Content uploaded by Syed Naimul Wadood
Author content
All content in this area was uploaded by Syed Naimul Wadood on Jun 26, 2020
Content may be subject to copyright.
Bangladesh Cancer Society (Dhaka, Bangladesh)
University of Dhaka, Department of Economics (Dhaka, Bangladesh)
Programme for Research, Advocacy and Capacity Building on Tobacco
Taxation (PROACTT), a collaboration between the Cancer Research United
Kingdom and the American Cancer Society, USA
The Economic Cost of
Tobacco Use in Bangladesh:
A Health Cost Approach
1
TOBACCO USE IS THE SINGLE MOST PREVENTABLE CAUSE OF
DEATH GLOBALLY. IT IS ONE OF THE GREATEST RISK FACTORS
FOR NONCOMMUNICABLE DISEASES INCLUDING CANCERS,
CARDIOVASCULAR AND RESPIRATORY DISEASES. GLOBALLY
SMOKING, CHEWING TOBACCO AND EXPOSURE TO SECONDHAND
SMOKE TOGETHER WERE RESPONSIBLE FOR THE LOSS OF MORE
THAN 8 MILLION LIVES AND 213 MILLION DISABILITY-ADJUSTED
LIFE YEARS (DALYS) IN 2017. BETWEEN 2007 AND 2017, THE
NUMBER OF TOBACCO-ATTRIBUTABLE DEATHS INCREASED
BY 0.8 MILLION AND DALYS INCREASED BY 13 MILLION. IN THE
20TH CENTURY, TOBACCO IS KNOWN TO HAVE KILLED ABOUT
100 MILLION PEOPLE, AND IT IS PROJECTED TO KILL 1 BILLION
PEOPLE IN THE 21ST CENTURY. THERE IS URGENT NEED TO
REVERSE THIS TREND BEFORE MORE LIVES AND HEALTHY LIFE
YEARS ARE LOST.
The global economic cost of diseases attributable to smoking,
including health expenditures and productivity losses, was estimated
at $1,436 billion in 2012, roughly equivalent to 1.8 percent of the
world’s annual Gross Domestic Product. Globally there are 1.1 billion
tobacco smokers aged 15 years or older, around 80 percent of whom
live in low- and middle-income countries (LMICs). As a result, the
enormous global health and economic burden of tobacco use will
be increasingly borne by these LMICs. The present study nds that
Bangladesh, a lower-middle income country, incurs substantial and
increasing tobacco-attributable economic costs consistent with the
global evidence.
According to the Global Adult Tobacco Survey (GATS), the prevalence
of tobacco use (including smoking and smokeless tobacco) among
adults 15 and older decreased from 43.3 percent in 2009 to 35.3
percent in 2017 overall. In males, there was a reduction from 58.0
to 46.0 percent; among females, a drop from 28.7 to 25.2 percent.
The youth (13-15) prevalence of tobacco use was 6.9 percent overall
(9.2 percent among boys; 2.8 percent among girls) in 2013. That
represented no signicant change in the prevalence overall and
among boys, and a decrease among girls from 5.1 percent since 2007.
Despite some positive changes in tobacco use prevalence among
adult men and women and girls in the recent past, tobacco use
remains one of the major risk factors for noncommunicable diseases
in Bangladesh, one that can be modied to prevent a substantial
number of deaths and disabilities.
The primary objective of the present study was to estimate the direct
and indirect health costs attributable to tobacco use and harm
from exposure to secondhand smoke. It also aimed at providing
a comparison between the current estimates and those obtained
from a 2004 WHO study on the impact of tobacco related illnesses
in Bangladesh. The ultimate objective was to compare the costs of
tobacco use with tobacco’s so-called economic benet, with a view
to shaping national tobacco control policy.
Tobacco control has been an explicit government priority, but
adoption and application of a fully functional and eective measure
have been constrained by the powerful lobby of the nancially
strong tobacco companies. The estimates of the economic burden
of tobacco use in this study provide the basis for a comprehensive
assessment of the social and economic impact of tobacco use in
Bangladesh and compelling evidence to support a strong public
health policy intervention. This would also help raise people’s
awareness about the hard reality tobacco users face and build public
opinion in support of tobacco control nationwide.
This study estimated that tobacco use caused nearly 126,000 deaths
accounting for 13.5 percent of deaths from any cause in Bangladesh
in 2018. In addition, the study revealed that approximately 1.5 million
adults were suering from diseases attributable to tobacco use and
nearly 61,000 children were suering from diseases due to exposure
to secondhand smoke. Tobacco-induced deaths and diseases alone
cost the economy of Bangladesh around BDT 305.6 billion ($3.61
billion) a year, which was equivalent to 1.4 percent of its national GDP
in the year 2017-18.
The total economic contribution of the tobacco sector (in terms
of household nal consumption expenditure, private and public
domestic investment and net export) to the GDP in Bangladesh was
estimated at BDT 229.1 in 2016-17 scal year in 2018 prices. This is
BDT 76.5 billion short of the estimated total cost of tobacco, BDT
305.6 billion. Tobacco thus appears to be causing net loss to the
economy of Bangladesh.
These ndings establish that tobacco use imposes a signicant
nancial burden on tobacco users, their families, and on the nation.
The breakdown of the total cost estimate of BDT 305.6 billion shows
that the direct healthcare costs attributable to tobacco use amounted
to BDT 83.9 billion annually, 76 percent of which was paid by tobacco
users’ households. The remaining 24 percent was nanced through
public health sector budget representing nearly 9 percent of total
government health expenditure in scal year 2018-19. The indirect
cost given by the annual productivity loss, due to morbidity and
premature mortality from tobacco-related diseases, was estimated to
be BDT 221.7 billion.
Generally, studies that measure the economic cost of tobacco use
apply diverse methods, and the estimates are not readily comparable
across populations, time and studies. The present study shows that
the annual estimate of total economic cost of tobacco in Bangladesh
more than doubled since 2004. By maintaining the cost-of-illness
approach followed in the 2004 WHO study in Bangladesh, this study
provided a consistent evidence base for comparison of the economic
costs of tobacco over time, measurement of progress in curbing the
tobacco epidemic and timely intervention to accelerate the progress
in tobacco control.
EXECUTIVE SUMMARY
2
Though this study is comprehensive insofar as including direct
medical costs and indirect productivity costs, it still misses a
signicant number of additional costs, and therefore remains an
underestimate of the total costs of the tobacco burden. We recognize
that this study has not accounted for the substantial costs of the
environmental and health damages from tobacco cultivation, loss of
food security caused by the use of scarce land resources for tobacco
growing, smoking-related re hazards, environmental pollution
from littering of cigarette butts, and so on. Had these costs been
estimated, the net loss from tobacco would have been even larger.
These costs can be measured in future research endeavors. Perhaps
more importantly from a human perspective, we will never be able
to gauge the pain and suering of tobacco victims and their families.
Moreover, the spending on tobacco and on health care attributable
to tobacco diverts resources away from necessities. The crowding-
out eect of tobacco consumption can lead to displacement of
basic needs among lower-income households and contribute to
impoverishment and overall economic and health inequity.
The need to reverse the adverse socioeconomic, environmental and
health consequences of tobacco use on individuals and on society is
urgent. The vision of Tobacco-Free Bangladesh by 2040 underscores
this need. The path to a tobacco-free state is, however, arduous and
can be reached only with aggressive and eective tobacco control
measures that:
• Raise tobacco taxes and prices
• Monitor tobacco use and prevention policies
• Protect people from tobacco smoke with smoke-free laws
• Oer help for tobacco use cessation
• Warn about the dangers with graphic warning labels on tobacco
packages
• Enforce bans on tobacco advertising, promotion and
sponsorship
• Tax tobacco farming land as industrial land to discourage
tobacco cultivation
• Provide incentive to tobacco farmers to shi to alternative and
more viable livelihoods
• Enforce ban of tobacco sales to minors.
