The impact of obesity and overweight on medical expenditures and disease incidence in Korea from 2002 to 2013
The impact of obesity and overweight on medical expenditures and disease incidence in Korea from 2002 to 2013
Hyun Jin Song 0 1
Jinseub Hwang 1
Seonmi Pi 1
Sena Ahn 0 1
Yoonseok Heo 1
Susan Park 0 1
Jin-Won Kwon 0 1
0 College of Pharmacy and Research Institute of Pharmaceutical Sciences, Kyungpook National University , Daegu , South Korea , 2 College of Pharmacy, University of Florida, Gainesville, Florida, United States of America, 3 Major in Statistics and Data Science, Daegu University , Gyeongbuk , South Korea , 4 Department of Surgery, Inha University College of Medicine , Incheon , South Korea
1 Editor: Jacobus P. van Wouwe, TNO , NETHERLANDS
Data Availability Statement: We used data from
the Korean National Health Insurance
ServiceHealth Screening Cohort (NHIS-HEALS) database
2 to 2013
. Raw data can be provided to
researchers upon reasonable request and with the
permission of the Korean National Health
Insurance Service (NHIS) Institutional Data Access
(http://nhiss.nhis.or.kr). The authors did not have
any special privileges in obtaining this data set.
Funding: This study used NHIS-NSC data
(NHIS2016-2-149) provided by the National Health
Few studies have assessed the long-term medical costs and incidence of obesity and
overweight in Asia. We evaluated the impact of body mass index (BMI) on medical expenditures
and disease incidence and prevalence over more than 10 years in South Korea.
data from the Korean National Claims Database, we analysed two
population sets (initial BMI in 2002±2003; consistent BMI in 2002±2003 and 201
was defined by Asian BMI criteria. Incremental medical expenditures or Charlson
Comorbidity Index (CCI) ratios for obese compared to normal weight individuals were calculated.
Medical expenditure over 11 years was estimated by BMI using a generalised linear model.
Individual obesity-related disease incidence was determined and adjusted hazard ratios
Data for 496,469 and 214,477 individuals were included in the entire and consistent BMI
level populations, respectively. Average CCI score change in normal weight and the obesity
III (BMI 35±59.99 kg/m2) group over 11 years were 0.94 and 1.56, respectively in the entire
population, and incremental ratio in the obesity III group was 66.0% compared to the normal
weight group. In consistent BMI level population, incremental ratio (92.1%) for obesity III
was higher than entire population. Medical costs in the obesity III groups versus the normal
weight group in the entire and consistent BMI level populations increased by 38.4% and
77.1%, respectively. Over 11 years, individuals with BMI 30 kg/m2 in the entire and
consistent BMI level populations had post-adjustment medical expenditures of 1.13±1.20 and
1.21±1.40 times the normal weight group, respectively. Incidence rate and adjusted hazard
Insurance Service (NHIS). This research was
supported by a grant from the Korea Health
Technology R&D Project through the Korea Health
Industry Development Institute (KHIDI), funded by
the Ministry of Health & Welfare, Republic of Korea
(grant number: HC15C1322 to Jin-Won Kwon)
(http://www.mohw.go.kr). The funders had no role
in study design, data collection and analysis,
decision to publish, or preparation of the
Competing interests: The authors declare no
potential conflicts of interest with the NHIS with
respect to the authorship and/or publication of this
ratio of obesity-related disease increased in the obesity groups compared to the normal
Our findings emphasize the importance of the effective and sustainable obesity
management strategies, considering the dramatic increase in obesity (BMI 30 kg/m2) in South
Since the World Health Organization (WHO) classified obesity as a disease, various measures
to decrease obesity have been considered worldwide . However, the prevalence of obesity has
not decreased, and it remains an urgent public health agenda in both developed and developing
countries. From 1980 to 2013, the prevalence of obesity or overweight increased worldwide by
27.5% and 47.1% in adults and children, respectively . The prevalence of obesity differs by
sex, age, and socioeconomic status . In particular, a rapid increase in the prevalence of
obesity among younger populations and those of lower socioeconomic status has been observed.
Obesity has been a major public health issue both in itself and as a result of its association
with the development of several chronic comorbidities. Obesity-related diseases are not
limited to cardiovascular diseases. There is abundant evidence of the association between obesity
and cancer, respiratory, and neurodegenerative diseases [4±5].
Many countries are faced with dramatic increases in medical expenditures owing to the
rising prevalence of chronic diseases. Accordingly, these countries seek to decrease medical costs
and increase healthy life expectancies. The increased prevalence of chronic diseases may be the
result of the increased number of elderly individuals. This factor, combined with the increasing
prevalence of obesity, may be responsible for a substantial proportion of the disease burden
. Developed countries track both the social and medical costs of obesity and prioritise
obesity prevention in health agendas in order to decrease medical expenditures.
In the US, medical costs related to obesity comprise more than 20% of the annual national
healthcare expenditure . The direct medical costs of obese adults are more than 42% higher
than those of adults of normal weight  and more than 81% higher in severely obese adults
(body mass index [BMI] >40 kg/m2) . The seriousness of obesity-related comorbidities or
costs has been reported in developed countries as well as the US [10±12].
Asian individuals experience obesity-related diseases at lower BMIs compared to those in
Western countries . However, few studies have assessed medical costs and disease
incidence by BMI in Asian countries, including South Korea. Furthermore, there have been few
studies regarding the long-term impact of obesity and overweight. Thus, the aim of this study
was to evaluate the impact of BMI on medical expenditures and disease incidence and
prevalence over more than 10 years in South Korea using a national longitudinal database from
Materials and methods
We used data from the Korean National Health Insurance Service-Health Screening Cohort
(NHIS-HEALS) database from 200
2 to 2013
[14±15]. Raw data can be provided to the
2 / 14
researchers upon reasonable request and with the permission of the Korean National Health
Insurance Service (NHIS) Institutional Data Access (http://nhiss.nhis.or.kr). We did not have
any special access privileges.
The NHIS-HEALS included all NHI claimed and screened medical information for 514,866
health screening participants who comprised a 10% random sample of all health screening
participants in 2002 and 2003. The participation rate of the general health screening among the
eligible population in 2014 was 74.8% . Although we could not rule out potential selection
bias in participation in the health screening, previous studies using Korean National Health
and Nutrition Examination Survey (KNHANES) data showed obesity prevalence similar to
our results [16±17]. For the direct comparison, we presented the obesity prevalence in our
database and analysed data for subjects aged 40±79 years from KNHANES II (2001) and III
(2005) (S1 Table). Our obesity prevalence was within the confidence interval of obesity
prevalence using KNHANES II data.
The NHIS-HEALS database contains four databases on insurance eligibility, medical
treatments, medical care institutions, and general health examinations. The database on insurance
eligibility includes information related to beneficiary qualifications (e.g. income level, death
record). Individuals 40 years of age or older are required undergoing general health
examinations every two years; these data were captured in the general health examination database.
The medical treatment database included disease diagnosis (International Codes of Disease
10th Edition Clinical Modification, ICD-10-CM), medical procedures, prescription drugs, and
medical expenditures covered under the NHI. The Institutional Review Board of Kyungpook
National University (KNU 2016±0077) approved this study. Informed consent was not
required because the patient records from the NHIS-HEALS were anonymised and
Among 514,350 health screening participants, individuals with cancer (n = 17,103, 3.3%) or
women who were pregnant (n = 754, 0.1%) were excluded because they might have a
disproportionate influence on weight. In addition, individuals with extreme BMI values, such as
more than 60 kg/m2 (n = 24, 0.005%), were excluded, as in another study in South Korea .
In our database, there were very few individuals with BMI 35 kg/ m2 (n = 870, 0.2%).
Therefore, obese individuals with BMI 60 kg/ m2 constituted a very small portion, and there was a
high probability of wrong coding data.
For this analysis, we used two populations. First, we selected individuals with weight and
height measurements performed in 2002±2003 from the general health examination data of
the NHIS-HEALS database; the first population set was the entire population. Second,
individuals whose BMI remained in the same category at baseline (2002±2003) and at the observation
end date (201
) were selected in order to investigate the weight effects in detail; this
group constituted the consistent BMI level population set. The number of two populations set
was specified in the S1 Fig. In the entire population set, some individuals died before 201
, but all individuals included in consistent BMI level population set lived for the entire
11-year follow-up period. The analysis of two populations was considered because we could
not fully capture BMI changes throughout the follow-up periods and many individuals had
only baseline BMI measurements. In the entire population set, the long-term effects of baseline
BMI on clinical outcomes (i.e., disease prevalence, incidence, and medical expenditures) can
be observed, regardless of weight change effects. Through analysis of the population with
consistent BMI levels, we can further investigate the impact on clinical outcome by BMI level
3 / 14
We observed the 11-year changes in Charlson Comorbidity Index (CCI) scores and medical
expenditure. Because general health examinations were conducted every two years, the
followup BMIs for individuals with first measurements in 2002 and 2003 were measured in 201
BMI categories. BMI was determined from weight and height measurements as weight
(kg) divided by the square of the height (m2). According to Asian criteria, BMI was defined as
underweight (<18.5 kg/m2), normal weight (18.5±22.99 kg/m2), overweight (23±24.99 kg/m2),
obesity I (25±29.99 kg/m2), obesity II (30±34.99 kg/m2), and obesity III (35±59.99 kg/m2) [19±
20]. There were some differences in the definition of obesity according to BMI level between
Asian and Western individuals. Underweight was the same according to both criteria;
however, normal weight and overweight according to Asian criteria are considered normal weight
by Western criteria. The obesity I, obesity II, and obesity III groups by Asian criteria were
defined as overweight, obesity I, and obesity II groups by Western criteria.
