Hindawi Publishing Corporation

BioMed Research International

Volume 2013, Article ID 413260, 18 pages
http://dx.doi.org/10.1155/2013/413260

Research Article
Modelling HIV/AIDS Epidemic among Men Who Have
Sex with Men in China

Xiaodan Sun,1 Yanni Xiao,1 Zhihang Peng,2 and Ning Wang3

1
Department of Applied Mathematics, School of Mathematics and Statistics, Xi’an Jiaotong University, Xi’an 710049, China
2
Department of Epidemiology and Biostatistics, Nanjing Medical University, Nanjing 210029, China
3
National Center for AIDS/STD Prevention and Control, Chinese Center for Disease Control and Prevention, 155 Changbai
Road, Beijing 102206, China
Correspondence should be addressed to Yanni Xiao; [email protected]

Received 9 April 2013; Accepted 19 July 2013

Academic Editor: Lucia Lopalco

Copyright © 2013 Xiaodan Sun et al. This is an open access article distributed under the Creative Commons Attribution License, which
permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

A compartmental model with antiviral therapy was proposed to identify the important factors that influence HIV infection
among gay men in China and suggest some effective control strategies. We proved that the disease will be eradicated if the
reproduction number is less than one. Based on the number of annual reported HIV/AIDS among MSM we used the Markov-
Chain Monte-Carlo (MCMC) simulation to estimate the unknown parameters. We estimated a mean reproduction number of
3.88 (95% CI: 3.69–4.07). The estimation results showed that there were a higher transmission rate and a lower diagnose rate
among MSM than those for another high-risk population. We compared the current treatment policy and immediate therapy
once people are diagnosed with HIV, and numerical studies indicated that immediate antiviral therapy would lead to few HIV
new infections conditional upon relatively low infectiousness; otherwise the current treatment policy would result in low HIV
new infection. Further, increasing treatment coverage rate may lead to decline in HIV new infections and be beneficial to
disease control, depending on the infectiousness of the infected individuals with antiviral therapy. The finding suggested that
treatment efficacy (directly affecting infectiousness), behavior changes, and interventions greatly affect HIV new infection;
strengthening intensity will contribute to the disease control.

1. Introduction in 2011 [2]. In mainland China MSM bear a high burden of

HIV since the cultural norm may cause them to wish to be
After the detection of the first acquired immunodeficiency hidden, which makes them very likely to be exposed to HIV,
syndrome (AIDS) patient, human immunodeficiency virus (HIV) and moreover causes much more difficulties to implement
spreads at an alarming rate worldwide. In recent years, though the intervention measures. At the same time, anal intercourse
number of people newly infected with HIV is decreasing, the between men, if unprotected, carries a high risk of HIV
prevalence of HIV among men who have sex with men (MSM) transmission [3]. There are studies that showed that HIV
has increased significantly in China since 2005. In China, the f irst infection rate via homosexual contact is much higher than
AIDS patient among MSM was found in 1989 and the proportion heterosexual contact [4]. So, how to identify the important
of reported cases resulting from homosexual contact in year 2005 factors which greatly influence HIV infection among MSM
to 2011 is 0.4%, 2.5%, 3.4%, 5.9%, 8.6%, 10.8%, and 13.0%, and control the quick increase in HIV epidemic among MSM
respectively [1]. The prevalence of HIV among MSM is became urgent in recent years. The primary objective of our
increasing year by year, has reached an average of 5% in large and study is to understand the HIV epidemic among MSM and to
medium cities, and is greater than 10% in the main cities of the try to suggest some control strategies.
Southwest [1]. Among all people living with HIV, the proportion Discovered in 1981, HIV is one of the few things that
of people infected by homosexual contact increases from 14.7% in draw attention from both mathematicians, medical scientists
2009 to 17.4% and behavioral scientists. Many models have been proposed in
2 BioMed Research International

order to predict and control the spread of HIV effectively. A

d+ I
basic SIR model was formulated by Gran et al. [5] in 2008.
Considering the feature of chronic disease, the authors 1 d+ T
1 I1 T1
generalized the basic model to a Markov model to represent
different infection rates in different stages [5]. In such model, 1 1
HIV/AIDS infected individuals are divided into 13 different U (t) d+ T
2 2
stages according to CD4+ level, and at each stage people have S I I2 T2
a specific infection rate. The idea of dividing the infected
d+ d+ I (1 − ) d+ I
group to different stages has been widely used in the study of 2 2
HIV dynamics [6–10]. In 2008, Zhou et al. [10] proposed a
A d+ A
staged discrete model to investigate HIV epidemic in mainland TA
China. We will also utilize the idea in this paper to formulate
our mathematical model. d+ A
In mainland China, the government initiated a large scale AIDS patients
of antiviral therapy since 2003, and free treatment has been with treatment
expanded to all HIV-positive individuals whose CD4+ T count
Figure 1: The flow diagram of model with antiviral therapy.
is less than 350 cells per L (microliter) of blood from 2007. In
recent years, antiviral therapy has become a measure to control
the epidemic since there are studies demonstrating that HIV-positive individuals without receiving treatment into
antiviral therapy can extend the life expectancy of HIV/AIDS two different stages: the HIV-positive individuals with >
infected individuals and reduce their infection rates [4, 13, 18,
350( 1), individuals with 200 < < 350 ( 2), where denotes
19]. Granich et al. [12] put forward a new therapy, starting
the CD4+ T level in the blood (i.e., the number of CD4+ T
antiviral therapy as soon as people are diagnosed with HIV.
cells per microliter of blood). HIV-positive individuals in
And he proved that this therapy can reduce the infection rate
and death rate greatly. Note that lower infection rate is group 1 and 2 will enter into 1 and 2 if they receive antiviral
beneficial to the disease control, but extended life span therapy, respectively, and finally progress to the AIDS stage (
increases the risk to infect others. For this high-risk group, ). The flow diagram is described in Figure 1. Let be AIDS
whether immediate antiviral therapy once people are patients who are diagnosed after onset of AIDS. Since all
diagnosed with HIV is beneficial to disease control or not is AIDS patients are given treatment at the present stage in
controversial since early treatment will inevitably result in China, here we do not distinguish between AIDS patients in
early occurrence of drug resistance due to bad adherence and and those in . The high-risk population size is represented
no additional drugs are available by ; that is, = + + 1+ 2 + + 1+ 2 + ,
[20]. So, to what extent antiviral therapy will be
implemented or when to initiate antiviral therapy is unclear. = − ( )− ( + ) ,
All these fall within the scope of this study.
This paper is divided into 6 sections. In Section 2, a = ( )− ( + + ) ,
mathematical model with antiviral therapy is formulated
following a modeling approach for stratification of the popu- 1= 1 − ( + + 1 + 1 ) 1,
lation according to the clinical progression of the disease and 2 = 2 + 11 − ( + + 2 + 2 ) 2 ,

epidemiological status of the individuals [6]. In Section 3, the (1)

unknown parameters and initial values involved in the model = (1 − ) +2 2 − ( + ) ,
are estimated. In Section 4, the main results of this paper are
1 = 1 1 − ( + + 1) 1,
given which include the threshold dynamics, prediction for the
disease, the effects of antiviral therapy, and the effects of 2 = 2 2 + 1 1 − ( + + 2 ) 2,
intervention measures. In Section 5, some sensitivity analyses
=22 −( + ) .
are conducted. At last, discussions are given in Section 6.
We assume that people enter into the susceptible class at a
2. Mathematical Model rate , exit high-risk population at a constant rate . Each one has
a natural death rate ( ). It is assumed that the transmission
Our model is formulated based on the key epidemiological probability for the undiagnosed HIV-positive individuals per
properties of HIV/AIDS and some implemented public heath high-risk behavior (i.e., male-male sex) is a constant, denoted
interventions such as condom use and antiviral therapy. The by . represents the contact rate per year, and represents the
underlying structure of the model comprises classes of indi- protection rate by interventions such as condom use and
viduals among MSM who are high-risk susceptibles ( ), HIV antiviral therapy. Since HIV/AIDS infected individuals in
infected but not yet diagnosed ( ), diagnosed HIV-positive different stages may have different transmission probabilities
individuals who have not yet progressed to AIDS stage, and per high-risk behavior due to different CD4 cell count,
those with AIDS clinical symptoms ( ). Further, according to antiviral therapy, behavior changes, and so forth, we then
the CD4+ T cell counts in the blood we divide the diagnosed introduce a modification factor for each
BioMed Research International 3

