoking by about 20%. (82,83) Of this reduction, about half (10%) is attributable to participation elasticity i.e. quitting by current smokers and half to demand elasticity resulting in less amount smoked. Consistent with the published literature showing greater price responsiveness in the young and among the poor (82,83) , we doubled the national among younger smokers (15-24 years old), and also applied this higher price elasticity to future smokers below 15 years old that have not yet started to smoke. (84,85) Similarly, we used a relative weighted price elasticity matrix by income and age drawn from existing studies with the smokers in the bottom fifth (20%) of the population being more price responsive compared to the top fifth. Therefore, the number of quitters is estimated by: Among persistent smokers, about half of prolonged smokers who do not quit are killed by smoking. This risk is particularly relevant to smokers below age 35 years in LMIC who are likely to have smoked from early in adult life. (86) Here, we conservatively assumed half of current and future smokers would be killed, given that smoking cessation rates in most LMICs are far lower than that in high-income countries . (86,87) Reductions in the excess (all-cause) mortality from smoking are greatest in smokers who quit early in life (and naturally in those who do not start). We applied age-specific benefits of cessation from epidemiological studies in the US and the UK among men and women, (77,88,89) corresponding roughly 97% of smokers avoided excess mortality by quitting by at 15-44 to about 25% avoided excess mortality by quitting by age 65 years. We adopted the risk reduction estimates ( ) by age group from Verguet et al. Further, we fitted a cubic spline to derive the age-specific life years gained from smoking cessation for all ages ( ). (81) To be conservative, we ignored the beneficial effects of reduced smoking amount. We proportioned the reductions in overall mortality across income groups and across four main causes of smoking-related mortality: chronic obstructive pulmonary disease (COPD), stroke, heart disease and tobacco attributable cancers from model-based estimates from the Global Burden of Disease. (15) For China and India, we were able to compare the GBD with direct large epidemiological studies, which yielded generally consistent results for male smoking deaths, but not for women where the GBD estimated wrongly that about 8% of Chinese adult female deaths are due to smoking when the prevalence of adult female smoking is only 2% and even lower in the cohort of women born after 1950. (89) This discrepancy did not, however affect the calculations for males. The total deaths averted are estimated by: Further, the life years gained (LYG) are estimated by: Treatment cost averted Data Sources: (1) treatment cost, insurance coverage rate, financial support, and healthcare utilization were obtained from peer-reviewed journals and country reports; (2) Purchasing Power Parity (PPP) adjustment factor, and Consumer Price Index were obtained from World Bank Procedure: We calculated the treatment cost averted by smokers who quit after price intervention. We obtained local treatment cost estimates, for each of the 4 disease conditions each country. To equalize the purchasing power of local currencies, we adjusted our cost estimates using a 2015 PPP conversion factor. We estimated the averted total healthcare expenditure (treatment cost), , , conditional to seeking health-care or being ill, using the following formula: We also derived the averted OOP health expenditure, , , by adjusting the treatment cost with coverage rate of the publicly-funded system, , probability of seeking health-care conditional on being ill, , and the percentage of total costs covered by the public healthcare system, : where, = (vi)

Individuals averting catastrophic health expenditures and poverty
Data Sources: (1) Gini Coefficient from the World Bank; (2) average household income capita (2015) were obtained from statistical offices of countries (PPP-adjusted).
Procedure: Individuals averting catastrophic health expenditures i.e. greater than 10% of their income, attributable to tobacco: We applied the World Bank definition of poverty i.e. earn less than US$ 1.9 /day/capita, World Health Organization's definition of catastrophic health expenditures meaning when out-of-pocket treatment costs exceed 10% of an individual's income for our analysis. We used average household income per capita obtained from statistics offices of respective countries and Gini Coefficient from World Bank to construct gamma distribution of per capita household income. (90) The probability , of individuals falling into poverty or incurring catastrophic health expenditures was derived from this distribution of household income. We estimated the total number of individuals having catastrophic health care expenditures attributed to out-of-pocket cost that would be averted by a 50% increase in price by following formula: , where; (viii) is the average number of sticks consumed by smokers in income group q, is the tax rate per pack of cigarettes at the baseline, and is the new tax rate post price increase. Thus, marginal tax revenues, gained is given by: