_NA'-"TU-'---'-'R-"E'--V-'---O'-'-L.-'3_1.c...8.c.2.c...8c...N--'.O_V.=EM'-'-"-BECCR'-'----'-l'---'98'-'--5------REVIEW ARTICLE------3~23 Vaccination and herd immunity to infectious diseases Roy M. Anderson & Robert M. May· Department of Pure and Applied Biology, Imperial College, London University, London SW7 2BB, UK * Biology Department, Princeton University, Princeton, New Jersey 08540, USA
An understanding of the relationship between the transmission dynamics of infectious agents and herd immunity provides a template for the design of effective control programmes based on mass immuniz ation. Mathematical models of the spread and persistence of infection provide important insights into the problem of how best to protect the community against disease.
MUCH attention is now focused on the development and com of acquired immunity and maternally derived protection, age munity-wide use of vaccines for the control of infectious dis related changes in the degree and intimacy of contacts among eases. This interest is a consequence of many factors; these people, and the prevailing levels of genetic, spatial and include the worldwide eradication of smallpox by vaccination behavioural heterogeneity in susceptibility/resistance to infec in the late 1970s1, the current success of child immunization tion. Intuition built on many years of practical epidemiological programmes for the control of measles, mumps, pertussis and experience often fails to predict the outcome of a particular rubella in the United States2·3, the failure of frequent appeals vaccination programme. Mathematical models which clearly by public health authorities in the United Kingdom to raise and accurately define the details of the association between vaccination rates that are at present low by developed-country disease agent and host (at the individual and population levels) 4 standards ·', the Expanded Programme on Immunization (EPI) can help to overcome such difficulties. This article reviews recent of the World Health Organisation (WHO) for the control of a theoretical and empirical work on the transmission dynamics variety of serious childhood viral and bacterial diseases in the of infectious disease. Emphasis is placed on the relevance of 8 12 developing world • , and the rapid advances in molecular bio such research to the design of disease control programmes based logy and biotechnology which promise new vaccines for the on vaccination. 13 16 future - • The main thrust of research today is at the molecular level, where it is stimulated by the urgent need to develop General theory vaccines for the major killing diseases of the Third World such Models for the transmission dynamics of directly transmitted 11 1 as malaria • 9, and to combat new infections such as the current viral and bacterial infections are typically compartmental in 23 35 epidemic of AIDS (acquired immune deficiency syndrome), structure · • They represent changes, with respect to age, a, caused by the recently isolated human T-lymphotropic virus and time, I, in the populations of infants protected by maternally 20 21 (HTL V-III) • • derived antibodies (M(a, t)), susceptible individuals (X(a, 1)), The development of a safe, effective and cheap vaccine, infected individuals who are not yet infectious (H(a, t)), infec however, is only a first step (albeit an essential one) towards tious people ( Y(a, t)), and people who have recovered and are community-wide control. Epidemiological, economic and moti immune from subsequent infection (Z,(a, t)). The total popula vational issues are at least as important as technological ones. tion is N(a, 1) = M(a, t) + X(a, t) + H(a, t) + Y(a, t) + Z(a, t). 12 A vaccine for measles (a very cost-effective immunization ) has The deterministic equations describing the flows from one class 24 25 27 been available since the late 1960s, yet the infection remains to another are of the general form • • 8 22 one of the world's major causes of child mortality • • Concomi aM(a, 1) aM(a, t) tant with research on vaccine development, there is a need for ---+--- -[µ(a)+d]M(a, t) an improved understanding ofhow best to use vaccines to protect at aa the community as well as the individual. The persistence of aX(a, t) aX(a, t) infectious disease within a population requires the density of ---+--- dM(a, t) susceptible individuals to exceed a critical value such that, on at aa average, each primary case of infection generates at least one -[v(a, t)+A(a, t)+ µ(a)]X(a, t) secondary case. It is therefore not necessary to vaccinate every aH(a, 1) aH(a, t) one within a community to eliminate infection; the level of herd ---+--- A(a, t) X(a, t) immunity must simply be sufficient to reduce the susceptible at aa fraction below the critical point. The central epidemiological -[µ(a)+a]H(a, 1) (1) questions are thus: what proportion of the population should be vaccinated to achieve elimination (in a local programme), a Y(a, t) a Y(a, t) ---+--- aH(a, t)-[µ(a)+y]Y(a, 1) eradication (in a global programme) or a defined level of con a1 aa trol? How is this affected by demographic factors (for example, az(a, 1) aZ(a, 1) birth rate)? What is the best age at which to immunize? How ---+---= yY(a, t)+ v(a, t) X(a, t) does mass immunization affect the age distribution of susceptible at aa individuals, particularly in those classes most at risk from serious -µ(a) Z(a, t) disease, and