infectious class, denoted by 1, 2, 1, 2, and , respec-tively. related death rate for AIDS patients with antiviral therapy
Thus, susceptible people become HIV infected at a is = 0.393 ∗ (1 − 65.3%) = 0.136. T he disease-related death
rate ( + 1 1+ 2 2+ 1 1+ 2 2+ ( + ))/ . 1 and 2 are the rate for HIV-positive individuals with antiviral therapy
proportions of HIV-positive individuals with > 350 is = 0.172 ∗ (1 − 65.3%) = 0.06.
and 200 < < 350when diagnosed, respectively. Let = 1 + 2, Recently, antiviral therapy is not provided to HIV-positive
and then 1 − is the proportion of individuals who have individuals with > 350, we then set 1 = 0. According to the
already progressed to AIDS stage when diagnosed. treatment policy of China, the HIV-positive individuals with >
Constants 1 and 2 represent the progression rates from 1 to 350; are given treatment since early 2007. Lou et al. [13] and
Xu et al. [14] estimated the treatment coverage rate for HIV-
2 and from 2 to , respectively. 1 and 2 denote the positive individuals with 200 < < 350 nowadays to be 0.2.
progression rates from 1 to 2 and from 2 to , respectively. Then we choose time-dependent function 2 = 0.2from year
( = , , ) and ( = 1,2) represent the disease related death rate
and antiviral therapy coverage rate for each stage. The 2007 to 2009 and 2 = 0from year 2005 to 2006. From the
model equations are given in (1), where ( ) = (1− ) ( + database we can get some initial values for the system: 1(0) =
1 1 + 2 2 + 1 1 + 2 2 + ( + ))/ . T he model is illustrated in 99, 2(0) = 49, and (0) = 53. Considering that there are no
Figure 1, and definitions of parameters are given HIV-positive individuals receiving antiviral therapy in year
in Table 2. We can verify that any solution of system (1) with 2005, thus we have 1(0) = 2(0) = (0) = 0.
nonnegative initial value is nonnegative. And it is easy to get In order to determine the relative infectiousness for each class
8
that = {( , 1,, 2, , 1, 2, ) ∈ + : + +1 + 2 + + 1 + 2 + ≤ / }is we follow the principle employed by Gran et al. [5]. They divided
positively invariant. the diagnosed HIV-positive individuals into 5 stages based on the
CD4+ T cell counts in the blood (above 500, 350–499, 200–349,
below 200 copies, and AIDS stage). The numbers of new
3. Data and Parameter Estimation infections caused by one individual, in each stage, per unit time in
3.1. Data. Based on the current surveillance system we can get a totally susceptible population are 0.031, 0.025, 0.017, 0.013, and
the number of annual reported HIV-positive cases and AIDS 0 for these 5 stages, respectively. Note that we combine the first
patients among MSM from year 1985 to 2009 in main-land two stages as one stage in our model. We calculate 1 as follows:
China. The first HIV infected case among MSM in China was the number of new infections in stage 1 is ((1/0.0455)/(1/0.0455
reported in 1989. However, the disease has spread very + 1/0.125)) ×
quickly among this population. The proportion of reported 0.031 + ((1/0.125)/(1/0.0455 + 1/0.125)) × 0.025 =
cases resulting from homosexual contact increases as follows: 0.0294, where 0.0455 and 0.125 are the progression rates
2.5% (2006), 3.4% (2007), 5.9% (2008), and 8.6% (2009) from stage 1 to stage 2 and from stage 2 to stage 3. By
[1]. Sentinel surveillance results have shown that the HIV- similar method we get that the numbers of new infections
positive rates among MSM population in different regions in and 2 are 0.0407 and 0.017, respectively. Then, we have
have been consistently greater than 1% and are increasing 1 : 1 : 2 = 0.0407 : 0.0294 : 0.017, which implies that 1 =
year by year. Homosexual transmission is becoming one of 0.7224, 2 = 0.4177. Other unknown parameters are
the most important drivers of the AIDS epidemic. We estimated on the basis of the real data.
choose year 2005 as a starting point since the consistent
surveillance and testing policy has been implemented in 3.3. MCMC Procedure. We utilize the Markov-Chain Monte-
mainland China since then [15]. Carlo (MCMC) simulation to estimate the main parameters,
and initial values of model (1). Metropolis-Hastings (M-H)
algorithm [21–26] are employed for MCMC simulation. The
3.2. Parameter Values Obtained from Data or the Literature. algorithm runs for 210000 iterations with a burn-in of 10000
The natural death rate was estimated to be = 0.0149 [11]. iterations. Using the rest 200000 samples, we obtain our
Generally speaking, if a person is a homoeroticism he tends to estimates [27]. The estimation procedure is carried out in two
be a homoeroticism, through all his life. So we roughly choose steps. Firstly, in order to reduce the number of parameters that
−1 need to be estimated to the minimum, we use the method
= 1/40 year . Zhou et al. [10] estimated the progression rate
from 1 to 2 is 1 = 1/6 and the progression rate from 2 to is 2 = proposed by Tang et al. [28]. Note that the size of high-risk
1/3. Gran et al. [5] estimated that the scale of progression rate population is very large compared with HIV/AIDS infected
population; / approximately equals 1. Assuming = , we can get
reduced by antiviral therapy is 1/2. So we choose 1 = 1/2 1 =
a reduced model (A.1), given in Appendix A. By f itting the
1/12 and 2 = 1/2 2 = 1/6. Because 1 = 1/2 ∗ 2 , we can reduced model A(.1) to the data on annual reported HIV/AIDS
suppose that one-third of the HIV-positive individuals are in cases from 2005 to 2009, we get the estimates for the
stage 2 and two-thirds are in stage 1 when diagnosed; that is, 1
transmission coefficient 0, proportion of diagnosed individuals
= 2 /3and 2 = /3.Xiao et al. [15] estimated the additional death
who are HIV-positive , diagnose rate , modification factors 2, ,
rate for the diagnosed HIV-positive individuals to be 0.172; and initial value of undiagnosed infected individuals (0). Here
then we choose = 0.172. The disease related death rate for we assume 1 = 2 since there are no significant difference
AIDS patients without receiving treatment is estimated to be between the infectiousness of people in these two groups.
0.393 by Zhang et al. [17]. The ministry of health, China [16], Using the simulation results of these parameters we can get the
estimated that the disease related death rate can be decreased Markov Chain of basic reproduction number 0 and its
by 65.3% after antiviral therapy, so the disease variation.
4 BioMed Research International
get the expression of = ∗ ∗ ∗ ∗ ∗ ∗
+ ( , , 1 , 2 , , 1 ,
Secondly, based on the estimation results we repeat the ∗ ∗
same procedure by fitting model (1) to our real data to 2 , ),
∗ ∗ ∗ ∗ ∗ ∗
estimate the rest of the unknown parameters (0)and . See where = ( − 1 )/( + ), 1 = 1 , 2 = 2 ,
∗
= ∗ ∗ ∗ ∗ ∗ ∗ = ∗ ∗ (0 −
Appendix A for details about parameter estimation. 3 , 1 =4 , 2 = 5 , 6 , =
1)/( 1 0+( + )(1+ 1+ 2+ 3+ 4+ 5+ 6−( 1/( + )))), 1 =
1 / 2, 2 = ( 2 / 3)+( 2 1/ 2 3), 3 = ((1 − ) / 4)+
4. Main Results ( 2 2/ 3 4)+( 1 1 2/ 2 3 4), 4 = 1 (1+ ) 1/ 2 5,
4.1. Threshold Dynamics. The basic reproduction number 0, 5 = ( 2 (1 + ) 2/ 3 6) + ( 1 1(1 + ) 2/ 2 3 6) +
the average number of secondary cases generated by a single ( 1 (1+ ) 1 1/ 2 5 6), and 6 =( 2 (1+ ) 2 2/ 3 6 4)+
primary case in a fully susceptible population during its ( 1 1(1+ ) 2 2/ 2 3 6 4)+( 1 (1+ ) 1 1 2/ 2 5 6 4).
average infectious period [29, 30], is a threshold parameter for Moreover, according to the persistence theorem devel-
the infectious disease and can help determine whether an oped by Smith and Zhao in [31] we can prove the system is
infectious disease will spread through a population. Follow- uniformly persistent if 0 > 1. To this end we verify that
ing the next-generation matrix method [29] we can get 0 for invariant sets in the boundary of the feasible region are not
system (1) (see Appendix B for details). For convenience, attractors (see details in Appendix C.2.) Hence we have the
we following theorem.
let 0 = (1 − , )1 = + + , 2 = + + 1 + 1,
3
= + + 2 + 2, 4 = +, 5 = + + 1 and Theorem 2. System (1) is uniformly persistent when 0 >
6 = + + 2. The expression of 0 is given as follows: 1, in the sense that there is a positive number 0 such that
for all initial values ( 0, 0, 10, 20, 0, 10, 20, 0) + × Int(
0 = 0 11 112 22 112 7
[1+ + + + + ), the solution of system (1) satisfies
1 2 23 3 234 lim inf ( ( ), ( ), 1 ( ), 2 ( ), ( ), 1 ( ), 2 ( ), ( ))
22
(1− ) 111
→∞

