A Note on Obesity as Epidemic in Korea

Article information

Osong Public Health Res Perspect. 2011;2(2):135-140
Publication date (electronic) : 2011 August 5
doi : https://doi.org/10.1016/j.phrp.2011.08.004
aDepartment of Mathematics, Kyungpook National University, Daegu, Korea
bDivision of Epidemic Intelligence Service, Korea Centers for Disease Control and Prevention, Osong, Korea
Corresponding author. Yongkuk@knu.ac.kr
Received 2011 May 23; Revised 2011 July 8; Accepted 2011 July 15.

Abstract

Objective

To analyze the incidence of obesity in adults aged 19–59 years in Korea and predict its trend in the future.

Methods

We considered a two-compartmental deterministic mathematical model Susceptible-Infected-Susceptible (SIS), a system of difference equations, to predict the evolution of obesity in the population and to propose strategies to reduce its incidence.

Results

The prevention strategy on normal-weight individuals produced a greater improvement than that produced by treatment strategies.

Conclusions

Mathematical model sensitivity analysis suggests that obesity prevention strategies are more effective than obesity treatment strategies in controlling the increase of adult obesity in Korea.

1 Introduction

Obesity is an unbalanced condition in which accumulated dietary calories exceed the body’s energy consumption. Currently, obesity is one of the most serious health problems all over the world, including in Korea. It is an ongoing problem for the whole of society, as well as for individuals. Obesity is associated with illnesses such as cardiovascular diseases, diabetes, musculoskeletal disorders such as osteoarthritis, and some cancers, and significantly devalues the quality of life [1]. From a societal perspective, the annual expenditure on controlling diseases related to obesity is astronomical [2].

Recently, obesity has been hypothetically regarded as a social epidemic in the sense that it can spread from one person to another [3,4]. There is consistent evidence that obesity is a health concern that spreads by social peer pressure or social contact [5]. Mathematical modeling of the progression of most infectious diseases is useful to discover the likely outcome of an epidemic or to help in their management. Indeed, mathematical studies indicate that obesity is transmitted socially [5–8].

In this paper, we applied a new difference equation model to study obesity in Korean adults aged 19–59 years and also projected the findings into the future. The main purpose of this study was to predict the future prevalence of obesity in 19–59-year-old Koreans and to propose some strategies to reduce obesity in Korea. Moreover, we identified important predictive factors of obesity in the Korean adult population.

2 Materials and Methods

2.1 Model description

We introduce a two-compartmental deterministic mathematical model, a system of difference equations, to predict the evolution of obesity in the Korean population and to propose strategies to reduce its incidence. Basically, the idea of modeling the obesity epidemic comes from the disease model Susceptible-Infected-Susceptible (SIS) [9].

As in [8], we divided the 19–59-year-old population in Korea into two subpopulations based on their body mass index {BMI; weight/height2 (kg/m2)}. The classes or subpopulations were individuals defined as normal weight St (BMI, <25) and obese individuals Ot (BMI, ≥25) according to the definition of WHO [10].

We regarded obesity as an infectious disease caused by social peer pressure or social contact that influences the probability of transmission of sedentary lifestyle and unhealthy nutritional habits. From this position, let us propose an epidemiological-type model to study the epidemic evolution of obesity. We adopted the following assumptions [8]:

  • (1) Once an adult starts an unhealthy lifestyle, he/she would continue it and develop obesity Ot because of this lifestyle. In class Ot, individuals who are able to stop his/her unhealthy lifestyle can move to class St.

  • (2) Populations of humans are homogeneously mixed, which means the rate of interaction between two different subpopulations is proportional to the product of the numbers in each of the subsets concerned [11,12]. The transit is modeled using the term βStOt [13].

  • (3) The subpopulation’s sizes and behaviors with time will decide the dynamic evolution of adulthood obesity.

The model can be described as shown in Figure 1.

Figure 1

Schematic diagram of the mathematical model.

Without loss of generality and for the sake of clarity, the 19–59-year-old adult population is normalized to unity; for all time t, St + Ot = 1. Under the above assumptions, we have the following nonlinear system of difference equations:

(2.1){St+1=StμSt+μS0βStOt+ρOtOt+1=OtμOt+μO0+βStOtρOt
The state variables of the model are shown in Table 1 and parameters used in the model in Table 2.

