Professional Documents
Culture Documents
1 s2.0 S0264410X14016077 Main
1 s2.0 S0264410X14016077 Main
Vaccine
journal homepage: www.elsevier.com/locate/vaccine
a r t i c l e i n f o a b s t r a c t
Article history: Background: The Measles & Rubella Initiative, a broad consortium of global health agencies, has provided
Received 31 August 2014 support to measles-burdened countries, focusing on sustaining high coverage of routine immunization of
Received in revised form children and supplementing it with a second dose opportunity for measles vaccine through supplemental
17 November 2014
immunization activities (SIAs). We estimate optimal scheduling of SIAs in countries with the highest
Accepted 27 November 2014
measles burden.
Available online 23 December 2014
Methods: We develop an age-stratified dynamic compartmental model of measles transmission. We
explore the frequency of SIAs in order to achieve measles control in selected countries and two Indian
Keywords:
Measles
states with high measles burden. Specifically, we compute the maximum allowable time period between
Immunization two consecutive SIAs to achieve measles control.
Supplementary immunization activities Results: Our analysis indicates that a single SIA will not control measles transmission in any of the
Low- and middle-income settings countries with high measles burden. However, regular SIAs at high coverage levels are a viable strategy
Child health to prevent measles outbreaks. The periodicity of SIAs differs between countries and even within a single
Mathematical modeling country, and is determined by population demographics and existing routine immunization coverage.
Conclusions: Our analysis can guide country policymakers deciding on the optimal scheduling of SIA
campaigns and the best combination of routine and SIA vaccination to control measles.
© 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND
license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
http://dx.doi.org/10.1016/j.vaccine.2014.11.050
0264-410X/© 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
1292 S. Verguet et al. / Vaccine 33 (2015) 1291–1296
population projections. The average infectious period of measles following the implementation of SIAs at regular time intervals from
was assumed to be 14 days [21]. MCV1 was assumed to be deliv- every single year to every 10 years (dotted vertical lines), with
ered before the first birthday, with the SIA measles dose being given MCV1 coverage of 74% and with the population structure of India.
to all children between six months and five years old, assuming for In this case, we see that there is resurgence of measles (occurrence
simplicity that vaccination could be given as maternally derived of incidence peaks after year 50) when the inter-SIA period is over
immunity had waned. In practice, this may be an approximation three years. Furthermore, the less frequent the SIA, the larger the
since the lower age limit is commonly 9 or 12 months, which is resurgence. Conversely, an average inter-SIA period of three years
a few months after these antibodies may have waned.1 The prob- or less suffices to control measles (no occurrence of incidence peaks
ability of receiving a MCV1 and SIA dose was assumed to be after year 50).
uncorrelated. MCV1 coverage by country was obtained from WHO Table 1 also lists the inter-SIA periods required in order to
and UNICEF [12]. Coverage figures for the SIAs were assumed to achieve measles control, estimated computationally, for selected
be of 90% among the 6–59 month-olds, a conservative estimate countries with high measles burden and 90% SIA coverage among
according to historical SIA coverage data collated by WHO [23]. 6–59 month-olds. We see that the maximum time period required
between two consecutive SIAs depends importantly on MCV1
2.3. Model analysis coverage (Pearson correlation coefficient of 0.71). For example,
countries like Ethiopia and Nigeria with lower coverage of MCV1
The dynamic compartmental model developed and presented (66% and 42%, respectively) would require SIAs about every 2 years.
above was run for 100 years with routine vaccination to reach a This national picture may however mask heterogeneities at
post-vaccination equilibrium (the last 50 years of which are shown the sub-national/state level and Table 1 also lists the periodicity
on Fig. 2), and finally run for a further 50 years with both rou- required between SIAs in order to achieve measles control with
tine and SIA vaccination. Using country-specific inputs (e.g. MCV1 current level of MCV1 coverage for two Indian states, Bihar and
coverage), we explored the potential periodicity of SIAs required Uttar Pradesh, where SIAs have been undertaken since 2010 and
in order to achieve measles control. Disease control refers to the 2011, respectively. For these two states, we see that the inter-SIA
reduction of morbidity and mortality to locally acceptable levels period required is 2 years compared with 3 years at the national
[24]. Here we define measles as being controlled if the incidence level.