At present, 35.3 percent of Bangladeshis 15 and older (an estimated
37.7 million adults) smoke and/or use smokeless tobacco. If we
start the clock now, 1.8 million tobacco users will have to quit
tobacco use every year to make the country tobacco-free by 2040.
A comprehensive tobacco tax reform with simplication of tax
structure, plus increases in tobacco taxes and prices can make this
target feasible. Yet there will be new tobacco users, unless we can
stop initiation of tobacco use among the youth completely. Global
evidence shows that tax and price increases are particularly eective
in deterring youth from starting to smoke. A tobacco tax increase also
can reduce tobacco consumption faster among the poor, who are
usually more price sensitive. The eect of tax and price increases in
reducing tobacco consumption can be stronger, if tax policy can be
aligned and combined with other measures.
Estimates of the tobacco-attributable health care costs and
productivity loss at the population level help to understand the
economic impact of tobacco use and provide the evidence base for
policymakers to implement comprehensive tobacco control policies
to curb the epidemic. The progress in tobacco control policies in
Bangladesh since the ratication of the WHO Framework Convention
on Tobacco Control (WHO FCTC) in 2004, followed by the passage
of the Tobacco Control Act in 2005 and the amendment to the
Tobacco Control Act in 2013, has been limited. Despite the reduction
of tobacco use, Bangladesh will not meet the target of a Tobacco-
Free Bangladesh by 2040 envisioned by the prime minister. It will
take much stronger tobacco control measures fortied with “best
practices” and stricter compliance with the guidelines of the articles
under the WHO FCTC, especially targeted at youth, who constitute the
generation of potential future smokers.
THE GOALS OF TOBACCO CONTROL ARE INTERTWINED WITH
THE UNITED NATIONS 2030 SUSTAINABLE DEVELOPMENT
GOALS TO ERADICATE EXTREME INCOME POVERTY, REDUCE
DEATHS FROM NONCOMMUNICABLE DISEASES BY ONE-
THIRD, AND ACHIEVE UNIVERSAL HEALTH COVERAGE TO HELP
PROTECT AGAINST IMPOVERISHMENT CAUSED BY ILLNESSES.
ELIMINATION OF TOBACCO USE CAN PREVENT THE DEATHS AND
DISEASES ATTRIBUTABLE TO TOBACCO USE AND SECONDHAND
SMOKE EXPOSURE ALTOGETHER, THEREBY CONTRIBUTING
SIGNIFICANTLY TO THE SUSTAINABLE DEVELOPMENT GOALS IN
BANGLADESH BY 2030.
3
Team Leader
Dr. Golam Mohiuddin Faruque, Professor and Former Head,
Department of Radiotherapy, Dhaka Medical College; Project
Director, Bangladesh Cancer Society
Other Members
Syed N Wadood, PhD, Associate Professor, Department of Economics,
University of Dhaka
Maruf Ahmed, Research Associate, Bangladesh Institute of
Development Studies
Rehana Parven, Joint Secretary, SEIP Project, Ministry of Finance,
Government of the People’s Republic of Bangladesh
lekharul Huq, PhD, Assistant Professor, Department of Economics,
East West University
Dr. Sohel Reza Chowdhury, Professor, Department of Epidemiology
and Research, National Heart Foundation Hospital & Research
Institute
Medical Team Members
Prof. Qazi Mushtaq Hussain, Head, Department of Radiotherapy,
Dhaka Medical College Hospital
Dr. Asish Kumar Shaha, Director, MIS, Directorate General of Health
Services
Dr. Afsana Habib Sheuly, Helen Keller International
Dr. Aparna Biswas, Medical Oicer, Coordination & Support Center,
Directorate General Health Services
Dr. Hasan Alif, Assistant Chief, MIS, Directorate General Health
Services
Dr. Ali Al Mahdi, Assistant Registrar, Uttara Adhunik Medical College
Hospital
Dr. Rafeea Nuhas, Lecturer, Anwar Khan Modern Medical College
Dr. Md. Azizur Rahman, Registrar, National Institute of Cancer
Research Hospital
Dr. Abdullah Al Noman Pavel, Medical Oicer, Radiotherapy, Dhaka
Medical College Hospital
Mr. Aminul Islam, Program Oicer, National Tobacco Control Cell
Research Assistants
Nahian Azad, MSS graduate, Department of Economics, University of
Dhaka
Biplob Hossain, MSS graduate, Department of Economics, University
of Dhaka
Masud Karim, data analyst, PROACTT Survey, University of Dhaka
Md. Golam Ahad, eld manager, PROACTT Survey
Resource Persons
AKM Ghulam Hussain, PhD, Professor, Department of Economics,
University of Dhaka
Md. Ashadul Islam, Secretary, Ministry of Health and Family Welfare,
Government of the People’s Republic of Bangladesh
Fowzul Azim, Chief Research Oicer (District Judge), Bangladesh Law
Commission
Technical Advisors
Nigar Nargis, PhD, Scientic Director, Economic and Health Policy
Research, American Cancer Society, USA
Jerey Drope, PhD, Scientic Vice President, Economic and Health
Policy Research, American Cancer Society, USA
Gregg Haiey, Director, Federal Relations, American Cancer Society
Cancer Action Network, USA
Additionally, 38 eld oicers were trained and sent to selected
locations throughout the country to conduct the primary survey
component.
The study team received technical support from the World Health
Organization Country Oice, Bangladesh, and are especially
thankful to the following oicials
Dr. Tara M Kessaram, Medical Oicer, Noncommunicable Diseases
Dr. M Mostafa Zaman, Advisor, Research and Publication
Dr. Syed Mahfuzul Huq, National Professional Oicer,
Noncommunicable Diseases
THE STUDY TEAM
4
FROM THE TEAM LEADER
The PROACTT Bangladesh Project has been a great journey for all of us. Our study team, medical team and our field research oicers visited
every district in the country to make this project a success. I appreciate all their eorts and the diiculties that they faced in the important work of
collecting evidence on the human and socioeconomic costs of tobacco use in Bangladesh.
The study grew out of a workshop in July 2017 at the University of Dhaka, followed by a formal agreement between the Bangladesh Cancer Society
and the American Cancer Society (ACS). The study was jointly supported by American Cancer Society (ACS) and Cancer Research United Kingdom
(CRUK) under the Programme for Research, Advocacy and Capacity Building on Tobacco Taxation (PROACTT). A multi-stage clustered survey was
designed for a nationally representative sample of 10,000 households. Aer developing survey questionnaires for disease prole and health costs,
38 inter viewers collected data from the sample households. The eld work for the 18-month-long study was conducted from January to April
2018. Aer data collection, processing and analysis, we released the key ndings of the study on 23 February 2019 in the presence of the State
Minister of Health and the Secretary of Health of the Government of Bangladesh, distinguished health professionals, civil society organizations,
tobacco control advocates and representatives of ACS and CRUK.
I am delighted to submit the full report. It is my sincerest hope that this study will contribute to the advancement of the tobacco control movement
in Bangladesh.
Cordially,
Professor Dr. Golam Mohiuddin Faruque
5
* Suggested citation: Faruque GM, Ahmed M, Huq I, Parven R, Wadood SN, Chowdhury SR, Hussain AKM G,
Haiey G, Drope J, Nargis N. The economic cost of tobacco use in Bangladesh: A health cost approach. Bangladesh
Cancer Society. March 1, 2020.
ACKNOWLEDGMENTS
The study team acknowledges and appreciates following institutions for their contributions
Technical support:
American Cancer Society
Bangladesh Bureau of Statistics
Cancer Research United Kingdom
World Health Organization Country Oce, Bangladesh
Funding:
Programme for Research, Advocacy and Capacity Building on Tobacco Taxation (PROACTT ),
a collaboration between Cancer Research United Kingdom and American Cancer Society, USA
Ethical Clearance:
Bangladesh Medical Research Council (BMRC).
