On the definition and the computation of the basic reproduction ratio R0 in models for infectious diseases in heterogeneous populations
Summary (3 min read)
1. Introduction
- Suppose the authors want to know whether or not a contagious disease can " invade" into a popula t ion which is in a steady (at the time scale o f disease transmission) demographic state with all individuals susceptible.
- To decide about this question the authors first o f all linearize, i.e. they ignore the fact that the density o f susceptibles decreases due to the infection process.
- The authors have followed the advice of I. Nfisell to use 'ratio' rather than 'number' in order to emphasize that R o does not even have a quasi-dimension (R o ~ cases/case!) individual during its entire period of infectiousness.
It is the aim of this note to demonstrate how these ideas extend to less simple
- (though probably still highly oversimplified) models involving heterogeneity in the population and to explain the meaning of "typical" in the "definition" of Ro above.
- Subsequently the authors shall deal with the actual computation of Ro in certain special cases, in particular the so-called "separable" or "weighted homogeneous mixing" case.
2. The definition
- Let the individuals be characterized by a variable 4, which the authors shall call the h-state variable (h for heterogeneity).
- The authors may call this quantity the next generation factor of ~. Remark.
- One could argue, as MacDonald did (see Bailey 1982, p. 100 and the references given there), that one should consider the average number of cases in the host population arising from one case in the host population via vector cases.
3.1. Separable mixing rate
- To compute the dominant eigenvalue of a positive operator is, in general, not an easy task.
- There is one special case in which the task is trivial: when the operator has one-dimensional range.
- Biologically this corresponds to the situation in which the distribution (over the h-state space f2) of the "offspring" (i.e. the ones who become infected) is independent of the state of the "parent" (i.e. the one who transmits the infection).
- Or weighted homogeneous mixing.the authors.
- This interpretation yields once more that Ro = fa b(tl)S(tl)a(tl) dtl. (3.6) (3) The special case in which a and b differ only by a multiplicative constant, is usually referred to as proportionate mixing (Barbour 1978).
3.2. Separable mixing rate with enhanced infection within each group
- A second case in which it is easy to derive an explicit threshold criterion, even if the authors cannot calculate R0 explicitly, occurs when individuals preferentially mix with their own kind and otherwise practise weighted homogeneous mixing.
- If the authors moreover assume that the h-state stays constant over epidemiological time (but see Example 4.3) then K(S) is of the form where c(4)S(~) is the number of first generation "offspring" produced "directly" in one's own group.
- The left hand side defines a decreasing function of 0 which tends to zero for.
Q ~ oo. The largest real root Ro is larger than one if and only if either
- When (i) holds a single just infected individual with h-state ~ will already start a full blown epidemic among its likes.
- If, on the other hand, c(~)S(¢)< 1 for all c f2 any epidemic has to be kept going by the additional cross infections among different types.
- The expected total number of cases, including its own, produced by an individual of h-state ~ through chains of infectives which stay wholly among its likes is ( 1 - S(Oc(O) - 1. Each of these produces an expected number of cross infections equal to b(O.
- One of us had derived the result (3.10) in the context of the geographical spread of plant diseases (think of foci within fields).
- Recently their attention for this special case was revived by Andreasen and Christiansen (1989) (in which they derive the same result in the context of a finite h-state space) and Blythe and Castillo-Chavez (1989).
3.3. Multigroup separable mixing
- An obvious mathematical generalization of a separable mixing rate is to assume that K(S) has a finite dimensional range.
- In general, however, this does not make biological sense.
- Therefore the authors restrict their elaboration to a special example in this category which does allow a biological interpretation.
4.1. Discrete and static h-state
- In this case the operator K(S) is represented by a matrix.
- The authors shall first show how this matrix can be derived in the special case of the conventional S-E-I-R compartment models.
- The fraction of the exposed individuals which enters I (before dying) is the diagonal of S(2; + M) -~.
- In the separable case the entries of T(S) are of the form aiSibj, and according to (3.6) R0 equals the trace of the matrix K(S).
- See Hethcote and Yorke (1984) for another derivation of this fact.
4.2. Sexually transmitted diseases
- Let the index 1 refer to females and the index 2 to males.
- Then the proportions of 0 (-'= free of d) and + ('.= having d) individuals that will be infected by D in the linear initial phase of an epidemic are described by the vector So ~, vS+} where So and S+ are the steady (with respect to d) state population sizes of 0 and + individuals.
- Without going into too much detail of how one could treat dynamic h-states in general, the authors explain some of the background of their calculations.