of measles remains below 1 per 100,000 individuals in the entire This computational analysis is further confirmed by the analyt-
population. ical results, also presented in Table 1. Based on Eq. (1) above, Fig. 3
shows the inter-SIA period required to achieve measles control for
2.4. Analytical approach a range of SIA coverage levels, as a function of MCV1 coverage
levels, using Indian demographic inputs and R0 of 16. In the last
The computer model presented above requires considerable column of Table 1, the results are shown for the inter-SIA period
technical capacity and computing power. To enable policymakers required in order to achieve measles control, estimated analytically
to use our model for SIA planning with greater ease, we also derived through Eq. (1) and rounded down,2 for 90% SIA coverage among
a simple formula relating SIA periodicity to immunization cover- 6–59 month-olds. We see that these analytical results match the
age and population demographics. To do this, we simplified the simulation results closely.
system of differential equations representing the model of measles Finally, in the simulation model, the use of mixing data from
transmission (Supplementary Data, Section 1), removing age strat- Vietnam [20] would lengthen the measles inter-epidemic cycles
ification in order to make the equations analytically tractable [25]. considerably and increase the height of the epidemic peaks, leading
This allowed us to compute the maximum time period allowed to unrealistic inter-epidemic cycles much larger than the 1–2 year
between two consecutive SIAs that would still achieve measles cycles usually expected in low- and middle-income countries with
control, depending on MCV1 coverage, demographic parameters, higher birth rates and lower MCV1 coverage [21,26,27]. Although
measles specific parameters, and vaccine effectiveness. The max- this is not clear, reasons for this difference may include less assorta-
imum inter-SIA period allowed to achieve measles control could tive mixing (due to differences in household structure and/or birth
then be estimated as (with the formula developed in Supplemen- rates) or recall bias.
tary Data, Section 2): In addition, we varied the basic reproduction number for sensi-
S S,eff tivity analysis. Table 2 reports on the inter-SIA period required in
Tmax ∼ (1) order to achieve measles control, estimated computationally and
R
˜ 0 (1 − R R )(1 − S S,eff /2)
analytically, for the selected countries with high measles burden,
where R0 is the basic reproduction number, ṽ is the birth rate with 90% SIA coverage among 6–59 month-olds, and with a lower
adjusted for population growth (birth rate multiplied by the ratio reproduction number R0 of 12, in order to highlight a better case
between under-five population and 5–9-year-olds population), R scenario. Expectedly, we see that the inter-SIA period to achieve
and S are the effectiveness of MCV1 and SIA measles vaccine measles control is increased on average for all countries.
respectively, R is the coverage of MCV1, and S,eff is the effec-
tive coverage of SIA immunization (90% coverage multiplied by the
fraction of measles cases among under-fives depending on MCV1 4. Discussion
coverage and as estimated by geographical region [6]). The com-
plete details of this approach are given in the Supplementary Data Both numerical simulation and mathematical analysis indicate
(Section 2). that a single SIA will not control measles outbreaks in any of the
countries with high burden of measles. However, regular SIAs at
3. Results high coverage are able to control measles transmission, with the
periodicity of SIA campaigns determined by population demo-
Fig. 2 shows results from the computer simulation showing graphics and existing MCV1 coverage.
change in the number of infected individuals (measles incidence)
2
Rounding down (e.g. 2.3∼2; 2.7∼2) the inter-SIA period estimated analytically
1
This choice does not materially affect the measles dynamics studied in the paper. leads to a more conservative estimate.
1294 S. Verguet et al. / Vaccine 33 (2015) 1291–1296
Fig. 2. Computer model predictions of annual measles incidence pre- and post-SIA in India, under different intervals between SIAs. Each SIA is indicated by a dotted vertical
line. R0 is 16 and SIAs are assumed to have 90% coverage among 6–59 month-olds. Note: The model was run for 100 years with routine vaccination to reach a postvaccination
equilibrium (the last 50 years of which are shown), and finally run for a further 50 years with both routine and SIA vaccination.
