Article Text

Economic burden from smoking-related diseases in Thailand
  1. Kanitta Bundhamcharoen1,
  2. Suchunya Aungkulanon1,
  3. Nuttapat Makka1,
  4. Kenji Shibuya2
  1. 1International Health Policy Program (IHPP), Ministry of Public Health, Nonthaburi, Thailand
  2. 2Department of Global Health Policy, Graduate School of Medicine, University of Tokyo, Tokyo, Japan
  1. Correspondence to Dr Kanitta Bundhamcharoen, International Health Policy Program (IHPP), Ministry of Public Health, Nonthaburi 11000, Thailand; kanitta{at}ihpp.thaigov.net

Abstract

Objective To assess economic burden attributable to smoking in Thailand in 2009.

Methods A prevalence-based, disease-specific cost of illness approach was used to estimate the direct medical costs, indirect medical costs, productivity loss due to premature deaths and absenteeism caused by smoking-related diseases. Direct healthcare costs were obtained from the inpatient and outpatient charge database at the National Health Security Office and the Central Office for Healthcare Information. Indirect healthcare costs were obtained from the Health and Welfare Survey. The household Socioeconomic Survey provided data on income of the population. Costs were estimated for 7 disease groups, namely, lung cancer, chronic obstructive pulmonary disease (COPD), cardiovascular disease (CVD), upper aerodigestive tract cancer, other cancer, other respiratory diseases and other medical conditions. Smoking Attributable Fractions were derived from the 2009 Thai Burden of Disease study.

Results Total economic burden of smoking amounted to 74.88 billion Thai Baht (THB) (95% CI 74.59 to 75.18) (US$2.18, 95% CI US$2.17 to US$2.19 billion). Of this, most of the burden resulted from productivity loss 62.24 billion THB (95% CI 62.05 to 62.44) (US$1.81, 95% CI US$1.81 to US$1.82 billion). Total medical cost was 12.64 billion THB (12.44 to 12.85) (US$0.37, 95% CI US$0.36 to US$0.37 billion). Excluding other medical conditions, the direct healthcare costs were highest for CVD, followed by COPD and other respiratory diseases, respectively. All together, the total cost of smoking accounted for 0.78% (95% CI 0.78% to 0.79%) of the national gross domestic product and about 18.19% (95% CI 18.12% to 18.27%) of total health expenditure.

Conclusions The total economic loss from smoking-related diseases highlights the significant loss to the society, health sector and the country's economy. Such information is crucial for informing national public health policy, particularly when a conflict arises between the economy and health.

  • Smoking Caused Disease
  • Public policy
  • Economics
  • Low/Middle income country
  • Health Services

Statistics from Altmetric.com

Background

Tobacco is among the leading contributors to disease burden, causing more than six million deaths per year worldwide1 and imposing a substantial burden on the global economy.2 The economic burden of a health behaviour is widely measured by the cost-of-illness (COI) approach, which estimates the monetary value that could be saved if a disease were to be eliminated. COI studies commonly estimate two major categories of costs associated with an illness: direct costs (medical costs) and indirect costs (eg, costs due to premature mortality and loss of productivity).3 ,4 The economic burden from smoking causes a considerable loss to society, both in terms of productivity loss and healthcare cost borne by patients with tobacco-caused diseases, and their families. Smoking-related healthcare costs have been estimated to be about 6–15% of healthcare costs in high income countries.5 In the USA, smoking is the leading cause of preventable deaths, and the estimated costs of smoking-related medical expenses and loss of productivity exceed US$167 billion annually.6

In Thailand, despite a declining trend in smoking prevalence among men, from 32% in 1991, to 21.4% in 2011,7 ,8 tobacco smoking accounted for 1 in 10 of all deaths in Thailand in 2009, and 7% of total disease burden in terms of disability-adjusted life years (DALYs) lost.9 Previous studies have estimated economic losses due to smoking in Thailand using a variety of approaches.10–12 The most recent published work by Leartsakulpanitch et al10 estimated that the medical cost of treatment of lung cancer, chronic obstructive pulmonary disease (COPD) and coronary heart disease (CHD) attributable to smoking in 2006, totalled approximately 9857 million Thai Baht (THB). A number of studies have estimated direct medical cost from selected tertiary hospitals using a bottom-up approach. Most of them estimated costs for certain major diseases, namely, lung cancer, COPD and CHD, but none calculated costs for all smoking-related diseases. Owing to poor quality of cause-of-death ascertainment in death registration,13 these studies did not estimate the overall economic loss due to premature death from all smoking-related diseases.