+ + +
≥( 0, 0, 0, 0, 0, 0, 0, 0).
34 4 25
(3)
+ 1112+1212+222 25623636
4.2. Prediction for the Disease. Based on the number of indi-
1112 1122 222 viduals living with HIV (not AIDS) or AIDS among MSM
+ + + ]. from year 2005 to year 2009, we estimate mean values of
2564 2364 364
parameters and their standard deviations which are listed in
(2) Table 2. We also derive the goodness of fit with data together
with the uncertainties (shown in Figure 2). The areas from
As the model shows, the primary infected individual light to dark mean the 50%, 90%, 95%, and 99% limits of the
will stay at the compartments , 1, 2, , 1, 2, and (if experiencing posterior uncertainty due to model parameters. In particular,
the stage) for mean times of 1/ 1, 1/ 2, we get the mean value of the basic reproduction number to be
1/ 3, 1/ 4, 1/ 5, 1/ 6, and 1/ 4, respectively. T he prob- 3.8840 with a estimation error 0.097, and the 95% confidence
abilities of the primary patient to experience each stage are interval is [3.69,4.07]. Figure 3 describes the Markov Chain
1, 1 / 1,( 1 1/ 1 2) + ( 2 / 1), ( 1 1 2/
for 0, which has a good convergency. From the estimation
1 2 3) + ( 2 2/ 1 3) + ((1 − ) / 1), 1 1/ 1 2, ( 1 1 1/ 1 2 5 ) +
( 1 2 1/ 1 2 3) + ( 2 2 / 1 3), and results we find that the value of transmission coefficient 0 is
( 1 1 1 2/ 1 2 5 6)+( 1 1 2 2/ 1 2 3 6)+( 2 2 2/ larger than the estimation for heterosexual transmission by Xu
1 3 6). And the transmission coefficients at each stage are et al. [14] and that for the general high-risk population without
0, 0 1, 0 2, 0 , 0 1, 0 2, and 0 . It is easy to see that 0/ 1, 0 1/ 2, considering transmission routs by Xiao et al. [15]. This implies
0 2/ 3, 0 / 4, 0 1/ 5, 0 2/ 6, that, on the one hand, HIV transmission probability among
and 0 / 4 represent the numbers of individuals the primary MSM are higher than that for other high-risk groups [4]; on
case infects at each stage, respectively. Using the total the other hand, the condom use rate or other interventions
probability formula we get the expression of 0. among MSM are a bit lower [32]. What is more, the diagnose
Meanwhile, system (1) has a disease-free equilibrium rate among MSM estimated is only 0.08, which is lower than
0 that for other high-risk groups. That is almost due to the
(DFE) 0 =( ,0,0,0,0, 0,0,0), where 0 = /( + ).From
[29] we can get that when 0 <1, the disease-free equilibrium of physiological specificity of MSM and the cultural norm of
system (1) is locally stable but unstable when 0 > 1. We can China. A study for Beijing was given in 2006, which pointed
out that testing and awareness of HIV infection among MSM
also prove that 0 is globally attractive when 0 < 1 (see details
in Appendix C.1). Thus, we get the following theorem. are very lacking [33].
We suppose that the treatment coverage rate will not
Theorem 1. When 0 < 1, the disease-free equilibrium of change and the current treatment policy will remain the same
(i.e., antiviral therapy starts when CD4+ T cell counts are less
system (1) is globally asymptotically stable, but unstable when
than 350 per L of blood). Our estimation shows that the total
0 >1. number of HIV/AIDS individuals among MSM in 2015 will
Besides the disease-free equilibrium the system has an reach 1.46×106, with 1.41×106 HIV-positive individuals and
4.41×104 AIDS patients if the current surveillance, testing,
endemic equilibrium + when 0 > 1. It is not dif f icult to
BioMed Research International 5

4
×10
3000 2
New diagnosed AIDS
cases

New diagnosed HIV

cases
2500
1.5
2000
1500 1
1000
0.5
500
0 2006 2007 2008 2009 2010 2011 0 2006 2007 2008 2009 2010 2011
2005 2005
Years Years

(a) (b)

Figure 2: Plots of data fitted results. (a) The number of annual reported AIDS patients. (b) The number of annual reported HIV-positive
individuals. Squares denote the real data. Areas from light to dark mean the 50%, 90%, 95%, and 99% predictive probability limits due to
parameter uncertainties.

4.3 lifespan of HIV/AIDS infected individuals, which conse-

quently makes them have more opportunities to infect
4.2 others. Some researchers supported the conclusion that
immediate antiviral therapy once people are diagnosed with
4.1 HIV is more effective for reducing the infection rate and
death rate, and finally controlling the infection [12],
Sample of R0

4 whereas other researchers worried that early initiating

3.9 treatment inevitably induces early occurrence of drug
resistance due to poor drug adherence and side effect and
consequently results in the decline in treatment efficacy
3.8 hence it is harmful to control HIV infection [20]. So when
antiviral therapy should be started is still controversial and
3.7 needs a further study. In order to investigate this issue we
consider the following two situations and compare the basic
3.6 reproduction numbers of these two situations.
0
2 4 6 8 10
Number of MCMC samples 4
×10 Situation 1. Start antiviral therapy when CD4+ T counts are
Figure 3: MCMC plots for less than 350 cells per L of blood (keep the current treatment
0. ∗
policy). Then we have 1 = 1 = 0. Let 2 = + + 1; we get the
expression of basic reproduction number in this situation,
∗
and interventions are unchanged (shown in Figure 4(a)). In which is denoted by 0 ; then
particular, our estimation gives that there are 6596 HIV/AIDS ∗
0 11 112 22 112
infected individuals among MSM receiving antiviral therapy 0 = [1+ ∗ + ∗ + + ∗
1 2 2 3 3 2 34
in 2011, which takes up to 55% of the diagnosed individuals
(1− ) (4)
among MSM with CD4+ T count less than 350 cells per L, 22 1212
+ + + ∗
whereas, estimation from Ministry of Health, people’s republic 34 4 2 36
of China stated that the antiviral therapy coverage rate has 2 22 1122 222
reached 73.5% in 2011 [2]. This difference agrees with the + + ∗ + ].
conclusion obtained by Tong that the treatment coverage rate 36 2 364 364

for HIV/AIDS cases infected sexually is lower than that for

Situation 2. Start antiviral therapy immediately once people
HIV/AIDS individuals infected by other routes [34].
are diagnosed with HIV. In this situation, the expression of
basic reproduction number 0 is given in Section 4.1.
4.3. Ef fects of Antiviral T herapy.Antiviral therapy can effec- Using the parameter values listed in Table 2 we get that
tively decrease the viral load in the blood of HIV/AIDS ∗
0 = 3.93, 0 = 3.88. This implies that immediate antiviral
infected individuals, thereby reduceing their infectivity. therapy once diagnosed may cause more new infections and
However, the decrease in viral load consequently alleviates the is not beneficial to the disease control if parameter values
symptom of HIV/AIDS infected individuals; then they may are chosen as those listed in Table 2. In such scenarios, the
become active or increase their high-risk behaviors. There are ef fect of expanded lifespan may overweight the effect of
some studies showed that HIV infected indi-viduals may reduced infectiousness caused by early treatment.
increase their high-risk behaviors since they believe that It is interesting to note that the value of 0 is greater than
∗
antiviral therapy can make them more healthy [4, 35–37]. 0 based on the parameter values shown in Table 2.
Moreover, antiviral therapy can prolong the However, immediate antiviral therapy may not always lead
6 BioMed Research International

6 6
×10 ×10
2 2

TotalHIV-positivecases
TotalHIV/AIDScases

1.5

1.5
1

1
0.5

0.5
0 2007 2009 2011 2013 2015

0 2007 2009 2011 2013 2015

2005 2005
Years Years
(a)
(b)
4 4
×10 ×10
6 8
TotalAIDScases

HIV/AIDScaseswithtreatment

4 6

3
4
2

1 2

0 2007 2009 2011 2013 2015

0 2007 2009 2011 2013 2015
2005 2005
Years Years
(c) (d)
Figure 4: Prediction of HIV/AIDS among MSM in China from 2005 to 2020 and the uncertainties of the model. Areas from light to dark
mean the 50%, 90%, 95%, and 99% predictive probability limits due to parameter uncertainties. (a) Total HIV/AIDS cases. (b) Total HIV-
positive cases. (c) Total AIDS cases. (d) Total HIV/AIDS cases with antiviral therapy. Parameters and initial values used are shown in
Table 2. Antiviral therapy started when CD4+ counts are less than 350 cells per L.

∗
to a greater value of the reproduction number. In fact, the In order to determine the sign of 0− 0 , we turn to consider
∗
relationship between 0 and 0 depends on many factors ∗
12 ( 0 −0 )
such as treatment coverage, infectiousness, and so forth. In
∗
the following we investigate the relation between 0 and 0 : 011
1 12 212
1 2 122
∗ =− ( + + + + )
∗ 1
0 −0
2 3 34 36 364
= 01 [( 1 + 12 + 12 +212 + 122 ) 1 12 12
+ + +
1 3 34 36 364 5 56 564
1 1 11
1
12
=− 1 ( + 12 )+ 11
×( − ∗ )+ + ∗ 1
2 2 25 256
2 3 5
112 1 21 2
+ ]. +( − ∗
)(
2 + ).
2564
5 2 3 46
(5) (6)
BioMed Research International 7

4 1
3.9 3.9

0.8 3.8
3.8 3.8

3.7 0.6 3.7

.7

R0 3.6
3
0.4 3.6
3

.6

3.5 3.5

0.2
.5
3.4 3.4
3

3.3 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1

0 0
(1= 2= )
∗
Current treatment policy (R0 )
Immediate treatment (R0)

(a) (b)
∗
Figure 5: (a) Plots of and 0 against factor V. (b) Contour plot of 0 varies the factor V and treatment coverage rate, where 1 = 2 = .
0
Parameters used are shown in Table 2.