The model also suggests that a single SIA may give the impres- immunization campaigns [25,36–38]. Here, we build on these mod-
sion of having controlled measles due to a post-vaccination els to explore the impact and optimal timing of periodic measles
‘honeymoon’ [28] (Fig. 2). Such potentially unexpected effects can immunization campaigns, which have been the main focus of
have disastrous consequences on plans for national elimination measles control efforts in the highest burden countries. We find
or even global eradication goals. This underscores the importance that impact and optimal timing of measles SIAs varied among
of using predictive models that account for transmission events, the highest burden countries, and even within a single country
rather than relying solely on surveillance based on reported cases, (Tables 1 and 2). This suggests that a country-specific approach
to inform decision making about vaccination. to SIA timing may be most effective, with the recommended
Several models of measles transmission have been previously length of time between SIAs dependent on local demographics and
developed in order to explore the potential impact of different routine coverage. Our computer model and the associated simpli-
vaccination strategies [27,29–35], including the impact of mass fied analytical formula offer a way for country policymakers to
S. Verguet et al. / Vaccine 33 (2015) 1291–1296 1295
Table 1
Inter-SIA period required to achieve measles control with 90% SIA coverage and R0 of 16, as a function of routine immunization coverage for the first dose of measles vaccine
and crude birth rate, for selected countries with high measles burden, and two Indian states.
Country Routine coverage for 1st Crude birth rate (per SIA period (years) SIA period (years)
dose of measles vaccine (%) 1000 population) (computational model) (analytical model)
Afghanistan 68 36 2 2
Burkina Faso 87 43 3 3
Ethiopia 66 35 2 2
India 74 23 3 3
State of Bihar 58 29 2 3
State of Uttar Pradesh 53 29 2 2
Indonesia 80 21 3 4
Madagascar 69 36 2 2
Mali 59 49 2 2
Niger 73 51 2 2
Nigeria 42 43 2 2
Table 2 within the range of the 2–4 years periods recommended by WHO
Supplemental immunization activity (SIA) period required to achieve measles con-
for countries with MCV1 coverage under 80% [8]. But, it also goes
trol with 90% SIA coverage and R0 of 12, for selected countries with high measles
burden, and two Indian states. beyond these recommendations as it provides a means to weigh
various combinations of coverage levels for MCV1 and SIAs with
Country SIA period (years) SIA period (years)
the dimension of SIA periodicity. This enables countries to identify
(computational model) (analytical model)
more locally appropriate SIA scheduling in accordance with finan-
Afghanistan 3 3 cial and operational capacity: this is especially important in the
Burkina Faso 5 5
context of resource-limited settings given that SIAs can be expen-
Ethiopia 3 3
India 3 4 sive.
State of Bihar 3 4 Nonetheless, our model has several limitations. First, we used an
State of Uttar Pradesh 2 3 average estimate of the basic reproduction number of measles, and
Indonesia 4 6
hence did not capture local variations in the intensity of measles
Madagascar 3 3
Mali 3 3 transmission (due to differences in cultural practices and social
Niger 3 3 gathering, for example). However, estimating this number sepa-
Nigeria 2 3 rately for each setting is hindered by the lack of reliable data for
measles case notifications due to underreporting and great dis-
parity in reporting quality. As sensitivity analysis, we reported
determine when measles SIAs should be implemented. They can
on the inter-SIA period required for the selected high burden
also further inform WHO guidelines on the conduct of SIAs at
countries with a lower reproduction number R0 of 12 (Table 2).
the country level, by recommending specific inter-SIA periods to
Also, in the absence of data, we did not account for potential
achieve measles control at the national and sub-national levels,
variations in vaccine effectiveness at the country/region level. Sec-
depending on a range of country-specific routine immunization
ond, due to lack of data, the model does not take into account
coverage rates and demography, and including within country
potential correlation between MCV1 and SIA coverage, and the
heterogeneities. Our analysis highlights country inter-SIA periods
fact that often children with better access to health systems and
hence more likely to have received MCV1 can be easier to reach
through SIA. Neither does it account for changes in immunization
10
70%
90% measles elimination (i.e. zero indigenous transmission events).
Elimination goals depend on monitoring measles importations,
vaccine coverage in hard-to-reach groups and stochastic one-off
SIA period (years)
complex and require considerable financial investments, optimi- [8] World Health Organization. Measles vaccines: WHO position paper. Wkly Epi-
zing country planning of measles activities is critical. This model demiol Rec 2009;35(84):349–60.