Design and Graphics:
Radish Lab
Nancy Inglis-Wesby (American Cancer Society)
6
TABLE OF CONTENTS
2.1. Cost of Illness (COI) Approach
2.2. Data Sources
2.2.1 Household Survey
2.2.2 Secondary Data
2.3. Measures
Study Population
Status of tobacco use
Exposure to secondhand smoke
Prevalence of Tobacco-related Diseases
Mortality from Tobacco-related Diseases
Risk Ratio (RR)
Population-attributable Risk (PAR)
Private Health Expenditures
Public Health Expenditures
Costs of Morbidity
Costs of Mortality
Costs of Secondhand Smoke Exposure
10
11
11
16
16
16
16
17
17
17
18
18
19
20
1
3
4
5
8
SECTION 1: INTRODUCTION
SECTION 2: METHODS AND ANALYTICAL FRAMEWORK
9
Executive Summary
The Study Team
From the Team Leader
Acknowledgements
List of Tables
List of Figures
7
21
23
24
26
27
28
28
30
31
32
35
37
38
42
44
45
46
47
59
88
3.1. Prevalence of tobacco-related diseases
3.2. Prevalence of tobacco use (adults aged 30 and older)
3.3. Excess risk of tobacco-related diseases attributable to tobacco use
and household exposure to secondhand smoke
3.4. National level estimates of the number of patients suering from
tobacco-attributable illnesses
3.5. National level estimates of tobacco-attributable mortality
3.6 Direct costs of tobacco-attributable Illnesses
3.6.1 Private health expenditures
3.6.2 Public health expenditures
3.6.3 Total health expenditure
3.7. Indirect costs of tobacco-attributable illnesses
3.7.1 Costs of morbidity
3.7.2 Costs of mortality
3.8. Comparison of the economic costs and benets of tobacco
3.9. Tobacco-attributable costs: 2004 and 2018
SECTION 3: FINDINGS
SECTION 4: DISCUSSION AND CONCLUSION
References
Glossary
List of Abbreviations
Appendices
Appendix 1. Household Disease Prole Questionnaire
Appendix 2. Household Health Cost Questionnaire
Appendix 3. Map of Bangladesh with All 64 Districts
8
LIST OF TABLES
Table 2.1
Geographic Distribution of Sample Households 12
Table 2.2
Sample characteristics of patients (30 and older) with proven medical records
of tobacco-related diseases 12
Table 3.2.1
Prevalence of tobacco use among adults 30 and older 23
Table 3.2.2
Prevalence of current tobacco use by product type among adults 30 and older 23
Table 3.3.1
Risk Ratio (RR) and Population Attributable Risk (PAR) due to tobacco use among 24
adults 30 and older
Table 3.3.2
Risk Ratio (RR) and Population Attributable Risk (PAR) due to household exposure
to secondhand smoke among children younger than 15 25
Table 3.4.1
Total number of patients suering from tobacco-related illnesses in Bangladesh, 2018. 26
Table 3.4.2
Total number of patients suering from tobacco-related illnesses by disease category for 26
age group 30 and older in Bangladesh, 2018
Table 3.6.1
Tobacco-attributable out-of-pocket health expenditure for inpatient care for patients 28
30 and older
Table 3.6.2
Tobacco-attributable out-of-pocket health expenditure for outpatient care for patients 29
30 and older
Table 3.6.3
Direct costs of tobacco-related illnesses in Bangladesh, 2018
31
Table 3.7.1 Value of time of patients 30 and older for attending health care services 32
Table 3.7.2 Household productivity loss of non-employed patients 30 and older 34
Table 3.7.3 Productivity loss due to premature mortality attributable to tobacco use
for the age group 30 and older 35
Table 3.7.4 Productivity loss due to premature mortality attributable to
secondhand smoke exposure 36
Table 3.8.1 The costs (in billion BDT) of tobacco-attributable illnesses in Bangladesh, 2018 37
Table 3.9.1 Comparison of the tobacco-attributable cost estimates for Bangladesh between 38
2004 and 2018 (in 2018 prices)
Figure 3.1.1 Prevalence of tobacco-related diseases among adults 30 and older 21
Figure 3.1.2 Prevalence of tobacco-related diseases among children younger than 15 22
Figure 3.3.1 Prevalence of tobacco-related diseases by tobacco use status of
adults 30 and older 24
LIST OF FIGURES
9
Tobacco use is the single most preventable cause of death globally.
It is one of the greatest risk factors for noncommunicable diseases
(NCDs) including cancers, cardiovascular and respiratory diseases.
Globally smoking, chewing tobacco and exposure to secondhand
smoke together were responsible for the loss of more than 8 million
lives and 213 million disability-adjusted life years (DALYs) in 2017
(Global Burden of Disease, Injuries, and Risk Factors Study, 2017).
Further, the deaths and disabilities attributable to tobacco are rising
— between 2007 and 2017, the number of tobacco-attributable
deaths increased by 0.8 million and DALYs increased by 13 million
(Global Burden of Disease, Injuries, and Risk Factors Study, 2017). In
the 20th century, tobacco is known to have killed about 100 million
people. It is projected to kill 1 billion people in the 21st century (World
Health Organization, 2008) (World Health Organization, 2017). There
is urgent need to reverse this trend before more lives and healthy life
years are lost.
The global economic cost of smoking-attributable diseases from
health expenditures and productivity losses was estimated at $ 1,436
billion in 2012, roughly equivalent to 1.8 percent of the world’s annual
Gross Domestic Product (Goodchild et al., 2018). Globally there are
1.1 billion tobacco smokers 15 years or older, around 80 percent of
whom live in low and middle-income countries (LMICs) (U.S. National
Cancer Institute; World Health Organization, 2016). As a result, the
enormous global health and economic burden of tobacco use will
be increasingly borne by these LMICs. The ndings from the present
study reveal that Bangladesh, a lower-middle income country, incurs
substantial and increasing tobacco-attributable economic costs
consistent with the global evidence.
According to the Global Adult Tobacco Survey (GATS), the prevalence
of tobacco use (including smoking and smokeless tobacco) among
adults 15 and older decreased from 43.3 percent in 2009 to 35.3
percent in 2017 overall (from 58.0 to 46.0 percent among males; from
28.7 to 25.2 percent among females (Bangladesh Bureau of Statistics,
2018)). The youth (13-15) prevalence of tobacco use was 6.9 percent
overall (9.2 percent among boys; 2.8 percent among girls) in 2013.
That represented no signicant change in the prevalence overall and
among boys, and a decrease among girls from 5.1 percent since 2007
(World Health Organization, Regional Oice for South-East Asia, 2015).
Despite some positive changes in tobacco use prevalence among
adult men and women and young girls in the recent past, tobacco
use remains one of the major risk factors for NCDs in Bangladesh, one
that can be modied to prevent a substantial number of deaths and
disabilities.
The primary objective of the present study was to estimate the direct
and indirect health costs attributable to tobacco use and harm
from exposure to secondhand smoke. It also aimed at providing
a comparison between the current estimates and those obtained
from 2004 WHO study on the impact of tobacco related illnesses in
Bangladesh (World Health Organization, Regional Oice for South-
East Asia, 2007). The ultimate objective was to compare the costs with
the so-called economic benet of tobacco, with a goal of setting out
policy implications for national tobacco control.
Tobacco control has been an explicit government priority, but
adoption and application of a fully functional and eective measure
from the part of the government have been systematically constrained
by the powerful lobby from the nancially strong tobacco companies.