Income level. Participant income levels were grouped using insurance type and level of
insurance at baseline. Beneficiaries were divided into district subscribers, employee
subscribers, and medical aids. In the case of district and employee subscribers, 10 levels in each group
are available based on insurance amount. Higher scores represent a higher income level. We
divided district and employee subscribers into three groups with similar proportions.
Comorbidities. Comorbidities were investigated using CCI scores. The CCI was
originally developed to evaluate the mortality risk of concurrent diseases. The diseases included in
the CCI are congestive heart failure (CHF), myocardial infarction (MI), cerebrovascular
disease, peripheral vascular disease, connective tissue disease, chronic lung disease, ulcer, chronic
liver disease, severe liver disease, dementia, diabetes, hemiplegia, moderate or severe kidney
disease, tumour, leukaemia, lymphoma, moderate or metastatic solid tumour, and acquired
immunodeficiency syndrome (AIDS). Each disease receives a score (i.e., diseases with higher
mortality risks have higher scores) and the scores are summed to show the mortality risk of an
individual. Quan et al. proposed methods to calculate CCI scores using ICD codes in
administrative data .
Hypertension and depression, diseases related to obesity that are not included in the CCI,
were also considered as comorbidities. Hypertension and depression were defined using
ICD10 codes I10-15 and F32-33, respectively.
Changes in CCI score and medical expenditures between 2002±2003 and 201
We investigated the changes in CCI scores and medical expenditures by BMI category for 11
years by comparing the average CCI score and medical expenditures between participants at
baseline and at the end of the observation period among survivors. For example, if the BMI of
a participant was available in 2002, the average CCI scores and medical expenditures in 2002
and 2012 were assessed.
The NHIS-HEALS database included all medical expenditures for participants covered
under the NHI from baseline to end of observation or death. The medical expenditures
included copayments by beneficiaries and reimbursement costs by the insurer for all medical
treatments including outpatient, inpatient, or emergency department visits under the NHI.
The incremental medical expenditures or CCI score ratios in the obesity I, II, and III groups
were calculated as follows. The 11-year changes in medical expenditures or CCI scores were
calculated for each obesity level. And the differences in changes between each obesity level and
4 / 14
those of the normal weight group were divided by the 11-year change in the normal weight
group and presented as percentages.
Ten-year disease incidence (2004±2013). The incidences of individual diseases associated
with obesity, such as congestive heart failure (CHF), cerebrovascular disease, chronic liver
disease, uncomplicated or complicated diabetes mellitus (DM), hypertension, and depression
were observed. For participants without an individual disease in 2002±2003, the development
of an individual disease was confirmed based on the presence of diagnosis codes in 2004±2013.
Continuous data are presented as means and standard deviation (SD), while categorical data
are shown as frequencies. Mean and SD of CCI scores and medical expenditures at baseline
(2002±2003) and at the end of the observation period (201
) are presented according to
BMI categories at baseline in entire population set. Data were also included from participants
who remained in the same BMI categories at baseline and at the end of the observation period
(consistent BMI level population set). Individuals with medical records in 2002 underwent
follow-up until 2012, while those with data in 200
3 were followed until 2013
. Thus, 11 years of
data were collected. To observe the association between disease incidence and obesity level,
adjusted hazard ratio (aHR) after adjustments for age, sex, income level, and CCI score were
also calculated using time-to-event hazard model. Death event or the end of study period was
considered as censoring in the disease incidence analysis.
To investigate the impact of BMI on medical expenditures after adjusting for confounding
variables such as demographic information and comorbidities, a generalized linear model
(GLM) was used. Log link and gamma distributions were selected in the GLM to reflect the
right-skewed distribution of the 11 years of medical expenditures . We considered
followup time as on offset variable because participants had different follow-up days owing to death.
The dependent variable for medical expenditure in GLM was the ratio of 2012/3 to 2002/3.
Cost ratio and 95% confidence intervals (CIs) using the GLM model are presented.
Data were analysed using SAS 9.4 (SAS Institute Inc., Seoul, Korea).
The entire population set included 496,469 individuals with BMI data in 2002±2003. A total of
214,477 people who remained in the same BMI category from baseline (2002±2003) to the end
of the observation period (201
) were selected as the consistent BMI level population
set. The percentages of participants according to BMI category were 2.3% (underweight),
35.3% (normal weight), 27.3% (overweight), 32.3% (obesity I), 2.7% (obesity II), and 0.2%
Entire participants' demographic, socioeconomic, and clinical data according to BMI
category are shown in Table 1. The average age was slightly lower in higher BMI categories.
