A Comparative Analysis of InfluenzaVaccination ProgramsShweta Bansal1, Babak Pourbohloul2, Lauren Ancel Meyers3,4*1 Computational and Applied Mathematics, University of Texas Austin, Austin, Texas, United States of America, 2 UBC Centre for Disease Control, University of BritishColumbia, Vancouver, British Columbia, Canada, 3 Section of Integrative Biology and Institute for Cellular and Molecular Biology, University of Texas Austin, Austin, Texas,United States of America, 4 External Faculty, Santa Fe Institute, Santa Fe, New Mexico, United States of AmericaFunding: We acknowledge thefinancial support of the CanadianInstitutes of Health Research, theSanta Fe Institute, and a NASAHarriett G. Jenkins Fellowship to SB.The funding agencies had no role instudy design, data collection andanalysis, decision to publish, orpreparation of the manuscript.Competing Interests: The authorshave declared that no competinginterests exist.Academic Editor: Bryan Grenfell,Pennsylvania State University,United States of AmericaCitation: Bansal S, Pourbohloul B,Meyers LA (2006) A comparativeanalysis of influenza vaccinationprograms. PLoS Med 3(10): e387.DOI: 10.1371/journal.pmed.0030387Received: October 13, 2005Accepted: July 13, 2006Published: October 3, 2006DOI:10.1371/journal.pmed.0030387Copyright: Ó 2006 Bansal et al. Thisis an open-access article distributedunder the terms of the CreativeCommons Attribution License, whichpermits unrestricted use,distribution, and reproduction in anymedium, provided the originalauthor and source are credited.Abbreviations: CDC, United StatesCenters for Disease Control andPrevention* To whom correspondence shouldbe addressed. E-mail: [email protected] threat of avian influenza and the 2004–2005 influenza vaccine supply shortage in theUnited States have sparked a debate about optimal vaccination strategies to reduce the burdenof morbidity and mortality caused by the influenza virus.Methods and FindingsWe present a comparative analysis of two classes of suggested vaccination strategies:mortality-based strategies that target high-risk populations and morbidity-based strategies thattarget high-prevalence populations. Applying the methods of contact network epidemiologyto a model of disease transmission in a large urban population, we assume that vaccinesupplies are limited and then evaluate the efficacy of these strategies across a wide range ofviral transmission rates and for two different age-specific mortality distributions.We find that the optimal strategy depends critically on the viral transmission level(reproductive rate) of the virus: morbidity-based strategies outperform mortality-ba sedstrategies for moderately transmissible strains, while the reverse is true for highly transmissiblestrains. These results hold for a range of mortality rates reported for prior influenza epidemicsand pandemics. Furthermore, we show that vaccination delays and multiple introductions ofdisease into the community have a more detrimental impact on morbidity-based strategiesthan mortality-based strategies.ConclusionsIf public health officials have reasonable estimates of the viral transmission rate and thefrequency of new introductions into the community prior to an outbreak, then these methodscan guide the design of optimal vaccination priorities. When such information is unreliable ornot available, as is often the case, this study recommends mortality-based vaccination priorities.The Editors’ Summary of this article follows the references.PLoS Medicine | www.plosmedicine.org October 2006 | Volume 3 | Issue 10 | e3871816PLoSMEDICINEIntroductionIn response to the 2004–2005 influenza vaccine shortage,the United States Centers for Disease Control and Prevention(CDC) restricted vaccines to those most at risk for hospital-ization and death — healthy infants, elderly individuals, andindividuals with chronic illnesses. This strategy may be limitedby the failure of vaccines to yield adequate protection forhigh-risk individuals [1,2] and the lesser roles played by infantsand the elderly in disease transmission—they typically do notintroduce influenza into households or other social groups.Influenza outbreaks are believed to hinge, instead, ontransmission by healthy school c hildren [3–6], collegestudents, and employed adults who have many daily contactsand are highly mobile [7]. Thus, epidemiologists havesuggested an alternative approach: vaccinate school-agechildren to slow the spread of disease and thereby indirectlydecrease mortality [8,9]. Several studies support this strategy.Monto et al. immunized school children in Tecumseh,Michigan, with inactivated influenza vaccine in 1968 andfound lower total morbidity than in a matching communityduring a wave of influenza A (H3N2) [10]. Reichart et al. arguethat mandatory influenza vaccination of school children inJapan from 1962 to 1987 reduced incidence and mortalityamong the elderly [11]. Recently, Longini et al. usedmathematical models to show that, under certain assump-tions, vaccinating 80% of all school-age children is almost aseffective as vaccinating 80% of the entire population [8].School-based vaccination programs have the additionalbenefits of high coverage, high efficacy, and minimal sideeffects [12].In a similar spirit, others have suggested contact-basedpriorities that target individuals with the highest numbers ofpotentially disease-causing contacts [13–15]. This assumesthat vulnerability is directly proportional to the number ofcontacts, and that removing the most vulnerable individualsfrom the transmission chain will maximally decrease diseasespread. Identifying high-contact individuals in a community,however, may be difficult in practice.Here we apply tools from contact network epidemiology[16–19] to evaluate vaccination strategies for a spectrum ofinfluenza strains when vaccine supplies are limited. We use arealistic model of contact patterns in an urban setting tocompare mortality- based strategies that target high -riskindividuals to morbidity-based strategies that target demo-graphics with high attack rates. We assess the efficacy of thesemeasures for two substantially different virulence patterns,one based on mortality estimates from annual influenzaepidemics and the other based on mortality estimates fromthe 1918 influenza pandemic. In addition, we consider theimpact of vaccination delay and multiple imported cases onthe relative effectiveness of the