The estimates of the economic burden of tobacco use made available
by this study provide the basis for a comprehensive assessment of
the overall social and economic impact of tobacco use in Bangladesh
and compelling evidence for strong public health policy intervention
by the government to curb the tobacco epidemic despite these
inuential lobbyists. This would also help raise people’s awareness
about the hard reality tobacco users face and build public opinion in
support of tobacco control nationwide.
1. INTRODUCTION
11
SECTION 2
METHODS AND
ANALYTICAL
FRAMEWORK
10
2.1 COST OF ILLNESS COI
APPROACH
This study followed the cost-of-illness approach to estimate the
economic cost of the adverse health eects of tobacco use (Rice,
Hodgson, Sinsheimer, Browner, & Kopstein, 1986). In this study, we
estimated the economic cost for seven tobacco-related diseases,
namely, ischemic heart disease, stroke, chronic obstructive
pulmonary disease (COPD), pulmonary tuberculosis, lung cancer,
laryngeal cancer, and oral cancer. There is signicant association
between tobacco use and the prevalence of or the risk of mortality
caused by these diseases. In addition, for capturing the eects of
exposure to secondhand smoke, the study considered additional
health conditions such as asthma, autism, lower respiratory infection,
low birth weight of the newborn, and sudden infant death syndrome
among the nonsmoker residents in the smokers’ households.
We used a prevalence-based, disease-specic approach to measure
the annual cost of tobacco-related illnesses and deaths caused by
both current and past tobacco use, both smoking and smokeless
tobacco. The economic cost of tobacco-related illnesses includes the
following components:
1. Private expenditure (out-of-pocket or insurance covered)
of patients on medical care for treating tobacco-related
diseases of the tobacco users and those exposed to
secondhand smoke: It includes household expenses on
inpatient hospitalizations, outpatient visits, hospital admission
costs, doctor fees, medicines and nutritious food costs,
diagnostic tests, accommodation costs, transportation to
health centers and hired caregivers’ costs;
2. Cost of the public health care system: It includes direct
government expenditure on health care ser vices oered to
patients suering from tobacco-attributable diseases in public
health care facilities, covering outpatient and inpatient care,
medical rehabilitation, publicly procured pharmaceuticals,
medical aids, emergency medical services, nursing, medical
diagnosis and procedures; and
3. Loss of productivity and income: It includes the eect of
disability and premature mortality on individual and household
level productivity and income due to tobacco-related illnesses.
The rst two components compose the “direct costs” to the patients
and the health care system. The third component captures the
“indirect costs” that the patients and their families incur because of
illnesses caused by tobacco use.
The cost-of-illness approach entails the determination of excess cost
that can be attributed to tobacco use. In the rst step, we determined
the relative risks (RR) based on the excess risk of the prevalence of
or mortality from these diseases among the tobacco users and the
nonsmokers exposed to secondhand smoke. More formally, RR is
the ratio of the probability of the outcome (prevalence of disease
or mortality from the disease) in the exposed group (tobacco users
or those exposed to secondhand smoke) to the probability of the
outcome in the unexposed group (non-tobacco users or those not
exposed to secondhand smoke).
The RR is required to identify the fraction of the total cost-of-illness
that can be attributed to tobacco use known as the population
attributable risk (PAR), as multiple factors including tobacco can
contribute to the disease prevalence and mortality. PAR is given by
the formula:
where P is the tobacco use prevalence in the population.
11
2.2 DATA SOURCES
2.2.1 Household Survey
SAMPLING DESIGN
The primary data on diseases of all household members
were collected during January-April 2018 from a nationally
representative survey of 10,000 households selected using
a multi-step clustered sampling design. The sample size of
households was pre-determined to ensure the availability of
enough patients suering from tobacco-related diseases to
successfully apply the prevalence-based disease-specic cost-
of-illness approach. Based on previous information on average
household size and age distribution from the h Population
and Housing Census of Bangladesh 2011, and the prevalence
of tobacco-related diseases observed in the WHO study in
2004, it was estimated that to obtain a sample of 2,500 patients
with a tobacco-related disease, a representative sample of at
least 10,000 households would be required.
The sampling frame used for the household survey was based
on the Population and Housing Census of Bangladesh 2011
and was obtained from the Bangladesh Bureau of Statistics.
At the time of the census, there were six divisions (Barisal,
Chittagong, Dhaka, Khulna, Rajshahi, and Sylhet), which are
the largest administrative units in Bangladesh. By 2018, when
the survey for the present study was carried out, two more
divisions had been created by dividing Dhaka division into
Dhaka and Mymensingh and Rajshahi division into Rajshahi
and Rangpur. For the purpose of this study, we isolated the
population statistics for the two new divisions based on the
population data from the districts (Zilla) covered within the
new divisions.
The sampling of households was done in three successive
stages:
1. In line with the latest Global Adult Tobacco Survey
conducted in Bangladesh in 2017, which is a global
surveillance system for systematically monitoring adult
tobacco use, a total of 496 Primary Sampling Units
(PSUs), Mauza in rural and Mohalla in urban areas, were
selected at the rst stage with probability proportional
to population size. The population size was measured by
the total number of households in each PSU in the 2011
Census. The 496 PSUs were allocated evenly across the
eight divisions, with 62 PSUs per division. Within each
division, 62 PSUs were distributed among all districts
(Zilla) with probability proportional to the population size
of each district.
2. At the second stage, one Secondary Sampling Unit (SSU),
village in rural areas and para in urban areas, was selected
randomly from each PSU.
3. At the third stage, households were selected randomly
(every 6th or 7th household based on the conditional
probability of selecting one household) from each SSU. To
represent all 64 districts in the country, the pre-determined
sample size of 10,000 households was distributed among
all districts proportionate to the number of households
in each district. To avoid losing this pre-determined
sample size due to nonresponse or missing information,
119 additional households were interviewed resulting
in 10,119 total number of households in the survey.
The sample size determined for each district was then
distributed among the selected SSUs proportionate to
the population size of each SSU. From each household,
the head of the household or other responsible adult
responded to the survey questions to provide household-
specic information as well as individual-specic
information on all residents of the household. In case of
non-response, a replacement household was selected
randomly to meet the target of pre-specied sample size.
Table 2.1 reports the geographic distribution of 10,119 sample
households. Information on the symptoms and diagnosed
cased of tobacco-related illnesses were collected for all
individuals 30 years and older and children below age 15
from these households. The tobacco use status of all adults
were collected at this stage as well. Based on the disease
prole of 10,119 sample households, 2,600 households were
identied with at least one member diagnosed with one of the
selected major tobacco related diseases. Among these 2,600
households, 998 households had conrmed cases of the said
diseases, with supporting medical records. The distribution of
these households by division is shown in the nal column of
Table 2.1.
The 998 households with conrmed cases received a health
cost questionnaire to collect detailed information on the use of
health care services and related expenses, employment status
and earnings of the patients, along with basic sociodemographic
and economic characteristics of households. In this round of
survey, age cohorts included adults 30 and older for diseases
attributable to tobacco use and children younger than 15
for diseases attributable to exposure to secondhand smoke.
There were 39 valid cases of tobacco-related diseases among
children younger than 15 in the sample households. The
sample characteristics of the individuals aged 30 and older are
presented in Table 2.2.
12
Table 2.1 Geographic Distribution of Sample Households
Division District (Zilla)
Primary Sampling
Unit (PSU)
(Mauza/Mohalla)
Secondary
Sampling Unit
(SSU) (village/
para)
Number
of sample
households
Number
of sample
households with
conrmed cases
of tobacco-related
diseases
Barisal 6 62 62 1,238 90
Chittagong 11 62 62 1,343 88
Dhaka 13 62 62 1,392 124
Khulna 10 62 62 1,191 142
Mymensingh 4 62 62 1,220 159
Rajshahi 8 62 62 1,273 128
Rangpur 8 62 62 1,242 150
Sylhet 4 62 62 1,220 116
Total 64 496 496 10,119 998
The 998 households with conrmed cases received a health cost questionnaire to collect detailed information on the use of health care
services and related expenses, employment status and earnings of the patients, along with basic sociodemographic and economic
characteristics of households. In this round of survey, age cohorts included adults 30 and older for diseases attributable to tobacco use and
children younger than 15 for diseases attributable to exposure to secondhand smoke. There were 39 valid cases of tobacco-related diseases
among children younger than 15 in the sample households. The sample characteristics of the individuals aged 30 and older are presented
in Table 2.2.