how important is genetic and spatial heterogeneity in suceptibility to infection (or response to immunization) to aN(a, t) aN(a, t) 23 26 ---+--- -µ(a) N(a, 1) the creation of effective herd immunity • ? a, aa To answer these questions, we need to understand the interac tion between the transmission dynamics of the disease agent Here µ(a) denotes age-related human mortality, 1/d is the and the level of naturally acquired (or artificially created) average duration of protection provided by maternal antibodies, immunity to infection. This relationship is complex and depends 1/ a and 1/ y are the average incubation and infectious periods on such factors as the precise course of infection within an respectively, and v( a, t) records the age- and time-specific individual, the demography of the host population, the duration vaccination schedule. Mortality explicitly associated with the
© 1985 Nature Publishing Group 3_24______REVIEWARTICLE------~N~A.c..cTUc...c..c.RE=-V-O'-'L'--.-'-3_18_2C-'8-'-N'--O'--V_E=MB~E=R'--1=9~85
Fig. 1 a, Time series analysis of the case so,ooo a resports for measles in England and Wales over the pre-vaccination period 1948-68. The top ~ 4 O, 0 0 0 e graph denotes the recorded cases and the bottom ; c 1.4 ~------, st 8 b graph the serial corrclogram • Note the sig- ~ 30,000 ., 1.2 nificant yearly (seasonal) and 2-yearly peaks in a. 0 incidence (the horizontal axis in the bottom "' 20,000 .; graph denotes weeks). b, Model predictions of 5 e the equilibrium (the state to which the system u 10,000 .! 0.8 .; settles in the long term) ratio of the number of 0.6 cases of mumps in the age range 14-50 yr (the ., 1953 1958 1963 1968 "' age range in which males are at risk to serious = 0.4 Year u disease resulting from infection) under the C 1.0 0.2 impact of a vaccination programme in which p .!2 o iii of each yearly cohort arc vaccinated at age 2 yr, 0.5 ~ 0.2 0.4 0.6 0.8 divided by the number of cases in the same age 0 0.0 range before vaccination. If the ratio is above u Proportion vaccinated unity, the control programme has a deterimental ~ -0.5 <( effect and vice versa. The average age at infection -1.0 Time lag infection has been ignored. The term A(a, t) denotes the 'force' unity. Within communities in which the infection is endemic, 23 37 or rate of infection, which under the 'mass action' assumption ~ is inversely correlated with A, the average age at infection - 24 30 adopts the form - A(a, t) = J~f3(a, a') Y(a', t) da'. Here (~= B/(A- D)). This relation holds both for developed coun {3(a, a') represents the transmission coefficient for infected tries (growth rates essentially zero) and for developing countries individuals in age class a' who come into contact with suscep (with their growing populations)52 . Here B is the reciprocal of 1 tible individuals in age class a (refs 25, 30). The boundary the finite birth rate (usually expressed as births (yr- ) per 1,000 conditions of equations ( 1) depend on the demography of the of population), and D is the duration of maternally derived community. The structure of the model may be modified to immunity. The value of ~ within a community is critically mimic the transmission of a wide variety of infections31 - 34, dependent on social and behavioural factors (the contact com including those with complex life cycles (for example, malaria ponent of /3 which influences A and hence A), on the biology 5 37 or yellow fever3 - ) and those involving carriers (such as tuber of the disease agent and host (the infectiousness and susceptibil 38 culosis and hepatitis B ) or macroparasites (such as helminths) ity components of /3) and on the demography of the community where note must be taken of the distribution of parasite numbers (value of B)25 . The marked difference between developed and 9 40 within the human community3 • • developing countries in the average ages at which children Insights into applied problems of disease control or into typically acquire infections such as measles in urban com general ecological and epidemiological questions can be munities is a consequence of variations in social, behavioural obtained through analytical or numerical studies of and demographic factors (Table lb). equations (1 ). To eradicate an infection by mass immunization it is necessary 3 25 53 to reduce the value of~ below unity2 - • . This can be achieved Cycles in incidence by vaccinating a proportion, p, of each cohort of the population For most common viral and bacterial infections of childhood, at age V (where V> D; vaccination during the period of mater the model predicts large-amplitude, slowly damping oscillations nally derived protection usually fails to protect the child 23 30 54 55 23 25 in disease incidence - • Several factors can tend to perpetuate adequately from subsequent infection • ) provided • p > 41 44 these oscillations indefinitely; these are stochastic effects - , [ 1 - A/ B]/[ 1 - V / B]. The infection can never be eradicated if 25 age-dependent transmission rates , incubation periods of a V> A (that is, p > 1). The value of pis minimized by vaccinating 45 46 27 47 50 discrete nature • and seasonality in contact • - • The average at as young an age as possible, given the constraint of V > D. 23 period of oscillations, T, is approximately given as : T= These conclusions are derived under the assumption that the 21r(AK)112, where A