Health, 2018, 10, 1044-1065

http://www.scirp.org/journal/health
ISSN Online: 1949-5005
ISSN Print: 1949-4998

Does High Systolic Blood Pressure Truly

Increase Medical Expenditure?
—An Empirical Analysis of the New 2017 ACC/AHA Hypertension Guideline

Kazumitsu Nawata1.2, Moriyo Kimura3

1
Graduate School of Engineering, University of Tokyo, Tokyo, Japan
2
Research Institute of Economy, Trade and Industry (RIETI), Tokyo, Japan
3
The Public Health Institute, Tokyo, Japan

How to cite this paper: Nawata, K. and Abstract

Kimura, M. (2018) Does High Systolic
Blood Pressure Truly Increase Medical Background: High blood pressure (BP) or hypertension is considered one of
Expenditure? Health, 10, 1044-1065. the top global disease burden risk factors. In November 2017, the ACC/AHA
https://doi.org/10.4236/health.2018.108079 and other organizations announced a new hypertension guideline of 130/80
mmHg. Data and Methods: We evaluate the effects of BP on increases in
Received: July 6, 2018
Accepted: August 7, 2018 medical expenditures using transformation tobit models and a dataset con-
Published: August 10, 2018 taining 175,123 medical checkups and 6,312,125 receipts from 88,211 indi-
viduals in three health insurance societies. The sample period was April 2013
Copyright © 2018 by authors and to March 2016. We first created a database of combined checkup results and
Scientific Research Publishing Inc.
This work is licensed under the Creative
medical expenditures. The power transformation tobit model was then used
Commons Attribution International to remove the effects of other variables, and we investigated the relation be-
License (CC BY 4.0). tween medical expenditures and BP, especially systolic BP (SBP). Results: We
http://creativecommons.org/licenses/by/4.0/ observed negative effects of SBP on medical expenditures. The results raise
Open Access uncertainty about the reliability of the new guideline, at least for SBP. Al-
though the simple correlation coefficient of medical expenditures and SBP
was positive, the sign of the SBP estimate became negative when a variable
representing obesity was included. In terms of other medical checkup items,
while LDL is considered the “bad” cholesterol, it reduced medical expendi-
tures. Conclusion: Our results did not support the new 2017 ACC/AHA
guideline for SBP. A wide and careful range of reviews not only for heart dis-
eases but also for other disease types will be absolutely necessary. New studies
to verify the guideline should also be conducted. Limitations: The dataset
was observatory, the sample period only 3 years, and we could not complete a
time-series analysis of individuals.

Keywords
Blood Pressure, Hypertension, 2017 ACC/AHA Hypertension Guideline,

DOI: 10.4236/health.2018.108079 Aug. 10, 2018 1044 Health

 K. Nawata, M. Kimura

Health and Medical Checkup

1. Introduction
The World Health Organization (WHO) [1] reports that, “Five of the top 10 se-
lected global disease burden risk factors identified by World Health Report 2002:
reducing risks, promoting healthy life—obesity, high blood pressure, high cho-
lesterol, alcohol and tobacco.” In terms of blood pressure (BP), the WHO [2]
states: “Worldwide, raised blood pressure is estimated to cause 7.5 million
deaths, about 12.8% of the total of all deaths. This accounts for 57 million disa-
bility adjusted life years (DALYS) or 3.7% of total DALYS.” It also specified
“complication of hypertension accounted for 9.4 million deaths in each year”
[3]. Lim et al. [4] also found that hypertension was one of the three leading risk
factors for the global disease burden (GDB) in 2010. Guidelines stated individu-
als should be treated for high BP or hypertension when their systolic BP (SBP;
during the heart beat) is 140 mmHg or more, or their diastolic BP (DBP, when
the heart is at rest) is 90 mmHg or more [5]. (For problems with BP measure-
ment errors due to the white-coat effect and other factors, see [6] [7] [8] [9]).
The National Heart, Lung and Blood Institute [10] classified hypertension as
Stage 1 (SBP of 140 - 159 mmHg; DBP of 90 - 99 mmHg) and Stage 2 (SBP of
160 mmHg or more; DBP of 100 mmHg). It has been reported that the risk of
cardiac diseases increased with BP level in all age groups [11] [12]. WHO and
the International Society of Hypertension (ISH) [13] provide guidelines for con-
trolling hypertension WHO-ISH classified hypertension into three categories:
Grade 1 (mild) - SBP of 140 - 159 mmHg, DBP of 90 - 99 mmHg; Grade 2
(moderate) - SBP of 160 - 179 mmHg, DBP of 100-109 mmHg; and Grade 3 (se-
vere) - SBP of 180 mmHg or more, DBP of 110 mmHg or more. Based on BP
and other risk factors, they concluded that the risk of a major cardiovascular
event over a 10-year period was about 20% - 30% in high-risk and 30% or more
in high- and very-high-risk groups.
The Prospective Studies Collaboration [14] performed a meta-analysis of in-
dividual data for 1,000,000 adults in 61 prospective analyses. They studied 12.7
million person-years at risk. They identified about 56,000 cases of vascular
death, including 12,000 stroke, 34,000 ischaemic heart disease (IHD), and other
vascular deaths. They found that IHD mortality increased in all age cohorts
(from 40 - 49 to 80 - 89) as SBP and DBP increased. Moreover, hypertension lo-
wered quality of life (QOL) [15] [16], and the true cost including indirect cost
such as effects on QOL is considered to be much higher than the direct cost. It is
estimated that hypertension caused 4.5% of the current GDB [17], suboptimal
BP cost $370 (US $) billion globally in 2001 [18], and hypertension cost the U.S.
about $51.2 billion per year in 2012-2013 [19] [20]. The Centers for Disease
Control and Prevention (CDC) [21] estimated high BP costs the U.S. $48.6 bil-

