1,664
Views
4
CrossRef citations to date
0
Altmetric
Original Articles

Global threshold dynamics of an SVIR model with age-dependent infection and relapse

, &
Pages 427-454 | Received 25 Mar 2016, Accepted 16 Aug 2016, Published online: 03 Sep 2016

ABSTRACT

A susceptible-vaccinated-infectious-recovered epidemic model with infection age and relapse age has been formulated. We first address the asymptotic smoothness of the solution semiflow, existence of a global attractor, and uniform persistence of the model. Then by constructing suitable Volterra-type Lyapunov functionals, we establish a global threshold dynamics of the model, which is determined by the basic reproduction number. Biologically, it is confirmed that neglecting the possibility of vaccinees getting infected will over-estimate the effect of vaccination strategies. The obtained results generalize some existing ones.

1. Introduction

Vaccination is the administration of antigenic material (a vaccine) to stimulate an individual's immune system to develop adaptive immunity to a pathogen. Vaccines can prevent or ameliorate morbidity from infection. Vaccination is the most effective method of preventing infectious diseases [Citation6]. In the last decades, much attention has been paid on formulating and analysing epidemic models with vaccination to gain insights into the role played by vaccination. For example, by incorporating continuous vaccination strategy into a classical susceptible-infectious-recovered (SIR) model, Liu et al. [Citation10] proposed a vaccination model described by a system of four ordinary differential equations (ODEs). It is found that vaccination has an effect of decreasing the basic reproduction number and it is also established that the basic reproduction number governs the global dynamics of the model. In this model, individuals are divided into susceptible, vaccinated, infective, and recovered. The two underlying important assumptions are as follows. On the one hand, vaccinated individuals still have the possibility of being infected by contacting with infected individuals. On the other hand, vaccinated individuals are thought to have partial immunity and hence the effective contacts with infectious individuals may decrease compared with those of susceptibles. In this sense, the continuous vaccination strategy can be evaluated by the basic reproduction number because of the two infection paths, susceptible infection path and vaccinated infection path.

Nowadays, it has been recognized that the transmission dynamics of certain diseases could not be correctly described by the traditional compartmental epidemic models with no age structure. Models with (continuous) age structures are described by a hybrid system of ODEs and partial differential equations (PDEs) [Citation32]. Following the sprit of the works [Citation10, Citation15, Citation19], Wang et al. [Citation26] formulated a susceptible-vaccinated-exposed-infectious-recovered model with the structure of infection age (time elapsed since the infection). Under the assumption that the removal rate from the infectious class is constant instead of a function of the infection age, the model can be reformulated as a differential equation with an infinite delay. It is shown that the global threshold dynamics determined by the basic reproduction number still holds. Duan et al. [Citation5] investigated the global stability of a susceptible-vaccinated-infectious-recovered (SVIR) model with vaccination age. Compared with the models in [Citation10, Citation26], the model studied in [Citation5] did not take into account the fact that vaccinated individuals still have the possibility of being infected by contacting with infected individuals.

Furthermore, the study of van den Driessche and Zou [Citation24] and van den Driessche et al. [Citation23] indicated that relapse is an important feature of some animal and human diseases such as tuberculosis and herpes. Relapse is characterized by the reactivation of removed individuals who have been previously infected and then reverting back to the infectious class. In [Citation23, Citation24], relapse takes forms of a constant relapse period and of a general relapse distribution, respectively. Then, Liu et al. [Citation11] studied a susceptible-exposed-infectious-recovered epidemic model with age-dependent latency and relapse and proved that the threshold dynamics is preserved. Here, the recurrent phenomenon is characterized by PDEs with age-dependent relapse rate (or varying relapse rate) instead of a constant relapse period or a general relapse distribution.

Following the line of  Liu et al. [Citation10] and Wang et al. [Citation26]and assuming that before obtaining immunity the vaccinees still have the possibility of being infected by contacting with infected individuals, Wang et al. [Citation27] investigated the dynamics of a hybrid system of the SVIR model with the infection age.

Motivated by the above discussion, in this paper, we structure infected and recovered individuals by age of infection and by age of recovery, respectively. We formulate an SVIR model with two continuous structuring variables, termed as infection age and relapse age for short. Denote by S(t) and V(t) the numbers of susceptibles and vaccinees at time t, respectively. Let i(t,a) and r(t,b) be the densities of infected individuals at time t with infection age a and of recovered individuals at time t with relapse age b, respectively. The model to be studied is (1) dS(t)dt=ΛβS(t)0p(a)i(t,a)da(μ+α)S(t),dV(t)dt=αS(t)β1V(t)0p(a)i(t,a)daμV(t),t+ai(t,a)=(μ+δ(a))i(t,a),t+br(t,b)=(μ+η(b))r(t,b)(1) with boundary conditions (2) i(t,0)=(βS(t)+β1V(t))0p(a)i(t,a)da+0η(b)r(t,b)db,r(t,0)=0δ(a)i(t,a)da,(2) and initial conditions S(0)=S0R+:=[0,),V(0)=V0R+,i(0,)=i0L+1(0,),r(0,)=r0L+1(0,), where L+1(0,) is the set of functions on (0,) that are nonnegative and Lebesgue integrable. The meanings of the parameters in (Equation1) are summarized in Table .

Table 1. Biological meanings of parameters in (Equation1).

We mention that if all p(), δ(), and η() are constant functions with the constants p, δ, and η, respectively, then Equation (Equation1) reduces to the following system of ODES: (3) dS(t)dt=ΛβpS(t)I(t)(μ+α)S(t),dV(t)dt=αS(t)β1pV(t)I(t)μV(t),dI(t)dt=(βpS(t)+β1pV(t))I(t)(μ+δ)I(t)+ηR(t),dR(t)dt=δI(t)(μ+η)R(t),(3) where I(t)=0i(t,a)da and R(t)=0r(t,b)db. The case where η=0 of Equation (Equation3) is one of the special cases of the model studied by Wang et al. [Citation26] and a threshold dynamics was established. The main purpose of this paper is to establish such a result for (Equation1) and hence our result generalizes some existing ones.

Throughout this paper, we make the following assumption.

Assumption 1.1

  1. Λ, μ, α, β, β1>0.

  2. p, δ, ηL+(0,) with finite essential upper bounds p¯, δ¯, η¯, respectively.

  3. p, δ, and η are Lipschitz continuous on R+ with Lipschitz constants Mp, Mδ, and Mη, respectively.

In recent years, epidemic models and viral infection models with age-dependent structures have been extensively studied. To name a few, see [Citation13] for a two-group model with infection age, [Citation14] for an SIR model with infection age, [Citation11, Citation16] for SEIR models with infection age, [Citation3] for an SIRS model with infection age, [Citation30] for an SVIER model with infection age, [Citation1, Citation31, Citation33] for models on cholera with infection age, and [Citation2, Citation8, Citation17, Citation18, Citation28, Citation29] for viral infection. Generally, it is not easy to analyse a hybrid system coupled of ODEs and PDEs, especially for the global stability/attractivity of equilibria. As mentioned earlier, the goal of this paper is to establish a threshold dynamics for Equation (Equation1).

The organization of the remaining part of this paper is as follows. Some preliminary results on Equation (Equation1) are presented in Section 2. In Section 3, we show the asymptotic smoothness of the semiflow of Equation (Equation1). Section 4 is devoted to the uniform persistence of the system when the basic reproduction number is larger than one. In Section 5 and Section 6, we show that the disease dies out if the basic reproduction number is less than one and the endemic equilibrium is globally attractive if the basic reproduction number is larger than one by the technique of Lyapunov functionals, respectively. The paper concludes with a brief discussion.