is the average age at infection and K is rate of transmission within the community is fair1 25 3 the generation time of the infection (K = 1/ u+ 1/ -y). homogeneous. In practice, this assumption is often violated • 6, The observed periods of such cycles (which agree well with but simple theory provides a useful guide, usually setting an prediction) can be determined accurately by time-series tech upper limit to the levels of vaccination required to eliminate 51 25 niques (Table la, Fig. la ); the longer-term cycles are usually many common childhood infections . It forms the basis of compounded with short-term seasonal effects. estimates suggesting that 92-96% of children must be vaccinated to eliminate measles and pertussis, 84-88% to eliminate rubella Vaccination and herd immunity and 88-92% to eliminate mumps in Western Europe and the The basic reproductive rate or transmission potential of an United States; these estimates assume that vaccination takes 23 25 - • infection, R0 , is the average number of secondary cases produced place just after the wane of maternal antibody protection 33 34 by one primary case in a wholly susceptible population • • Recent experience in the United States suggests that high levels Clearly, an infection cannot be maintained unless R0 exceeds of cohort immunization are required (as indicated by theory) © 1985 Nature Publishing Group ~-'-'A'-'-TU-"-R'-"E~VO.c..Lcc.c·...c.3.c..18'-=28'--N'--O=--V....cE=--Mc.c:Bc=E.c.cR....clc...98:..;;.5 ______REVIEW AITTICLE------"3=25 Table I Infections reponed throughout the world ,, " Infectious disease Inter-epidemic Geographical location Infectious disease Average Geographical location period (yr) and time period age A and time period (yr) Measles 2 England & Wales, 1948-68 2 Aberdeen, Scotland, 1883-02 Measles 5-6 USA, 1955-58 2 Baltimore, USA, 1900-27 4-5 England & Wales, 1948-68 2 Paris, France, 1880- 10 2-3 Morocco, 1962 I Yaounde, Cameroun, 1968-7 5 2-3 Ghana, 1960-68 1 llesha, Nigeria, 1958-61 2-3 Pondicherry, India, 1978 Rubella 3-5 Manchester, UK, 1916-83 1-2 Senegal, 1964 3-5 Glasgow, Scotland, 1929-64 1-2 Bankok, Thailand, 1967 Parvovirus (HPV) 3-5 England & Wales, 1960-80 Rubella 9-10 Sweden, 1965 Mumps 3 England & Wales, 1948-82 9-10 USA, 1966-68 2-4 Baltimore, USA, 1928-73 9-10 Manchester, UK, 1970-82 Poliomyelitis 3-5 England & Wales, 1948-65 6-7 Poland, 1970-77 Echo virus ( type II) 5 England & Wales, 1965-82 2-3 Gambia, 1976 Smallpox 5 India, 1868-48 Chickenpox 6-8 USA, 1912-28 Chickenpox 2-4 New York City, USA, 1928-72 Poliovirus 12-17 USA, 1955 2-4 Glasgow, Scotland, 1929-72 Penussis 4-5 England & Wales, 1948-68 Coxsackie virus 2-3 England & Wales, 1967-82 4-5 USA, 1920-60 (type 82) Mumps 6-7 England & Wales, 1975-77 Scarlet fever 3-6 England & Wales, 1897-78 6-7 Netherlands, 1977-79 Diphtheria 4~ England & Wales, 1897-79 Penussis 3-4 England & Wales, 1948-85 C Mycoplasma pneumoniae 4 England & Wales, 1970-82 Force of Age groups (yr) Proponion to infection 0-5 5-10 10-15 15-20 20-75 be immunized 1 A (yr- ) for eradication ( p) Case 1 0 .2 0.2 0 .2 0.2 0.2 0.94 Case 2 0.18 0.56 0.2 0.1 0.1 0.89 99 a. The inter-~idemic period, T, for various viral and bacterial infections"·"·"· ·' ". b, Average age at infection (yr), A, for different diseases in different 24 34 116 localities"· • • ' . c, The imr,act of age-related changes in the force of infection, A (yr- •), on the critical proportion of the population p, to be vaccinated to eradicate 2 measles in England and Wales '· • (case l, constant; case 2 age dependent). to substantially reduce the incidences of common childhood bility is lost by vaccination or by natural infection. Reduction infections. of transmission via vaccination, however, usually increases the The vaccination of cohorts must be continued for many inter-epidemic cycle period and the average age at infec decades before the full impact of a programme is manifest; tion23-26·51. This latter manifestation can be worrying if the risk long-term suppression or interruption of transmission requires of serious disease resulting from infection increases with age continual cohort vaccination as long as the infection remains ( Fig. 1c ). Whether this issue is important depends on the quanti endemic in other communities or countries. The only infection tative details of such factors as how risk changes with age, the to be eradicated worldwide is smallpox. This success (announ average age at which the vaccine is administered, and how the ced by WHO in 1977) generated some optimism concerning the force of infection changes with age before widespread immuniz feasibility of eradicating other major infections, such as measles ation25·26 ( Fig. 3 a). Rubella and mumps are of particular concern 89 56 58 and pertussis · • - . Such optimism, however, may be mis because of the risk of congenital rubella syndrome (CRS) in placed, given the low communicability of smallpox (low R0 and infants born to mothers who contracted rubella in their first 64 65 high average age at infection, A, before widespread immuniz trimester of pregnancy • and the occurrence of orchitis and ation) in comparison with infections such as measles, the ease the associated risk of sterility in post-pubertal males following with which smallpox infection can be diagnosed, the practical mumps infection66 (Fig. lb, c). The problems surrounding mass simplicity of