DOI: 10.4236/health.2018.108079 1045 Health

K. Nawata, M. Kimura

lion per year. The medical expenditure for hypertension and related diseases
reached as high as 1.85 trillion yen in fiscal year 2015 in Japan [22]. Almost all
studies agree that hypertension is a major health risk factor and costly disease.
Race, genetic and environmental factors, and health administrative activities
are important factors for BP [7] [23] [24]. For example, Rose [25] compared the
distribution of SBP in Kenyan nomads and London civil servants, and found
that the former was much lower than the latter. Nawata et al. [26] reported that
various factors such as age, gender, health conditions and lifestyle strongly af-
fected BP levels in Japan. They found that SBP increased about 5 mmHg with 10
years of increased age, and was 4 mmHg higher in males. These facts raised
questions about the results of previous studies. For example, the Prospective
Studies Collaboration study [14] selection criteria for the analyses did not clarify
the reasons for the 10-year age cohort interval, or the methods for removing ef-
fects of individual characteristics other than BP.
Many hypertension patients have few or no subjective symptoms. The distri-
bution of BP is close to the normal distribution, and even becomes a statistical
example of normal distribution [27], while the definition of hypertension is ar-
tificial and may be altered in the future by the accumulation of new medical
knowledge.
More recently, the American College of Cardiology (ACC), American Heart
Association (AHA), and nine other organizations [28] published the “2017
ACC/AHA/AAPA/ABC/ACPM/AGS/APhA/ASH/ASPC/NMA/PCNA Guideline
for the Prevention, Detection, Evaluation, and Management of High Blood Pres-
sure in Adults” (hereafter, 2017 ACC/AHA guideline). Under the new guideline,
the threshold for hypertension, requiring treatment with lifestyle changes and
medication, is 130/80 mmHg. The 2017 ACC/AHA guideline replaces the term
“prehypertension” with “elevated BP” (SBP 120 - 129 mmHg and DBP below 80
mm Hg) and “stage 1 hypertension” (SBP 130 - 139 mmHg or DBP 80 - 89
mmHg). Stage 2 hypertension is defined as SBP of 140 mmHg or more, or DBP
of 90 mmHg or more (replacing BP of 160/100 mmHg or more [28] [29] [30]).
However, the American Academy of Family Physicians (AAFP), one organiza-
tion asked to join the publication, announced on December 12, 2017, that
“AAFP Decides to Not Endorse AHA/ACC Hypertension Guideline, Academy
Continues to Endorse JNC8 Guideline” [31].
As a reason for withholding its endorsement, the AAFP stated that the new
guideline lacked a systematic review, and would classify 46% of the U.S. adult
population as having hypertension, a 32% increase. Dr. O’Gurek, chair of the
AAFP’s Commission on Health of the Public and Science added: “although the
guideline’s recommendations were given an evidence quality grade, they weren’t
grounded in an assessment of the background resources. Finally, substantial
weight was given to the Systolic Blood Pressure Intervention Trial (SPRINT),
but other trials were minimized.” Actually, the word “SPRINT” was used many
times in the main text of the 2017 ACC/AHA guideline. In the SPRINT [32],

DOI: 10.4236/health.2018.108079 1046 Health

 K. Nawata, M. Kimura

9361 individuals with a SBP of 130 mmHg or higher and an increased cardi-
ovascular risk, but without diabetes, were randomly assigned to a SBP target of
less than 120 mmHg (intensive treatment) or a target of less than 140 mmHg
(standard treatment). The trial period was from November 2010 to March 2013,
and was stopped early after a median follow-up of 3.26 years owing to a signifi-
cantly lower rate in the intesive treatment group.
As Nawata, Sekizawa and Kimura [33] point out, however, biases such as a
publication [34] [35] [36], conflict of interest [37] [38] [39], and termination (or
endpoint) biases [40] might exist in randomized clinical trails (CRT). The
change of the hypertension guideline will affect the medical system worldwide.
As suggested by the AAFP [31], careful and wide-ranging reviews are needed to
determine the hypertension guideline. The Action to Cardiovascular Risk in Di-
abetes (ACCORD) Study Group [41] did a similar trial for 4733 participants
with type 2 diabetes. In that study, 491 were recruited from January-June 2001,
and an additional 4242 were recruited from January 2003 through October 2005.
The researchers concluded that targeting SBP below 120 mmHg did not reduce
the rate of composite outcome of fatal and major nonfatal major cardiovascular
events compared to below 140 mmHg. The SPRINT research group [32] them-
selves admitted: “Rates of serious adverse events of hypertension, ... and acute
kidney injury, … were higher in the intensive-treatment group than in the stan-
dard-treatment group.” In other words, further investigation as to whether the
new guideline is appropriate or not are absolutely necessary. Such research must
include not only cardiovascular diseases but also all other (serious) diseases.
Since the influence of the AHA/ACC Guideline is so vast, these studies must be
done as soon as possible.
In this paper, we analyze the effects of BP (especially SBP) on annual medical
expenditures using a dataset containing 175,123 medical checkups and 6,312,125
receipts obtained from 88,211 individuals. The power transformation tobit mod-
el [42] is used in the analysis, because the distribution of medical expenditure
shows a heavy tail on the right side and many “zeroes” are observed. Although
we did not directly analyze heart diseases, medical expenditure is a very impor-
tant indicator representing the health conditions of an individual. The sample
period is from fiscal year 2013 to 2015 (i.e., April 2013 to March 2016).

2. Data and Methods

In this paper, we used an anonymized dataset combining medical checkups and
receipts. First, we compared the distributions of medical expenditures for each
fiscal year. Various characteristics and health conditions, including BP, affect
medical expenditures. To measure the effects of BP correctly, it is necessary to
remove the effects of other variables; we therefore employed a regression-type
analysis. However, there are problems, as previously identified by Gregori et al.
[43]. One is that medical expenditures take many zero values (about 20%). The
other is that the distribution is asymmetric and has a very heavy tail, and the va-

DOI: 10.4236/health.2018.108079 1047 Health

K. Nawata, M. Kimura

riance becomes very large. It may not be best to use the ordinary least squares
methods in such cases. We therefore used the power transformation tobit model
for the analysis.

2.1. Data
Japan employs a public health insurance system for the whole nation , and eve-
ryone must join some type of public health insurance organization. Corpora-
tions form heath insurance societies for employees and their family members.
Most employees 40 or older are required to have a medical checkup once a year
by the Industrial Safety and Health Act [44], and family members can voluntari-
ly have such checkups. The dataset was created with the cooperation of three
health insurance societies (Societies 1 - 3).
Society 1 was formed at a large Japanese corporation with offices and opera-
tional centers throughout Japan. Societies 2 and 3 were formed by groups of
corporations. The dataset contained information regarding 175,123 medical
checkups from 88,211 individuals between fiscal years 2013-2015 (i.e., April
2013 to March 2016). The monthly reports of medical treatments and payments,
called “receipts”, are sent from medical institutes to the health insurance associ-
ations. Payments are made to the medical institutes after checking the receipts.
(According to the Health Insurance Claims Review & Reimbursement Services
[45], nearly 99% of receipts were paid as requested in March 2018.) Receipts
were classified into five categories: dental; inpatients of DPC hospitals; outpa-
tients and inpatients of non-DPC hospitals; and pharmacies. Of these, we used
the sum of DPC, outpatient & non-DPC hospital, and pharmacy receipts as the
medical expenditure.
Japan measures medical expenditures in points, paying 10 yen per point to
medical institutes. Moreover, the same points are allotted for the same treat-
ments and medicines determined by the government regardless of region and
medical institution, with a few exceptions [46]. This means that medical expend-
iture is a good indicator of the health condition of an individual in Japan. A total
of 6,312,125 receipts were summed, and medical expenditures in each fiscal year
were calculated. In the analysis, we used the dataset containing 175,123 observa-
tions for which both the results of checkups and medical expenditures were
available in the same fiscal year.