It is obvious that sign( 0 − 0∗ ) = sign( 1 2( 0 − 0∗ )/ 0 1 1), where postponing antiviral therapy is better in terms of causing
sign( ) denotes the sign function and indi-cates the sign of . So relatively few new infections.
if 1 and 1 satisfy the following: Should We Increase the Treatment Coverage Rate? In the
1
1 21 2
1 12 following we investigate whether increasing the treatment
+( − ∗
)(
2 + )< ∗
(
1
+ ), coverage rate is beneficial or harmful to control HIV epi-
5 5 23 46 2 demic. To this end, we examine the variation in 0 with
3 (7) treatment coverage rate. We simply suppose 1 = 2 = .
∗
we have 0 − < 0, which means that immediate antiviral
0 Solving the partial derivative of 0 with respect to we have
therapy once people are diagnosed with HIV will result in a
lower value of basic reproduction number and hence lead to 0 0 1 2 11 ( 2 + 3)
= [− −( + )
fewer new infections. 21 2 2 2 2

From (7) we obtain that the difference between 0 and 1 2 3 2 3 (8)

∗
0 is determined by many parameter values. For simplicity, + 1 + 2 + ],
we introduce a factor V which varies from 0 to 1 to let the
modification factors vary simultaneously. To this end, let V
∗ where
multiply 1, 2, and ; we then analyze the value of 0 and 0 as
a function of V. T hus,V = 0 denotes that the infected 1 ( + + 1) (9)
= 2 ,
individuals are not infectious anymore and V = 1 means the 2 5
modification factors 1, 2, and take values as those listed in 11( + + 1) 2 ( + + 2)
Table 2. In order to compare the two situations (antiviral = 2 + 2
therapy started when CD4+ count less than 350 cells per L, 2 5 6 3 6 (10)
antiviral therapy started immediately once people are 11 11 ( 2 + 3)
∗ + − ,
diagnosed with HIV) we draw 0 and 0 in Figure 5(a). It 236
2 2
2 3 6
∗ ∗
shows that there is a critical value V = 0.81such that 0 < 0 for V < ( + )3
∗ ∗ ∗ 22 112 2
V and 0 > 0 for V > V . T hat is to say, if =− 2 − 2 2
we can reduce the modification factors to 81%, immediate 34 23 4
antiviral therapy will be more beneficial in terms of inducing a 112 ( + + 1) 22 ( + + 2)
+ 2 + 2 (11)
less reproduction number. Summing up the above, if the 2 456 3 46
modification factors for people with antiviral therapy are ( + 3)
112 112 2
relatively small, the earlier to start antiviral therapy the better; + −
2 2
.

in contrast, if the modification factors are relatively large, 2346 2 3 46

8 BioMed Research International

We can verify that > 0, > (see0 Appendix D for details if the diagnose rate increases to 0.3, 0.5, and 0.8, the total
about the proof of > 0). Let 0/ = 0; we have number of HIV/AIDS infected individuals will decrease by
32.74%, 52.14%, and 68.59% in year 2015, respectively, and
1 + 2 the number of individuals with treatment will increase by
92.07%, 90.06%, and 63.00% in year 2015, as shown in
1 11 ( 2 + 3) 2
Figure 8. It is interesting to note that relatively large testing rate
= 2 +[2+ 2 ] 2 − . leads to a slow increase of the number of individuals with
2 2 2
treatment (shown in Figure 8(d)). Meanwhile, low recruitment
(12)
rate and large exit rate will also slow down the spread of HIV
however their effects are very limited (as shown in Figure 7). T
Thus, there exists a critical value for a combination of
∗ his is because the number of HIV infected individuals is very
parameters 1, 2, , and so forth such that 0 / 2 < 0 small compared with the total high-risk population among
∗
for 1 + 2 < and 0 / 2 > 0 for 1 + 2 > . Note that and are MSM; thus the variation of / is very limited. This conclusion
both positive, which implies that is in agreement with that for general high-risk population
if the modification factors for HIV-positive individuals are obtained by Xiao et al. [15].
relatively small, then a large treatment coverage rate will In order to investigate the effect of antiviral therapy, we
lead to decline in new infections and hence is beneficial to study the variation in the incidence of HIV with different
disease control; otherwise increasing treatment coverage rate antiviral therapy efficacy represented by infectiousness ( 1, 2,
would result in an increase in new infections. T hat is to say, ), disease progression rates ( 1, 2), and disease-related death
given unchanged behavior of HIV-positive individuals, the rates ( and ). In order to reduce the number of variables we
infectiousness due to antiviral therapy plays a vital role in introduce two factors V (0 < V ≤ 1) and (0 <
determining HIV new infections. In particular, when ≤ 1)to describe variation in infectiousness and disease
infectiousness due to antiviral therapy is below a critical progression rates, respectively. During the analysis, V is set to
value, increasing treatment coverage rate is effective to be 0.75, 0.5 and 0.25 to represent that infectiousness is
reduce the new infections. Whereas infectiousness is greater decreased to 75%, 50%, and 25%, respectively. We suppose
than a critical value, increasing treatment coverage rate that antiviral therapy decreased the progression rates to 5%.
induces more new infections, which implies the more Figure 9 shows that HIV incidence will peak in around year
therapy for the HIV infected individuals, the more new
2016, and the better drug efficacy the lower HIV incidence. It
infections produced (see Figure 5(b) for details). It follows
indicates that although the drugs are persistently effective,
from Figure 5(b) that 0 decreases as treatment uptake rate HIV incidence will increase in the recent years. That is
increases if V is less than a critical value (around 0.6), which because effect of antiviral therapy on HIV epidemic between
implies that increasing treatment coverage is beneficial to hosts is not instantaneous but delays some time.
disease control in the case of better treatment efficacy;
otherwise, increasing treatment is harmful to it.
5. Sensitivity Analysis
4.4. Effects of Intervention Measures. In recent years many Note that in our model some parameters are known with
intervention measures have been implemented to control uncertainties or have large variances, which may greatly
the quick increase of HIV epidemic among MSM. The affect outcomes. It is then necessary to do the uncertainty
intervention measures include (1) strengthening education and sensitivity analysis such that the sensitive parameters
to the high-risk population, which decreases the constant can be detected. To examine the sensitivity of our results to
recruitment rate , the contact rate per year and increases the parameter variations, we use latin hypercube sampling
condom use rate; (2) increasing surveillance and testing, (LHS) and partial rank correlation coef f icients (PRCCs)
which results in an increase in the diagnose rate . Note that 38][ to examine the dependence of the reproduction number
both decreasing in contact rate and increasing in condom 0 and the expected number of total HIV/AIDS individuals
use rate can lead to the decline of transmission coefficient in 2015 on each parameter.
0. We initially examine the sensitivity of basic reproduction
To address the impact of each intervention measure on number 0 to parameter variations. We choose the sample size =
HIV infection among MSM, we investigate variation in 4000, parameters interested as the input variables, and the
number of HIV/AIDS infected individuals with varying value of 0 as the output variable. Parameter values and ranges
transmission coefficients, treatment uptake rate. Figure 6 are listed in Table 1. Figure 10(a) shows the PRCC value of
shows that strengthening education to high-risk population each parameter. Parameters with star above the bar are the
(i.e., smaller 0) and increasing surveillance and testing (i.e., significant ones, and the significance level we choose
larger ) slow down the spread of HIV signif icantly. In here is = 0.05. It shows that 0, 1, 2, , 1, and 2 have positive
particular, if 0 decreases by 30% and increases by 25%, the impact upon 0, whilst , 1, 2, , , and have negative impact.
total number of HIV/AIDS infected individuals among From the PRCC results we know
5 that 0 is not sensitive to parameters 1, 2, 1, and 2. So,
MSM will decrease to 2.2959 × 10 , which is decreased by
84.2%. Similarly, we could increase the diagnose rate only assumptions of 1 = 2 /3, 2 = /3, and 1, 2 being half of 1 and 2
to examine the effect of improving HIV diagnose and testing are justif ied and hence have little inf luence on our main
strategy on HIV transmission among MSMs. In particular, results. Further, we can get that the most influential
BioMed Research International 9

5 5
×10 ×10
15 15

Total HIV-positive cases

Total
HIV/AI 10 10
DS
cases

5 5

0 0 2007 2009 2011 2013 2015

2005 2006 2007 2008 2009 2010 2005
Years Years
(a) (b)
4 4
×10 ×10
5 6
Total
AIDS 5
Total HIV/AIDS with treatment

4
cases

4
3
3
2
2

1
1

0 0 2007 2009 2011 2013 2015

2005 2007 2009 2011 2013 2015 2005
Years Years

0 = 0.8578, = 0.08 0.75 0, 1.2 0 = 0.8578 , = 0.08 0.75 0, 1.2

0.9 0, 1.1 0.7 0, 1.25 0.9 0, 1.1 0.7 0, 1.25
0.85 0, 1.15 0.85 0, 1.15
(c) (d)

Figure 6: Plots of estimated number of HIV/AIDS cases vary with transmission coefficient 0 and diagnose rate . (a) Total number of
HIV/AIDS cases. (b) Total HIV-positive cases. (c) Total AIDS cases. (d) Total HIV/AIDS with treatment. Other parameters used are
shown in Table 2.