inoculation procedures and the stability of the vaccination against rubella (to control CRS) have resulted in 59 60 smallpox vaccine under primitive storage conditionst· · • the adoption of different vaccination programmes in different Successful immunization programmes may themselves create countries. In the United Kingdom, girls only are vaccinated at health problems, since vaccination almost invariably carries an average age of around 12 yr, so as to allow rubella virus to some risk (usually extremely small) to the individual. At the circulate and create naturally acquired immunity in the early start of a mass-immunization programme, the probability of years. By contrast, in the United States both boys and girls are serious disease arising from vaccination is usually orders of vaccinated at around 2 yr of age, with the aim of eliminating magnitude smaller than the risk of serious disease arising from rubella. Theory suggests that the US policy is best if very high natural infection. As the point of eradication is approached, the levels of vaccination (80-85% of each yearly cohort) can be relative magnitude of these two probabilities must inevitably be achieved at a young age, while the UK policy is better if this 24 67 reversed. The optimum strategy for the individual (not to be cannot be guaranteed • • A mixed US and UK policy is predic vaccinated) therefore becomes at odds with the needs of society ted to be of additional benefit over the UK single-stage policy (to maintain herd immunity). This issue-which was central to if moderate to high levels of vaccine uptake among boys and the decline in uptake of whooping cough vaccine in the United girls can be achieved at a young age (>60% )67 (Fig. 1d). Kingdom during the mid- to late-1970s-can be overcome by legislation to enforce vaccination (as in the United States), but Epidemiological data its final resolution is only achieved by global eradication of the The average age at infection, A, is central to most estimates of 6 disease agent (so that routine vaccination can cease) t. the intrinsic transmission potential of an infection, to the calcula Mass immunization at levels below that required for elimina tion of the desired level of vaccination coverage for elimination, tion acts to reduce the number of cases of infection (and hence and to theoretical assessment of how the risk of serious disease 23 26 the force of infection,,\). However, both theory - and observa caused by infection may change with age under the impact of tion62'63 suggest that it has little, if any, impact on the total mass vaccination. For infections which are thought to induce number of susceptible individuals. If the infection is endemic, life-long immunity against disease (for example, measles, its effective reproductive rate is unity-each case produces, on mumps, rubella and pertussis), the value of A can be estimated average, one other case-and the fraction susceptible remains (see Table 1 b) either from age-stratified case notification roughly constant ( -1/ R0 ), independently of whether suscepti- records or, more accurately, from serological profiles (Fig. 2). © 1985 Nature Publishing Group J~ 26~------REVIEW ARTICLE------.:..:.N:..cATU-=---=R=E:..._V.:..cO=--=L=---..::_3:..:18_:2=8-.:..Nc.-=O:..:V-=E::.:M=B=ER:.:_:_19=85 100 ---- its impact on the level of vaccine-induced herd immunity 0 sojh 25 67 75 so required to eliminate an infection • • • The high values of A 60: observed in the 5-15-yr-old classes (Fig. 3a) are thought to arise 60 I as a consequence of frequent and intimate contacts within school 40 40 environments. 20 Analyses of the significance of age-related changes in contact 2: with infection are crude at present, due to the absence of detailed 25 O O 1 1.5 2 2.5 3 3.5 4 4.5 ::, information on 'who acquires infection from whom' • Models ,5' ... ,5' which incorporate high degrees of contact within school-age 100 C 100:d classes (and low contact in the very young and in adult age 80 so; groups), however, yield predictions of temporal changes in 60 disease incidence and of age-related changes in serology in vaccinated versus unvaccinated communities which closely 40 parallel observed trends25 (Figs 3c, d, 4a-c). Recent numerical 20 studies suggest that predictions based on the assumption of a, > homogenous mixing tend to overestimate the critical level of ~ 0 0.5 7.5 17.5 27 5 45 65 vaccination coverage for elimination by roughly 5% for infec 0. ~ 100~------~ tions such as measles and pertussis in the United Kingdom and 25 30 v, , e 9oT United States (Table lc) • • For these infections, analyses ;fl a.01 ~ based on apparent age-related changes in contact with infection 60 (Fig. 3a) yield estimates of approximately 90% coverage at 25 • 40 1-2 yr of age These predictions, however, must be accepted 30, 0 ~ with caution at present, because the observed low rates of 20 • %~ ~ infection in late teenage and adult age groups may be artefacts 10· ~~ ;J0, o' ,%,,,-, ~ arising from biases in case reporting, from the insensitivity of 1 2 3 4 5 6 7 8 9 1012 ~"" 0 2 6 6 2 ,5-"6' > s "'s s s 'o, o serodiagnostic tests, or from genetic variability in susceptibility ', 1s '2s 78 to infection . These caveats are underlined by the fact that no 10 o11 ------~ 1h such marked age-related changes in the per capita rate of infec 807 tion are found in epidemics within virgin populations