2.2. Power Transformation Tobit Model

Many studies of medical expenditures use various types of regression analyses
[43] [47]. Medical expenditures do not become negative values (left censored),
and there exist many zero values. Moreover, the distribution is asymmetric,
having a very heavy tail on the right side, and variance is very large. Therefore,
analysis by a standard regression model and ordinary least squares methods may
not be the best in this case. We therefore used the power transformation model
in this analysis. We briefly explain the power transformation tobit model.

DOI: 10.4236/health.2018.108079 1048 Health

 K. Nawata, M. Kimura

1) Tobit model
The standard tobit model (or censored regression model) is given by
( )
xi′β + ui , ui ~ N 0, σ 2 , i =
yi* = 1, , n (1)

 yi* if yi* > 0

yi =  *
0 if yi ≤ 0

where yi* is a latent variable and its value is not observable when it is negative,
xi is a vector of explanatory variables, β is a vector of unknown parameters,
and ui is an error term following the normal distribution with mean 0 and va-
riance σ 2 . For more detail, see Amamya [48].
2) Power transformation
In the tobit model, normality of the error term is assumed. However, the
medical expenditures follow the distribution with a heavy tail on the right side,
and do not follow normal distribution. Since the medical expenditure takes zero
values, we cannot use the log transformation. In this study, we use the power
transformation to make the error term cross to the normal distribution. Gregori
et al. [43] considered the Box-Cox transformation [49], including both log and
power transformations. However, the log transformation is not included because
of zero values; the power transformation is sufficient in this study. The power
transformation is given by:
y Mα , 0 <α ≤1
= (2)

where M is the medical expenditure and α is the transformation parameter.

3) Power transformation tobit model
In this paper, we use the power transformation tobit model that combines the
tobit model and power transformation. Here,
dy
= α M α −1 . (3)
dM
Therefore, the likelihood function to be maximized is obtained by
 1 ( y − x′β ) 
2

log L (θ ) = ∑ M >0 − log ( 2π ) − log (σ ) − + log ( α ) + ( α − 1) log ( M ) 

 2 2σ 2 
  x′β  
+ ∑ M =0 log 1 − Φ  
  σ 
(4)
where Φ is the standard normal distribution function. We consider several
different models, and they are explained with the estimation results. Note that
when α is given, β and σ 2 are obtained by the standard tobit model. The
maximum likelihood estimator (MLE) is calculated by the following scanning
method [50]:
1) Choose α1 < α 2 <  < α m from 0.01-1.0 at intervals of 0.01.
2) Calculate β̂ and σˆ 2 for each αby the tobit maximum likelihood me-
thod.

DOI: 10.4236/health.2018.108079 1049 Health

K. Nawata, M. Kimura

3) Choose α̂ that maximizes the BC likelihood function.

4) Choose α i in the neighborhood of α̂ with an interval of 0.0001, and re-
peat steps (2) and (3).
5) Determine the final estimator.

3. Distributions of Medical Expenditures and Blood

Pressures
3.1. Medical Expenditures
Figure 1 shows the distribution of medical expenditures. The distribution is
skewed and has a very heavy tail on the right side. The basic statistics (points)
are as follows: mean: 13,356, median: 4061, standard deviation (SD): 39,241,
skewness: 11.0, kurtosis: 174.0, and maximum: 1,212,291. A total of 20.2% of all
observations of medical expenditures are zero. On the other hand, 1.9, 0.4, and
0.16% used more than 100,000, 300,000 and 500,000 points, and their medical
expenditures accounted for 30.3, 14.3 and 7.8%, respectively, of total medical
expenditures.

3.2. SBP and DBP

Figures 2-4 present the distributions of SBP and DBP, respectively. Excluding
observations of BP that are too high (SBP > 300 or DBP > 200) or too low (SBP,
DBP < 30), the basic statistics of SBP and DBP of 175,083 observations are given
in Table 1. Under the 140/90 criterion, 22.8% are diagnosed with hypertension.
Under the new guideline of 130/80, this value jumps up to 51.1%, more than a
half of observations, suggesting the effect of changing the criterion is quite large.

3.3. Relation between SBP and Medical Expenditures

Figure 5 shows the relation of SBP to average medical expenditures. Average
medical expenditures are averages of SBP at intervals of 5 mmHg (i.e., for a SBP

No. of cases

50000
45000
40000
35000
30000
25000
20000
15000
10000
5000
0

medical expenditure (points)

Figure 1. Distribution of medical expenditures.

DOI: 10.4236/health.2018.108079 1050 Health

 K. Nawata, M. Kimura

No. of observations

25000

20000

15000

10000

5000

-80
80-85
85-90
90-95
95-100
100-105
105-110
110-115
115-120
120-125
125-130
130-135
135-140
135-145
145-150
150-155
155-160
160-165
165-170
170-175
175-180
180-185
185-190
mmHg

Figure 2. Distribution of SBP.

No. of observations

35000
30000
25000
20000
15000
10000
5000
0
-40
40-45
45-50
50-55
55-60
60-65
65-70
70-75
75-80
80-85
85-90
90-95
95-100
100-105
105-110
110-115
115-120
120-125
125-130
130-135
135-140
140-
mmHg

Figure 3. Distribution of DBP

Table 1. Summaries of SBP and DBP.

SBP DBP

Mean 125.7 77.7

Median 125 78
Maximum 278 172
Minimum 38.5 30

SD 17.1 12.0

Skewness 0.521 0.282

Kurtosis 4.128 3.454

Observations 175,083

DOI: 10.4236/health.2018.108079 1051 Health

K. Nawata, M. Kimura

Figure 4. Relationship of SBP and DBP, and 130/80 and 140/90 criteria

Average medical expenditure(points)

25000
20000
15000
10000
5000
0
80 100 120 140 160 180
SBP (mmHg)

Figure 5. Relationship of average medical expenditure and SBP.

value of 130 mmHg, average medical expenditures of observations between

127.5 - 132.5 mmHg). The figure shows an upward trend, and the correlation
coefficient between SBP and average medical expenditure between 80 - 180
mmHg is 0.843. This result seems to support the new guideline. But the question
is whether this relation is a true or spurious one.
As already mentioned, various factors affect medical expenditures and BP.
Figure 6 shows the relationships between the average medical expenditures at
intervals of 5 years by gender. As an individual ages, medical expenditure in-
creases, and there is a difference between males and females. Nawata et al. [26]
pointed out that BP is strongly affected by age and gender. SBP increases by 5
mmHg over a 10-year aging period, and the SBP of males is about 4 mmHg
higher than that of females. BP becomes higher as an individual grows older. In
the next section, we conduct the analysis by the transformation tobit model.

DOI: 10.4236/health.2018.108079 1052 Health

 K. Nawata, M. Kimura

points
35000
30000 Male
25000 Female
20000
15000
10000
5000
0
40 50 60 70
Age

Figure 6. Medical expenditures by age and gender.