Table 2.2 Sample characteristics of patients (30 and older) with proven medical records of tobacco-
related diseases
Urban Rural Total
Number of patients by disease category
Ischemic heart disease 144 220 364
Stroke 130 194 324
COPD 70 91 161
Pulmonary tuberculosis 23 28 51
Lung cancer 17 14 31
Laryngeal cancer 10 11 21
Oral cancer 3 5 8
Total 397 563 960
Gender (%)
Men 66.7 76.2 72.3
Women 33.3 23.8 27.7
Age group (%)
30-39 years 14.9 11.9 13.1
40-49 years 27.0 22.2 24.2
50-59 years 26.4 25.8 26.0
13
Urban Rural Total
60-69 years 20.4 24.9 23.0
70 years and older 11.3 15.3 13.6
Marital status (%)
Married 85.0 90.4 88.3
Single/widowed/separated 15.0 9.6 11.7
Relationship to household head (%)
Head 90.2 91.8 91.1
Other members 9.8 8.2 8.9
Education (%)
No formal education 30.0 35.6 33.4
Less than primary 9.8 14.0 12.3
Primary 14.6 19.0 17.2
Less than secondary 38.3 28.2 32.4
Secondary and above 7.3 2.8 4.7
Employment status (%)
Not Employed 39.1 31.3 34.5
Employed 60.9 68.7 65.5
Primary occupation (%)
Self-employed in non-farm businesses 24.6 17.7 20.6
Laborer 10.1 10.2 10.1
Farmers 10.4 26.8 20.0
Professional 11.1 6.3 8.3
Retired, students, others 4.4 6.8 5.8
Homemaker 26.7 22.5 24.3
Unemployed 12.7 9.6 10.9
Annual per capita household income
1st Quartile (less than BDT 28,800) 17.4 29.0 24.5
2nd Quartile (BDT 28,801 to 45,000) 25.1 24.2 24.6
3rd Quartile (BDT 45,001 to 88,000) 31.0 21.8 25.4
4th Quartile (above BDT 88,000) 26.5 24.9 25.5
14
SAMPLING WEIGHT
To ensure the representativeness of the sample measures at the
national level, sample weight was derived considering the multi-
stage probability sampling design. It was given by the inverse of the
unconditional probability of selection for each household, which
is the product of the probabilities of selection at each stage of the
sample design.
At the rst stage, the probability of the i-th PSU (Mauza in rural and
Mohalla in urban areas) being selected, p_i, was:
At the second stage, the probability of the j-th SSU (village in rural
areas and para in urban areas) within the i-th PSU being selected,
p_(j(i)), was:
The unconditional joint probability of selecting the i-th PSU and the
j-th SSU from the i-th PSU is,
At the third stage, the conditional probability of the h-th household
from the j-th SSU in the i-th PSU being selected, p_(h(i(j))), was:
Finally, the unconditional joint probability of selecting the h-th
household from the j-th SSU in the i-th PSU into the sample was:
Thus, the associated sampling design weight for a respondent
household was:
The total sample of 10,000 households was allocated to a selected
SSU in a selected PSU, given by N_h(i(j)) above, according to the
population proportion of the total number of households in a
selected SSU in a selected PSU within each district in each division.
Thus,
= ∗
∑
where
= number of PSUs chosen from each district (Zilla),
= number of households in the i-th PSU as of the census of 2011, and
∑ = total number of households in all PSUs in a district as of the census of 2011.
() =()
∑()()
where
()
= number of SSUs selected by simple random sampling (without replacement), and
∑
()()
= total number of SSUs in the i-th PSU.
()
=1 for all PSUs in the sample.
where
=
number of households selected by simple random sampling from the j-th SSU in the
i-th PSU, and
=
total number of households in the j-th SSU in the i-th PSU as of the census of 2011.
where
= total number of households at the national level in the census of 2011,
= total number of households in each division in the census of 2011,
= total number of households in each district in the census of 2011,
= total number of households in a selected PSU in the census of 2011,
= total number of households in all the selected PSUs in a district in the census of 2011,
= total number of house holds in a selected SSU in the census of 2011,
= total number of households in all the selected SSUs in a PSUs in the census of 2011,
The ratio
, because only one SSU was selected per PSU.
15
SURVEY TOOLS
Household Disease Prole Questionnaire: First, an in-depth
survey named “Disease Prole” was carried out over 10,000 randomly
selected households throughout Bangladesh to identify the selected
tobacco–related illnesses. The respondents were screened initially
with a questionnaire regarding specic symptoms related to the
selected tobacco-related diseases, and tobacco-use status of the
members within the respondent households. This questionnaire was
developed by members of the study team, and it was nalized on the
basis of pre-testing and opinions of experts (including a number of
physicians) (see Appendix 1).
Household Health Cost Questionnaire: The household “Health
Cost” questionnaire was administered on 998 households that were
conrmed to have the selected tobacco-related diseases out of
10,000 disease-prole intervie wed households. It collected extensive
information on health-seeking behavior and a variety of costs of
treatment incurred in last one year before the survey. More specically,
household expenditures on outpatient care included primary
care, specialist consultations, health services provided in hospital
emergency departments, outpatient surgeries, etc. Household
expenditures on inpatient care included costs of treatment during
hospital stays. Distinct sections in the questionnaire were devoted to
collecting information only on morbidity, mortality, health insurance,
purchase pattern and use of tobacco products, etc. (see Appendix 2).
In addition, it obtained information on household characteristics,
e.g. household yearly/monthly income-expenditure, assets, tobacco
use by each member, occupation and the educational achievement
of each member, etc. The dra questionnaires were reviewed by the
technical advisors and pretested before they were nalized.
Diagnosis of Diseases: The Disease Prole questionnaire provided
the scope for screening of those who were found to have tobacco-
related illnesses. Only those who had valid documents (e.g.,
prescription from a qualied doctor (at least MBBS), hospital records)
concerning any of the selected diseases, as checked by the survey
enumerators, were considered diagnosed or conrmed cases, and
subsequently they were surveyed with the Health Cost questionnaire.
The in-depth medical questionnaire named “Disease Prole” in
the rst stage was originally designed to identify probable cases of
tobacco-related diseases. Those detected as probable cases (per self-
reported) but who did not have valid supporting documents were
excluded from the survey.
SURVEY MANAGEMENT AND QUALITY
CONTROL
The primary data collection for this study through a nationwide
household level survey involved eicient planning and management
of the eldwork including placement of eld enumerators at
their respective positions, movement from one survey location to
another, provision of food and lodging facilities to the enumerators,
supervision at dierent tiers, and coordination with the headquarters.
For eight divisions, 19 survey teams (16 for outside the Dhaka city
and 3 for the Dhaka city) were formed. Each team consisted of two
enumerators (both male and each at least a university graduate).
Though 38 enumerators were employed throughout the survey,
the preparatory training was provided to a total of 58 prospective
enumerators. All had to attend a comprehensive compulsory training
session for two long oicial days before they were sent to the eld.
During the training session, the prospective enumerators were sent
to the nearby clusters within the Dhaka city for one day to assess
the respondents’ ability to understand the questionnaire in dierent
situations. Finally, the enumerators were recruited on the basis
of sincerity and aptitudes to ll out the questionnaires in varying
circumstances, taking into consideration their participation in the
training sessions and performance on the eld day.