2. Preliminaries

2.1. The solution semiflow

Denote Y=R+×R+×L+1(0,)×L+1(0,), the positive cone of the Banach space R×R×L1(0,)×L1(0,) equipped with induced product norm (x,y,ϕ,ψ)=|x|+|y|+ϕL1+ψL1for (x,y,ϕ,ψ)R×R×L1(0,)×L1(0,). If any initial value X0=(S0,V0,i0,r0)Y satisfies the coupling equations i(0,0)=(βS0+β1V0)0p(a)i0(a)da+0η(b)r0(b)db and r(0,0)=0δ(a)i0(a)da, then (Equation1) is well-posed under Assumption 1.1 due to Iannelli [Citation9] and Magal [Citation12]. In fact, for such solutions, it is not difficult to show that (S(t),V(t),i(t,),r(t,))Y for each tR+. In the sequel, we always assume that the initial values satisfy the coupling equations. Then, we obtain a continuous solution semiflow Φ:R+×YY defined by Φ(t,X0)=Φt(X0):=(S(t),V(t),i(t,),r(t,)),tR+,X0Y.

For the sake of convenience, we introduce the following notations: Ω(a)=e0a(μ+δ(τ))dτandΓ(b)=e0b(μ+η(τ))dτfor a,bR+. By Assumption 1.1, we have (4) 0Ω(a)eμaand0Γ(b)eμbfor all a,bR+.(4) For any solution of Equation (Equation1), we define the following functions on R+: P(t)=0p(a)i(t,a)da,Q(t)=0δ(a)i(t,a)da,R(t)=0η(b)r(t,b)dbfor tR+. Then the boundary conditions given in Equation (Equation2) can be rewritten as i(t,0)=[βS(t)+β1V(t)]P(t)+R(t) and r(t,0)=Q(t). Integrating the third and the fourth equations of (Equation1) along the characteristic lines ta=const. and tb=const. , respectively gives (5) i(t,a)=[(βS(ta)+β1V(ta))P(ta)+R(ta)]Ω(a)=i(ta,0)Ω(a)if 0at,i0(at)Ω(a)Ω(at)if 0ta,(5) and (6) r(t,b)=Q(tb)Γ(b)=r(tb,0)Γ(b)if 0bt,r0(bt)Γ(b)Γ(bt)if 0tb.(6) The boundedness of the solution semiflow of Equation (Equation1) is given in the following result.

Proposition 2.1

Define Ξ:=X0=(S0,V0,i0,r0)YX0Λμ. Then Ξ is a positively invariant subset for Φ, that is, Φ(t,X0)Ξfor all t0andX0Ξ. Moreover, Φ is point dissipative and Ξ attracts all points in Y.

Proof.

First, we have ddtΦt(X0)=dS(t)dt+dV(t)dt+ddti(t,)L1+ddtr(t,)L1. On the one hand, it follows from Equation (Equation5) and changes of variables that i(t,)L1=0ti(ta,0)Ω(a)da+ti0(at)Ω(a)Ω(at)da=0ti(σ,0)Ω(tσ)dσ+0i0(τ)Ω(t+τ)Ω(τ)dτ. Then di(t,)L1dt=i(t,0)Ω(0)+0i0(τ)Ω(τ)ddtΩ(t+τ)dτ+0ti(σ,0)ddtΩ(tσ)dσ. Note that Ω(0)=1 and (d/da)Ω(a)=(μ+δ(a))Ω(a) for almost all aR+. Using Equation (Equation5) and changes of variables, one can get di(t,)L1dt=i(t,0)0i0(τ)Ω(τ)(μ+δ(t+τ))Ω(t+τ)dτ0ti(σ,0)(μ+δ(tσ))Ω(tσ)dσ=i(t,0)0(μ+δ(a))i(t,a)da. Similarly, we can get dr(t,)L1dt=r(t,0)0(μ+η(b))r(t,b)db. These, combined with the first and the second equations of (Equation1), give us ddtΦt(X0)=ΛμΦt(X0)for tR+. It follows that (7) Φt(X0)=ΛμeμtΛμX0Λμ(7) for tR+ if X0Ξ. In summary, we have shown that Ξ is positively invariant with respect to Φ.

Lastly, it follows from Equation (Equation7) that lim suptΦt(X0)Λ/μ for any X0Y, that is, Φ is point dissipative and Ξ attracts all points in Y. This completes the proof.

The following result is a direct consequence of Proposition 2.1.

Proposition 2.2

Let AΛ/μ be given. If X0Y satisfies X0A, then the following statements hold for all tR+.

  1. S(t), V(t), i(t,)L1, r(t,)L1A;

  2. P(t)p¯A, Q(t)δ¯A and R(t)η¯A;

  3. i(t,0)β¯A and r(t,0)δ¯A, where β¯=βp¯A+β1p¯A+η¯.

It follows from Proposition 2.2, Assumption 1.1 and [Citation27, Proposition 4.1], we obtain the following basic properties of the functions P(t), Q(t) and R(t).

Proposition 2.3

For any solution of Equation (Equation1), the associated functions P(t), Q(t) and R(t) are Lipschitz continuous on R+ with Lipschitz constants, LP=(p¯β¯+p¯(μ+δ¯)+Mp)A,LQ=(δ¯β¯+δ¯(μ+δ¯)+Mδ)A,LR=(η¯δ¯+η¯(μ+η¯)+Mη)A, respectively.

2.2. Equilibria and the basic reproduction number

For Equation (Equation1), there always exists the infection-free equilibrium E0=(S0,V0,i0,r0):=Λμ+α,αΛμ(μ+α),0,0. Moreover, if an equilibrium is not infection-free then it must be endemic.

An endemic equilibrium E=(S,V,i,r) satisfies the following algebraic equations: (8) ΛβS0p(a)i(a)da(μ+α)S=0,αSβ1V0p(a)i(a)daμV=0,di(a)da=(μ+δ(a))i(a),dr(b)db=(μ+η(b))r(b),i(0)=(βS+β1V)0p(a)i(a)da+0η(b)r(b)db,r(0)=0δ(a)i(a)da.(8) From the third and fourth equations of (Equation8), we obtain i(a)=i(0)Ω(a) and r(b)=r(0)Γ(b). Substituting them into the last two equations of (Equation8) gives (9) 1=(βS+β1V)0p(a)Ω(a)da+0δ(a)Ω(a)da0η(b)Γ(b)db.(9)

In order to obtain an endemic equilibrium, we define a key threshold parameter. According to Diekmann et al. [Citation4] and van den Driessche and Watmough [Citation22], the expected number of secondary cases produced by a typical infectious individual during its entire period of infectiousness is defined as the basic reproduction number 0, which is given by 0=(βS0+β1V0)H+LK:=β+β1αμΛHμ+α+LK, where H=0p(a)Ω(a)da,L=0η(b)Γ(b)db,K=0δ(a)Ω(a)da. Note that Lη¯/(μ+η¯)<1 and Kδ¯/(μ+δ¯)<1. As we will see later, 0 serves as a sharp threshold parameter for Equation (Equation1), which completely determines the global behaviour of Equation (Equation1).