4. Results of Analysis by the Power Transformation Model

Models and Explanatory Variables
Regression models are used to remove the effects of various factors. We first
consider the following the power transformation tobit model in Equation (1).
Model A:
yi* = β1 + β 2 Age + β3 Female + β 4 Height + β5 BMI + β 6 SBP + β 7 DBP
+ β10 HDL + β9 LDL + β10Triglyceride + β11GGP + β12 AST + β13 ALT
(5)
+ β14 Boold _ Sugar + β14Urine_sugar + β15Urin_protein
+ β16 F_year14 + β17 F_year15 + β18 Society 2 + β19 Socitey3 + u

Besides SBP and DBP (mmHg), the following explanatory variables are used.
Age, Female (1: if female, 0: otherwise), Height (cm), BMI ( = height (m)/weight
(kg)2), HDL (high density lipoprotein cholesterol blood, mg/dL), LDL
(low-density lipoprotein cholesterol, mg/DL), Triglyceride (mg/dL), GGP
(γ-glutamyl transferase, U/L), AST (aspartate aminotransferase, U/L), ALT (ala-
nine aminotransferase, U/L), Blood_sugar (mg/dL), Urine_sugar (integers of
1-5, sugar in urine increasing with number; 1 is normal, 5 is worst),
Urine_protein (same as Urine_sugar), F_year 14 (1: fiscal year 2014, 0: other-
wise), F_year 2015 (1: fiscal year 2015), Society 2 (1: Society 2, 0: otherwise) and
Society3 (1: Society 3, 0: otherwise) where U/L is units per liter.
For all explanatory variables, objectively measured values could be obtained
from medical checkup data. This model did not include variables related to
anamnesis, currently treated diseases, or individual lifestyles. For example,
hypertension is an important risk factor of diabetes [51]. Suppose that the rela-
tion may be “hypertension = > diabetes = > medical expenditure”. In this case, if
a variable representing diabetes is included, the relation of “hypertension = >
medical expenditure” could not be observed. In econometric terms, we used the
reduced form so as not to miss any possible effects of BP.
Age, Female and Height represent basic individual characteristics; BMI

DOI: 10.4236/health.2018.108079 1053 Health

K. Nawata, M. Kimura

represents obesity; while HDL, LDL and Triglyceride represent lipid concentra-
tion in the blood. If lipid concentration is abnormal (too high or too low), an in-
dividual is diagnosed as dyslipidemia. Lipoproteins are proteins that carry cho-
lesterol through the blood. LDL cholesterol makes up most of the body’s choles-
terol, and HDL cholesterol absorbs cholesterol and carries it back to the liver
[52]. Triglyceride is the most common type of fat in the body, and stores excess
energy [53]. Although our bodies need lipids to build cells, too much could be a
problem [54].
Currently, dyslipidemia is mainly hyperlipidemia, where the lipid concentra-
tion is too high. WHO [55] warned: “Raised cholesterol increases the risks of
heart disease and stroke. Globally, a third of ischemic heart disease is attributa-
ble to high cholesterol. Overall, raised cholesterol is estimated to cause 2.6 mil-
lion deaths (4.5% of total) and 29.7 million disability adjusted life years
(DALYS), or 2.0% of total DALYS.” LDL and HDL cholesterols are classified as
“bad” and “good”. LDL (bad) cholesterol contributes to fatty buildups in arte-
ries, and raises the risk factor for chronic coronary heart disease, heart attack
and stroke. On the other hand, HDL (good) cholesterol removes LDL cholesterol
from the arteries [52] [53]. GGP, AST and ALT are mainly related to liver func-
tions; Blood_sugar and Urine_sugar are important indicators of diabetes; and
Urine_protein represents the condition of the kidneys [56].
We first excluded observations with missing values in explanatory variables.
We then excluded the following observations: BMI too high (over 100); SBP too
high (over 300) or too low (under 30); DBP too high (over 200) or too low (un-
der 30); SBP-DBP becomes zero or negative; HDL too high (over 500); LDL too
high (over 500); Triglyceride too high (over 1000); GGT too high (over 1000);
ALT too high (over 500); AST too high (over 500); and Blood_sugar too high
(500). Excluding observations with missing values in explanatory variables, we
used 173,498 (M > 0: 138,407, and M = 0: 35,091) observations for the estimation
of the model. Among these observations, 20.2% of medical expenditures were
zero, and 79.8% were positive values.
Model A assumes that the effects of BP are continuous. However, it is possible
that BP affects health conditions only if it becomes higher than certain threshold
values (hereafter, threshold value hypothesis; criteria such as 140/90 and 130/80
are obviously based on this hypothesis). Therefore, we consider the model using
dummy variables of SBP. Note that we analyzed only SBP as the SPRINT.
Model B:
yi* = β1 + β 2 Age + β3 Female + β 4 Height + β5 BMI + β 6 SBP130
+ β 7 SBP140 + β8 SBP160 + β9 SBP180 + β10 DBP + β11 HDL
+ β12 LDL + β13Triglyceride + β14GGP + β15 AST + β16 ALT (6)
+ β17 Boold _ Sugar + β18Urine_sugar + β19Urin_protein
+ β 20 F_year14 + β 21 F_year15 + β 22 Society 2 + β 23 Socitey3 + u

SBP130 (1: if SPB ≥ 130, 0:otherwise), SBP140 (1: if SPB ≥ 140, 0:otherwise),
SBP160 (1: if SPB ≥ 160, 0:otherwise) and SBP180 (1: if SPB ≥ 180, 0:otherwise)

DOI: 10.4236/health.2018.108079 1054 Health

 K. Nawata, M. Kimura

are dummy variables representing threshold values. Table 2 presents a summary

of the explanatory variables.
Table 3 lists the result of estimations for Model A. Figure 7 shows the distri-
bution of the medical expenditures after the power transformation ( y = M 0.4088 ).
The distribution is much closer to the normal distribution, suggesting usefulness
of the model for analyzing this dataset. Since the sample size was quite large, all
variables except F_year 14 were significant at the 1% level. The estimates of Age,
Female, Height, BMI, Triglyceride, GGT, AST, ALT, Blood_suger, Urine_suger,
Urine_protein, andF_year15 were positive, with these variables making medical

No. of observations

40000
35000
30000
25000
20000
15000
10000
5000
0

Figure 7. Distribution medical expenditures after the transformation (y = M 0.4088

).

Table 2. Explanatory variables.

Variable Mean SD Variable Mean SD

Age 50.25 7.76 AST 23.46 10.56

Female 1:25.0%, 0:75.0% ALT 24.77 17.40

Height 166.86 8.16 1:96.49%, 2:0.71%, 3:0.94%,

Urine_sugar
BMI 23.69 3.77 4:0.74%, 5:1.12%

1:91.58%, 2:4.78%, 3:2.60%,

SBP 125.72 17.09 Urine_protein
4:0.79%, 5:0.25%
2013: 27.7%, 2014:31.2%,
DBP 77.70 11.95 Fiscal year
2015: 41.1%
HDL 61.31 16.59 Societies 1: 65.5%, 2:16.1%, 18.4%

LDL 124.57 31.70 BP130 1:37.9%, 0:62.1%

Triglyceride 126.74 94.37 BP140 1:18.4%, 0:81.6%

Blood_sugar 63.41 48.81 BP160 1:3.55%, 0:96.43%

GGT 44.44 49.17 BP180 1:0.55%, 0:99.45%

SD: Standard deviation.