Whereas the average age of the underweight group (<18.5 kg/m2) was 56.3 years (SD 11.8), it
was 52.6 years (SD 9.1) in the very severe obesity group (35±59.99 kg/m2). The baseline CCI
score was higher in the underweight (BMI <18.5 kg/m2) and overweight and obesity ( 23 kg/
m2) groups compared to the score in the normal weight (18.5±22.99 kg/m2) group. The CCI
score increased with increased BMI in the overweight and obese group. The prevalence of
hypertension and depression according to BMI category showed similar trends to those of
CCI scores. In addition, the descriptive statistics for the BMI level consistent population set
were presented in S2 Table.
The change in average CCI score over 11 years was 0.94 for the normal weight group (0.90
in 2002±2003 and 1.84 in 201
) in the entire population (participants with BMI data in
5 / 14
Western criteria are presented in parentheses by the near Asian criteria: Underweight (Underweight), Normal weight (Normal weight), Overweight (Normal weight),
Obesity I (Overweight), Obesity II (Obesity I), and Obesity III (Obesity II).
CCI: Charlson Comorbidity Index, NHI: National Health Insurance, SD: standard deviation
2002±2003), while those in the overweight and obesity groups were 1.07 to 1.56. This trend
was similar for the consistent BMI level population set (participants who remained in the same
BMI category as in 201
) (Table 2). The incremental CCI score ratios in the obesity I to
III groups were 30.9% to 66.0% in the entire population set and 32.6% to 92.1% in the
consistent BMI level population set compared to the normal weight group (Fig 1a).
The change in average medical expenditures over 11 years was larger in the underweight
and obesity groups (Table 2) and was greater with increasing obesity level. This trend was also
similar between the entire and consistent BMI level population sets. The incremental medical
expenditure ratios for the obesity I, II, and III groups were 14.1%, 34.3%, and 38.4%,
respectively, in the entire population set and 14.6%, 38.4%, and 77.1% in the consistent BMI level
population set (Fig 1b). The medical expenditures in the follow-up period (maximum 11
years) according to BMI category were calculated after adjusting for sex, age, income level, and
comorbidities (CCI score, hypertension, and depression) using the GLM model with log link
and gamma distribution (Table 3). The expenditures were significantly higher in the obese
groups (obesity I, obesity II, and obesity III) compared to those of the normal weight group in
the entire or consistent BMI level population sets. Individuals with BMI 35 kg/m2 spent
about 1.40 times more than did the reference group of normal weight individuals in the
consistent BMI level population set.
6 / 14
Entire population: participants with available BMI data in 2002±2003
²Population with consistent BMI level for 11 years: participants who remained in their baseline BMI category in 201
³1,000 South Korea Won = 0.92 US$ (based on 30 November 2017)
#Western criteria are presented in parentheses by the near Asian criteria: Underweight (Underweight), Normal weight (Normal weight), Overweight (Normal weight),
Obesity I (Overweight), Obesity II (Obesity I), and Obesity III (Obesity II).
BMI: body mass index, SD: standard deviation
Changes in CCI scores and medical expenditures in the subgroup analysis of men or
women revealed that individuals with higher obesity levels showed worse health outcome in
both groups (S3 Table). Medical expenditures after adjustments for confounding variables
were significantly higher in the overweight and obese groups (obesity I, obesity II, and obesity
III) in women, while they were significantly higher in the obesity II and obesity III groups in
men (S4 Table).
Among participants without individual disease in 2002±2003, the incidence of several
diseases increased in 2004±2013 with increasing BMI (Fig 2). Among participants with BMI
30 kg/m2 in the entire population set, the incidence rates of CHF, cerebrovascular disease,
chronic liver disease, uncomplicated DM, complicated DM, hypertension, and depression
were 8.1%, 12.5%, 32.2%, 41.0%, 17.2%, 49.4%, and 9.2%, respectively, over approximately
10 years. In the consistent BMI level population set, the incidence of CHF, chronic liver
disease, uncomplicated DM, hypertension, and depression among participants with BMI 30
kg/m2 was higher than that in the entire population set. In particular, the highest aHRs of
individuals with BMI 30 kg/m2 in the entire population in comparison to normal weight
individuals were 2.33 (95% CI 2.26±2.39) for hypertension, 2.22 (95% CI 2.14±2.31) for
complicated DM, 1.85 (95% CI 1.81±1.90) for uncomplicated DM, 1.58 (95% CI 1.51±1.65) for
CHF, 1.30 (95% CI 1.27±1.34) for chronic liver disease, 1.08 (95% CI 1.05±1.12) for
cerebrovascular disease, and 0.96 (95% CI 0.93±1.00) for depression. In the consistent BMI level
population set, the aHRs for those diseases in individuals with BMI 30 kg/m2 were greater
than those for in the entire population set. The aHR for depression was 1.18 (95% CI 1.00±
7 / 14
Fig 1. Incremental Charlson Comorbidity Index (CCI) score ratios and medical expenditures ratio over 11 years. Comparison of 2002±2003 and 201
for entire population set (participants with available BMI data in 2002±2003) and consistent BMI level population set (participants who remained in their
baseline BMI categories in 201
) according to BMI a. Incremental CCI score ratio based on normal weight b. Incremental medical expenditure ratio
based on normal weight. Incremental CCI or medical expenditure ratio = (11-year change of each obesity level± 11-year change in normal weight) / (11-year
change in normal weight) 100.