Table 1: Parameter values and ranges.

Parameters Ranges Initial values Parameters Ranges Initial values

10000, 100000] [ 402450 (0) [70000, 150000] 763240
0 [0.1, 1] 0.8578 0.01, 1] [ 0.0816
1 [0.01, 1] 0.5879 2 [0.01, 1] 0.2939
1 [0.02, 1] 1/12 2 [0.02, 1] 1/6
1 [0.02, 1] 0.2 2 [0.02, 1] 0.2
1 [0.01, 1] 0.7224 2 [0.01, 1] 0.4177
[0.01, 1] 0.2495 [0.02, 1] 0.172
1 [0.01, 1] 0.4177 [0.02, 1] 0.06
2 [0.01, 1] 0.3507 [0.02, 1] 0.1364
10 BioMed Research International
Table 2: Parameters and initial values.

Parameters Definition Value Std Source

Transmission probability of HIV per high-risk behavior — — —
Contact rate per year — — —
Protection rate by intervention measures (condom use) — — —
0 Transmission coefficient, 0 = (1 − ) 0.8578 0.013 MCMC
1 Modification factor for HIV infected individuals with > 350 0.7224 — [5]
2 Modification factor for HIV infected individuals with 200 < < 350 0.4177 — [5]
1 Modification factor for HIV infected individuals with > 350and 0.3507 — —
receiving antiviral therapy
2 Modification factor for HIV infected individuals with 200 < < 350 0.3507 0.028 MCMC
and receiving antiviral therapy
Modification factor for AIDS patients with antiviral therapy 0.2495 0.028 MCMC
Recruitment rate of susceptible 402450 20764 MCMC
Natural death rate 0.0149 — [11]
Exit rate of susceptible 0.025 — —
Diagnose rate 0.0799 0.020 MCMC
Proportion of diagnosed HIV-positive individuals = 1 + 2 0.8820 0.006 MCMC
1 Proportion of diagnosed HIV-positive individuals with > 350 0.5879 — [10]
2 Proportion of diagnosed HIV-positive individuals with 200 < < 350 0.2939 — [10]
1 Progression rate from 1 to 2 1/6 — [10]
2 Progression rate from 2 to 1/3 — [10]
1 Progression rate from 1 to 2 1/12 — [12]
2 Progression rate from 2 to 1/6 — [12]
1
Antiviral therapy coverage rate for HIV-positive individuals with 0 — —
> 350
2
Antiviral therapy coverage rate for HIV-positive individuals with 0.2 — [13, 14]
200 < < 350
Disease-related death rate for HIV infected individuals 0.172 — [15]
without receiving antiviral therapy
0.06 — [15, 16]
Disease-related death rate for HIV infected individuals with
antiviral therapy
0.136 — [16, 17]
Disease-related death for AIDS patients with antiviral
(0) therapy Initial value of (high-risk susceptible) 763240 87285 MCMC
(0) Initial value of (individuals living with undiagnosed HIV) 5988 1274 MCMC
1(0) Initial value of 1 (HIV-positive individuals with > 350) 99 — Database
2(0) Initial value of 2 (HIV-positive individuals with 200 < < 350) 49 — Database
Initial value of (AIDS patients diagnosed at onset of AIDS 53 — Database
(0)
symptoms)
0 — Database
Initial value of 1 (HIV-positive individuals with > 350and
1(0)
receiving therapy)
0 — Database
Initial value of 2 (HIV-positive individuals with 200 < < 350and
2(0)
receiving therapy)
(0) Initial value of (AIDS patients diagnosed before onset of AIDS 0 — Database
symptoms)
0 The basic reproduction number 3.8840 0.097 Calculated

parameters are 0, since the PRCC values of them are larger Table 1 except for 1, 1, and 1 are chosen as the input
than 0.8. In particular, simple calculation indicates that the variables. The total HIV/AIDS cases in year 2015 is output
basic reproduction number can decrease to less than 1 if 0 variable. The PRCC values are shown in Figure 10(b), from
reduces to 0.2. which we can find that both and (0)have little impacts on
We also investigate the sensitivity of the expected number the number of people living with HIV/AIDS in year 2015.
of people living with HIV/AIDS to parameter variations. The influences of other parameters on total HIV/AIDS
Given the current treatment policy, parameters described in cases are similar to their influences on 0.
BioMed Research International 11

×10
5 5
×10
15 15

Total HIV-positive cases

Total
HIV/AID 10 10
S cases

5 5

0 0 2007 2009 2011 2013 2015

2005 2007 2009 2011 2013 2015 2005
Years Years
(a) (b)
4 4
×10 ×10
Total 5 6

AIDS 5
Total HIV/AIDS with treatment

4
cases
4
3
3
2
2

1
1

0 0 2007 2009 2011 2013 2015

2005 2007 2009 2011 2013 2015 2005
Years Years
5
U = 4.0245 × 10 , = 0.025 0.8U, 1.2 5
U = 4.0245 × 10 , = 0.025 0.8U, 1.2
0.9U, 1.1 0.7U, 1.25 0.9U, 1.1 0.7U, 1.25
0.85U, 1.15 0.85U, 1.15
(c) (d)

Figure 7: Plots of estimated number of HIV/AIDS cases vary with constant recruitment and exit rate . (a) Total number of HIV/AIDS cases. (b)
Total HIV-positive cases. (c) Total AIDS cases. (d) Total HIV/AIDS with treatment. Other parameters used are shown in Table 2.

6. Discussion We studied the effect of antiviral therapy in two situations:

antiviral therapy started immediately once people are diag-
According to the CD4+ T cell counts in the blood, we divided nosed with HIV and antiviral therapy started when CD4+ T
the HIV-positive individuals to several stages and formulated a counts are less than 350 cells per L (current policy). We found
mathematical model with antiviral therapy. The unknown that, given unchanged behaviors after treatment, there exists a
parameters involved in this model were estimated using the
critical value for the infectiousness below which immediate
MCMC simulation basing on the real data (i.e., the number of
annual reported HIV/AIDS among MSM). We defined the treatment is better than the current policy in terms of the
reproduction number, whereas current policy exhibits better
threshold value (the basic reproduction number 0) which results than immediate treatment if the infectiousness is
determines whether the epidemic goes to extinction or not. By greater than the critical level. It indicates that if the treat-ment
using the stability theories and methods of ordinary differential
efficacy is really good (i.e., relatively low infectiousness), our
equation, we proved that the disease-free equilibrium is
conclusion suggests immediate treatment; otherwise the
globally asymptotically stable when 0 < 1, whilst the system is current policy is recommended. Similarly, when the infec-
uniformly persistence when 0 > 1. tiousness is relatively low (relatively good treatment efficacy),
12 BioMed Research International

5 5
×10 ×10
15 15

Total HIV-positive cases

Total
HIV/AI 10 10
DS
cases

5 5

0 2010 0 2010 2015

2005 2015 2005
Years Years
(a)
(b)
4
×10 4
×10
10 15
Total
8
Total HIV/AIDS with treatment

AIDS
6
cases 10

2 5

0 2010

0 2010 2015
2005 2015 2005
Years Years
= 0.08 = 0.5 = 0.08 = 0.5

= 0.3 = 0.8 = 0.3 = 0.8

(c) (d)

Figure 8: Plots of estimated number of HIV/AIDS cases vary with diagnose rate . (a) Total number of HIV/AIDS cases. (b) Total HIV-
positive cases. (c) Total AIDS cases. (d) Total HIV/AIDS with treatment. Other parameters used are shown in Table 2.

increasing treatment coverage will decrease the reproduction and UK (3.38–3.96) [39]. From the estimated parameters we
number and lead to decline in new HIV infection, whilst know that the transmission coefficient 0 is much larger than the
increasing treatment coverage will result in an increase in new estimation for heterosexual transmission [14] and general high-
HIV infection for the relatively great infectiousness, which is risk population [28]. This result is associated with the conclusion
in agreement with that for heterosexual transmission by Lou et al. [13] that MSM are 19 times more likely to be
[14]. Summing up the above, if treatment efficacy is relatively infected with HIV than general population. In fact, there are a lot
good, our conclusions suggest immediate treatment with high of money boys in China, also called male sex workers especially
uptake rate; otherwise the current policy is reasonable. in bathhouse, bars, and clubs [34, 40]. Recent surveys showed that
Using the data on the number of new reported HIV/AIDS the condom use rate among MSM is very low, only less than 30
infected individuals by year among MSM, we obtained esti- percent [32]. Anotherstudy showed that anal intercourse between
mates of the reproduction number, intervention parameter men, if unprotected, carries a high-risk of HIV transmission [3].
values, and the high-risk population size. Our estimated All these factors lead to the high probability for HIV transmission
reproduction number is 3.88 (95% CI 3.69–4.07) which is in among MSM, which is in agreement with our estimation. Our
the ranges of estimates for Western Germany (3.43–4.08) estimation
BioMed Research International 1
3
5
×10
5.5 Appendix
5
A. Parameter Estimation
4.5
Since the number of susceptible cases is large compared to
Incidence