DOI: 10.4236/health.2018.108079 1055 Health

K. Nawata, M. Kimura

Table 3. Result of estimation (Model A).

Variable Estimate SE t-value

α 0.4088 0.0007 623.62**

Constant −59.9797 2.0011 −29.974**

Age 0.9576 0.0105 91.299**

Female 11.0988 0.2111 52.571**

Height 0.0857 0.0100 8.565**

BMI 1.3286 0.0199 66.796**

SBP −0.0566 0.0054 −10.425**

DBP 0.0316 0.0079 3.997**

HDL −0.0427 0.0042 −10.144**

LDL −0.1128 0.0020 −55.931**

Triglyceride −0.0026 0.0007 −3.638**

Blood_sugar 0.0236 0.0012 19.913**

Urine_sugar 5.0889 0.1067 47.692**

Urine_protein 4.2282 0.1007 41.986**

GGT 0.0179 0.0013 13.650**

AST 0.0770 0.0080 9.576**

ALT 0.0653 0.0057 11.463**

F_year14 0.1263 0.1608 0.786

F_year15 0.9614 0.1460 6.585**

Society2 −5.5580 0.1736 −32.016**

Society3 −5.3787 0.1618 −33.243**

σ 26.9985 0.2294 117.67**

logL −1,552,289

M > 0: 138407, M = 0:3501,

No. of observations
Total: 173,498

SE: standard error, **: significant at 1% level.

expenditures higher. The effects of most of these variables were as expected. On

the other side, the estimates of LDL, HDL, Society 2 and Society 3 were negative.
Although LDL Cholesterol is called “bad” and HDL “good” [53], higher levels of
both LDL and HDL cholesterols reduced medical expenditures in our study.
Hence further studies are necessary to determine the roles and functions of
cholesterols, especially LDL cholesterol. This is one important finding of this
study.
The medical expenditures of Societies 2 and 3 were lower than those of Society
1. Society 1 was formed by one large corporation, while Societies 2 and 3 were
formed by groups of smaller corporations. Although the reason why cannot be

DOI: 10.4236/health.2018.108079 1056 Health

 K. Nawata, M. Kimura

elucidated, it might be necessary to check and revise the healthcare system in

Society 1. Although the sign of DBP was positive and significant, the estimate of
SBP was −0.0566, and its t-values were −10.42 and significant at any reasonable
significance at any reasonable level. This means that higher SBP reduced medical
expenditures.
Table 4 presents the estimation results of Model B, which contained the thre-
shold value dummies for SBP. The values of estimations for variables other than
BP were very similar to those of Model A. For the SBP dummy variables,
SBP130, SBP140 and SBP180 dummies were not significant even at the 5% level,
despite the fact that the sample size was quite large. Only the SBP160 dummy
was significant at the 1% level, but the estimated value was negative. Although it
was not significant at the 5% level, the estimate of DBP becomes a negative value
in this model. These findings do not support the threshold value hypothesis, at
least for SBP.

5. Discussion
The effects of BP on medical expenditures are mixed. Higher DBP makes them
higher, but higher SBP makes them lower. We evaluated the relations between
medical expenditures and high SBP or SBP hypertension. As shown in Figure 5,
there exists an upward trend between SBP and average medical expenditures.
We consider a simple regression model of Equation (1) that is given by:
Model C:
yi* = β1 + β 2 SBP + u (7)

Then we get (standard errors are in parentheses),

0.4094 ( 0.0007 )
13.264 + 0.1308SBP, αˆ =
yi* =
(8)
( 0.466 ) ( 0.00367 )
The estimate of SBP is positive, and its t-value is 35.70 and significant at any
reasonable significance level. As shown in Figure 6, age and gender might affect
medical expenditures. We add Age and Female, and consider the model:
Model D:
yi* = β1 + β 2 Age + β3 Female + β 4 SBP + u (9)

The estimation results of this model are given by:

αˆ = 0.4088 ( 0.0007 )
yi* =
−24.016 + 0.8508 Age + 5.8150 Female + 0.0747 SBP (10)
( 0.5665) ( 0.0100 ) ( 0.15468) ( 0.00359 )
Although the size is almost half that of the previous case, the estimate of SBP
is still positive, and the t-value is 20.86 and significant at any reasonable level.
We then add BMI, representing obesity, and consider the model,
Model E:
yi* = β1 + β 2 Age + β3 Female + β 4 SBP + β5 BMI + u

DOI: 10.4236/health.2018.108079 1057 Health

K. Nawata, M. Kimura

Table 4. Result of estimation (Model B).

Variable Estimate SE t-value

α 0.4089 0.0007 625.23**

Constant −62.0945 2.0300 −30.588**

Age 0.9446 0.0104 90.919**

Female 11.1871 0.2111 52.991**

Height 0.0868 0.0100 8.680**

BMI 1.3088 0.0198 66.067**

SBP130 0.0421 0.1644 0.256

SBP140 −0.2291 0.1970 −1.162

SBP160 −1.3446 0.3487 −3.856**

SBP180 0.3567 0.7030 0.507

DBP −0.0188 0.2287 −0.082

HDL −0.0440 0.0042 −10.456**

LDL −0.1124 0.0020 −55.857**

Triglyceride −0.0029 0.0007 −3.935**
Blood_sugar 0.0240 0.0012 20.277**
Urine_sugar 5.0627 0.1065 47.556**

Urine_protein 4.2229 0.1007 41.942**

GGT 0.0179 0.0013 13.595**

AST 0.0759 0.0080 9.432**

ALT 0.0664 0.0057 11.642**

F_year14 0.1205 0.1644 0.733

F_year15 0.9640 0.1460 6.604**

Society2 −5.5515 0.1736 −31.981**

Society3 −5.3967 0.1618 −33.362**

α 27.031 0.2287 118.21**

logL −1552309

M > 0:138407, M = 0:3501,

No. of observations
Total:173498

SE: standard error, **: significant at 1% level.