Biologically, Ω(a) is the probability of an infected individual still in the infected class at the infection age a and hence 0p(a)Ω(a)da is the total infection force. It follows that the first term in 0 is the average number of secondary cases directly produced by an infected individual introduced into a population with susceptibles and vaccinees only. Furthermore, 0δ(a)Ω(a)da represents the chance of recovery for an infected individual and 0η(b)Γ(b)db is the chance of a recovered individual can be infectious after relapse. Therefore, LK in 0 is the average number of secondary cases produced by infected individual after recovery through relapse.

Now, from the first and second equations of (Equation8), we have S=Λμ+α+βHi(0),V=αΛ(μ+α+βHi(0))(μ+β1Hi(0)). Plugging them into (Equation9) yields (10) G(i(0))=a0(Hi(0))2+a1Hi(0)+a2=0,(10) where a0=ββ1(1LK)>0,a1=βμ+β1(μ+α)ββ1ΛH(μ+α)β1LKμβLK,a2=μ(μ+α)(10). If 01, then a1>0 and G(0)=a20. It follows from the relationship between zeros and coefficients, G has no positive zeros and hence there is no endemic equilibrium. Now, suppose that 0>1. Then a2<0. By the relationship between zeros and coefficients, G has a unique positive zero and hence there is a unique endemic equilibrium (11) E=(S,V,i,r):=Λμ+α+βHi(0),αΛ(μ+α+βHi(0))(μ+β1Hi(0)), i(0)Ω(),Ki(0)Γ(),(11) where i(0) is the only positive zero of G defined by Equation (Equation10). In summary, we have obtained the following result.

Theorem 2.1

  1. System (Equation1) always has an infection-free equilibrium E0.

  2. If 0>1, then Equation (Equation1) also has a unique endemic equilibrium E=(S,V,i,r) defined by Equation (Equation11).

3. Asymptotic smoothness

A semiflow is asymptotically smooth if each forward invariant bounded closed set is attracted by a nonempty compact set. Since L1+(0,) is a part of Y, which belongs to infinite dimension, we cannot deduce precompactness only from boundedness. On the other hand, the global attractivity results will utilize the Lyapunov functional technique combined with the invariance principle. Due to [Citation25, Theorem 4.2 of Chapter IV], we need the asymptotic smoothness of the orbit {Φ(t,X0)|tR+} in Y in order to use the invariance principle. To this end, we first decompose Φ:R+×YY into the following two operators Θ, Ψ:R+×YY defined respectively by Θ(t,X0):=(0,0,ϕ~i(t,),ϕ~r(t,)),Ψ(t,X0):=(S(t),V(t),i~(t,),r~(t,)), where ϕ~i(t,a)=0if t>a0,i(t,a)if at0;ϕ~r(t,b)=0if t>b0,r(t,b)if bt0;i~(t,a)=i(t,a)if t>a0,0if at0;r~(b,t)=r(t,b)if t>b0,0if bt0. Then Φ(t,X0)=Θ(t,X0)+Ψ(t,X0) for tR+. Note that i~(t,a) and r~(t,b) can also be written as (12) i~(t,a)=[(βS(ta)+β1V(ta))P(ta)+R(ta)]Ω(a)if t>a0,0if at0,(12) and r~(t,b)=Q(tb)Γ(b)if t>b0,0if bt0, respectively.

Theorem 3.1

For X0Ξ, the orbit {Φ(t,X0)|t0} has a compact closure in Y if the following two conditions hold.

  1. There exists a function Δ:R+×R+R+ such that, for any r>0, limtΔ(t,r)=0 and if X0Ω with X0r then Θ(t,X0)Δ(t,r) for tR+.

  2. For tR+, Ψ(t,) maps any bounded sets of Ξ into sets with compact closure in Y.

Theorem 3.1 is proved by similar method used in [Citation32, Proposition 3.13] and [Citation21], that is, verifying the conditions (i) and (ii) of Theorem 3.1 in the following two lemmas.

Lemma 3.1

For r>0, let Δ(t,r)=eμtr. Then Δ satisfies condition (i) of Theorem 3.1.

Proof.

Obviously, limtΔ(t,r)=0. By Equations (Equation5) and (Equation6), ϕ~i(t,a)=0if t>a0i0(at)Ω(a)Ω(at)if at0andϕ~r(t,b)=0if t>b0,r0(bt)Γ(b)Γ(bt)if bt0. Then, for X0Ξ satisfying X0r and for tR+, we have Θ(t,X0)=|0|+|0|+ϕ~i(t,)L1+ϕ~r(t,)L1=ti0(at)Ω(a)Ω(at)da+tr0(bt)Γ(b)Γ(bt)db=0i0(σ)Ω(σ+t)Ω(σ)dσ+0r0(σ)Γ(σ+t)Γ(σ)dσ=0|i0(σ)eσσ+t(μ+δ(τ))dτ|dσ+0|r0(σ)eσσ+t(μ+η(τ))dτ|dσeμ0ti0L1+eμ0tr0L1eμtX0eμtr. This completes the proof.

Lemma 3.2

For tR+, Ψ(t,) maps any bounded sets of Ξ into sets with compact closure in Y.

Proof.

First, S(t) and V(t) remain in the compact set [0,Λ/μ][0,A] from Proposition 2.1. Thus, it suffices to show that i~(t,a) and r~(t,b) remain in a precompact subset of L+1(0,), which is independent of X0Ξ. To this end, we verify the following conditions for i~(t,a) and similar ones for r~(t,b) (see, for example, [Citation21, Theorem B.2]).

  1. The supremum of i~(t,)L1 with respect to X0Ξ is finite;

  2. limhhi~(t,a)da=0 uniformly with respect to X0Ξ;

  3. limh0+0|i~(t,a+h)i~(t,a)|da=0 uniformly with respect to X0Ξ;

  4. limh0+0hi~(t,a)da=0 uniformly with respect to X0Ξ.

It follows from Equations (Equation5) and (Equation12) that 0i~(t,a)=i(ta,0)Ω(a)if 0at,0if 0t<a. This, combined with Proposition 2.2 and Equation (Equation4), produces i~(t,a)β¯Aeμa=(βp¯A+β1p¯A+η¯)Aeμa, from which conditions (i), (ii) and (iv) directly follows.

Now, we are in a position to verify condition (iii). For sufficiently small h(0,t), it follows from the definition of i~ and Proposition 2.2 that (13) 0|i~(t,a+h)i~(t,a)|da=0th|i(t,a+h)i(t,a)|da+tht|0i(t,a)|da=0th|i(tah,0)Ω(a+h)i(ta,0)Ω(a)|da+tht|i(ta,0)Ω(a)|daΔ1+Δ2+β¯Ah,(13) where Δ1=0thi(tah,0)|Ω(a+h)Ω(a)|da and Δ2=0th|i(tah,0)i(ta,0)|Ω(a)da. We estimate Δ1 and Δ2 separately as follows.

We claim that Δ1β¯Ah. In fact, it follows from the fact that 0Ω(a)1 and Ω is a non-increasing function, we have 0th|Ω(a+h)Ω(a)|da=0th(Ω(a)Ω(a+h))da=0thΩ(a)da0thΩ(a+h)da=0thΩ(a)dahtΩ(a)da=0thΩ(a)dahthΩ(a)dathtΩ(a)da=0hΩ(a)dathtΩ(a)dah. This, combined with Proposition 2.2, gives Δ1β¯Ah.