This study quantified the impact of BMI on medical expenditures over 11 years in a Korean
population 40±79 years of age. The incremental medical expenditure ratio showed a
doseresponse relationship with increasing obesity level. The incremental medical expenditure
ratios of the obesity I, II, and III groups over 11 years ranged from 14.6% to 77.1% compared
to the normal weight group in subjects who were remained in the same category (i.e., the
consistent BMI level population set). In the entire population set (i.e., those with BMI data in
2002±2003), similar results (14.1% to 38.4%) were observed even though the differences were
Furthermore, the impact of obesity on medical expenditures was quantified based on the
follow-up periods using a GLM model. Because some individuals died during the observation
8 / 14
Entire population: participants with available BMI data in 2002±2003
² Population with consistent BMI level for 11 years: participants who remained in their baseline BMI categories in 201
# Western criteria are presented in parentheses by the near Asian criteria: Underweight (Underweight), Normal weight (Normal weight), Overweight (Normal weight),
Obesity I (Overweight), Obesity II (Obesity I), and Obesity III (Obesity II).
BMI: body mass index, CCI: Charlson Comorbidity index, NHI: National Health Insurance
period, this factor was also considered in analysis of medical expenditures. Obese individuals
with BMI 30 kg/m2 had medical expenditures of 1.21±1.40 times those of normal weight
individuals after adjusting for age, sex, income level, and comorbidities in the consistent BMI
level population set over approximately 11 years.
9 / 14
Fig 2. Disease incidence over 10 years (2004±2013) a. Entire population set (participants with available BMI data in 2002±2003) b. Consistent BMI level population
set (participants who remained in their baseline BMI categories in 201
Our study results were in accordance with those of previous studies that were limited in
estimating the attributable costs of obesity by the cross-sectional design [10±12]. In a previous
systematic review, the respective incremental costs of overweight and obesity were 9.9% and
42.7% higher than those of normal weight . It is very well known that obesity is a major
cause of chronic diseases. Our study also showed that higher CCI scores with higher baseline
BMI were associated with a greater change in CCI score after 11 years. Increases in CCI scores
in the normal BMI group over 11 years may be explained by the effects of age. The incremental
CCI score changes in the obese group over those of the normal group might be attributed to
the effects of obesity on CCI. In the obesity III (35±59.99 kg/m2) group, the CCI score change
was about twice that of the normal group. The consistent BMI level population set experienced
a greater change in CCI score. In addition, our study determined the obesity-related disease
incidence according to the increase in BMI level over 10 years. The aHRs for obesity-related
disease in the overweight and obesity groups were significantly higher than that in the normal
weight group. Even though these findings were in accordance with those of previous studies
[6, 23±24], the results of our study are meaningful in that they show the long-term effects of
prevalence or incidence of obesity-related comorbidities in an Asian population.
In the present study, no increased medical expenditure in the overweight group (23±24.99
kg/m2) was observed compared to the normal weight group (cost ratio 1.00, 95% CI 0.99±1.01
and 1.02, 95% CI 1.01±1.03 in the entire and consistent BMI level population sets, respectively)
in both sexes. In a separate analysis of men and women, women showed increased medical
expenditures in the overweight group, but this trend was not observed in men. Even though
there was no difference in medical costs in the overweight group, the change in CCI score and
disease incidence in this group also showed increasing trends when compared to those of the
10 / 14
normal weight group. The findings in the overweight group were consistent with those of
previous studies on mortality according to obesity level . The overweight group may not show
the adverse effects of final outcomes such as mortality or medical expenditure compared to the
normal weight group, especially for men. Different BMI criteria for definition of obesity by sex
should be discussed and studied in the future.