- Assume that the infectivity towards an individual in h-state ~ then depends on 0, but not on the history of h-state transitions by which it reached 0 nor on the time elapsed since infection z (i.e., the influence of h-state on the, otherwise constant, infectivity is only through present h-state).
- If this operator has a one-dimensional range the authors can just as in the case of Eq. (2.1) find an explicit expression for Ro.
4.3. Age dependence
- The authors now turn their attention to a continuous dynamic h-state variable.
- Recalling Remark 6 at the end of Sect. 2 the authors shall now consider an endemic steady state.
- Let 2(a) = age specific force of infection, (4.14) i.e. the age specific probability per unit of time of becoming infected.
- One can now use data about the endemic state to estimate f , Q and ~ and subsequently calculate whether or not a given ~v suffices to eradicate the disease.
- This condition allows an interpretation similar to the one of (3.11).
Did you find this useful? Give us your feedback
Citations
7,106 citations
Cites background or methods from "On the definition and the computati..."
...The basic reproduction number, denoted R0, is ‘the expected number of secondary cases produced, in a completely susceptible population, by a typical infective individual’ [2]; see also [5, p. 17]....
[...]
...Following Diekmann et al. [2], we call FV 1 the next generation matrix for the model and set R0 ¼ qðFV 1Þ; ð4Þ where qðAÞ denotes the spectral radius of a matrix A....
[...]
...This criterion, together with the definition of R0, is illustrated by treatment, multigroup, staged progression, multistrain and vector–host models and can be applied to more complex models....
[...]
5,915 citations
3,324 citations
1,485 citations
Cites background from "On the definition and the computati..."
...This matrix, usually denoted by K, is called the next-generation matrix (NGM); it was introduced in Diekmann et al. (1990) who proposed to define R0 as the dominant eigenvalue of K....
[...]
...In a natural way this growth factor is then the mathematical characterization of R0 (Diekmann et al. 1990)....
[...]
1,327 citations
References
2,024 citations
683 citations
"On the definition and the computati..." refers background in this paper
...(4.5) (See Hethcote and Yorke (1984) for a "discrete" version of this result.)...
[...]
...See Hethcote and Yorke (1984) for another derivation of this fact....
[...]
...Combinations like (3.7) can also be found in Nold (1980), Hethcote and Yorke (1984), Hyman and Stanley (1988) (where it is called biased mixing), and Jacquez et al. (1988) (where it is called preferred mixing)....
[...]
409 citations
"On the definition and the computati..." refers background in this paper
...9 of the paper by Jacquez et al. (1988), a special case of this matrix is introduced with T(S) written out in some more detail.)...
[...]
...Combinations like (3.7) can also be found in Nold (1980), Hethcote and Yorke (1984), Hyman and Stanley (1988) (where it is called biased mixing), and Jacquez et al. (1988) (where it is called preferred mixing)....
[...]
283 citations
260 citations
"On the definition and the computati..." refers background in this paper
...Combinations like (3.7) can also be found in Nold (1980), Hethcote and Yorke (1984), Hyman and Stanley (1988) (where it is called biased mixing), and Jacquez et al. (1988) (where it is called preferred mixing)....
[...]
Related Papers (5)
Frequently Asked Questions (6)
Q2. What is the eigenvalue problem for K(S)?
Note that linearization at the trivial solution 2 - 0 andthe transformation 4, ~ ~ 2 lead us back to the eigenvalue problem for K(S), as to be expected.
Q3. What is the definition of a 'irreducibility hypothesis'?
(3) In order to guarantee that any introduction of infectivity in the population leads to an epidemic when R0 > 1 the authors need to make an irreducibility hypothesis (Schaefer 1974).
Q4. What is the way to prove that a disease leads to a higher birth rate?
If one makes the obvious assumption that the disease leads to a lower (or equal) birth rate and to a higher (or equal) death rate one can use the linearized problem to obtain upper estimates for the nonlinear problem.
Q5. What is the expected infectivity towards x at time z?
The expected infectivity towards ~'s of x at time z after infection is thenA(z, ~, ,1) = [ a(~, 0)P(~, 0, q) dO,where P(z, O, r/) is the conditional probability that the h-state of x at time z is 0 given that x is still alive at z and that its h-state at time 0 was t/. For K(S) the authors find( K ( S ) d P ) ( ~ ) = S ( ~ ) f a f o ~ [ f a(~,O)P('c,O, tl) dO3dp(tl)drdtl.
Q6. How do the authors calculate the R0 for a disease?
The authors want to calculate R0 for a disease D, assuming that the susceptibility to D is, for individuals having d, v times as large as for individuals without d.