State variables for population dynamics

Time-invariant parameters for population dynamics

Model (2.1) is analyzed qualitatively to investigate the existence and stability of equilibria. First, we identify the equilibrium of nonlinear system of difference equations (2.1). We denoted by (S¯,O¯) the equilibrium points of system (2.1). Computing the steady state, we obtain two equilibrium points; the disease-free equilibrium (DFE) for (2.1) is:

(2.2)(S¯,O¯)=(1,0)
and the endemic equilibrium (EE) for (2.1) is:
(2.3)(S¯,O¯)=(ωω2+4βμO02β+1,ω+ω2+4βμO02β)

DFE (1,0) is locally asymptotically stable if 0 < ρ + μ − β < 2 and the endemic equilibrium (S¯,O¯) (2.3) is locally asymptotically stable if 0<ρ+μβ(S¯O¯)<2. On the one hand, model (2.1) can be simplified to the following one difference equation using the assumption St + Ot = 1.

(2.4)Ot+1=β(Ot)2+(1+βρμ)Ot+μO0
O¯=(ρ+μβ)+(ρ+μβ)2+4βμO0/(2β) is a locally asymptotically stable equilibrium of (2.4) if 0 < (ρ + μ − β)2 + 4βμO0 < 4.

Additionally, DFE (1,0) is locally asymptotically stable of (2.1) if and only if O¯ is a locally asymptotically stable equilibrium of (2.4) and the endemic equilibrium (S¯,O¯) (2.3) is locally asymptotically stable of (2.1) if and only if O¯ is a locally asymptotically stable equilibrium of (2.4). See Appendix for details.

2.2 Numerical computation

Parameter estimation was performed to produce the model of the current Korean situation.

First, we need to know the parameters μ, S0, and O0 of the model. μ is inversely proportional to the mean time spent by an adult in the system. The total 40 years from age 19 to 59 is 2080 weeks (accepting 1 year = 52 weeks) so μ should be 1/2080. S0 and O0 are the proportions of the normal weight and obese populations, respectively, in the 18 years age group. According to the statistical data of the Korea Institute of Sport Science in 2007 [14], the obesity rate of 18-year-old Koreans was 16.4%.

Parameters β = 0.0009038 and ρ = 0.0000336 were estimated by least-square method using obesity data (from 1998 to 2008) of the fourth Korea National Health and Nutrition 2008 (Figure 2) [15].

Figure 2

Parameter estimation for β and ρ using least square method.

Recall Ot+1 = −β(Ot)2 + (1 + β − ρ − μ)Ot + μO0 (2.4). To know the trend of the obese population of 19–59-year-old Koreans, using the parameter in Table 3, we simulate our model (2.4) (Figure 3).

Estimated parameters

Figure 3

Numerical simulation of the mathematical model.

From Figure 3, we see that the trend of obese population of persons aged 19 to 59 years in Korea increases.

Now, we compare prevention strategy versus treatment by present simulations of the mathematical model and varying some of the parameters. The aim of varying the parameters is to observe how the final prediction can be affected by these changes. This perturbation allows us to propose obesity prevention strategies.

Figures 4 and 5 suppose an increase of physical activity in the obese class Ot. Consequently, we tried to increase gradually the parameter ρ from 0% to 2000% initially given ρ = 0.0000336. In this case, this treatment strategy on obese individuals produces a small improvement. Figures 6 and 7 suppose a decrease in the social transmission parameter β. In this case, we tried to reduce gradually the parameter β from 0% to 100% initially given β = 0.0009.38. The prevention strategy on the normal-weight individuals produced a bigger improvement than that produced by the previously mentioned treatment strategies.

3 Results

Data of the fourth Korea National Health and Nutrition 2008 show that there has been a noticeable increase in the rate of obesity in Korea [15]. Performing the simulations with the above model showed the same increases in coming years. In other words, the numerical simulations carried out with our proposed mathematical model and with the estimated parameters indicate an increasing trend in obesity among the 19–59-year-old population in future.

From the results of the numerical simulations (Figures 4–7), we conclude that the prevention strategy on normal-weight individuals to adopt physical activity and healthy nutritional habits is better than the treatment strategy that motivates physical activity and healthy nutritional habits in the obese populations due to the finding that the reduction observed in the incidence of obesity is greater in the former case.