While the government is concerned about economic growth, and since the tobacco industry is perceived as an important contributor to the gross domestic product (GDP), evidence of the economic impact resulting from health loss attributable to tobacco is needed. In consideration of the recent studies to verify Thailand's mortality data14–16 as well as the national Burden of Disease (BOD) study,9 the aim of this study was to provide a more systematic and comprehensive assessment of the economic burden attributable to tobacco smoking in the Thai population in 2009.

Methods

A prevalence-based annual cost approach was applied to estimate the costs of illness due to smoking.17 Our study included the costs incurred from health-related impacts, those being, direct medical cost, indirect medical cost, and productivity loss due to premature deaths and absence from work. A framework of the estimation approach is provided in online supplementary figure S1. Details of key parameters used in the study are provided in online supplementary tables S1–S10.

Sources of data

Mortality data for smoking-related diseases were taken from the national BOD study,9 which followed the Global Burden of Disease (GBD)18 method to estimate deaths and DALYs lost due to disease, injury and health risk factors. Hospital utilisation was analysed from three main sources of health claim data: the Ministry of Finance (MOF), the National Health Security Office (NHSO) and the Social Security Office (SSO), responsible for civil servants’ medical benefits, universal coverage and social security schemes, respectively.

Data from three nationally representative, self-reported interview surveys from the National Statistical Office (NSO), namely, the 2009 Health and Welfare Survey (HWS), 2009 Socioeconomic Status Survey (SES) and Labour Force Survey, were analysed to provide certain parameters used in the estimation. The 2008–2009 National Health Examination Survey (NHES), the fourth NHES conducted every 5 years, provided prevalence of tobacco use.

Estimation of the prevalence of smoking, health outcomes and risks in the population

Smoking prevalence by age and sex was obtained from the 2008–2009 NHES. The survey asked about current smoking among those over 15 years of age in the population. The questionnaire includes a set of questions about smoking, including whether a respondent has ever smoked 100 cigarettes in his/her lifetime. In the survey, a current smoker was defined as someone who smokes regularly and smoked at least 100 cigarettes in his/her lifetime. Following the GBD approach, health outcomes attributable to smoking were estimated for people over 30 years of age. The health outcomes included seven disease groups: (1) lung cancer; (2) COPD; (3) cardiovascular disease (CVD); (4) upper aerodigestive tract cancer (mouth, oesophageal and pharyngeal cancer); (5) other cancer; (6) other respiratory diseases (including tuberculosis) and (7) other medical conditions. Deaths attributable to smoking from these diseases were derived from the Thai BOD study.

Population attributable fractions (PAFs) were adopted from the Thai BOD study. PAF is the fraction by which the occurrence of a disease of interest would be reduced if there were no risk exposure.18 PAFs were applied to estimate the effects of tobacco on the risk of disease or mortality in a population. It was calculated from smoking prevalence and disease-specific relative risks. This is similar to SAF (Smoking Attributable Fraction), as defined by Levin19 and applied in the Rice method.17 As the Thai BOD study group uses the PAF from current smokers compared with non-smokers, we do not take into account the ex-smoker population.

Estimation of direct medical costs

Direct medical cost covered inpatient (IP) and outpatient (OP) costs due to illness related to smoking. The OP care includes medicine, and laboratory and surgical services, but does not cover care at home, or private purchase of medicine, which we estimated from out-of-pocket payments. IP care does not include long-term care, as hospitals in Thailand do not provide long-term care, and, at present, the health security insurance does not cover long-term care.

Total direct medical care cost by disease, sex and age group was estimated from IP and OP claim data. IP data were collected from all Ministry of Public Health (MOPH) hospitals, selected university hospitals, and other public and private hospitals. The data cover all patients reimbursed by the NHSO and the MOF during 2009. Owing to lack of access to claim data from Social Security Insurance (SSI) in 2009, we estimated the values based on the assumption that the proportion of claims under SSI in 2009, relative to claims under NHSO and MOF, would be identical to the proportion in 2008, for which data were available. Total charge for each disease outcome was derived from principal diagnosis coded in the databases.

For OP data, the coverage is considerably less than for the IP data. The OP data were obtained from the MOPH; most of these data came from their healthcare facilities. We interpolated the data to account for data incompleteness by each type of healthcare facility.

The number of visits and admissions were derived from the claim data, assuming no cross-utilisation among hospitals. Average length of stay for each disease, sex and age group was also derived from these databases. The total number of visits and admissions was also extrapolated to the total with the same correction factors as described above.

Direct medical costs of IP and OP due to smoking were then estimated by applying the PAFs by age group, sex and disease to the charges of patients, accordingly.