Now we show that there exists M such that Δ2Mh/μ. Actually, it follows from Equation (Equation12) that Δ2=0th|[(βS(tah)+β1V(tah))P(tah)+R(tah)][(βS(ta)+β1V(ta))P(ta)+R(ta)]|Ω(a)da0th|(βS(tah)P(tah)(βS(ta)P(ta))|Ω(a)da+0th|(β1V(tah)P(tah)β1V(ta)P(ta))|Ω(a)da+0th|(R(tah)R(ta))|Ω(a)da. From Proposition 2.2 and the first equation of (Equation1), we can conclude that S(t) and V(t) are Lipschitz continuous on R+ with Lipschitz constants MS=Λ+(μ+α)A+βA2p¯ and MV=αA+(μ+α1)A+β1p¯A2, respectively. Furthermore, it follows from Proposition 2.3 that P(t) and R(t) are Lipschitz continuous with Lipschitz constants LP and LR, respectively. Thus S(t)P(t) and V(t)P(t) are Lipschitz continuous with Lipschitz constants MSP=ALP+p¯AMS and MVP=ALP+p¯AMV, respectively. Hence taking M=βMSP+β1MVP+LR, we have (14) Δ2Mh0theμadaMhμ.(14) It follows from Equations (Equation13)–(Equation14) that condition (iii) holds.

According to results on the existence of global attractors in [Citation7, Citation21], the following result is a consequence of Proposition 2.1, Theorem 3.1, Lemmas 3.1 and 3.2.

Theorem 3.2

The semiflow Φ has a global attractor A in Y, which attracts any bounded subset of Y.

4. The uniform persistence

The aim of this section is to show that Equation (Equation1) is uniformly persistent.

Define ρ:YR+ by ρ(S,V,i,r)=(βS+β1V)0p(a)i(a)da+0η(b)r(b)dbfor (S,V,i,r)Y. Let Y0={(S0,V0,i0,r0)Y:ρ(Φ(t0,(S0,V0,i0,r0)))>0 for some t0R+}. Obviously, if (S0,V0,i0,r0)YY0, then Φ(t,(S0,V0,i0,r0))E0 as t.

Definition 4.1

[Citation21, p. 61]

System (Equation1) is said to be uniformly weakly ρ-persistent (respectively, uniformly strongly ρ-persistent) if there exists an ϵ>0, independent of the initial conditions, such that lim suptρ(Φ(t,(S0,V0,i0,r0)))>ϵ (respectively, lim inftρ(Φ(t,(S0,V0,i0,r0)))>ϵ) for (S0,V0,i0,r0)Y0.

Let iˆ(t):=i(t,0) and rˆ(t):=r(t,0). Note that iˆ(t)=ρ(Φ(t,X0)). Then, Equations (Equation5) and (Equation6) can be rewritten as (15) i(t,a)=iˆ(ta)Ω(a)if ta0,i0(at)Ω(a)Ω(at)if at0,(15) and r(t,b)=rˆ(tb)Γ(b)if tb0,r0(bt)Γ(b)Γ(bt)if bt0, respectively. Substitute them into the boundary condition (Equation2) to obtain (16) iˆ(t)=(βS(t)+β1V(t))0tp(a)Ω(a)iˆ(ta)da+tp(a)Ω(a)Ω(at)i0(at)da+0tη(b)Γ(b)rˆ(tb)db+tη(b)Γ(b)Γ(bt)r0(bt)db(16) and (17) rˆ(t)=0tδ(a)Ω(a)iˆ(ta)da+tδ(a)Ω(a)Ω(at)i0(at)da.(17)

Lemma 4.1

If 0>1, then Equation (Equation1) is uniformly weakly ρ-persistent.

Proof.

We first get an estimate on iˆ(t) as follows. By Equation (Equation16), we have (18) iˆ(t)(βS(t)+β1V(t))0tp(a)Ω(a)iˆ(ta)da+0tη(b)Γ(b)rˆ(tb)db.(18)

By way of contradiction, for any ε>0, there exists an XεY0 such that lim suptρ(Φ(t,Xε))ε. Since 0>1, there exists a sufficiently small ϵ0>0 such that (19) βΛϵ0μ+αϵ0+β1α(Λϵ0μ+αϵ0)ϵ0μϵ00p(a)Ω(a)da+0η(b)Γ(b)db0δ(a)Ω(a)da>1.(19) In particular, there exists Xϵ0/2Y0 (for simplicity, denoted by X0 in the remaining of the proof) such that lim suptρ(Φ(t,X0))ϵ02. A contradiction is arrived at as follows.

Firstly, there exists TR+ such that ρ(Φ(t,X0))ϵ0for all tT. Without loss of generality, we assume that T=0 by replacing X0 with Φ(T,X0). Then it follows from the first equation of (Equation1) that dS(t)dtΛϵ0(μ+α)S(t)for all tR+. It follows that lim suptS(t)(Λϵ0)/(μ+α). As before, with possible replacing of the initial value, we can assume that (20) S(t)Λϵ0μ+αϵ0for all tR+.(20) Similarly, with Equation (Equation20) and the second equation of (Equation1), we can get (21) V(t)α(Λϵ0μ+αϵ0)ϵ0μϵ0for all tR+.(21) Combining Equations (Equation17), (Equation18), (Equation20) and (Equation21), we have iˆ(t)βΛϵ0μ+αϵ0+β1α(Λϵ0μ+αϵ0)ϵ0μϵ00tp(a)Ω(a)iˆ(ta)da+0tη(b)Γ(b)0tbδ(a)Ω(a)iˆ(tab)dadb for all tR+. Taking the Laplace transforms of both sides of the above inequality, we obtain L[iˆ]βΛϵ0μ+αϵ0+β1α(Λϵ0μ+αϵ0)ϵ0μϵ00p(a)Ω(a)eλadaL[iˆ]+0η(b)Γ(b)eλbdb0δ(a)Ω(a)eλadaL[iˆ]. Here, L[iˆ] denotes the Laplace transform of iˆ, which is strictly positive. Dividing both sides of the above inequality by L[iˆ] and letting λ0 give us 1βΛϵ0μ+αϵ0+β1α(Λϵ0μ+αϵ0)ϵ0μϵ00p(a)Ω(a)da+0η(b)Γ(b)db0δ(a)Ω(a)da, which contradicts with Equation (Equation19). This proves the claim and hence completes the proof.

In order to apply a technique used by Smith and Thieme [Citation21, Chapter 9], we consider a total Φ-trajectory. φ:RY is a total Φ-trajectory if Φ(s,φ(t))=φ(t+s) for all tR and sR+. A Φ-total trajectory φ=(S,V,i,r) satisfies i(t,a)=i(ta,0)Ω(a)=iˆ(ta)Ω(a)for tR and aR+,r(t,b)=r(tb,0)Γ(b)=rˆ(tb)Γ(b)for tR and bR+. Therefore, (22) dS(t)dt=ΛβS(t)0p(a)Ω(a)iˆ(ta)da(μ+α)S(t),dV(t)dt=αS(t)β1V(t)0p(a)Ω(a)iˆ(ta)daμV(t),iˆ(t)=(βS(t)+β1V(t))0p(a)Ω(a)iˆ(ta)da+0η(b)Γ(b)rˆ(tb)db,rˆ(t)=0δ(a)Ω(a)iˆ(ta)da(22) for tR. Substituting the fourth equation of (Equation22) into the third one, we have (23) iˆ(t)=(βS(t)+β1V(t))0p(a)Ω(a)iˆ(ta)da+0η(b)Γ(b)rˆ(tb)db=(βS(t)+β1V(t))0p(a)Ω(a)iˆ(ta)da+0η(b)Γ(b)0δ(a)Ω(a)iˆ(tab)dadb(23) for tR.