4 Discussion

To reduce the incidence of adulthood obesity among the 19–59-year-old population in Korea in future, from Figures 4–7 we could see that it would be more beneficial to focus on the promotion of physical activity and healthy nutritional habits in normal-weight individuals than on implementing similar campaigns for obese individuals. To achieve lower obesity prevalence, these technical suggestions would produce positive effects on reducing the incidence of obesity based on the above numerical simulations.

5 Conclusions

Mathematical model sensitivity analysis suggests that obesity prevention strategies are more effective than obesity treatment strategies in controlling the increase of adulthood obesity in Korea. Consequently, prevention strategy is recommended for tackling the obesity social epidemic.

References

1. Kopelman P.G.. Obesity as a medical problem. Nature 404(6778)2000 Apr;:635–643. 10766250.
2. Eric A.F., Ian C.F., Guijing W.. National medical spending attributable to overweight and obesity: how much, and who’s paying? Health Aff 22(3)2003 May;:w219–w226.
3. James P.T.. Obesity: the worldwide epidemic. Clin Dermatol 22(4)2004 Jul–Aug;:276–280. 15475226.
4. James P.T., Leach R., Kalamara E., Shayeghi M.. The worldwide obesity epidemic. Obes Res 9(4)2001 Nov;:228–233.
5. Christakis N.A., Fowler J.H.. The spread of obesity in a large social network over 32 years. N Engl J Med 357(4)2007 Jul;:370–379. 17652652.
6. Evangelista A.M., Ortiz A.R., Rios-Soto K.R., Urdapilleta A.. USA the fast food nation: obesity as an epidemic. 2004. Los Alamos National Laboratory, T-7, MS B284, Theoretical Division.
7. Jodar L., Santonja F.J., Gonzalez-Parra G.. Modeling dynamics of infant obesity in the region of Valencia, Spain. Comp Appl Math 56(3)2008 Aug;:679–689.
8. Santonja F.J., Villanueva R.J., Jodar L., Gonzalez Parra G.. Mathematical modelling of social obesity epidemic in the region of Valencia, Spain. Comp Appl Math 6(1)2010;:23–34.
9. Brauer F., Castillo-Chavez C.. Mathematical models in population biology and epidemiology. 2001. Springer. New York:
10. World Health Organization (WHO). The World Health Report, 1995. Available from: http://www.who.int/whr/1995. . [accessed 07.03.08].
11. Emilia V., Richard G.W.. An introduction to infectious disease modelling. 2010. Oxford University Press. Oxford:
12. Murray J.D.. Mathematical biology I. An introduction. 3rd ed.th ed. 2007. Springer. Berlin:
13. Daley D.J., Gani J.. Epidemic modelling: an introduction. 2000. Cambridge University Press. Cambridge:
14. Korea Institute of Sport Science, Seoul. Available from: http://ssps.sportskorea.net/service/pub/IntroServlet. 2008. [accessed 08.01.11, in Korean].
15. Korea Ministry of Health and Welfare . The fourth Korea health and nutrition examination survey. Seoul. 2008. [in Korean].
16. Wikipedia. Jacobian matrix and determinant. Available from: http://en.wikipedia.org/wiki/Jacobian_matrix_and_determinant. .
17. Edelstein-Keshet L.. Mathematical models in biology. 2005. SIAM. Philadelphia:
18. Jury E.I.. "Inners" approach to some problems of system theory. IEEE Trans Automat Contr 31971 Jun;:233–240.
19. Allen L.. An introduction to mathematical biology. 2006. Pearson Prentice Hall. New Jersey:
20. Neuhauser C.. Calculus for biology and medicine. 2nd ed.th ed. 2003. Pearson Prentice Hall. New Jersey:
Appendix

To identify the equilibrium of nonlinear system of difference equations (2.1), we set

(A1)S¯=S¯μS¯+μS0βS¯O¯+ρO¯,
(A2)O¯=O¯μO¯+μO0+βS¯O¯ρO¯.

Since S¯+O¯=1, (A2) can be expressed by

(A3)β(O¯)2+(ρ+μβ)O¯μO0=0.

By putting ω = ρ + μ − β we can express (A3) as

(A4)β(O¯)2+ωO¯μO0=0.