The estimation result is given by:

αˆ = 0.4085 ( 0.0007 )
yi* =
−58.963 + 1.0416 Age + 9.0088 Female − 0.0250 SBP + 1.7028 BMI (11)
( 0.6973) ( 0.0101) ( 0.1554 ) ( 0.0035) ( 0.0186 )
In Model E, the coefficient of SBP becomes negative and significant at the 1%
level. Muntner et al. [57] analyzed data from the US National Health and Nutri-

DOI: 10.4236/health.2018.108079 1058 Health

 K. Nawata, M. Kimura

tion Examination Survey (NHANES). They pooled data from the 2011-2012 and
2013-2014 NHANES cycles of adult participants, 20 years of age and older (n =
10,907). They declared that, “Implementation of the 2017 ACC/AHA hyperten-
sion guideline has the potential to increase the prevalence of hypertension and
use of antihypertensive medication among U.S. adults. This should translate into
a reduction in CVD events.” Although age, gender, race, smoking, total and
HDL-cholesterol, and diabetes were included, “obesity” was not considered in
their analysis. The correlation coefficient of SBP and BMI is 0.307 in this study.
The relation between obesity and hypertension has been recognized for more
than a half century [58], and many studies have been conducted. For details, see
the review works of Kotchen [59], Jiang et al. [60] and Leggio et al. [61]. Jiang et
al. [60] declared that: “The mechanisms underlying obesity-associated hyperten-
sion or other associated metabolic diseases remains to be adequately investi-
gated.” They furthermore contended that, “There is no single cause to explain all
the cases of obesity worldwide.” The relation between BP and obesity should be
carefully studied.
The results of this study suggest that the risks of hypertension might be spu-
rious, and other factors such as obesity might be affecting health condition.
Moreover, BP has been found to affect not only heart diseases but also various
other health conditions such as kidney diseases [32] [62]. The influences of the
new guideline of 130/80 are so large that careful reviews of various studies in-
cluding analyses of various factors and diseases (not only heart diseases) affected
by BP levels are absolutely necessary to determine whether or not the new guide-
line is appropriate.

6. Conclusions
In this study, we mainly evaluated the effects of BP on medical expenditures by
the transformation tobit models using a dataset containing 175,123 medical
checkups and 6,312,125 receipts obtained from 88,211 individuals obtained from
three health insurance societies. Medical expenditure is a very good indicator of
an individual’s health condition, because under the current Japanese national
health insurance system, most medical institutes receive the same amount for
the same treatments and medicines, independent of region. We first considered
a model that included various heath information factors for individuals obtained
in yearly medical examinations. Although the estimate of DBP had a positive
value, that of SBP became negative and the absolute t-value was larger than 10,
suggesting that the new guideline for SBP was not supported.
We then theorized that threshold values and BP might affect health condition
only if BP exceeded those values (threshold value hypothesis). We used SBP
dummies to check the threshold value hypothesis, but the results did not support
the hypothesis for SBP. While the estimates of most other variables had expected
signs, LDL cholesterol, considered “bad”, showed the opposite result. It is likely
we will need additional studies for the evaluation of cholesterols.

DOI: 10.4236/health.2018.108079 1059 Health

K. Nawata, M. Kimura

We then evaluated the relation between medical expenditures and SBP. Med-
ical expenditures and SBP were positively correlated, and if the simple model
only contained SBP, the estimate became a positive value. Although the size of
the coefficient was almost cut in half, the sign did not change if age and gender
variables were considered in the model. However, when BMI, representing obes-
ity, was added, the estimate of SBP became negative and significant at the 1%
level.
It is possible that the relation between SBP and medical expenditures might be
spurious, and the correlation of SBP and BMI might affect the result. The rela-
tion between BP and obesity should be carefully studied. Moreover, the effect of
the new 2017 ACC/AHA guideline, the first comprehensive hypertension clinical
practice guideline since 2003 [30], is so large that a careful and wide range of re-
views of various studies, not only of heart diseases but for other types of diseases
as well, are absolutely necessary. New studies verifying the guideline should also
be conducted.
In this paper, we evaluated medical expenditures, not the risks of BP on heart
diseases. Evaluation of the effects of BP on heart diseases and other important
diseases is needed. It will also be necessary to analyze a larger and longer
time-range dataset from various insurance societies to make the analysis more
precise. These are subjects to be studied in future.

Acknowledgements
This study was supported by a Grant-in-Aid for Scientific Research, “Analyses of
Medical Checkup Data and Possibility of Controlling Medical Expenses (Grant
Number: 17H22509),” from the Japan Society of Science, and by a research
grant, “Exploring Inhibition of Medical Expenditure Expansion and Health-
oriented Business Management Based on Evidence-based Medicine” from the
Research Institute of Economics, Trade and Industry (RIETI). The dataset was
anonymized at the health insurance societies. This study was approved by the
Institutional Review Boards of the University of Tokyo (number: KE17-30). The
authors would like to thank the health insurance societies for their sincere coop-
eration in providing us the data. We would also like to thank an anonymous re-
feree for his/her helpful comments and suggestions.

Conflicts of Interest
The authors declare no conflicts of interest regarding the publication of this pa-
per.

References
[1] World Health Organization (WHO) (2013) Obesity, High Blood Pressure, High
Cholesterol, Alcohol and Tobacco: The World Health Organization’s Response.
World Health Report. http://www.who.int/whr/media_centre/factsheet3/en/
[2] WHO (2016) Raised Blood Pressure.

DOI: 10.4236/health.2018.108079 1060 Health

 K. Nawata, M. Kimura

http://www.who.int/gho/ncd/risk_factors/blood_pressure_prevalence_text/en/
[3] WHO (2013) A Global Brief on Hypertension.
http://www.who.int/cardiovascular_diseases/publications/global_brief_hypertensio
n/en/
[4] Lim, S.S., Vos, T., Flaxman, A.D., et al. (2012) A Comparative Risk Assessment of
Burden of Disease and Injury Attributable to 67 Risk Factors and Risk Factor Clus-
ters in 21 Regions, 1990-2010: A Systematic Analysis for the Global Burden of Dis-
ease Study 2010. Lancet, 380, 2224-2260.
https://doi.org/10.1016/S0140-6736(12)61766-8
[5] Blood Pressure Association, UK (2008) Blood Pressure Chart for Adults.
http://www.bloodpressureuk.org/BloodPressureandyou/Thebasics/Bloodpressurech
art
[6] Wohlfahrt, P., Cifková, R., Movsisyan, N., et al. (2016) Threshold for Diagnosing
Hypertension by Automated Office Blood Pressure Using Random Population.
Journal of Hypertension, 2180-2186.
https://doi.org/10.1097/HJH.0000000000001076
[7] Omiboni, S., Aristizabal, D., De la Sierra, A., et al. (2016) Hypertension Types De-
fined by Clinic and ambulatory Blood Pressure in 14142 Patients Referred to
Hypertension Clinics Worldwide. Data from ARTEMIS Study. Journal of Hyper-
tension, 34, 2187-2198. https://doi.org/10.1097/HJH.0000000000001074
[8] AACN Practice Alert (2016) Obtaining Accurate Noninvasive Blood Pressure Mea-
surements in Adults. Critical Care Nurse, 36, e12-e16.
https://doi.org/10.4037/ccn2016590
[9] Nawata, K. and Kimura, M. (2017) Reliability of Blood Pressure Measurements: An
Analysis of the White Coat Effect and Its Fluctuations. Health, 9, 506-519.
https://doi.org/10.4236/health.2017.93036
[10] National Heart, Lung and Blood Institute, US Department of Health & Human Ser-
vices (2015) Description of High Blood Pressure.
[11] Vasan, S.R., Larson, M.G., Leip, E.P., et al. (2001) Impact of High-Normal Blood
Pressure on the Risk of Cardiovascular Disease. New England Journal of Medicine,
345, 1291-1297. https://doi.org/10.1056/NEJMoa003417
[12] Fujiyoshi, A., Ohkubo, T., Miura, K., et al. (2012) Blood Pressure Categories and
Log-Term Risk of Cardiovascular Disease According to Age Group in Japanese Men
and Women. Hypertension Research, 35, 947-953.
https://doi.org/10.1038/hr.2012.87
[13] Guidelines Subcommittee (1999) 1999 World Health Organization - International
Society of Hypertension Guidelines for Management of Hypertension. Journal of
Hypertension, 17, 151-183.
[14] Prospective Studies Collaboration (2002) Age-Specific Relevance of Usual Blood
Pressure to Vascular Mortality: A Meta-Analysis of Individual Data for One Million
Adults in 61 Prospective Studies. Lancet, 360, 1903-1913.
https://doi.org/10.1016/S0140-6736(02)11911-8
[15] Blanchflower, D.G. and Oswald, A.J. (2008) Hypertension and Happiness across
Nations. Journal of Health Economics, 27, 218-233.
https://doi.org/10.1016/j.jhealeco.2007.06.002
[16] Mojon-Azzi, S. and Sousa-Pazo, A. (2011) Hypertension and Life Satisfaction: An
Analysis Using Data from the Survey of Health, Aging and Retirement in Europe.
Applied Economics Letters, 18, 183-187.