Lemma 4.2

For a total Φ-trajectory φ, S(t) is strictly positive on R and iˆ(t)=0 for all tR+ if iˆ(t)=0 for all t(,0].

Proof.

First, we show that S(t) is strictly positive on R. By way of contradiction, if we suppose that S(t)=0 for some number tR. Then dS(t)/dt=Λ>0 follows from the first equation of (Equation22), which implies that S(tη)<0 for sufficiently small η>0 and it contradicts with the fact that the total Φ-trajectory φ remains in Y. This completes the proof of S(t) is strictly positive on R.

Next, we show that iˆ(t)=0 for all tR+ if iˆ(t)=0 for all t(,0]. By changing the variables, we can rewrite Equation (Equation24) as iˆ(t)=(βS(t)+β1V(t))tp(ta)Ω(ta)iˆ(a)da+tη(tb)Γ(tb)bδ(bc)Ω(bc)iˆ(c)dcdb for tR. If iˆ(t)=0 for all t(,0] then iˆ(t)(βS¯+β1V¯)p¯0tiˆ(a)da+η¯δ¯0t0biˆ(c)dcdbfor tR, where S¯ and V¯(0,) are upper bounds for S and V , respectively. This is a Gronwall-like inequality and hence iˆ(t)=0 for all tR+. In fact, let Iˆ(t):=0tiˆ(a)da+0t0biˆ(c)dcdb,tR+. Then, for tR+, dIˆ(t)dt=iˆ(t)+0tiˆ(a)da(βS¯+β1V¯)p¯0tiˆ(a)da+η¯δ¯0t0biˆ(c)dcdb+0tiˆ(a)damax(η¯δ¯,(βS¯+β1V¯)p¯+1)Iˆ(t), which implies Iˆ(t)Iˆ(0)emax(η¯δ¯,(βS¯+β1V¯)p¯+1)t=0,tR+.

According to Lemma 4.2, we have the following result that a total Φ-trajectory φ enjoys.

Lemma 4.3

For a total Φ-trajectory φ, iˆ is strictly positive or identically zero on R.

Proof.

From Lemma 4.2, we can conclude that iˆ(t)=0 for all tt if iˆ(t)=0 for all tt by performing appropriate shifts, where tR is arbitrary. Hence, we have that either iˆ is identically zero on R or there exists a decreasing sequence {tj}j=1 such that tj as j and iˆ(tj)>0. In the latter case, letting iˆj(t)=iˆ(t+tj) for t0, we have from Equation (23) that, for tR+, iˆj(t)(βS_+β1V_)0tp(a)Ω(a)iˆj(ta)da+jˆj(t), where S_=inftRS(t)>0, V_=inftRV(t)>0 and jˆj(t)=(βS(t+tj)+β1V(t+tj))tp(a)Ω(a)iˆj(ta)da+0η(b)Γ(b)0δ(a)Ω(a)iˆj(tab)dadb. Since jˆj(0)=iˆ(tj)>0 and jˆj(t) is continuous at 0, it follows from Corollary B.6 of Smith and Thieme [Citation21] that there exists a number t>0 (which depends only on (βS_+β1V_)p(a)Ω(a)) such that iˆj(t)>0 for all t>t or iˆ(t)>0 for all t>t+rj. Since tj as j, we obtain that iˆ(t)>0 for all tR by letting j. This completes the proof.

According to Theorem 3.2, Lemmas 4.1–4.3, and the Lipschitz continuity of iˆ (which immediately follows from Proposition 2.3), we can apply results as in [Citation21, Theorem 5.2] to Φ. The precise results are as follows.

Theorem 4.1

If 0>1, then the semiflow Φ is uniformly (strongly) ρ-persistent.

When 0>1, the uniform persistence in Y0 of (Equation1) immediately follows from Theorem 4.1. In fact, it follows from Theorem Equation15 that i(t,)L10tiˆ(ta)Ω(a)da and hence from a variation of the Lebesgue–Fatou lemma [Citation20, Section B.2], we get lim infti(t,)L1iˆ0Ω(a)da, where iˆ=lim inftiˆ(t). Under Theorem 4.1, there exists a positive constant ϵ>0 such that iˆ>ϵ if 0>1 and hence the persistence of i(t,a) with respect to L1 follows. By a similar argument, we can prove that S(t) and V(t) are persistent with respect to || and r(t,) is persistent with respect to L1. To summarize, we get the following result.

Theorem 4.2

If 0>1, then the semiflow Φ is uniformly persistent in Y0, that is, there exists a constant ϵ>0 such that, for each X0Y0, lim inftS(t)ϵ,lim inftV(t)ϵ,lim infti(t,)L1ϵ,lim inftr(t,)L1ϵ.

5. Global stability of the infection-free equilibrium

The aim of this section is to establish the global stability of the infection-free equilibrium E0 when 0<1. We first show the local stability of E0.

Theorem 5.1

The infection-free equilibrium E0 is locally asymptotically stable if 0<1 and is unstable if 0>1.

Proof.

Linearizing (Equation1) at E0 by introducing the perturbation variables x1(t)=S(t)S0,x2(t)=V(t)V0,x3(t,a)=i(t,a),x4(t,a)=r(t,a), we obtain the linearized system dx1(t)dt=(μ+α)x1(t)βS00p(a)x3(t,a)da,dx2(t)dt=αx1(t)β1V00p(a)x3(t,a)daμx2(t),t+ax3(t,a)=(μ+δ(a))x3(t,a),t+bx4(t,b)=(μ+η(b))x4(t,b),x3(t,0)=(βS0+β1V0)0p(a)x3(t,a)da+0η(b)x4(t,b)db,x4(t,0)=0δ(a)x3(t,a)da. Set x1(t)=x10eλt, x2(t)=x20eλt, x3(t,a)=x30(a)eλt, and x4(t,b)=x40(b)eλt. After some calculation, we get the characteristic equation at E0, which is C(λ)(βS0+β1V0)0p(a)eλaΩ(a)da+0η(b)eλbΓ(b)db0δ(a)eλaΩ(a)da1=0. First, suppose that 0>1. Then C(0)=01>0. This, combined with limλC(λ)=1 and the intermediate value theorem, implies that C(λ) has a positive zero. Therefore, E0 is unstable if 0>1. Now assume that 0<1. We claim that C(λ) has no zeros with nonnegative real parts. Otherwise, C(λ)=0 for some λC with Re(λ)0. Then, we have 1=|C(λ)|0, a contradiction to 0<1. This proves the claim and hence E0 is locally asymptotically stable if 0<1.

In the following, we obtain the global attractivity of E0 by using the Lyapunov technique. During the discussion, we need the function g:xx1lnx,x(0,). It is easy to check that g is continuous and concave up and has the global minimum 0 only at x=1.

Theorem 5.2

The infection-free equilibrium E0 of Equation (Equation1) is globally asymptotically stable if 0<1.

Proof.