4 the number of HIV/AIDS infected cases, / approximately

equals 1. Thus we can get the following reduced model:
3
3.5 ̃
= ( )− ( + + ) ,
2.5

1 = 1 − ( + + 1 + 1 ) 1,

2 = 2 + 1 1 − ( + + 2 + 2 ) 2,
2

1.5
(A.1)
= (1 − ) +2 2 − ( + ) ,
1
1 = 1 1 − ( + + 1) 1,
2010 2015 2020 2025
Years
2 = 2 2 + 1 1 − ( + + 2 ) 2,
= 1, u = 1, T = 0.06, A = 0.1364
= 0.75, u = 0.05, T = A = 0 =22 −( + ) ,
= 0.50, u = 0.05, T = A = 0
= 0.25, u = 0.05, T = A =0
̃
where ( ) = (1− )( + 1 1+ 2 2+ 1 1+ 2 2+ ( + ))and = 1
Figure 9: Plots of incidence against factor V when antiviral + 2.
therapy started immediately after diagnose. 1 = 2 = 0.8. Other For ensuring good convergence of Markov Chain, some
parameters used are shown in Table 2. prior information is given. Xiao et al. [15] built a mathemat-
ical model which did not consider the different transmission
routes and estimated several parameters involved in the model.
also shows that the diagnose rate among MSM is much lower They got the transmission coefficient is 0.386, the diagnose
than that for other high-risk population [15]. Meanwhile, we rate for HIV-positive individuals to be 0.304, and the
estimated that the antiviral therapy coverage rate among MSM proportion of HIV-positive individuals when diagnosed to be
in 2011 is less than the estimation by Ministry of Health, 0.864. Now we use them as the prior information for our
People’s Republic of China [2]. This agrees with the estimation. That is to say, the initial values for 0, , and are
conclusions obtained by Tong [34] that the antiviral therapy 0.386, 0.304, and 0.864, respectively. What is more, the
coverage rate for individuals infected sexually is lower than ranges for each parameter and initial value are given: 0 ∈
other infected groups. (0,1), ∈ (0.01,0.5), ∈ (0,1), 2 ∈ (0.3,0.4177), ∈
Simulation results show that strengthening education to (0,0.5), and (0) ∈ (1000,8000). Here we assume 1 = 2 since
high-risk population and increasing surveillance and testing there are no significant differences for the infectiousness
can slow down the spread of disease. Further, sensitivity of people in these two groups. Let ̃ and ̃ denote ( ) ( )
analysis implies that the most influential parameters are the real data, the number of annual reported HIV-positive
infection rate 0 and disease related death rate for HIV-positive individuals and AIDS patients from year 2005 to 2009, where
= 2005,2006,...,2009. The proposal density is chosen to be
individuals . Note that high ef f icacy drug can reduce the
transmission probability of HIV per high-risk behavior [18], multivariate normal distribution. Let = ( 0 , , , , (0)), and 1(
and the education may reduce the contact rate and increase , ) denotes the numerical solution of the first equation of
the condom use rate. This means that a high effective drug reduced model (A.1) at time . T hen the real value
of ̃ and ̃ can be taken as random variables from Gaus-
and timely education may effectively control HIV epidemic. ( ) ( )
sian distributions with mean 1( , ) and (1 − ) 1( , ),
In this paper, we concluded that if the infectiousness for
respectively. During the MCMC simulation, we will
HIV/AIDS infected cases is relatively small, treatment started minimize
immediately once diagnosed is more beneficial to disease =2009 2 2
+( ( )−(1− ) 1( , )). We
=∑
=2005 ( ( )− 1( , ))
control. It should be mentioned that we have not considered
use a small MCMC package to achieve the estimation
costs of antiviral therapy. However, early antiviral therapy will which provided in http://www.helsinki.fi/∼mjlaine/mcmc/.
increase the financial burden of the government of China and By fitting this reduced model to the annual reported
may increase high risk of occurrence of drug resistance. We HIV/AIDS cases from 2005 to 2009 we get the estimates of
will consider these factors in the future study.
parameters 0, , , 2, , and (0) and their standard deviations.
Based on these parameters, and (0) are estimated by
fitting model (1) to the data.
 14 BioMed Research International
1 ∗ 1.5

0.8 1
0.6 ∗
0.4 0.5
∗ ∗ ∗
0.2 ∗ ∗ ∗ ∗ ∗
∗

0 0
∗

∗ ∗
−0.2 ∗ ∗ ∗

−0.4 −0.5 ∗
∗
−0.6 ∗
−1 ∗
−0.8
−1 ∗ −1.5

01 2 I1 I2 A T1 T2 1 2 1 2 I T A U S(0) I(0) 01 2 I1 I2 A T2 2 2 I T A

(a) (b)

Figure 10: (a) Partial rank correlation coefficients (PRCC) results for the dependence of 0 on each parameter. (b) Partial rank correlation
coefficients (PRCC) results for the dependence of total HIV/AIDS cases in year 2015 on each parameter. ∗ denotes the value of PRCC is
not zero significantly, where the significance level is 0.05.

B. Basic Reproduction Number 0

+ 111+1112+1212+222 2525623636
The model has a disease-free equilibrium 0 = ( 0 ,0,0,
0,0,0,0), where 0 = /( + , )which can be obtained easily 1112 1122 222
by setting the right side of model (1) to zero. Following the + + + ].
2564 2364 364
next-generation operator method of [28] we can get 0 of
(B.2)
system (1). Let 0 = (1 − , 1 ) = + + , 2 = + + 1+ 1, 3 =
+ + 2+ 2, 4 = + , 5 = + + 1, and 6 = + + 2; then
C. Theoretical Analysis of System (1)
01 02 01 02
0 0 0
0 0 0 0 0 0 0 C.1. DFE Is Globally Attractive When 0<1
0
( 0 0 0 0 0 0) Proof. Since ≥ , so ( ) ≤ (1 − )[ + 1 1 + 2 2 +
(
= 0 0 0 0 0 0 0 ) ,
( )
0 0 0 0 0 0 0
1
1 + 2 2 + ( + )]; then, ≤ 0 [ + 1 1 + 2 2 +
0 0 0 0 0 0 0 1
1+ 2 2+ ( + )]− 1 . T hen we get the comparison
0 0 0 0 0 0 0) system of system (1):
(
1 0 0 0 0 0 0
− 1 2 0 0 0 0 0 1 =0[ 1 + 2 2 + 2 3 + 1 5
( − 2 1−
3 0 0 0 0)
( 0
= −(1 − ) 2 −4 0 0 0 ), + + ( + 7)]− 1 1,
( ) 2 6 4
0 − 0 0 0 0
1 5
0 0 − 2 0− 1 6
0
2 = 1 1 − 2 2,
( 0 0 000− 2 4)

(B.1)
such that 3 =21+ 12 − 3 3,
(C.1)
−1 4 = (1− ) 1 + 2 3 − 4 4,
0 = ( )
= 5 = 1 2 − 5 5,
0 11 112 22
[1+ + +
1 2 23 3 6 =23+ 15 − 6 6,
(1− )
112 22
+ + + 7 = 2 6 − 4 7.
234 34 4
BioMed Research International 15
Construct the Lyapunov function as
8 8 8
point dissipative on + , and : + → + is compact. By [7], it
1 1 2 1 follows that admits a global attractor.
12
Define = {( 0 , 0, 10, 20, 0, 10, 20, 0) ∈ 0 : ( 0,
= 1+( + +
0, 10, 20, 0, 10, 20, 0) ∈ 0 , for all ≥ 0}. Now we will
0 2 23 234
2
verify that
21 221
+ + = {( ,0,0,0,0,0,0,0) : ≥ 0}. (C.5)
236 2364
1 2
In fact, it is obvious that {( ,0,0,0,0,0,0,0) : ≥ 0} ⊆ , for
1 11 121
+ + + ) 2
any ( 0, 0, 10, 20, 0, 10, 20, 0) ∈ 0 \{( ,0,0,0,
25 256 25 64 (C.2) 0,0,0,0) : ≥ 0}.
2 2 2 2 22 (1) If 0 > 0, then at least one of 10, 20, 0, 10 , 20,
+( + + + ) 3

3 34 36 364
and 0 is equal to zero. Let us assume that 10 = 0,
then 1(0) = (0) > 0. Similarly, when anyone
1