[9] Pan American Health Organization. Measles elimination: field guide (Scien-
provides a valuable and practical tool for high measles-burdened tific and Technical Publication no. 605). Washington, DC: Pan American Health
countries toward efficient resource allocation in order to achieve Organization; 2005.
measles control. [10] Wolfson LJ, Strebel PM, Gacic-Dobo M, Hoekstra EJ, McFarland JW, Hersh
BS, Measles Initiative. Has the 2005 measles mortality reduction goal been
achieved? A natural history modelling study. Lancet 2007;369:191–200.
Funding [11] Morris SK, Awasthi S, Kumar R, Shet A, Khera A, Nakhaee F, et al.
Measles mortality in high and low burden districts of India: estimates
from a nationally representative study of over 12,000 child deaths. Vaccine
This work was supported by the WHO’s Initiative for Vaccine 2013;31(41):4655–61.
Research. The authors alone are responsible for the views expressed [12] World Health Organization UNICEF., Available at: http://www.who.int/
in this publication and they do not necessarily represent the deci- immunization monitoring/routine/immunization coverage/en/index4.html
[accessed 29.10.14] WHO/UNICEF estimates of national immunization
sions, policy or views of the World Health Organization.
coverage. Geneva: World Health Organization; 2014.
[13] de Quadros CA, Izurieta H, Venczel L, Carrasco P. Measles eradication in the
Conflicts of interest Americas: progress to date. J Infect Dis 2004;189(Suppl. 1):S227–35.
[14] Otten M, Kezaala R, Fall A, Masresha B, Martin R, Cairns L, et al. Public-health
impact of accelerated measles control in the WHO African region 2000–2003.
The authors declare that they do not have any conflict of inter- Lancet 2005;366:832–9.
ests. [15] Masresha BG, Fall A, Eshetu M, Sosler S, Alleman M, Goodson JL, et al. Measles
mortality reduction and pre-elimination in the African region, 2001–2009. J
Infect Dis 2011;204(Suppl. 1):S198–204.
Acknowledgments [16] Schenzle D. An age-structured model of pre- and post-vaccination measles
transmission. Math Med Biol 1984;1(2):169–91.
[17] Sudfeld CR, Navar AM, Halsey NA. Effectiveness of measles vaccination and
The study could not have been completed without the sup-
vitamin A treatment. Int J Epidemiol 2010;39(Suppl. 1):48–55.
port of Raymond Hutubessy and the WHO’s Initiative for Vaccine [18] Stover J, Garnett G, Seitz S, Forsythe S. The epidemiological impact of an
Research. We thank Peter Horby for providing data on contact HIV/AIDS vaccine in developing countries. World Bank Policy Research Dis-
patterns in Vietnam, and Arindam Nandi, Ashvin Ashok, Eliza- cussion Paper no. 2811. Washington, DC: Development Research Department,
World Bank; 2002.
beth Brouwer, and Zachary Olson for sourcing relevant data. We [19] Mossong J, Hens N, Jit M, Beutels P, Auranen K, Mikolajczyk R, et al. Social
received valuable comments from Laurel Ohm and participants of contacts and mixing patterns relevant to the spread of infectious diseases. PLoS
a workshop on the broader economic benefits of vaccination (WHO, Med 2008;5(3):e74.
[20] Horby P, Thai PQ, Hens N, Thi Thu Yen N, Mai LQ, Thoang DD, et al. Social contact
Geneva, 2012) and of the fourth meeting of the SAGE working group patterns in Vietnam and implications for the control of infectious diseases. PLoS
in measles and rubella (WHO, Geneva, 2013), particularly John ONE 2011;6(2):e16965.
Edmunds, Peter Strebel, Robert Perry and Susan Reef. SV was also [21] Anderson RM, May RM. Infectious diseases of humans: dynamics and control.
New York, NY: Oxford University Press; 1991.
partially supported by the Disease Control Priorities Network grant [22] United Nations, Department of Economic and Social Affairs, Population Divi-
to the University of Washington. Finally, we would like to thank sion. World population prospects: the 2010 revision. New York, NY: United
two anonymous reviewers for providing valuable and constructive Nations; 2011.
[23] World Health Organization, Available at: http://www.who.int/immunization
comments on the manuscript.
monitoring/data/data subject/en/index.html [accessed 27.07.13] Retrospec-
tive measles data on supplementary immunization activities 2000–2012; 2013.