DOI: 10.4236/health.2018.108079 1061 Health

K. Nawata, M. Kimura

https://doi.org/10.1080/13504850903508291
[17] World Health Organization, International Society of Hypertension Writing Group
(2003) 2003 World Health Organization (WHO)/International Society of Hyperten-
sion (ISH) Statement on Management of Hypertension. Journal of Hypertension,
11, 1983-1992.
[18] Gaziano, T.A., Bitton, A., Anand, S., et al. (2009) The Global Cost of Nonoptimal
Blood Pressure. Journal of Hypertension, 27, 1472-1477.
https://doi.org/10.1097/HJH.0b013e32832a9ba3
[19] Benjamin, E.J., Blaha, M.J., Chiuve, S.E., et al. (2017) Heart Disease and Stroke Sta-
tistics—2017 Updates: A Report from the American Heart Association. Circulation,
135, e146-e603. https://doi.org/10.1161/CIR.0000000000000485
[20] Wang, G., Grosse, G.S.D. and Schooley, M.W. (2017) Conducting Research on the
Economics of Hypertension to Improve Cardiovascular Health. American Journal
of Preventive Medicine, 53, S115-S117.
https://doi.org/10.1016/j.amepre.2017.08.005
[21] Centers for Disease Control and Prevention (CDC) (2016) High Blood Pressure
Fact Sheet.
https://www.cdc.gov/dhdsp/data_statistics/fact_sheets/fs_bloodpressure.htm
[22] Ministry of Health, Welfare, and Labor (2017) Heisei 27 nendo kokumin iryouhi no
gayou (Summary of National Medical Expenditures). (In Japanese)
[23] Ibrahim, M.M. and Damasceno, A. (2012) A Hypertension in Developing Coun-
tries. The Lancet, 380, 611-619. https://doi.org/10.1016/S0140-6736(12)60861-7
[24] Gorman, B.K. and Porter, J.R. (2011) Social Networks and Support, Gender, and
Radical/Ethic Disparities in Hypertension among Older Adults. Population Re-
search & Policy Review, 20, 885-911. https://doi.org/10.1007/s11113-011-9215-4
[25] Rose, G. (2001) Sick Individuals and Sick Populations. International Journal of Ep-
idemiology, 30, 427-432. https://doi.org/10.1093/ije/30.3.427
[26] Nawata, K., Matsumoto, A., Kajihara, K., et al. (2017) Evaluation of the Distribution
of and Factors Affecting Blood Pressure Using Medical Checkup Data in Japan.
Health, 9, 124-137. https://doi.org/10.4236/health.2017.91009
[27] Surfstat.australia: An Online text in Introductory Statistics (2018) Summarising and
Presenting Data, Normal Distributions.
https://surfstat.anu.edu.au/surfstat-home/1-3-1.html
[28] Whelton, P.K., Carey, R.M., Arrow, W.S., et al. (2018) 2017 ACC/AHA/AAPA/ABC
/ACPM/AGS/APhA/ASH/ASPC/NMA/PCNA Guideline for the Prevention, Detec-
tion, Evaluation, and Management of High Blood Pressure in Adults: A Report of
the American College of Cardiology/American Heart Association Task Force on
Clinical Practice Guidelines. Journal of the American College of Cardiology, 71,
e13-e115.
[29] Whelton, P.K., Carey, R.M., Arrow, W.S., et al. (2018) 2017 Guild Line for Blood
Pressure in Adults. American College of Cardiology.
http://www.acc.org/latest-in-cardiology/ten-points-to-remember/2017/11/09/11/41/
2017-guideline-for-high-blood-pressure-in-adults
[30] Carey, R.M. and Whelton, P.K. for the 2017 ACC/AHA Hypertension Guideline
Writing Committee (2018) Prevention, Detection, Evaluation, and Management of
High Blood Pressure in Adults: Synopsis of the 2017 American College of Cardiol-
ogy/American Heart Association Hypertension Guideline. Annals of Internal Medi-
cine, 168, 351. https://doi.org/10.7326/M17-3203