1 2 ̇̇
1 12
+ 4 +( + + ) 5 of 20 , 0, 10 , 20, and 0 is equal to 0, we can get
4 5 56 564 ̇̇ ̇̇ ̇ ̇ ̇ . 2(0), (0), 1(0), 2(0), (0) > 0
2
2
+( + ) 6 + 7. (2) If 0 = 0, then at least one of 10, 20, 0, 10, 20, 0 is
6 64 4 greater than 0. Let us assume that , then ̇ 10 > 0 (0) ≥
What is more, 0 (0)( 1 1(0)/ (0)) > 0, when any one of 20, 0,
=( − (C.3) 10, 20, 0 is greater than 0, we can get the similar
01 1) =( 0 − 1) 1 1. conclusion.
0 0 0 According to the above verification, we can get that for >
0sufficiently small, we have ( ( ), ( ), 1( ), 2( ), ( ),
It is easy to see that, when 0 < 1, we have ≤ 0, if and only
1( ), 2( ), ( )) ∉ 0; then ( ( ), ( ), 1( ), 2( ), ( ), 1(
if 1 = 0, = 0. That is to say system (C.1)’s largest invariant ), 2( ), ( )) ∉ ; that is to say, for any ( 0, 0, 10, 20, 0,
set is = { 1 = 0}. According to Lasalle invariant set 10 20, 0) ∉ {( ,0,0,0,0,0,0,0) : ≥ 0} there must be ( 0,
,
principle, when → ,0solutions of system (C.1) satisfy 1 → 0, 10, 20, 0, 10, 20, 0) ∉ ; then ⊆ {( ,0,0,0,0,0,0,0) : ≥
0. Then we get the terminal system of system (C.1): 0}. So, = {( ,0,0,0,0,0,0,0) :
≥ 0}.
2 = − 2 2, Obviously, 0( 0,0,0,0,0,0,0,0) is a fixed point of
3 = 1 2 − 3 3, in . If ( ( ), ( ), 1( ), 2( ), ( ), 1( ), 2( ), ( )) is a
4 = 2 3 − 4 4, solution initiating from , then when → ∞,we have ( ) →
(C.4) 0, ( ) → 0, 1( ) → 0, 2( ) → 0, and ( ) → 0, 1( ) → ,
5 = 1 2 − 5 5,
0 2( ) → 0, ( ) → 0. Since every solution initiating from
6 = 2 3 + 1 5 − 6 6, will stay in forever, so 0 is an isolated invariance.
Since 0 > 1, there exists a small enough such that
7 = 2 6 − 4 7.
∗
− (C.6)
Obviously, all the eigenvalues of (C.4) are negative. Since 1− ∗ 0 < 0.
+8
system (C.4) is a linear differential system with constant
coefficients, so the equilibrium (0,0,0,0,0,0,0) is globally The following proof is that ( 0) ∩ 0 = . T hat is to say, there
asymptotic stable. Then, the solutions of system (C.1) satisfy exists a positive constant 0, such that for any solution ( ,
0 0
1 → 0, 2 → 0, 3 → 0, 4 → 0, 5 → 0, 6 → 0, 7 → 0. )initiating from 0, lim →∞sup ‖ ( , )− 0‖ ≥ 0. Suppose
From the comparison principle, we get that the disease-free that for any positive 0 (let 0 = ),
equilibrium of system (1) is globally attractive. lim 0
)− 0 < . (C.7)
sup ( ,

→∞

C.2. The Uniform Persistence of the System

Then, there exists 0 > 0 large enough such that when >
8 7
Proof of Theorem 2. Let = + , 0 = + × Int( + ), and 0 := \ 0,0−< ( )< 0 + , ( ) < , 1( ) < , 2( ) < , ( ) < , 1( )

0. < , 2( ) < , ( ) < . It follows that 0 − < ( ) <

+ 8 . What is more, we have
8 8 0 0
Define a map : + → + , which satisfies ( ) = ( ,
0 0
), where ( , ) is the unique solution starting from the ≥0⋅( − )
0 0 0
initial value (0, ) = . T hat is to say, is the solution ×+ 1 1 + 2 2 + + 1 +
semiflow. Firstly, we show that is uniformly persistent with 1 2 2 +
−1 .
respect to ( 0, 0). It is easy to see that and 0 are positively 0 +8
invariant and the solutions of system (1) are uniformly and (C.8)
ultimately bounded. Thus the semi-flow is
16 BioMed Research International

Now we get the comparison system: 2

We only need to verify that ( 1( + + 1 )/ 2 ) + (( + +
=0⋅( − ) 2 )/ 3) − ( 1 / 2 3) > 0. Reducing it to a common
1 0
denominator, we get the numerator which is denoted by .
× 1 +12+23+ 4 +15+26+ 40 +8 We just verify that >0:

− 1 1, = 1( + + 1)( + + 2 )
2
2 = 1 1 − 2 2, +( + + 1 + ) ( + + 2)− 1 ( + + 1 )

3 = 2 1 + 1 2 − 3 3,
= 1( + + 1)( + + 2 + )
4 = (1 − ) 1 + 2 3 − 4 4, 2
+( + + 1 + ) ( + + 2)− 1 ( + + 1 )
5 = 1 2 − 5 5,
2
6 = 2 3 + 1 5 − 6 6, = 1( + )+ 1( 1 + 2 + )( +

2
7 = 2 6 − 4 7. + 11 ( 2 + )+( + + 1 + )( + )
(C.9)
2
+( + + + )2
The characteristic equation of (C.9) is 1

7 6 5 4 2 2
=+1+2+3 − 1 ( + )− 1 − 1
(C.10)
3 2 1 = 1( + )( + + 1 + 2 + )
+4 +5 +6 +7,
where 7 = 1 2 3 4 5 6 4(1− 0(( 0 − )/( 0 +8 ))) < 0. 2
1234567 +( + + 1 + ) ( + )+ 11 ( 2 + )
Since = − 7 > 0, so it has at least one
∗ 2
of positive real eigenvalues. Let us assume that > 0, +( + ) 2 +2 2 ( 1 + )( +
( 1, 2, 3, 4, 5 , 6, 7) is the eigenvector. Suppose 1 > 0;
∗ ∗ 2 2 2
then 2 =11 /( + 2) > 0, 3 = ( 2 1 + 1 2)/( + + 2( + ) − 1 ( + )− 1 − 1
∗ 1
3) > 0, 4 = ((1 − ) 1 + 2 3)/( + 4 ) > 0, 5 =
∗ ∗
1 2/(+5) > 0, 6 = ( 2 3 + 1 5)/( + 6 ) > 0, = 1( + )( + + 1 + 2 + )
∗
and 7 = 2 6/( > 0. That is to say there exists a
+7)
2 2
positive real eigenvalue which has a positive eigenvector. +( + + 1 + ) ( + )+ 2( + )
This implies that the solution tends to infinity as → 2
∞,which contradicts (C.7). This completes the proof. +(2 1 +2 1 − 1 )( + )