By Theorem 5.1, it suffices to show that E0 is globally attractive and we show it by the Lyapunov method. Consider the Lyapunov functional candidate, L(t)=L1(t)+L2(t)+L3(t), where L1(t)=S0gS(t)S0+V0gV(t)V0,L2(t)=0φ1(a)i(t,a)da,L3(t)=0ψ(b)r(t,b)db. Here, the nonnegative kernel functions φ(a) and ψ(b) will be determined later. The derivative of L1 along the solutions of Equation (Equation1) is calculated as follows: dL1(t)dt=1S0SΛβS0p(a)i(t,a)da(μ+α)S+1V0VαSβ1V0p(a)i(t,a)daμV=1S0S(μ+α)S0βS0p(a)i(t,a)da(μ+α)S+1V0VαSβ1V0p(a)i(t,a)daμV=μS02SS0S0S+αS03VV0SV0S0VS0S+βS0+β1V00p(a)i(t,a)da(βS+β1V)0p(a)i(t,a)da=dS02SS0S0S+αS03VV0SV0S0VS0S+(βS0+β1V0)0p(a)i(t,a)dai(t,0)+0η(b)r(t,b)db. Using integration by parts, we have dL2(t)dt=0φ1(a)i(t,a)tda=0φ1(a)(μ+δ(a))i(t,a)+i(t,a)ada=φ1(a)i(t,a)|0+0φ1(a)i(t,a)da0φ1(a)(μ+δ(a))i(t,a)da=φ1(0)i(t,0)+0(φ1(a)φ1(a)(μ+δ(a)))i(t,a)da. Similarly, dL3(t)dt=ψ(0)r(t,0)+0(ψ(b)ψ(b)(μ+η(b)))r(t,b)db. Now we choose ψ(b)=bη(u)ebu(μ+η(ω))dωdu and φ1(a)=a[(βS0+β1V0)p(u)+ψ(0)δ(u)]eau(μ+δ(ω))dωdu. Then ψ(0)=0η(b)Γ(b)db=L, φ1(0)=0, and ψ and φ1 satisfy ψ(b)ψ(b)(μ+η(b))+η(b)=0,φ1(a)φ1(a)(μ+δ(a))+(βS0+β1V0)p(a)+ψ(0)δ(a)=0. Therefore, we have dL(t)dt=dS02SS0S0S+αS03VV0SV0S0VS0S+βS0+β1V00p(a)i(t,a)dai(t,0)+0η(b)r(t,b)db+φ(0)i(t,0)+0(φ(a)φ(a)(μ+δ(a)))i(t,a)da+ψ(0)r(t,0)+0(ψ(b)ψ(b)(μ+η(b)))r(t,b)db=dS02SS0S0S+αS03VV0SV0S0VS0S+(01)i(t,0)0. Notice that dL(t)/dt=0 implies that S=S0 and V=V0. It can be verified that the largest invariant set where dL(t)/dt=0 is the singleton {E0}. Therefore, by the invariance principle, E0 is globally attractive.

6. Global attractivity of the endemic equilibrium

As at the beginning of Section 5, we can get the characteristic equation at E, which is (24) 0=λ+μ+α+βHi(0)0βSHˆ(λ)0αλ+μ+β1Hi(0)β1VHˆ(λ)0βHi(0)β1Hi(0)1(βS+β1V)Hˆ(λ)Lˆ(λ)00Kˆ(λ)1,(24) where Hˆ(λ)=0eλap(a)Ω(a)da, Kˆ(λ)=0eλaδ(a)Ω(a)da, and Lˆ(λ)=0eλbη(b)Γ(b)db. Though we believe that when 0>1, all roots of Equation (24) have negative real parts and hence E is locally asymptotically stable. Unfortunately, it is difficult to confirm it. In the following, we just show that E is globally attractive. To achieve this, we need the following properties of solutions to Equation (Equation1).

Lemma 6.1

Suppose that 0>1. Then, for any solution (S(t),V(t),i(t,),r(t,)) of Equation (Equation1) with the initial value in Y0, the following equalities hold: (25) 0δ(a)i(a)1i(t,a)r(0)i(a)r(t,0)da=0,(25) (26) βS0p(a)i(a)1i(t,a)i(0)Si(a)i(t,0)Sda+β1V0p(a)i(a)1i(t,a)i(0)Vi(a)i(t,0)Vda+0η(b)r(b)1i(0)r(t,b)i(t,a)r(b)db=0.(26)

Proof.

The proofs of Equations (25) and (26) are quite similar and we only give that for Equation (25) as an illustration. In fact, 0δ(a)i(a)1i(t,a)r(0)i(a)r(t,0)da=0δ(a)i(a)dar(0)r(t,0)0δ(a)i(t,a)da. This, combined with Equations (Equation2) and (Equation8), immediately gives Equation (25).

Theorem 6.1

The unique endemic equilibrium E=(S,V,i,r) of Equation (Equation1) defined by Equation (Equation11) is globally attractive when 0>1.

Proof.

Define G[x,y]=xyylnxy. It is easy to see that G is non-negative on (0,)×(0,) with the minimum value 0 only when x=y. Furthermore, it is easy to verify that xGx[x,y]+yGy[x,y]=G[x,y].

Consider a candidate Lyapunov functional, H(t)=H1(t)+H2(t)+H3(t), where H1(t)=G[S(t),S]+G[V(t),V], H2(t)=0φ2(a)G[i(t,a),i(a)]da, H3(t)=0ψ(b)G[r(t,b),r(b)]db. Similar to the one as in the Proof of Theorem 5.2, ψ(b) and φ2(a) are given as ψ(b)=bη(u)ebu(μ+η(ω))dωdu and φ2(a)=a[(βS+β1V)p(u)+ψ(0)δ(u)]eau(μ+δ(ω))dωdu. It is easy to see that φ2(0)=1, ψ(0)=0η(b)Γ(b)db=L, and (27) ψ(b)ψ(b)(μ+η(b))=η(b),(27) (28) φ2(a)φ2(a)(μ+δ(a))=[(βS+β1V)p(a)+ψ(0)δ(a)].(28) In what follows, we shall calculate the derivative of H(t) along solutions of Equation (Equation1).