on islands 79 8 601 (largely susceptible to infection before the epidemic • t) (Fig. 3b). At present it seems prudent to accept the higher vaccine coverage levels estimated by homogeneous mixing 20· 25 models • 0 Genetic factors. Genetic factors are important in determining 0.096 0.2884 0 481 0.673 0 788 8 host susceptibility and resistance to infectious agents 7 • Geneti cally based variability in the level and duration of antibody Fig. 2 Age-stratified (horizontal) serological profiles of the percentages of production following infection would seriously complicate the 82 people in different age groups with antibodies to various infectious disease interpretation of age-serological profiles • Associations agents in developed and developing countries. a, Measles in the United between HLA type and antibody titres following vaccination States in 1955-58 (New Haven, Connecticut)'l7_ b, Measles in lndia118 (left-hand histograms) and Nigeria119 (right-hand histograms). Note the against rubella and measles provide evidence for the existence 83 84 rapidity with which the proportion seropositive rises with age when com of such complications • • Little is understood, at present, about pared with a. c, Rubella in England and Wales24 in the 1960s and 1970s. d, 120 the importance of such factors in the design of vaccination Rubella in the Gambia in the late 197Os . Again note the rapidity of the programmes. In the future, however, community-based diagnos rise in seropositivity with age when compared with c. e, Mumps in England and Wales (left-hand histograms) and the Netherlands (right-hand his tic programmes aimed at the identification of immunodeficiency 34 tograms)121·122. /, Polio in the United States in 1955 • g, Hepatitis B (HBV) could make it possible to target vaccination, for certain less in Senegal 123. The vertical axis denotes the percentage with HBV markers. common but serious viral and bacterial diseases, to those child h, Malaria (Plasmodiumfalciparam) in Nigeria"'. The vertical axis denotes 85 ren genetically predisposed to infection and severe morbidity • the cumulative percentage of children who have experienced infection. Note that in a-h the scales of the x-axes are not always identical and in certain Spatial factors. Inhomogeneity in the spatial distribution of cases record unequal age class divisions (a nonlinear scale). hosts, with some people living in dense aggregate and others living in isolated or small groups, can lead to heterogeneity in transmission rates. This, in turn, can result in the transmission The interpretation of the data, however, is beset with many potential of an infection ( R0) being greater on average than 68 69 problems, such as age-specific biases in notification records • , suggested by estimation procedures which assume spatial 6 87 and a reduced ability to detect antibodies in adults who experien homogeneity8 • . Under these circumstances, the optimal sol 70 ced infection in early childhood • ution appears to involve 'targeting' vaccination coverage in relation to group size, with dense groups receiving the highest 4 Heterogeneity in transmission rates of vaccination7 • The optimal programme is defined as that Conventional epidemiological theory is founded on the 'mass minimizing the total, community-wide number of immunizations action' principle of transmission (see equations (1)), in which needed for elimination or a defined level of control. This strategy infection is assumed to spread via contacts within a reduces the overall proportion that must be vaccinated to achieve 71 homogeneously mixing community • Recent research has begun elimination, compared with that estimated on the assumption to take account of the complications induced by heterogeneity of spatial homogeneity. This conclusion has practical sig in transmission arising from age-related, genetic, spatial or nificance for the control of infections such as measles and 25 30 72 77 behavioural factors • • - • pertussis in some developing countries, where rural-urban Age-dependent factors. Analyses of case notification records and differences in population density tend to be much more marked serological profiles suggest that, for many common infections than in developed countries. It is probable that in many regions (measles, rubella and pertussis), the per capita rate of infection of Africa and Asia, diseases such as measles cannot persist ( A ( a, t)) depends on the ages of susceptible individuals, chang endemically in rural areas without frequent movement of people ing from a low level in the 0-5-yr age classes, via a high level between rural and urban regions (Ro< 1 in rural areas). Under in the 5-15-yr age classes, back to a low level in the adult age these circumstances, disease control might be achieved in both classes (Fig. 3a)25. This is of interest both because it reflects types of region by high levels of mass immunization in the urban behavioural attributes of human communities and because of centres alone. © 1985 Nature Publishing Group NATURE VOL. 318 28 NOVEMBER 1985 REVIEW ARTICLE 327 0.6 100 C 'i 0.5 >- 80 Fig. 3 a, Age-specific change in the rate, or force, of measles infection ~A) in England and Wales before C: 0.4 2 -~ 1·;;; 60 mass vaccination • b, Percentages seropositive for ti ., 0 rubella antibodies in various age classes on St Paul's 0.3 Cl. 0 Island, Alaska, before (left-hand histograms) and after ~ ;;; 40 (right-hand historgrams) an epidemic of rubella in 0 0.2 (/) 79 81 1963 - • The population (-400 people) had not ., .... 