If O¯=0, then O0 = 0. Therefore, the disease-free equilibrium (DFE) for (2.1) is

(A5)(S¯,O¯)=(0,1)

Now, suppose that O¯0. Then, the quadratic equation (A4) has roots O¯=ω±ω2+4βμO0/(2β). For looking up the positive equilibrium, we need for O¯>0, then we get O¯=ω+ω2+4βμO0/(2β). Here, clearly O¯<1, for ω+ω2+4βμO0/(2β)<1 if and only if −S0μ < ρ. Since S¯+O¯=1, we have S¯=ωω2+4βμO0/(2β)+1. Therefore, the endemic equilibrium (EE) for (2.1) is

(A6)(S¯,O¯)=(ωω2+4βμO02β+1,ω+ω2+4βμO02β)

Next, we check the stability of the equilibrium points (S¯,O¯).

Theorem 1

DFE (S¯,O¯) in (A5) and the endemic equilibrium (S¯,O¯) in (A6) are locally asymptotically stable if 0 < ρ + μ − β < 2 or if 0<ρ+μβ(S¯O¯)<2, respectively.

Proof

The Jacobian matrix evaluated at equilibrium (S¯,O¯) is of the form [16]:

J(S¯,O¯)=(1μβO¯βS¯+ρβO¯1μ+βS¯ρ)

By Jury Test for n = 2 [17,18], if |Tr(J)|<1+det(J)<2 holds then the equilibrium point (S¯,O¯) is locally asymptotically stable. Otherwise, the equilibrium (S¯,O¯) is unstable.

It is easy to check the condition |Tr(J)|<1+det(J)<2 is equivalent to the conditions in Theorem for each case.

Remark

Since St + Ot = 1, the system of difference equations (2.1) can be simplified to one difference equation:

(A7)Ot+1=Ot+μO0+β(1Ot)OtρOtμOt=β(Ot)2+(1+βρμ)Ot+μO0=f(Ot)

Now, we may check the stability of the equilibrium point O¯ in (A7) as follows:

Theorem 2

O¯=ω+ω2+4βμO0/(2β) is a locally asymptotically stable equilibrium of (A7) if 0 < ω2 + 4βμO0 < 4, where ω = ρ + μ − β.

Proof

Put f(x) = −βx2 + (1 + β − ρ − μ) x + μO0. We need to find the condition |f(O¯)|<1 [19,20]. Since ω = ρ + μ − β, we see that f(O¯)=2βO¯+(1+βρμ)=2β(ω+ω2+4βμO0/(2β))+(1+βρμ)=ωω2+4βμO0+1ω=1ω2+4βμO0, whence we have that |f(O¯)|<1|1ω2+4βμO0|<11<1ω2+4βμO0<12<ω2+4βμO0<00<ω2+4βμO0<2. Squaring both sides, we get 0 < (ρ + μ − β)2 + 4βμO0 < 4.

Remark

By simplifying the condition ρ+μβ(S¯O¯), one may see that the DFE (S¯,O¯) in (A5) and the endemic equilibrium (S¯,O¯) in (A6) are locally asymptotically stable if and only if O¯ in (A7) is a locally asymptotically stable equilibrium.

Notes

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

Article information Continued

Figure 1

Schematic diagram of the mathematical model.

Figure 2

Parameter estimation for β and ρ using least square method.

Figure 3

Numerical simulation of the mathematical model.

Figure 4

Numerical simulation of the mathematical model when parameter ρ = 0.0000336 is increased from 0% to 2000%.

Figure 5

Dynamics of obesity depending on perturbation of ρ.

Figure 6

Numerical simulation of the mathematical model when parameter β = 0.0009038 is reduced by 0–100%.

Figure 7

Dynamics of obesity depending on perturbation of β.

Table 1

State variables for population dynamics

St The proportion of normal weight individuals at time t
Ot The proportion of obese individuals at time t

Table 2

Time-invariant parameters for population dynamics

1/μ Average time of stay in the system of 19–59-year-old adults
ρ Rate at which an obese individual moves to the normal weight subpopulation
β Transmission rate because of social pressure to adopt an unhealthy lifestyle
S0, O0 Proportions of normal weight and obese populations in the 18-year-old age group

Table 3

Estimated parameters

Parameter Value
μ 1/2080
β 0.0009038
ρ 0.0000336
S0 0.8365
O0 0.1635