The results of the present study indicate that obesity influenced higher medical
expenditures after adjustment for demographic factors and baseline chronic diseases. This finding
may be explained by the incidence of additional diseases due to obesity, as observed in our
study (Fig 2). This trend was more prominent in the consistent BMI level population set,
which remained in the same BMI category. The medical expenditure or CCI score change
showed similar trends in the entire and consistent BMI level population sets, even though the
incremental medical expenditure and CCI score ratio in the BMI 30 kg/m2 group were
slightly higher in the consistent BMI level population set. We could not fully examine weight
gain or loss during the follow-up periods owing to data limitations. Indeed, when we compare
the results between subgroups of consistent BMI and changed BMI (increase or decrease), the
subgroup of BMI decrease showed higher CCI score and medical cost than that of consistent
BMI, especially in normal BMI range (S5 and S6 Tables). Unintentional weight loss is regarded
as a sign of serious illness . In many epidemiological studies, weight loss was associated
with high mortality risk, which was explained by the preexisting disease [27±29]. Thus, it was
possible that the BMI change, particularly BMI decrease, could confound the association
between baseline BMI and medical cost. However, through comparison of the entire and
consistent BMI level population sets, adverse effects of obesity on medical expenditure or disease
incidence or prevalence were robustly confirmed. The influence of weight gain or loss may
have been present in the entire population set, but could not be observed in the consistent
BMI level population set. When considering the higher cost ratio of obesity in the consistent
BMI level population set, the impact of obesity after excluding confounding factors due to
weight change is more obvious.
The findings of the study imply the negative health influence of persistent obesity. Recently,
the prevalence of BMI 30 kg/m2 has been increasing in Korea . Our study results suggest
a dramatic increase in disease burden and medical costs due to obesity in the near future.
Thus, health policy regarding obesity, especially severe obesity, is required.
The present study accomplished several notable achievements. First, our results showed
long-term follow-up results for obesity-related comorbidities and medical expenditures
according to BMI over approximately 10 years in an Asian region. This study was original and
unique in that no previous study has attempted to investigate medical expenditure in relation
to obesity level as well as the incidence or prevalence of obesity-related disease in South Korea.
Second, the quantification of the impact of obesity on medical expenditures offered valuable
insights. We analysed the medical expenditures associated with obesity after adjusting for age,
sex, income level, and comorbidities. Most previous studies on medical costs due to obesity
did not control for those confounding factors. By determining the attributable costs related to
obesity, the importance of obesity prevention and improved intervention strategies for obesity
management may be highlighted from a public health perspective. Third, the incremental
ratios of CCI scores were comprehensively analysed. It is expected that CCI score will increase
with age. Therefore, the change in CCI score in the obesity group was compared with that of
the normal weight group, i.e., the ªincremental CCI score ratioº. With this measurement, we
identified the additional disease burden attributable to obesity without the effect of aging.
This study had several limitations. First, the analysis included only data at baseline and after
11 years, so BMI trends between these time points were not reflected. However, we attempted
to maintain the reliability of the results by analysing a consistent BMI level population set of
11 / 14
individuals who remained at their baseline BMI category after 11 years. Second, the medical
costs in the present study only included health insurance reimbursement costs and copayment
under the NHI scheme. So, we did not consider other costs including out-of-pocket payments
by the uninsured from the patient perspective, or indirect costs from the societal perspective.
Therefore, the absolute costs could be underestimated. However, we can estimate the relative
value of obesity compared to normal weight. In this regard, the results of our study may be
Further research into the impact of weight gain or loss on the cost and disease incidence or
prevalence would be helpful to investigate various benefits or adverse effects on weight-related
public health issues. Additional studies considering the comprehensive costs including indirect
costs from the societal perspective would be beneficial to understand the costs attributable to
The present study estimated the incremental medical expenditure and chronic disease burden
due to obesity over an 11-year follow-up period in an Asian region. The results could be a
reference in Asian countries such as South Korea, which may have similar trends in increasing
obesity prevalence (BMI 30 kg/m2), and may provide evidence for the development of
effective and sustainable obesity management strategies.
S1 Table. The proportion by BMI level in our database and KNHANES.
S2 Table. Baseline characteristics of consistent BMI level population.
S3 Table. Change in Charlson Comorbidity Index (CCI) scores and medical expenditures
between 2002±2003 and 201
in men and women.
S4 Table. Eleven-year medical expenditure ratios by BMI category and sex after
S5 Table. Change in Charlson Comorbidity Index (CCI) scores and medical expenditures
between 2002±2003 and 201
in people whose BMI increased or decreased.
S6 Table. Eleven-year medical expenditure ratios by BMI category after adjustments in
people whose BMI increased or decreased.
S1 Fig. Flow chart of study population selection.
This study used NHIS-HEALS data (NHIS-2016-2-149) provided by the National Health
Insurance Service (NHIS).
12 / 14
Conceptualization: Hyun Jin Song, Sena Ahn, Jin-Won Kwon.
Data curation: Jinseub Hwang, Seonmi Pi.
Formal analysis: Jinseub Hwang, Seonmi Pi.
Methodology: Hyun Jin Song, Jinseub Hwang, Yoonseok Heo.
Project administration: Jin-Won Kwon.
Supervision: Yoonseok Heo.