Firstly, differentiating H1 along solutions of Equation (Equation1) yields dH1(t)dt=1SSΛβS0p(a)i(t,a)da(μ+α)S+1VVαSβ1V0p(a)i(t,a)daμV=1SS(μ+α)S+βS0p(a)i(a)daβS0p(a)i(t,a)da(μ+α)S+1VVαSβ1V0p(a)i(t,a)daμV=μS2SSSS+μV3VVSVSVSS+βS01Si(t,a)Si(a)SS+i(t,a)i(a)p(a)i(a)da+β1V02Vi(t,a)Vi(a)SSSVSV+i(t,a)i(a)p(a)i(a)da. Secondly, using Equation (Equation5), we have H2(t)=0tφ2(a)G[i(ta,0)Ω(a),i(a)]da+tφ2(a)G[i0(at)eata(μ+δ(ω))dω,i(a)]da=0tφ2(tr)G[i(r,0)Ω(tr),i(tr)]dr+0φ2(t+r)G[i0(r)ert+r(μ+δ(ω))dω,i(t+r)]dr. Differentiating H2(t) and using i(a)=i(0)e0a(μ+δ(ω))dω produce dH2(t)dt=φ2(0)G[i(t,0),i(0)]+0tφ2(tr)G[i(r,0)e0tr(μ+δ(ω))dω,i(tr)]dr0tφ2(tr)(μ+δ(tr))[i(r,0)e0tr(μ+δ(ω))dω×Gx[i(r,0)e0tr(μ+δ(ω))dω,i(tr)]+i(tr)Gy[i(r,0)e0tr(μ+δ(ω))dω,i(tr)]]dr+0φ2(t+r)G[i0(r)ert+r(μ+δ(ω))dω,i(t+r)]dr0φ2(t+r)(μ+δ(t+r))[i0(r)ert+r(μ+δ(ω))dω×Gx[i0(r)ert+r(μ+δ(ω))dω,i(t+r)]+i(t+r)Gy[i0(r)ert+r(μ+δ(ω))dω,i(t+r)]]dr=G[i(t,0),i(0)]+0[φ2(a)φ2(a)(μ+δ(a))]G[i(t,a),i(a)]da=(βS+β1V)0p(a)i(a)i(t,a)+i(a)lni(t,a)i(a)da+L0δ(a)i(a)i(t,a)+i(a)lni(t,a)i(a)da+i(t,0)i(0)i(0)lni(t,0)i(0)(using~Equation (28)). The second last equality follows from (Equation5) and the fact that xGx[x,y]+yGy[x,y]=G[x,y]. Similarly, using Equations (Equation6) and (27), we have dH3(t)dt=ψ(0)G[t(t,0),r(0)]+0[ψ(b)ψ(b)(μ+η(b))]G[r(t,b),r(b)]db=0η(b)r(b)r(t,b)+r(b)lnr(t,b)r(b)db+Lr(t,0)r(0)r(0)lnr(t,0)r(0). It follows that dHdt=μS2SSSS+μV3VVSVSVSS+βS02Si(t,a)Si(a)SS+lni(t,a)i(a)p(a)i(a)da+β1V03Vi(t,a)Vi(a)SSSVSV+lni(t,a)i(a)p(a)i(a)da+L0δ(a)i(a)i(t,a)+i(a)lni(t,a)i(a)da+i(t,0)i(0)i(0)lni(t,0)i(0)+0η(b)r(b)r(t,b)+r(b)lnr(t,b)r(b)db+Lr(t,0)r(0)r(0)lnr(t,0)r(0). Using Equation (Equation2) and the fifth and sixth equations of (Equation8), we have (29) dH(t)dt=μS2SSSS+μV3VVSVSVSS+βS02Si(t,a)Si(a)SS+lni(t,a)i(a)p(a)i(a)da+β1V03Vi(t,a)Vi(a)SSSVSV+lni(t,a)i(a)p(a)i(a)da+βS0p(a)i(t,a)da+β1V0p(a)i(t,a)da+0η(b)r(t,b)dbβS0p(a)i(a)da+β1V0p(a)i(a)da+0η(b)r(b)db×1+lni(t,0)i(0)+L0δ(a)i(a)i(t,a)+i(a)lni(t,a)i(a)da+L0δ(a)i(t,a)i(a)i(a)lnr(t,0)r(0)da+0η(b)r(b)r(t,b)+r(b)lnr(t,b)r(b)db.(29) Collecting the terms of Equation (29) yields (30) dH(t)dt=μS2SSSS+μV3VVSVSVSS+βS01SS+lni(t,a)i(a)lni(t,0)i(0)p(a)i(a)da+β1V02SSSVSV+lni(t,a)i(a)lni(t,0)i(0)p(a)i(a)da+L0δ(a)i(a)lni(t,a)i(a)lnr(t,0)r(0)da+0η(b)r(b)lnr(t,b)r(b)lni(t,0)i(0)db.(30) It follows from Lemma 6.1 that putting Equations (25) and (26) into Equation (27) gives dH(t)dt=μS2SSSS+μV3VVSVSVSS+βS01SS+lnSS+1i(t,a)i(0)Si(a)i(t,0)S+lni(t,a)i(0)Si(a)i(t,0)Sp(a)i(a)da+β1V01SS+lnSS+1SVSV+lnSVSV+1i(t,a)i(0)Vi(a)i(t,0)V+lni(t,a)i(0)Vi(a)i(t,0)Vp(a)i(a)da+L0δ(a)i(a)1i(t,a)r(0)i(a)r(t,0)+lni(t,a)r(0)i(a)r(t,0)da+0η(b)r(b)1r(t,b)i(0)r(b)i(t,0)+lnr(t,b)i(0)r(b)i(t,0)dbβS01i(t,a)i(0)Si(a)i(t,0)Sp(a)i(a)da+β1V01i(t,a)i(0)Vi(a)i(t,0)Vp(a)i(a)da+0η(b)r(b)1r(t,b)i(0)r(b)i(t,0)db+L0δ(a)i(a)1i(t,a)r(0)i(a)r(t,0)da.

Note that 1xlnx0 for all x>0 with the equality holds only when x=1. It follows that dH(t)dt=μS2SSSS+μV3VVSVSVSSβS0g(SS)p(a)i(a)daβ1V0g(SS)+g(SVSV)p(a)i(a)daβS0gi(t,a)i(0)Si(a)i(t,0)Sp(a)i(a)daβ1V0gi(t,a)i(0)Vi(a)i(t,0)Vp(a)i(a)da0η(b)r(b)gr(t,b)i(0)r(b)i(t,0)dbL0δ(a)i(a)gi(t,a)r(0)i(a)r(t,0)da0 and dH(t)/dt=0 implies that S=S, V=V and i(t,a)i(a)=i(t,0)i(0)=r(t,b)r(b)=r(t,0)r(0)for all a,b0. It is not difficult to check that the largest invariant subset where dH(t)/dt=0 is the singleton {E}. By the invariance principle, E is globally attractive.

7. Discussion

In this paper, we first integrate solutions along the characteristic line to obtain an equivalent integral equation, which is developed by Webb [Citation32] and Walker [Citation25] for age-dependent models. Secondly, the asymptotic smoothness of the semiflow generated by the system is proved by the method in [Citation21]. Thirdly, in order to make use of the invariance principle, we establish the uniform persistence and the existence of a compact global attractor of the system similarly as in [Citation16, Citation30, Citation31]. Finally, the global stability of the equilibria is derived by constructing suitable Volterra-type Lyapunov functionals. Theorems 5.2 and 6.1 imply that the basic reproduction number 0 completely governs the global dynamics of system (Equation1).

In the following, we discuss the biological implications of the continuous vaccination strategy. According to  Liu et al. [Citation10] and Wang et al. [Citation26, Citation27], we also assume that β1β as the vaccinees may have some partial immunity and hence the effective contacts with infected individuals becomes smaller.

Case for vaccinees without infection. Let β1=0 and denote 1:=0|β1=0=βΛμ+αH+LK. It is easy to see that 10. In this case, 1α<0andlimα1=LK<1. According to Theorems 5.2 and 6.1, the disease always can be eradicated by some suitable vaccination strategy.

Case for vaccinees with infection. Note that 0α=ΛμH(ββ1)μ(μ+α)2<0, which implies that vaccination always has a positive effect on disease control by increasing the vaccination rate. It is obvious to see that 00|α=0:=0¯. Suppose that 0¯>1. Denote limα0 by 2, the average number of secondary cases produced by one infected individual introduced to a population with vaccinees only during the lifespan. If 2<1, then there is a unique α such that 0=1 for α=α and 0<1 for α>α, which implies that the disease can be eliminated by some suitable vaccination strategies with α>α by Theorem 5.2. If 21, then 0>21, which implies that disease cannot be eradicated by any vaccination strategies (any values of α) according to Theorem 6.1. It follows that 2<1 is a key condition for disease elimination when vaccinees can be infected (with small β1). Thus, neglecting the possibility for vaccinees to be infected will over-estimate the effect of vaccination strategy.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

J. Wang was supported by the National Natural Science Foundation of China [Nos. 11401182, 11471089], Science and Technology Innovation Team in Higher Education Institutions of Heilongjiang Province [No. 2014TD005]. Y. Chen was supported partially by the Natural Sciences and Engineering Research Council (NSERC) of Canada.