0 20 experienced rubella infection (before 1963) for 22 yr. 0 0.1 Note that in the age range 0-19 yr, the rate of infection u. of susceptible individuals was similar in each age class. 0 0 c, Comparison of model predictions ( equations (I)) 0-4 5-9 10-14 15-19 0 4 8 12 16 20 24 of the age-serological profile for measles antibodies in England and Wales with observed patterns. The 100 observed data (squares) were collected in the period d 1979-82. Model predictions were based on an average 80 0.0 age of infection of 4.5 yr before control and the age- ~ 1 "'0 specific vaccination rates in the United Kingdom over Cl. "'0 60 0 0.6 the period 1968-82". d, Comparison of model predic- Cl. 0 tions of the age-serological profile for rubella anti- ;;; ~ bodies in females in England and Wales with the (/) 40 C: 0.4 67 observed pattern (crosses) in 1984 • Predictions were .... -~ based on an average age of infection of 10 yr before 0 20 Cl. 0.2 control and the age-specific vaccination rates in the 0 United Kingdom (vaccination of teenage girls between a: the ages of 10 and 15 yr) over the period 1970-84. 0 0 o,<1 s,9 'o,, 's 0 5 10 15 20 25 30 35 40 45 Behavioural factors. These are of great importance as deter waned ( 6 months) and before the age of 2 yr to achieve elimina minants of disease transmission and provide a source of sig tion. Calculations suggest an R0 value of around 17, which is nificant heterogeneity. The sexually transmitted infections pro similar to that for measles in England and Wales (with a birth vide a good example because the probability distribution of rate of 12 per 1,000 and an average age of infection around sexual promiscuity ( defined as the number of different sexual 4.5-5 yr). Simple theory suggests that something like 96% of partners per unit of time) is highly skewed. It seems very likely each yearly cohort of children would have to be effectively that the highly sexually active individuals (R0 » 1) in the tail of immunized by age 1 yr ( V = l) to attain elimination. Coverage the distribution are responsible for maintaining transmission could be lower in rural areas but, as suggested in the section 88 90 within the community as a whole - . Drug treatment and on spatial heterogeneity, an optimal control programme would surveillance focused on these 'superspreaders' is therefore highly have to focus on the urban centres. These estimates question beneficial, given that R0 may be below unity for the remaining the feasibility of measles elimination in many developing coun population. If vaccines are developed for HTLV-111, selective tries at this time and argue that, as a first priority, worldwide 91 93 immunization of those most at risk · (haemophiliacs, programmes should concentrate on controlling morbidity and 52 99 intravenous drug abusers, persons of Central African or Haitian mortality among those most at risk • • Moderate to low levels origin and homosexual males) and those predisposed to serious of vaccination coverage, however, are beneficial; they raise the clinical infection, may be a practical option for community-wide average age at infection and thereby lengthen the age 'window' 99 control. in which vaccine can be administered , and reduce morbidity Behavioural factors are also important in helping to cause and mortality (young infants appear to be more at risk than seasonal fluctuations in the incidence of many common viral older children). 49 75 94 and bacterial infections • • • The timings of school terms and vacation periods has a marked influence on temporal patterns New vaccines of measles and pertussis cases in the United Kingdom and West The coming decade will see the development and use of a number Germany, for example. of new vaccines, along with the expansion of community-based programmes of mass immunization using the currently available Developing countries vaccines. The development of safe, effective and cheap vaccines Many infections that in developed countries are primarily a is clearly the first priority, but, once these objectives are achieved, source of morbidity are major cause of mortality in the develop the problem of community-wide control requires careful atten ing world (mainly as a consequence of malnutrition, although tion. In particular, age-specific serological profiles of the com 9 22 95 genetic components may also be involved) • • . Measles, for munity need to be obtained both before and during any pro example, is estimated to cause more than one million child gramme of mass vaccination. 