DOI: 10.4236/health.2018.108079 1062 Health

 K. Nawata, M. Kimura

[31] American Academy of Family Physicians (AAFP) (2017) AAFP Decides to Not En-
dorse AHA/ACC Hypertension Guideline, Academy Continues to Endorse JNC8
Guideline. AAFP News.
https://www.aafp.org/news/health-of-the-public/20171212notendorseaha-accgdlne.
html
[32] Wright, J.T., Williamson, J.D., Whelton, P.K., et al. (2015) A Randomized Trial of
Intensive versus Standard Blood-Pressure Control. New England Journal of Medi-
cine, 373, 2103-2116. https://doi.org/10.1056/NEJMoa1511939
[33] Nawata, K., Sekizawa, Y. and Kimura, M. (2018) Evaluation of Blood Pressure Con-
trol Medicines Using Health and Medical Checkup Data in Japan: Alternative Me-
thods for Randomized Controlled Trials. Health, 10, 587-603.
https://doi.org/10.4236/health.2018.105047
[34] Dickersin, K., Chan, S.S, Chalmersx, T.C., et al. (1987) Publication Bias and Clinical
Trials. Controlled Clinical Trials, 8, 343-353.
https://doi.org/10.1016/0197-2456(87)90155-3
[35] Hopewell, S., Loudon, K., Clarke, M.J., et al. (2009) Publication Bias in Clinical Tri-
als Due to Statistical Significance or Direction of Trial Results. Cochrane Database
of Systematic Reviews, No. 1, MR000006.
http://sverigesradio.se/diverse/appdata/isidor/files/83/5775.pdf
[36] Turner, E.H., Matthews, A.M., Linardatos, E., et al. (2008) Selective Publication of
Antidepressant Trials and Its Influence on Apparent Efficacy. New England Journal
of Medicine, 358, 252-260. https://doi.org/10.1056/NEJMsa065779
[37] Lexchin, J., Bero, L.A., Djulbegovic, B.B., et al. (2003) Pharmaceutical Industry
Sponsorship and Research Outcome and Quality: Systematic Review. BMJ, 326,
1167. https://doi.org/10.1136/bmj.326.7400.1167
[38] Perlis, R.H., Perlis, C.S., Wu, Y., et al. (2005) Industry Sponsorship and Financial
Conflict of Interest in the Reporting of Clinical Trials in Psychiatry. American
Journal of Psychiatry, 162, 1957-1960. https://doi.org/10.1176/appi.ajp.162.10.1957
[39] Lundh, A., Lexchin, J., Mintzes, B., et al. (2017) Industry Sponsorship and Research
Outcome. Cochrane Database of Systematic Reviews, No. 12, MR000033.
http://cochranelibrary-wiley.com/doi/10.1002/14651858.MR000033.pub3/full
https://doi.org/10.1002/14651858.MR000033.pub3
[40] Park, T.R., Rodriguez, D.M. and Roth, F.P. (2015) Why Clinical Trials Are Termi-
nated.
https://www.biorxiv.org/content/biorxiv/early/2015/07/02/021543.full.pdf
[41] The ACCORD Study Group (2010) Effect of Intensive Blood-Pressure Control in
Type 2 Diabetes Mellitus. New England Journal of Medicine, 362, 1575-1585.
https://doi.org/10.1056/NEJMoa1001286
[42] Nawata, K. and Kimura, M. (2017) Evaluation of Medical Costs of Kidney Diseases
and Risk Factors in Japan. Health, 9, 1734-1749.
https://doi.org/10.4236/health.2017.913127
[43] Gregori, D., Petrinco, M., Simona, B., et al. (2011) Regression Models for Analyzing
Costs and Their Determinants in Health Care: An Introductory Review. Interna-
tional Journal for Quality in Health Care, 23, 331-341.
https://doi.org/10.1093/intqhc/mzr010
[44] Japanese Law Translation (2009) Industrial Safety and Health Act.
http://www.japaneselawtranslation.go.jp/law/detail/?id=1926&vm=&re
[45] Health Insurance Claims Review & Reimbursement Services (2018) Sinsa joukyou

DOI: 10.4236/health.2018.108079 1063 Health

K. Nawata, M. Kimura

heisei 30 nen 3 gatsu (Results of Checking, March 2018). (In Japanese)

[46] Nawata, K., Ii, M., Toyama, H., et al. (2009) Evaluation of the Inclusive Payment
System Based on the Diagnosis Procedure Combination with Respect to Cataract
Operations in Japan. Health, 1, 93-103. https://doi.org/10.4236/health.2009.12016
[47] Deng, L., Lou, W. and Mitsakakis, N. (2018) Modeling Right-Censored Medical
Cost Data in Regression and the Effects of Covariates. Statistical Methods & Appli-
cations, 1-13. https://doi.org/10.1007/s10260-018-0428-0
[48] Amemiya, T. (1985) Advanced Econometrics. Harvard University Press, Cam-
bridge.
[49] Box, G.E.P. and Cox, D.R. (1964) An Analysis of Transformation. Journal of the
Royal Statistical Society B, 26, 211-252.
[50] Nawata, K. (1994) Estimation of Sample Selection Bias Models by the Maximum
Likelihood Estimator and Heckman’s Two-Step Estimator. Economic Letters, 45,
33-40. https://doi.org/10.1016/0165-1765(94)90053-1
[51] Nawata, K. and Kimura, M. (2017) An Analysis of the Medical Costs of and Factors
Affecting Diabetes Using the Medical Checkup and Payment Dataset in Japan: Can
We Reduce the Prevalence of Diabetes? Health, 9, 1113-1127.
https://doi.org/10.4236/health.2017.98081
[52] CDC (2017) LDL and HDL Cholesterol: “Bad” and “Good” Cholesterol.
https://www.cdc.gov/cholesterol/ldl_hdl.htm
[53] American Heart Association (AEA) (2017) HDL (Good), LDL (Bad) Cholesterol
and Triglycerides.
http://www.heart.org/HEARTORG/Conditions/Cholesterol/HDLLDLTriglycerides/
HDL-Good-LDL-Bad-Cholesterol-and-Triglycerides_UCM_305561_Article.jsp#.W
xUKUkjRCUk
[54] AEA (2017) About Cholesterol.
http://www.heart.org/HEARTORG/Conditions/Cholesterol/AboutCholesterol/Abo
ut-Cholesterol_UCM_001220_Article.jsp#.Wtrw2ojFKUk
[55] WHO (2018) Raised Cholesterol.
http://www.who.int/gho/ncd/risk_factors/cholesterol_text/en/
[56] U.S. National Library of Medicine (2018) Medical Encyclopedia.
https://medlineplus.gov/encyclopedia.html
[57] Muntner, P., Carey, R.M., Gidding, S., et al. (2018) Potential U.S. Population Impact
of the 2017 ACC/AHA High Blood Pressure Guideline. Journal of the American
College of Cardiology, 71, 109-118. https://doi.org/10.1016/j.jacc.2017.10.073
[58] Kennel, W.B., Brand, N., Skinner, J.J., et al. (1967) The Relation of Adiposity to
Blood Pressure and Development of Hypertension: The Framingham Study. Annals
of Internal Medicine, 67, 48-59. https://doi.org/10.7326/0003-4819-67-1-48
[59] Kotchen, T.A. (2010) Obesity-Related Hypertension: Epidemiology, Pathophysiol-
ogy, and Clinical Management. American Journal of Hypertension, 23, 1170-1178.
https://doi.org/10.1038/ajh.2010.172
[60] Jiang, S.-Z., Lu, W., Zong, X.-F., et al. (2016) Obesity and Hypertension. Experi-
mental and Therapeutic Medicine, 12, 2395-2399.
https://doi.org/10.3892/etm.2016.3667
[61] Loggi, M., Lombardi, M. and Caldarone, E. (2017) The Relationship between Obes-
ity and Hypertension: An Comprehensive Overviews on Twins. Hypertension Re-
search, 40, 947-963. https://doi.org/10.1038/hr.2017.75

DOI: 10.4236/health.2018.108079 1064 Health

 K. Nawata, M. Kimura

[62] The National Institute of Diabetes and Digestive and Kidney Diseases, Health In-
formation Center (2014) Blood Pressure & Kidney Disease, What Is High Blood
Pressure?
https://www.niddk.nih.gov/health-information/kidney-disease/high-blood-pressure

DOI: 10.4236/health.2018.108079 1065 Health