3 2 22
D. The Proof for >0 + 1+2 1 + 12 + 11

2 2
Proof. Suppose 1 = 2 , 2 = 1 , where > 1. Since 1 < 2, so 2 <
+ 11 − −1
1

3. Meanwhile, 1 < 1 < 2 and < , so 5 < 3. Consequently, 1/

= 1( + )( + + 1 + 2 + )
2 > 1/ 3 and 1/ 5 > 1/ 3. T hen we have
2 2
1 ( + +1) ++2 +( + + 1 + ) ( + )+ 2( + )
= 2 + 2
1 2 56 3 6 2
1 +(2 1 +2 1 − 1 )( + )
1 ( 2 + 3)
3 2 22
+ − 2 2 + 1+ 1 + 11 + 11
236
2 36
>0.
1 ( ++1) ++2 1
> 2 + 2 + (D.2)
236
23 6 36
− − This completes the proof.
1 1
2 2
236 236
Acknowledgments
1 ( ++1) ++2 1
> 2 + 2 − 2 The authors are supported by the National Megaproject of
23 6 36 23 6 Science Research no. 2012ZX10001-001, by the National Nat-
= 1 [ 1 ( + + 1) + + + 2
− 1
]. ural Science Foundation of China (NSFC, 11171268 (YX)), by
2 the Fundamental Research Funds for the Central Universities
36 2 3 23
(GK 08143042 (YX)), and by the International Development
(D.1)
Research Center, Ottawa, Canada (104519-010).
BioMed Research International [15] Y. Xiao, S. Tang, Y. Zhou et al., “Predicting an HIV/AIDS
epi-demic and measuring the effect on it of population
mobility in
References
[1] Ministry of Health, People’s Republic of China, Joint
United Nations Programme on HIV/AIDS, World Health
Organiza-tion, 2009 Estimates for the HIV/AIDS
Epidemic in China, Ministry of Health China, UNAIDS
and WHO, Beijing, China, 2010.
[2] Ministry of Health, People’s Republic of China, Joint
United Nations Programme on HIV/AIDS, World Health
Organi-zation, 2011 Estimates for the HIV/AIDS
Epidemic in China, Ministry of Health China, UNAIDS
and WHO, Beijing, China, 2011.
[3] L. A. Valleroy, D. A. Mackellar, J. M. Karon et al., “HIV
prev-alence and associated risks in young men who have
sex with men,” Journal of the American Medical
Association, vol. 284, no. 2, pp. 198–204, 2000.
[4] D. P. Wilson, M. G. Law, A. E. Grulich, D. A. Cooper,
and J. M. Kaldor, “Relation between HIV viral load and
infectiousness: a model-based analysis,” The Lancet, vol.
372, no. 9635, pp. 314– 320, 2008.
[5] J. M. Gran, L. Wasmuth, E. J. Amundsen, B. H. Lindqvist,
and
O. O. Aalen, “Growth rates in epidemic models:
application to a model for HIV/AIDS progression,”
Statistics in Medicine, vol. 27, no. 23, pp. 4817–4834,
2008.
[6] R. M. Anderson, R. M. May, M. C. Boily, G. P. Garnett,
and J. T. Rowley, “The spread of HIV-1 in Africa: sexual
contact patterns and the predicted demographic impact of
AIDS,” Nature, vol. 352, no. 6336, pp. 581–589, 1991.
[7] J. M. Hyman and J. Li, “T he reproductive number for an
HIV model with differential infectivity and staged
progression,” Linear Algebra and Its Applications, vol.
398, no. 1–3, pp. 101–116, 2005.
[8] J. M. Hyman, J. Li, and E. Ann Stanley, “T he dif ferential
infectivity and staged progression models for the
transmission of HIV,” Mathematical Biosciences, vol. 155,
no. 2, pp. 77–109, 1999.
[9] C. C. McCluskey, “A model of HIV/AIDS with staged
progres-sion and amelioration,” Mathematical
Biosciences, vol. 181, no. 1, pp. 1–16, 2003.
[10] Y. Zhou, Y. Shao, Y. Ruan et al., “Modeling and
prediction of HIV in China: transmission rates structured
by infection ages,” Mathematical Biosciences and
Engineering, vol. 5, no. 2, pp. 403– 418, 2008.
[11] J. Li, C. Luo, and N. de Klerk, “Trends in infant/child
mortality and life expectancy in Indigenous populations
in Yunnan Province, China,” Australian and New
Zealand Journal of Public Health, vol. 32, no. 3, pp.
216–223, 2008.
[12] R. M. Granich, C. F. Gilks, C. Dye, K. M. de Cock, and B.
G. Williams, “Universal voluntary HIV testing with
immediate antiretroviral therapy as a strategy for
elimination of HIV transmission: a mathematical model,”
The Lancet, vol. 373, no. 9657, pp. 48–57, 2009.
[13] J. Lou, J. Wu, L. Chen, Y. Ruan, and Y. Shao, “A sex-
role-preference model for HIV transmission among men
who have sex with men in China,” BMC Public Health,
vol. 9, supplement 1, article S10, 2009.
[14] X. Xu, Y. Xiao, and N. Wang, “Modeling sexual
transmission of HIV/AIDS in Jiangsu province, China,”
Mathematical Methods in the Applied Sciences, vol. 36, no.
2, pp. 234–248, 2013.
 17

mainland China,” Journal of T heoretical Biology, vol. 317, pp.

271– 285, 2013.
[16] Ministry of Health, People’s Republic of China, Joint United
Nations Programme on HIV/AIDS, World Health Organiza-
tion, 2007 Estimates for the HIV/AIDS Epidemic in China,
Ministry of Health China, UNAIDS and WHO, Beijing, China,
2007.
[17] F. Zhang, Z. Dou, Y. Ma et al., “Effect of earlier initiation of
antiretroviral treatment and increased treatment coverage on
HIV-related mortality in China: a national observational cohort
study,” The Lancet Infectious Diseases, vol. 11, no. 7, pp. 516–
524, 2011.
[18] P. L. Vernazza, B. L. Gilliam, M. Flepp et al., “Effect of
antiviral treatment on the shedding of HIV-1 in semen,” AIDS,
vol. 11, no. 10, pp. 1249–1254, 1997.
[19] M. S. Cohen, I. F. Hoffman, R. A. Royce et al., “Reduction of
concentration of HIV-1 in semen after treatment of urethritis:
implications for prevention of sexual transmission of HIV-1,”
The Lancet, vol. 349, no. 9069, pp. 1868–1873, 1996.
[20] R. F. Baggaley, G. P. Garnett, and N. M. Ferguson, “Modelling
the impact of antiretroviral use in resource-poor settings,”
PLoS Medicine, vol. 3, no. 4, pp. 493–504, 2006.
[21] S. Tang and Y. Xiao, Biological Dynamic System for Single-
Species, Science Publishing House, Beijing, China, 2008.
[22] W. R. Gilks, S. Richardson, and D. J. Spiegelhalter, Markov Chain
Monte Carlo in Practice, Chapman and Hall, London, UK, 1996.
[23] A. E. Gelfand, S. E. Hills, A. Racine-Poon, and A. F. M. Smith,
“Illustration of Bayesian inference in normal data models using
gibbs sampling,” Journal of the American Statistical
Association, vol. 85, no. 412, pp. 972–985, 1990.
[24] D. J. Lunn, N. Best, A. Thomas, J. Wakefield, and D. Spiegel-
halter, “Bayesian analysis of population PK/PD models: general
concepts and software,” Journal of Pharmacokinetics and Phar-
macodynamics, vol. 29, no. 3, pp. 271–307, 2002.
[25] Y. Huang, D. Liu, and H. Wu, “Hierarchical Bayesian methods for
estimation of parameters in a longitudinal HIV dynamic system,”
Biometrics, vol. 62, no. 2, pp. 413–423, 2006.
[26] H. Putter, S. H. Heisterkamp, J. M. A. Lange, and F. de Wolf,
“A Bayesian approach to parameter estimation in HIV
dynamical models,” Statistics in Medicine, vol. 21, no. 15, pp.
2199–2214, 2002.
[27] H. Haario, M. Laine, A. Mira, and E. Saksman, “DRAM: ef f
icient adaptive MCMC,” Statistics and Computing, vol. 16, no.
4, pp. 339–354, 2006.
[28] S. Tang, Y. Xiao, Y. Yang, Y. Zhou, J. Wu, and Z. Ma, “Com-
munity-based measures for mitigating the 2009 H1N1 pandemic in
China,” PLoS One, vol. 5, no. 6, Article ID e10911, 2010.
[29] P. van den Driessche and J. Watmough, “Reproduction numbers
and sub-threshold endemic equilibria for compartmental mod-
els of disease transmission,” Mathematical Biosciences, vol.
180, no. 1, pp. 29–48, 2002.
[30] O. Diekmann, J. A. Heesterbeek, and J. A. Metz, “On the
definition and the computation of the basic reproduction ratio
R0 in models for infectious diseases in heterogeneous
populations,” Journal of Mathematical Biology, vol. 28, no. 4,
pp. 365–382, 1990.
[31] H. L. Smith and X.-Q. Zhao, “Robust persistence for semi-
dynamical systems,” Nonlinear Analysis: Theory, Methods &
Applications, vol. 47, no. 9, pp. 6169–6179, 2001.
[32] Y. Xiao, J. Sun, C. Li et al., “Prevalence and correlates of HIV and
syphilis infections among men who have sex with men in
18 BioMed Research International

seven Provinces in China with historically low HIV

prevalence,” Journal of Acquired Immune Deficiency
Syndromes, vol. 53, supplement 1, pp. S66–S73, 2010.
[33] K.-H. Choi, H. Lui, Y. Guo, L. Han, and J. S. Mandel, “Lack
of HIV testing and awareness of HIV infection among men
who have sex with men, Beijing, China,” AIDS Education
and Prevention, vol. 18, no. 1, pp. 33–43, 2006.
[34] G. Tong, MSM Volunteers Participation in HIV/AIDS
Prevention Activities, Public Publication, 2001.
[35] J. W. Dilley, W. J. Woods, and W. McFarland, “Are advances in
treatment changing views about high-risk sex?” The New
England Journal of Medicine, vol. 337, no. 7, pp. 501–502, 1997.
[36] J. A. Kelly, R. G. Hoffmann, D. Rompa, and M. Gray,
“Protease inhibitor combination therapies and perceptions of
gay men regarding AIDS severity and the need to maintain
safer sex,” AIDS, vol. 12, no. 10, pp. F91–F95, 1998.
[37] D. E. Ostrow, K. J. Fox, J. S. Chmiel et al., “Attitudes
towards highly active antiretroviral therapy are associated
with sexual risk taking among HIV-infected and uninfected
homosexual men,” AIDS, vol. 16, no. 5, pp. 775–780, 2002.
[38] S. Marino, I. B. Hogue, C. J. Ray, and D. E. Kirschner, “A
meth-odology for performing global uncertainty and
sensitivity analysis in systems biology,” Journal of
Theoretical Biology, vol. 254, no. 1, pp. 178–196, 2008.
[39] H. Nishiura, “Correcting the actual reproduction number: a
simple method to estimate R0 from early epidemic growth
data,” International Journal of Environmental Research and
Public Health, vol. 7, no. 1, pp. 291–302, 2010.
[40] G. Tong, An Inquiry into Commercial Sex in the Community
of Men Who Have Sex with Men in China, Gender Health
Education Institute, Beijing, China, 2007.