References

  • F. Brauer, Z. Shuai, and P. van den Driessche, Dynamics of an age-of-infection cholera model, Math. Biosci. Eng. 10 (2013), pp. 1335–1349. doi: 10.3934/mbe.2013.10.1335
  • C.J. Browne, A multi-strain virus model with infected cell age structure: Application to HIV, Nonlinear Anal.: RWA 22 (2015), pp. 354–372. doi: 10.1016/j.nonrwa.2014.10.004
  • Y. Chen, J. Yang, and F. Zhang, The global stability of an SIRS model with infection age, Math. Biosci. Eng. 11 (2014), pp. 449–469. doi: 10.3934/mbe.2014.11.449
  • O. Diekmann, J. A. P. Heesterbeek, and J. A. J. Metz, On the definition and the computation of the basic reproduction ratio R0 in models for infectious diseases in heterogeneous populations, J. Math. Biol. 28 (1990), pp. 365–382. doi: 10.1007/BF00178324
  • X. Duan, S. Yuan, and X. Li, Global stability of an SVIR model with age of vaccination, Appl. Math. Comput. 226 (2014), pp. 528–540. doi: 10.1016/j.amc.2013.10.073
  • A. Gabbuti, L. Romano, P. Blanc, F. Meacci, A. Amendola, A. Mele, F. Mazzotta, and A.R. Zanetti, Long-term immunogenicity of hepatitis B vaccination in a cohort of Italian healthy adolescents, Vaccine 25 (2007), pp. 3129–3132. doi: 10.1016/j.vaccine.2007.01.045
  • J.K. Hale, Asymptotic Behavior of Dissipative Systems, Mathematical Surveys and Monographs, Vol. 25, American Mathematical Society, Providence, RI, 1988.
  • G. Huang, X. Liu, and Y. Takeuchi, Lyapunov functions and global stability for age-structured HIV infection model, SIAM J. Appl. Math. 72 (2012), pp. 25–38. doi: 10.1137/110826588
  • M. Iannelli, Mathematical theory of age-structured population dynamics, Applied Mathematics Monographs, Vol. 7, Consiglio Nazionale delle Ricerche (C.N.R.), Giardini Pisa, 1995. comitato Nazionale per le Scienze Matematiche.
  • X. Liu, Y. Takeuchi, and S. Iwami, SVIR epidemic models with vaccination strategies, J. Theoret. Biol. 253 (2008), pp. 1–11. doi: 10.1016/j.jtbi.2007.10.014
  • L. Liu, J. Wang, and X. Liu, Global stability of an SEIR epidemic model with age-dependent latency and relapse, Nonlinear Anal.: RWA 24 (2015), pp. 18–35. doi: 10.1016/j.nonrwa.2015.01.001
  • P. Magal, Compact attractors for time periodic age-structured population models, Electron. J. Differ. Equ. 65 (2001), pp. 1–35.
  • P. Magal and C.C. McCluskey, Two-group infection age model including an application to nosocomial infection, SIAM J. Appl. Math. 73 (2013), pp. 1058–1095. doi: 10.1137/120882056
  • P. Magal, C.C. McCluskey, and G.F. Webb, Lyapunov functional and global asymptotic stability for an infection-age model, Appl. Anal. 89 (2010), pp. 1109–1140. doi: 10.1080/00036810903208122
  • C.C. McCluskey, Global stability for an SEIR epidemiological model with varying infectivity and infinite delay, Math. Biosci. Eng. 6 (2009), pp. 603–610. doi: 10.3934/mbe.2009.6.603
  • C.C. McCluskey, Global stability for an SEI epidemiological model with continuous age-structure in the exposed and infectious classes, Math. Biosci. Eng. 9 (2012), pp. 819–841. doi: 10.3934/mbe.2012.9.819
  • P.W. Nelson, M.A. Gilchrist, D. Coombs, J.M. Hyman, and A.S. Perelson, An age-structured model of HIV infection that allows for variations in the production rate of viral particles and the death rate of productively infected cells, Math. Biosci. Eng. 1 (2004), pp. 267–288. doi: 10.3934/mbe.2004.1.267
  • L. Rong, Z. Feng, and A.S. Perelson, Mathematical analysis of age-structured HIV-1 dynamics with combination antiretroviral therapy, SIAM J. Appl. Math. 67 (2007), pp. 731–756. doi: 10.1137/060663945
  • G. Röst and J. Wu, SEIR epidemiological model with varying infectivity and infinite delay, Math. Biosci. Eng. 5 (2008), pp. 389–402. doi: 10.3934/mbe.2008.5.389
  • H.R. Thieme, Mathematics in Population Biology, Princeton University Press, Princeton, 2003.
  • H.L. Smith and H.R. Thieme, Dynamical Systems and Population Persistence, American Mathematical Society, Providence, 2011.
  • P. van den Driessche and J. Watmough, Reproduction numbers and sub-threshold endemic equilibria for compartmental models of disease transmission, Math. Biosci. 180 (2002), pp. 29–48. doi: 10.1016/S0025-5564(02)00108-6
  • P. van den Driessche and X. Zou, Modeling relapse in infectious diseases, Math. Biosci. 207(1) (2007), pp. 89–103. doi: 10.1016/j.mbs.2006.09.017
  • P. van den Driessche, L. Wang, and X. Zou, Modeling diseases with latency and relapse, Math. Biosci. Eng. 4(2) (2007), pp. 205–219. doi: 10.3934/mbe.2007.4.205
  • J.A. Walker, Dynamical Systems and Evolution Equations, Plenum Press, New York, 1980.
  • J. Wang, G. Huang, Y. Takeuchi, and S. Liu, SVEIR epidemiological model with varying infectivity and distributed delays, Math. Biosci. Eng. 8 (2011), pp. 875–888. doi: 10.3934/mbe.2011.8.733
  • J. Wang, R. Zhang, and T. Kuniya, The dynamics of an SVIR epidemiological model with infection age, IMA J. Appl. Math. (2015), doi:doi:10.1093/imamat/hxv039.
  • J. Wang, R. Zhang, and T. Kuniya, Global dynamics for a class of age-infection HIV models with nonlinear infection rate, J. Math. Anal. Appl. 432 (2015), pp. 289–313. doi: 10.1016/j.jmaa.2015.06.040
  • J. Wang, R. Zhang, and T. Kuniya, Mathematical analysis for an age-structured HIV infection model with saturation infection rate, Electron. J. Differential Equ. (33) (2015), pp. 1–19.
  • J. Wang, R. Zhang, and T. Kuniya, The stability analysis of an SVEIR model with continuous age-structure in the exposed and infectious classes, J. Biol. Dyn. 9 (2015), pp. 73–101. doi: 10.1080/17513758.2015.1006696
  • J. Wang, R. Zhang, and T. Kuniya, A note on dynamics of an age-of-infection cholera model, Math. Biosci. Eng. 13(1) (2016), pp. 227–247. doi: 10.3934/mbe.2016.13.227
  • G.F. Webb, Theory of Nonlinear Age-Dependent Population Dynamics, Marcel Dekker, New York, 1985.
  • J. Yang, Z. Qiu, and X .-Z. Li, Global stability of an age-structured cholera model, Math. Biosci. Eng. 11 (2014), pp. 641–665. doi: 10.3934/mbe.2014.11.641