9 22 deaths annually • • Although there is much current interest in An example is provided by the recent proposal to initiate a worldwide immunization programmes for the control of such mass-vaccination campaign for the control of hepatitis B virus infections, the problems of creating sufficient levels of herd (HBV) in Taiwan14 by a serum-derived vaccine (a genetically 100 101 immunity for elimination should not be underestimated, given engineered vaccine is also available • ; Fig. 2). The epi the logistic difficulty of vaccine storage and delivery in poor demiology of HBV is made complicated by chronic infection in countries. Even if such practical issues are resolved by effective immunodeficient individuals who become carriers of the virus cold-chains, the development of heat-stable vaccines, and aero ( with a high associated risk of inflammatory disease and malig sol delivery techniques ( which are more effective in the immuniz nancy) 102, and who constitute an important reservoir of infec 96 98 ation of infants with maternal antibodies) - , the feasibility of tion. The desirability of mass immunization is clear when viewed blocking transmission by mass vaccination in areas where the as a method for reducing the incidence of carriers (the risk of average age of infection is often as low as 1-2 yr of age is becoming a carrier appears to be much greater in children than 103 questionable (Table 1b ). in adults ), but less clear given the observation that serious Consider a large urban centre in India or Africa, where the symptoms of disease are more likley to occur in adults than in annual birth rate is around 40 per 1,000 population and the children. Serological surveys in high-risk regions (South-East average age of measles infection is 2 yr (Fig. 2). Vaccination Asia and Africa) indicate that HBV typically has a low R0 within must take place after the maternally derived protection has the population as a whole (when compared with measles). © 1985 Nature Publishing Group ....;328~------REVIEW ARTICLE------'-N...:.A:..:TIJ-=R=E_V....;O:..::L::....-=--31:..:8....;2:..:8....;N....;Oc_V....;Ec.c_M.:;:B:..::E::.cR....;1:..:.9=85 Moderate levels of vaccination coverage should therefore have a 23 102 104 a significant impact on disease incidence • • (Fig. ld). The calculation of Ro is made complicated by the presence of carriers (R0 = RoJ + Ro2 (1 - f), where f is the fraction of carriers and ( 1 -f) the fraction of non-carriers) and a component of perinatal 42 ( =vertical) transmission • The maintenance of transmission ( R0 > 1) is probably due to the carriers alone ( R01 » 1, Ro2 < 1) given that they are often a significant fraction of the community (with f values ranging from 0.1 % to 2% in the United States and most European countries, to as high as 50% in certain 0 C) • Pacific islands (determined in part by the ethnic composition c 102 ::, of a community )). E Vaccines for viral infections such as cytomegalovirus, .§ rhinovirus, adenovirus, rotavirus, respiratory syncytial virus and .gC parainfluenza virus may provide sufficient protection to the 0 individual a. 0 to be considered for community-wide use in the 0 • - future, provided their public health significance 0:: 0 0 is of sufficient b importance relative to other priorities in health care. The existence of many distinct antigenic types within certain groups of these viruses (for example, adenovirus and rhinovirus), however, present problems for vaccine development. Recent work has raised the hope that vaccines may be developed for the major parasitic infections such as malaria 5 (Plasmodium falciparum) and certain helminth species 1° • The problems, however, are formidable because of the antigenic complexity of protozoa and helminths (in comparison with ~- 106 ., viruses and bacteria) , sequential changes in surface antigens C ::, during development within the host (for example, the filarial E 107 worms) , antigenic variation (whether due to the parasite's .§ ability to express many variable antigen types, as in the case of C 108 109 the African trypansomes • , or as a consequence of popula 1 0 tion genetic ( =strain) variability, as in the case of the malaria a. 110 0 parasites ), and the failure of man naturally to develop fully 0:: protective immunity under conditions of repeated exposure to C 111 infection • Vaccines must be more effective in stimulating the host's immunological defences than the parasite itself if they are to overcome the evasion mechanisms that certain parasitic 112 species have evolved • Combined with these problems is the suspicion that heterogeneity in immunological responsiveness is an important determinant of the observed aggregation in the distribution of parasite numbers per person (such that 15% of the hosts will often harbour 90% or more of the ., 0,., 39 40 113 114 C parasites) • • • • Predisposition to heavy or repeated para ::, sitic infection is often a consequence of behavioural or social E factors but laboratory studies increasingly point to the import .§ 112 ance of genetic and nutritional components • An effective vaccine must therefore be able to protect both immunocom so petent and immunodeficient individuals, the latter being those most likely to suffer serious disease and to maintain parasite transmission. The ideal malaria vaccine will probably contain antigens from Fl&, 4 a, Model predictions ( equations ( 1)) of the impact of mass vaccina different malaria species, strains and developmental stages tion on the age-serological profiles for measles antibodies (the proportion (sporozoite, asexual and gamete) so as to reduce mortality and in each age class who are immune ei1her as a result of maternally derived 18 19 protection (in the infants), recovery from natural infection, or vaccination) morbidity, and to block transmission • . But once the technical 25 in England and Wales • Mass vaccination was introduced in year I (year problems in vaccine development have been overcome, the 0 denotes the pre-vaccination profile, based on the age-specific forces of issues of community control must take priority. The creation