Writing ± original draft: Hyun Jin Song, Jin-Won Kwon.
Writing ± review & editing: Hyun Jin Song, Susan Park, Jin-Won Kwon.
13 / 14
1. World Health Organization (WHO). Obesity and overweight . Fact sheet 2016 . http://www.who.int/ mediacentre/factsheets/fs311/en/
2. Ng M , Fleming T , Robinson M , Thomson B , Graetz N , Margono C , et al. Global, regional, and national prevalence of overweight and obesity in children and adults during 1980±2013: a systematic analysis for the Global Burden of Disease Study 2013 . Lancet. 2014 ; 384 ( 9945 ): 766 ± 781 . https://doi.org/10. 1016/S0140- 6736 ( 14 ) 60460 - 8 PMID: 24880830
3. Kang HT , Shim JY , Lee HR , Park BJ , Linton JA , Lee YJ . Trends in prevalence of overweight and obesity in Korean adults, 1998 ± 2009 : the Korean National Health and Nutrition Examination Survey . J Epidemiol . 2014 ; 24 ( 2 ): 109 ± 16 . https://doi.org/10.2188/jea.JE20130017 PMID: 24531001
4. National Institute for Health and Clinical Excellence. Obesity: the prevention, identification, assessment and management of overweight and obesity in adults and children . NICE Clinical Guidelines. London: National Institute for Health and Clinical Excellence (UK) , 2006 .
5. Apovian CM , Aronne LJ , Bessesen DH , McDonnell ME , Murad MH , Pagotto U , et al.; Endocrine Society . Pharmacological management of obesity: an endocrine Society clinical practice guideline . J Clin Endocrinol Metab . 2015 ; 100 ( 2 ): 342 ± 362 . https://doi.org/10.1210/jc.2014-3415 PMID: 25590212
6. Li Q , Blume SW , Huang JC , Hammer M , Ganz ML . Prevalence and healthcare costs of obesity-related comorbidities: evidence from an electronic medical records system in the United States . J Med Econ . 2015 ; 18 ( 12 ): 1020 ± 1028 . https://doi.org/10.3111/13696998. 2015 .1067623 PMID: 26134917
7. Spieker EA , Pyzocha N. Economic Impact of Obesity . Prim Care . 2016 ; 43 ( 1 ): 83 ± 95 . https://doi.org/10. 1016/j.pop. 2015 . 08 .013 PMID: 26896202
8. Finkelstein EA , Trogdon JG , Cohen JW , Dietz W. Annual medical spending attributable to obesity: payer-and service-specific estimates . Health Aff (Millwood) . 2009 ; 28 ( 5 ):w822± 831 .
9. Arterburn DE , Maciejewski ML , Tsevat J . Impact of morbid obesity on medical expenditures in adults . Int J Obes (Lond) . 2005 ; 29 ( 3 ): 334 ± 339 .
10. Tsai AG , Williamson DF , Glick HA . Direct medical cost of overweight and obesity in the USA: a quantitative systematic review . Obes Rev . 2011 ; 12 ( 1 ): 50 ± 61 . https://doi.org/10.1111/j. 1467 - 789X . 2009 . 00708 . x PMID : 20059703
11. Rudisill C , Charlton J , Booth HP , Gulliford MC . Are healthcare costs from obesity associated with body mass index, comorbidity or depression? Cohort study using electronic health records . Clin Obes . 2016 ; 6 ( 3 ): 225 ± 231 . https://doi.org/10.1111/cob.12144 PMID: 27097821
12. Moriarty JP , Branda ME , Olsen KD , Shah ND , Borah BJ , Wagie AE , et al. The effects of incremental costs of smoking and obesity on health care costs among adults: a 7-year longitudinal study . J Occup Environ Med . 2012 ; 54 ( 3 ): 286 ± 291 . https://doi.org/10.1097/JOM.0b013e318246f1f4 PMID: 22361992
13. Ramachandran A , Snehalatha C , Viswanathan V , Viswanathan M , Haffner SM . Risk of noninsulin dependent diabetes mellitus conferred by obesity and central adiposity in different ethnic groups: a comparative analysis between Asian Indians, Mexican Americans and Whites . Diabetes Res Clin Prac . 1997 ; 36 ( 2 ): 121 ± 125 .