Estimation of indirect medical costs

Indirect medical costs included expenses patients incurred due to transportation and any extra medical payments not covered in the insurance. Average transportation cost per visit was obtained from the Health Intervention and Technology Assessment Program (HITAP) database.20 The out-of-pocket payment of service not covered in the insurance per utilisation was derived from the 2009 HWS, which is a self-report of the costs incurred by patients using the health services. The average transportation cost per visit was 89.16 (95% CI 80.56 to 97.76) THB and the out-of-pocket payments for hospitalisation and OP visit were 697.3 (95% CI 521.83 to 872.76) and 38.78 (95% CI 28.85 to 48.72) THB, respectively. These costs were then applied to the total number of visits and admissions derived from the IP and OP database.

Estimation of productivity loss

Productivity loss was estimated to account for future loss due to premature deaths and work loss days due to illnesses. Annual average income by age and sex analysed from the SES was applied to reflect lifetime productivity loss due to mortality, with a 3% discount rate. Income loss due to absenteeism from work for OP and IP utilisation was estimated from the number of visits per year and length of stay multiplied by the average income loss of patients aged 15–59 years, and that of caregivers for patients aged 60 years and older. The average income loss per visit for patients and caregivers was obtained from the HITAP database. For conditions resulting in hospital admission, the number of workdays lost was estimated from the length of stay derived from the IP database.

Productivity loss due to premature deaths was calculated from cumulative lifetime income by age group and sex multiplied by total number of deaths in 2009. Life expectancy for the productivity lost calculation was also obtained from the Thai BOD study.

Productivity loss from loss of workdays due to illnesses was estimated for patients aged 15–60 years and caregivers of patients aged over 60 years. It was estimated by multiplying average income per day by age group and sex by total length of stay combined with average days absent from work for hospitalised patients. For non-admitted patients, the average length of hours lost in the hospital visit was applied as the duration of income loss.

Uncertainty analysis

To assess an effect of uncertainty in the input parameters applied in the estimation, we undertook uncertainty analysis to provide the range of possible values of our findings. Monte Carlo simulation was employed whereby probability distributions of input parameters were generated.

Results

Cost of smoking by direct and indirect medical and indirect costs

Our estimates revealed that smoking resulted in an annual economic loss of 74 884 (95% CI 74 586 to 75 182) million THBi (table 1). Approximately 83% (62 billion THB) of this was due to productivity loss from mortality and work absence. Direct medical costs from healthcare utilisation accounted for 15% of the total economic loss. IP cost was approximately 59% of total direct medical care cost, a little higher than OP cost. Transportation and patients’ out-of-pocket payment for non-insured benefits equalled 1168 (95% CI 1159 to 1177) million THB (1.6% of total). Income loss due to work absence totalled 1025 (95% CI 1018 to 1033) million THB for patients and caregivers. In total, economic loss due to smoking accounted for 0.78% of Thailand's GDP and amounted to 18% of total health expenditure. This was 5% more than the total budget of the Health Ministry.

Table 1

Cost attributable to smoking by direct and indirect medical cost, and indirect costs

Smoking prevalence in Thai men was much higher than that in women, therefore the cost attributable to smoking in men is much higher, totalling more than 10 times the cost attributable to women (table 2). Smoking cost in men amounted to 69 371 (95% CI 69 078 to 69 665) million THB and 5513 (95% CI 5461 to 5565) million THB for women. Comparing across diseases, the cost was highest for CVD, equalling 20 859 (95% CI 20 704 to 21 014) million THB, followed by other medical conditions, lung cancer, COPD, other cancer, other respiratory disease and upper aerodigestive cancer. Cost of smoking in CVD shared approximately 28% of the total; other medical conditions, lung cancer, COPD, other cancer, other respiratory disease and upper aerodigestive cancer equalled approximately 18%, 16%, 15%, 9%, 8% and 6% of the total, respectively. However, if the costs of all cancers are aggregated, cancer will share the largest proportion of the total cost.

Table 2

Cost of illness attributable to smoking by sex (million THB)

Further analysis of the cost by disease and cost category is shown in table 3. Direct and indirect medical costs were highest in other medical conditions, followed by CVD, COPD, other respiratory disease, other cancer, lung cancer and upper aerodigestive cancer, respectively. Productivity loss due to premature death was found highest in CVD, followed by lung cancer, COPD, other medical conditions, other cancer, other respiratory disease and upper aerodigestive cancer. Productivity loss due to patient work absenteeism was highest in other medical conditions, followed by CVD, COPD, other respiratory disease, lung cancer, other cancer and upper aerodigestive cancer. The highest productivity loss due to work absenteeism of a caregiver was found in other medical conditions, followed by COPD, CVD, other respiratory diseases, lung cancer, other cancer and upper aerodigestive cancer.