of infection defined in Fig. 3a) and maintained for 16 yr with the age-specific a level of herd immunity high enough to eliminate malaria vaccination rates recorded in England and Wales (mean age at vaccination appears impractical in endemic regions where the average age -2.2 yr; -50-60% of each yearly cohort was immunized by age 5 yr). The age axis denotes the intervals 0, 0.5, 1.0, 2.5, . ... etc. yr. Note how mass at first infection is typically around 3-6 months of age (Ro= vaccination (in the age range 1-3 yr) acts to decrease the proportion seroposi 50-100, which implies 99% coverage of the community before tive in the older age classes (by reducing the overall force of infection within the age of 3 months with a vaccine which gives lifelong protec the community below the level in the pre-vaccination community). The tion) (Fig. le). Low to moderate levels of immunization, valley of seronegativity running diagonally across the surface denotes the ageing of the last cohorts of young children to be excluded from mass however, will greatly help to control mortality and morbidity, immunization (see ref. 25 for further details). b, c, Similar to a but denoting which is of greatest significance in the 0-5-yr-old age classes. model predictions of the impact of vaccination on the age-serological profiles Care must, however, be exercised to pr~vent an associated for rubella antibodies over the period 1984-2030 in the United Kingdom in build-up of susceptible individuals in teenage and adult age the male (c) and female (b) segments of the population. The basis of the predictions is described in ref. 67. In the period 1970-84 a single-stage UK classes; at present, such individuals have acquired a degree of policy was in operation (vaccination of teenage girls between the ages of immunity to infection from repeated exposure in childhood. 10-15 yr, with the observed vaccination rates for England and Wales). In Once the properties of a potential vaccine are more clearly 1985 a two-stage (UK-US) policy was begun (vaccination of teenage girls defined, mathematical studies can help to quantify such risks. at 1984 rates, vaccination of60% of2-yr-old boys and girls) to be maintained up to 2030. Note how the introduction of the two-stage policy alters the In the case of helminth infections, vaccines will probably only level of herd immunity in both che male and female segments of the provide protection of ~hort duration against re-infection, and population. The age axis records che intervals 0, 0.5, 1.5, 2.5, . . . 49.5 yr. © 1985 Nature Publishing Group '-NA'-'-TU-'--"--'R-"'E'--VO=L=.-'3'-'-1~8 =28~N=O_V=EM~BE=R~l9-'8~5------REVIEW ARTICLE------32_9 hence the creation of herd immunity may require repeated cohort wide control of many of the world's major infections is likely immunization. Cost factors and other practical issues are there to depend on integrated approaches, involving some combina fore likely to be of great importance in developing countries tion of selective or targeted vaccination and chemotherapy when considering the benefits of mass vaccination versus mass ( directed towards those most predisposed to infection and chemotherapy. morbidity), vector control and efforts to improve nutrition, hygiene, sanitation and education. Insights founded on an Conclusions understanding of the population dynamics of infection will, Many difficulties surround the attainment of sufficient levels of therefore, be of great importance in any quantitative assessment herd immunity to eradicate common infections in developed of control policy. The recent convergence of mathematical and developing countries. Theory can define the level of vaccina theory and observation in epidemiology has created a powerful tion coverage required for elimination, but success in practice set of tools for the design and evaluation of community-based depends on economic and motivational issues_. Vaccines against programmes of disease control, provided they are used sensibly. protozoan and helminth parasites will undoubtedly be of import At present the potential value of these techniques is not widely ance in the future (certainly to the Western traveller to tropical appreciated. regions, but less certainly to those who live in these areas), and We thank the Rockefeller Foundation (R.M.A. and R.M.M.), research towards their development must remain a priority. This the Department of Health and Social Security, UK (R.M.A.), emphasis, however, should not distract attention from the and the NSF (grant BSR83-03772 to R.M.M.) for financial importance of epidemiological research. Effective community- support. 1. Final Rtport of tht Global Commission for tht Certification of Smallpox Eradication (WHO, 65. Hansow, J. 8. & Dudgeon, J. A. Viral Dinans of tht Foetus and Newborn (Saunders, Geneva. 1980). London, 1978). 2. Kau, S. L. hdiatric, 71, 653-654 (1983). 66. Brunell, P. A. in Pediatric lnfectiou, Di,ra.,, (eds Feigin, R. D. & Cherry, J. D.) 123-1235 3. Centres for Disease Control. Morbidity Mortality Wukly R,p. 32, 1-125 (1983). (Saunder.;, London, 1981 ). 4. Noah, N. D. Br. m,d. J. 289, 1476 (1984). 67. Anderson, R. M. & Grenfell, 8 . T. J. Hyg., Camb. (in the press). 5. Broolt, C. G.D. Br. m,d. J. 286, 1082-1083 (1983) . 68. Sydenstriclter, E. & Hedriclt, A. W. US pubL Hlth Rep. 44, 1537-1543 (1928). 6. Editorial Lane © 1985 Nature Publishing Group