The trainers who were involved in developing the questionnaire
conducted the training of the enumerators. Every question was
claried. Field-like condition was simulated in the classroom and
mock interviews were conducted to make the questionnaire clear
to the enumerators. Aer every mock interview, various challenges
arose, and the trainers claried every issue for every question.
Many of the suggestions from the trainee enumerators to make the
questionnaire clearer were also incorporated in the nal printed
version. One day eld practice was organized in the urban area of
Dhaka before the beginning of the actual survey.
For eld administration, a temporary oice was set up in Dhaka city
(the Head Oice was in the University of Dhaka, Arts Building; one
other oice was at the Bangladesh Cancer Society Headquarters,
Mirpur, Dhaka). A number of trained personnel were involved in
coordination of the eld, in dierent capacities, some with personnel
management responsibilities and some with data management
duties. The designated eld manager administered and coordinated
the activities of enumerators through a hierarchy of supervisors.
Enumerators were expected to ll out all information sought in
the household questionnaire, and they identied probable cases
of tobacco-related illnesses. They also marked suspected cases
(households that claimed to have at least one member who had any
of the selected diseases but failed to show necessary documents or
households that reported a number of symptoms of the diseases)
and passed on the to the physician working in the Dhaka Oice. The
conrmation of these cases required expensive diagnostic tests which
were beyond the purview of the timeline and budget of the study.
In the evening, each enumerator would produce a list of information
that was lacking from the households surveyed. Each pair of
enumerators would exchange their surveyed questionnaires to some
other pair of enumerators for cross-examination and then send it to
the Dhaka Oice. Three people (each of them recent graduates from
the Department of Economics, University of Dhaka) were appointed
full time to check entry errors by the enumerators and logical aws
(if any) in the completed questionnaire. Each of them re-checked 10
questions from each pair of enumerators at the household level for
consistency. One additional full day was allotted for the base work
in each cluster to re-check the whole questionnaire and collect
the missing information. If any information from the respondent
household or an individual household member was incomplete
or confusing, the three research team members would contact
the households immediately to collect accurate and complete
information.
16
DATA MANAGEMENT
Before handing over the questionnaires for coding, each team
checked the questionnaire again in the presence of the supervisor
(student) working at the Head Oice. Coders then worked on
the questionnaire under the guidance of the data management
supervisor and coordinator of the household survey. The data entry
supervisor was present as the data were entered. A trained pool of
data entry personnel did the work under one supervisor. The data
entry supervisor and the investigators did consistency checks.
DATA STORAGE
Data were preserved in secure locations, in secure computers. Using
soware packages (e.g. EXCEL, STATA and SPSS), descriptive analysis
was carried out to guide data checks and cleaning. The identication
codes were separated from the database for maintaining anonymity
and privacy of the respondents and strict condentiality of data.
2.2.2 Secondary Data
The prevalence rates of current, former and non-tobacco users came
from the Global Adult Tobacco Survey Bangladesh 2009 (World
Health Organization, Country Oice for Bangladesh, 2009). Data on
government expenditures and health care utilization rate on inpatient
and outpatient care provided in public health care facilities came from
the national health sector budget, available in the Ministry of Finance’s
Budget Brief and the national Health Bulletin 2017 published by the
Management Information System of the Directorate General of Health
Services, Ministry of Health and Family Welfare (Ministry of Health and
Family Welfare, Government of Bangladesh, 2017). Average annual
salary per employed person and the employment rate of working
age individuals were drawn from the Labor Force Survey 2016-17
published by the Bangladesh Bureau of Statistics.
Supplementary national level data on cause-specic mortality,
age-specic mortality rates attributable to tobacco, and probability
of survival between successive age groups were obtained from
Bangladesh Sample Vital Statistics 2016, Global Burden of Disease
Study 2017 and WHO Life Table for Bangladesh. Data on annual
tobacco tax revenue was obtained from the National Board of
Revenue, Ministry of Finance.
2.3 Measures
Study Population
Because the average age at initiation of daily smoking is below
20 and the health eects of tobacco use result from many years of
exposure, studies evaluating the burden of tobacco use generally
focus on adults 30 years and older. Exposure to secondhand smoke
can, however, aect nonsmokers of any age, particularly children. The
study population in the present case included adults 30 and older for
estimating the health burden of tobacco use and children below age
15 for estimating the health burden of secondhand smoke exposure.
Status of tobacco use
If adult household members age 15 and older responded that
they currently smoked cigarette, biri and/or hukka, and/or used
smokeless tobacco products such as jarda, gul and sadapata, they
were identied as current tobacco users. Those who were not using
any tobacco product at the time of the survey were asked about their
past tobacco use and were identied as former tobacco users if they
answered yes. Current and former tobacco users were identied
as “ever” tobacco users in subsequent analysis. The rest of the
individuals were identied as never tobacco users.
Exposure to secondhand smoke
Nonsmokers, specically children younger than 15 and pregnant
women, living with current smokers in the same household where
smoking was allowed indoors were identied as regularly exposed
to secondhand smoke. Otherwise, they were considered as not
exposed to secondhand smoke. This criterion excluded occasional
exposure of children and pregnant women to secondhand smoke in
public or workplaces, which can be further restricted by smoke-free
laws. However, smoke-free laws do not protect children or pregnant
women who share common airspace with smokers residing in the
same household. The health consequences of such exposure go
unaccounted for in the existing literature. The present study aimed
to measure the health impact of this unrestricted exposure to indoor
secondhand smoke on children and pregnant women in particular.
Prevalence of Tobacco-related Diseases
The population-level prevalence of tobacco-related diseases was
estimated using the weighted sample proportion of individuals
who reported currently suering from at least one tobacco-related
disease. The prevalence was estimated by:
Age groups: 30 years and older, below 15 years
Gender: Male, female
Disease categories:
• Adults 30 and older: Ischemic heart disease, stroke, COPD,
pulmonary tuberculosis, lung cancer, laryngeal cancer, and
oral cancer
• Children younger than 15: Autism, asthma and lower
respiratory infection for children younger than 15, in addition
to the seven diseases mentioned above for adults 30 and
older
• Newborn: Low birth weight, sudden infant death syndrome
(SIDS)
• Tobacco use status among adults: Ever tobacco user, never
tobacco user
• Exposure to secondhand smoke: Residents in households
with smokers, residents in households without smokers
The prevalence of disease in each population sub-group was
multiplied with corresponding population size to obtain the total
number of patients suering from these diseases at the national level.
17
Mortality from Tobacco-related Diseases
Households reported on the premature deaths of family members
retrospectively for the ve years before the survey. The causes
of death were classied as natural death due to aging, accident,
injuries and illnesses. The tobacco-use status of the deceased was
also reported to help identify the mortality rates among ever and
never tobacco users. The mortality of children and nonsmoking
adults was dierentiated by the status of households with smokers
and without smokers, to identify deaths attributable to secondhand
smoke exposure. The mortality rates obtained from the health cost
survey were calibrated against the corresponding national level
death rates reported in the latest issue of Bangladesh Sample Vital
Statistics 2016. The mortality rates, thus determined from this study,
were applied to national population size of corresponding age
groups to calculate the number of deaths attributable to tobacco-
related diseases both from tobacco use and secondhand smoke
exposure.
Risk Ratio (RR)
Disease-specic RRs for tobacco use were estimated using the
following ratio:
When RR is statistically signicantly greater than 1, it indicates
excess risk of morbidity associated with tobacco use, either current
or former.
For estimating the excess risk of premature mortality from tobacco-
related diseases, a separate RR was calculated as:
In addition, separate RRs were calculated for measuring the excess
risk from secondhand smoke exposure based on disease prevalence
and cause-specic mortality among children younger than 15
reported by sample households as below:
The RR for exposure to secondhand smoke among adult nonsmokers
age 15 and older was found statistically not signicant and hence
this age group was not covered in subsequent analysis.