Table 3

Cost attributable to smoking by disease and cost category (million THB)

For a total population of 63.5 million in 2009, we estimated smoking attributed cost per capita in table 4. Total smoking-related cost per capita was about 1180.07 (95% CI 1175.37 to 1184.77) THB. The cost per capita was highest in CVD, approximately 328.71 (95% CI 326.26 to 331.16) THB, followed by other medical causes, lung cancer, COPD, other cancer, other respiratory diseases and upper aerodigestive cancer, respectively. For direct medical cost, other medical cost was the highest, followed by CVD.

Table 4

Cost per capita attributable to each smoking-related disease (THB)

As for individual victims, it could well mean that a male suffering from lung cancer at 45 years of age is expected to have an annual cost of 51 925 THB, which in turn amounts to 7519 THB spent, on average, on OP visits, 44 406 THB on IP care, 1271 THB spent on annual transportation and out-of-pocket payment, and 3250 THB of lost wages and productivity. If this person dies, his lifetime loss of income would be 2.13 million THB.

Discussion

The total costs of smoking in Thailand for 2009 amounted to an estimated 74.9 billion THB (95% CI 74.6 to 75.2), or approximately 0.78% of the country's GDP. The cost contributed to 18% of total health expenditure, approximately one-fourth of total public health expenditure and 5% over the MOPH's budget. The direct medical cost shares about 4% of total public health expenditure.

Of the total economic costs attributable to tobacco, productivity loss constituted the highest loss while the medical cost contributed 15% of the total. The direct medical cost we calculated for Thailand was much less than that of California, where a study in the state reported the direct medical cost to be 54% of the total cost. The lower share of direct medical care cost in our study may be due to several factors, including unit cost of care, access and quality of medical care, and the fact that the California study did not separate indirect medical cost.

Among developing countries, the proportions of medical costs to the total costs are quite varied. In China, direct medical cost due to smoking contributed 21–25% of the total cost.21 The healthcare cost in Vietnam was much higher, about 51.5%.22 Such differences are likely to be a result of differences in healthcare costing approaches. The costing approach in Vietnam was based on a cross-sectional survey of healthcare while our study used a large claim database. In addition, the number of deaths in developing countries may be underestimated due to poor quality of cause-of-death information.

However, our estimate of direct cost in relation to national health expenditure is fairly similar to other developing countries. The direct cost of smoking in our study accounted for 3.4% of total health expenditure compared with 3.0% in China,21 3.1% in Taiwan23 and 4.7% in India.5

Since the share of medical cost in our study was comparatively lower than in other studies, indirect cost due to premature death shared a higher proportion of the total than those in other studies.3–6 Nevertheless, the direct medical cost attributed to smoking-related diseases was about 16% of the MOPH's total budget. It is notable that the per capita cost of smoking of 1180 THB is equivalent to more than half of the per capita budget for the national health security scheme (2202 THB), which provides services for two-thirds of the population.

We found that, when excluding other groups of diseases, direct medical cost attributable to smoking was highest for CVD, followed by COPD and lung cancer. These results are similar to a study in the USA, which evaluated four smoking-related diseases whereby total medical expenditure was highest for ischaemic heart disease, and second highest for COPD.24 Likewise, we found that cost of treatment for lung cancer per claim was the most expensive.

Our estimate of the annual economic loss attributable to smoking is different from other Thai studies for several reasons. Leartsakulpanitch et al10 estimated out-of-pocket medical cost for lung cancer, COPD and CHDs to be about 0.48% of GDP. Their estimates were based on Pongpanich's11 medical cost per patient per year. Data from that study were collected from each of the biggest hospitals (number of beds), both government and private, in each region, and extrapolated to the national prevalence of the three diseases. Our study estimated costs incurred from hospitalisation and hospital visits from the charge data from medical insurance. Although their study did not include indirect costs from premature death, as was included in our study, they estimated the number of patients from prevalence cases regardless of their healthcare utilisation, while we captured only patients utilising healthcare services. In addition, the costing approaches in the Pongpanich study used a survey that likely produced higher cost than an approach using a claim database.

Difference in medical cost estimation is another reason for the variation in results. Wattanasap25 estimated the medical cost of patients with COPD by severity of disease, whereas Patumanond et al26 assumed that the medical cost was constant for all stages of illness. As a result, the medical cost of COPD by Wattanasap is 16 388, while Patumanond et al's is 6457 THB per person, per year. Our approach, on the other hand, is derived from the claim database that covered most healthcare providers. Although using claim data tends to underestimate the real cost, it has some advantages. First, it consumes fewer resources than a costing survey. Second, it can provide comparable cost estimates for other diseases of interest.