14. National Health Insurance Sharing Service (NHIS) . https://nhiss.nhis.or.kr/bd/ab/bdaba006cv.do#
15. Seong SC , Kim YY , Park SK , Khang YH , Kim HC , Park JH , et al. Cohort profile: the National Health Insurance Service-National Health Screening Cohort (NHIS-HEALS) in Korea . BMJ Open . 2017 ; 7 ( 9 ): e016640. https://doi.org/10.1136/bmjopen-2017 -016640 PMID: 28947447
16. Song HJ , Lee EK , Kwon JW . Gender Differences in the Impact of Obesity on Health-Related Quality of Life . Asia Pac J Public Health . 2016 ; 28 ( 2 ): 146 ± 56 . https://doi.org/10.1177/1010539515626267 PMID: 26809970
17. Shin HY , Kang HT . Recent trends in the prevalence of underweight, overweight, and obesity in Korean adults: The Korean National Health and Nutrition Examination Survey from 1998 to 2014 . J Epidemiol . 2017 ; 27 ( 9 ): 413 ± 419 . https://doi.org/10.1016/j.je. 2016 . 08 .014 PMID: 28420559
18. Yi SW , Ohrr H , Shin SA , Yi JJ . Sex-age-specific association of body mass index with all-cause mortality among 12.8 million Korean adults: a prospective cohort study . Int J Epidemiol . 2015 ; 44 ( 5 ): 1696 ± 705 . https://doi.org/10.1093/ije/dyv138 PMID: 26208807
19. Guh DP , Zhang W , Bansback N , Amarsi Z , Birmingham CL , Anis AH . The incidence of co-morbidities related to obesity and overweight: a systematic review and meta-analysis . BMC Public Health . 2009 ; 9 : 88 . https://doi.org/10.1186/ 1471 -2458-9-88 PMID: 19320986
20. Korean Endocrine Society. Management of Obesity , 2010 Recommendation. Endocrinol Metab . 2010 ; 25 ( 4 ): 301 ± 304 .
21. Quan H , Li B , Couris CM , Fushimi K , Graham P , Hider P , et al. Updating and validating the Charlson comorbidity index and score for risk adjustment in hospital discharge abstracts using data from 6 countries . Am J Epidemiol . 2011 ; 173 ( 6 ): 676 ± 682 . https://doi.org/10.1093/aje/kwq433 PMID: 21330339
22. Manning WG , Mullahy J . Estimating log models: to transform or not to transform? J Health Econ . 2001 ; 20 ( 4 ): 461 ± 494 . PMID: 11469231
23. Mokdad AH , Ford ES , Bowman BA , Dietz WH , Vinicor F , Bales VS , et al. Prevalence of obesity, diabetes, and obesity-related health risk factors , 2001 . JAMA. 2003 ; 289 ( 1 ): 76 ± 9 . PMID: 12503980
24. Yatsuya H , Li Y , Hilawe EH , Ota A , Wang C , Chiang C , et al. Global trend in overweight and obesity and its association with cardiovascular disease incidence . Circ J . 2014 ; 78 ( 12 ): 2807 ± 2818 . PMID: 25391910
25. Flegal KM , Kit BK , Orpana H , Graubard BI . Association of all-cause mortality with overweight and obesity using standard body mass index categories: a systematic review and meta-analysis . JAMA . 2013 ; 309 ( 1 ): 71 ± 82 . https://doi.org/10.1001/jama. 2012 .113905 PMID: 23280227
26. Harrington M , Gibson S , Cottrell RC . A review and meta-analysis of the effect of weight loss on allcause mortality risk . Nutr Res Rev . 2009 ; 22 ( 1 ): 93 ± 108 . https://doi.org/10.1017/S0954422409990035 PMID: 19555520
27. Adams KF , Leitzmann MF , Ballard-Barbash R , Albanes D , Harris TB , Hollenbeck A , Kipnis V . Body mass and weight change in adults in relation to mortality risk . Am J Epidemiol . 2014 ; 179 ( 2 ): 135 ± 144 . https://doi.org/10.1093/aje/kwt254 PMID: 24173550
28. Corrada MM , Kawas CH , Mozaffar F , Paganini-Hill A . Association of Body Mass Index and Weight Change with All-Cause Mortality in the Elderly . American Journal of Epidemiology . 2006 ; 163 ( 10 ): 938 ± 949 . https://doi.org/10.1093/aje/kwj114 PMID: 16641311
29. Droyvold WB , Lund Nilsen TI , Lydersen S , Midthjell K , Nilsson PM , Nilsson JA , et al. Weight change and mortality: the Nord-Trondelag Health Study . J Intern Med . 2005 ; 257 ( 4 ): 338 ± 345 . https://doi.org/ 10.1111/j.1365- 2796 . 2005 . 01458 . x PMID : 15788003
30. National Statistics for obesity prevalence . http://www.index.go.kr/potal/enaraIdx/idxField/userPageCh. do?chkURL=/potal/stts/idxMain/selectPoSttsIdxMain. do%3Fclas_div%3DC%26idx_cd%3D2705% 26bbs%3DINDX_001&idx_cd=2705 & Title =%EB%B9%84%EB%A7%8C%EC%9C%A8&playurlstr = http://www.index.go.kr/potal/main/EachDtlPageDetail.do%3Fidx_cd%3D2932