Population-attributable Risk (PAR)
Based on the RRs measured from household survey data gathered
in 2018 and the estimates of national prevalence of tobacco use
and exposure to indoor secondhand smoke obtained from the
Global Adult Tobacco Survey conducted in 2009, separate PARs
were calculated for relative morbidity and mortality risks of tobacco
use and exposure to secondhand smoke at the aggregate level
using equation (1). We were thus able to maintain a nine-year lag to
relate the health outcomes reported in 2018 to the rate of tobacco
consumption observed in 2009. Gender- or age group-specic
PARs were not used in subsequent analysis because of inadequate
information on the breakdown of health care costs, morbidity and
premature mortality by these population sub-groups.
Private Health Expenditures
Private health expenditure for each of the tobacco-related diseases
was collected from the health cost survey based on self-reported
out-of-pocket health spending under three headings:
1. Inpatient care: Disease-specic average out-of-pocket inpatient
health expenditure was obtained by multiplying the average
out-of-pocket expenditure per day in inpatient hospitalization,
average number of days per hospitalization and average number of
hospitalizations per patient in last 12 months. The disease-specic
average out-of-pocket inpatient health expenditure was multiplied
with corresponding total number of patients at the national level
who sought medical care and PAR to obtain the total out-of-
pocket inpatient health expenditure for each disease. According to
the survey, only 55% of the patients who identied themselves as
suering from one of the tobacco-related diseases sought medical
care from health care facilities (private or public) in the last 12
months. The disease-specic total out-of-pocket inpatient health
expenditures were added together to calculate the total tobacco-
attributable inpatient private health care expenditure.
2. Outpatient care: Disease-specic average out-of-pocket
outpatient health expenditure was found by multiplying the average
out-of-pocket expenditure per outpatient visit and average number
of outpatient visits per patient last 12 months. The disease-specic
average out-of-pocket outpatient health expenditure was multiplied
with corresponding total number of patients at the national level
who sought medical care and PAR to obtain the total out-of-pocket
outpatient health expenditure for each disease. The disease-specic
total out-of-pocket outpatient health expenditures were added
together to calculate the total tobacco-attributable outpatient
private health care expenditure.
3. Treatment received abroad: The total tobacco-attributable cost
of treatment received abroad was reached by multiplying average
out-of-pocket expenditure per visit for treatment received abroad,
average number of visits per patient for treatment received abroad
in last 12 months, total number of patients at the national level who
sought medical care and the PAR for all diseases combined. Because
of the very small number of observations by disease categories, it
was not feasible to obtain disease-specic cost for treatment
received abroad.
RR =Prevalence of tobacco − related disease among ever tobacco users age ≥30
Prevalence of tobacco − related disease among never tobacco users age ≥30
RR =Mortality rate from tobacco − related disease among ever tobacco users age ≥30
Mortality rate from tobacco − related disease among never tobacco users age ≥30
RR =Prevalence of tobacco − related disease among children in smokers′households
Prevalence of tobacco − related disease among children in nonsmokers′households
RR = Mortality rate from tobacco − related disease among children in smokers
′
households
Mortality rate from tobacco − related disease among children in nonsmokers
′
households
18
The above three components constituted the total tobacco-
attributable private health care expenditure (PRHCE). There were
no reported health expenditures covered by health insurance, either
private or public, in the household survey. More formally,
PRHE = [PRINP * INPDAY * NINP + PROUTP * NOUTP + TABR *
NTABR] * POP * PAR ( 2)
where
PRINP = average private expenditure per day in inpatient care;
INPDAY = average number of days per hospitalization;
NINP = average number of hospitalizations per person;
PROUTP = average private expenditure per outpatient visit;
NOUTP = average number of outpatient visits per person;
TABR = average private expenditure per visit for treatment
received abroad;
NTABR = average number of visits per patient for treatment
received abroad;
POP = total number of patients who sought medical care.
Public Health Expenditures
The government health sector budget for the 2018-19 scal year
was BDT 233.38 billion, which includes operating and development
expenses for inpatient, outpatient and emergency medical services
in public health care facilities (e.g., Upazila Health Complex, District
Hospitals, Medical College Hospitals, Specialized Institutes). The total
public health expenditure budget was allocated into inpatient and
outpatient (including emergency) services based on the information
provided by household survey respondents. The total public
health expenditure for each type of service was then divided by the
annual national health care utilization rate (number of attendees)
of inpatient, outpatient and emergency services in public health
care facilities to estimate the average public expenditure per day in
inpatient services and average cost per patient in outpatient and
emergency departments.
The average public expenditure per day in inpatient services was
multiplied with the average number of hospitalizations per patient,
average number of days per hospitalization, total number of patients
suering from tobacco-related diseases who sought medical care,
and the PAR for all diseases combined (as obtained from household
survey data) to nd the aggregate tobacco-attributable public health
expenditure for inpatient care. Similarly, the average public health
expenditure per outpatient visit was multiplied with the average
number of outpatient visits per patient, total number of patients
suering from tobacco-related diseases who sought medical care,
and the PAR for all diseases combined (as obtained from household
survey data) to obtain the aggregate tobacco-attributable public
health expenditure for outpatient care. The sum of the public health
expenditure estimated for inpatient and outpatient care provided the
total tobacco-attributable public health care expenditure (PUHCE).
More formally,
PUHE = [ PUINP * INPDAY * NINP + PUOUTP * NOUTP] * POP * PAR
(3)
where PUINP = average public expenditure per day in inpatient care;
PUOUTP = average public expenditure per outpatient visit;
and the rest of the notations follow the interpretation in equation (2).
Costs of Morbidity
The costs of morbidity (COSTMORB) attributable to tobacco use
include the following four components that were estimated from
household survey data:
1. Value of lost time of patients 30 and older attending health
care services
The value of lost work days of employed patients who sought health
care services was reached by adding the product of the average
number of work days lost for attending health care services, average
daily income per employed patient, proportion of patients employed,
total number of patients getting health care in last 12 months, and
PAR, together over all diseases.
The value of the lost time of non-employed patients for getting health
care was calculated using the average number of days spent per non-
employed patient for getting health care, average daily reservation
wage of non-employed patients imputed using the average daily
income of employed patients, proportion of patients not employed,
total number of patients getting health care in last 12 months, and
PAR, added together over all diseases.
2. Value of time of caregivers
The cost for caregivers for treatment received domestically was
reached using the product of the average cost per patient for
caregivers, the proportion of patients attended by caregivers when
they received treatment in domestic health care facilities, total
number of patients attending health care in last 12 months, and PAR,
added together over all diseases.
The cost for caregivers for treatment received abroad was found
using the product of the average cost per patient for caregivers, the
proportion of patients attended by caregivers when they received
treatment abroad, total number of patients attending health care
in last 12 months, and PAR for all diseases combined. Because of
the limited number of observations, this estimate was done at the
aggregate level for all diseases together.
19
3. Expected market productivity loss of employed patients
Expected annual market productivity loss of employed patients
due to morbidity was computed by multiplying the loss of expected
annual income per employed patient due to morbidity by the total
number of patients suering from tobacco-related diseases and
PAR for all diseases combined. The loss of expected annual income
considers both the reduction in average annual income per employed
patient and the reduction in employment probability of patients due
to morbidity.
The average monthly salary per employed person reported in the
national Labor Force Survey (LFS) 2016-17 was BDT 13,258, which was
equivalent to average annual salary of BDT 159,096 in 2016 prices or
BDT 178,155 in 2018 prices aer adjustment for annual ination. Let
the annual average income of employed patients in the household
survey be BDT X in 2018. If the national prevalence of the diseases is
p and the annual average income per employed person among those
without the disease is Y, the national annual average income is given
by:
p * BDT X + (1-p) * BDT Y = BDT 178,155.