Santisart12 estimated direct medical care cost and productivity loss from absenteeism for COPD and lung cancer in 1999 to be 236 million THB, approximately 0.1% of total health expenditure in that year. As more data becomes available, our study can include up to seven diseases and productivity loss due to mortality, which is a major contributor to damage of the economy.

Although our estimates covered almost all types of costs of illness, ranging from direct and indirect medical care cost, and indirect cost due to workdays loss and mortality, they were still conservative. We did not include cost due to quality of work loss owing to an illness (presenteeism), which is usually higher than cost due to absenteeism. Our medical cost estimate, mentioned earlier, did not take into account the cost of long-term care at patients’ own homes. Additionally, we assumed a proportion of incompleteness of the medical database due to absence of the data, particularly in private settings.

In addition, we were aware of the assumption on the estimation of productivity loss from caregivers. Caregivers of patients aged over 60 years may be unemployed and aged over 60 years as well (as are their spouses), or they may be males or females of any age, or the patients may not need caregivers. We do not have information about the caregivers nor the amount of time the caregivers spend with each patient, particularly if the patient needs more time to be cared for at home than at the hospital. This should be refined in the future once information becomes available.

On coverage of diseases related to smoking, we were limited to seven disease categories. Although the list covered major diseases, there are other conditions not included in our studies. The US Surgeon General's Report27 identifies a substantial number of diseases found to be caused by smoking, including the reproductive effect of low birth weight and a higher risk of mortality for smokers in accidents,28 however, neither of these conditions was included in our study.

This study gives the most recent estimates of economic loss due to illnesses attributable to smoking in Thailand. We limited our estimates to active smoking because the Thai BOD estimates did not include secondhand smoking estimates. In Hong Kong, it was estimated that passive smoking accounted for 23% of total economic loss in 1998.29 Had we taken secondhand smoking into account, the total cost would be considerably higher than what is presented here.

In 2009, tobacco product manufacturing accounted for 47.88 billion THB,30 or 0.50% of GDP. However, tobacco-related illnesses could cost the economy 74.88 billion THB (74.59 to 75.18) or 0.78% of GDP, a total of 0.28% more than tobacco manufacturing contributed. Given that our estimate fell in the range of 0.17–1.19% of GDP found in other studies,31 ,32 better coverage of healthcare data would improve the accuracy of the results. However, even when underestimating the real cost due to conservative coverage as discussed above, the results provide strong evidence for policymakers.

Cost of illness studies translate adverse effects of disease into monetary terms.33 Our estimate of healthcare costs was about 4% of the public health expenditure. This can inform decision-makers as to how much they spend each year on providing healthcare services to patients suffering from preventable smoking-related illnesses. These costs could eventually be saved and used for other needs by preventing people from smoking. As for the government, smoking has caused a loss of 0.78% of GDP, compared with the annual GDP growth during 2010–2013, which ranged between 0.1% and 7.8%,34 such loss should be seriously considered. While the tobacco industry generates considerable revenues for the government, our study shows that its business costs more to the country's economy than it produces.

What this paper adds

  • Smoking, a leading cause of illness and death, can cause considerable burden on healthcare and national productivity.

  • There is no national and comprehensive estimation of the cost of smoking-related illnesses in Thailand.

  • Smoking in Thailand is a loss for the economy, creating costs of 74.88 billion THB (74.59 to 75.18) (US$2.18, US$2.17 to US$2.19 billion), approximately 0.78% of GDP, while the tobacco industry only produced 0.50% of GDP.

Acknowledgments

The authors thank Professor Virasakdi Chongsuvivatwong for his guidance, and Candyce Silva and Wit Wichaidit for their comments and assistance.

References

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

Footnotes

  • Contributors KB contributed to the design and conceptual approach, data analysis and drafting the initial manuscript. SA and NM carried out the analysis and data compilation. NM assisted in the analysis and coordination of the study. All the authors jointly contributed to the revision of the manuscript and discussion of the results. KS oversaw the study, guided the manuscript drafting and finalised the manuscript. All the authors approved the final version of the manuscript.

  • Funding This work was supported by the Thai Health Promotion Foundation (ThaiHealth 54-00-0820) and National Health Security Office 56A00193/2555.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement Access to the additional data from the study are available on request to KB.

  • i 100 THB=US$2.91 in 2009 (Bank of Thailand reference rate).

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.