[24] Dowdle WR. Perspectives for the elimination/eradication of disease with vac-
Appendix A. Supplementary data
cines. In: de Quadros, editor. Vaccines. Preventing disease and protecting
health. Washington, DC: Pan American Health Organization; 2004.
Supplementary data associated with this article can be [25] Shulgin B, Stone L, Agur Z. Pulse vaccination in the SIR epidemic model. Bull
found, in the online version, at http://dx.doi.org/10.1016/j.vaccine. Math Biol 1998;60:1123–48.
[26] Keeling MJ, Rohani P. Modeling infectious diseases in humans and animals.
2014.11.050. Princeton: Princeton University Press; 2007.
[27] Earn DJD, Rohani P, Bolker BM, Grenfell BT. A simple model for complex dynam-
ical transitions in epidemics. Science 2000;287:667–9.
References [28] Jit M, Brisson M. Modelling the epidemiology of infectious diseases for decision
analysis. Pharmacoeconomics 2011;29:371.
[1] United Nations Children’s Fund. World Health Organization, World Bank, [29] Anderson RM, May RM. Vaccination against rubella and measles: quantitative
United Nations. Levels and trends in child mortality – Report 2013 – estimates investigation of different policies. J Hyg 1983;90:259–325.
developed by the UN Inter-agency Group for Child Mortality Estimation. New [30] Bolker BM, Grenfell BT. Chaos and biological complexity in measles dynamics.
York, NY: United Children’s Fund; 2013. Proc R Soc Lond B 1993;251:75–81.
[2] Institute for Health Metrics and Evaluation. Global Burden of Disease Study [31] Grenfell BT, Kleczkowski A, Ellner SP, Bolker BM. Measles as a case study in
2010. Seattle: Institute for Health Metrics and Evaluation; 2013. nonlinear forecasting and chaos. Philos Trans R Soc Lond 1994;348:515–30.
[3] Liu L, Johnson HL, Cousens S, Perin J, Scott S, Lawn JE, et al. Global, regional, and [32] Keeling MJ, Grenfell BT. Understanding the persistence of measles: reconciling
national causes of child mortality: an updated systematic analysis for 2010 theory, simulation and observation. Proc R Soc Lond B 2002;269:335–43.
with time trends since 2000. Lancet 2012;379:2151–61. [33] Conlan AJ, Grenfell BT. Seasonality and the persistence and invasion of measles.
[4] Lopez AD, Begg S, Bos E. Demographic and epidemiological characteristics of Proc R Soc B 2007;274:1133–41.
major regions, 1990–2001. In: Lopez AD, Mathers CD, Ezzati M, Jamison DT, [34] Ferrari MJ, Grais RF, Bharti N, Conlan AJK, Bjornstad ON, Wolfson LJ, et al. The
Murray CJL, et al., editors. Global burden of disease and risk factors. Washington, dynamics of measles in sub-Saharan Africa. Nature 2008;451:679–84.
DC: World Bank; 2006. [35] Ferrari MJ, Grenfell BT, Strebel PM. Think globally, act locally: the role of local
[5] Bryce J, Boschi-Pinto C, Shibuya K, Black RE, the WHO Child Health Epidemiol- demographics and vaccination coverage in the dynamic response of measles
ogy Reference Group. WHO estimates of the causes of death in children. Lancet infection control. Philos Trans R Soc B 2011;368:1–7.
2005;365:1147–52. [36] Agur Z, Cojocaru L, Mazo G, Anderson RM, Danon YL. Pulse mass vaccination
[6] Simons E, Ferrari M, Fricks J, et al. Assessment of the 2010 global measles across age cohorts. Proc Natl Acad Sci U S A 1993;90(24):11698–702.
mortality reduction goal: results from a model of surveillance data. Lancet [37] Nokes D, Swinton J. Vaccination in pulses: a strategy for global eradication of
2012;379:2173–8. measles and polio? Trends Microbiol 1997;5(1):14–9.
[7] Orenstein WA, Hinman AR, Strebel PJ. Measles: the need for 2 opportunities for [38] Bauch CT, Szusz E, Garrison LP. Scheduling measles vaccination in low-income
prevention. Clin Infect Dis 2006;42:320–1. countries: projections of a dynamic model. Vaccine 2009;27:4090–8.