With p and X estimated from the household survey conducted for this
study, we solved for Y.
In the same manner, we obtained the national employment rate of
0.65 among population aged 30 and older from the LFS 2016-17 and
calculated the employment rate E among the patients of the same
age group in the household survey. If the employment rate among
the 30+ age group without the disease is Z, the national employment
rate is reached by:
p * E + (1-p) * Z = 0.65
with p and E estimated from the health cost survey, we solved for Z.
The expected annual market productivity loss of employed patients
due to morbidity was then calculated as the dierence in the
expected annual income between a patient and the expected annual
income of a person if he/she did not have the disease, that is, Z * Y - E
* X. This method closely resembles the case-control analysis where
cases (participants with a particular health condition or health risk)
are matched to controls (participants without the condition or risk)
to measure health-related productivity loss and the analysis groups
dier only by the presence of a particular health condition or risk
(Mitchell & Bates, 2011).
4. Household productivity loss of non-employed patients
Annual household productivity loss of non-employed patients due to
morbidity was found using the number of days ill in last 12 months,
average daily household productivity of non-employed patients
imputed by the average daily income reported by employed patients,
proportion of non-employed patients, total number of patients at
the national level and PAR by disease categories. These products
were added across all disease categories to obtain the total annual
household productivity loss from tobacco-attributable morbidity.
Costs of Mortality
The economic cost of mortality (COSTMORT) due to tobacco-related
illnesses involved estimation of income loss from premature mortality
in the following stages:
1. The years of potential working life lost per death was determined
for seven age groups in ve-year intervals beginning at 30 and ending
at the potential age of retirement at 64 (e.g., 30-34, 35-39, 40-44, 45-
49, 50-54, 55-59, and 60-64). For example, the potential working years
lost to a person expiring at an age in the interval 30-34 was calculated
as the number of years from the midpoint of the age group 32 to the
endpoint of working age at 64, which is 32 years.
2. The number of tobacco-attributable deaths in each age group
was determined by applying the proportion of tobacco-attributable
deaths in the corresponding age group reported in the Global Burden
of Disease Study 2017 to the total number of tobacco-attributable
deaths determined from the health cost survey for this study. Due
to limited number of observations on deaths reported in sample
households, it was not possible to work out age-group specic death
rates from the present survey. The age distribution of deaths from
the Global Burden of Diseases Study was simulated for the present
purpose.
20
3. The expected annual income of an employed person without a
disease, as calculated in the Costs of Morbidity section described
above, was imputed as the potential annual productivity of a
prematurely deceased person. This expected annual income was
assigned to each successive year, aer adjustment for expected
income growth at 6 percent annual rate (average annual growth rate
of per capita GDP in Bangladesh) and discounting for future income
stream at 3 percent annual rate, in the potential work life lost to that
person. In contrast to the existing method of assigning economic
value to only the potentially employed individuals, we assigned
economic value to the working age lost to all individuals regardless of
their potential employment status.
4. The sum of the discounted annual income over the potential
working life years lost provided the present value of foregone
lifetime income for each age group, which was then weighted by the
age-specic probability of survival from one to the next age group
obtained from the WHO Life Table for Bangladesh.
5. Finally, the age-specic present value of foregone lifetime
income per person was multiplied by the total number of tobacco-
attributable deaths in the corresponding age group to obtain the
aggregate economic cost of premature mortality. Thus,
where SURV(m) is the probability that a person would survive to age
m, max a is the maximum age group, Y is the average annual income
of an individual without a disease mentioned in the Cost of Morbidity
measure above, g is the growth rate of per capita income, V is the
discount rate, and a is the age at death.
Costs of Exposure to Secondhand Smoke
As the excess risk of disease prevalence from secondhand smoke
exposure was statistically signicant for the age group below 15,
the health expenditures and indirect morbidity costs attributable to
secondhand smoke exposure were estimated for children only. The
private and public health expenditures for treating the secondhand
smoke exposure caused illnesses among children were measured
following the same method described above. Since children are
below the working age, there is no estimate of current productivity
loss due to morbidity. The cost of morbidity in this case includes only
the costs incurred for caregiving.
The excess risk of death from secondhand smoke exposure was
measured using the reported deaths among nonsmokers of all ages
who were living in smoker’s households compared with the reported
deaths in the corresponding group who were living in nonsmokers’
households. Thus, the cost of mortality was measured for all age
groups. For premature deaths among children due to exposure to
secondhand smoke, the full potential working life (assumed to begin
at age 22, the midpoint of age group 20-24, and end at age 64) was
considered lost and was accounted for in the same process as for
adults. The combined economic cost from tobacco use and
exposure to secondhand smoke was nally given by:
TOTALCOST = PRHE + PUHE + COSTMORB + COSTMORT (5)
= ∑ [()][
max
∗ (1 + )−/(1 + )− (4)
23
SECTION 3
FINDINGS
21
3.1 PREVALENCE OF TOBACCO
RELATED DISEASES
Figure 3.1.1
Prevalence of tobacco-related diseases among adults aged 30 and older
Note: The prevalence of disease is weighted based on multistage probability sample design.
5.2%
2.3%
1.6%
0.3% 0.2% 0.2% 0.1%
9.1%
0.0%
2.0%
4.0%
6.0%
8.0%
10.0%
12.0%
Ischemic
heart
disease
Stroke COPD Pulmonary
tuberculosis
Lung cancer Laryngeal
cancer
Oral cancer All
Male Female Both
The overall prevalence of tobacco-related diseases stood at 9.1
percent (10.1 percent among males and 7.8 percent among females)
for adults aged 30 years and older (Figure 3.1.1). The disease
prevalence was generally higher among males than among females
for all tobacco-related diseases, except laryngeal cancer for which
the prevalence was the same (0.2 percent) across gender. Ischemic
heart disease (IHD) was found to be the most prevalent disease
5.2 percent), followed by stroke (2.3 percent), chronic obstructive
pulmonary disease (COPD) (1.6 percent), pulmonary tuberculosis
(0.3 percent), lung cancer (0.2 percent), laryngeal cancer (0.2
percent) and oral cancer (0.1 percent), as shown in Figure
3.1.1. The prevalence of the three types of cancer combined
(lung, laryngeal and oral cancers) was 0.7 percent overall
(0.8 percent among males and 0.4 percent among females).
industry 1 industry 2 Solutions 1
FIG 3.1.1 PREVALENCE OF TOBACCO-RELATED DISEASES AMONG
22
Figure 3.1.2
Prevalence of tobacco-related diseases among children younger than 15
Note: The prevalence of disease is weighted based on multistage probability sample design.
1.01%
21.7%
2.6%
0%
5%
10%
15%
20%
25%
All tobacco-related
diseases
Low birth weight SIDS
Male Female Both
Tobacco-related diseases and other secondhand smoke exposure
induced diseases such as autism, asthma and lower respiratory
infection were prevalent among 1.01 percent of children (0.94
percent among males and 1.09 percent among females) younger
than 15 (Figure 3.1.2). Among 424 live births recorded in the
survey for the last 12 months before the survey, 21.7 percent
were reported to be low birth weight (22.2 percent among male
live births and 17.4 percent among female live births) and 2.6
percent suered from Sudden Infant Death Syndrome (SIDS) (2.9
percent among male infants and none among female infants).
The rate of SIDS is a close approximation to the national level
infant mortality rate (deaths before less than one year of age) at
28 per 1,000 live births, including neo-natal mortality rate (deaths
before less than one month of age) at 19 per 1,000 live births
and post-neo-natal mortality rate (deaths between one month
and 11 months) at 9 per 1,000 live births in 2016 (Bangladesh
Bureau of Statistics, 2017).
industry 1 industry 2 Solutions 1
FIG 3.1.2 Prevalence